Corporate Governance and Finance, Volume 8 (Advances in Financial Economics, Volume 8)

LIST OF CONTRIBUTORS Christopher W. Anderson

School of Business, University of Kansas, USA

Gurmeet S. Bhabra

University of Otago, New Zealand

Terry L. Campbell II

College of Business and Economics, University of Delaware, USA

Stephen P. Ferris

Department of Finance, University of Missouri-Columbia, USA

Kathleen P. Fuller

Department of Banking and Finance, University of Georgia and University of Michigan Business School, USA

Michael B. Glatzer

University of Georgia, USA

Mark Hirschey

School of Business, University of Kansas, USA

Narayanan Jayaraman

DuPree College of Management, Georgia Institute of Technology, USA

Teresa A. John

Stern School of Business, New York University, USA

Gershon N. Mandelker

Katz Graduate School of Business, University of Pittsburgh, USA

John D. Martin

Hankamer School of Business, Baylor University, USA

Jeffry M. Netter

Department of Banking and Finance, University of Georgia, USA

Annette B. Poulsen

Department of Banking and Finance, University of Georgia, USA vii

viii

Akin Sayrak

Joseph M. Katz Graduate School of Business, University of Pittsburgh, USA

Susan Scholz

School of Business, University of Kansas, USA

Nilanjan Sen

Nanyang Technological University, Singapore

Gopala K. Vasudevan

School of Management, Northeastern University, USA

Peng Peck Yen

Nanyang Technological University, Singapore

BANK MONITORING, FIRM PERFORMANCE, AND TOP MANAGEMENT TURNOVER IN JAPAN Christopher W. Anderson, Terry L. Campbell II, Narayanan Jayaraman and Gershon N. Mandelker ABSTRACT An inverse relation between performance and managerial turnover at Japanese firms suggests that bank monitoring substitutes for other governance mechanisms (Kaplan, 1994; Kang & Shivdasani, 1995). Morck and Nakamura (1999), however, report that Japanese banks protect their self-interests as creditors rather than the interests of shareholders when appointing corporate directors. We re-examine data on top management changes at Japanese firms and find results consistent with this latter notion. Specifically, management turnover is conditionally related to a firm’s ability to meet its short-term obligations rather than profitability or stock returns. Bank monitoring is therefore not a substitute for mechanisms that directly serve shareholders’ interests.

1. INTRODUCTION U.S.-style mechanisms for corporate control are relatively weak or non-existent in Japan. For example, the ownership of many Japanese firms is characterized by cross holding of shares by industrial and financial groups, and the market for Corporate Governance and Finance Advances in Financial Economics, Volume 8, 1–27 © 2003 Published by Elsevier Science Ltd. ISSN: 1569-3732/PII: S1569373203080010

1

2

CHRISTOPHER W. ANDERSON ET AL.

takeovers is relatively inactive (Aoki, 1992; Kester, 1991). Outside directorships are also rare in Japan (Ballon & Tomita, 1988; Kaplan, 1994), as are individual block holders (Nishiyama, 1984; Prowse, 1992). An inactive takeover market, the prevalence of firm grouping, and the absence of individual blockholders imply that management is insulated from monitoring and control by shareholders (Nishiyama, 1984). On the other hand, it has been argued that monitoring by commercial banks substitutes for shareholder governance (Aoki, 1990, 1992; Sheard, 1989). Specifically, the intimacy of bank-client relationships, coupled with the bank’s authority to intervene during periods of financial distress, may serve as a mechanism that monitors firm performance and disciplines management. Kaplan (1994, 1997), Kaplan and Minton (1994), and Kang and Shivdasani (1995) report an inverse relation between Japanese firm performance and managerial turnover that is seemingly consistent with this notion. Since the collapse of asset prices in Japan in the early 1990s, the ensuing recession, and the onset of the banking crisis, the role of Japanese banks in the affairs of their client firms has been critically re-examined. Kang and Stulz (2000) find that bank-dependent Japanese firms experience worse stock price performance than other firms during the 1990s. Gibson (1995) provides evidence that troubled Japanese banks inefficiently ration credit among their borrowers. Kang, Shivdasani and Yamada (2000) report that benefits to bank monitoring of investment decisions by client firms appear to dissipate in the early 1990s when banks themselves became troubled. Finally, Morck and Nakamura (1999) suggest that bank intervention by means of appointments to boards of directors of client firms is in the short-term self-interest of the bank rather than the interests of shareholders of such firms. These studies question whether monitoring by Japanese banks effectively substitutes for mechanisms of corporate control that directly safeguard shareholder wealth. The purpose of this study is to examine these two conflicting views of Japanese corporate governance mechanisms by investigating the determinants of top management change. Specifically, we investigate the extent to which measures that are directly related to creditors’ interest, such as liquidity and leverage, explain top management changes at Japanese firms as opposed to profitability or stock returns. We examine data on top management changes and performance at 207 Japanese firms from 1984 to 1989, a period that precedes the collapse of asset prices and ensuing recession in the 1990s but which matches sample periods of earlier studies of management incentives in Japan (Kaplan, 1994; Kang & Shivdasani, 1995). We first confirm the findings of these earlier studies that report an inverse relation between profitability and managerial turnover for similar sample periods. We then focus on measures of a firm’s ability to meet its short-term obligations and the degree to which they predict top management changes. Our findings suggest

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

3

that firms with difficulties meeting short-term obligations display a higher rate of top management change. Furthermore, we find that liquidity measures largely dominate unconditional measures of profitability when examined simultaneously. Specifically, while measures of liquidity and indebtedness are correlated with profitability, when we isolate susceptibility to creditor monitoring in particular from poor earnings performance in general, we find that it is the former notion that matters most in influencing the likelihood of top management changes. For example, independent of industry-adjusted earnings, firms with low interest coverage ratios have significantly higher rates of top management change than firms with high interest coverage ratios. In contrast, firms with high interest coverage ratios but with relatively poor earnings performance display rates of managerial change similar to those of profitable firms. These results suggest that top management change at Japanese firms is triggered specifically by endangerment of creditors’ interests rather than poor profitability in general. Conversely, managers of firms with sufficient liquidity to appease their creditors are not disciplined for poor firm performance. Bank monitoring, therefore, does not substitute for other governance mechanisms that directly safeguard the interests of shareholders. Instead, our findings are consistent with the hypothesis that the Japanese corporate governance system lacks mechanisms that vigorously protect shareholder interests. The remainder of this paper is organized as follows. Section 2 briefly reviews features of the Japanese economy and their implications for corporate governance. Section 3 discusses our data and methodology. Section 4 presents our results, and Section 5 concludes.

2. CORPORATE GOVERNANCE OF JAPANESE FIRMS Japanese corporations seem to lack several governance mechanisms that function in the U.S. to mitigate problems associated with the separation of ownership from management. First, large individual or family shareholders are rare in Japan (Kaplan, 1994; Prowse, 1992), and large corporate shareholders, especially banks and insurance companies, are themselves diffusely held (Nishiyama, 1984). Second, outside directors are observed relatively infrequently; Ballon and Tomita (1988) claim that 43.5% of large Japanese companies have no outside directors, while Kaplan (1994) cites 60% for his sample of 119 firms. Coupled with an absence of independent corporate boards and powerful individual shareholders is an inactive takeover market.1 One obstacle to takeovers among large Japanese firms is corporate grouping or keiretsu. A keiretsu is a group of affiliated companies linked by crossholding of shares and the predominance of a common large bank in the financial dealings of member firms. The stable cumulative

4

CHRISTOPHER W. ANDERSON ET AL.

shareholdings of keiretsu members, including the so-called main bank, are large enough to render hostile takeovers impossible (Aoki, 1990). These features of the Japanese corporate system have led some observers to question whether Japanese managers face any effective control mechanisms (Nishiyama, 1984). In contrast, many authors cite the critical role played by banks in Japan and suggest that creditor monitoring is an effective alternative mechanism that promotes corporate efficiency (Aoki, 1990, 1992; Sheard, 1989). Close ties between firms and banks may help resolve information asymmetries and conflicts of interest between firms and suppliers of capital (Horiuchi, Packer & Fukada, 1988; Hoshi, Kashyap & Scharfstein, 1990a, 1991; Prowse, 1990), and provide an efficient means for resolving financial distress (Hoshi, Kashyap & Scharfstein, 1990b; Suzuki & Wright, 1985). Most importantly, it is argued that monitoring by financial institutions effectively substitutes for the missing takeover market as a mechanism to discipline top management. In the case of keiretsu firms, for example, Sheard (1989) characterizes the main bank as exploiting its informational assets derived from being a primary supplier of capital to the firm to serve as a specialized monitor of managerial actions. In situations where corporate performance is inadequate, the bank may intervene by stipulating policy adjustments, limiting managerial discretion, lending managerial talent, and even calling for top management changes.2 Evidence on changes in Japanese boards of directors also suggests that banks place directors representing their interest on corporate boards in response to poor performance (Kaplan & Minton, 1994). Studies by Kaplan (1994, 1997) and Kang and Shivdasani (1995) find an inverse relation between managerial turnover and stock returns or accounting income.3 On the basis of this empirical relation, Kaplan and Ramseyer (1996) conclude that the notion of Japanese managers acting without regard for shareholders is “another fable for the academy.” The role of creditor monitoring in Japan, as discussed by Aoki (1990, 1992), for example, raises several questions regarding the extent to which monitoring by banks substitutes for mechanisms that directly protect shareholders’ welfare. First, performance benchmarks for managers may be those that are of keen interest to creditors such as the firm’s ability to meet its interest and principal payment schedules. Second, the link between corporate performance and top management change may be weak for firms performing above a certain threshold if only firms in financial distress experience creditor-induced managerial turnover. Finally, firms with a high exposure to creditor monitoring, i.e. firms which are highly leveraged in general and heavily indebted to banks in particular, may display the strongest relation between performance and turnover. Morck and Nakamura (1999) test several implications of this perspective by examining appointments of directors to Japanese boards. They report that appointments of directors by banks are primarily related to the short-term

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

5

interests of the bank, measured, for instance, by liquidity. Kaplan and Minton (1994) suggest that such appointments presage executive turnover, but Morck and Nakamura do not directly investigate management turnover. The objective of our study is to examine the hypothesis that exposure to creditor monitoring triggered by poor liquidity better predicts top management changes rather than unconditional corporate performance.

3. SAMPLE AND DATA DESCRIPTION We construct a sample of Japanese firms by cross-referencing Japanese company listings in Moody’s International with companies listed on the University of Rhode Island’s Pacific Basin Capital Markets Database (PACAP) for fiscal years from 1983 through 1989. This sample comprises 207 firms, although for any year the number of firms represented is less than 207 due to data deficiencies in Moody’s. By nature of the selection criteria for coverage by Moody’s, the sample firms are among the largest and best known firms in Japan. The sample contains many firms that are smaller than the 119 firms composing the sample of a similar study by Kaplan (1994). For fiscal year 1986, the 207 sample firms have mean (median) sales of $3.74 billion ($1.14 billion), total assets of $2.73 billion ($1.27 billion), and market value of equity of $2.06 billion ($1.22 billion).

3.1. Top Management Change We identify executives with the title chairman, president, or chief executive officer as listed in Moody’s for 1983 through 1989. We classify the firm as having experienced top management change if any previously listed top executive ceases to be listed as a top executive in the following year.4 A mere shuffling of titles among executives without any executive ceasing to be listed is not classified as a management change. This classification scheme is similar to that of Warner, Watts and Wruck (1988). The resulting sample of firm years totals 1,086 observations. Table 1 reports management changes by year, industry, and keiretsu versus non-keiretsu classification. Panel A reports the frequency of management changes by fiscal year. The annual frequency is relatively constant, with only 1986 having an unusually high number of management changes. The average annual frequency of 13.4% is similar to the turnover frequencies reported for samples of U.S. firms by various authors as well as the 14.5% rate cited by Kaplan (1994) and the 12.9% reported by Kang and Shivdasani (1995) for samples of Japanese firms. Panel B demonstrates that the sample is well distributed across industry

6

CHRISTOPHER W. ANDERSON ET AL.

Table 1. Observed Frequency of Top Management Change.a Fiscal-year Observations

Management Changes

Annual Frequency

164 181 186 185 189 181 1,086

19 20 36 25 26 20 146

11.6% 11.0% 19.4% 13.5% 13.8% 11.0% 13.4%

B. By industry Textiles Pulp & paper Chemicals Petroleum & rubber Glass & ceramics Iron & steel Other metals Machinery Electrical machinery Transportation equipment Precision equipment Other manufacturing Wholesale Retail Transportation & shipping

63 35 168 23 42 34 47 108 173 112 52 72 50 79 28

9 6 22 2 10 5 8 10 25 17 6 2 11 6 7

14.3% 17.1% 13.1% 8.7% 23.8% 14.7% 17.0% 9.3% 14.5% 15.2% 11.5% 2.8% 22.0% 7.6% 25.0%

C. By keiretsu classification Keiretsu (69 firms) Non-keiretsu (138 firms)

385 701

53 93

13.8% 13.3%

A. By fiscal year 1984 1985 1986 1987 1988 1989 All years

a This table reports observed managerial change for 207 firms composing a sample of 1,086 fiscal years

from 1984 to 1989. Management change is defined as the delisting of any top executives (chairman, president, or CEO) previously listed as such by Moody’s International. The number of observations per fiscal year reported in panel A varies according to data availability for sample firms in Moody’s. Industry classification for panel B is according to the PACAP Database for Japan. Keiretsu classification in panel C is by Nakatani (1984).

groups, but it also suggests considerable industry variation in the incidence of top management change. Analysis of these patterns rejects the hypothesis that turnover likelihood is identical across all industries. All analyses were reproduced after including industry-specific dummy variables with results similar to those reported. For brevity, further discussion of industry patterns is omitted. Panel C of Table 1 compares top management changes in the keiretsu sample to that of the

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

7

non-keiretsu firm sample.5 The keiretsu and non-keiretsu firm samples comprise 69 and 138 firms, with 385 and 701 firm fiscal years, respectively. The rates of management change are similar (13.8% for keiretsu members compared to 13.3% for non-keiretsu firms), and a chi-squared test fails to reject the hypothesis that they are identical.

3.2. Firm Performance We compute performance measures for sample firms using the PACAP Database for Japan. First, we calculate return on assets (ROA), defined as earnings before interest and taxes divided by total assets. We then subtract the equally-weighted matching industry average return on assets from this measure for each observation and derive an industry-adjusted return on assets referred to subsequently as IROA. Weisbach (1988), Barro and Barro (1990), and Kaplan (1994) also use performance measures based on return on assets in studies concerning top management incentives. We also compute profit margin on sales (IPM), defined as operating income scaled by sales, adjusted by industry averages. Scaling earnings by sales may be more appropriate to the extent that land and financial asset appreciation, accelerated depreciation, and other factors distort book value of assets. Stock returns are the performance measures of choice for a host of studies focusing on U.S. firms (Coughlan & Schmidt, 1985; Gibbons & Murphy, 1990; Jensen & Murphy, 1990; Warner, Watts & Wruck, 1988). We calculate a number of alternative stock return-based performance measures, including raw returns, multiple-period cumulative raw returns, returns net of value weighted or equally weighted return indices, and returns net of industry average returns. Among these, the measures with the strongest relation to management changes are raw returns. We report results using raw returns, cumulated over the two years prior to the observation year (RET2) and the four years prior to the observation year (RET4). We examine whether variables that measure exposure to creditor pressure better explain the incidence of top management change in Japan than profitability or stock returns. This hypothesis is motivated by descriptive characterizations of Japanese corporate governance (e.g. Aoki, 1992) and the findings of Morck and Nakamura (1999) with respect to director appointments.6 We utilize several alternative measures of corporate liquidity and leverage. First, we calculate the interest coverage ratio (COV) as operating income scaled by interest expense. Hoshi, Kashyap and Scharfstein (1990b) use such a measure as an indicator of financial distress. We also compute a firm’s quick ratio (QUICK) as current assets minus inventory divided

8

CHRISTOPHER W. ANDERSON ET AL.

Table 2. Summary Statistics for Firm-specific Variables.a Variable

Mean

Median

Std. Dev.

Sales (Yen billion) Assets (Yen billion) Market Value Equity (Yen billion) ROA – return on assets IROA – industry adjusted return on assets PM – profit margin on sales IPM – industry adjusted profit margin RET2 – stock return, prior two years RET4 – stock return, prior four years COVb – operating income to interest expense QUICK – current assets minus inventory to current liabilities DTV – total debt to value (debt + market equity) LTV – loans to value (debt + market equity)

944 534 422 0.079 0.009 0.057 0.010 0.488 1.057 6.287b 1.311

282 242 215 0.071 0.003 0.050 0.004 0.344 0.715 3.642 1.039

2,429 802 589 0.041 0.037 0.053 0.048 0.673 1.357 6.103b 0.876

0.428 0.143

0.399 0.087

0.216 0.156

a This

table provides the mean, median, and standard deviation of key variables for a sample of 1,086 fiscal years from 207 firms over 1984–1989. All data is from the PACAP Database for Japan. Industry adjusted variables are calculated by subtracting the fiscal-year industry mean value of a given variable from the raw value. Industry classification is by the PACAP two-digit code and the entire PACAP universe of firms is sampled to derive industry means. b To reduce distortions by extreme outliers, interest coverage observations above the 5% tail of the distribution have been set to the 95% level of the distribution, or about 20.0. Without truncation the mean and standard deviation of COV are 128.3 and 1050.9, respectively.

by current liabilities. The quick ratio is a stock measure of liquidity as opposed to a flow measure such as the interest coverage ratio. We hypothesize that less liquid firms will experience higher rates of management change. In addition to these measures of liquidity, we also calculate two measures of leverage: debt to firm value (DTV), defined as book value of debt divided by total firm value (market value of equity plus book value of debt), and loans to value (LTV), defined as total loans outstanding divided by firm value. Firms that have high leverage and especially high bank loan leverage are more likely to be actively monitored by creditors (Aoki, 1992; Hodder & Tschoegl, 1992). In contrast, less highly leveraged firms are unlikely candidates for active monitoring and direct intervention. Table 2 provides the means, medians, and standard deviations for these performance variables, as well as for sales, total assets, and market value of equity. Table 3 reports correlations among variables. As discussed above, the average size of sample firms is relatively large, as mean (median) sales, assets, and market value of equity are 944 (282) billion yen, 534 (242) billion yen, and 422 (215) billion yen, respectively. Measures of earnings performance are well behaved, with

Sales Assets MVE ROA IROA PM IPM RET2 RET4 COV QUIK DTV LTV

Sales

Assets

MVE

ROA

IROA

PM

IPM

RET2

RET4

COV

QUIK

DTV

LTV

– 0.736 0.298 −0.098 −0.065 −0.197 −0.133 −0.054 −0.068 −0.120 −0.107 0.311 0.228

0.736 – 0.588 −0.153 −0.092 −0.191 −0.126 −0.032 −0.059 −0.179 −0.188 0.372 0.284

0.298 0.588 – 0.119 0.157 0.086 0.083 0.188 0.217 0.152 −0.014 −0.217 −0.191

−0.098 −0.153 0.119 – 0.941 0.806 0.773 0.170 0.288 0.722 0.305 −0.529 −0.411

−0.065 −0.092 0.157 0.941 – 0.778 0.820 0.135 0.255 0.693 0.360 −0.504 −0.376

−0.197 −0.191 0.086 0.806 0.778 – 0.955 0.208 0.318 0.593 0.376 −0.497 −0.330

−0.133 −0.126 0.083 0.773 0.820 0.955 – 0.164 0.278 0.572 0.388 −0.423 −0.265

−0.054 −0.032 0.188 0.170 0.135 0.208 0.164 – 0.720 0.105 −0.081 −0.233 −0.063

−0.068 −0.059 0.217 0.288 0.255 0.318 0.278 0.720 – 0.163 −0.044 −0.319 −0.132

−0.120 −0.179 0.152 0.722 0.693 0.593 0.572 0.105 0.163 – 0.566 −0.686 −0.619

−0.107 −0.188 −0.014 0.305 0.360 0.376 0.388 −0.081 −0.044 0.566 – −0.515 −0.457

0.311 0.372 −0.217 −0.529 −0.504 −0.497 −0.423 −0.233 −0.319 −0.686 −0.515 – 0.842

0.228 0.284 −0.191 −0.411 −0.376 −0.330 −0.265 −0.063 −0.132 −0.619 −0.457 0.842 –

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

Table 3. Correlation Matrix for Firm-specific Variables.

9

10

CHRISTOPHER W. ANDERSON ET AL.

industry-adjusted return on assets (IROA) and profit margin (IPM) averaging close to zero without excessively wide dispersion. The return measures presented (RET2 and RET4) seem quite large (i.e. median return for the two prior years is 34.4%, median return for the four prior years is 71.5%). However, given that our sample period includes the dramatic mid-to-late 1980s rise in the Japanese stock market, the magnitude of these measures is not surprising. Variables which proxy for exposure to creditor monitoring are also reasonably distributed with the exception of interest coverage (COV), which has a skewed sample distribution due to a small number of extreme, upper tail outliers. Due to the presence of these outliers (i.e. the 5% of the observations with interest coverage ratios which exceed about 20), we set 20.0 as the upper limit for the COV variable in all reported analyses. Not surprisingly, the four measures for creditor monitoring are highly correlated as reported in Table 3.

4. DETERMINANTS OF TOP MANAGEMENT CHANGE We utilize the sample of firm-year management changes and corresponding firm performance measures to estimate several permutations of a logit equation of the form:
 Pr(management change) ln (1) = a + X B + e 1 − Pr(management change) where Pr(·) indicates probability, a is a scalar, X is a vector of performance measures and firm characteristics, B is the corresponding vector of coefficients, and e is a zero mean error term. We use maximum likelihood techniques to estimate Eq. (1) (see Judge, Griffiths, Hill, Lutkepohl & Lee, 1985). Results from an ordinary least squares estimation of a linear probability model are similar to the reported logit results. To illustrate the economic significance of the estimated coefficients from Eq. (1), we estimate the implied probability of top management change conditional on certain values of the independent variables via the following transformation of the estimated form of Eq. (1): Pr(management change/X) =

exp(a + X  B) 1 + exp(a + X  B)

(2)

To gauge the economic significance of the estimated empirical relation we frequently calculate the implied change in probability of top management change that results from a one standard deviation change (plus or minus) in a performance measure from its mean.

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

11

4.1. Impact of Profitability and Stock Returns on Top Management Change Table 4 presents estimated logit equations of top management change conditional on firm performance as measured by return on assets, profit margin, and stock returns. The first two columns of Table 4 provide estimated logit equations that explain the likelihood of top management change as a function of industry adjusted return on assets (IROA) and industry adjusted profit margin (IPM). These two measures are calculated for the fiscal year immediately prior to the observation year. Similar results are obtained when observation-year measures are used or when the measures are averaged across these two years. The results in these columns indicate that top management changes in Japan are significantly related to earnings performance, as the null hypotheses that management changes are unrelated to IROA or IPM is rejected at the 1% level. The estimated relations appear economically significant as well; the standardized coefficient estimates imply that a one standard deviation increase (decrease) in IROA from the sample mean results in a decrease (increase) in management change probability of 3.5% (4.6%) from the 12.9% rate predicted for a firm with mean IROA performance.7 The analogous probabilities associated with a one standard deviation change in IPM from the mean are 4.0% (5.5%) from a base level of 12.8% for a firm with mean IPM. In order to gauge the fit of the estimated logit models and the economic significance of the observed relation between performance and top management change, Table 5 presents actual and estimated incidence of top management changes across performance quintiles. Panels A and B of Table 5, for example, present the actual and predicted rates of top management change by IROA and IPM quintiles. A comparison of actual and predicted rates of management change by performance quintile indicates that the logit model estimations in columns (1) and (2) of Table 4 are not driven by outliers and, in fact, fit the data rather well. For IROA, the observed frequency of management change increases monotonically as performance falls. For IPM, the relation is nearly monotonic. Chi-squared tests for differences across performance quintiles confirm these inferences. For both IROA and IPM, the incidence of top management change is significantly different: (i) between the best performing quintile and the worst performing quintile; (ii) between firms with performance above the median and firms with below median performance; and (iii) across all performance quintiles. The observed and model-implied differences in the rate of management changes between the best and worst performing firms are greater for IPM than for IROA, perhaps implying that scaling earnings by annual sales captures relative performance better than scaling by total assets, but the high degree of correlation between the variables impedes further investigation of this question.

12

Table 4. Likelihood of Top Management Change Based on Profitability and Stock Returns.a Variable INTERCEPT

a This

−1.821 [−1.906] (0.0000) −9.612 [−0.359] (0.0004) –

–

–

13.53 (0.0002) 1,086

(2) −1.830 [−1.919] (0.0000) –

−8.832 [−0.421] (0.0001) –

–

17.67 (0.0000) 1,086

(3) −1.787 [−1.867] (0.0000) –

–

−0.164 [−0.111] (0.2544) –

1.37 (0.2411) 1,079

(4) −1.716 [−1.855] (0.0000) –

–

–

−0.131 [−0.178] (0.1049) 3.01 (0.0826) 1,062

(5) −1.780 [−1.907] (0.0000) −9.341 [−0.347] (0.0007) –

−0.097 [−0.065] (0.5037) –

13.78 (0.0010) 1,079

(6) −1.743 [−1.887] (0.0000) −8.776 [−0.324] (0.0019) –

–

−0.068 [−0.092] (0.3957) 13.30 (0.0013) 1,062

(7)

(8)

−1.799 [−1.921] (0.0000) –

−1.768 [−1.90] (0.0000) –

−8.760 [−0.418] (0.0001) −0.071 [−0.048] (0.6303) –

−8.400 [−0.398] (0.0002) –

18.21 (0.0001) 1,079

−0.049 [−0.067] (0.5355) 17.56 (0.0002) 1,062

table presents estimated logit equations where the dependent variable is equal to one when a top executive (chairman, president, or CEO) as reported by Moody’s International ceases to be listed over 1984–1989. Independent variables are: IROA – industry adjusted return on assets in the prior fiscal year; IPM – industry adjusted profit margin on sales in the prior fiscal year; RET2 – cumulative stock return in the two years prior to the current fiscal year; and RET4 – cumulative stock return over the four prior years. Numbers in square brackets are the coefficients on standardized independent variables. Numbers in parentheses are p-values associated with t-tests of the coefficient estimates.

CHRISTOPHER W. ANDERSON ET AL.

IROA (industry adjusted return on assets) IPM (industry adjusted profit margin) RET2 (stock return, year t − 2 to t − 1) RET4 (stock return, year t − 4 to t − 1) Chi-squared (p-value) Observations

(1)

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

13

Table 5. Predicted and Observed Managerial Change Based on Profitability and Stock Returns.a Quintile Median

Management Change Predicted

Observed

A. Management change conditional on IROA IROA Quintile High 0.049 9.2% 10.1% 2 0.019 11.9% 10.6% 3 0.003 13.5% 13.4% 4 −0.009 15.0% 15.2% Low −0.026 17.2% 18.0% Chi-squared tests of observed turnover: high quintile = low: 5.67∗∗ ; above median = below: 6.24∗∗ ; high = 2 = 3 = 4 = low: 7.96∗ B. Management change conditional on IPM IPM Quintile High 0.058 8.8% 6.4% 2 0.022 11.7% 13.4% 3 0.004 13.4% 12.4% 4 −0.011 15.0% 16.1% Low −0.030 17.3% 18.9% Chi-squared tests of observed turnover: high quintile = low: 15.91∗∗∗ ; above median = below: 6.93∗∗∗ ; high = 2 = 3 = 4 = low: 17.62∗∗∗ C. Management change conditional on RET2 RET2 Quintile High 1.264 12.0% 11.1% 2 0.656 13.1% 11.6% 3 0.344 13.7% 16.7% 4 0.093 14.2% 15.8% Low −0.147 14.6% 12.0% Chi-squared tests of observed turnover: high quintile = low: 0.09; above median = below: 1.32; high = 2 = 3 = 4 = low: 4.92 D. Management change conditional on RET4 RET4 Quintile High 2.520 11.4% 11.7% 2 1.226 13.3% 11.8% 3 0.715 14.1% 12.2% 4 0.329 14.7% 17.0% Low −0.040 15.3% 15.6% Chi-squared tests of observed turnover: high quintile = low: 1.33; above median = below: 3.53∗ ; high = 2 = 3 = 4 = low: 4.23 ∗∗∗ ∗∗ ∗

,

, significantly different from zero at the 1%, 5%, and 10% levels, respectively.

a Each

panel displays the sample distribution of a performance measure and turnover probabilities across performance quintiles. Performance measures are: IROA – industry adjusted return on assets in prior fiscal year; IPM – industry adjusted profit margin on sales in prior fiscal year; RET2 – cumulative stock return in the two prior fiscal years; RET4 – cumulative stock return over the four prior fiscal years. Predicted turnover is obtained by using the estimated logit equations in Table 4 conditional on quintile median performance. Tests statistics for differences in observed turnover rates are chi-squared tests with one, one, and four degrees of freedom, respectively.

14

CHRISTOPHER W. ANDERSON ET AL.

Columns (3) and (4) in Table 4 provide estimates of the relation between the likelihood of top management change and stock return performance. In addition to raw returns cumulated over the two-year (RET2) and four-year (RET4) periods preceding an observation year, we also investigated market-adjusted returns, industry-adjusted returns, and multiple-year adjusted returns. The only results that approach statistical significance are those for raw returns. In Table 4, for example, the coefficient estimate on raw returns cumulated over the four-year period prior to an observation year (RET4) borders on statistical significance ( p-value = 10.5%). The economic significance of the impact of stock returns on top management change is low, however. The standardized coefficients on two-year returns (−0.111 as reported in column 3 of Table 4) and four-year returns (−0.178 as reported in column 4) are smaller than the standardized coefficients for the earnings based variables (e.g. −0.421 for IPM as reported in column 2). This implies that a one standard deviation change in stock return has a much smaller impact on probability of top management change than a one standard deviation change in IROA or IPM. Second, when we examine observed top management change across performance quintiles based on RET2 and RET4 (panels C and D of Table 5) we find that the differences in rates of management change across quintiles are narrow, prone to non-monotonicities, and, with the exception of the above versus below median performance test for RET4 in panel D, statistically insignificant. Third, when we include both an earnings based measure of performance and a stock return based measure of performance (columns (5)–(8) of Table 4), estimated coefficients on stock return measures, although of the hypothesized sign, are smaller in magnitude and are statistically indistinguishable from zero. In short, we find little evidence in support of a relation between stock returns and top management change. These results on stock returns seem at first to contrast with those of Kaplan (1994) and Kang and Shivdasani (1995), who find a significant relation between top management change and raw stock returns for different samples of Japanese firms. However, upon closer examination our results can be reconciled. First, Kaplan’s sample is composed of the largest 119 Japanese firms with average annual sales of $5.6 ($3.5) billion, while our sample comprises 207 firms with mean (median) annual sales of $3.7 ($1.1) billion. When we restrict our sample to include only the largest 119 firms in any given year or only firms with above sample medium sales, we find that the coefficient on RET4, which borders on conventional significance in Table 4, becomes statistically significant at the 10% level. Second, our results are consistent with Kaplan’s in that only earnings-based variables have robust explanatory power. Specifically, in both Kaplan’s tables and ours the magnitude of the coefficients on earnings variables is much larger than that for stock return variables. Furthermore, Kaplan finds that stock returns are not significant in a multivariate framework in which earnings variables are also included. This latter finding is

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

15

consistent with columns (5)–(8) of Table 4. Finally, differences in definitions of turnover and sample construction partially explain the difference in results. Kaplan finds a significant relation between top management change and stock returns for a very narrow definition of top management change – instances where a Japanese president ceases to be president yet does not assume the chairmanship, a category that contains only about one fourth of all of Kaplan’s observations of top management change. When Kaplan utilizes a broader definition of turnover, top executive change is not significantly related to stock returns. Our definition of top management change is more refined than Kaplan’s broad definition, yet not as restrictive as the “president does not become chairman” definition. For example, our definition does not include instances where chairman/president relinquishes the presidency to a newcomer but keeps the chairmanship; Kaplan’s broad definition regards this as turnover, his narrow definition does not. In short, while we can replicate Kaplan’s finding regarding stock returns on a restricted sample of larger firms and using a particular return measure, our conclusion is that stock returns have marginal explanatory power when compared to earnings-based measures of performance. Kang and Shivdasani (1995) suggest that the weak relation between returns and top management change may be partially due to imperfections in identification of the turnover year due to reliance on Moody’s or failure to distinguish between routine and non-routine managerial turnover. These suggestions do not explain, however, why we fail to find economic and statistical significance for stock returns yet find significant results for accounting-based measures. Second, while we consider a departing chairman succeeded by a sitting president as top management change, we do not identify a situation where a chair/president relinquishes only the latter title as a management change. Consequently, the routine versus non-routine distinction made by Kang and Shivdasani is somewhat, but not fully, reflected in our definition of management change. Finally, Kang and Shivdasani themselves do not report multivariate results. We suspect that such analysis would confirm our finding that the role of earnings-based measures dwarfs that of stock returns. Finally, the weak relation between stock returns and management change could be peculiar to our sample period. Japanese security prices increased rapidly without parallel growth in earnings during the late 1980s. Consequently, stock returns were imprecise indicators of managerial performance during this period. This characterization of stock return performance as a relatively noisy indicator of managerial effort and decision-making quality may explain the important role seemingly assumed by accounting measures such as industry adjusted return on assets or profit margin. This explanation is consistent with Lambert and Larcker (1987), who find that the relative importance of stock returns versus accounting measures in evaluating and rewarding managerial performance in the U.S. is positively related to the degree to which such measures are likely to be informative about

16

CHRISTOPHER W. ANDERSON ET AL.

managerial actions. In a similar vein, Jarrell and Dorkey (1992) find that accounting measures of performance tend to be highly correlated with long-run stock return performance. They argue for the use of such measures in managerial incentives because they are less likely to be affected by macroeconomic “noise” such as that coincident with the extraordinary run-up of Japanese stock prices in the late 1980s.

4.2. Top Management Change and Exposure to Creditor Monitoring The prior section finds a relation between firm performance, especially when measured by accounting income, and managerial change. This finding is comparable to those of Kaplan (1994) and Kang and Shivdasani (1995). As discussed in Section 2, bank intervention during periods of poor performance may be a unique and effective governance mechanism in the Japanese economy. Since active creditor monitoring and intervention are most likely to occur when firms are highly indebted or experiencing severe liquidity problems, we next estimate the likelihood of top management change as a function of alternative measures of liquidity and leverage. Table 6 presents estimated logit equations of top management change conditional on variables that proxy for exposure to creditor monitoring: interest coverage (COV), quick ratio (QUICK), debt to value (DTV), and loans to value (LTV). Each coefficient has the hypothesized sign and is statistically significant in the univariate equation estimates reported in columns (1)–(4) of Table 6. The standardized coefficient estimates are all of a magnitude comparable to those for accounting performance variables in Table 4. For example, a one standard deviation increase (decrease) in COV results in a decreased (increased) probability of top management change of 4.2% (5.8%) from the 12.7% rate of top management change for firms with mean coverage ratios.8 The multivariate equation estimates provided in columns (5) and (6) suggest that the interest coverage ratio (COV) is the most important of these variables, as the debt to value and loan to value coefficients are insignificant when COV enters the equation. When either debt to value (DTV) or loans to value (LTV) enter the estimated equations, QUICK is rendered insignificant (columns (7) and (8)). Table 7 reports predicted and observed frequency of top management change across sample quintiles based on liquidity and leverage. The actual incidence of top management change is significantly related to each of the four variables examined. Interest coverage (COV – panel A) is a particularly interesting variable. Firms in the highest coverage quintile have annual management change of just 6.9%, while firms in the middle three quintiles have turnover rates that are close to the unconditional mean, and firms in the lowest coverage display an annual top

Variable INTERCEPT

COV (operating income over interest) QUICK (current assets – inv. to current liabilities) DTV (total book debt to firm value) LTV (total loans to value) Chi-squared ( p-value) Observations a This

(1) –1.471 [–1.928] (0.0000) –0.073 [–0.444] (0.0001) –

(2) –1.427 [–1.893] (0.0000) –

–0.355 [–0.311] (0.0105)

–

–

–

–

18.27 (0.0000) 1,086

8.08 (0.0045) 1,086

(3) –2.516 [–1.898] (0.0000) –

–

1.444 [0.312] (0.0003) –

12.78 (0.0003) 1,086

(4)

(5)

–2.117 [–1.886] (0.0000) –

–1.786 [–1.928] (0.0000) –0.058 [−0.356] (0.0141) –

–

–

1.615 [0.251] (0.0022) 8.95 (0.0028) 1,086

0.524 [0.113] (0.3319) –

19.21 (0.0001) 1,086

(6) –1.552 [–1.926] (0.0000) –0.067 [–0.407] (0.0035) –

–

0.310 [0.048] (0.6493) 18.48 (0.0001) 1,086

(7)

(8)

–2.191 [–1.915] (0.0000) –

–1.769 [–1.898] (0.0000) –

–0.162 [–0.142] (0.2809)

–0.223 [–0.196] (0.1309)

1.165 [0.252] (0.0140) –

14.05 (0.0009) 1,086

–

1.149 [0.179] (0.0588) 11.55 (0.0031) 1,086

table presents estimated logit equations where the dependent variable is equal to one when a top executive (chairman, president, or CEO) listed by Moody’s International ceases to be listed over 1984–1989. Independent variables are: COV – ratio of operating income divided by interest expense; DTV – total book value of debt divided by sum of book debt and market value of equity; LTV – book value of loans divided by sum of book debt and market value of equity; QUICK – total current assets minus inventory divided by total current liabilities. To improve model fit and reduce the influence of extreme outliers, COV observations above the upper 5% tail have been set to the 95% level of the distribution. Numbers in square brackets are the coefficients on standardized independent variables. Numbers in parentheses are p-values associated with t-tests of the coefficient estimates.

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

Table 6. Likelihood of Top Management Change Based on Liquidity and Leverage.a

17

18

CHRISTOPHER W. ANDERSON ET AL.

Table 7. Predicted and Observed Managerial Change Based on Liquidity and Leverage.a Quintile Median

Management Change Predicted

Observed

A. Management change conditional on COV COV Quintile High 18.770 5.6% 6.9% 2 7.100 12.1% 12.4% 3 3.640 15.0% 14.4% 4 2.040 16.5% 14.7% Low 1.240 17.4% 18.9% Chi-squared tests of observed turnover: high quintile = low: 14.47∗∗∗ ; above median = below: 12.80∗∗∗ ; high = 2 = 3 = 4 = low: 15.21∗∗∗ B. Management change conditional on QUICK QUICK Quintile High 2.283 9.6% 9.2% 2 1.363 12.9% 7.8% 3 1.039 14.2% 17.5% 4 0.835 15.1% 16.1% Low 0.634 16.1% 16.6% Chi-squared tests of observed turnover: high quintile = low: 4.62∗∗ ; above median = below: 12.80∗∗∗ ; high = 2 = 3 = 4 = low: 16.41∗∗∗ C. Management change conditional on DTV DTV Quintile High 0.166 9.3% 7.4% 2 0.286 10.9% 12.0% 3 0.399 12.6% 13.8% 4 0.534 14.9% 14.3% Low 0.747 19.2% 19.7% Chi-squared tests of observed turnover: high quintile = low: 14.63∗∗∗ ; above median = below: 5.37∗∗ ; low = 2 = 3 = 4 = high: 15.2∗∗∗ D. Management change conditional on LTV LTV Quintile High 0.000 10.8% 10.6% 2 0.028 11.2% 9.7% 3 0.087 12.2% 12.9% 4 0.181 13.9% 15.2% Low 0.398 18.6% 18.8% Chi-squared tests of observed turnover: high quintile = low: 5.91∗∗ ; above median = below: 4.58∗∗ ; low = 2 = 3 = 4 = high: 10.00∗∗ ∗∗∗ ∗∗ ∗

,

, significantly different from zero at the 1%, 5%, and 10% levels, respectively.

a Each

panel displays the sample distribution of a performance measure and turnover probabilities across quintiles for measures of liquidity and indebtedness. Measures are: COV – ratio of operating income divided by interest expense; DTV – total book value of debt divided by sum of book debt and market value of equity; LTV – book value of loans divided by sum of book debt and market value of equity; QUICK – total current assets minus inventory divided by total current liabilities. Predicted turnover is obtained by using the estimated logit equations in Table 6 conditional on quintile median performance. Tests statistics for differences in observed turnover rates are chi-squared tests with one, one, and four degrees of freedom, respectively.

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

19

management change rate of 18.9%. A similar pattern across quintiles is evident for debt to value (DTV) in panel C. The results for quick ratio (QUICK – panel B) indicate some non-monotonicities in top management change with respect to this variable and a relatively poor model fit in comparison to the other variables. Loans to value (LTV – panel D) displays a similar non-monotonicity, but only for the second quintile. The results in Tables 6 and 7 suggest that changes in Japanese top management are significantly related to measures of liquidity and indebtedness. This suggests that creditor monitoring is the mechanism that disciplines top management. However, the evidence presented thus far does not distinguish creditor monitoring per se from a simple hypothesis of performance-related managerial turnover. For example, while there is an empirical relation between alternative measures of leverage and observed top management change, these measures are correlated with earnings performance (see Table 3). The relation between interest coverage and alternative measures of corporate earnings such as IROA and IPM is obvious, but the other leverage measures are also highly correlated with corporate profitability (see Table 3). Prowse (1990), for example, documents that leverage for Japanese firms is negatively related to past profitability. Consequently, the results attributed to earnings performance per se may actually be due to exposure to creditor monitoring, or the reverse. Furthermore, discussions of creditor monitoring in Japan suggest that management is subject to active monitoring and potential discipline in times of financial distress. To further investigate the creditor-monitoring hypothesis, we segment the sample by both firm profitability (measured by IROA and IPM) and exposure to creditor monitoring (measured by COV, QUICK, DTV and LTV). Specifically, in Table 8 we divide the sample into quadrants based on the median profitability and median liquidity or leverage and investigate top management change across the resulting four quadrants. Similar analyses were conducted using return-based measures; consistent with our findings in Tables 4 and 5, conditioning top management change on the basis of returns had no meaningful effect. Panels A and B of Table 8 present persuasive evidence that it is exposure to creditor monitoring per se and not mere earnings performance that influences top management change in Japan. Panel A breaks the sample into four quadrants based on industry-adjusted return on assets (IROA) and interest coverage (COV). These two measures are highly correlated (their correlation coefficient, reported in Table 3, is 0.69), yet there are still many observations with above median IROA yet below median interest rate coverage (n = 142). Similarly, there are 142 firm-years with below median IROA and above median coverage. For firms with above median coverage (COV = high) there is no significant difference in the incidence of top management change for firms with above median IROA (top management change of 38/401 = 9.5%) and firms with below median IROA (top management

20

CHRISTOPHER W. ANDERSON ET AL.

Table 8. Management Change Based on Exposure to Creditor Monitoring Versus Profitability.a High

Low

Chi-squared

14.8% (21/142) 18.0% (72/401) 0.76 {0.3835}

2.89 {0.0894} 4.58 {0.0323}

14.3% (26/182) 18.6% (67/361) 1.60 {0.2065}

3.59 {0.0583} 4.61 {0.0319}

16.7% (32/192) 17.4% (61/351) 0.04 {0.8329}

9.86 {0.0017} 1.39 {0.2388}

15.6% (33/212) 18.1% (60/331) 0.60 {0.4375}

8.46 {0.0036} 2.35 {0.1250}

9.4% (35/372) 14.6% (25/171) 3.10 {0.0781}

2.49 {0.1145} 0.37 {0.5432}

9.4% (34/361) 14.3% (26/182) 2.82 {0.0932}

1.75{0.1859} 0.76 {0.3844}

A. IROA × COV Quadrants COV IROA High Low Chi-squared

9.5% (38/401) 10.6% (15/142) 0.14 {0.7097}

B. IPM × COV Quadrants COV IPM High Low Chi-squared

8.9% (32/361) 11.5% (21/182) 0.96 {0.3273}

C. IROA × QUICK Quadrants QUICK IROA High Low Chi-squared

7.7% (27/351) 13.5% (26/192) 4.64 {0.0312}

D. IPM × QUICK Quadrants QUICK IPM High Low Chi-squared

7.6% (25/331) 13.2% (28/212) 4.58 {0.0324}

E. IROA × DTV Quadrants DTV IROA High Low Chi-squared

14.0% (24/171) 16.7% (62/372) 0.62 {0.4312}

F. IPM × DTV Quadrants DTV IPM High Low Chi-squared

13.2% (24/182) 17.2% (62/361) 1.48 {0.2237}

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

21

Table 8. (Continued ) High

Low

Chi-squared

10.4% (38/367) 13.1% (23/176) 0.86 {0.3593}

0.30 {0.5830} 1.74 {0.1873}

9.4% (31/330) 14.1% (30/213) 2.80 {0.0942}

1.44 {0.2304} 1.18 {0.2776}

G. IROA × LTV Quadrants LTV IROA High Low Chi-squared

11.9% (21/176) 17.4% (64/367) 2.84 {0.0918}

H. IPM × LTV Quadrants LTV IPM High Low Chi-squared

13.7% (29/213) 17.6% (58/330) 2.41 {0.1208}

a The table provides incidence of top management change based on measures of profitability, liquidity,

and leverage. Each panel splits the sample of 1,086 observations into four quadrants based on a measure of profitability, industry-adjusted return on assets (IROA) or industry-adjusted profit margin (IPM), and then by measures of liquidity or indebtedness, i.e. interest coverage ratio (COV), quick ratio (QUICK), debt to value ratio (DTV), or loans to value ratio (LTV). The notation “high” and “low” denotes above median and below median, respectively. The rate of top management change per quadrant, the number of instances of and sample size per quadrant (in parentheses), and chi-squared tests for equality of turnover frequency across selected quadrants are provided. p-Values for the chi-squared statistics are presented in scrolled brackets.

change of 15/142 = 10.6%). Similarly, for firms with below median coverage ratios, there is no significant difference in top management change with respect to IROA classification (above median rate of 21/142 = 14.8%, below median rate of 18.0%, chi-squared statistic insignificant at 0.76). However, firms with above median IROA but below median coverage have a significantly higher rate of managerial change than firms with high IROA and high coverage (14.8% vs. 9.5%, chi-squared of 2.89). Similarly, firms with low coverage and low IROA have significantly more management changes than firms with high coverage and low IROA (18.0% vs. 10.6%, chi-squared of 4.58). The results using industry-adjusted profit margin (IPM) to split the sample (panel B) parallel those of panel A. After controlling for coverage there is no significant difference in management change by IPM, yet after controlling for IPM there are significant differences by interest coverage. These results suggest that top management change in Japanese firms is related to exposure to creditor monitoring, measured by interest coverage, and not to unconditional earnings performance. These findings buttress results from Kang and Shivdasani (1995), who report that negative income (and presumably negative cash flow and interest coverage) dramatically raises the odds of turnover,

22

CHRISTOPHER W. ANDERSON ET AL.

not merely below average performance.9 Panels C and D of Table 8, which present results based on segmenting the sample by quick ratio and, alternatively, IROA and IPM, provide further evidence in support of creditor monitoring. First, firms with high levels of asset liquidity, i.e. above median quick ratios, display significantly more management changes when profitability is low (13.5% vs. 7.7%, chi-square of 4.64, in panel C; 13.2% vs. 7.6%, chi-square of 4.58, in panel D). However, the differences are even larger when high profit margin liquid firms are compared to high profit illiquid (i.e. low quick ratio) firms (7.7% vs. 16.7% in panel C; 7.6% vs. 15.6% in panel D). This suggests that a firm’s ability to meet its short-term obligations is at least as important as relative profitability in influencing top management change. Not surprisingly, the worst possible quadrant, the one associated with low quick ratio and low profitability, displays the highest rate of top management change (17.4% in panel C; 18.1% in panel D), but for firms with poor profitability turnover does not significantly increase when QUICK is below median. Panels E–H of Table 8 suggest that the relation between profitability and top management change is not conditional on aggregate leverage or bank leverage. In all four panels, high performance, low leverage firms display the lowest rates of managerial change. However, management changes increase when performance declines for both high and low leverage samples, although these differences are not always significant. These results do not suggest that high leverage firms have a stronger turnover-performance relation than less leveraged firms. We also estimate logit equations with the leverage measures, performance measures, and interaction terms as independent variables. Consistent with the breakdown in Table 8, we find that estimated coefficients on the leverage measures and performance measures are statistically significant, but that the coefficients on the interactions of these variables are not. For brevity we omit these results. In general, the results reported in Tables 6–8 indicate that liquidity variables significantly affect the level of top management change at Japanese firms. In particular, low interest coverage ratios seem more important in explaining top management change than return of assets or profit margin. Specifically, firms with below median coverage ratios have significantly higher rates of top management change regardless of industry-adjusted profitability. A similar result holds when we measure liquidity with the firm’s quick ratio. On the other hand, top management change is not significantly higher for firms with below median profitability but relatively high coverage ratios compared to firms with above median profitability. These results suggest that management change occurs disproportionately more often when the interests of creditors are threatened and not when firms perform poorly yet have sufficient cash to appease creditors.

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

23

4.3. Additional Results on Keiretsu Firms Versus Non-keiretsu Firms The role of Japanese banks in monitoring client firms and intervening in times of financial distress is often hypothesized to be especially pronounced for members of keiretsu corporate groups. The long-standing ties between lenders and borrowers, extraordinary bank access to information on the borrower, and implicit control rights afforded the so-called main bank under extraordinary circumstances suggest that creditor monitoring is more likely to influence top management change for keiretsu member firms. To test this implication we conduct a logit estimation of top management change with a keiretsu dummy variable intercept and an interactive dummy on measures of performance, liquidity, and indebtedness. A negative coefficient estimate for the interactive term would imply that keiretsu member firms display top management change that is more sensitive to firm performance. The keiretsu specific intercept term is not significantly different from zero in all specifications, confirming the prima facie evidence in Table 1 that the incidence of managerial change is identical for keiretsu and non-keiretsu firms. The coefficients on interactive dummy variables are also insignificant for all specifications, suggesting that the turnover-performance relations documented earlier do not vary by keiretsu affiliation. The strongest statement we can make about differential sensitivity of management changes to performance concerns four-year cumulative stock returns (RET4): the relation between top management change and stock returns is significantly negative for keiretsu firms and insignificantly negative for non-member firms, but this difference itself is not significant. These findings are similar to that reported by Morck and Nakamura (1999) with respect to bank appointments to boards of directors. Additional specifications, involving alternative variables and interactions of variables, also failed to distinguish keiretsu from non-member firms. Consequently, we find no evidence that the relation between top management change and Japanese firm performance differs by keiretsu affiliation. For brevity, we omit full reporting of these results.

5. SUMMARY AND CONCLUSIONS Recent empirical findings of an inverse relation between firm performance and managerial turnover at Japanese firms are construed as evidence that bank monitoring substitutes for other corporate governance mechanisms (Kaplan, 1994; Kang & Shivdasani, 1995). In contrast, Morck and Nakamura (1999) report that banks act to protect their narrow self-interest as creditors when appointing corporate directors rather than the interests of shareholders. We examine these conflicting views by investigating the extent to which top management changes at Japanese firms are

24

CHRISTOPHER W. ANDERSON ET AL.

explained by measures of liquidity and leverage that are of direct concern to creditors as opposed to measures of interest to shareholders such as profitability or stock returns. Specifically, we examine the relation between top management change and profitability, stock-price performance, and variables that reflect a firm’s ability to meet its short-term obligations for a sample of 207 Japanese firms in the 1980s. First, we find that top management changes are significantly related to accounting measures of firm performance but are less sensitive to stock-price performance measures. These results are consistent with evidence on the performance-turnover relation as reported by Kaplan (1994) and Kang and Shivdasani (1995). Next, we find that top management changes are significantly related to measures of liquidity and leverage, and these measures largely dominate profitability per se. For instance, we investigate management changes for firms with high (low) industry adjusted rates of profitability but differential exposure to potential creditor monitoring as measured by liquidity and leverage variables. Firms with high industry-adjusted profitability but with low liquidity experience significantly higher rates of management change than similarly profitable yet liquid firms. Similarly, relatively unprofitable yet liquid firms have fewer management changes than other unprofitable firms. In short, liquidity measures of keen interest to creditors seem to drive top management changes to a greater degree than earnings or stock return performance. Consistent with the evidence on appointments to corporate boards reported by Morck and Nakamura (1999), our findings counter prior inferences that a performance-turnover relation is evidence of effective corporate governance by means of bank monitoring (Kaplan, 1994; Kang & Shivdasani, 1995; Kaplan & Ramseyer, 1996). Instead, Japanese corporate governance mechanisms appear skewed to protect creditors’ interests rather than those of shareholders. Specifically, our results suggest that monitoring by creditors is driven by their own short-term interests and is therefore an imperfect substitute for other mechanisms of corporate control that directly protect the interests of shareholders of Japanese firms.

NOTES 1. Kester (1991) describes the Japanese takeover market as confined to only small, privately negotiated deals, with the purchase price for publicly announced mergers and acquisitions between 1982 and 1987 averaging less than the equivalent of $4 million. 2. Pascale and Rohlen (1983) illustrate how such main bank intervention facilitated the turnaround of the Mazda Corporation in the 1970s. 3. Popular press discussion of Japanese restructuring in the face of the most recent recession suggests that Japanese stakeholders hold managers accountable for poor performance. See, for instance, The Wall Street Journal, July 8, 1993, “Japanese CEOs Find Life is Getting Tougher at the Top.”

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

25

4. Lacking biographical data or the official reason for executive departure, our measure of management change probably includes executive deaths in addition to departures and retirements. This biases against a relation between firm performance and management change. Kaplan (1994) indicates that the annual rate of executive death is less than 1%, or approximately about one out of every 15 observed top management changes, so the effects are unlikely to be significant. 5. We utilize Nakatani (1984) to classify sample firms as keiretsu or non-keiretsu firms. Nakatani identifies a firm as a keiretsu member if one of the following conditions is met: (1) a keiretsu’s main bank has been the largest lender to the firm for the prior three years and the equity share held by keiretsu member firms exceeds 20%; (2) a keiretsu main bank has been the lender of at least 40% of the firm’s debt for the prior three years; (3) the firm has been historically identified as a member of a particular keiretsu. Nakatani also provides a list of ostensibly independent firms. We identify only 18 sample firms (87 firm fiscal years) as independent firms. The results do not differ significantly when we compare keiretsu firms to these ostensibly independent firms. 6. Gilson (1989, 1990) also reports that financial distress tends to increase the likelihood of management turnover at U.S. firms. 7. These probabilities are derived using Eq. (2) and the standardized coefficient estimates from Table 4 column (1). The predicted rate of management change for firms with mean IROA is calculated as exp(−1.906)/[1 + exp(−1.906)] = 12.9%. The implied probability for a firm with a IROA one standard deviation above the mean is exp(−1.906−0.359)/[1 + exp(−1.906 − 0.359)] = 9.4%. The implied probability for a firm with IROA one standard deviation below the mean is exp(−1.906 + 0.359)X/[1 + exp(−1.906 + 0.359)] = 17.6%. Analogous calculations lead to the implied probabilities of management change associated with variation in IPM. 8. These implied probabilities are derived from Eq. (2) using the standardized coefficient estimates from column (1) of Table 6. An example of such a calculation is presented in note 7. 9. In Kang and Shivdasani’s Table 6 turnover for negative income firm years is 13% for firms with bank ties and 8.3% for firms without such ties compared to 1.2 and 2.9%, respectively, for firm years with positive income.

ACKNOWLEDGMENTS We are grateful for the comments and suggestions on previous incarnations of this study. In particular, we thank Peter Abken, Mark Hirschey, Chuan Yang Hwang, Jonathon Karpoff, Ken Lehn, Anil Makhija, Bob Nachtmann, David Nachman, Tom Noe, Steve Smith, Larry Wall, Jerold Warner, and workshop participants at the University of Pittsburgh and Georgia Tech.

REFERENCES Aoki, M. (1990). Toward an economic model of the Japanese firm. Journal of Economic Literature, 28, 1–27.

26

CHRISTOPHER W. ANDERSON ET AL.

Aoki, M. (1992). Ex post monitoring by the main bank and corporate governance structure in Japan. Presentation at the Center for Economic Policy Research, Stanford University (May). Ballon, R., & Tomita, I. (1988). The financial behavior of Japanese corporations. New York, NY: Kodansha International. Barro, J., & Barro, R. (1990). Pay, performance and turnover of bank CEOs. Journal of Labor Economics, 8, 448–481. Coughlan, A., & Schmidt, R. (1985). Executive compensation, management turnover, and firm performance: An empirical investigation. Journal of Accounting and Economics, 7, 43–66. Gibbons, R., & Murphy, K. (1990). Relative performance evaluation for chief executive officers. Industrial and Labor Relations Review, 43, 30s–50s. Gibson, M. (1995). Can bank health affect investment? Evidence from Japan. Journal of Business, 68, 281–308. Gilson, S. (1989). Management turnover and financial distress. Journal of Financial Economics, 25, 241–262. Gilson, S. (1990). Bankruptcy, boards, banks, and blockholders: Evidence on changes in corporate ownership and control when firms default. Journal of Financial Economics, 27, 355–387. Hodder, J., & Tschoegl, A. (1992). Corporate finance in Japan. Policy paper, Center for Economic Policy Research, Stanford University. Horiuchi, A., Packer, F., & Fukada, S. (1988). What role has the ‘main bank’ played in Japan? Journal of the Japanese and International Economies, 2, 159–180. Hoshi, T., Kashyap, A., & Scharfstein, D. (1990a). Bank monitoring and investment: Evidence from the changing structure of Japanese corporate banking relationships. In: R. G. Hubbard (Ed.), Asymmetric Information, Corporate Finance, and Investment. Chicago, IL: University of Chicago Press. Hoshi, T., Kashyap, A., & Scharfstein, D. (1990b). The role of banks in reducing financial distress in Japan. Journal of Financial Economics, 27, 67–88. Hoshi, T., Kashyap, A., & Scharfstein, D. (1991). Corporate structure, liquidity, and investment: evidence from Japanese industrial groups. Quarterly Journal of Economics, 106, 33–60. Jarrell, G., & Dorkey, F. (1992). The longer-term relation between accounting performance and stock returns. University of Rochester. Working paper. Jensen, M., & Murphy, K. (1990). Performance pay and top-management incentives. Journal of Political Economy, 98, 225–264. Judge, G., Griffiths, W., Hill, R., Lutkepohl, H., & Lee, T. (1985). The theory and practice of econometrics. London: Cambridge University Press. Kang, J., & Shivdasani, A. (1995). Firm performance, corporate governance, and top executive turnover in Japan. Journal of Financial Economics, 38, 29–58. Kang, J., Shivdasani, A., & Yamada, T. (2000). The effect of bank relations on investment decisions: An investigation of Japanese takeover bids. Journal of Finance, 55, 2197–2218. Kang, J., & Stulz, R. (2000). Do banking shocks affect borrowing firm performance? An analysis of the Japanese experience. Journal of Business, 71, 1–23. Kaplan, S. (1994).Top executive rewards and firm performance: A comparison of Japan and the U.S. Journal of Political Economy, 102, 510–546. Kaplan, S. (1997). Corporate governance and corporate performance: A comparison of Germany, Japan, and the U.S. In: D. Chew (Ed.), Studies in International Corporate Finance and Governance (pp. 251–258). Oxford: Oxford University Press. Kaplan, S., & Minton, B. (1994). Appointments of outsiders to Japanese boards: Determinants and implications for managers. Journal of Financial Economics, 36, 225–258.

Bank Monitoring, Firm Performance, and Top Management Turnover in Japan

27

Kaplan, S., & Ramseyer, M. (1996). Those Japanese firms with their disdain for shareholders: Another fable for the academy. Washington University Law Quarterly, 74(2), 403–418. Kester, W. (1991). Japanese takeovers: The global contest for corporate control. Boston, MA: Harvard Business School Press. Lambert, R., & Larcker, D. (1987). An analysis of the use of accounting and market measures of performance in executive compensation contracts. Journal of Accounting Research, 25(s), 85–125. Morck, R., & Nakamura, M. (1999). Banks and corporate control in Japan. Journal of Finance, 54, 319–339. Nakatani, I. (1984). The economic role of financial corporate grouping. In: M. Aoki (Ed.), The Economic Analysis of the Japanese Firm. New York, NY: Elsevier Science Publishers. Nishiyama, T. (1984). The structure of managerial control: Who owns and controls Japanese businesses? In: K. Sato and Y. Hoshino (Eds), The Anatomy of Japanese Business. New York, NY: M. E. Sharpe. Pascale, R., & Rohlen, T. (1983). The Mazda turnaround. Journal of Japanese Studies, 9, 219–263. Prowse, S. (1990). Institutional investment patterns and corporate financial behavior in the United States and Japan. Journal of Financial Economics, 27, 43–66. Prowse, S. (1992). The structure of corporate ownership in Japan. Journal of Finance, 47, 1121–1140. Sheard, P. (1989). The main bank system and corporate monitoring and control in Japan. Journal of Economic Behavior and Organization, 11, 399–422. Suzuki, S., & Wright, R. (1985). Financial structure and bankruptcy risk in Japanese companies. Journal of International Business Studies, 9, 97–110. Warner, J., Watts, R., & Wruck, K. (1988). Stock prices and top management changes. Journal of Financial Economics, 20, 461–492. Weisbach, M. (1988). Outside directors and CEO turnover. Journal of Financial Economics, 20, 431–460.

CORPORATE GOVERNANCE IN SINGAPORE: THE IMPACT OF DIRECTORS’ EQUITY OWNERSHIP Gurmeet S. Bhabra, Stephen P. Ferris, Nilanjan Sen and Peng Peck Yen ABSTRACT We examine whether the curvilinear relationship between directors’ equity ownership and firm performance exists in a non-Western economy such as Singapore. We find that it does, although the inflection points are much higher than that generally cited for U.S. firms. We then compare this relationship across two kinds of firms that are not common to the U.S. marketplace. We observe for founder-controlled firms that the impact of director ownership is insignificant. We also examine government-linked corporations and in spite of the presence of a government blockholder, find that the pattern of alignment, entrenchment and then alignment remains operative.

1. INTRODUCTION An important characteristic of the modern corporation is the divergence in identities between its managers and owners. Because of this separation, an agency relationship develops between the firm’s owners and managers. In the corporate business organization, managers who initiate and implement important decisions

Corporate Governance and Finance Advances in Financial Economics, Volume 8, 29–46 Copyright © 2003 by Elsevier Science Ltd. All rights of reproduction in any form reserved ISSN: 1569-3732/PII: S1569373203080022

29

30

GURMEET S. BHABRA ET AL.

are not the major residual claimants. Therefore, they do not bear a major share of the wealth effects of their decisions. Hence, Jensen and Meckling (1976) observe that managers are likely to take actions that deviate from the wealth-maximizing interests of their shareholders. Among the many mechanisms designed to alleviate this agency conflict has been the granting of equity ownership to the firm’s managers and directors. Jensen and Meckling (1976) develop a formal model that establishes the relation between corporate value and managerial equity ownership. Morck, Shleifer and Vishny (1988), McConnell and Servaes (1990) and Short and Keasey (1998) show that the relationship between managerial equity ownership and firm value is non-linear. This non-linearity is the result of two conflicting influences resulting from equity ownership: the convergence of interest and managerial entrenchment effects. The convergence of interest effect (Denis, Denis & Sarin, 1997; Jensen & Meckling, 1976; McConnell & Servaes, 1990; Morck, Shleifer & Vishny, 1988) recognizes that large equity holdings by corporate insiders such as managers or directors can be associated with a high market valuation for the firm. This is due to the incentive effect of such ownership. As managers hold more equity, their personal wealth is more directly tied to the performance of the firm’s stock. Consequently, managers are more likely to make those decisions that enhance shareholder value and are less likely to dissipate corporate value by shirking or perquisite consumption. The entrenchment effect (Demsetz, 1983; Fama & Jensen, 1983) occurs when managers who hold a large portion of the firm’s equity possess sufficient voting power to ensure that their internal positions are secure. Consequently, they become insulated from external disciplining forces such as hostile takeovers or the managerial labor market. Such managers have adequate voting power to guarantee their continued employment with the firm and to maintain high levels of compensation. It is also likely that these managers will spend increased time shirking and perquisite consumption rather than engaged in value-maximizing activities. Stulz (1988) develops a model consistent with managerial entrenchment whereby high ownership by managers effectively precludes the possibility of an unfriendly takeover. Consistent with this, Weston (1979) finds that firms reporting managerial ownership in excess of 30% have never been acquired in a hostile takeover. Morck, Shleifer and Vishny (1988) examine the relationship between managerial ownership and firm value for a set of Fortune 500 firms. They find that firm value increases as managerial ownership rises between 0% and 5%. Beyond the 5% ownership level, increases in managerial ownership appears to be associated with conditions conductive to managerial entrenchment. However, as board ownership rises beyond the 25% level, firm value also increases in relation to rising managerial equity ownership.

Corporate Governance in Singapore

31

McConnell and Servaes (1990) also find that the relation between managerial equity ownership and firm value is curvilinear. They report that Tobin’s q, their proxy for firm value, initially increases, then decreases as share ownership becomes concentrated among corporate managers and board directors. This relationship attains its maximum prior to a 50% level of managerial ownership. Short and Keasey (1998) examine the international nature of the relationship between firm value and managerial ownership by studying a set of U.K. firms. They likewise find evidence that management moves from alignment, to entrenchment, and back to alignment as its ownership in the firm increases. They also report that U.K. managers require higher levels of equity ownership than their U.S. counterparts before they become entrenched. The findings of Short and Keasey (1998) demonstrate that national culture and governance systems influence the threshold at which managerial entrenchment occurs. Hence, it suggests that an international examination of the relationship between firm value and managerial ownership would provide new insights. In this study we analyze the effect that equity ownership by the board of directors has on firm value for a set of Singaporean corporations. Recently, the government of Singapore has actively promoted the country as a gateway into the Asian economy for U.S., European, and other international firms.1 Since it is at the crossroads of capital inflows from so many different investors, we believe that Singapore is an interesting market in which to study the relationship between insider ownership and firm value. Additionally, Singaporean firms demonstrate a diversity in ownership structure that is lacking in many Western economies. In addition to the diffusely-held corporation, the Singapore economy contains a number of founder-controlled firms and government-private hybrids referred to as government-linked corporations. Our analysis of the relationship between firm value and insider ownership is made richer by the presence of these other kinds of firms in our sample. Because Singapore differs from the West in terms of regulation and culture, entrenchment might or might not occur. If entrenchment is found in Singapore, then it might occur at a threshold different from that of Western firms. Harmony and cooperation are preferred over disagreement and competition in Asian society. There are also strict rules in Singapore to prevent raiders from accumulating large ownership interests in a target without adequate disclosure. Under the Singapore Companies Act, investors holding 5% or more of a firm’s outstanding shares must disclose all equity transactions within two days of the transaction. Consequently, the takeover market in Singapore is largely inactive. The threat of takeover as a control mechanism is essentially non-existent. These regulatory and cultural differences suggest that entrenchment might occur at a different level in Singapore relative to other markets.

32

GURMEET S. BHABRA ET AL.

The contribution of this study resides in three areas. First, we examine whether the curvilinear relationship between managerial equity ownership and firm performance documented for U.S. and U.K. firms, exists in a non-Western economy such as Singapore. Second, if such a relationship exists, does the threshold at which managerial entrenchment occurs differ from that reported in the literature for other nations. These results will provide insights into the international nature of agency costs and how corporate governance mechanisms can be designed to better overcome them. Third, we exploit the ownership diversity present in the Singaporean economy by comparing the relationship between firm performance and insider equity holdings across different kinds of firms. This analysis will allow us to better understand the interplay between the equity ownership structure and firm performance by studying firm types generally absent from Western economies or relatively uncommon.

2. INSTITUTIONAL BACKGROUND In this section we provide a discussion of several important aspects of Singaporean finance and business practices that are different from that prevailing in the West. These characteristics are important for understanding international differences regarding the impact that director equity ownership might exert on firm performance. Further, we discuss the three general types of business ownership that exist for firms listing on the Singapore Exchange (SGX). The SGX represents the 1999 merger of the Stock Exchange of Singapore (SGX) and the Singapore International Monetary Exchange. 2.1. Differences Between Western and Singapore Business Practices 2.1.1. Institutional Equity Ownership In Singapore, the Singapore Banking Act of 1970 precludes banks from holding large amounts of common stock. Neither mutual nor pension funds are major shareholders in Singaporean equities. This occurs despite the presence of a large public pension fund in Singapore known as the Central Provident Fund (CPF).2 Nor are international equity mutual funds widely invested in Singapore, perhaps due to the small capitalization of the SGX. Therefore, institutional investors have only a limited role to play in the Singaporean equity market. 2.1.2. Shareholder Voting U.S. pension funds are governed by the 1974 Employee Retirement Income Security Act that legally obliges the funds to exercise their voting rights. Hence, such funds are typically more active than their international counterparts.

Corporate Governance in Singapore

33

Singaporean businesses, like many Asian firms, place a significant importance on external appearances. Hence, open confrontation is rare. Shareholders generally exercise their power in private meetings with the firm’s management. Singaporean shareholders typically fail to exercise their rights to vote at company general meetings. 2.1.3. Disclosure of Equity Holdings The U.S. has strict rules to prevent raiders from accumulating large stakes without disclosing their holdings to the target firm. Existing U.S. law requires that an announcement is made within 10 days following a 5% or greater acquisition. Thereafter, any subsequent trading must be promptly disclosed. In Singapore, the disclosure of a 5% or greater holding must be disclosed within two days. Singapore listed companies are also subjected to a mandatory bid threshold of 30% of outstanding equity. Any shareholder breaching the 30% threshold is required to make an offer to all shareholders at the highest price paid by the offerer during the last twelve months.

2.2. The Ownership Structure of Singaporean Firms There are three general types of ownership structure that differentiate firms that list on the SGX. 2.2.1. Founder In these firms, one or a few individuals hold a substantial portion of the outstanding equity. These individuals are the founders of the company and they provide the daily operational management for the firm. There is little, if any, difference between the managers and owners of these firms. The CEO is almost always the chairman of the board of directors. 2.2.2. Government-Linked Corporations (GLC) Shortly after gaining independence, the Singaporean government adopted an industrialization strategy to reduce unemployment and to diversify the economy. As part of this approach, Singapore established companies with a high level of government equity ownership in select industries. These firms are referred to as government-linked corporations or GLCs. During the 1980s, GLCs were established in higher value-added industries in response to changing international economic opportunities. Our sample of twenty GLCs has a mean (median) level of government equity ownership of 52.63% (60.59%). The maximum government ownership for our sample firms is 75.5% of the outstanding equity while the minimum is 9.6%.

34

GURMEET S. BHABRA ET AL.

2.2.3. Corporate These firms resemble the typical publicly-traded firm in the U.S. The ownership of these firms is typically dispersed and managers provide the supervision necessary for the firm to satisfy its operational requirements. The board chair and CEO position are typically separate for these firms.

3. SAMPLE CONSTRUCTION AND DATA DESCRIPTION 3.1. Sample Construction We begin construction of our sample by selecting the 100 largest and 50 smallest firms listed on the SGX as of 2001. This represents nearly 40% of the 384 firms listed on the SGX. We use the value of the firm’s equity market capitalization as our measure for size. Our selection procedure is driven by a desire to construct a sample that contains corporate, GLC, and founder firms. The largest firms consist mainly of GLCs and corporate firms. The smallest firms are mostly founder-owned firms and reflect their generally lower level of capitalization. From these 150 firms, we exclude 15 financial institutions because of their different accounting procedures for measuring income relative to firms in the manufacturing and service sectors. We eliminate ten foreign firms from our sample because of unavailability of their annual reports. Finally, 37 firms are dropped because of data insufficiency or unavailability on Datastream International. We are left with a final sample of 88 firms. We then create three sub-samples consisting of 28 founder-controlled firms, 40 corporate firms and 20 GLC firms. We extract the financial data for our sample firms from Datastream International. We obtain all other information from company annual reports for the years ending 1998, 1999 and 2000. 3.2. Data Description 3.2.1. Measurement of Firm Value Tobin’s q is used to measure the value of a firm. Consistent with Morck, Shleifer and Vishny (1988) and McConnell and Servaes (1990), we calculate q as follows: MVE + PS + BORROWING TA The market value of equity (MVE) is the product of a firm’s share price and the number of common shares outstanding. Preferred stock value (PS) is the book value of the firm’s outstanding preferred shares while BORROWING is the value of the firm’s short term liabilities net of its short term assets, plus the book value of the firm’s long term debt. Total assets (TA) is the book value of the firm’s total assets. q=

Corporate Governance in Singapore

35

3.2.2. Measurement of Insiders’ Equity Ownership We measure insiders’ equity ownership as the percent of the total shares outstanding held by the directors of the firm. This measure of ownership is consistent with Morck, Shleifer and Vishny (1988) and Short and Keasey (1998). Equity ownership data in Singapore however is restricted to individuals who hold the position of corporate director or hold a 5% or greater stake in the firm. The equity holdings of senior executives are not routinely disclosed as is common in other nations. Hence, we limit our analysis to the equity controlled by the firm’s board of directors. We manually extract all director ownership data from the annual reports of our sample firms. The degree of director equity ownership in Singapore clusters at two extremes. At one end, we observe that 41.4% of our sample (36 firms) report mean board ownership of 53%. Not surprisingly, these firms are mostly founder-controlled enterprises. At the other extreme, we find that GLCs constitute the majority of the 44 firms (50% of the sample) having a mean director equity ownership of 1.5% or less. Corporate firms report a mean directorship ownership of 14.5%. 3.2.3. Measurement of the Control Variables We include several additional variables in our analysis to control for other potential influences on firm value. These variables are firm size, firm growth and the level of firm debt. Two advantages due to firm size can positively affect firm value. The first occurs because larger firms tend to have greater access to external capital markets. This allows them to raise the capital necessary to finance positive NPV projects. Second, larger firm size provides economies of scale that can either generate additional corporate profitability or serve as a barrier to entry for potential competitors. We measure firm size (Size) as the natural logarithm of the firm’s equity market capitalization. Firm value will also be influenced by its growth in earnings. We anticipate that firms with a stronger growth in sales and earnings will generate higher market valuations. This, in turn, produces greater q values. Hence, we include a variable, Growth, in our model to capture this possibility. We define Growth as the mean annual change in sales over the years 1998 through 2000. Debt, which we define as the book value of total debt divided by the firm’s total assets, is included to control for a number of factors. First, it controls for the possibility that the creditors exert an influence over the decisions and operations of the firm. Stiglitz (1985), for instance, argues that control over managerial actions is effectively exercised, not by shareholders, but by lenders, particularly banks. Second, as suggested by Grossman and Hart (1982) and Jensen (1986), debt might be used by management to signal that they have bonded themselves to achieve the levels of cash flow necessary to meet the

36

GURMEET S. BHABRA ET AL.

debt repayments. Debt might therefore be used to resolve agency costs and enhance equity values since it reduces management’s ability to consume excessive perquisites. 3.2.4. Comparison Across Sub-samples As shown in Table 1, the founder sub-sample has the highest mean (median) level of director equity ownership with 53.50% (55.67%). The GLC sub-sample exhibits the lowest percent of director ownership with a mean (median) of 1.5% (0.08%). The corporate sub-sample falls in between the two extremes with an insider shareholding mean (median) of 14.90% (1.88%). Table 1. Descriptive Statistics Across Samples. Variable

Mean

Median

Standard Deviation

6.94% 12.77 0.56 0.47% 0.24

27.19% 1.99 0.64 43.06% 0.19

0% 9.26 0 (46.46%) 0

82.64% 17.54 3.30 312.14% 0.86

Panel B: Government-linked corporations sub-sample Dir 1.50% 0.08% 4.95% Size 14.16 13.78 1.41 Tobin’s q 1.05 0.68 0.89 Growth 15.61% 7.48% 38.83% Debt 0.25 0.21 0.22

0% 12.37 0.13 (34.34%) 0

22.00% 17.54 3.30 133.05% 0.86

Panel C: Founder controlled sub-sample Dir 53.5% 55.67% Size 10.49 10.38 Tobin’s q 0.63 0.56 Growth 13.58% (1.99%) Debt 0.22 0.20

15.32% 0.95 0.43 63.4% 0.18

21.22% 9.26 0.08 (46.46%) 0

75.75% 13.37 2.29 312.14% 0.60

Panel D: Corporate sub-sample Dir 14.9% Size 12.46 Tobin’s q 0.65 Growth 0.65% Debt 0.27

22.49% 1.77 0.57 22.6% 0.17

0% 9.78 0 (41.85%) 0

82.64% 15.68 3.02 85.26% 0.75

Panel A: Aggregate sample Dir 24.14% Size 12.22 Tobin’s q 0.74 Growth 8.17% Debt 0.25

1.88% 13.01 0.51 (2.49%) 0.28

Minimum

Maximum

Note: DIR is the directors’ shareholding as a percentage of the total shares outstanding. Size is the log of the firm’s equity market capitalization; GROWTH is the average annual sales growth from 1998 to 2000; DEBT is the total debt of the firm divided by the book value of total assets.

Corporate Governance in Singapore

37

As expected, GLCs are the largest firms as measured by market capitalization. They are followed in size by the corporate enterprises. As anticipated, the foundercontrolled firms are the smallest companies in our sample. This sequencing reflects the fact that GLCs enjoy government support and investment, corporate enterprises have a longer operating history while the founder-controlled firms are usually a first generation establishment. The GLC sub-sample possesses the highest mean (median) Tobin’s q with a value of 1.05 (0.68). The mean (median) q values for the founder and corporate sub-samples are comparable in magnitudes, suggesting that these firms are equally valued. The high q values for the GLC sub-sample might be attributable to the fact they enjoy a financial stability resulting from government investment. Consistent with its higher q values, the GLC sub-sample demonstrates the highest average annual sales growth with a mean (median) of 15.61% (7.48%). This is followed by the founder sub-sample, with a mean (median) sales growth of 13.58% (−1.99%). The corporate sub-sample has the lowest annual sales growth with a mean (median) of 0.65% (−2.49%).

4. HYPOTHESIS DEVELOPMENT 4.1. Firm Value and Directors’ Equity Ownership Shleifer and Vishny (1986) observe that after an initial period of interest alignment, increasing managerial equity ownership results in entrenchment. As managers increase their level of equity ownership, the threat of discharge correspondingly decreases and managers are more capable of consuming agency perquisites. But at still higher level of ownership however, the incentive effect re-establishes itself, with managers now having to pay a larger share of their indulgence in non-value maximizing activities. For our aggregate sample of Singaporean firms, we test whether this pattern is present with our first hypothesis: H1. There is a non-linear relation between firm value and the level of director equity ownership in Singaporean firms. We test this hypothesis by estimating the following model and examining the coefficients on the director equity ownership variables: Performance = ␣ + ␤1 DIR + ␤2 DIR2 + ␤3 DIR3 +

3
i=1

␤i Control Variables

(1)

38

GURMEET S. BHABRA ET AL.

We define the model’s variables as follows: DIR is the percentage of the outstanding shares owned by the firm’s directors while DIR2 (DIR3 ) is the corresponding square (cube) value. The control variables are the firm’s level of debt, its growth rate in sales, and its equity market capitalization. The level of managerial ownership where entrenchment occurs will be computed by differentiating Tobin’s q with respect to directors’ equity ownership, assuming that all other variables are constant. We then let (dq)/(d DIR) = 0 and solve for q. To determine whether this is a maximum or minimum inflection point, the value of (d2 q)/(d DIR2 ) is also calculated. If (d2 q)/(d DIR2 ) > 0, then the inflection point is determined to be a maximum. Alternatively, if (d2 q)/(d DIR2 ) < 0, we determine that the inflection point is a minimum. 4.1.1. Director Equity Ownership and Founder-Controlled Firms In a founder-controlled firm, one or a few individuals control a substantial percent of the firm’s outstanding equity. Given the high level of director ownership in such firms, the costs associated with any non-maximizing behavior due to entrenchment will correspondingly fall more heavily on the directors. Consequently, the relationship between firm value and director equity ownership might be different for these firms relative to our aggregate sample. Thus, we hypothesize that the relation between director ownership and corporate value is linear for these firms, with the convergence of interest effect dominating any tendency towards entrenchment: H2. The value of a founder-controlled company is linearly related to the percentage of equity held by its directors. 4.1.2. Director Equity Ownership and GLC Firms GLCs are firms owned by the investing arm of Singapore government. Director equity ownership in such firms is generally low. As indicated in Section 2.2, the mean level of government ownership in these firms is nearly 53%. The firm is subject to government intervention and corporate business strategies can be influenced by the government’s choice of social or political goals. These might conflict with the firm’s desire for profit maximization, thus having a negative impact on corporate values. But government ownership might enhance investor confidence in the firm since it provides monitoring comparable to that of a large bock-holder. Mikkelson and Ruback (1985), Shleifer and Vishny (1986) and Brickley, Lease and Smith (1988) report evidence consistent with greater monitoring by blockholders and consequent positive influences on firm value. Thus, government monitoring might exert a positive impact on firm value and neutralize whatever

Corporate Governance in Singapore

39

adverse influences result from government ownership. Thus, we specify our third hypothesis: H3. Because of the conflicting effects of government ownership, government affiliation will exert no impact on the relation between firm value and director equity ownership.

5. EMPIRICAL RESULTS 5.1. Firm Value and Director Equity Ownership Table 2 presents the results from the test of our first hypothesis that the relationship between firm value and director equity ownership is non-linear. We find that for the aggregate sample of firms the coefficients on the director equity ownership variables (i.e. Dir, Dir2 and Dir3 ) are statistically significant and possesses the Table 2. Relationship Between Firm Value and Director Equity Ownership. Variable Intercept Dir Dir2 Dir3 Size Growth Debt Adjusted R2 F-statistic Inflection points

Coefficient

t-Statistic

−1.695 4.115 −14.020 12.792 0.185 0.315 −0.083 0.287 6.840 20.34% 52.73%

−3.40∗∗ 2.16∗ −2.13∗ 2.18∗∗ 5.17∗∗ 2.27∗∗ −0.25

Note: The dependent variable is firm performance, measured by Tobin’s q. Dir is the percentage of shares outstanding held by directors. Dir2 (Dir3 ) is the percentage of shares held by directors squared (cubed). Size is the natural log of the firm’s equity market capitalization. Growth is the average annual growth in sales from 1998 to 2000. Debt is the total value of debt divided by the book value of total assets. The inflection points of a cubic function are calculated by assuming that all other variables are constant: Tobin’s q = 4.115606Dir − 14.020903(Dir2 ) + 12.79241(Dir3 ) The inflection points are determined by differentiating Tobin’s q with respect to Dir, letting dq/d Dir = 0 and solving for q. If d2 q/d Dir2 > 0, the inflection point is the maximum, if d2 q/d Dir2 < 0, the inflection point is the minimum. Statistical significance at the one and five percent level is indicated by ∗∗ and ∗ respectively.

40

GURMEET S. BHABRA ET AL.

expected sign. These results are consistent with the claim that management moves from alignment, to entrenchment, and back to alignment as their ownership level in the firm increases. We also note that the statistical significance of Size and Growth are consistent with the results that Short and Keasey (1998) report for their sample of U.K. firms. 5.2. Where Does Entrenchment Occur? We also examine in Table 2 whether Singaporean managers become entrenched at an ownership level that is different from those documented for U.S. and U.K. firms. We calculate that the inflection points for the cubic function of director ownership occurs at 52.73% (maximum) and at 20.34% (minimum). Table 3 presents our findings and compares them to others in the existing literature. The inflection point where initial alignment turns to entrenchment in Singapore is most comparable to that reported by Short and Keasey (1998) for a set of U.K. firms. It is lower than the 37.6% reported by McConnell and Servaes (1990), but higher than the 5% level in the widely cited study by Morck, Shleifer and Vishny (1988). Kole (1994) explains the difference between these studies by the difference in the sample size of the firms. Kole notes that Morck, Shleifer and Vishny used only large firms whereas McConnell and Servaes’s sample consists of smaller firms. To the extent that Singaporean firms are smaller than those in the U.S. or the U.K., our results are more appropriately comparable to the values that McConnell and Servaes report. The second inflection point captures the transition from entrenchment back to alignment. We find that this point is again approximately equivalent between Singaporean and U.K. firms, but is higher than the corresponding value for U.S. firms estimated by Morck, Shleifer and Vishny (1988). Table 3. Comparison of Inflection Points in the Relationship Between Firm Value and Insider Equity Ownership. Country

U.S.

Study

Morck, Shleifer and Vishny (1988) 5% 25%

Minimum inflection point Maximum inflection point

U.S. McConnell and Servaes (1990) 37.6% NA

U.K.

Singapore

Short and Keasey (1998)

Bharba, Ferris, Peck and Sen (2003) 20.34% 52.73%

15.58% 41.84%

Note: This table is a comparison of inflection points in the relationship between firm value and insider equity ownership as determined by various researchers across different national economies.

Corporate Governance in Singapore

41

Our results suggest that firm performance is positively related to the ownership held by directors over the 0% to 20.34% range, but then becomes negatively related when ownership falls within the 20.34% to 52.73% range. It again becomes positive when the level of director equity ownership exceeds 52.73%.

5.3. The Incentive Effect of Ownership in Founder Firms In Table 4 we test our second hypothesis by examining whether the relationship between firm value and director equity ownership is different for founder controlled firms. Specifically, we test for evidence of a positive relation between firm value and the level of equity ownership held by directors. We observe in Table 4 that the impact of director ownership is statistically insignificant for all measures. Most importantly, there is no statistically meaningful relation with the linear directorship ownership term. We conclude that our second hypothesis does not hold and that the equity ownership of founder-managers of firms does not result in the generation of higher firm values. Several phenomena might explain these results. Casson (1999) and Chami (1999) propose that founding families view their firms as assets to bequeath rather than wealth to be consumed. Hence, founder-managers might be more interested in firm survival rather than shareholder wealth maximization. Morck, Schleifer and

Table 4. Relationship Between Firm Value and Director Equity Ownership: Founder Sub-sample. Variable Intercept Dir Dir2 Dir3 Size Growth Debt Adjusted R2 F-statistic

Coefficient

t-Statistic

−1.807 7.435 −18.668 13.539 0.162 0.251 −0.232 0.327 3.194

−0.78 0.46 −0.55 0.60 1.83∗ 2.09∗ −0.48

Note: The dependent variable is firm performance, measured by Tobin’s q. Dir is the percentage of shares outstanding held by directors. Dir2 (Dir3 ) is the percentage of shares held by directors squared (cubed). Size is the natural log of the firm’s equity market capitalization. Growth is the average annual growth in sales from 1998 to 2000. Debt is the total value of debt divided by the book value of total assets. Statistical significance at the five percent level is indicated by ∗ .

42

GURMEET S. BHABRA ET AL.

Vishny (1988) find that in older firms, Tobin’s q is lower for firms with founding members as CEOs. Johnson, Magee, Nagarajan and Newman (1985) suggest that founders might exert a negative influence on firm performance by noting positive abnormal returns to the announcement of a founder’s sudden death. Gomez-Mejia, Nunez-Nickel and Gutierrez (2001) extend this line of reasoning and argue that founder ownership leads to greater managerial entrenchment. Founders with strong equity positions in their firms are also more capable of placing a family member in senior executive positions such as CEO. By holding the CEO position, the founder/family can more closely align the firm’s actions with their own interests which may or may not align with those of shareholders. Further, choosing the CEO and other senior managers from the restricted pool of family members potentially excludes more qualified external personnel which can lead to poorer operating performance.

5.4. The Impact of Government Ownership In this section, we examine the extent to which government affiliation influences the relation between firm value and the level of equity ownership held by directors. Government ownership might provide monitoring comparable to that of a blockholder and thereby offset whatever negative influences that its investment might exert. In Table 5 we present our findings from a regression analysis that introduces a dummy variable into Eq. (1). The dummy assumes a value of one for a GLC firm and is zero otherwise. We find in panel A of Table 5 that the director equity ownership terms exhibits a pattern and statistical significance comparable to that observed for the aggregate sample. The dummy variable itself is statistically insignificant. Thus, the results in panel A are consistent with the proposition that government ownership exerts no impact on the relation between firm performance and the level of director equity holdings. In panel B we introduce a set of slope dummy variables to test for the possibility that the influence of government ownership on firm value might be more subtle. This model specification allows us to test whether the impact of government ownership is localized and occurs at select levels of director ownership. Our findings are inconsistent with government ownership exerting a significant impact at any level of director equity holdings. We conclude from Table 5 that government equity ownership in a firm does not change the nature of the relationship between firm performance and director equity holdings. We continue to observe a progression from an initial incentive effect to entrenchment and then back to alignment. It appears that the

Corporate Governance in Singapore

43

Table 5. Relationship Between Firm Value and Director Equity Ownership: GLC Sub-sample. Variable Panel A: Intercept dummy Intercept Dir Dir2 Dir3 Size Growth Debt GLC Dummy Adjusted R2 F-statistic Panel B: Intercept and slope dummies Intercept GLC Dummy Dir (Dir × GLC Dummy) Dir2 (Dir2 × GLC Dummy) Dir3 (Dir3 × GLC Dummy) Size Growth Debt Adjusted R2 F-statistic

Coefficient

t-Statistic

−1.619 4.401 −14.699 13.263 0.175 0.303 −0.095 0.122 0.282 5.901

−3.17∗∗ 2.25∗ −2.21∗ 2.24 4.59∗∗ 2.17∗ −0.28 0.71

−1.600 0.146 4.315 −11.758 −14.522 113.025 13.169 −252.295 0.174 0.292 −0.075 0.259 4.055

−3.07∗∗ 0.70 12.04∗∗ −0.19 −2.05∗∗ 0.07 2.12∗∗ −0.04 4.47∗∗ 2.02∗∗ −0.22

Note: The dependent variable is firm performance, measured by Tobin’s q. Dir is the percentage of shares outstanding held by directors. Dir2 (Dir3 ) is the percentage of shares held by directors squared (cubed). Size is the natural log of the firm’s equity market capitalization. Growth is the average annual growth in sales from 1998 to 2000. Debt is the total value of debt divided by the book value of total assets. GLC is a dummy variable that assumes a value of 1 if the firm is a government linked corporation and is zero otherwise. Dir×GLC Dummy (Dir2 ×GLC Dummy) [Dir3 ×GLC Dummy] represents the slope dummy for the linear (square) [cubic] directors’ equity ownership variable. Statistical significance at the one, five and ten percent level is indicated by ∗∗ and ∗ respectively.

presence of a government investor, regardless of the extent of that investment, is unable to circumvent the entrenchment that occurs with middling levels of share ownership. But neither does government ownership frustrate the incentive effects of equity ownership that operate at both low and high levels of equity ownership.

44

GURMEET S. BHABRA ET AL.

6. CONCLUSIONS This study examines the nature of the relationship between firm value and director equity holdings in an international environment. Our findings are both consistent with earlier results for the Western firms and different. We find that the relationship between firm value and director equity ownership is non-linear and hence consistent with that observed in previous studies for U.S. and U.K. firms. We find that the effect of director equity ownership is to initially align management’s interests with those of its shareholders and then a transition to entrenchment as higher levels of ownership provide insulation against hostile takeovers and labor market discipline. But at still higher levels of ownership, the entrenchment effect is reversed within Singaporean firms and director/shareholder interests are realigned. We find however that the points at which the effect of director ownership transitions between alignment and entrenchment effects differ for Singaporean firms compared to that documented for U.S. or U.K. firms. We find that firm value is positively related to director ownership in the 0% to 20.34% range, negatively related in the 20.34% to 52.73% range, and again positively related when directors’ ownership exceeds 52.73%. These inflection points are much higher than those reported in the frequently cited study by Morck, Shleifer and Vishny (1988) of U.S. firms. As part of this analysis we examine a set of founder-controlled firms to see if the incentive effects of director ownership dominate any entrenchment effects that might occur. We find for these firms that the relationship between firm value and director ownership is insignificant. We interpret these results as consistent with the argument that the incentive effects of equity ownership are offset by the negative factors associated with founder control. These include a focus on firm survival rather than shareholder wealth maximization and managerial entrenchment whereby individuals are promoted to senior positions based on bloodline rather than merit or performance. We conclude our analysis of the effect of director equity ownership on the value of Singaporean firms by examining government linked corporations. Our findings indicate that the presence of the Singaporean government as a blockholder in the firm does not change the relationship between firm value and director equity holdings. We continue to observe a progression from alignment to entrenchment with an ultimate return to alignment. Our findings are consistent with the claim that the impact of insider equity ownership on firm value is not static, but rather varies with the level of that ownership. The finding that the relationship holds in a non-Western economy suggests that there are more similarities than differences across the international capital

Corporate Governance in Singapore

45

markets regarding the challenges in designing an effective corporate governance architecture.

NOTES 1. In the 1999 Global Competitiveness Report conducted by the Geneva-based World Economic Forum (WEF), Singapore continued to be ranked as the world’s most competitive economy despite the currency turmoil present in the Asian region. Singapore’s GDP per capita for the period 2000–2008 is projected to grow according to the WEF by 5.02%, making it the fastest growing economy in the world. 2. The Central Provident Fund (CPF) was introduced as a nationally funded pension on 1 July 1955 under the British colonial government. The purpose of this program was to provide financial security for workers upon retirement or when they are unable to work. Following independence in 1965, the CPF remained intact as it conformed to the government’s development strategy, which relied on high levels of savings and investment. However, the People’s Action Party government made gradual and innovative changes to the scheme. From 1968, contribution rates for employees and employers were gradually increased, resulting in a peak contribution of 25% by 1984.

REFERENCES Brickley, J. A., Lease, R. C., & Smith, C. W., Jr. (1988). Ownership structure and voting on antitakeover amendments. Journal of Financial Economics, 20(1/2), 267–292. Casson, M. (1999). The economics of the family farm. Scandinavian Economic History Review, 47, 10–23. Chami, R. (1999). What’s different about family businesses? Working paper, University of Notre Dame and the International Monetary Fund. Demsetz, H. (1983). The structure of ownership and the theory of the firm. Journal of Law and Economics, 26, 375–390. Denis, D. J., Denis, D. K., & Sarin, A. (1997). Ownership structure and top executive turnover. Journal of Financial Economics, 45(2), 193–221. Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of Law and Economics, 26, 301–325. Gomez-Mejia, L., Nunez-Nickel, M., & Gutierrez, I. (2001). The role of family ties in agency contracts. Academy of Management Journal, 44, 81–95. Grossman, S., & Hart, O. (1982). Corporate financial structure and managerial incentives. In: J. McCall (Ed.), The Economics of Information and Uncertainty (pp. 107–137). Chicago, IL: University of Chicago Press. Jensen, M. (1986). Agency cost of free cash flow, corporate finance and takeovers. American Economic Review, 76, 323–329. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 4, 305–360. Johnson, B., Magee, R., Nagarajan, N., & Newman, H. (1985). An analysis of the stock price reaction to sudden executive deaths: Implications for the management labor market. Journal of Accounting and Economics, 7, 151–174.

46

GURMEET S. BHABRA ET AL.

Kole, S. R. (1994). Measuring managerial equity ownership: A comparison of sources of ownership data. Journal of Corporate Finance, 1, 413–435. McConnell, J. J., & Servaes, H. (1990). Additional evidence on equity ownership and corporate value. Journal of Financial Economics, 27, 595–612. Mikkelson, W. H., & Ruback, R. S. (1985). Takeovers and managerial compensation: A discussion. Journal of Accounting and Economics, 7(1/3), 233–238. Morck, R., Shleifer, A., & Vishny, R. W. (1988). Managerial ownership and market valuation: An empirical analysis. Journal of Financial Economics, 20, 292–315. Shleifer, A., & Vishny, R. W. (1986). Large shareholders and corporate control. Journal of Political Economy, 95, 461–488. Short, H., & Keasey, K. (1998). Managerial ownership and the performance of firms: Evidence from the U.K. Journal of Corporate Finance, 5, 79–101. Stiglitz, J. E. (1985). Credit markets and the control of capital. Journal of Money, Credit and Banking, 17(2), 133–152. Stulz, R. M. (1988). On takeover resistance, managerial discretion and shareholder wealth. Journal of Financial Economics, 20, 25–54. Weston, J. F. (1979). The tender takeover. Mergers and Acquisitions, 74–82.

METHOD-OF-PAYMENT CHOICE FOR INTERNATIONAL TARGETS Kathleen P. Fuller and Michael B. Glatzer ABSTRACT Though cross-border acquisitions have grown dramatically in value and frequency in the last ten years, little is known about returns to acquirers or their method-of-payment choice. This paper studies returns to U.S. bidders and their method-of-payment choice for acquisitions of foreign targets. Results indicate that bidder returns are higher for cash offers, for offers to private and subsidiary targets, if there is high insider ownership, and if there is high exchange rate variation. The method-of-payment choice for these bidders is linked to the target country’s legal regime and accounting standards, insider ownership, target type, and value uncertainty.

1. INTRODUCTION International takeover activity increased three-fold from 1998 to 1999, and European takeover activity alone (measured in dollar volume) overtook U.S. takeover activity for the first time in 2000. Interestingly, a significant portion of this growth involved cross-border deals. Of the cross-border deals in 1999, over $145 billion involved U.S. acquirers and over $293 billion involved U.S. targets.1 Given the benefits to cross-border mergers including entry into new markets, reduction of risk, and exploitation of technology from the home market, this growth is not surprising.2 Interestingly, unlike the U.S. takeover market where stock dominated Corporate Governance and Finance Advances in Financial Economics, Volume 8, 47–64 Copyright © 2003 by Elsevier Science Ltd. All rights of reproduction in any form reserved ISSN: 1569-3732/PII: S1569373203080034

47

48

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

as a method of payment in the 1990s, cash was used more often when a U.S. bidders acquired a foreign target. On average, 41% of these transactions were financed with cash while only 30% of domestic target mergers and acquisitions were financed with cash.3 Myers and Majluf (1984) argue that a bidder will use stock as the medium of exchange if the board believes that its own shares are overvalued. Since target shareholders know this, they are not inclined to accept a stock offer.4 However, if there is uncertainty regarding the target’s value, the bidder may not want to offer cash as the target will only accept a cash offer greater than its true value and the bidder will have overpaid. Hansen (1987) solves this dilemma by suggesting that since stock offers have a “contingency pricing effect,” the target is forced to share part of the risk if the bidder overpays when evaluating a stock offer. Thus, bidders offer stock when there is high target uncertainty.5 Martin (1996) empirically tests these predictions for U.S. mergers and finds that stock offers are more likely if there is greater uncertainty about the bidder and the target. Given the dramatic growth of the U.S. stock market during the 1990s, arguably many firms were significantly overvalued (as hindsight would confirm). Thus, the method-of-payment literature would predict that a domestic bidder should have offered stock regardless of the target’s nationality. Further, if foreign targets have more value uncertainty, due to weaker accounting standards, corporate governance structure, foreign tax laws, etc., one might expect more stock offers than cash offers. Yet, U.S. acquirers offered cash to foreign targets and stock to domestic targets. Thus, other factors must have influenced the method-of-payment choice for international mergers beyond valuation asymmetries. This paper examines what factors influence the cross-border payment method choice by U.S. bidders. This study is warranted for several reasons. First, the pattern of payment methods provides an area to further the understanding of corporate strategy, which is becoming increasingly more global and complex. Second, as the economy becomes more global, most opportunities for growth will involve expanding into foreign markets. Thus, firms will be concerned with which payment method they choose for this growth. Finally, both target and bidder investors will be concerned not only with returns associated with the various offer types but also the ownership consequences of such an offer. Fourth, Black (2000) predicts that future merger waves will be global in nature and no longer a U.S.-only phenomenon. Therefore, the more understanding corporations have regarding international mergers, the better their strategic decision-making abilities. We begin by calculating the wealth impact for a sample of 421 mergers by U.S. bidders for foreign targets between 1995 and 2001. We find that bidders receive significant and positive announcement-day abnormal returns when making cash

Method-of-Payment Choice for International Targets

49

offers and insignificant returns for stock offers. However, bidders making offers for private and subsidiary targets have significantly positive abnormal returns while those bidders making offers for public targets have insignificant returns. If we partition both on offer type and target type, bidders have significant and positive returns for private and subsidiary targets acquired with cash but significant and negative returns for public targets acquired with stock. Further, these returns are increasing in the relative size of the merger, the level of insider ownership, and the past exchange rate variability. Next, we examine what factors influence the method-of-payment choice for international mergers. Using a logistic regression of cash versus stock, we test to see whether asymmetric information, exchange rate uncertainty, tax codes, corporate governance structure, and insider ownership impact the method-ofpayment choice. Results indicate that bidders are more likely to use stock if the target is a private firm or subsidiary and if there is a high bidder or target uncertainty. However, the bidder is less likely to use stock if the insider ownership is high, if the target is located in a civil law country, and if there is less value-relevant accounting information. Thus, U.S. bidders for foreign targets are concerned not only by asymmetric information but also with the corporate governance structure in the target’s country. Previous research on cross-border deals is limited and has focused solely on the wealth effects of these deals. Markides and Ittner (1994), Doukas and Travlos (1988), Cakici, Hessel and Tandon (1996), Eckbo and Thorburn (2000), Kiymaz and Mukherjee (2000), and Black, Carnes and Jandik (2001) all examine announcement day returns to cross-border acquisitions.6 Most of these studies find that U.S. bidders have insignificant or significantly negative returns for foreign targets. However, similar to our results, Doukas and Travlos and Black, Carnes and Jandik find that U.S. bidders have significantly positive announcement day returns for foreign targets. One reason for the inconclusive evidence may be the sample period used. Previous studies have focused on mergers from the 1980s, with the exception of Black, Carnes and Jandik who examined mergers from the late 1980s and early 1990s. Further, we partition our sample by offer type and target type, while other studies do not make these distinctions. Besides examining the wealth effects of cross-border deals, our analysis focuses on the method-of-payment chosen. This helps increase the understanding of why firms may make potentially value-destroying offers. The remainder of the paper is organized as follows. Section 2 discusses the factors that may impact the method-of-payment choice for international mergers. Section 3 describes the data and tests. Section 4 presents the results, and Section 5 concludes.

50

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

2. FACTORS IMPACTING INTERNATIONAL METHOD OF PAYMENT Several possible reasons exist for the dominant use of cash to purchase foreign targets. First, similar to domestic mergers, asymmetric information about the bidder or target may impact the method-of-payment choice. Myers and Majluf (1984) predict that if the bidder is overvalued, it should be more likely to offer stock than cash. This would have been more likely in the 1990s when the U.S. stock market experienced huge positive growth. Further, if the target’s value is unknown, the bidder may be more likely to use stock to force the target to make an efficient acceptance decision, as suggested by Hansen (1987). If foreign targets are difficult to value, due to accounting differences, political issues, or greater informational asymmetries, bidders should offer stock more often than cash. Second, and maybe most importantly, the corporate governance structure in the target firm’s country may influence the method-of-payment choice in many ways. For example, French law requires an investor who obtains more than a third of the outstanding voting shares of a corporation to bid on the remaining outstanding shares, so as to attain two-thirds of the voting shares. Although a stock transfer, or offre publique d’´echange, is an option, this choice generally requires an extraneous meeting of the shareholders to agree on this tender. “This disadvantage might lead the stock exchange authorities to impose a cash tender offer when launching of a take-over bid is made compulsory” (S. G. Warbourg Group, 1990). If this forced cash tender offer is the norm, then it is not surprising that a large number of U.S. acquisitions of French firms are cash transactions. La Porta et al. (1998) find that civil-law countries provide investors the weakest protection and weakest enforcement while common-law countries provide the highest protection and high enforcement. Since we examine only U.S. bidders, theoretically the legal protection and enforcement in the target country should not matter. Thus, U.S. bidders should be indifferent between offering cash or stock. Yet, if the merger were funded by stock and the U.S. bidder expected that the foreign legal system could adversely influence its operations or its new foreign shareholders, the bidder might wish to use cash instead. That is, U.S. acquirers may offer cash to foreign target shareholders in civil-law countries so to avoid weaker levels of protection and enforcement. Further, the country’s accounting standards (often government set or strongly influenced) may influence the method-of-payment choice. Alford et al. (1993) and Ali and Hwang (2000) suggest that countries with little demand for information from financial reports will have accounting practices that generate less relevant financial data. If the accounting data is value-relevant, U.S. bidders should be more likely to price the targets accurately. However, as the value-relevance declines, so too does the price accuracy. Thus, U.S. acquirers may be more likely to make stock

Method-of-Payment Choice for International Targets

51

offers for targets in countries with less value-relevant accounting information so as to force targets to share in any overpayment. Taxation of capital gains and dividends may also impact the target shareholder’s likelihood of accepting a stock offer relative to a cash offer. Stockholders who receive dividends from French-based corporations can obtain a 50% tax credit if the corporation has already been taxed on their income. In other words, dividends are not fully doubly taxed. However, if the corporation distributing the dividends is not a French-based company, then the dividends are taxed at the ordinaryincome rate. Depending on the dividend value this tax law alone could deter shareholders from accepting stock offers. The United Kingdom (U.K.), which had the greatest number of cross-border mergers between 1995 and 2000 (1,103), has a similar requirement for dividend taxation. U.K. shareholders are fully taxed on dividend payments of foreign corporations, but companies can bypass this tax by establishing a U.K. subsidiary that distributes the dividends. Yet, this technique could be prohibitively costly. Thus, if U.S. firms are unwilling to establish U.K. subsidiaries to pay out non-taxable dividends, then U.K. shareholders may not be willing to accept a stock tender offer as future dividends are highly taxed. Further, some countries force a mandatory dividend payment, e.g. Brazil.7 If a U.S. bidder acquired a Brazilian target with stock, strong pressure (by target shareholders or the Brazilian government) could be applied to the U.S. bidder to start paying dividends or increase existing dividends. Thus, U.S. bidders may prefer to offer cash and not stock. Third, though exchange rate fluctuations are irrelevant for domestic targets, they may influence the payment choice for international mergers. All else equal, the target knows exactly how much it will receive for its shares and the bidder knows how much it pays when a cash offer is made. When the bidder makes a stock offer, the target and bidder know how many shares will be exchanged but given stock price movements, the precise value of the deal will be unknown. To reduce this price uncertainty the target and bidder may prefer cash offers. However, if the exchange rate between the U.S. dollar and the foreign currency is highly volatile, the value of a cash deal also becomes uncertain. Yet, in this case, a stock deal would protect target shareholders from exchange rate movements, as they could continue to hold the bidder’s stock until exchange rates move favorably. Thus, one would expect bidders to make cash (stock) offers when exchange rates are less (more) volatile or exchange rates are expected to move in their favor (against them). If the currency movements can be easily hedged, then cash would once again dominate as the preferred method of payment. However, the cost to individual investors of hedging against current exchange rate movements may be prohibitively costly. Finally, insider ownership could potentially explain the preference towards making cash offers for foreign targets. If bidder insider ownership is significant,

52

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

then the firm may be less likely to make a stock offer so as to maintain control of the company, especially if the foreign target is large relative to the U.S. bidder. However, Martin (1996) suggests that for very low and very high levels of ownership, the impact of dilution is irrelevant. Yet, in some middle range, managers will be very concerned with maintaining control and, thus, be less likely to offer stock.

3. DATA AND METHODOLOGY We collect from Securities Data Corporation’s (SDC) Mergers and Acquisitions Database a list of successful mergers for foreign targets announced between January 1, 1995 and December 31, 2001.8 To be included in the sample, the following conditions must be satisfied: (1) The target is a public firm, a private firm, or a subsidiary of a public firm. (2) The target firm has a disclosed dollar value and the bidder acquires more than 50% of the target firm. (3) The deal value is one million dollars or more. Deal value is defined as the total value of consideration paid by the acquirer, excluding fees and expenses. The dollar value includes the amount paid for all common stock, common stock equivalents, preferred stock, debt, options, assets, warrants, and stake purchases made within six months of the announcement date of the transaction. (4) Acquiring firms are U.S. firms publicly traded on the Amex, NASDAQ, or NYSE and have 230 days of return data around the takeover announcement listed on the Center for Research in Security Prices (CRSP) file. (5) The target’s country’s exchange rate data is available for six-months prior to merger announcement. (6) Neither the acquirer nor the target is a utility or a financial institution. (7) The U.S. firm’s insider ownership is available from proxy statements or annual reports prior to the merger announcement. To avoid bid-ask bias in the announcement-period abnormal returns, we exclude bids where the bidder stock price is below two dollars. Our final sample includes 382 unique acquirers making 421 bids. We group the method of payment into two categories, cash and stock. Cash financing includes any combination of cash, debt, and/or liabilities. Stock offers include payments with common stock or any combination of common stock, options, and/or warrants. Table 1 reports the summary statistics for the firms making multiple acquisitions and their targets. As expected the bidder’s market and book values are much larger than the target’s.

Number of Observations Target Market Value (or Offer Value) Target Book Value Target Market-to-Book Ratio Bidder Market Value Bidder Book Value Bidder Market-to-Book Ratio Exchange Rate Variance Insider Ownership for Bidder

421 421 421 421 421 421 421 421

Mean

$331.167 $112.386 7.652 $8,959.140 $10,329.330 14.325 0.030 14.579%

Standard Deviation

Minimum

747.291 212.484 17.617 25,521.330 35,785.990 95.942 0.067 0.166

$6.726 $0.900 0.335 $29.091 $5,700 0.049 0.003 1.000%

Maximum

Median

$4,561.800 $1,142.920 131.565 $215,725.500 $213,016,000 860.398 0.587 88.400%

$98.057 $41.840 2.682 $1,855.790 $664.900 2.123 0.016 7.950%

Method-of-Payment Choice for International Targets

Table 1. Summary Statistics.

Note: Summary statistics for 424 acquisitions announced between January 1, 1995 and December 31, 2001. ($ reported in millions.)

53

54

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

Table 2. Target Countries and Number of Mergers from Each Country. Argentina Australia Belgium Brazil Canada Denmark Finland France Germany Hong Kong Ireland-Rep Israel Italy Japan Mexico Netherlands New Zealand Norway Singapore South Africa Spain Sweden Switzerland Thailand United Kingdom

4 20 3 4 125 8 2 15 16 7 6 12 2 2 6 8 2 4 2 3 3 10 4 1 152

Table 2 indicates the number of targets and the countries of each target. Though a disproportionate number of mergers occur in the U.K. and Canada, a wide range of countries is represented. To calculate the bidder’s abnormal returns we use the market-model methodology outlined in Brown and Warner (1980, 1985). The ordinary-least-squares coefficients of the market-model regression are estimated over the period t = −200 to t = −30. We then estimate the abnormal returns as follows: AR i = r i − [␣ˆ i − ␤ˆ i r m ]

(1)

where ri is the return on firm i, ␣ˆ i and ␤ˆ i are the ordinary-least-squares estimates of the market-model parameters, and rm is the equally weighted market index return. Finally, we calculate cumulative abnormal returns (CARs) for the three-day period (−1, 1) around the announcement date. Since univariate results do not control for numerous factors that conceivably contribute to CARs, we conduct multivariate tests to control for these factors. The

Method-of-Payment Choice for International Targets

55

following regression is estimated using weighted least squares with the weight as the reciprocal of each firm’s forecast variance, CAR = ␣ + ␤1 RELSIZE + ␤2 OFFER + ␤3 TARGET + ␤4 EU + ␤5 SIC + ␤6 EXVAR + ␤7 BOWN + ␤8 GOV + ␤9 DIV + ␤10 VALREL + ␧ (2) where CAR is the firm’s three-day cumulative abnormal return, RELSIZE is the log of the relative size of the merger, OFFER is a dummy variable that equal one if the offer is made with stock and zero if made with cash, TARGET is equal to one if the target is either a private target or a subsidiary and zero if a public target, EU is a dummy variable for whether the merger was announced after the euro was introduced, SIC is a dummy variable that equals one if the target and bidder have the same two-digit SIC code and zero otherwise, EXVAR is the six-month variance in the target’s country exchange rate one week prior to announcement, BOWN is a dummy variable that equals one if the insider ownership is between 5 and 25% and zero otherwise,9 GOV is a dummy variable that equals one if the target is located in civil-law country and zero if located in a common-law country, DIV is a dummy variable that equals one if the target firm is located in a country with mandatory dividends or adverse dividend tax regulation and zero otherwise, RELVAL is a dummy that equals one if the target is in a country that has less value-relevant accounting information and zero otherwise, and ␧ is the error. The relative size of the merger is the target’s market value divided by the bidder’s market value where market values are measured one month prior to the merger announcement. For private firms and subsidiaries no market value exists, so we use the offer price for the target’s market value. To measure value-relevant accounting information, we use data from Alford et al. (1993), Mueller, Gernon and Meek (1984), and Ali and Hwang (2000) to create a series of dummy variables. Alford et al. and Ali and Hwang find that value relevance is lower for countries with bank-oriented rather than market-oriented financial systems, where private sector bodies are not involved in the standard-stetting process, where accounting practices follow the Continental model as opposed to the BritishAmerican model, where tax rules have greater influence on financial accounting measurement, and where spending on auditing services is relatively low. We do not include the amount spent on auditing services since this information was unavailable for all countries in our sample. Further, we do not include a dummy variable for bank-oriented versus market-oriented capital systems since this would be almost identical to the dummy variable for Continental versus BritishAmerican model. Finally, following La Porta et al. (1998) the target country’s legal regime is classified.10

56

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

To examine the characteristics that influence the method-of-payment choice, we utilize a logistic function. The logit procedure denotes xi as the vector of values of the characteristics for firm i. The probability that firm i chooses method-of-payment j, where j = 0 for cash offers, and j = 1 for stock offers is 

Prob[Y = 1] =

e␤ x 

1 + e␤ x

(3)

where: ␤ x = ␣ + ␤1 RELSIZE + ␤2 TARGET + ␤3 EXVAR + ␤4 TUNCRT + ␤5 BUNCRT + ␤6 BOWN + ␤7 EU + ␤8 SIC + ␤9 GOV + ␤10 DIV + ␤11 VALREL where BUNCRT and TUNCRT are measures of the bidder and target’s value uncertainty, respectively, and all other variables remain as defined above. The firm’s market-to-book ratio for the year prior to the merger is used as a proxy for the firm’s value uncertainty.11 Since the market-to-book is a proxy for growth opportunities of the firm and Chung and Charoenwong (1991) find that higher growth opportunities are associated with riskier common stock, firms with higher market-to-book ratios are assumed to have higher uncertainty. The sample size for the logistic regression falls to 350 mergers because target market and book values are not available for the entire sample. We expect that RELSIZE would be positively related to the bidder’s CARs and the likelihood of a stock offer. At a basic level, the larger the target relative to the bidder, the greater the effect the acquisition has on the bidder, and the more likely a greater market reaction. Further, as the relative size of the merger increases, the likelihood the bidder will have enough cash (given a reasonable constraint on debt capacity or cash on hand) for a cash offer falls. TARGET should be positively related to CARs but inversely related to the likelihood of a stock offer. If the target is a private firm or subsidiary, we would expect the bidder’s CARs to be higher than for a public target due to availability of information, competition, etc.12 Further, since private firms and subsidiaries tend to be smaller and cheaper, a bidder may be more likely to make a cash offer. For reasons discussed earlier DIV, GOV, and BOWN should be negatively related to the likelihood of stock offers, while SIC, RELVAL, EXVAR, TUNCRT, and BUNCRT should be positively related to stock offers. DIV and GOV should both be negatively related to returns since any regulations that dictate the bidder’s ability to manage the merged entity effectively will have a negative impact on future profitability. However, insider ownership should be positively related to CARs. If the bidder ownership is high, then managers would be less likely to take value-destroying mergers and

Method-of-Payment Choice for International Targets

57

returns to any merger announcements should be higher for firms with more insider ownership. We predict SIC will be positively related to returns since if the bidder and target are in the same industry, the bidder already has expertise in that industry thus making the integration of the target’s operations easier resulting in higher profitability. RELVAL should be positively related to returns since as more valuerelevant accounting information means the bidder is less likely to misprice the target, to under or overestimate synergies, or to misevaluate operating performance. We do not include BUNCRT and TUNCRT in Eq. (2) since we hypothesize that the offer type is highly correlated with these two variables.13 We expect EXVAR will have a negative impact on returns. If a country has more variation in exchange rates, then there is more risk, and more risk would lead to lower abnormal returns. Finally, we include EU to control for any fundamental shift caused by the adoption of one currency in Europe; however, we have no predicted relation between EU and CARs or the method-of-payment choice.

4. RESULTS Table 3 reports the three-day CARs to bidders classified by type of target and method of payment. For all bids, the CAR is a statistically significant positive 0.36%. However, when we differentiate the returns on the basis of whether the target was public or non-public, we find that the CAR is a significantly negative 0.54% for public targets, significantly positive 0.71% for private targets, and significantly positive 1.67% for subsidiaries. If the target is public, bidder returns are insignificant if the bid is made with cash but a negative and significant 1.51% if the bid is made with stock. The CARs are positive and significant for the private target sample for cash offers (1.38%) and insignificant (−0.22%) for stock offers. The market also views acquisitions of subsidiaries positively if cash is used as the method-of-payment (2.07%) but insignificant if stock is used (−0.21%). Our results differ from Fuller, Netter and Stegemoller (2002) in that they find bidder returns for private and subsidiary targets are significantly positive regardless of offer type. The results for foreign targets are not surprising, though. First, in Fuller, Netter and Stegemoller the majority of targets are U.S. firms that would have less overall risk than foreign targets and therefore higher bidder returns. Second, foreign targets will have higher negotiation and information acquisition costs than U.S. targets which will reduce bidder returns. Table 4 reports the multivariate results of Eq. (2). Results indicate that the relative size of the merger, the offer type, the target public status, the exchange rate variation, and insider ownership are all significant determinants of bidder CARs. As we found in the univariate results, bidders that made stock offers

58

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

Table 3. Cumulative Abnormal Returns of Acquirers. Offer Type All

Cash

Stock

All targets

0.363%a [1.02%] 421

0.993%a [0.910%] 250

−0.602% [−0.401%] 171

Public targets

−0.540%c [−0.057%] 156

0.152% [0.339%] 105

−1.508%a [−1.877%] 51

Private targets

0.705%a [0.154%] 190

1.383%b [0.632%] 87

−0.219% [−0.260%] 103

Subsidiary targets

1.665%a [1.712%] 75

2.071%a [2.001%] 60

−0.209% [0.175%] 15

Note: Cumulative abnormal returns for bidders that acquired 421 foreign targets (public, private or subsidiary targets) between January 1, 1995 and December 31, 2001. Cumulative abnormal returns are calculated for the three days (−1, 1) around the announcement (day 0) of a takeover. All acquirers are publicly traded firms listed on the NYSE, NASDAQ, or AMEX with a stock price of two dollars or greater in the month of the takeover announcement. Cash offers include cash only and mixtures of cash and debt. Stock offers include common stock only or a combination of common stock and options, warrants, or rights. The median CAR is reported in brackets below the average CAR, and the number of bids is reported below the median. a Denotes significance at 1% level. b Denotes significance at 5% level. c Denotes significance at 10% level.

have significantly lower returns that those making cash offers. Also, if the target is a private or subsidiary of a firm, the bidder’s returns are significantly higher. As predicted, as the relative size of the merger increases so do bidder returns. Interestingly, if the exchange rate exhibits higher past variability, bidder returns are significantly higher. One possible explanation is as follows. Merger announcements can be viewed as call options on the target assets. Any increase in the volatility of those assets’ value (higher exchange rate volatility) would increase the value of the call option and thus increase returns. Insider ownership is also positive and significant.14 The market rewards higher insider ownership with higher returns. Intriguingly, none of the corporate governance variables (dividend regulations, legal regimes, or value-relevant accounting information) significantly impact returns. The value relevance results are consistent with Black, Carnes and Jandik (2001) who find no significant relation between long-run returns to

Method-of-Payment Choice for International Targets

59

Table 4. Ordinary Least Squares Regression Analyses of Cumulative Abnormal Returns of International Bidders. Independent Variables Intercept RELSIZE OFFER TARGET EU SIC EXVAR BOWN GOV DIV VALREL = GAAP

−0.0400 [0.172] 0.0104 [0.032] −0.0263 [0.037] 0.0345 [0.012] −0.0079 [0.516] −0.0012 [0.925] 0.0119 [0.001] 0.0250 [0.054] −0.0150 [0.420] −0.0065 [0.612] 0.0098 [0.607]

−0.0342 [0.362] 0.0103 [0.034] −0.0269 [0.031] 0.0346 [0.012] −0.0076 [0.545] 0.0000 [0.998] 0.0119 [0.001] 0.0250 [0.056] −0.0125 [0.592] −0.0069 [0.589]

−0.0164 [0.649] 0.0102 [0.035] −0.0255 [0.041] 0.0345 [0.012] −0.0081 [0.514] −0.0007 [0.958] 0.0118 [0.001] 0.0259 [0.046] −0.0250 [0.230] −0.0079 [0.536]

−0.0050 [0.841]

VALREL = ACC

−0.0242 [0.290]

VALREL = FIN VALREL = ‘Worse Regime’ F-statistic N Adjusted R2

−0.0404 [0.165] 0.0096 [0.050] −0.0276 [0.027] 0.0329 [0.017] −0.0065 [0.601] 0.0003 [0.979] 0.0120 [0.001] 0.0260 [0.045] −0.0018 [0.907] −0.0088 [0.496]

3.65 [0.000] 421 10.20%

3.62 [0.000] 421 10.10%

3.76 [0.000] 421 10.58%

−0.0232 [0.273] 3.77 [0.000] 421 10.62%

Note: Ordinary least squares regression of the bidder’s three-day cumulative abnormal return on the following variables. RELSIZE is the relative size of the target is the log of target market value, or deal value as reported by SDC, divided by acquirer market value as of the month before the announcement date. OFFER is a dummy variable equals zero if the target is acquired cash and one if stock. Cash offers include cash only and mixtures of cash and debt. Stock offers include common stock only or a combination of common stock and options, warrants, or rights. TARGET is a dummy variable equals one if the firm was either private or a subsidiary target and zero if the target was a public firm. EU is a dummy variable that equals one if the merger occurred after 1999 or zero if before 1999. SIC is a dummy variable for whether the target and bidder were in the same industries, measured by two-digit SIC code. EXVAR is the six-month variance in the target’s country exchange rate one week prior to announcement. BOWN is the ownership percentage of bidder insiders prior to the merger. GOV is a dummy variable that equals one if the target country’s legal system is based on civil law and zero if based on common law. DIV is a dummy variable that equals one if the target firm is located in a country with mandatory dividends or adverse dividend tax regulation and zero otherwise. VALREL can be one of four dummy variables based on the nationality of the target. VALREL = GAAP is a dummy variable that equals one if national accounting standards are set by governmental bodies only and 0 if private-sector bodies are also involved. VALREL = ACC is a dummy variable that equals one if the country is classified as British-American model or zero if the country is a Continental model country. VALREL = FIN is a dummy variable that equals one if the level of alignment of financial and tax accounting is low and zero if it is high. VALREL = ‘Worse Regime’ is a dummy variable that equals one GAAP generates less timely and valuerelevant information than U.S. GAAP and zero if either more or equally timely and value-relevant. p-values are reported in brackets below to the parameter estimates.

60

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

Table 5. Logistic Regression Analysis of Factors Affecting Method of Payment Choice. Independent Variables Intercept RELSIZE TARGET EXVAR TUNCRT BUNCRT BOWN EU SIC GOV DIV VALREL = GAAP VALREL = ACC VALREL = FIN VALREL = ‘Worse Regime’

Model 1 −0.8116 [0.144] 0.0102 [0.914] 1.0341 [0.026] 0.3089 [0.222] 0.0038 [0.340] 0.0011 [0.175] −0.7288 [0.026] 0.2757 [0.396] 0.2541 [0.446] −0.5036 [0.046] −0.1178 [0.733] −1.0372 [0.045]

−1.3313 [0.132] 0.0184 [0.844] 1.0920 [0.018] 0.2769 [0.269] 0.0032 [0.422] 0.0009 [0.241] −0.7120 [0.029] 0.3085 [0.340] 0.2600 [0.431] −0.8006 [0.076] −0.0809 [0.813]

−2.1733 [0.011] 0.0286 [0.763] 1.0605 [0.023] 0.3040 [0.231] 0.0037 [0.361] 0.0011 [0.152] −0.7745 [0.019] 0.3393 [0.296] 0.2853 [0.392] −0.2990 [0.098] −0.0200 [0.954]

Model 2 −0.9016 [0.101] −0.0017 [0.986] 1.0761 [0.019] 0.2763 [0.272] 0.0032 [0.422] 0.0011 [0.170] −0.6764 [0.037] 0.3030 [0.345] 0.2709 [0.412] −0.9873 [0.018] −0.1584 [0.643]

0.4571 [0.502]

−1.1233 [0.132] 0.1156 [0.264] 0.7572 [0.100] 0.2191 [0.561] 0.8706 [0.067] 2.0553 [0.001] −0.7871 [0.022] 0.1178 [0.732] −0.0506 [0.875] −0.5846 [0.100] −0.1033 [0.778] −1.0120 [0.057]

−1.8040 [0.071] 0.1215 [0.238] 0.8059 [0.081] 0.2182 [0.558] 0.7968 [0.091] 2.0246 [0.001] −0.7841 [0.021] 0.1692 [0.622] −0.0309 [0.930] −0.7984 [0.087] −0.0416 [0.558]

−2.2319 [0.024] 0.1249 [0.227] 0.8304 [0.073] 0.2076 [0.557] 0.7832 [0.096] 1.9946 [0.001] −0.8004 [0.019] 0.1711 [0.617] −0.0033 [0.992] −0.5567 [0.112] −0.0144 [0.968]

−1.2787 [0.086] 0.1157 [0.271] 0.8317 [0.071] 0.18990 [0.547] 0.7792 [0.098] 2.0255 [0.001] −0.7602 [0.026] 0.1330 [0.696] −0.0203 [0.954] −1.1973 [0.008] −0.0803 [0.826]

0.5443 [0.432] 1.2963 [0.111]

0.9719 [0.138] −0.6061 [0.319]

−0.0082 [0.989]

Note: Logistic regression of the offer type, cash = 0 and stock = 1 on the following variables. RELSIZE is the relative size of the target is the log of target market value, or deal value as reported by SDC, divided by acquirer market value as of the month before the announcement date. TARGET is a dummy variable equals one if the firm was either private or a subsidiary target and zero if the target was a public firm. EXVAR is the six-month variance in the target’s country exchange rate one week prior to announcement. BUNCRT and TUNCRT are measures of the bidder and target’s value uncertainty, respectively. In Model 1 TUNCRT and BUNCRT are measured as the market-to-book ratio for the year prior to the announcement. Model 2 also uses the market-to-book ratio as a proxy for uncertainty but as a dummy variable. If the firm’s market-to-book ratio is greater than one, TUNCRT or BUNCRT equals one and zero otherwise. BOWN is the ownership percentage of bidder insiders prior to the merger. EU is a dummy variable that equals one if the merger occurred after 1999 or zero if before 1999. SIC is a dummy variable for whether the target and bidder were in the same industries, measured by two-digit SIC code. GOV is a dummy variable that equals one if the target country’s legal system is based on civil law and zero if based on common law. DIV is a dummy variable that equals one if the target firm is located in a country with mandatory dividends or adverse dividend tax regulation and zero otherwise. VALREL can be one of four dummy variables based on the nationality of the target. VALREL = GAAP is a dummy variable that equals one if national accounting standards are set by governmental bodies only and 0 if private-sector bodies are also involved. VALREL = ACC is a dummy variable that equals one if the country is classified as British-American model or zero if the country is a Continental model country. VALREL = FIN is a dummy variable that equals one if the level of alignment of financial and tax accounting is low and zero if it is high. VALREL = ‘Worse Regime’ is a dummy variable that equals one GAAP generates less timely and value-relevant information than U.S. GAAP and zero if either more or equally timely and value-relevant. p-values are reported in brackets below to the parameter estimates.

Method-of-Payment Choice for International Targets

61

U.S. bidders of foreign targets and value-relevant accounting information in their regression analysis. Finally, the introduction of the euro had no significant impact on returns, nor did whether the two firms were in the same industry. Table 5 presents the results from the logistic regression in Eq. (3). Model 1 presents the results using the target’s and bidder’s market-to-book ratio as the proxy for uncertainty. Model 2 also uses the market-to-book ratio as a proxy for uncertainty but as a dummy variable. If the firm’s market-to-book ratio is greater than one, the firm is classified as having high uncertainty, and if it is less than one, it has low uncertainty. That is, TUNCRT (BUNCRT) equals one if the target’s (bidder’s) market-to-book ratio is greater than one and zero otherwise. For both Model 1 and 2, BOWN and GOV are negative and significant. Thus, for targets located in civil law countries with less investor protection and enforcement, bidders are less likely to use stock. When insiders own a significant portion of the firm’s shares and wish to maintain control of the company, they are also less likely to use stock as the method of payment. Contrary to our hypothesis, TARGET is positive and significant. Thus, bidders are more likely to offer stock to private firms and subsidiaries than cash. It may be that the owner(s) of the private firm or key employees of the subsidiary have vital knowledge or expertise that the bidding firm wants to keep, so it offers stock to keep this knowledge or expertise in the merged entity. The value-relevant accounting information is negatively related to stock when measured as GAAP. However, for the other definitions of value relevance, the variables are insignificant. Since Ali and Hwang (2000) find all these measures highly correlated, our results weakly suggest that stock offers are less likely when there is less value-relevant accounting information available. In Model 1 and Model 2, RELSIZE, DIV, EXVAR, EU, and SIC have no significant impact on the method-of-payment choice. In Model 1, TUNCRT and BUNCRT are also insignificant. However, for Model 2, both TUNCRT and BUNCRT are positive and significant. Thus, as predicted by the method-of-payment literature and found in Martin (1996) for domestic mergers, stock offers are more likely if there is greater uncertainty about target or bidder values.

5. CONCLUSION We examine the returns to U.S. bidders acquiring foreign targets. We find these bidders have significantly higher returns when using cash than stock for public, private and subsidiary targets. Further, these returns are increasing in the relative size of the merger, the level of insider ownership, and the past exchange rate variability. Delving into the method-of-payment choice, we find that bidders are more likely to use stock if the target is a private firm or subsidiary and if there is

62

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

high bidder or target uncertainty. However, the bidder is less likely to use stock if the insider ownership is high, if the target is located in a civil law country, and if there is less value-relevant accounting information. Overall the results suggest that the method-of-payment choice by U.S. bidders for foreign targets is impacted not only by the same variables that impact the method-of-payment choice for U.S. targets, but is also strongly affected by the corporate governance structure of the target’s country.

NOTES 1. See Black (2000), Wessel (2000), and Black, Carnes and Jandik (2001). 2. See, for example, Madura and White (1990), Davis, Shore and Thompson (1991), Morck and Yeung (1991), and Heston and Rouwenhorst (1994) for motivations of international mergers. 3. See Andrade, Mitchell and Stafford (2001) and Mergerstat Review 2000. 4. Berkovitch and Narayanan (1990), Fishman (1989), and Eckbo, Giammarino and Heinkel (1990) expand on this idea and show that higher-valued bidders will use cash or a higher proportion of cash to signal their value to the market. 5. Eckbo and Thorburn (2000) develop a model showing that when the bidder’s value is known and the target’s value is uncertain, the “expected overpayment cost of cash” is greater than the “expected overpayment cost of stock.” Therefore, bidders prefer to make stock offers. 6. Black, Carnes and Jandik (2001) focus mainly on the long-run returns to U.S. bidders of foreign targets. They find U.S. bidders have significantly negative returns for both three-year and five-year windows after the merger. 7. See La Porta et al. (1998, 2000) for a discussion of dividends around the world. 8. We verified the announcement dates listed on SDC. Only 28 mergers had the incorrect announcement date that we changed. Usually, the merger was announced within three days of the SDC announcement date. 9. The ownership percentages of 5 to 25% are chosen somewhat arbitrarily, though commonly used. See Martin (1996) and Morck, Shleifer and Vishny (1998). However, we conducted robustness tests using percentages between 5 and 75% using 5% increments with qualitatively similar results. 10. La Porta et al. (1998) find that German-civil-law and Scandinavian countries have weaker laws than common-law countries but the highest quality of enforcement. As a robustness check, we allow GOV to equal one if the target is located in a French-civil-law country and zero for common-law, German-civil-law, or Scandinavian countries. Results are qualitatively similar. 11. For the private and subsidiary targets the offer value is used as the market value and the book value of assets is provided by SDC. 12. See Fuller, Netter and Stegemoller (2002) for a discussion of how bidder returns differ depending on the target’s public status. 13. Martin (1996) found this relation held for domestic mergers. Thus, to the extent that returns to bidders for domestic targets are similar to those for foreign targets, we would be misestimating the regression.

Method-of-Payment Choice for International Targets

63

14. To ensure the monotonic and increasing relation between insider ownership and returns, we used the percentage insider ownership as the independent variable (BOWN) in Eq. (2). Results were qualitatively the same as using the dummy variable for ownership ranges.

ACKNOWLEDGMENTS We are grateful to Jeff Bacidore, Jay Coughenour, seminar participants at the University of Georgia, and participants at the 2001 University of Georgia CURO Symposium, the 2001 National Conference on Undergraduate Research in Lexington, KY, and the Eastern Finance Association Conference in Baltimore, MD. All remaining errors are solely our responsibility.

REFERENCES Alford, A., Jones, J., Leftwich, R., & Zmijewski, M. (1993). The relative informativeness of accounting disclosures in different countries. Journal of Accounting Research, 31, 183–233. Ali, A., & Hwang, L. (2000). Country-specific factors related to financial reporting and the value relevance of accounting data. Journal of Accounting Review, 38, 1–21. Andrade, G., Mitchell, M., & Stafford, E. (2001). New evidence and perspectives on mergers. Journal of Economic Perspectives, 15, 103–120. Berkovitch, E., & Narayanan, M. (1990). Competition and the medium of exchange in takeovers. Review of Financial Studies, 3, 153–174. Black, B. (2000). The first international merger wave (and the fifth and last U.S. wave). University of Miami Law Review, 54, 799–818. Black, E., Carnes, T., & Jandik, T. (2001). The long-term success of cross-border mergers and acquisitions. Working Paper, University of Arkansas. Brown, S., & Warner, J. (1980). Measuring security price performance. Journal of Financial Economics, 8, 205–258. Brown, S., & Warner, J. (1985). Using daily stock returns. Journal of Financial Economics, 14, 3–31. Cakici, N., Hessel, C., & Tandon, K. (1996). Foreign acquisitions in the United States: Effect on shareholder wealth of foreign acquiring firms. Journal of Banking and Finance, 20, 307–329. Chung, K., & Charoenwong, C. (1991). Investment options, assets in place, and the risk of stocks. Financial Management, 23, 70–74. Davis, E., Shore, G., & Thompson, D. (1991). Continental mergers are different. Business Strategy Review, 2, 49–70. Doukas, J., & Travlos, N. (1988). The effect of corporate multinationalism on shareholders’ wealth: Evidence from international acquisitions. Journal of Finance, 43, 1161–1175. Eckbo, E., Giammarino, R., & Heinkel, R. (1990). Asymmetric information and the medium of exchange in takeovers: Theory and tests. Review of Financial Studies, 3, 651–675. Eckbo, E., & Thorburn, K. (2000). Gains to bidder firms revisited: Domestic and foreign acquisitions in Canada. Journal of Financial and Quantitative Economics, 35, 1–25.

64

KATHLEEN P. FULLER AND MICHAEL B. GLATZER

Fishman, M. (1989). Pre-emptive bidding and the role of the medium of exchange in acquisitions. Journal of Finance, 44, 41–57. Fuller, K., Netter, J., & Stegemoller, M. (2002). What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions. Journal of Finance, 57, 1763–1793. Hansen, R. (1987). A theory for the choice of exchange medium in mergers and acquisitions. Journal of Business, 60, 75–95. Heston, S., & Rouwenhorst, K. (1994). Does industrial structure explain the benefits of international diversification? Journal of Financial Economic, 36, 3–27. Kiymaz, H., & Mukherjee, T. (2000). The impact of country diversification on wealth effects in cross-border mergers. Financial Review, 35, 37–58. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., & Vishny, R. (1998). Law and finance. Journal of Political Economy, 106, 1113–1155. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., & Vishny, R. (2000). Agency problems and dividend policies around the world. Journal of Finance, 60, 1–33. Madura, J., & White, A. (1990). Diversification benefits of direct foreign investment. Management International Review, 30, 73–85. Markides, C., & Ittner, C. (1994). Shareholder benefits from corporate international diversification. Journal of International Business Studies, 25, 343–360. Martin, K. (1996). The method of payment in corporate acquisitions, investment opportunities, and management ownership. Journal of Finance, 51, 1227–1246. Mergerstat Review (2000). Applied financial information LP, Los Angeles, 14–16 and 98–101. Morck, R., Shleifer, A., & Vishny, R. (1998). Management ownership and market valuation: An empirical analysis. Journal of Financial Economics, 20, 293–316. Morck, R., & Yeung, B. (1991). Foreign acquisitions: When do they make sense? Managerial Finance, 17, 10–18. Mueller, G., Gernon, H., & Meek, G. (1984). Accounting: An international perspective. New York: McGraw-Hill/Irwin. Myers, S., & Majluf, N. (1984). Corporate financing and investment decisions when firms have information investors do not have. Journal of Financial Economics, 87, 355–374. S. G. Warbourg Group (1990). Mergers and acquisitions in Europe (Vol. I & II). Professional Publishing, England. Wessel, D. (2000). Cross-border mergers soared last year – most deals were struck by Western Europeans: Britain overtook U.S. Wall Street Journal (July 19), A18.

ORGANIZATION STRUCTURE AND CORPORATE GOVERNANCE: A SURVEY Mark Hirschey ABSTRACT During recent years, financial economists have made a significant contribution to the rapid development of a vibrant and growing literature on organization structure and corporate governance. In reviewing the development of this literature, it becomes easy to see how the seminal contributions of Ronald Coase (awarded the Nobel Prize in Economics in 1991) have become the cornerstone of a new institutional economics. In particular, researchers following in Coase’s footsteps have clarified the conditions under which voluntary contracts between private agents can resolve a wide variety of so-called “agency problems.” More than just representing an important discovery of the significance of transaction costs and property rights for the institutional structure and functioning of the economy, Coase’s work has become an important foundation for the theory of contracts and for the whole field of “organization economics.”

1. INTRODUCTION Optimal boundaries and the profit-maximizing structure of the firm are not static. Both are dynamic and responsive to the changing needs of the marketplace.1 Corporate Governance and Finance Advances in Financial Economics, Volume 8, 65–112 Copyright © 2003 by Elsevier Science Ltd. All rights of reproduction in any form reserved ISSN: 1569-3732/PII: S1569373203080046

65

66

MARK HIRSCHEY

Before the mid-nineteenth century, hierarchal organization structures were virtually non-existent. Transportation and information technology had not yet progressed to the point where large-scale consolidation in business was possible. This changed radically when steamship and railroad transportation and the telegraph and telephone made for speedy transport and near-instantaneous communication. Suddenly, the means for coordination in large-scale enterprise became available. At the start of the twentieth century, company sizes grew rapidly to take advantage of economies of scale in production to better serve nations hungry for industrial and consumer products. This process of industrial consolidation flourished throughout the early post-World War II era. The emerging information age of the twenty-first century promises to further refine corporate boundaries. Whereas the post-World War II corporate environment was designed for the efficient mass production of standardized industrial and consumer goods, present-day customers are more interested in unique combinations of goods and services to meet specialized needs. Just as the industrial revolution transformed the competitive landscape of the twentieth century, so too is the information revolution transforming corporate boundaries. Highly skilled managers no longer need to locate in New York, Chicago, or Los Angeles to find exciting career opportunities. Powerful desktop computers, the Internet and low-cost satellite-based communications make it possible to work in Los Angeles while living on Catalina Island; in Sante Fe, New Mexico; or Durango, Colorado. Centralized authority supported by layers of managerial staff is being replaced by desktop computers and sophisticated software that make line personnel instantly accountable for bottom-line performance. Chief financial officers do not want detailed reports filed by cumbersome staffs of accounting and financial personnel; they want immediate access to the data with user-friendly software. Ramifications of this ongoing revolution in information technology for corporate organizations should not be underestimated.2 Lei and Slocum (2002) point out that the growing importance of knowledge workers, rapid innovation of substitute products, and rising importance of information costs are redefining how firms compete. Traditional strategies and organization designs are often ill-suited for firms seeking to learn new technologies and prepare themselves to meet new competitive challenges.3 Key strategic imperatives include learning new ways to pursue self-cannibalization of older technologies, buy-out emerging entrants, identify the skill sets of potential competitors, and foster new product innovation through parallel development teams. In knowledge-based organizations, front-line employees need to know that new ideas, which may challenge an existing set of practices, will not be stifled or suppressed. It is essential that firms design and implement effective reward systems. As Madhok (2002) argues, understanding the organization of

Organization Structure and Corporate Governance

67

economic activity requires a sensitivity to the interdependence of production and exchange relations. In an efficient setting, an alignment exists between organization structure, corporate governance, and the substance of exchange. The purpose of this essay is to provide an interpretive survey of recent research on organization structure and corporate governance. In this literature, corporate governance is broadly defined as the system of controls that helps corporations and other organizations effectively manage, administer, and direct economic resources. This literature deals with the role played by boards of directors, organization design and corporate governance mechanisms inside the firm, franchise agreements, and strategic alliances, among other topics. An emerging area of inquiry delves into the affects of mandatory corporate governance mechanisms outside the firm, such as auditor-client relationships, and the influences of federal, state, and local laws and regulations that govern corporate behavior. Studies of the corporate governance implications of the ownership structure of the firm, or the complex array of divergent claims on the value of the firm, are another important area of inquiry.4 Before considering problems encountered in arriving at an optimal corporate design, it is worth considering briefly the very nature of the firm.

2. TRANSACTION COSTS AND THE NATURE OF FIRMS 2.1. Nature of the Firm Nobel laureate Ronald Coase is justly famous for his work on defining the nature of the firm. At its core, Coase (1937) saw the firm as a nexus of contracts, or a collection of binding agreements among owners, managers, workers, suppliers, and customers. It has no physical presence; it exists only as a legal device. A new and still-growing appreciation for what is sometimes described as “organization economics” started with Coase’s (1937) explicit introduction of transaction costs into economic analysis. Still, the seminal contributions by Coase have been augmented and strengthened by the contributions of Oliver Williamson, Harold Demsetz, George Stigler, and a host of colleagues in law, anthropology, sociology, political science, sociobiology and other disciplines. Williamson (1998, 2000) argues persuasively that institutions matter, and institutions are susceptible to analysis. The proposition that institutions matter is now embraced by most economists. What distinguishes the new approach to organization economics is the concept that institutions are susceptible to analysis within the context of a positive research agenda. As Demsetz (1997) suggests, traditional neoclassical theory’s objective is to understand price-guided, not

68

MARK HIRSCHEY

management-guided, resource allocation. The firm does not play a central role in neoclassical theory. At the core of neoclassical theory, perfect competition is based upon the assumption of known technology and prices. Management has no real influence. In present-day economic theory, economists have followed Coase in relaxing key assumptions of neoclassical theory. No longer is it taken for granted that all markets function freely, prices and technology are known by all parties, and owners are freely effective in controlling the uses of their assets. In contemporary theory, economists consider the effects of positive transactions costs, and a productive role for management in dealing with agency problems tied to shirking, opportunism, reputation problems, and quality uncertainty. The late George Stigler, a noted colleague of Coase, was awarded the Nobel Prize in economics for 1982. Stigler remains widely known for his study of markets, the structure of industries, and the economics of regulation, with special reference as to how regulations are developed within the context of the political system. According to Stigler, the foundation of Coase’s contribution to modern economic theory rests on two fundamental propositions: (1) There are transaction costs and they are large; (2) people tend to arrange their affairs so as to minimize the sum of their transaction costs and deadweight losses arising from their failures to exploit gains from trade (see Ellickson, Stigler & Donohue, 1989). Coase taught that, when it is to the benefit of people to reach an agreement, they will seek to do so. Reaching agreements is typically costly in terms of time and other resources. When agreements fail to materialize, an unfavorable comparison between small benefits or the large costs is to blame. Predictions based upon the assumption of zero transaction costs will seldom hold in the real world. From this perspective, the efficiency of firms depends upon the ability of participants to find effective means to minimize the transactions costs of coordinating productive activity. Important categories of transactions costs encountered within the firm include information costs, decision costs, and enforcement costs. To appreciate the importance of transaction costs in defining the nature of firms it will prove useful to consider each category in some detail. Search or information costs encompass expenses encountered in discovering the type and quality of goods and services demanded by consumers. Information costs also include expenses encountered in securing necessary raw materials, attracting and training a skilled workforce, developing brand name recognition, and so on. Bargaining or decision costs include expenditures involved with successfully negotiating production agreements. For example, labor bargaining is sometimes referred to as a continuously renegotiable agreement between management and labor. Even with a long-term union contract, the labor bargain is continuously renegotiable in the sense that the trade-off between pay and effort varies daily according to the level of effort expended. Reluctant workers do not work very hard. Policing

Organization Structure and Corporate Governance

69

or enforcement costs include charges necessary to make sure that all parties live up to their contractual commitments. Supervisory costs are an example of enforcement costs. Some transactions costs involve cash outlays for insurance, legal fees, and so on. However, many transaction costs involve implicit or opportunity costs. Necessary delays to facilitate consensus building, group meetings, and constant communication among team members are all manifestations of transaction costs. In the early post-World War II period, notable transaction costs associated with coordinating large-scale production and distribution of industrial and consumer goods were more than offset by significant economies of scale in production. As a result, top firms in the industrial world reached gigantic size in terms of assets, employment, sales revenues, and profits (see Hannah, 1998). More recently, consumers have tended to prefer distinctive goods and services tailored to specific needs. Economies of scale in production tend to be much less important when production needs are specialized and are often nonexistent in the provision of services. At the same time, the costs of coordinating activity among increasingly specialized and highly educated professionals tends to be much higher than the costs of coordinating effort among less specialized workers. Therefore, it is not surprising to note the recent downsizing trend among giant corporations throughout the industrial world and the coincident growth in importance of entrepreneurship and small business.

2.2. Coase Theorem According to Coase, firms exist as an economic force because they offer an effective means for minimizing transaction costs. It would be prohibitively expensive for each of us to organize production of all goods and services that we desire. For example, most consumers are unprepared to deal with the complexity of building a home. As a result, most consumers contract with real estate developers, who contract with builders who, in turn, contract with carpenters, electricians, plumbers, landscapers, and so on. If sufficient ongoing demand exists, builders employ their own staff of professionals; they form a firm. If demand is sporadic, contracting will ensue prior to the start of each independent home building project. In deciding whether or not to retain their own staff of building professionals, builders compare the costs of negotiating for the completion of individual home projects with the costs of maintaining a less than fully utilized staff during slack periods. In the same way, law firms, medical practices, and professional consultants consider transaction costs in decisions concerning the scope and scale of professional enterprises.

70

MARK HIRSCHEY

Similarly, Coase sees firms as well-equipped to deal with traditional problems caused by negative externalities, such as pollution. Because individuals respond to economic incentives, they seek out and undertake mutually beneficial trades. For example, suppose a power plant reduced recreational opportunities by fouling a nearby river. Such a company might appease local residents by purchasing land for general recreation use upriver or by matching public tax spending to reduce or eliminate down river pollution. The power plant might also be willing to pay some other local polluter, say, a municipal water treatment plant, to reduce its pollution, and hence the total amount of emissions, to an acceptable level. This latter alternative might prove especially attractive if the marginal cost of reducing pollution by the municipal water facility is much less than the marginal cost of reducing power plant emissions. In fact, some local jurisdictions award “pollution rights” that are traded among polluters who seek the lowest cost method for reducing total emissions. According to what is now referred to as the Coase Theorem, resource allocation will be efficient so long as transaction costs remain low and property rights can be freely assigned and exchanged. As conventionally described, Schroeder (1998) notes that the perfect or ideal market concept describes a strange world without time, space, objects, or subjects. Indeed, in the absence of transactions costs or market frictions of any kind, no transactions will occur. Coase seeks to describe how and why transactions occur in real world markets where transactions costs and market frictions are common. The transaction cost concept is important in financial economics because it provides the context necessary for considering the economic significance of organization structure. For example, Anderson, Glenn and Sedatole (2000) offer evidence on the correlates of sourcing decisions in the U.S. auto industry to see whether adoption of new contracting terms and early involvement of suppliers in design activities yields different results as compared to previous findings. Using data on 156 sourcing decisions for process tooling of a new car program, the authors find that transaction cost theory offers a useful explanation for sourcing decisions. In a similar vein, Dobrev, Kim and Carroll (2002) argue that while the niche concept figures prominently in contemporary theories of organization, economists often fail to tie micro processes within the niche to long-term changes in the broader environment. They argue that a significant relationship exists between an organization’s niche and evolution in the structure of its organizational population over time. Dobrev, Kim and Carroll (2002) focus on technological niches and the process of positioning and crowding among firms in the niche space. Both are related to the level of concentration among all firms in the market. Dobrev, Kim and Carroll (2002) show that niches and concentration interact in complex ways, and together play an important role in organizational evolution over time.

Organization Structure and Corporate Governance

71

In a less sympathetic framework, Dixit and Olson (2000) ask if the very nature of voluntary participation undermines the Coase Theorem. Whereas the Coase Theorem states that the enforcement of voluntary agreements yields efficient outcomes, Dixit and Olson (2000) argue that previous treatments fail to recognize the full meaning of the voluntary concept. They argue that voluntary agreements require a two-stage game involving a non-cooperative participation decision, followed by Coaseian bargaining among those who choose to participate. Using a simple public-goods model, outcomes ranging from extremely inefficient to fully efficient are illustrated. However, their efficient equilibriums are not robust to even very small transaction costs, thus causing Dixit and Olson (2000) to cast doubt on Coase’s claim of universal efficiency through voluntary bargaining. Such considerations have given rise to a number of dynamic models of coalition formation in cooperative games. For example, Arnold and Schwalbe (2002) posit a model in which players decide which of the existing coalitions to join, and demand payoffs at each step along the way. These decisions are determined by a best-reply rule, given the coalition structure and allocation in the previous period, and myopically suboptimal strategies whenever there are potential gains from trade. Alternatively, Barr and Saraceno (2002) view firms as learning algorithms. The costs and benefits of processing information are linked to the structure of the firm and its relationship with the environment. Within this framework, the firm is modeled as a type of artificial neural network. Increased vertical separation within the firm as implied, for instance, by subcontracting or by additional vertical layers of management yields investment in a more flexible technology. In contrast, increased horizontal separation as implied by lack of cooperation among different (horizontal) divisions within the firm has ambiguous implications on flexibility. When divisions of the organization confront significantly different levels of uncertainty, horizontal separation enhances technological flexibility. Otherwise, when the extent of uncertainty confronting different divisions is comparable it is cooperation that yields greater flexibility of the technology. The attributes of the technology selected by a given firm may depend upon the internal structure of its competitor if those attributes can be observed by the competing firm. In particular, the firm chooses a less flexible technology if its competitor is vertically separated rather than integrated. To understand the importance of organization structure in the ongoing success of firms, it is necessary to learn how organization structure is dictated by the nature of the firm’s economic environment and competitive strategy. As a first step in this process, it is necessary to ask: Why are there economic conflicts within firms?

72

MARK HIRSCHEY

3. AGENCY PROBLEMS: SOURCES OF CONFLICT WITHIN FIRMS 3.1. What Is the Firm’s Agency Problem? A large and growing corporate governance literature has built upon the work of Jensen and Meckling (1976) who consider the role of corporate control mechanisms as means for helping ameliorate any divergence of interests between managers and stockholders. Problems in corporate governance exist to the extent that unresolved material conflicts exist between the self-seeking goals of (agent) managers and the value maximization goal of (principal) stockholders. Agency costs incurred by stockholders are reflected in expenses for managerial monitoring, the over-consumption of perquisites by managers, and lost opportunities due to excessive risk avoidance. While Jensen and Meckling’s (1976) agency cost characterization of the corporate governance issue is fairly recent, modern concern with the topic began more than sixty years ago when Berle and Means (1932) predicted that managers with little direct ownership interest, and thus having “own” rather than stockholder interests in mind, would come to run the bulk of U.S. business enterprise by the latter part of the 20th century. Before them, economists’ concern with the “other people’s money” problem dates from Adam Smith (1776), who noted that people tend to look after their own affairs with more care than they use in looking after the affairs of others. Agency problems exist because of conflicts between the incentives and rewards that face owners and managers. Such conflicts commonly arise given ownermanager differences in risk exposure, investment horizons, and/or familiarity with investment opportunities. Problems mount because differing economic incentives between owners and managers can cause predictable, but hard to correct, hurdles that must be overcome if the shareholder’s value maximization objective is to be achieved. 3.2. Risk Management Problems Significant differences in the risk exposure of managers and stockholders often leads to an excessive risk-taking problem. In 1995, for example, London’s famous Barings Bank, a bank founded in 1762 that had helped fund Britain’s war effort against Napoleon, was brought to its knees given excessive risk taking by a single individual. A trader by the name of Nick Leeson in Baring’s Singapore branch had been charged with the responsibility of overseeing the branch’s risk-free arbitrage business. In risk-free arbitrage, banks and other financial institutions seek to profit

Organization Structure and Corporate Governance

73

by taking advantage of price discrepancies in different markets. By instantaneously buying and selling the same security in different markets, arbitrage by banks and other financial institutions can result in small but risk-free profits. Unfortunately, when imperfect hedging led to losses rather than profits, Leeson began to engage in the far riskier business of foreign currency market speculation. Leeson guessed wrong and lost $1.4 billion of the bank’s money. Although Leeson ended up in jail, Barings was sold to ING, the large Dutch financial institution, for a mere £1. The Barings bank episode is an obvious manifestation of the “other people’s money” problem. It stemmed from the fact that Leeson was not speculating with his own resources; he was gambling with Barings Bank funds. Other related agency problems tied to differences between the investment horizons of stockholders and management can also emerge. Salary and bonus payments tied to short-term performance often constitute a large part of the annual total compensation package earned by management. Thus, managers typically have huge personal incentives to turn in favorable year-to-year growth in revenues, profits, and earnings per share. This can sometimes have the unfortunate effect of focusing managerial attention on near-term accounting performance to the detriment of long-term value maximization. To combat such myopic behavior, more and more companies are insisting that managerial compensation be directly tied to long-term performance. An efficient means for establishing this link is to demand that top management hold a significant stock position that cannot be sold until some time after retirement. The Barings bank fiasco, and more recent corporate debacles at Enron, Inc., WorldCom, Inc., and Tyco International, among others, point to the need for better risk management theory and practice. In an interesting survey of risk management practice, Yu (2002) acknowledges that while strategy formation frequently delves into modern portfolio theory to explore frameworks for factoring risk into strategic decision making, the gap between theory and management practice remains large. Surveys of management practice find that many managers are simply not familiar with models useful for risk-assessment, management, and control. A large number of managers appear to believe their experience and intuition are more reliable than models based on forecasts, or add fudge factors based upon incomplete sensitivity analyses. Similarly, Raz, Shenhar and Dvir (2002) argue that even in this era of heightened competition and globalization, many projects suffer delays, overruns, and even failure because risk management tools and techniques are used too little by project managers. Raz, Shenhar and Dvir (2002) base this assessment on data collected on over 100 projects performed in Israel in a variety of industries, the extent of usage of some risk management practices, such as risk identification, probabilistic risk analysis, planning for uncertainty and trade-off analysis, the difference in application across different types of projects, and their impact on various project success dimensions. Raz, Shenhar and Dvir’s (2002) findings suggest that

74

MARK HIRSCHEY

risk management practices are still not widely used. At the same time, when risk management and control practices are employed, Raz, Shenhar and Dvir (2002) find that their use appears related to project success. These findings on the lack of widespread use for common approaches to risk assessment, management, and control, suggest the need for new and better approaches. One such method is offered by Berkowitz (2001) who argues that the forecast evaluation literature has traditionally focused on methods of assessing point forecasts. In the context of many models of financial risk, however, interest centers on more than just a single point of the forecast distribution. For example, value-at-risk models that are currently in extremely wide use form interval forecasts. Many other important financial calculations also involve estimates not summarized by a point forecast. Although some techniques are currently available for assessing interval and density forecasts, existing methods tend to display low power in the small sample sizes typically available. Berkowitz (2001) suggests a new approach to evaluating such forecasts. It requires evaluation of the entire forecast distribution, rather than a scalar or interval. The information content of forecast distributions combined with ex post realizations is enough to construct a powerful test even with sample sizes as small as 100. More such approaches are needed.

3.3. Investment Horizon Problems The typical CEO of a giant U.S. corporation is 55 to 60 years old, has been with the company 20 to 25 years, and looks forward to a term in office of 8 to 10 years. With this demographic background, most top executives can only expect to see the fruits of their actions at the top of the organization if such benefits accrue quickly as opposed to slowly. Similarly, many managers pin their hopes for promotion on the basis of results achieved during fairly short time frames. As a result, top executives and many other managers tend to have fairly short time horizons within which they and others within the corporation evaluate investment and operating decisions. This tendency toward focusing on short-term versus long-term results is both reflected in and reinforced by compensation plans that rely on near-term corporate performance.5 During 2002, Business Week reported that the average CEO at a giant U.S. corporation pulled in a stupendous $11 million in total compensation. On average, more than 75% of this total is tied to near-term company and stock-price performance in the form of salary, bonus pay and other compensation. Salary and other compensation (such as health and retirement benefits) is often tied to longevity; bonus pay is typically tied to accounting performance as measured by

Organization Structure and Corporate Governance

75

the annual growth in earnings per share, profitability as captured by the annual rate of return on stockholders’ equity, or other such annual measures of firm performance. Despite the obvious benefits of bonus plans and the incentive-based pay, short-term compensation plans can narrow the focus of top executives to near-term versus long-term corporate performance. To combat the potential for shortsighted operating and investment decisions, sometimes referred to as the managerial myopia problem, most corporations now tie a significant portion of total compensation for top executives and other managers to the company’s long-term stock-price performance. Stock options and other payments tied to stock-price appreciation now account for a significant portion of top executive pay. Similarly, a wide variety of employee stock ownership plans give managers and other employees strong incentives to consider the long-term implications of present-day decisions. For example, Ang, Lauterbach and Schreiber (2002) study how U.S. banks compensates their top management teams. They observe two tiers of compensation. Chief Executive Officers (CEOs) receive not only greater pay in absolute dollar terms, they are also rewarded more in relation to performance because a relatively greater portion of their pay is determined by corporate performance. Such plans work best if they consider stock-price gains over extended periods, say 10 years. To reflect the marginal contribution of a given top executive or managerial team, managers must only be rewarded for above-average stock-price appreciation. Finally, given the advanced age of most top executives and many senior managers, firms must be on guard against what is sometimes referred to as the end-of-game problem. The end-of-game problem is the most serious manifestation of myopic decision making, or inefficient risk avoidance, and reflects the fact that it becomes difficult to discipline poorly performing managers at the end of their career. Young managers have lucrative future job opportunities both inside and outside the firm as an incentive for hard work and honest dealings with their current employer. Older managers enjoy no such future opportunities and face correspondingly weaker incentives for hard work and honest dealing. Like any such problem involving managerial myopia, most firms deal with the end-of-game problem by insisting that senior managers take a significant portion of total compensation in the form of pay tied to long-term stock-price appreciation. Wiersema (2002) argues that, for better or worse, investors tend to view CEOs as a key determinant of corporate performance. When the company does well, CEOs get the credit and cash in lucrative incentive compensation. When the company does poorly, the CEO gets the blame and the boot. Interestingly, Wiersema (2002) reports that many companies perform no better after they dismiss their CEOs than they did in the years leading up to the dismissals. Worse still, organizational disruption created by a rushed CEO firing can lead to a significant and lasting

76

MARK HIRSCHEY

downturn in corporate performance. Wiersema (2002) places the blame for such poor results with boards of directors. Boards often lack the strategic understanding of the company’s business necessary to select an appropriate replacement CEO. In many cases, corporate boards are simply driven to make a quick CEO replacement decision to restore investor confidence, rather than ensuring the replacement’s strategic fit with the needs of the organization. Within this context, CEO failures and related failures of the CEO search process can be interpreted as failures of boards of directors to successfully fulfill their mission. Mace (2002) argues that boards of directors seldom mount an appropriate response to corporate disasters. To ensure more effective boards, Mace (2002) proposes eliminating all insiders from corporate boards other than the chairman and the president, and favors establishing criteria by which the board is required to formally evaluate the annual performance of the president. In a similar vein, Sonnenfeld (2002) argues that good boards are high-functioning groups distinguished by a climate of respect, trust, and candor among board members and between the board and management. Information is shared openly and on time; emergent political factions are quickly eliminated. Members feel free to challenge one another’s assumptions and conclusions, and management encourages lively discussion of strategic issues. Individual directors feel a responsibility to contribute meaningfully to the board’s performance. In addition, good boards assess their own performance, both collectively and individually, on a regular basis.

3.4. Information Asymmetry Problems Yet another potential source of agency problems is tied to management’s inherently superior access to information inside the firm, or the information asymmetry problem. When compared with outside shareholders, management inside the firm has access to more and better information concerning the firm’s relative performance and investment opportunities. By definition, “insiders” know more than “outsiders” about the firm’s performance, prospects, and opportunities. The problem is made more serious by the fact that, despite the conspicuous limitations of accounting earnings information, most indicators of firm and top executive performance typically rely upon accounting measures that are collected and reported by management. Because managers are themselves responsible for the collection and processing of accounting information, they control the reporting mechanism designed to monitor managerial and firm performance. Incentive pay plans linked to accounting performance offer more than just incentives for efficient operating and financial decisions. Such plans also give inducements for accounting earnings manipulation and bias.6

Organization Structure and Corporate Governance

77

Earnings manipulation and bias can occur when managers choose accounting methods that lead to income inflation, income smoothing, or both. Managers have incentives for income inflation when higher reported earnings boost managerial compensation, provide greater job security, or both. Incentives exist for income smoothing to the extent that spectacular short-run performance creates expectations that are difficult or impossible to satisfy and, therefore, leads to stockholder disappointment and sanctions. Incentives for income smoothing are also present if stockholder risk aversion leads to an asymmetry of managerial rewards and penalties following short-term earnings “success” versus “failure.” Recent studies in financial economics document the effects of executive incentive pay plans on the reporting strategy of top managers. Their results confirm that the compensation schemes chosen by shareholders and/or boards of directors have a direct influence on the degree of earnings manipulation and bias. In the modern corporate environment, it is shortsighted to focus concern about poor firm performance on the tendency of inefficient managers to waste money on fancy offices, corporate aircraft, and other such perquisites. Knowledgeable critics of managerial inefficiency are much more troubled about the inherent difficulty of gauging excess compensation, unprofitable empire building, and other elements of managerial malfeasance when managers control the flow of information about firm and managerial performance. Cohan (2002) discusses internal dynamics of firms that contribute to information processing problems. In the wake of the Enron scandal, directors are likely to face much greater demands of accountability. Possible solutions to enhance the effective management of internal and external information include programs to encourage employees to expose wrongdoing without fear of retribution, new communication systems that enable important information to move upward in the organization without distortion, and newly independent audit committees. This recently renewed emphasis on the importance of the auditing function echos Coase’s (1990) long-term interest in the “difficulties” in economic investigation caused by the lack of uniformity in the practice of accountants. Coase (1938) was among the first to formally recognize that if financial economists are to fruitfully examine the costs and benefits of organizing activities within firms, they will need the assistance of accountants. Indeed, Bushman and Smith (2001) propose that research on corporate governance must be extended to comprehensively explore the use of financial accounting information and the effects of financial accounting information on economic performance.7 For example, Klein (2002) contemplates whether audit committee and board characteristics are related to earnings management practices. Klein (2002) finds a negative relation between audit committee independence and abnormal accruals. A negative relation is also found between board independence and abnormal accruals. Reductions in board

78

MARK HIRSCHEY

or audit committee independence tend to be accompanied by large increases in abnormal accruals. The most pronounced effects occur when either the board or the audit committee is comprised of a minority of outside directors. Klein’s (2002) results suggest that boards structured to be more independent of the CEO tend to be more effective in monitoring the corporate financial accounting process.

3.5. Business Ethics Daboub (2002) contends that organizational structure has evolved from vertical integration to the multidivisional form, to the matrix, to the network, and now to the cellular form of organization. Like a living cell, an organization can live alone but can perform more complex functions if it acts in concert with older cells. Daboub (2002) argues that, in many cases, the most efficient way to act in concert is not through mergers but rather through alliances and networks. One obvious downside is that contracts and ethics are more difficult to monitor in these relationships, given that responsibility may often be fragmented. To counter this problem and avoid material ethical lapses, companies need to select ethical partners at the outset and then foster appropriate legal and ethical norms at the individual level and across the relationship. Brickley, Smith and Zimmerman (2002) assert that economic theory and methodology can offer important insight concerning business ethics issues on at least three basic dimensions. First, economics provides a theory of how individuals make choices; including choices that have ethical dimensions. Second, business ethics and the internal structure of the organization are inextricably linked and they establish important incentives for those individuals who together compose the firm. Third, a company’s reputation for ethical behavior is part of its brandname capital. As such, it is reflected in the value of its securities. Some argue that if competitors have adopted low ethical standards, it would be unprofitable for a firm to adopt high standards. However, economic theory and methodology document that this presumption is plainly incorrect. Potential customers discount their demand prices where there is uncertainty about the quality of the product to be supplied. By credibly promising to act ethically, a firm can differentiate their product and increase their demand, potentially by a substantial amount. Similarly, bondholder and stockholders make their cost of capital calculations in light of the company’s ongoing reputation for fair dealing. High quality firms enjoy a relatively lower cost of capital as one of the many benefits tied to their reputations for fair dealing. Low quality firms can expect a relatively higher cost of capital, and higher supplier costs generally, as typical disadvantages tied to their poor reputation.

Organization Structure and Corporate Governance

79

In short, ethical considerations are simply one of the many important concerns that must be effectively addressed through the organization design process. Costly breakdowns in business ethics, like any costly breakdown in firm operations, reflect an important failure of the firms corporate governance process.

4. ORGANIZATION DESIGN 4.1. Factors Dictating an Appropriate Design The value-maximizing organization design minimizes unproductive conflict within the firm. Masten (2002) argues that an optimal organization structure design involves a balanced consideration of how to allocate decision-making authority, how to monitor and evaluate performance, and how to reward productive behavior. Goold and Campbell (2002) provide a practical framework to guide executives through the complexities of organization design. They present nine tests of organizational design that can be used to evaluate an existing structure or to create a new one. The first four tests are called “fit” tests that provide an initial screen for design alternatives; the next five tests are “good design” tests that can help a company find an appropriate balance between empowerment and control. According to Goold and Campbell (2002), an appropriate organization design establishes the necessary amount of hierarchy, control, and process for the design to work smoothly, but not so much so as to dampen personal initiative, flexibility and networking opportunities. According to Stigler (1951), vertical disintegration should be the typical organizational design in growing industries, vertical integration in declining industries. The basic argument is that firms will spin off production stages subject to increasing returns to scale in response to market growth. Elberfeld (2002) re-examines Stigler’s hypothesis within an equilibrium model of industrial structure in which the organization of firms is endogenous. Stigler’s hypothesis is confirmed when entry into markets is free and firms compete. However, when entry into the intermediate good market is restricted, or intermediate good producers collude, vertical integration increases with market size. Rindova and Kotha (2001) emphasize the fact that hotly competitive environments tend to undermine the stability of successful organizational forms. Organizational form, function, and the competitive advantage of firms tend to dynamically evolve over time. In a similar fashion, Insead and Eisenhardt (2001) see corporate divisions as combinations of capabilities and product-market responsibilities in a “dynamic community” that may be recombined in various ways according to the interplay of economic and social imperatives that motivate such recombinations.

80

MARK HIRSCHEY

4.2. Resolving Unproductive Conflict Within Firms Successful firms get close to the customer, efficiently identify customer needs and expectations, and then exceed those expectations. Such firms are effective production and distribution systems. Similarly, they are effective means for collecting and processing a vast array of sometimes conflicting information about customer demands, technology, input prices, raw material supplies, and so on. The information-processing function of firms is sometimes overlooked. Also neglected are the costs and benefits of effective information processing. However, these are vital concepts and key determinants of firm success. Useful information has a price. It must be identified, combined with complementary information, and effectively communicated within the organization. This is made difficult by the fact that bits of useful information are typically dispersed among several different individuals on several different levels up and down the organization. For example, research scientists may encounter problems communicating with engineering staff about problems with the commercial applications of a given discovery. Scientists often share a common language, and it might conflict with the common language of engineers. In turn, both scientists and engineers may confront obstacles in communicating with accounting staff, who themselves share a language that is common to accountants, but foreign to many others. However, the problem is much more serious than just a failure to share a common vocabulary. It is not just that various specialists and subspecialists within the firm talk different languages; communication is made difficult by the fact that many “owners” of valuable information within the firm have conflicting incentives to hide useful details. Constructive communication within the firm is essential if customer needs are to be met effectively. The design of the organization is appropriate if it facilitates the firm’s function as an information-processing mechanism. In considering the design of any organization, basic needs must be met to facilitate worthwhile activity. As shown in Fig. 1, three prime considerations are the needs to allocate decision authority, monitor and evaluate performance, and reward productive behavior. Authority is simply the ability to command, control, or influence. An effective organization design is one that allocates decision authority to that person or team of persons best able to perform a given task or influence a particular outcome. Decision authority, or decision rights, allows individual employees to determine how and when to best deploy the productive resources and valuable information at their disposal. With decision authority comes responsibility. Mishra (2002), for example, considers how the choice of effort and honesty tend to be influenced by various incentive schemes and organizational structures. While clear vertical hierarchies tend to induce less corruption than horizontal organization structures,

Organization Structure and Corporate Governance

81

Fig. 1. Organization Design must Facilitate Effective Decision Making and Efficient Performance.

organizational structure matters most when there are constraints on the rewards and penalties that can be administered. It is therefore imperative to monitor and evaluate performance. Managers and all employees must be held accountable for outcomes tied to individual decisions. Accountability can only be measured in terms of the tangible and intangible rewards derived from productive activity and in terms of the penalties or sanctions tied to unproductive behavior. To minimize the costs of unproductive conflict with the firm, it is essential that the design of the organization effectively allocate decision-making authority, monitor and evaluate performance, and reward productive behavior (see Bates, 1997).

4.3. Centralization versus Decentralization A fundamental question facing management of all companies is the basic issue of when and how to centralize versus decentralize decision-making authority.8 With centralized decision authority, detailed judgments concerning how to best manage corporate resources, deal with suppliers and customers, and so on, are handled by top-line executives within the organization. Many corporations with a centralized, or “top-down,” management style employ a significant central office staff. With decentralized decision authority, front-line employees, often those in direct communication with customers, are empowered to make fundamental judgments concerning how to best serve customer needs. Most companies that feature a decentralized, “bottom-up” or “flat,” organization structure employ a minimal decision support staff in the central office. The decision to feature a centralized or

82

MARK HIRSCHEY

decentralized management structure is often an evolutionary process. At some companies, when the size and scale of operation grows, effective resource management requires greater centralization of decision-making authority. At others, the need to remain nimble in responding to customer needs requires ongoing employee empowerment. Decentralized decision authority works best when local managers or other employees working directly with customers have valuable up-to-the-minute information about the needs and price sensitivity of customers. Decentralized decision-making authority and local or customer-specific knowledge can be a powerful combination when seeking to quickly satisfy local customers. Centralized decision making that requires local managers to seek permission to change prices or reconfigure products takes time that can result in costly delay. An important benefit of decentralized decision making is that it trains local managers and other employees to put the customer first and to accept personal responsibility for satisfying customer needs. On the other hand, decentralized decisions that fail to account for indirect or company-wide consequences can sometimes prove counterproductive (see Nault, 1998). For example, a price cut in one region or for a single product line may divert sales from other regions or products. Local managers often ignore important interaction effects. Similarly, information about customer needs or product quality considerations may be more reliable and used more effectively when pooled from various local operations. At most firms, strategic planning decisions are the responsibility of the chairman and board of directors, as supported by the chief executive officer and central office staff. At the same time, corporations that feature highly centralized strategic planning may emphasize significant decentralization and employee empowerment when it comes to operating decisions. Harris and Raviv (2002) explain that the organization structure actually employed is one based upon an optimal coordination of interactions among activities. While each individual manager is capable of detecting and efficiently coordinating interactions within his or her limited area of expertise, only CEOs can coordinate company-wide interactions. The optimal design of the organization trades off the costs and benefits of various configurations of managers.

4.4. Assigning Decision Rights The question of centralization versus decentralization focuses on defining the appropriate numbers of levels within the organization. A flat organization has few or a single level of decision-making authority; a vertical organization features multiple ascending levels of decision-making authority. Irrespective of which

Organization Structure and Corporate Governance

83

organizational form is appropriate, it becomes necessary to choose where in a given organization structure any given decision should be made. In all instances, the issue is one of making effective use of local or customer-specific knowledge, while at the same time maintaining an effective use of centralized information. In an interesting study, Chompalov, Genuth and Shrum (2002) conduct an empirical analysis of 53 multi-institutional collaborations in physics and allied sciences to investigate generalizations about the essentially informal and collective social organization of collaborative projects in science. This analysis suggests that the variety of organizational formats used in collaborative projects can be grouped into four types, ranging from bureaucratic to participatory. The structure of leadership tends to be related to the character of interdependence in data acquisition, analysis, and communication. In general, greater interdependence among co-workers leads to decentralization of leadership and less formalization. At its most basic level, the production process encompasses a sequence of related tasks, or assignments necessary to effectively meet customer needs. In turn, related tasks are bundled into jobs, when such packaging facilitates cost savings and a more productive use of firm resources. Complex jobs involve the completion of a large number or wide variety of tasks. Simple jobs involve only one elementary chore or a few related responsibilities. Both simple and complex jobs can feature a mix of decision authority. Management trainees may be assigned fairly simple jobs with little or no decision authority; senior management routinely take on complex jobs with complete decision authority. Thus, an appropriate organization design features a constructive mapping of individual skills with task assignments and a correct allotment of decision rights (see Hoogeweegen, Teunissen, Vervest & Wagenaar, 1999). In a simple model, Prat (2002) shows how the amount of complementarity among jobs determines the amount of workforce homogeneity. With positive complementarities, the workforce will tend to be comprised of workers with similar capabilities. With negative complementarities, workforce heterogeneity becomes optimal. As in the case of deciding upon a centralized versus decentralized organization design, judgments regarding the correct bundling of tasks into jobs involve a trade-off between the costs and benefits of coordination and control. When the distribution of useful knowledge is broad and coordination costs are high, jobs tend to be complex and encompass significant authority to make important decisions about how to best serve customers. When the distribution of useful knowledge is narrow and coordination costs are modest, jobs tend to be simple and embody minimal decision authority. During the 1990s, the explosion of small business, especially small professional corporations, implies rapid growth in the number of complex jobs with significant decision authority. The sophisticated types of goods and services offered by small professional corporations typically involve

84

MARK HIRSCHEY

levels of complexity and customer-specific knowledge that make large-scale coordination uneconomic, even in the face of impressive recent advances in communications technology, like e-mail. In terms of professional career opportunities, this increasing complexity places a high premium on motivated individuals capable of independent self-directed cognitive activity. Many larger organizations facing the need to successfully deal with increasingly complex tasks have turned to team concepts. Here the word team refers to worker groups given shared responsibility for making and distributing products, managing activity, or providing recommendations. Production teams are often given complete responsibility for managing product flow, setting work schedules, maintaining quality control, and so on. In many instances, production teams are an effective means of empowering and motivating front-line employees. The team concept has also proven successful in helping firms bring together input from a variety of functional areas. Teams are especially adept at sharing local or customer-specific information that is sufficiently complex or far reaching so as to be beyond the grasp of any single individual (see Prather & Middleton, 2002). Teams are less effective in dealing with problems where the process of consensus building is unduly slow, or where monitoring and motivating underperforming team members is prohibitively expensive. Because communication costs and shirking problems rise with team size, most effective teams involve just a handful of participants, often less than a dozen or so members. In all cases, an optimal team size is defined by the trade-offs involved between the marginal costs and marginal benefits of adding additional team members.

4.5. Decision Management and Control Decision management is the vital process of generating, choosing, and implementing management decisions. Decision control is the essential process of assessing how well the decision management process functions. As illustrated in Fig. 2, it is useful to characterize the decision management and control process as consisting of four distinct parts. To be successful, this process must effectively generate attractive decision proposals, choose the best decision, implement the best decision, and assess decision success. A recurring problem in decision management and control is that it is often difficult to generate a large number of attractive choice alternatives. Managers and other employees from a given work environment tend to have similar perspectives based upon a common work experience or educational background. This makes “groupthink” and the inability to take a fresh look at problems from new perspectives a common problem. To avoid this, successful management teams often look for input

Organization Structure and Corporate Governance

85

Fig. 2. The Decision Management and Control Process.

from different functional areas within the firm or from individuals who base their proposals on unconventional underlying assumptions. If the decision process is to represent a legitimate search among desirable alternatives, choices must represent valid options with the reasonable potential to offer an attractive trade-off between marginal costs and benefits. If the best decision alternative is to be chosen, the decision process must settle on the option that offers the best marginal net benefit after all direct and indirect costs and benefits to the firm have been considered. Finally, constructive management demands an ongoing assessment of the decision-making process and the success, or lack thereof, of past decisions. It is important to remember that correct investment and operating decisions are made on the basis of economic expectations, or a reasonable before-the-fact forecast of monetary implications. Moreover, judging the wisdom of past decisions involves much more than a simple after-the-fact analysis of economic realizations, or financial outcomes. For example, a decision to buy fire insurance is sound if the chance for material loss is present and insurance is obtained in a cost-effective manner. This is true whether or not a fire loss is actually incurred. Similarly, a given managerial decision is sound if expected profits yield a reasonable risk-adjusted return. Of course, if after-the-fact realizations consistently exceed or fall short of before-the-fact expectations, it becomes necessary to seek out and correct the biases or errors responsible for decision failures.

86

MARK HIRSCHEY

5. INCENTIVE COMPENSATION 5.1. Individual Pay for Performance Incentive compensation has the potential to act as a powerful catalyst for a convergence of interests and effort between employees, top management and the owners of a corporation. As noted by Nagar (2002), top management and the board of directors face two key challenges in the optimal design of a corporation. First, top management must decide how much authority to delegate to employees and lower-level managers. Then, top management must design an appropriate incentive compensation plan to ensure that employees and lower-level managers do not misuse their discretion. In addition, to maximize effectiveness, such incentive compensation plans must be both designed correctly and communicated effectively.9 In a recent study, Mulvey, LeBlanc and Henema (2002) find that worker’s detailed knowledge of the pay process can beat out the amount of pay as an important determinant of employee retention and organizational effectiveness. Surveying more than 6,000 managers and employees in 26 major U.S. and Canadian organizations, Mulvey, LeBlanc and Henema (2002) found close connections between employee knowledge of the organization pay structure and the employee’s ability to align their own behaviors and goals to those of the organization. When employees have a detailed understanding of the organization’s pay structure, Mulvey, LeBlanc and Henema (2002) found that employees tend to feel a greater sense of company loyalty and tend to make a more effective contribution to organizational effectiveness. Many compensation plans fail to achieve desired ends because they focus on input-oriented measures rather than output-oriented criteria.10 In the regulation of public utilities, where utilities are typically promised a “fair” rate of return on necessary investment, charges of excessive investment and “gold-plated” service are often heard. Similarly, Department of Defense procurement policies that promise government contractors a price equal to cost plus a fixed profit percentage have led to significant cost overruns. In these and other cases, the obvious challenge is to base compensation on measures of productivity that are output oriented. Pay for performance simply means to adjust the level of compensation according to measurable indicators of worker productivity (Shaw, Gupta & Delery, 2002). When each worker is paid according to his or her marginal revenue product, the level of compensation is tied directly to the amount of economic value derived from employment. The economic value derived from employment is said to originate from two distinct sources. Productivity that stems from the worker’s natural dexterity, intelligence, or general education is derived from the

Organization Structure and Corporate Governance

87

worker’s general human capital, where human capital is the capitalized value of future productivity derived from worker skills. General human capital can be applied in a wide variety of work settings. Therefore, workers need not depend upon the availability of any specific job opportunity in order to earn a return on general human capital. General human capital can be contrasted with specific human capital, which is any special aptitude, education, or skill that gives rise to added productivity in a unique work setting. To realize the full benefits of specific human capital, workers must have access to specific job opportunities where such skills can be best applied (see Hanushek, Heckman & Derek, 2002). Workers that seek jobs outside the firm are able to command a general labor market wage that offers a fair risk-adjusted rate of return on general human capital. Because the marginal revenue product derived from employment fairly reflects the beneficial effects of general human capital, the outside labor market will offer workers job opportunities that provide compensation that fairly reflects general human capital. The wage rate that could be obtained in any worker’s next-best employment opportunity, sometimes called the reservation wage rate, is the opportunity cost of continuing to work for the worker’s current employer. To attract valued employees, the wage rate inside the firm must at least equal the outside labor market opportunity cost, or reservation wage. To retain valued employees, the wage rate inside the firm must at least equal the reservation wage plus some percentage of the value derived from firm-specific human capital. Thus, firms can justify paying more than the price any competing employer would pay, so long as the wage premium offered to retain valued employees is a fair reflection of the value generated by such employees.11 At the other end of the wage scale, remember that the typical CEO of a giant U.S. corporation has been with the company 20 to 25 years (see Shen & Cannella, 2002). It is quite rare to find a successful top executive recruited from outside the firm because successful CEOs typically require a plethora of detailed knowledge about the firm, top employees, competitors, industry regulation, and so on. Louis Gerstner, former CEO of IBM was recruited from IBM customer American Express and is a prominent exception to this rule. Much more typical is the case of Coca-Cola’s former CEO Roberto Goizueta, who spent an entire business career of 45 years learning and exploiting the nuances of the soda business. It is therefore unsurprising that top executives of major U.S. corporations earn total compensation far in excess of what they might reasonably earn in some alternative employment. The typical top executive generates far more value for his or her current employer than in any other outside employment opportunity. Similarly, senior managers and other long-term employees are often able to use a wealth of firm-specific knowledge to the great net benefit of their employers, even after giving full consideration to their above-average compensation.

88

MARK HIRSCHEY

5.2. Divisional Pay for Performance Organizations are often divided into subunits, or divisions, with independent decision-making authority, operating budgets, and performance evaluations. As in the case of deciding to empower employees with a team-based operating strategy, a divisional approach stems from the perception that transaction costs can be minimized when a group-based method of production or distribution is adopted. For maximum effectiveness, it is essential that divisional performance evaluation and reward systems be fashioned in a manner that is consistent with the decision authority granted. This is made difficult by the fact that outside monitoring of divisional performance suffers from many of the same informational disadvantages common to all “outsider” evaluations of group performance, plus some unique problems tied to divisional performance evaluations. In an interesting theoretical study, Anctil and Dutta (1999) show that managerial compensation contracts can generally be described as linear functions of divisional and firm-wide profits. If managers are compensated solely on the basis of divisional profits, they will have an incentive to invest less than the optimal amount. While compensation contracts based on firm-wide profits alone can induce first-best investments, they have a tendency to impose extra risk on risk-averse managers. Therefore, Anctil and Dutta (1999) find that optimal linear compensation contracts contain both divisional and firm-wide components. Their analysis also identifies a feature of negotiated transfer-pricing, namely interdivisional risk sharing, and characterizes its impact on the design of optimal contracts. Performance evaluation and reward setting is relatively easy when a single individual has complete decision authority for completing a given task. After controlling for the influence of extraneous factors, if any, assessments of individual productivity can be tied directly to output-oriented performance criteria. On the other hand, assessments of the productivity of an entire firm are made on a daily basis in the stock market. For publicly-traded firms, a rise in stock price is consistent with an increase in the present value of future profits; a fall in stock price is compatible with a decline in prospects for future earnings. Because all stockholders share in the value created by a rise in stock prices, it is easy to see why many publicly-traded firms tie an enormous share of top executive and employee compensation to stock-price appreciation. However, although assessments can be quite straightforward in the case of individual and firm-wide performance evaluations, this is seldom the case with divisional performance evaluation. As in the case of any evaluation of group-based performance, monitoring costs are typically much higher in the case of divisional versus individual performance evaluation. Divisional performance evaluation suffers from the additional problem that there are few widely accepted benchmarks for

Organization Structure and Corporate Governance

89

judging divisional performance. There are no divisional stock-price performance measures. Keating (1997) offers evidence on factors affecting firm use of three types of performance metrics commonly used to evaluate division managers: division accounting metrics, firm accounting metrics and firm stock price. Keating’s (1997) survey data reveal that division accounting metric use increases with the degree of correlation observed between the divisional and industry price-earnings ratios. Division accounting metric use decreases with divisional growth opportunities. Firm accounting metric use increases with the divisional manager’s “span of control,” as measured by the manager’s ability to impact the performance of other divisions. Firm accounting metric use decreases with growth opportunities. Divisional manager performance is more often measured in terms of the resulting impact on share prices for larger divisions of high-beta firms that exhibit a strong correlation between firm stock returns and market-wide returns. The great variety of division types can also lead to performance evaluation difficulty. Cost centers, given responsibility for the cost-effective production of goods or services, are typically judged according to their ability to minimize costs for a target level of output or proficiency in maximizing output for a given budget. Like cost centers, revenue centers have only partial responsibility for generating a profit. Marketing efforts organized into revenue centers are frequently given responsibility for maximizing revenues given a certain price or quantity of output. These efforts are consistent with overall profit maximization, provided correct pricing and output decisions are made at the firm level. Although cost centers are an effective means for organizing manufacturing units, other non-revenue-producing services provided within the firm are typically organized as expense centers. Accounting, capital budgeting, human resources, investor relations, and legal services are all typical examples of expense centers. A proven way for controlling costs in expense centers is to adopt an internal market within the firm for expensecenter services. In this approach, operating units are free to decide whether or not they wish to “purchase” services from the firm’s own expense centers. Operating units can decide whether or not to employ internal legal services, for example, and will do so when the price-performance trade-off for internal legal services is more attractive than for services provided by outside law firms. Thus, under a “chargeback” system, the level of internally provided services is optimal when internal supply and demand conditions are in exact balance. Finally, some divisions are set up as independent profit centers. As in the case of investment project evaluation, profit centers must be careful to adjust accounting cost figures to allow for a clear assessment of incremental costs and benefits. Profit centers must also be cautious to properly account for goods and services transferred among divisions, and to allow for any interdependence between divisional performance measures.

90

MARK HIRSCHEY

Fig. 3. The Goal of Incentive Compensation Is to Match Worker Incentives with Managerial Motives.

Figure 3 shows the necessary link between effective individual and divisional pay for performance plans. In all instances, a suitable divisional pay for performance plan sets bonus payments based upon the relative productivity of individual employees or groups of employees as measured by their contribution to the amount of EVA generated by the division. It is also important that the share of divisional EVA paid out in the form of bonuses to divisional employees, as opposed to the amount of divisional EVA retained by the corporation, represent a valid risk-sharing arrangement with stockholders. If 100% of divisional EVA were paid out in the form of divisional bonuses, stockholders would not benefit from superior performance and receive no compensation for taking on the risk associated with poor relative performance. On the other hand, if none of divisional EVA is paid out in the form of divisional bonuses, division employees would have little incentive for superior performance. For any divisional pay for performance plan to be both fair and effective, the benefits of superior performance must be shared among the corporation’s stockholders and divisional employees in a manner that adequately compensates both stockholders’ risk-taking activity and exceptional worker productivity.12

6. CORPORATE GOVERNANCE 6.1. Role Played by Boards of Directors Corporate governance is the system of controls that helps the corporation effectively manage, administer, and direct economic resources. If any corporation fails

Organization Structure and Corporate Governance

91

to effectively command its economic resources, this corporate failure can often be blamed on a similar failure of its corporate governance mechanisms. The most important and closely monitored corporate governance mechanism is the company board of directors. The board of directors is a group of people legally charged with the responsibility for governing a corporation. In a for-profit corporation, it is generally accepted that the board of directors is responsible to stockholders. Some adopt the broader perspective that the board is responsible to “corporate stakeholders”; that is, to everyone who is interested and/or can be affected by the corporation. Corporate stakeholders include stockholders, customers, employees, the community at large, and so on. In a non-profit corporation, the board generally reports to stakeholders, particularly the local communities in which the non-profit serves. Boards of directors provide continuity for the organization by maintaining a legal existence. A primary responsibility is to select, appoint, and review the performance of a chief executive officer responsible for day-to-day decision making and administration of the organization. Table 1 shows five important attributes of effective corporate boards of directors and corresponding limitations of ineffectual corporate boards. The National Center for Non-profit Boards, in their booklet Ten Basic Responsibilities of Non-profit Boards, itemize 10 responsibilities for non-profit boards: determine the organization’s mission and purpose, select the executive, support the executive and review his or her performance, ensure effective organizational planning, ensure adequate resources, manage resources effectively, determine and monitor the organization’s programs and services, enhance the organization’s public image, serve as a court of appeal, assess its own performance. Though intended as a guide to non-profit institutions, these responsibilities often apply to the boards of directors for profit-seeking corporations as well.

6.2. Corporate Governance Mechanisms Inside the Firm Corporate control mechanisms inside the firm are useful means for helping ameliorate the potential divergence of interests between managers and stockholders. Organization design, including the degree of vertical integration and the horizontal scope of the corporation, are examples of essential corporate governance mechanisms inside the firm. When inputs can be reliably obtained from suppliers operating in perfectly competitive markets, it is seldom attractive to produce such components in-house. Simple market procurement tends to work better. Similarly, when important economies of scale in production are operative, it is preferable to obtain inputs from large specialized suppliers. A high degree of vertical integration only makes sense when input production is within the firm’s core

92

MARK HIRSCHEY

Table 1. Attributes of Effective and Ineffectual Boards of Directors. Attribute

Effective Board

Ineffectual Board

Integrity

Board members must take pains to ensure that words and actions are in the best long-term interests of large and small shareholders. Training and business experience of board members must be up to the task of providing value-added oversight to managerial decisions. Through words and deeds, it must be clear that the board of directors effectively represents shareholder interests in its oversight of managerial decisions. Clear lines of authority and responsibility must be drawn. Both the board and top management must be held accountable for corporate performance. Actions must be carefully and completely disclosed in timely shareholder reports and SEC filings.

Board decisions and management compensation plans are sometimes structured to favor entrenched management at the expense of stockholders. Board members with little or no relevant training or business experience are too often encountered. Celebrities are great at cocktail parties, but make poor board members.

Competence

Independence

Accountability

Transparency

Cronies of top management, or cozy consulting arrangements between companies and board members, can compromise if not undermine board member independence. Some boards are too large by design. When the board of directors has more than a dozen members, it can quickly become impossible to wield effective decision making power. Too many boards of directors allow management to hide subpar corporate performance through misleading corporate communications or endless restructuring.

competency and supply is erratic or suppliers charge excessive markups. In such instances, vertical integration can result in better coordination of the production process and thereby protect the firm’s tangible and intangible investments. Vertical integration is sometimes seen as a useful means for deterring competitor entry into a company’s primary market. Years ago, IBM made a huge strategic error in licensing Intel Corp. to manufacture key components for the personal computer. Intel started out as simply a microprocessor manufacturer supplying the “brains” to manufacturers of personal computers and other “smart” electronics. Today, Intel dominates that business with a market share approaching 90%, and enjoys sky-high margins and an enviable rate of return on investment. To spur future growth, Intel is now branching out into the production of other PC components like modems, networking equipment, and so on. Soon, the famous trademark “Intel

Organization Structure and Corporate Governance

93

Inside” may have to be replaced with “Intel Inside and Outside.” Meanwhile, IBM earns only anemic returns in the PC business. All manufacturers would do well to contemplate IBM’s experience with Intel before licensing to others the production of key components. Another useful means for controlling the flow of corporate resources is provided by internal markets established among divisions to better balance the supply and demand conditions for divisional goods and services (see Kachelmeier & Towry, 2002). So, too, is incentive compensation perhaps the most obvious corporate governance mechanism inside the firm. In many circumstances, the proper design and implementation of a appropriate incentive pay plan is the most fundamental determinant of whether or not corporate resources will be administered effectively and equitably. Like any effective corporate governance mechanism inside the firm, such arrangements must further the objective of minimizing transaction costs by effectively joining decision authority with the system of performance evaluation and rewards.

6.3. Franchise Agreements A wide variety of monitoring mechanisms outside the firm work together with mechanisms inside the firm to establish an optimal set of restrictions on firm activity. Interestingly, firms themselves often suggest such restrictions. Commercial bank loan covenants, financial audits by independent auditors, and performance scrutiny by independent security analysts are all common examples of outside monitoring mechanisms agreed to by firms. Franchise agreements are a prime example of voluntary contractual arrangements outside the firm that can be viewed as corporate governance mechanisms. Franchise agreements give local companies the limited right to offer goods or services developed or advertised on a national basis. Franchise agreements are especially popular in instances where personalized customer service is crucial to success and when the performance of local managers is hard to measure over short periods of time (see Boyle, Dwyer, Robicheaux & Simpson, 1992). For example, McDonald’s Corp. operates an extensive franchise system of fastfood restaurants throughout the world. Local franchise owners make a credible commitment to the company by paying for the construction of local restaurants and undergoing extensive training at “Hamburger University,” as McDonald’s likes to call its Oak Brook, Illinois, training facility. In turn, McDonald’s makes a credible commitment by awarding local franchisees exclusive rights to market McDonald’s food in a given trade area. In this way, McDonald’s can be assured that local outlets will be run effectively, and local franchisees have the assurance that McDonald’s

94

MARK HIRSCHEY

will continue its aggressive advertising and franchise development programs. Both parties hold viable threats over the other. The valuable right to sell McDonald’s food products can be taken away from local franchisees if food quality or store cleanliness falters. Local managers could choose to withhold their support for corporate marketing and pricing policies if they deem that insufficient support has been provided for local markets. In addition to the fast-food business, franchise agreements are common in the automotive repair business. In automotive repair, performance is hard to measure because shoddy repair service shows up only over long periods of time. The quality assurance offered to repair service customers improves when local managers of auto service outlets have an owner’s incentive to stand behind work quality. Similarly, franchise agreements give local managers and owners incentive to develop the economic potential of local markets. Like the local customers of gas and service stations, car buyers benefit from a network of new car dealers who promote and stand behind the quality of products produced by Ford, DaimlerChrysler, GM and other major automobile manufacturers. Without an owner’s incentive to build and maintain customer loyalty, the new car business would suffer from many of the same image problems that have dogged the used-car business for decades.

6.4. Strategic Alliances Strategic alliances are formal operating agreements between independent companies that can also be viewed as corporate governance mechanisms (see Koka & Prescott, 2002). Killen, Hunt, Ayres and Janssen (2002) point out that these combinations are increasingly used to improve foreign marketing. In an analysis of 476 strategic alliances, Sengupta and Perry (1997) find that equity joint ventures are preferred to contracts when cultural differences between partner firms are greater, and when alliances involve upstream rather than downstream value chain activities. On the other hand, contracts tend to be preferred to equity joint ventures in cross-industry alliances and when the technological intensity of the industry sector of cooperation is higher. Cereal Partners Worldwide, a strategic alliance between General Mills, Inc., and Nestl`e, is used to market breakfast cereal products. Snack Food Ventures Europe, a partnership between General Mills and Pepsico, markets snack foods in Belgium, France, Holland, Spain, Portugal, and Greece. Another proposed alliance between British Airways and American Airlines was abandoned after serious antitrust concerns were raised in Britain and the United States because such an arrangement has the potential to allow the companies to squeeze out competition and dominate the lucrative U.S.-London business and tourist travel market.

Organization Structure and Corporate Governance

95

Cisco Systems, Inc. is one of the most aggressive proponents of corporate strategic alliances. At Cisco, strategic alliances are designed to help deliver a customer-centric, total solutions approach to solving problems. Cisco and its partners have found strategic alliances to be an effective means for exploiting business opportunities and creating sustainable competitive advantages for its customers. For example, IBM Global Services and Cisco Systems have a strategic alliance to drive development of the global Internet economy by combining their mutual strengths in the rapid deployment of networking applications and joint creation and delivery of end-to-end e-business solutions. The IBM-Cisco strategic alliance is helping businesses successfully migrate to an Internet infrastructure and thereby enabling higher levels of customer satisfaction. In another alliance, Hewlett-Packard & Cisco are collaborating to deliver end-to-end, network-enabled solutions that will allow new and joint customers to optimize and reduce the complexity of their networks. Strategic alliances also arise when participating companies enjoy complementary capabilities. For example, Oracle Corp. develops, manufactures, and distributes computer software that helps corporations manage and grow their businesses. Oracle software products can be categorized into two broad areas: systems software and Internet business applications software. Systems software is used to deploy applications on the Internet and corporate intranets. It includes database management software that allows users to create, retrieve, and modify the various types of data stored in a computer system. Internet business applications software allows users to access information or use applications through a simple Internet browser. In a collaborative alliance, Oracle and Cisco, a networking equipment supplier, have agreed to develop network-enhanced database technology and enterprise applications. Oracle and Cisco plan to deliver flexible and secure solutions to maximize customer advantage and satisfaction. Another longtime proponent of strategic alliances is TRW Inc., a global manufacturing and services company focused on supplying advanced technology. The Cleveland, Ohio-based company uses strategic alliances to acquire technology, create synergy, extend its own manufacturing capabilities, enter new markets, and build existing customer relationships. TRW strategic alliances produce automotive occupant safety systems, chassis systems, electronics, and engine components for spacecraft and space communications, defense systems, telecommunications products, information technology, public safety systems, and other complex integrated systems. Whipple and Frankel (2000) suggest that the growing popularity of strategic alliances follows recognition that competition is shifting from a “firm versus firm perspective” to a “supply chain versus supply chain perspective.” Corporations are increasingly seeking competitive advantage through cooperative supply chain

96

MARK HIRSCHEY

arrangements because such strategic alliances allow companies to quickly combine individual strengths and unique resources. Stuart (1998) points out that strategic alliances are an especially attractive means for collaboration among firms that enjoy significant reputation capital. 6.5. Mandatory Corporate Governance Mechanisms Outside the Firm Other outside monitoring mechanisms can be mandatory. Such compulsory mechanisms include the wide variety of federal, state, and local laws and regulations that govern corporate behavior. Indeed, the potential exploitation of stockholders, bondholders, and other residual claimants by opportunistic decision agents is often reflected in arguments leading to the establishment of broad regulatory initiatives, such as those stemming from establishment of the SEC. Hirschey and Jones (2001) study how calculated or inadvertent violations of federal laws have the potential to impose significant costs on shareholders and other residual claimants. The pursuit of illegal short-term strategies can represent a form of self-dealing by managers who seek to reap short-term personal gain while escaping detection. Actual or suspected violations of federal laws have the potential to result in significant costs measured in terms of investigation expenditures, litigation expenses, fines and seizures, and lost reputational capital for the firm – all of which can measurably reduce future cash flows and current market values. Within this context, federal laws can be seen as part of the institutional framework that contributes to the range of control mechanisms that originate inside and outside the firm to comprise an effective system of corporate governance. Because short-term “hit and run” managers may possess incentives to “cut” legal and ethical corners, the design and administration of federal laws can be seen as a means of outside monitoring designed to ensure a coincidence of managerial incentives, stockholder interests, and broader social objectives.

7. OWNERSHIP STRUCTURE AS A CORPORATE GOVERNANCE MECHANISM 7.1. Ownership Structure and Performance The field of finance has long placed great emphasis on the firm’s debt versus equity financing decision (see Jensen, 1986; Myers & Majluf, 1984). More recently, economists in general have become interested in the corporate governance implications of the ownership structure of the firm, or the complex array of divergent claims on the value of the firm.

Organization Structure and Corporate Governance

97

In financial economics, the capital structure of the firm has been traditionally described in terms of the share of total financing obtained from equity investors versus lenders (debt). Today, interest has shifted from capital structure to ownership structure, as measured along a number of important dimensions, including inside equity, institutional equity, widely dispersed outside equity, bank debt, and widely dispersed outside debt. Among these, the percentage of inside equity financing receives the most attention. Inside equity is the share of stock closely held by the firm’s CEO, other corporate insiders including top managers, and members of the board of directors. Employees are another important source of inside equity financing, perhaps as part of an employee stock ownership plan, or ESOP. The balance of equity financing is obtained from large single-party outside shareholders, mutual funds, insurance companies, pension funds, and the general public. When the share of insider holdings is “large,” a similarly substantial selfinterest in the ongoing performance of the firm can be presumed. Managers with a significant ownership interest have an obvious incentive to run the firm in a value-maximizing manner. Similarly, when ownership is concentrated among a small group of large and vocal institutional shareholders, called institutional equity, managers often have strong incentives to maximize corporate performance. On the other hand, when the amount of closely-held stock held is “small,” and equity ownership is instead dispersed among a large number of small individual investors, that top management can sometimes become insulated from the threat of stockholder sanctions following poor operating performance. Empirical research has long documented the fact that the largest corporations tend to be run by managers with relatively little direct ownership interest (Larner, 1970). Ownership structure is conventionally described by the share of stock that is closely held by the firm’s chief executive officer (CEO), other corporate insiders including top managers and members of the board of directors, employees (perhaps as part of an employee stock ownership plan, or ESOP), large single-party outside shareholders, institutions, and the general public. When the share of closely-held stock held is “large,” a similarly substantial self-interest in the on-going performance of the firm can be presumed. As such, managers with a substantial share interest in the firm have an important incentive to run the firm in a value-maximizing manner. Similarly, when managers enjoy contractual incentives to engage in value-maximizing activity, a convergence in managerial and stockholder interests can be presumed. To this point, little emphasis has been placed upon the well-established world-wide trend toward replacement of small atomistic shareholders by large institutional investors. However, there are reasons to believe that large institutional holdings have the potential to affect a meaningful change in managerial incentives. Substantial ownership provides institutional investors with correspondingly

98

MARK HIRSCHEY

material incentives for the discovery of costly information. As such, the probability that outside investors will discover evidence of managerial inefficiency or malfeasance is increased when institutional ownership is substantial. For example, many institutional investors are forced to liquidate their holdings in the event of dividend omissions or bankruptcy filings. As a result, institutional investors are especially sensitive to such possibilities. Fiduciary responsibility also forces many institutional investors to tender their shares in the event of an above-market tender offer or takeover bid. At the same time, when institutional share ownership is high, the costs of proxy solicitations are reduced. Managers of firms with high institutional ownership are, therefore, relatively more susceptible to replacement following an unfriendly takeover bid. Therefore, fiduciary responsibility and the dynamics of ownership concentration have the potential to make institutional stockholders especially effective in the managerial monitoring process. The level of independent monitoring by outside shareholders depends upon the costs and benefits such activity entail. All else equal, higher relative share ownership by institutional stockholders leads to greater monitoring incentives and monitoring activity. Similarly, when a relatively large share of stock is held by the general public, relatively lower incentives for managerial monitoring can exist. Small investors have neither the incentive nor the resources to expend the substantial effort necessary for effective managerial monitoring. Thus, when the number of small outside stockholders is large, a reduction in the scope of effective managerial monitoring can be anticipated. A number of studies support the basic propositions of the agency cost literature with respect to the advantages tied to high ownership concentration. In an especially influential study, Morck, Shleifer and Vishny (1988) explore the relationship between management ownership and the market valuation of the firm. Piecewise linear regressions are estimated, allowing for changes in the slope coefficient on board ownership. Evidence is found of a significant non-monotonic relationship. The market value of the firm, takes as an estimate of the discounted net present value of future profits, first increases, then decreases, and then increases slightly as ownership by the board of directors rises. In a related paper, McConnell and Servaes (1990) find a significant curvilinear relation between the market value of the firm and the fraction of common stock owned by corporate insiders. The curve slopes upward until insider ownership reaches approximately 40% to 50% and then slopes slightly downward. A significant positive relation is also found between the market value of the firm and the fraction of shares owned by institutional investors. McConnell and Servaes (1990) results are consistent with the hypothesis that corporate value is a function of the structure of equity ownership. McConnell and Servaes (1995) investigate the relation between corporate value, leverage, and equity ownership. For high-growth firms, McConnell and Servaes

Organization Structure and Corporate Governance

99

(1995) find that corporate value is negatively correlated with leverage. For lowgrowth firms corporate value is positively correlated with leverage. McConnell and Servaes (1995) results also hint that the allocation of equity ownership among insiders, institutions, blockholders, and atomistic outside shareholders is of marginally greater significance in low-growth than in high-growth firms. The overall interpretation of the results is that debt policy and equity ownership structure matter. More recently, a positive relationship between ownership concentration and firm performance has been extensively documented in both the United States and abroad. For example, Gedajlovic and Shapiro (2002) examine the relationship between ownership structure and financial performance for 334 Japanese corporations over the 1986–1991 period. Even in Japan, where corporate governance mechanisms are often thwarted by convention and government fiat, a positive relationship exists between ownership concentration and financial performance as predicted by the agency theory literature. Similarly Morck, Nakamura and Shivdasani (2000) report that the value of Japanese firms tends to rise with increased managerial ownership and a greater importance for large blockholders.13

7.2. Is Ownership Structure Endogenous? Demsetz and Lehn (1985) consider four general forces affecting corporate ownership structure, including the: (1) value-maximizing size of the firm; (2) profit potential from exercising more effective control; (3) amenity potential of firm ownership; and (4) systematic regulation. Among these forces, the influences of value-maximizing firm size and the profit potential from exercising more effective control are more likely to broadly shape the ownership structure of the modern corporation. Amenity potential and systematic regulation are prone to have more narrow influences. When Demsetz and Lehn (1985) refer to the amenity potential of a firm’s output, they point to the utility consequences of being able to influence the type of goods produced, not to any general benefits derived from leadership or managerial perquisites. Such benefits can be derived from ownership of mass media and professional sports teams, for example, and explain why such endeavors tend to be tightly controlled by top management. On the other hand, the diffuse ownership structure of regulated utilities can be explained by extensive rate-of-return regulation that limits the capacity of managers to influence firm performance. As a result, public utility regulation has the potential to supplant the need for extreme ownership concentration and close oversight by shareholders. As argued by Demsetz and Lehn (1985), the value-maximizing size of firms in product and input markets tends to vary within and among industries. In the case of

100

MARK HIRSCHEY

firms that produce easily identifiable products subject to economies of scale in production, a relatively diffuse ownership structure can be consistent with shareholder wealth maximization. In the case of firms that produce goods and services with the potential for high quality variation, a more concentrated ownership structure may be required to give shareholders the amount of control necessary to mollify other suppliers and customers. The control potential of the firm is the wealth gain achievable through more effective monitoring of managerial performance. Demsetz and Lehn (1985) argue that control potential is directly related to the “noisiness” of the operating environment. For example, heavy advertisers or high-tech firms that depend on hard-to-monitor advertising or R&D activity for their success may require more concentrated ownership structure than would be true of low-tech firms that produce easily monitored goods and services. When the quality of output is hard to measure, and production involves inputs with little collateral value, a high degree of ownership concentration gives outside investors, suppliers and customers the collateral value needed for quality assurance. Large brand name advertisers like Coca-Cola, for example, may feature highly concentrated ownership because outside investors, suppliers and customers would be otherwise reluctant to deal with the firm given their informational disadvantage vis-`a-vis management. Similarly, outside investors, suppliers and customers may demand a high degree of ownership concentration among high-tech firms like Microsoft, given the high degree of outsider uncertainty regarding the quality of hard-to-monitor R&D programs. In essence, Demsetz and Lehn (1985) assert that ownership structure is endogenously determined by economic forces. Depending on the firm’s operating environment, either a diffuse or concentrated ownership structure might be entirely appropriate. This contrasts with the more traditional approach of Berle and Means (1932) who suggest that diffuse ownership implies a harmful diminution in shareholder control. To test these conflicting hypotheses, it is necessary to first consider the extent to which corporate ownership structure is in fact endogenously determined. The Demsetz and Lehn (1985) hypothesis is supported to the extent that ownership structure appears to vary systematically across firm-size classes and among industries in ways that are consistent with value maximization. In that event, it may be relevant to argue that observed ownership structure differences are an optimal reflection of economic forces. Clearly, the probability that outside investors will discover evidence of managerial inefficiency or malfeasance is increased when institutional ownership is substantial. Many institutional investors are forced to liquidate their holdings in the event of dividend omissions or bankruptcy filings. As a result, institutional investors are especially sensitive to such possibilities. Fiduciary responsibility also forces many institutional investors to tender their shares in the event of an

Organization Structure and Corporate Governance

101

above-market tender offer or takeover bid. At the same time, when institutional share ownership is high, the costs of proxy solicitations are reduced. Thus, managers of firms with high institutional ownership are relatively more susceptible to unfriendly takeover bids. Fiduciary responsibility and the dynamics of ownership concentration have the potential to make institutional stockholders especially effective in the managerial monitoring process. Insider and institutional stock ownership represent alternative forms of ownership concentration that combine to form an effective method for monitoring managerial decisions. Remember, a relatively high concentration of insider plus institutional ownership is descriptive of the modern corporation. Chauvin and Hirschey (1996) provide theory and evidence in support of the hypothesis that closely-held ownership, institutional holdings and the degree of financial leverage are jointly and endogenously determined. An inverse relation between the percentage of closely-held shares and institutional holdings is clearly apparent, as is a compatible inverse link between institutional holdings and leverage. A somewhat more complex link between closely-held ownership and leverage is suggested. High financial leverage generally reduces the percentage of closely-held shares, but already high closely-held ownership can work to increase the optimal degree of financial leverage. In a related paper, Pagano and Roell (1998) argue that the choice of stock ownership structure depends upon a tradeoff between agency costs and monitoring expenses. From the viewpoint of a company’s controlling shareholder, the optimal ownership structure generally involves some measure of dispersion, to avoid excessive monitoring by other shareholders. The optimal dispersion of share ownership can be achieved by going public, but this choice also entails some costs in terms of listing costs and costs due to loss of control over the shareholder register. If the controlling shareholder sells shares privately instead of going public, some of the costs of going public can be avoided, but private external shareholders may tend to monitor the controlling shareholder too closely. Thus, the optimal ownership structure entails a trade-off between the cost of providing a liquid market for minority shareholders and the monitoring costs they engender. The relationship between monitoring costs and ownership structure is also investigated by Denis, Denis and Sarin (1997) who report that ownership structure significantly affects the likelihood of a change in the top executive. Controlling for stock price performance, Denis, Denis and Sarin (1997) find that the probability of top executive turnover is negatively related to the ownership stake of officers and directors and positively related to the presence of an outside blockholder. In addition, the likelihood of a change in the top executive is significantly less sensitive to stock price performance in firms with higher managerial ownership. Finally, an unusually high rate of corporate control activity in the twelve months preceding

102

MARK HIRSCHEY

top executive turnover is documented. It is concluded that ownership structure has an important influence on internal monitoring efforts and that this influence stems in part from the effect of ownership structure on external control threats. Mayer (2002) argues that a distinguishing characteristic of the financing of high technology firms is their evolving pattern of control by different investor groups. While stock markets are an important component of the development of the most successful firms, they are not the most common. Regulation is another significant influence on institutional structure. The degree of risk taking by financial institutions and the diversity of their investments are affected by trade-offs between competition and stability and the emphasis placed on minority investor protection. In a similar vein, Lehn (2002) argues that deregulation of the telecommunications industry is likely to result in significant changes in the governance structures of telecommunications firms. In particular, with deregulation, the ownership structures of telecoms are expected to become more concentrated, the level of executive compensation is expected to increase, executive pay is expected to become more sensitive to performance, and board sizes are expected to become smaller. Evidence from the airline industry shows that these governance changes occurred after this industry was deregulated in the United States during 1978. Preliminary evidence on telecommunications firms indicates that similar changes are beginning to occur in the governance structures of telecom firms. In short, ownership structure appears to vary systematically across firm size classes, among industries, and across regulatory environments in ways that are consistent with value maximization. It thus appears relevant to argue that observed differences in corporate ownership structure are an optimal reflection of economic forces.

8. THE RESEARCH AGENDA 8.1. Accounting and Control To this point, I have described the evolution of a vibrant and growing literature on organization structure and corporate governance. It is now worth considering what may be the most exciting venues for future developments. Among the most obvious fruitful directions for further research is investigation of the use and or abuse of the accounting standards setting process. Abdel-khalik (2002) argues that one potentially positive outcome of Enron’s demise could be the resulting improvement in the process by which auditors are selected, retained and compensated. Abdel-khalik (2002) argues that one way of improving audit independence might be to allow shareholders to directly decide on auditor choice

Organization Structure and Corporate Governance

103

and compensation. Abdel-khalik (2002) calls for establishing a Shareholders’ Board of Trustees (SBT) that is independent from the board of directors and vest it with the responsibility for selecting, retaining and compensating external auditors. Yeo, Tan, Ho and Chen (2002) argue that management typically has discretion over the recognition of accruals and this discretion can be used to signal private information or to opportunistically manipulate accounting earnings. To the extent that management uses their discretion to manipulate accruals, earnings will become less informative. Yeo, Tan, Ho and Chen (2002) examine how managerial ownership and external unrelated blockholdings affect the informativeness of earnings. A non-linear relation appears to exist between managerial ownership and the information content of earnings. The information content of earnings tends to increase with managerial ownership at low levels. Consistent with a monitoring role for large blockholders, a strong positive relationship also appears to exist between external unrelated blockholdings and the information content of earning. In a related paper, Fan and Wong (2002) examine the earnings informativeness, measured by the earnings-return relation, and the ownership structure of 977 companies in seven East Asian economies. In Asia, Fan and Wong (2002) find that concentrated ownership creates a reporting conflict between controlling owners and outside investors. In Asia, concentrated ownership is associated with low earnings informativeness as ownership concentration prevents leakage of proprietary information about the firms’ rent-seeking activities. Going forward, financial economists need to carefully consider the agency cost implications of managerial discretion in the processing of accounting information about firm performance. Accounting statements about firm performance have the potential to give investors, consumers, and the general public useful insight concerning firm and managerial performance. From a corporate governance perspective, it becomes important that this information processing function is conducted in an efficient manner. From an academic perspective, agency problems tied to the information processing function of management are well worth considering.

8.2. Special Purpose Entities (SPE) Economic and accounting policy for Special Purpose Entities (SPEs) has risen to prominence as a result of well-publicized abuse by Enron and other corporations (see Hartgraves & Benston, 2002). A primary accounting issue regarding SPEs is whether they should be consolidated into the sponsor’s (or primary beneficiary’s) financial statements or left “off-balance sheet.” Correspondingly, issues have arisen as to whether the sponsor should be able to treat gains and losses resulting from transactions with SPEs as independent, arm’s-length transactions. The problem at

104

MARK HIRSCHEY

Enron, among others, was that even though its SPEs were thinly capitalized and held assets that posed considerable risks, the accounting performance and financial condition of Enron’s SPEs were not consolidated on Enron’s financial statements. In other words, Enron maintained the facade that its dealings with its own SPEs represented arm’s-length transactions despite the fact that Enron assumed virtually all of the risks associated with failure. Benston and Hartgraves (2002), among others, have investigated how Enron used non-consolidated special-purpose entities (SPE) to substantially inflate reported revenue, net income and stockholders’ equity, while significantly understating liabilities. More than just involving reporting errors, Enron’s use of SPEs led to a significant misallocation of corporate resources. In some cases, outright fraud was involved. The abuse of SPEs at Enron and other such companies represents an important failure by company boards of directors, audit committees, outside attorneys, and public auditors. As such, the abuse of SPEs at Enron and other such companies represents an important failure of the corporate governance system. Looking ahead, enlightened accounting policy may require consolidation by business enterprises of many activities now conducted under the guise of SPEs. From an economic standpoint, if a business enterprise has a controlling financial interest in an SPE, the assets, liabilities, and results of SPE activities should be included in the consolidated financial statements of the business enterprise. Under this perspective, most off-balance sheet leasing arrangements and structured financing arrangements should be consolidated by primary beneficiaries. Similarly, transparency and reporting efficiency suggest that off-balance sheet and off-income statement SPEs that largely reflect the assets and business operations of a given business enterprise should be included in consolidated financial statements with those of the business enterprise. At a minimum, current SPE abuses provide a fertile area for investigation of the unethical manipulation of accounting standards.

8.3. Diverse Ownership Claims Amoako-Adu and Smith (2001) analyze changes in the capitalization and control of firms with capital structures that feature securities with differential voting rights, sometimes called “dual class firms.” Results indicate that the combination of a large controlling shareholder with family interests, rather than concentrated ownership per se, leads to dual class capitalization. During the first 15 years post-IPO, voting leverage continuously increases as the dual class firms issue more restricted than superior voting shares. However, control changes are equally frequent for dual and single class firms suggesting that dual class capitalization is not used to unduly entrench management. Disputes between restricted and

Organization Structure and Corporate Governance

105

superior voting shareholders are documented to illustrate potential corporate governance problems which are associated with dual class capitalization. Investor interest in dual class equity has decreased recently, and there is a current trend toward reclassification back into single class equity. Harper and Madura (2002) argue that tracking stock for a unit of a firm allows the market to value and monitor that unit independently of the rest of the firm. Announcement of the creation of tracking stock appears to elicit an abnormal share price response of 2.17% on average over a two-day period. The share price response at the time of the announcement is more favorable: when the voting rights of the tracking stock are based on a market valuation; when the parent company’s debt ratio is relatively low; when the parent’s previous stock performance is relatively poor; and when the parent is not engaging in an acquisition. These results are consistent with reduction of agency problems. At the same time, firms that create tracking stock do not experience higher long-term valuations, suggesting that agency problems are not resolved with the creation of tracking stock. However, unlike a spin-off or equity carve-out, a tracking stock does not create a new legal entity. When tracking stock is issued, the parent firm does not relinquish any claims on the assets of the separate business; rather, the transaction can be viewed as a type of a stock split. In an equity carve-out, a parent firm takes a subsidiary public through an IPO of shares in the subsidiary. A new legal entity is created, although in a majority of cases the parent firm retains a controlling interest in the newly traded subsidiary. Billett and Mauer (2000) note that diversified firms trade at a discount relative to comparable portfolios of stand-alone firms. One explanation is that these firms have inefficient internal capital markets. Billett and Mauer (2000) examine a link between firm value and the value of internal capital markets using a new form of corporate restructuring called tracking stock. A model is presented that illustrates that the announcement effect of a tracking stock equity restructuring conveys information about the market’s assessment of the value of a firm’s internal capital markets. Billett and Mauer (2000) develop a measure of the profitability of the internal capital market, and a strong positive relation is found between it and tracking stock announcement effects. Dual voting rights, non-voting stock, and tracking stock are all examples of diverse ownership claims on future cash flows. More than just considering the valuation effects of issuing debt versus equity, financial economists need to carefully consider the advantages and disadvantages of the recent proliferation in diverse ownership claims. From a positive perspective, it becomes interesting to consider the conditions under which diverse ownership claims are observed. From a normative perspective, it is also interesting to ask if such diverse ownership claims serve a useful economic purpose.

106

MARK HIRSCHEY

NOTES 1. Both positive (what is) and normative (what should be) issues are relevant. Years ago, the make-or-buy decision was seen as a fairly simple, yet vital determinant of the firm’s ongoing success. It was then thought that decisions regarding the organization structure of individual firms would simply revolve around questions concerning optimal production scale. If economics of scale were important, large corporations would evolve to minimize production costs. If economies of scale were slight to non-existent, small and nimble corporations would evolve to exploit niche markets. Today, answering the question of how to achieve an optimal corporate design has become much more complicated. When economies of scope are relevant, it becomes attractive to offer customers bundles of related products and services, despite the fact that the returns to scale characteristic may be quite different for each individual product. 2. Invention of the airplane did not make it important to become a faster walker; it changed the nature of travel. Similarly, invention of the Internet did not just increase the importance of computer literacy in the workplace; it promises to fundamentally change the size, scope, and viability of organizations throughout the economy. 3. The optimal design of the firm is the organization type that most successfully meets customer demands. This design composition of the firm is traditionally referred to as its structure. Organization structure is described by the vertical and horizontal relationships among the firm, its customers and suppliers. A vertical relation is a business connection between companies at different points along the production-distribution chain from raw materials, to finished goods, to delivered products. An example of a vertical relation is the business linkage between Wal-Mart Stores, Inc., the giant discount retailer, and Rubbermaid, the leading manufacturer of rubberized food- and household-storage products. By way of contrast, a horizontal relation is a business affiliation between companies at the same point along the production-distribution chain. The commercial relationship between Wal-Mart Stores, Inc., and Target Corp., the parent company of Target discount stores, Marshall Field’s and Mervyn’s, is an example of a horizontal relation. 4. When I described this project, a good friend and prominent financial economist snickered: “You must be smoking dope instead of reading the newspapers.” In response, I laughed and said: “Most of the roughly 10,000 large publicly-traded companies in the United States have demonstrated effective corporate governance mechanisms. This is despite the spectacular fraud that has emerged at Enron Corp., Tyco International Ltd., WorldCom, Inc., and a handful of other major corporations. In any event, the successes and failures of organization structure and corporate governance are both worthy subjects for academics.” 5. When it was an independent entity, investment bank Salomon, Inc., now part of Citigroup, Inc., established a compensation plan whereby managers were required to take a significant portion of total compensation in the form of common stock that could not be sold until five years after the employee left the bank. Over time, traders managing directors, top management and other employees at Salomon came to own roughly 35% of the company. Most impressively, this significant employee ownership was achieved through direct stock purchases in the open market; no outright grants of employee or executive stock options were involved. At Salomon, a giant boost to employee stock ownership was being accomplished in such a way as to preserve the ownership position of other stockholders. This contrasts with the much more common situation where generous grants of stock options to top management realizes the desired end of aligning managerial and shareholder interests,

Organization Structure and Corporate Governance

107

but only at the significant cost of a material dilution in the ownership interest of outside shareholders. Many credit the resurgence in Salomon’s performance during the mid-1990s to its enlightened management compensation and stock ownership practices. Traveler’s Group, Inc., chairman and CEO Sandy Weill was sufficiently impressed to buy Salomon, Inc., during late 1997. Interestingly, other investment banks, like Lehman Brothers Holdings, Inc., have also come to appreciate the advantages of encouraging top executives to make open market purchases of the firm’s common stock. It is a trend that seems to be spreading. 6. Recent studies document the importance of accounting information by showing market valuation effects tied to accounting announcements that have no direct cash flow implications (see Docking, Hirschey & Jones, 1997; Hirschey & Richardson, 2002). 7. For perspective on the “value relevance” literature, see Hirschey, Richardson and Scholz (2001). 8. Over the last three decades, privatization programs in a variety of countries have radically reduced the role of the state as a major owner of productive assets. This trend can be thought of as support for the argument that recent improvements in telecommunications have dramatically reduced the historical advantages of centralized as opposed to decentralized methods of production. In an interesting study of this trend, Munari (2002) analyzes seven cases of privatization in Italy and France to gauge how R&D activities may be affected by privatization in terms of objectives and organization. Munari (2002) shows that R&D units within privatized companies are subject to profound restructuring designed to boost efficiency and to create value for customers. 9. Years ago, overnight document delivery giant FedEx Corp. faced nagging delays in getting parcels sorted and processed at its main terminal in Memphis, Tennessee. All across the country, FedEx’s red, white, and blue trucks pick up customer documents and parcels for delivery to homes and businesses within a 24-hour or two-day time frame. Because of enormous economies of scale in sorting and distribution, all packages from across the United States are flown to Memphis, sorted, flown to regional airports, and then taken by courier truck to their eventual destinations. Even packages sent from Los Angeles to San Francisco have a stop in Memphis in between. Both speed and accuracy are of paramount importance in this sorting process. Timing is critical. A slow sort in Memphis leads to costly delays, late shipments, and irate customers; a quick sort means cost efficiency, timely deliveries, and happy customers. The problem was; How do you motivate employees to quickly accomplish this vital function? When paid on an hourly basis, costly delays were a recurring management nightmare. Problems continued until someone came up with the idea of paying teams of employees by the job performed and not according to the number of hours taken in accomplishing it. Whenever the sort was completed, all employees were done and could go home. Until the sort was completed, the job was not finished, and nobody could go home. Not only were individual employees motivated to complete their tasks as quickly and efficiently as possible, but monitoring employee performance also became less of a management headache and more of a team responsibility. Costly delay became a thing of the past, and FedEx’s business boomed as it gained a well-deserved reputation for efficient and timely deliveries. In a similar fashion, efficient mail delivery and garbage collection services can be provided when employees are paid for completing service along specific routes (“task completion”) rather than by more conventional hourly compensation. FedEx Corp. accomplished three important objectives when it revised its employee compensation methods at its Memphis distribution center. First, by switching from hourly

108

MARK HIRSCHEY

compensation to a task completion compensation scheme, FedEx is better able to align employee incentives with managerial objectives for timely and accurate service. This greatly reduces management’s need for monitoring. Second, by tying compensation to output generated rather than input hours, FedEx empowers front-line employees to design the method of task completion to suit worker needs preferences. By allocating important decision rights to front-line employees, these workers are better able to serve customers by quickly responding to unpredictable problems tied to weather, scheduling snafus, and so on. Finally, by empowering front-line workers, FedEx is able to attract and retain workers who seek jobs that demand and reward individual initiative. 10. In the case of FedEx, an hourly compensation plan not only failed to provide incentives for timely and accurate sorting of parcels and packages, but it actually rewarded employees if scheduling holdups resulted in delayed deliveries. 11. To illustrate the role of the outside labor market in setting lower bounds for worker compensation, consider the case of municipal police and fire workers in Madison, Wisconsin. Years ago, a controversy arose as to what level of pay would result in a “fair” rate of relative compensation for police and fire workers. Industrial psychologists retained by the City argued persuasively that the level of stress, responsibility, and demands placed upon fire workers were quite comparable with those placed upon law enforcement personnel. As a result, they recommended that comparable rates of pay be set for police and fire personnel. Others in the community contended that demands placed upon the police were far greater than those placed upon fire workers and that the police deserved a significant wage premium. This controversy was resolved in favor of a wage premium for the police when it was pointed out that at similar wages a significant shortage of qualified police recruits was being experienced, although an excessively large number of qualified fire worker applicants had applied for positions. The local labor market was telling the city that the police had the far tougher job and that it had to raise police wages to find sufficient qualified applicants. Conversely, fire workers were overpaid, at least on a relative basis. 12. See Harter and Harikumar (2002) for an interesting perspective on the economic cost approach to the accounting for option-based compensation. 13. In a related stream of research, beneficial changes in corporate performance have been positively associated with changes in managerial stock ownership (see, for example, Cole & Mehran, 1998; Mayers & Smith, 2002).

REFERENCES Abdel-khalik, A. R. (2002). Reforming corporate governance post Enron: Shareholders’ board of trustees and the auditor. Journal of Accounting & Public Policy, 21, 97–103. Amoako-Adu, B., & Smith, B. F. (2001). Dual class firms: Capitalization, ownership structure and recapitalization back into single class. Journal of Banking & Finance, 25, 1083–1111. Anctil, R., & Dutta, S. (1999). Negotiated transfer pricing and divisional vs. firm-wide performance evaluation. Accounting Review, 74, 87–104. Anderson, S. W., Glenn, D., & Sedatole, K. L. (2000). Sourcing parts of complex products: Evidence on transactions costs, high-powered incentives and ex-post opportunities. Accounting, Organizations & Society, 25, 723–749. Ang, J., Lauterbach, B., & Schreiber, B. Z. (2002). Pay at the executive suite: How do us banks compensate their top management teams? Journal of Banking & Finance, 26, 1143–1163.

Organization Structure and Corporate Governance

109

Arnold, T., & Schwalbe, U. (2002). Dynamic coalition formation and the core. Journal of Economic Behavior & Organization, 49, 363–380. Barr, J., & Saraceno, F. A. (2002). A computational theory of the firm. Journal of Economic Behavior & Organization, 49, 345–361. Bates, K. A. (1997). The role of coercive forces in organization design adoption. Academy of Management Review, 22, 849–851. Benston, G. J., & Hartgraves, A. L. (2002). Enron: What happened and what we can learn from it? Journal of Accounting & Public Policy, 21, 105–127. Berkowitz, J. (2001). Testing density forecasts, with applications to risk management. Journal of Business & Economic Statistics, 19, 465–474. Berle, A. A., & Means, G. C. (1932). The modern corporation and private property. New York: The Macmillan Co. Billett, M. T., & Mauer, D. C. (2000). Diversification and the value of internal capital markets: The case of tracking stock. Journal of Banking & Finance, 24, 1457–1490. Boyle, B., Dwyer, F. R., Robicheaux, R. A., & Simpson, J. T. (1992). Influence strategies in marketing channels: Measures and use in different relationship structures. Journal of Marketing Research, 29, 462–473. Brickley, J. A., Smith, C. W., Jr., & Zimmerman, J. L. (2002). Business ethics and organizational architecture. Journal of Banking & Finance, 26, 1821–1835. Bushman, R. M., & Smith, A. J. (2001). Financial accounting information and corporate governance. Journal of Accounting & Economics, 32, 237–333. Chauvin, K. W., & Hirschey, M. (1996). Ownership structure and capital atructure. In: M. Hirschey & M. W. Marr (Eds), Advances in Financial Economics (Vol. 2, pp. 25–46). Greenwich, CN: JAI Press. Chompalov, I., Genuth, J., & Shrum, W. (2002). The organization of scientific collaborations. Research Policy, 31, 749–767. Coase, R. (1937). The nature of the firm. Economica, 4, 386–405. Coase, R. (1990). Accounting and the theory of the firm. Journal of Accounting & Economics, 12, 3–13. Cohan, J. A. (2002). “I didn’t know” and “I was only doing my job”: Has corporate governance careened out of control? A case study of Enron’s information myopia. Journal of Business Ethics, 40, 275–299. Cole, R. A., & Mehran, H. (1998). The effect of changes in ownership structure on performance: Evidence from the thrift industry. Journal of Financial Economics, 50, 291–317. Daboub, A. J. (2002). Strategic alliances, network organizations, and ethical responsibility. S.A.M. Advanced Management Journal, 67, 40–48. Demsetz, H. (1997). The firm in economic theory: A quiet revolution. American Economic Review, 87, 426–429. Demsetz, H., & Lehn, K. (1985). The structure of corporate ownership: Causes and consequences. Journal of Political Economy, 93, 1155–1177. Denis, D. J., Denis, D. K., & Sarin, A. (1997). Ownership structure and top executive turnover. Journal of Financial Economics, 45, 193–221. Dixit, A., & Olson, M. (2000). Does voluntary participation undermine the Coase Theorem? Journal of Public Economics, 76, 309–335. Dobrev, S. D., Kim, T., & Carroll, G. R. (2002). The evolution of organizational niches: U.S. automobile manufacturers, 1885–1981. Administrative Science Quarterly, 47, 233– 264.

110

MARK HIRSCHEY

Docking, D. S., Hirschey, M., & Jones, E. (1997). Information and contagion effects of bank loan-loss reserve announcements. Journal of Financial Economics, 43, 219–239. Elberfeld, W. (2002). Market size and vertical integration: Stigler’s hypothesis reconsidered. Journal of Industrial Economics, 50, 23–42. Ellickson, R. C., Stigler, G. J., & Donohue, J. J., III (1989). The case for Coase and against “Coaseanism” two notes on the Coase theorem reply. Yale Law Journal, 99, 611–636. Fan, J. P. H., & Wong, T. J. (2002). Corporate ownership structure and the informativeness of accounting earnings in East Asia. Journal of Accounting & Economics, 33, 401–425. Gedajlovic, E., & Shapiro, D. (2002). Ownership structure and firm profitability in Japan. Academy of Management Journal, 45, 565–575. Goold, M., & Campbell, A. (2002). Do you have a well-designed organization? Harvard Business Review, 80, 117–124. Hannah, L. (1998). Survival and size mobility among the world’s largest 100 industrial corporations, 1912–1995. American Economic Review, 88, 62–65. Hanushek, E., Heckman, J., & Neal, D. (2002). Introduction to JHR’s special issue on designing incentives to promote human capital. Journal of Human Resources, 37, 693–695. Harper, J. T., & Madura, J. (2002). Sources of hidden value and risk within tracking stock. Financial Management, 31, 91–109. Harris, M., & Raviv, A. (2002). Organization design. Management Science, 48, 852–865. Harter, C. I., & Harikumar, T. (2002). The economic cost approach. Journal of Business Finance & Accounting, 29, 1007–1022. Hartgraves, A. L., & Benston, G. J. (2002). The evolving accounting standards for special purpose entities and consolidations. Accounting Horizons, 16, 245–258. Hirschey, M., & Jones, E. (2001). The role of federal law enforcement actions in corporate governance. In: M. Hirschey, K. John & A. Makhija (Eds), Advances in Financial Economics (Vol. 6, pp. 117–141). Amsterdam, The Netherlands: Elsevier Science. Hirschey, M., & Richardson, V. J. (2002). Information content of accounting goodwill numbers. Journal of Accounting & Public Policy, 21, 173–191. Hirschey, M., Richardson, V. J., & Scholz (2001). Value relevance of non-financial information: The case of patent data. Review of Quantitative Finance & Accounting, 17, 223–235. Hoogeweegen, M. R., Teunissen, W. J. M., Vervest, P. H. M., & Wagenaar, R. W. (1999). Modular network design: Using information and communication technology to allocate production tasks in a virtual organization. Decision Sciences, 30, 1073–1103. Insead, D. C. G., & Eisenhardt, K. M. (2001). Architectural innovation and modular corporate forms. Academy of Management Journal, 44, 1229–1249. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review, 76, 323–329. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3, 305–360. Kachelmeier, S. J., & Towry, K. L. (2002). Negotiated transfer pricing: Is fairness easier said than done? Accounting Review, 77, 571–593. Keating, A. S. (1997). Determinants of divisional performance evaluation practices. Journal of Accounting & Economics, 24, 243–273. Killen, C. P., Hunt, R., Ayres, B., & Janssen, C. (2002). Strategic alliances for world competitiveness. International Journal of Technology Management, 24, 569–582. Klein, A. (2002). Audit committee, board of director characteristics, and earnings management. Journal of Accounting & Economics, 33, 375–400.

Organization Structure and Corporate Governance

111

Koka, B. R., & Prescott, J. E. (2002). Strategic alliances as social capital: A multidimensional view. Strategic Management Journal, 23, 795–816. Larner, R. J. (1970). Management control and the large corporation. New York: Dunellen Publishing Co. Lehn, K. (2002). Corporate governance in the deregulated telecommunications industry: Lessons from the airline industry. Telecommunications Policy, 26, 225–242. Lei, D., & Slocum, J. W. (2002). Organization designs to renew competitive advantage. Organizational Dynamics, 31, 1–18. Mace, M. (2002). Directions for directors. Harvard Business Review, 80, 136. Madhok, A. (2002). Reassessing the fundamentals and beyond: Ronald Coase, the transaction cost and resource-based theories of the firm and the institutional structure of production. Strategic Management Journal, 23, 535–550. Masten, S. E. (2002). Modern evidence on the firm. American Economic Review, 92, 428–432. Mayer, C. (2002). Financing the new economy: Financial institutions and corporate governance. Information Economics & Policy, 14, 311–326. Mayers, D., & Smith, C. (2002). Ownership structure and control: Property-casualty insurer conversion to stock charter. Journal of Financial Services Research, 21, 117–144. McConnell, J. J., & Servaes, H. (1990). Additional evidence on equity ownership and corporate value. Journal of Financial Economics, 27, 595–612. McConnell, J. J., & Servaes, H. (1995). Equity ownership and the two faces of debt. Journal of Financial Economics, 39, 131–157. Mishra, A. (2002). Hierarchies, incentives and collusion in a model of enforcement. Journal of Economic Behavior & Organization, 47, 165–178. Morck, R., Nakamura, M., & Shivdasani, A. (2000). Banks, ownership structure, and firm value in Japan. Journal of Business, 73, 539–567. Morck, R., Shleifer, A., & Vishny, R. W. (1988). Management ownership and market valuation. Journal of Financial Economics, 20, 293–315. Mulvey, P. W., LeBlanc, P. V., & Heneman, R. (2002). Study finds that knowledge of pay process can beat out amount of pay in employee retention, organizational effectiveness. Journal of Organizational Excellence, 21, 29–42. Munari, F. (2002). The effects of privatization on corporate R&D units: Evidence from Italy and France. R & D Management, 32, 223–232. Myers, S., & Majluf, N. S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 13, 187– 221. Nagar, V. (2002). Delegation and incentive compensation. Accounting Review, 77, 379–395. Nault, B. R. (1998). Information technology and organization design: Locating decisions and information. Management Science, 44, 1321–1335. Pagano, M., & Roell, A. (1998). The choice of stock ownership structure: Agency costs, monitoring, and the decision to go public. Quarterly Journal of Economics, 113, 187–225. Prat, A. (2002). Should a team be homogenous? European Economic Review, 46, 1187–1207. Prather, L. J., & Middleton, K. L. (2002). Are N + 1 heads better than one? The case of mutual fund managers. Journal of Economic Behavior & Organization, 47, 103–120. Raz, T., Shenhar, A. J., & Dvir, D. (2002). Risk management, project success, and technological uncertainty. R & D Management, 32, 101–109. Rindova, V. P., & Kotha, S. (2001). Continuous “morphing”: Competing through dynamic capabilities, form, and function. Academy of Management Journal, 44, 1263–1280.

112

MARK HIRSCHEY

Schroeder, J. L. (1998). The end of the market: A psychoanalysis of law and economics. Harvard Law Review, 112, 483–558. Sengupta, S., & Perry, M. (1997). Some antecedents of global strategic alliance formation. Journal of International Marketing, 21, 31–50. Shaw, J. D., Gupta, N., & Delery, J. E. (2002). Pay dispersion and workforce performance: Moderating effects of incentives and interdependence. Strategic Management Journal, 23, 491–512. Shen, W., & Cannella, A. A., Jr. (2002). Revisiting the performance consequences of CEO succession: The impacts of successor type, post succession senior executive turnover, and departing CEO tenure. Academy of Management Journal, 45, 717–733. Smith, A. (1776). The wealth of nations. New York: Cannan Edition, Modern Library, 1937, 700. Sonnenfeld, J. A. (2002). What makes great boards great. Harvard Business Review, 80, 106–113. Stigler, G. J. (1951). The division of labor is limited by the extent of the market. Journal of Political Economy, 59, 185–193. Stuart, T. (1998). Network positions and propensities to collaborate: An investigation of strategic alliance formation in a high-technology industry. Administrative Science Quarterly, 43, 668– 698. Whipple, J. M., & Frankel, R. (2000). Strategic alliance success factors. Journal of Supply Chain Management, 36, 21–28. Wiersema, M. (2002). Holes at the top: Why CEO firings backfire. Harvard Business Review, 80, 70–77. Williamson, O. E. (1998). The institutions of governance. American Economic Review, 88, 75–79. Yeo, G. H. H., Tan, P. M. S., Ho, K. W., & Chen, S. (2002). Corporate ownership structure and the informativeness of earnings. Journal of Business Finance & Accounting, 29, 1023–1046. Yu, L. (2002). Risk management in practice. MIT Sloan Management Review, 44, 9.

OWNERSHIP STRUCTURE AND SHAREHOLDER VOTING ON BOARD STRUCTURE CHANGES Teresa A. John and Gopala K. Vasudevan ABSTRACT We examine voting outcomes on shareholder governance proposals that seek annual elections for all the directors on the corporate board. We relate these voting outcomes to different ownership structure characteristics and a series of variables that are publicly available. The pattern of support indicates that proposals are generally successful when they are supported by large activist groups and when institutions hold a significant fraction of shares outstanding. Our evidence casts some doubt on the efficacy of the Rule 14A-8 mechanism, which limits the amount of information that can be provided to shareholders as part of the proposal.

1. INTRODUCTION The recent decline in the market for corporate control has led to an increased role for internal corporate control mechanisms such as shareholder activism and corporate boards intervention.1 Proposals that call for changes in board structure, composition and compensation have become very popular and are gaining increasing shareholder support. Corporate boards have played a very active role in the restructuring of companies such as General Motors, K Mart and Kodak. Academics, legal experts, and the popular press view internal control Corporate Governance and Finance Advances in Financial Economics, Volume 8, 113–125 © 2003 Published by Elsevier Science Ltd. ISSN: 1569-3732/PII: S1569373203080058

113

114

TERESA A. JOHN AND GOPALA K. VASUDEVAN

mechanisms to be superior to external control mechanisms (see Pound, 1988; Roe, 1994). These internal control mechanisms are more flexible and can correct mistakes at an early stage, thereby avoiding the drastic changes which accompany a takeover.2 In this study we examine the factors which influence the success of shareholder proposals for a change in board structure. We examine voting outcomes on 169 shareholder governance proposals during the 1990–1995 period under which shareholders sought to declassify the board. We relate the voting outcomes to a series of different ownership structure characteristics, including ownership by institutions, outside blockholders and insider ownership. We also relate the voting outcomes on these proposals to a series of variables that are publicly available such as the stock price performance of the firm and identity of the sponsor. The evidence in this paper fills a void in the literature. Most studies on voting outcomes have examined proxy contests and management-sponsored resolutions to change corporate governance structure. An exception is Gordon and Pound (1993), whose study examines the shareholder proposals to change corporate governance structure. However, their study did not include any proposals that were successful in declassifying the board. Shareholder proposals to change the board structure are becoming more common, and are passing more frequently. In this study, we provide some evidence on when these proposals will be successful. Our evidence also provides an explanation for the surprising lack of any stock market reaction to the successful passage of shareholder resolutions that could increase shareholder wealth (see, for example, John & Klein, 1996; Karpoff, Malatesta & Walking, 1996). We find that the market is able to anticipate the outcome of the proposal, hence, there is no market reaction when these proposals are passed. Our evidence indicates that the percentage of votes cast for shareholdersponsored proposals to change the board structure vary systematically as a function of the ownership structure and the identity of the proposal sponsor. Proposals receive more votes when they are sponsored by activist organizations, such as the United Shareholders Association (USA) or the Investor’s Rights Association of American (IRAA), and when there is large institutional ownership. We do not find any relation between the votes received and variables such as the economic performance of the firm, insider ownership, and growth opportunities. Our evidence casts some doubt on the efficacy of Rule 14A-8, which limits the amount of information that can be provided to shareholders as part of the proposal. The paper is organized as follows: Section 2 provides background on the shareholder voting process and discusses the literature on corporate boards. Section 3 describes the sample and the methodology. Section 4 presents our

Ownership Structure and Shareholder Voting on Board Structure Changes

115

evidence on the determinants of the voting on changes in board structure. Section 5 summarizes and concludes.

2. BACKGROUND Shareholder proposals to make changes in the corporate governance structure have become increasingly common and more successful in recent years. One of the reasons for their success could be that they are being used more often by large fiduciary institutions such as the California Public Employees Retirement Systems (Calpers), Wisconsin State Investment Board, and the College Retirement Equity Fund. These shareholder proposals are very specific and generally advisory. They relate to issues such as abolishing greenmail, establishing confidential voting, relaxing supermajority provisions to changes in board structure (such as changing from a staggered board to a non-staggered board), and abolishing pension plans for directors. Although these proposals are advisory rather than binding, their passage puts pressure on management to adopt them. In most cases, management adopts a proposal if it receives the required majority of votes. In the few cases when a proposal is passed but not accepted, the proposer includes it in the following year’s agenda putting more pressure on the management to adopt its measures. Sometimes, management will put forward its own proposal the following year, and accept it once it is passed. Shareholder proposals are quite different from the choices shareholders have when they participate in proxy contests or in management proposals. Shareholder proposals on governance are made under SEC’s Rule 14A-8, established in 1942, commonly referred to as “shareholder proposal rule.” Under this rule, shareholders can make proposals on suggested changes to corporate governance structure of up to 500 words in length. Management is required to include these proposals in their proxy solicitation materials so that shareholders can indicate their preferences. The factor that distinguishes proposals under Rule 14A-8 from other non-routine proposals is the 500-word limit imposed on the amount of information that can be provided.3 This means shareholders must use their own resources to become informed about these issues. Until the late 1980s it was standard practice to elect all the directors at the same time during the annual meeting of shareholders. With the coming of the takeover era, raiders found they could take over a company by proposing their own board of directors who competed with the management’s nominees. To prevent this from happening, companies began to nominate directors for three staggered sets of three year terms.4 Under this scheme one third of the seats on the board would be elected every year. A raider who wanted to control the board would have

116

TERESA A. JOHN AND GOPALA K. VASUDEVAN

to put up nominees for three succeeding years, a very long time to maintain a hostile bid. Management generally argues that adopting a staggered board ensures a continuity of service. However, shareholder advocates argue that shareholders should have the right to vote for the entire board every year which enables them to remove the directors who are not performing their job. They believe that staggered boards insulate both management and the board from shareholders. While staggered boards make it extremely difficult for raiders to take over a company, the practice also gives fewer choices to shareholders in their efforts to remove existing management or to restructure the company. Studies that have examined the stock price reaction to the announcement of poison pill amendments, such as the adoption of a staggered board, find that the adoption of a staggered board reduces firm value (Dann & De Angelo, 1983). The majority of recent studies focus on the benefits of outside directors and small boards.5 Rosenstein and Wyatt (1990), for example, find that shareholder wealth increases with the addition of outsiders to the board. Brickley, Coles and Terry (1994) find that the market response to management-initiated control actions is positively related to the percentage of outsiders on the board. Booth and Deli (1996) examine the factors affecting the number of outside directorships held by CEOs. They find that the number of outside directorships held by CEOs increases as they get closer to the end of their tenure and transfer decision-making to their eventual successors. Booth and Deli also find that the number of outside directorships is higher for CEOs of mature firms and increases with the CEO’s tenure. Yermack (1996) finds that firms with small boards tend to perform better. Prior empirical work on voting has examined proxy contests in which outside dissident shareholders try to get board control, and management-sponsored proposals to enact protections against proxy contests. Pound (1988) shows that dissidents are more likely to win proxy contests when the target firm has higher institutional ownership, when managerial and non-dissident block ownership is lower, and when dissidents are offering a higher premium to buy the shares conditional on winning the proxy contest. Brickley, Lease and Smith (1988) show that management-sponsored anti-takeover amendments receive more support from pressure-sensitive institutional investors, such as insurance companies, and less support from money managers. Jarrell and Poulsen (1987) show that valuedecreasing anti-takeover amendments, such as staggered board amendments, are adopted at firms with lower institutional and higher insider ownership. Gordon and Pound (1993) examine shareholder-sponsored proposals to change corporate governance structure. They find that the percentage of votes cast for shareholder-sponsored proposals varies systematically as a function of the type of proposal, the identity of the sponsor, and the economic performance of the firm.

Ownership Structure and Shareholder Voting on Board Structure Changes

117

The percentage of votes cast decreases with insider ownership and increases with institutional ownership. Gillian and Starks (2000) examine shareholder proposals during the 1987–1994 period. They find that the sponsor identity, issue type and prior firm performance are useful in predicting the outcome of the proposal. Proposals sponsored by institutions gain more support than proposals sponsored by individuals.

3. SAMPLE AND METHODOLOGY Our sample consists of 169 shareholder-sponsored proposals to declassify the board. The period covered is 1990–1995. Our source for these proposals is the Investor Responsibility Research Center (IRRC), which lists the outcomes on these proposals. For each proposal, IRRC provides information on the voting date, outcome, percentage of votes cast in favor of the proposal, and the sponsor of the proposal. We also obtain data on the corporate stock price performance of the firm from Compustat. Data on ownership structure is taken from disclosure and growth opportunities measured by the market-to-book value of equity from Compustat. The data shows that among the 169 shareholder proposals to declassify the board, only eight were successful. Table 1 provides information on the sample of firms that passed the proposals to declassify their boards. It is interesting to note that although the sample is from 1990–1995, only recently have such proposals begun to pass. The first case in which such a proposal passed was in 1993 for Lockheed Martin; six of the eight successful proposals were made in 1995. One of the reasons why these governance proposals are becoming more successful could be because they are being used more Table 1. Description of the Eight Sample Firms that Adopted Resolutions to Abolish Staggered Boards. Company Name

Shareholder Meeting Date

Lockheed Martin Mead Corporation Alumax, Inc. Chase Manhattan Host Marriott K Mart Oryx Energy Stride Rite

4/22/93 4/28/94 5/4/95 4/18/95 5/11/95 5/23/95 5/4/95 4/12/95

Sponsor E. Davis UPI IRAA E. Davis E. Davis G. Switzer Calpers IRAA

Percentage of Vote for Proposal 0.508 0.539 0.677 0.520 0.510 0.605 0.561 0.506

118

TERESA A. JOHN AND GOPALA K. VASUDEVAN

Table 2. Characteristics of the Firms for which the Proposal was not Adopted. Variables STKRET MKTBOOK INSTOWNER INSDOWNER BLOCKOWNER

No. of Obs.

Mean

Standard Dev.

Minimum

Maximum

161 161 161 161 161

0.0343 1.0089 0.5204 0.0963 0.1858

0.1011 14.3833 0.1756 0.1612 0.2390

−0.2341 −170.5300 0.1139 0.0003 0.0000

0.4877 22.5800 0.9647 0.8511 0.9999

Notes: PROBPASS = the percentage of the votes for the proposal at the shareholder meeting; STKRET = the stock return for the last year; MKTBOOK = the ratio of the market value of equity to the book value of equity at the end of the previous fiscal year; INSTOWNER = the percentage ownership of institutional shareholders; INSDOWNER = the percentage ownership of insider shareholders; BLOCKOWNER = the percentage ownership of the shareholders who own at least 5% of shares.

often by large fiduciary institutions such as Calpers, Wisconsin State Investment Board, and the College Retirement Equity Fund. Table 2 presents summary statistics of the 161 firms that did not adopt such proposals. These firms had a return of 3.43% in the year prior to the issue; the market-to-book ratio of equity was very close to one. Institutions held 52.04% of the shares in these firms; insider and block owners held on average 9.63% and 18.58% of the total shares outstanding. We relate the voting outcomes to a series of different ownership structure characteristics, including ownership by institutions, outside blockholders and insider ownership, the stock price performance of the firm and identity of the sponsor using a weighted least squares logit regression where the dependent variable is the percentage of votes for the proposal, measured as a percentage of total votes cast. We use weighted least squares because it has the same asymptotic distribution as the maximum likelihood estimator (see Greene, 1990, pp. 666–671). In our regressions we use only the sample of outcomes that were not successful. We use methodology which applies to “proportions data,” in which the sample is a homogenous group which had the same outcome (i.e. all the proposals ended in failure). We then use the estimates from this regression to make predictions about the successful outcomes.

4. DETERMINANTS OF THE VOTING OUTCOME In this section we determine the variables that influence the outcome of a shareholder proposal to declassify the board. We use a weighted least squares logit regression in which the dependent variable is the percentage of votes cast

Ownership Structure and Shareholder Voting on Board Structure Changes

119

in favor of the proposal and the weights are the standard deviation of the error terms. As firm performance variables, we use the stock returns of the target firm during the one-year period prior to the issue (STKRET) and the market-to-book value of equity (MKTBOOK). Firms with poor stock price performance might have poor incumbent management and disgruntled shareholders might be willing to join with dissident shareholders in bringing about changes in the firms. We use the market-to-book value of equity (MKTBOOK) to proxy for future expected performance of the firm. As ownership structure variables, we use institutional, insider and block holder ownership. We use the percentage held by institutional owners (INSTOWNER) as an independent variable. Pozen (1994) argues that it is in the interest of institutional investors to vote for proposals that promote good governance structures, since these act as safeguards in times of crises. Firms with a larger fraction of institutional owners are more likely to adopt such proposals. The second ownership variable we use is the percentage held by insider owners (INSDOWNER). Proxy proposals in firms with larger insider ownership are less likely to succeed because insiders are likely to vote against these proposals. Our last ownership variable is the percentage held by block owners (BLOCKOWNER), which we define as those investors who own more than 5% of the total shares. In their role as monitors, block owners should support shareholder initiatives more frequently. We use five dummies for the identity of the sponsor of the resolution: The first dummy variable is set equal to one when the proposal is made by an institution such as a pension fund. Shareholder resolutions sponsored by professional investors who have significant investments in the firm should receive more support than proposals made by small activists. The second dummy is set equal to one if the proposer is a union: Some unions hold significant stakes in firms and hence can determine the outcome of a proposal. Dummy 3 is set equal to one if the proposal is sponsored by activist organizations with a large following such as the USA or IRRA. These institutions have played a major part in putting forward shareholder proposals; since they are more organized, these proposals have a higher chance of passing. Dummy 4 is set equal to one if the sponsor is an unaffiliated investor, such as Gilbert or the Davis brothers. The proposals put forward by such unaffiliated investors usually have a low chance of passing. Dummy 5 is set equal to one if the shareholder proposal is sponsored by Calpers. Calpers has been very active regarding corporate governance issues. Each year it targets a group of firms that have been performing poorly. Smith (1996) finds that the long-term stock price performance of the firms targeted by Calpers improves following the targeting. Table 3 reports the results of the regressions. In Model 1, we omit the dummy variables that proxy for the identity of the sponsor of the resolution. The only

120

TERESA A. JOHN AND GOPALA K. VASUDEVAN

Table 3. Determinants of the Support for Governance Proposals to Abolish Staggered Board. Independent

Model 1

INTERCEPT

−0.6224 (−19.812)

Model 2

0.0783 (0.821) 0.0017 (0.649)

−0.6089 (−19.301) −0.0233 (−0.172) 0.0715 (0.952) 0.3040 (3.190) −0.1086 (−0.727) 0.1701 (0.576) −0.0117 (−0.115) 0.0014 (0.542)

1.3005 (10.709) −0.0942 (−0.486) −0.1091 (−0.779) 25.235 0.431

1.2277 (9.601) −0.0904 (−0.472) −0.1784 (−1.260) 14.302 0.454

DUMMY1 DUMMY2 DUMMY3 DUMMY4 DUMMY5 STKRET MKTBOOK INSTOWNER INSDOWNER BLOCKOWNER F-statistic Adj. R-square

Variables

Note: Our results are based on 161 observations for which the proposal was not adopted at the shareholders’ meeting. We use WLS (weighted least squares) method to estimate the logit regressions and use the standard deviation of the error terms as weight (t-values are in parentheses) Model 1: Logit (PROBPASS) = ␣0 INTERCEPT + ␤1 STKRET + ␤2 MKTBOOK + ␤3 INSTOWNER + ␤4 INSDOWNER + ␤5 BLOCKOWNER + u, Model 2: Logit (PROBPASS) = ␣0 INTERCEPT + ␣1 DUMMY1 + ␣2 DUMMY2 + ␣3 DUMMY3 + ␣4 DUMMY4 + ␣5 DUMMY5 + ␤1 STKRET + ␤2 MKTBOOK + ␤3 INSTOWNER + ␤4 INSDOWNER + ␤5 BLOCKOWNER + u, where PROBPASS = the percentage of the votes for the proposal at the shareholder meeting; INTERCEPT = intercept; DUMMY1 = 1 if the proposer is such an institution as pension fund, 0 otherwise; DUMMY2 = 1 if the proposer is a union, 0 otherwise; DUMMY3 = 1 if the proposer is United Shareholders Association or IRAA, 0 otherwise; DUMMY4 = 1 if the proposer is unaffiliated individual, 0 otherwise; DUMMY5 = 1 if the proposer is CALPERS, 0 otherwise; STKRET = the stock return for the last year; MKTBOOK = the ratio of the market value of equity to the book value of equity at the end of the last fiscal year; INSTOWNER = the percentage ownership of institutional shareholders; INSDOWNER = the percentage ownership of insider shareholders; BLOCKOWNER = the percentage ownership of the shareholders who own at least 5% of shares; and u = disturbance terms.

Ownership Structure and Shareholder Voting on Board Structure Changes

121

explanatory variable that is significant is the coefficient of institutional ownership, which is positive (1.3) and significant at the 1% level. This indicates that with every 1% increase in institutional ownership, there is a 1.3% increase in the number of votes cast in favor of the resolution. The intercept is negative and significant at the 1% level. The adjusted R-square is 43.1%, indicating that these variables can explain 43.1% of the voting outcome. In Model 2, we also include the dummies representing the sponsor of the resolution. As with the earlier regression, the coefficient of institutional ownership is positive and significant at the 1% level. The only other variable that is significant is the coefficient of the dummy when the sponsor is USA or IRAA. This is positive and significant at the 1% level. The adjusted R-square is 45.4%. As we have noted, the pattern of support for shareholder proposals indicates that proposals tend to be successful when they are sponsored by large activist groups and when institutions hold a significant fraction of the shares outstanding. Since these changes are initiated under Rule 14A-8, information regarding these proposals can be scarce. Large activist organizations such as the USA and IRAA can spend their own resources to give outside investors more information about their proposals. Similarly, institutions can spend their own resources and generate more information about these firms. Surprisingly, we do not find that one-year prior returns are a significant explanatory factor. The information conveyed by the one-year return could contain too much noise. Hence, investors may not be using it as a good predictor of corporate performance. Table 4 presents the results of our models where we predict the percentage of votes in favor of the proposal for the eight successful outcomes. In Model 1 we use stock returns for the prior year, the market-to-book value of equity at the end of the last fiscal year, the percentage of shares held by institutional owners, insider owners and block owners as the independent variables. Column 2 reports the results of our prediction model. For every event the model is able to correctly predict the outcome of the proposal (success or failure). In the second specification we also include the dummies for the identity of the sponsor. The model is able to predict the outcome for all the eight successful events in our sample. It is interesting to note that when we include the identity of the sponsor the predicted percentage of votes is slightly lower, indicating that the identify of the sponsor is an important factor which influences the outcome. The variables we use are observable to the market participants at the time of the voting and hence the market will be able to correctly anticipate the outcome of the voting. This could explain the surprising lack of any stock price reaction to the successful passage of shareholder resolutions that could increase shareholder wealth (see, for example, John & Klein, 1996; Karpoff, Malatesta & Walking, 1996).

122

TERESA A. JOHN AND GOPALA K. VASUDEVAN

Table 4. Predicted versus Actual Percentage of Votes for the Eight Firms that Adopted the Proposal. Company Name

Lockheed Martin Mead Corporation Alumax, Inc. Chase Manhattan Host Marriott K Mart Oryx Energy Stride Rite

Predicted Prob. Model 1

Model 2

63.0 59.5 59.7 55.0 54.8 53.2 54.0 55.7

61.8 57.5 57.8 53.2 53.3 52.5 53.0 55.0

Actual Percent of Votes for Proposal 50.8 53.9 67.7 52.0 51.0 60.5 56.1 50.6

Model 1: Logit (PROBPASS) = ␣0 INTERCEPT + ␤1 GROWTH + ␤2 MKTBOOK + ␤3 INSTOWNER + ␤4 INSDOWNER + ␤5 BLOCKOWNER + u, Model 2: Logit (PROBPASS) = ␣0 INTERCEPT + ␣1 DUMMY1 + ␣2 DUMMY2 + ␣3 DUMMY3 + ␣4 DUMMY4 + ␣5 DUMMY5 + ␤1 GROWTH + ␤2 MKTBOOK + ␤3 INSTOWNER + ␤4 INSDOWNER + ␤5 BLOCKOWNER + u, where PROBPASS = the percentage of the votes for the proposal at the shareholder meeting; INTERCEPT = intercept; DUMMY1 = 1 if the proposer is such an institution as pension fund, 0 otherwise; DUMMY2 = 1 if the proposer is a union, 0 otherwise; DUMMY3 = 1 if the proposer is United Shareholders Association or IRAA, 0 otherwise; DUMMY4 = 1 if the proposer is unaffiliated individual, 0 otherwise; DUMMY5 = 1 if the proposer is CALPERS, 0 otherwise; GROWTH = the stock return for the last year; MKTBOOK = the ratio of the market value of equity to the book value of equity at the end of the last fiscal year; INSTOWNER = the percentage ownership of institutional shareholders; INSDOWNER = the percentage ownership of insider shareholders; BLOCKOWNER = the percentage ownership of the shareholders who own at least 5% of shares; and u = disturbance terms.

5. SUMMARY AND CONCLUSIONS Internal corporate control mechanisms such as shareholder activism and corporate board intervention have become major components of the corporate landscape. Academics view these internal control mechanisms as being more flexible and better able to correct mistakes at an early stage thus avoiding the drastic changes that accompany external control mechanisms such as takeovers. In this study, we examine the factors that influence the success of shareholder proposals for a change in board structure from a staggered to a non-staggered board. The evidence in this paper fills a void in the literature. Most studies on voting outcomes have examined proxy contests and management-sponsored resolutions

Ownership Structure and Shareholder Voting on Board Structure Changes

123

to change corporate governance structure. We provide some evidence on which of these proposals will be successful and the factors influencing the success of these proposals. Our evidence also explains the surprising lack of any stock market reaction to the successful passage of shareholder resolutions that could potentially increase shareholder wealth. We find that since the market is able to anticipate the outcome of the proposal it does not react when the proposal is passed. We examine voting outcomes on 169 shareholder proposals from the period 1990–1995 under which shareholders sought to have annual elections for all the directors of the corporate board (non-staggered board). We relate the voting outcomes on these proposals to a series of variables that are publicly available, such as the firm’s stock price performance and the identity of the sponsor. We also relate the voting outcomes to a series of different ownership structure characteristics, including ownership by institutions, outside blockholders and insider ownership. The pattern of support indicates that proposals are generally successful when they are supported by large activist groups and when institutions hold a significant fraction of the shares outstanding. Since these changes are initiated under Rule 14A-8, information may be scarce which explains the voting pattern observed. Large activist organizations such as the USA and IRAA, can spend their own resources to inform outside investors about their proposals. Similarly, voting institutions can spend their own resources and generate more information about these firms that enables the institutional investors to make informed choices. Surprisingly, we do not find that long-term returns are a significant explanatory factor. The information conveyed by the one-year return could contain too much noise. Hence investors may not be using it as a good predictor of corporate performance. Our evidence casts some doubt on the efficacy of Rule 14A-8 which limits the amount of information that is provided to shareholders as part of the proposal.

NOTES 1. In the case of GM, the board replaced CEO Robert C. Stemple with John Smale, an outside member of the board. The board also played an important role in the restructuring of K Mart, where the board ousted CEO Antonio and appointed Donald S. Perkins, an outside member of the board. 2. The empirical literature on takeovers shows that acquisitions are a mixed blessing for acquiring companies and very good for target companies. Bradley, Desai and Kim (1988) and Morck, Shleifer and Vishny (1990) find that the stock price reaction to the announcement of an acquisition is close to zero and mostly negative. Agrawal, Jaffe and Mandelkar (1992) find that the post-acquisition stock returns are not very good for acquiring companies. The stock holders of acquiring companies suffer a loss of 10% over the five-year post-merger

124

TERESA A. JOHN AND GOPALA K. VASUDEVAN

period. Jensen and Ruback (1983) in their survey on takeovers also note that acquirers have poor stock price performance during the three-year period following the acquisition. 3. In non-routine voting decisions, the management and dissident shareholders have to provide detailed information about their proposals and their impact on the corporation. Typically these proposals run to dozens and even hundreds of pages. 4. This is sometimes called a classified board. Proposals to change the board from a staggered to a non-staggered board are called proposals to de-classify the board. 5. A notable exception is Gordon and Pound (1993).

REFERENCES Agrawal, A., Jaffee, J. F., & Mandelkar, G. N. (1992). The post-merger performance of acquiring firms: A re-examination of an anomaly. Journal of Finance (September), 1605–1621. Booth, J. R., & Deli, D. N. (1996). Factors affecting the number of outside directorships held by CEOs. Journal of Financial Economics, 81–104. Bradley, M., Desai, A., & Kim, E. H. (1988). Synergistic gains from corporate acquisitions and their division between the stockholders of target and acquiring firms. Journal of Financial Economics (May), 3–40. Brickley, J., Coles, J., & Terry, R. L. (1994). Outside directors and the adoption of poison pills. Journal of Financial Economics, 35, 371–390. Brickley, J., Lease, R., & Smith, C. (1988). Ownership structure and voting on anti-takeover amendments. Journal of Financial Economics (December), 267–293. Dann, L., & De Angelo, H. (1983). Standstill agreements, privately negotiated stock repurchases and the market for corporate control. Journal of Financial Economics (March), 273–300. Gillian, S., & Starks, L. (2000). Corporate governance proposals and shareholder activism: The role of institutional investors. Journal of Financial Economics, 42, 275–305. Greene, W. H. (1990). Econometric analysis. New York, NY: MacMillan. Gordon, L. A., & Pound, J. (1993). Information, ownership structure and shareholder voting: Evidence from shareholder-sponsored governance proposals. The Journal of Finance (June), 697–718. Jarrell, G., & Poulsen, A. (1987). Shark repellents and stock prices: The effects of anti-takeover amendments since 1980. Journal of Financial Economics, 19, 127–168. Jensen, M. C., & Ruback, R. S. (1983). The market for corporate control. Journal of Financial Economics (March), 5–50. John, K., & Klein, A. (1996). Shareholder proposals and corporate governance. New York University Working Paper (January). Karpoff, J., Malatesta, P., & Walking, R. (1996). Corporate governance and shareholder initiatives: Empirical evidence. Journal of Financial Economics, 42, 365–395. Morck, R., Shleifer, A., & Vishny, R. (1990). Do managerial objectives drive bad acquisitions? Journal of Finance (March), 31–48. Pound, J. (1988). Proxy contests and the efficiency of shareholder oversight. Journal of Financial Economics, 20, 237–266. Pozen, R. C. (1994). Institutional investors: The reluctant activists. Harvard Business Review (January–February), 140–149. Roe, M. R. (1994). Strong managers, weak owners. The Political Roots of American Corporate Finance. Princeton, NJ: Princeton University Press.

Ownership Structure and Shareholder Voting on Board Structure Changes

125

Rosenstein, S., & Wyatt, J. G. (1990). Outside directors, board independence and shareholder wealth. Journal of Financial Economics, 26(March), 175–192. Smith, M. (1996). Shareholder activism by institutional investors: Evidence from Calpers. Journal of Finance (March), 227–252. Yermack, D. (1996). Higher market valuation of companies with a small board of directors. Journal of Financial Economics (March), 185–212.

DID EARNINGS MANAGEMENT CONTRIBUTE TO THE OVERVALUATION OF ENRON’S STOCK? John D. Martin and Akin Sayrak ABSTRACT This paper asks whether market fundamentals can explain the run-up and collapse of Enron’s stock price and price-earnings ratio. We use a variant of the discounted cash flow model proposed by Miller and Modigliani (1961) to show that the growth rates implied by the stock’s valuation have rarely been achieved in recorded business history. We also provide evidence of earnings management by the company that may have contributed to extravagant investor expectations of earnings growth. Between 1990 and 2000 the firm’s reported earnings met or exceeded analysts’ earnings forecasts 77% of the time. Furthermore, beginning in 1997 Enron used asset sales (often to related parties) to generate as much as 83% of its annual earnings.

1. INTRODUCTION Enron’s common stock closed at $83.13 a share in December 31, 2000 and just eleven months later traded for $0.26. Was this precipitous decline the result of changes in investor expectations related to the underlying company fundamentals Corporate Governance and Finance Advances in Financial Economics, Volume 8, 127–145 Copyright © 2003 by Elsevier Science Ltd. All rights of reproduction in any form reserved ISSN: 1569-3732/PII: S156937320308006X

127

128

JOHN D. MARTIN AND AKIN SAYRAK

combined with bad market conditions, or the deflation of a price bubble created by the overzealous efforts of Enron executives to meet and exceed equity analysts’ earnings forecasts?1 We provide evidence suggesting that Enron was not only “laser focused on earnings” as its 2000 annual report states, but that it was extremely adept at meeting and exceeding analysts’ earnings forecasts. Furthermore, we now know that Enron’s share price rose to levels that reflected expectations that proved totally unfounded. In fact, the growth rate expectations underlying the valuation of Enron’s shares had been achieved by only a handful of firms in the recorded history of publicly traded firms. Although this is not conclusive evidence that Enron’s shares were overvalued, it is suggestive that they were at the very least “optimistically” priced. The implications of under-valuation have been dealt with extensively in the investment community; however, the consequences of over-valuation are relatively unstudied. Recently, Fuller and Jensen (2002) argued that from the CEO’s perspective, over-valuation is potentially more dangerous and difficult to address than under-valuation. What’s a CEO to do? Should she/he try to let the air out of the company’s stock or continue to try to meet analysts’ increasingly more unrealistic expectations?2 Fuller and Jensen go so far as to suggest that over-valuation can arise naturally out of the dynamics of the interplay between the firm’s executives and Wall Street analysts, i.e. CEOs are in a difficult bind with Wall Street. Managers up and down the hierarchy work hard at putting together plans and budgets for the next year and quite often when those plans are completed top management discovers that the results fall far below what Wall Street expects. CEOs and CFOs are therefore left in a difficult situation. They can stretch to try to meet Wall Street’s expectations or prepare to be punished if they fail. All too often top managers react to the situation by encouraging or mandating middle and lower level managers to redo their forecasts, plans and budgets to get them in line with external expectations. In some cases, fearing the results of missing the Street’s expectations, managers start the budgeting process with the consensus expectations and mandate that internal budgets and plans be set so as to meet them. Either way this sets the firm and its managers up for failure if external expectations are, in fact, impossible for the firm to meet (p. 1). The Fuller and Jensen (2002) thesis seems to suggest that corporate CEOs are pawns of the Wall Street analyst community.3 However, this begs the question, “Why are CEOs so focused on stock price?” The answer may be as simple as, “That’s what they’re paid to do.” At worst, recent history shows that share-price obsession has driven some executives to improperly, and perhaps illegally, misrepresent their companies’ financial condition to investors, who later lost many millions when the true picture was revealed. But even at companies that continued to prosper, executives report

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

129

that the general fixation on daily and weekly stock-price movements drives them to manage for the short term, sacrificing the long-term interest of shareholders and employees. There was nothing inevitable about this development – it didn’t use to be that way, executives say, and in other countries it still isn’t. But over the past 20 years, as American capitalism has morphed into shareholder capitalism, virtually all the incentives and disincentives that have been brought to bear on the executive suite are linked directly or indirectly to short-term movements in the company stock price (Pearlstein, 2002, p. H01). We hypothesize that Enron’s share price became increasingly detached from underlying fundamentals in the years just prior to its failure. Furthermore, we conjecture that management’s attempts to manage reported earnings was an important factor that drove a wedge between observed market prices and the intrinsic value of the firm’s shares. The paper is organized as follows: In Section 2 we address the challenge that over-valuation poses to the efficient market hypothesis and investigate the nature of market expectations that would justify the valuations placed on Enron throughout the nineties. During this period the stock sold for more than $80 a share and had a price-earnings multiples as high as 67.4 To carry out this analysis we use a discounted cash flow model patterned after Miller and Modigliani (1961) to show that the underlying growth rates implied by Enron’s lofty price-earnings ratio were extremely high when compared to the historical growth rates realized by other firms. However, since growth expectations and required returns are unobservable, this analysis is more suggestive than conclusive. Section 3 presents our investigation into whether Enron managed its earnings in order to meet the expectations of equity analysts. Powers et al. (2002) (hereafter the Powers Report)5 documents Enron’s use of special purpose entities to hedge its losses from some of its investments, and the financial press has alleged that the firm engaged in elaborate swap agreements that were also intended to manage reported earnings.6 This prompts us to look for evidence of earnings management more generally. We compare quarterly earnings performance with analysts’ consensus forecasts and find that Enron was successful in meeting or beating analysts’ expectations in 29 of 44 quarters, or 77% of the time, during the last decade. Finally, Section 4 contains some concluding remarks and suggestions for future research.

2. WAS ENRON’S STOCK OVERVALUED? In his 1986 presidential address to the American Finance Association, Fisher Black (1986) suggested that U.S. equity markets might be efficient within a factor of say

130

JOHN D. MARTIN AND AKIN SAYRAK

two. By this he meant that prices were not more than twice their intrinsic value, nor were they less than half. Although the number two is arbitrary, Black’s description of the pricing process highlights the fact that even well functioning capital markets exhibit noisy pricing. Furthermore, a noisy price mechanism opens the possibility that market prices may differ from intrinsic values (value based on the facts) for extended periods of time with some firms being under-valued while others are over-valued.7

2.1. Behavioral Finance, Overvaluation, and the Efficient Market Hypothesis The very idea that a firm’s shares can be over- and under-valued flies in the face of the efficient market hypothesis (EMH), which has been a central tenet of finance for more than three decades. So it behooves us to take some time to explain how this might come about in a reasonably well functioning capital market. Shleifer (2000) suggests that the theoretical case for the EMH rests on three progressively weaker assumptions. The first is that investors are rational and that they value securities in a rational way. Second, to the extent that some investors might not be rational (sometimes referred to as noise traders), their trades are random and therefore tend to cancel each other out without having a material impact on prices. Finally, to the extent that investors are irrational in similar ways, they are met in the market by rational arbitrageurs who work to counter their influence on prices. The developing field of behavioral finance has documented a number of theoretical and empirical challenges to the fundamental assumptions underlying the EMH.8 Of particular interest to this paper is the notion that investors are not rational in the way they formulate expectations. Specifically, individuals systematically violate Bayes rule and other maxims of probability theory when formulating their expectations concerning future uncertain outcomes. They tend to predict an uncertain future by taking a short history of data and asking what broader picture this data reflects.9 This type of behavior can generate a mistaken guide to the future if the pattern being observed is simply a random occurrence. For example, investors and even analysts may erroneously extrapolate a brief history of rapid growth in earnings far into the future, thus resulting in overvaluation of the firm’s stock. The second assumption underlying the EMH may place boundaries on the preceding type of irrationality. That is, if the tendency to extrapolate brief periods of high earnings growth into the distant future does not result in herding of large numbers of investors, then the effects on stock prices may be mitigated. However, if equity analysts engage in this form of irrational behavior and, as Fuller and Jensen (2002) suggest, company executives comply by trying to meet

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

131

these expectations (if only for a brief period of time), then investors may become convinced that the firm’s short-term growth foretells better long-term prospects and consequently bid the firm’s shares up to unrealistic levels. The third and final line of defense against over- and under-valuation of company shares lies in the actions of arbitrageurs. If there are a sufficient number of rational investors who are willing to stand against the market herd, rationality of pricing will prevail. The argument is that individuals who are not subject to psychological flaws in their behavior will stem the pressures of noise traders to overreact and drive share prices above or below their intrinsic values. Shleifer (2000) argues that the principal limitation of arbitrage lies in the fact that arbitrage is itself risky where there are no perfect substitutes in the market which can be used to formulate the arbitrage position.10 That is, to lay off their risks, arbitrageurs who sell or short over-priced securities must be able to buy essentially the same security that is not over-priced. In some instances, for example publicly traded derivatives, this is possible. However, where close substitutes do not exist, and imperfect substitutes must be used, arbitrage becomes risky.

2.2. Evidence of Over-valuation Can market and firm fundamentals explain the run-up and decline of Enron’s share price and corresponding price-earnings ratio? Enron’s price-earnings ratio rose from 18.33 times earnings to 67 times from June 1996 through May 2001 while industry competitors realized an increase in price-earnings from 17.60 to only 22.55.11 Figure 1 illustrates the relatively high valuation accorded Enron’s shares when compared to a set of competitors for the period beginning with June 1998 and ending with May 2001. With the exception of August 1998 Enron’s P/E ratio was always higher than the average of the competitor firms. Furthermore, with the exception of the period from June 1998 through November of 1998, Enron’s P/E ratio was higher than the maximum of any of the competitor firms up until November 2001, after which time the firm declared bankruptcy. The equity market clearly accorded Enron a premium share price. The comparable companies used in the preceding analysis are from the energy industry; however, it can be argued that Enron was in process of being transformed from an energy firm to Wall Street type trading company and as such, it should have been valued accordingly. However, an analysis of the P/E ratios for Wall Street trading firms reveals that they were even lower than the energy companies used here. For example, the median P/E ratios for Bear Stearns, Goldman Sachs, Lehman, and Merrill Lynch for year-end 1998–2001 were 8.6, 11.9, 13.6, and 18.6, respectively.

132

JOHN D. MARTIN AND AKIN SAYRAK

Fig. 1. Enron Valuation Ratios. Enron Price/Earnings, Price/Cash Flow, and Price/Book Relative to the Mean for a Comparison Group of Competitor Firms Including Duke Power, Dynegy, Entergy, Reliant, and the Southern Company. The Data Include all Months Beginning with June 1998 and Ending with May 2001.

2.2.1. Stock Prices and Supernormal Growth Expectations The model we use to investigate the role of investor expectations regarding growth rates in earnings and the investor’s required rate of return on equity was developed by Miller and Modigliani (1961).12 They show that for a firm that has opportunities to earn supernormal returns, r∗ , over a finite period of T years, P/E ratio can be expressed as follows:

   1 1 + kr ∗ T k(r ∗ − r) P/E = 1− (1) 1+ r r − kr ∗ 1+r where k is the fraction of firm earnings invested in supernormal return projects and r is the discount rate. To study the market valuations of dot.com stocks, Ofek and Richardson (2001) assume that k = 1 for T years, and that after T years the firm’s P/E returns to the levels of old economy firms, P/E old . Under these conditions (1) reduces to   1 + r∗ T P/E old (2) P/E = 1+r The primary appeal of Eq. (2) is its simplicity. For example, given the term of the supernormal investment opportunities, T, and an assumption about the firm’s long

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

133

Table 1. Implied Excess Returns and Supernormal Growth Rates. Growth Period in Years (T )

Terminal P/E levels 10

15

Panel A. Implied excess returns on capital {[(1 + 10 25.89% 20.89% 15 16.59% 13.48% 20 12.20% 9.95% 25 9.65% 7.88% 30 7.98% 6.53%

20

25

30

r∗ )/(1

+ r)] − 1}, where the current P/E ratio is 100 17.46% 14.87% 14.87% 11.33% 9.68% 9.68% 8.38% 7.18% 7.18% 6.65% 5.70% 5.70% 5.51% 4.73% 4.73%

Panel B. Implied growth rates, g, assuming a weighted average cost of capital, r, of 20% 10 51.1% 45.1% 41.0% 37.8% 37.8% 15 39.9% 36.2% 33.6% 31.6% 31.6% 20 34.6% 31.9% 30.1% 28.6% 28.6% 25 31.6% 29.5% 28.0% 26.8% 26.8% 30 29.6% 27.8% 26.6% 25.7% 25.7%

term P/E (i.e. that it will revert to that of the market as a whole as represented by P/E old ), we can estimate how large the firm’s supernormal investment returns, r∗ , have to be relative to required returns, r, in order to justify current P/E levels.13 That is,     1 + r∗ P/E 1/T (3) = 1+r P/E old Thus, to justify a P/E ratio of 100 where P/E old is equal to 10 and the anticipated period of supernormal returns is 10 years, (1 + r ∗ ) must be 125.89% of (1 + r) implying an excess return on capital of 25.89%. Furthermore, if r is 20% this implies that the supernormal rate of return on invested capital, r∗ , must be equal to 51.1% per year over a 10 year period of supernormal growth.14 Panel A of Table 1 contains implied excess returns on capital for various combinations of supernormal growth periods and terminal P/E multiples (P/E old ). For example, if the terminal price-earnings ratio (P/E old ) is 20 and the growth period (T) is 10 years, the implied excess return on capital is 17.46%. Panel B converts the implied excess return on capital calculations from Panel A into estimates of supernormal growth rates where the discount rate (r) is assumed to equal 20%. Using the preceding example (i.e. P/E old = 20 and T = 10) this produces an estimate of r∗ equal to 41.1%. Note that the supernormal growth rate required to support a particular P/E ratio declines for longer supernormal growth periods, and for bigger terminal P/E multiples.

134

JOHN D. MARTIN AND AKIN SAYRAK

Although the assumption that the reinvestment rate was 100% (i.e. k = 1) suited Ofek and Richardson’s (2001) analysis of internet stocks very well, a more general analysis must allow firms with supernormal growth opportunities to pay some cash to stockholders such that k < 1. Where k < 1 the relationship between observed and terminal P/E ratios is more complex than Eq. (2), but is still tractable, i.e.15

     1 + kr ∗ T 1 + kr ∗ T (1 − k)(1 + kr ∗ ) 1 − + P/E = P/E old (4) r − kr ∗ 1+r 1+r Equation (4) demonstrates that the P/E ratio for a firm facing a finite period of supernormal growth is a weighted average of the P/E ratio for a firm with an infinite growth period, P/E =

(1 − k)(1 + kr ∗ ) r − kr ∗

and the P/E ratio corresponding to the end of the super normal growth period, i.e. P/E old . The weight attached to the P/E old is [(1 + kr ∗ )/(1 + r)]T and one minus this quantity is the weight attached to the P/E ratio for the firm with an infinite growth period.16 Table 2 contains calculations of the P/E ratio using Eq. (4) for a firm that faces a finite period of supernormal growth ranging from 5 to 20 years and that enjoys a return premium over its discount rate ranging from 6.2% up to 24.2%. The remaining parameters used to solve Eq. (4) correspond roughly to Enron’s position in 2000–2001. Note that the assumed retention rate (k) of 60% and discount rate (r) of 15% implies a growth rate of 6.2%, and for an infinite period of supernormal growth this produces a P/E multiple of 20 (the terminal P/E ratio). Consequently, for the 6.2% growth rate the P/E multiples are the same for all super normal growth periods. Table 2 shows, for example, that to justify a P/E ratio of 60.83 (given the other parameters in the model) the firm must have a super normal growth period of 15 years, and earn a return premium (r ∗ − r) of 22.2%. This return premium implies a return on reinvested funds, r∗ , of 22.2% + 15% = 37.2% (given our assumed 15% discount rate) and an annual growth rate in earnings, kr∗ , of 0.60 × 37.2% = 22.3%. Achieving a P/E multiple of at least 60 times for shorter periods of super normal growth require even more lofty expectations. For example, to support a P/E ratio as high as 60 where the period of super normal growth is only 10 years where the retention ratio is 0.60 (not shown in Table 2) requires a return premium of 29.58% which corresponds to an earnings growth rate of 26.7% a year. Furthermore, to justify a P/E of 60 with a supernormal growth period of only five years requires a return premium of 53.93% and an annual rate of growth in earnings

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

135

Table 2. P/E Ratios for Firms Facing a Finite Period of Super Normal Growth. Growth Rate (g = kr∗ ) 12.7% 13.9% 15.1% 16.3% 17.5% 18.7% 19.9% 21.1% 22.3% 23.5%

Excess Returns (r∗ − r)

5

Growth Period (T) 10

15

20

6.2% 8.2% 10.2% 12.2% 14.2% 16.2% 18.2% 20.2% 22.2% 24.2%

20.00 21.04 22.13 23.27 24.45 25.68 26.96 28.29 29.67 31.11

20.00 22.04 24.28 26.73 29.41 32.35 35.55 39.05 42.86 47.01

20.00 23.00 26.44 30.40 34.95 40.17 46.15 53.00 60.83 69.77

20.00 23.91 28.62 34.30 41.13 49.36 59.24 71.10 85.31 102.33

Note: Assumptions: Retention ratio (k) = 60.0%; Discount rate (r) = 15.0%; Terminal P/E ratio = 20; Implied Terminal g = 12.75%; Implied Terminal r ∗ = 21.2%; Implied Terminal (r ∗ − r) = 6.2%. At the time of Enron’s peak market capitalization its stock price ratio was over 60 times earnings and the firm was retaining and reinvesting about 60% of its earnings. In the P/E ratio calculations found below we assume a required rate of return of 15%. The P/E ratios correspond to firms that face a finite period of supernormal growth (T) during which time they earn excess returns (r ∗ − r). After this period the firm’s stock is assumed to be priced at a normal P/E multiple for a firm that is not earning a supernormal rate of growth in earnings (the Terminal P/E ratio is assumed to be 20 for purposes of these calculations). P/E ratios in excess of 60 are shaded to highlight the combinations of excess returns and growth periods that would justify a P/E of 60 or higher.

of 41.4% (also not reported in Table 2). Clearly, in the context of Eq. (4), justifying a P/E for Enron of 60 or more suggests that investors are anticipating a lengthy period of supernormal growth and a very high return premium over Enron’s cost of capital. 2.2.2. Evolving Growth Expectations and the Valuation of Enron Shares To this point we have only considered hypothetical cases, albeit situations that closely mimicked the pricing of Enron’s shares at their peak. Let’s now turn to Enron’s actual situation over the period June 1998 through September 2001 to see the implications of the market’s pricing of Enron shares for the firm’s anticipated rate of growth and how these expectations evolved over the period. Table 3 contains a time series analysis of Enron’s P/E ratios for the period beginning with the second quarter of 1998 and ending with the third quarter of 2001. A comparison of columns two and three reveals that Enron’s price-earnings ratio exceeded that of its competitors and the margin grew dramatically over the

136

Table 3. Time Series Analysis of Enron’s Price-Earning Ratio (June 1998–June 2001). Quarter

Terminal Enron P/E Ratio

25.622 23.265 25.056 30.184 34.063 34.883 31.031 50.879 47.354 54.461 55.337 55.471 44.473

22.333 20.819 17.280 15.208 17.271 17.544 17.391 20.652 22.294 23.911 21.961 20.145 19.663

Enron Beta

0.535 0.730 0.810 0.805 0.857 0.902 0.835 0.444 0.472 0.447 0.552 0.660 0.741

Enron Required Return (r) (%)

Retention Ratio (k) (%)

5.94 5.61 5.38 5.66 6.22 6.47 6.59 6.46 6.42 6.06 5.81 5.56 5.87

9.68 10.72 11.05 11.30 12.22 12.78 12.43 9.57 9.72 9.19 9.67 10.18 11.05

55 55 55 63 63 63 63 59 59 59 59 60 60

Super-normal Growth Period (T ) 5

10

15

Implied Growth Rates 10.9 11.2 17.3 25.3 26.5 27.3 24.1 29.4 25.8 27.1 30.2 33.1 28.8

9.4 9.9 13.0 17.1 18.2 18.9 17.2 18.3 16.8 17.1 18.8 20.4 18.8

8.9 9.5 11.7 14.5 15.6 16.3 15.0 14.8 13.9 14.0 15.2 16.4 15.6

20 (kr∗ )

(%) 8.6 9.3 11.0 13.2 14.3 15.0 13.9 13.1 12.5 12.5 13.5 14.5 14.1

Note: Excess returns and corresponding super-normal growth rates are calculated for 5, 10, 15, and 20 year growth periods using the average P/E ratio for Enron over the quarter, a terminal P/E ratio equal to the average for Enron competitors (Duke Energy, Dynegy, Entergy, and the Southern Company). Enron’s required rate of return is calculated using the capital asset pricing model based on the contemporaneous yield on the 20-year U.S. Treasury Bond, and a market risk premium of 7%.

JOHN D. MARTIN AND AKIN SAYRAK

Jun-98 Sep-98 Dec-98 Mar-99 Jun-99 Sep-99 Dec-99 Mar-00 Jun-00 Sep-00 Dec-00 Mar-01 Jun-01

P/E Ratio

20-year T-Bond Yield (%)

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

137

period.17 In fact, Enron’s price-earnings ratio was a factor of two or more times the industry price earnings ratio by the end of the first quarter of 1999. Clearly, the investment community changed its perception of Enron over this period to reflect a much higher set of growth expectations than those of its competitor firms. To calculate the anticipated growth rate in earnings inferred by Enron’s price-earnings ratio using Eq. (4) we must first estimate the remaining parameters of the model. This includes the retention ratio, k; equity discount rate, r; the length of the period of supernormal growth, T; and the P/E ratio that will be used to value the firm’s shares at the end of the supernormal growth period, P/E old . We estimate k as one minus the ratio of dividends to earnings for the year. The required equity return, r, is estimated using the capital asset pricing model where the contemporaneous yield on the 20 year Treasury Bond is used to proxy the risk free rate, Enron’s beta coefficient is taken from the COMPUSTAT database, and the market risk premium is estimated to be 7%. To estimate the terminal P/E ratio for each quarter we average the corresponding ratios for each of Enron’s principal competitors in the energy business including Duke Power, Dynegy, Entergy, Reliant Energy, and the Southern Company. Finally, we consider supernormal growth periods of 5, 10, 15, and 20 years.18 Our estimates of the equity discount rate using the capital asset pricing model appear to be quite low, ranging from 9.19% in September 2000 to 12.78% in September 1999. These low figures are a direct result of the beta coefficient estimates that range from 0.444 to 0.810 over the period. Note, however, that should the costs of capital estimates be too low, the impact of the error is to reduce the implied supernormal growth rate estimates. That is, if the discount rates should be underestimated, correcting the estimates would have the effect of making the implied growth rates even higher than the ones we report in Table 3. The implied growth rates found in the last four columns of Table 3 suggest that investor expectations increased dramatically over the period we analyze, peaking in the first quarter of 2001, when Enron’s P/E ratio was just over 55. This lofty P/E implied a supernormal growth rate of 33.1% for a five-year growth period and 14.5% for a 20-year growth horizon. In fact, the implied supernormal growth rates for the five-year growth period hovered near 30% for the entire period bracketed by March 2000 through June 2001. 2.2.3. Were Enron’s Implied Growth Rates Reasonable? How reasonable is it to assume that Enron’s earnings could grow at the rates required to justify a P/E multiple of 60? Specifically, what are the prospects that Enron could grow its earnings by a rate of 26.7% per year over a period of 10 years, or 41.4% over five years? Chan et al. (2001) discuss what other firms have been

138

JOHN D. MARTIN AND AKIN SAYRAK

able to accomplish, and thus provide evidence regarding the likelihood that Enron might be able to sustain these rates of growth. They find that for the five-year period ending in 1998, the top 2% of all firms on the COMPUSTAT database achieved an average rate of growth in earnings of 57.4%. Similarly, the top 2% of all firms achieved an average growth rate in earnings of 36.4% over the 10 years ended in 1998. Further evidence regarding the growth experience of U.S. firms suggests that only 2.6% realize a rate of growth in earnings that is above the median for their industry over the five-year period ended in 1998. This drops to 0.0% when the period is extended to 10 years. Since these results correspond to all firms, it is possible that the experiences of technology or glamour stocks are different. Looking at these two groups of stocks, Chan et al. find that 4.1% and 2.9% achieve above median growth rates in earnings for the technology and glamour groups, respectively. The proportion achieving 10 years of above median annual growth in earnings drops to 0.4% and 0.1%, respectively, for technology and glamour stocks. It appears that using the general population of stocks as the universe, the likelihood of achieving the very high growth rates required to support Enron’s lofty P/E is extremely unlikely. However, Enron had exhibited a very high rate of growth in the past. Perhaps successful firms exhibit persistence in earnings not exhibited by firms in the general population. Chan et al. (2001) tests this proposition by looking at the fraction of firms that had above-median growth in earnings for the past five years and above median growth for the subsequent five year period. They find that 5.1% of the firms were successful in repeating above median performance over the subsequent five years.

3. DID ENRON MANAGE EARNINGS? The internal report of Enron’s board of directors (i.e. the Powers’ Report, 2002) alleges that Enron’s management engaged in a number of activities designed to buffer the firm’s reported earnings from the effects of adverse performance of some of its more significant merchant investments.19 In addition, Enron’s management may well have been engaged in the process of managing its earnings so as to meet or beat equity analyst forecasts for an extended period of its history.20 Although it is difficult to establish with certainty whether earnings management was being practiced, we offer several pieces of evidence that indicate that Enron was very adept at not only meeting, but also beating analysts’ earnings expectations. Panel A of Fig. 2 contains a plot of the difference in Enron’s actual quarterly earnings per share and the consensus forecast of earnings for the period beginning with the first calendar quarter of 1990 and ending with the last calendar quarter of 2000. During this 44 quarter period Enron met or exceeded analysts’ consensus

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

Fig. 2. Accuracy of Analysts’ Earnings Projections.

139

140

JOHN D. MARTIN AND AKIN SAYRAK

earnings forecasts 29 times, met them five times, and fell below the forecast only 10 times. If we concentrate on the quarters in which Enron met or exceeded expectations, we find that the firm met or exceeded the consensus earnings forecast 77% of the time, which is statistically different from 50% at one percent significance level. In the preceding analysis we assumed that the analysts’ consensus forecast was an unbiased estimate of actual quarterly earnings. However, prior research indicates that analysts’ tend to overestimate earnings (O’Brien, 1988).21 This suggests that Enron’s record over 1990–2000 is even more remarkable. In Panel B, we provide the results of a similar set of tests for five of Enron’s energy industry competitors. In two of the five cases (Dynegy and Entergy) we find evidence of a disproportionately large number of quarters in which reported earnings met or exceeded the analyst consensus forecast. Note, however, that Enron’s earnings exceeded the consensus forecast 29 quarters and fell below them only 10 times. This compares to 11 above and 8 below for Dynegy and 28 above and 12 below for Entergy. It appears that when compared to its competitors, Enron was particularly adept at beating analysts’ forecasts of its earnings except for Entergy, perhaps, which was relatively close. Still another piece of evidence that suggests Enron was actively engaged in managing its earnings relates to the way that the firm used asset sales to boost reported profits. In a footnote to its 1999 annual report Enron notes that it booked “pretax gains from sales of merchant assets and investments totaling $756 million, $628 million, and $136 million” in 1999, 1998, and 1997. Profits from these asset sales constituted 56%, 61%, and 83% of Enron’s pre-tax profits in the respective years. Clearly, the ability to sell merchant investments and recognize imbedded gains as profits was critical to Enron’s success in meeting analyst earnings forecasts. UBS Warburg analyst Ron Barone said, as much when he noted, “This is an enormous earnings vehicle, which can often be called upon when and if market conditions require.”22 The fact that Enron relied on asset sales to produce a significant portion of its earnings is not evidence of earnings management, in and of itself. However, if those asset sales were not the result of arms length agreements, then we have evidence of earnings management of the worst kind. At the time of this writing, the courts have not yet determined whether fraud was involved, however, it has been alleged. For example, Qwest and Enron are accused of having engaged in a swap of fiber optic network capacity and services at exaggerated prices in an effort to improve their earnings. Details of the transaction revealed in Enron bankruptcy filings indicate that the two companies raced to complete a transaction valued at more than $500 million that analysts suggest would be hard to justify because of the glut of fiber optic capacity.23 According to published reports the deal called for Qwest to overpay for the assets in exchange for services for which Enron would

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

141

overpay. The net result was that Enron could avoid recording a loss by selling an asset whose value had dropped in the market.24 Some Enron executives have been quoted in the financial press as saying that the company manipulated its prudence reserves set aside to cover potential losses from its energy trading reserves to keep the company from having to report very sizeable profits in 2000 and 2001.25 It has been alleged in the financial press that the reserves helped Enron hide the substantial trading profits it was making as a result of the California energy crisis so as to reduce their political ramifications, and also provided the company with a way to smooth out its reported earnings growth. Alternatively, the increased reserves could have been a realistic response to the fact that the economic value of energy trades – particularly deals that provide for the delivery of electric power over an extended contract period – is highly subjective. As a consequence, increasing reserves to offset future losses related to some of those trades and the associated legal costs seems very reasonable. For example, when Pacific Gas and Electric sought bankruptcy protection in April 2001 it owed Enron $500 million. Earnings management for its own sake is inconsistent with generally accepted accounting practice. However, the accounting rules explicitly allow firms to set aside reserves for a wide variety of transactions. The subjectivity of the potential losses that provide the basis for these reserves, however, means that they offer the potential for abuse. Was Enron using reserves and transactions with related parties to either create or hide profits in an effort to manage a smooth growth pattern in earnings? The circumstantial evidence we have presented here would certainly lead a suspicious person to believe so. Although we cannot say with certainty that Enron’s management was deliberately managing its reported earnings, we can say that the firm’s record of matching and exceeding analyst earnings projections is remarkable. Furthermore, the firm evidently used special purpose entities (some of which involved related parties) to mark sufficient profits to meet analysts’ forecasts during the 1997–2000 period. Finally, where trading profits grew to outlandish levels during the California energy crisis it appears that Enron used reserves set aside for potential losses to offset as much as half their trading profits.

4. CONCLUDING REMARKS This paper investigates whether market fundamentals can explain the run-up and collapse of Enron’s stock price. However, we cannot expect to arrive at a definitive response to the question as to whether Enron’s stock overvalued. Any time series of prices can be justified by an appropriate set of assumptions about investor

142

JOHN D. MARTIN AND AKIN SAYRAK

expectations. However, we are able to show that the growth rates implied by the Enron stock’s market valuation have rarely been achieved in recorded business history. We provide evidence that Enron was actively engaged in the management of its reported earnings, and they were very good at it. Between 1990 and 2000 Enron’s reported earnings met or exceeded analysts’ earnings forecasts 77% of the time. This latter statistic is particularly impressive in light of recent studies that have documented the optimism that characterizes equity analysts’ earnings forecasts. Furthermore, beginning in 1997 Enron used asset sales (often to related parties) to generate as much as 83% of its annual earnings. Although this evidence is not conclusive with respect to the motives of Enron’s management, they do suggest that Enron was very concerned about meeting analysts’ earnings forecast. Our results raise very serious questions regarding the financial reporting process as well as the efficient pricing of traded shares of common stock. These questions are critical to the study and practice of corporate finance where market prices of securities are presumed to provide unbiased estimates of underlying fundamentals. If investors rely on reported earnings as a signal of these fundamentals, and reported earnings is subject to manipulation, then we have good reason to distrust market prices. Consequently, decisions that the firm makes that rely on information derived from market prices such as the firm’s opportunity cost of capital, may be seriously compromised, and the resource allocation decisions based on the capital costs inferred from these prices may be flawed.

NOTES 1. Both Rangan (1998) and Teoh et al. (1998) provide evidence consistent with the notion that investors can be fooled (at least temporarily) by earnings management. They find that investors naively extrapolate the pre-offering earnings of firms that engage in a seasoned equity offering without fully adjusting for potential manipulation of reported earnings. This results in a temporary over-valuation of the issuing firm’s shares followed by an adjustment to reflect the declines in earnings caused by earnings management. 2. See Jenkins (2002). 3. Bratton (2002) suggests that Enron’s management may have simply asked too much of its strategy in terms of immediate increases in earnings per share, and suggests that the firm’s management may have mistaken its own inflated stock valuation for fundamental value. 4. Behavioral finance scholars now contend that equity prices can and do deviate from intrinsic values in both a positive and negative direction for extended periods of time. See Shleifer (2000) for an excellent review of this research. 5. This is the internal report prepared by three members of Enron’s board of directors. 6. Barboza and Feder (2002) report that Enron and Qwest engaged in “a deal last fall to swap fiber optic network capacity and services at exaggerated prices in an effort to improve

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

143

each company’s financial picture . . ..” The authors quoted an Enron executive who described the transaction as follows: “Qwest said we will overpay for the assets, and you will overpay me on the contract.” 7. Lamont and Thaler (2001) study a number of cases in which apparent market pricing errors persist for extended periods of time. They conclude that prices can and do deviate from intrinsic value, but that these apparent pricing errors do not present exploitable arbitrage opportunities because of the costs of arbitrage involving short sales in the particular instances they study. 8. Shleifer (2000) provides an excellent overview of this literature. 9. See Kahneman and Tversky (1973). 10. John Maynard Keynes put forth another argument for the riskiness of arbitrage when he suggested that “Markets can remain irrational longer than you can remain liquid.” 11. Energy firms included here as competitors in this analysis include Duke, Dynegy, Entergy, Reliant, Mirant, and the Southern Company. 12. French and Poterba (1991) used this model to investigate the level of Japanese equity prices in the 1980s and Ofek and Richardson (2001) used it to study the rise and fall of dot.com prices during the period 1998–2000. 13. Note that the growth rate in firm earnings is equal to kr∗ such that where the proportion of firm earnings that are reinvested, k, equals 100%, then g = r ∗ and Eq. (2) can equivalently be written as follows:   1+g T P/E old P/E = 1+r 14. That is, (1 + r ∗ )/(1 + 0.20) = 1.2580 such that r ∗ = 0.511 or 51.1%. 15. Miller and Modigliani (1961) provide the following approximation for the P/E ratio where k < 1, T is not infinite, and (1 + kr ∗ ) and (1 + r) are both close to unity: P/E ∼ = (1/r) + k((r ∗ − r)/r)T. This approximation provides a straightforward interpretation. The P/E ratio for a growing firm is equal to the sum of the P/E of a firm with no growth (i.e., 1/r) plus an increment reflecting the firm’s special growth opportunities. This latter term is an increasing function of the fraction of the firm’s earnings that it retains, k, the relative excess return it earns over the required return, (r ∗ − r)/r, and the duration of the period of growth opportunities, T. 16. In the limit where T → ∞ and kr ∗ < r Eq. (4) reduces to the familiar form, P/E = ((1 − k)(1 + kr ∗ ))/r − kr ∗ . 17. We use Duke Energy, Dynegy, Entergy, and the Southern Company as the basis for industry comparisons. 18. Sharpe (2002) estimates that the stock market utilizes analyst’s long-term growth prospects out an average of 8–9 years. Thus, our analysis of 5–20 year periods of supernormal growth appears to be sufficient to capture the determinants of observed market prices. 19. Enron followed a practice of accounting for the results of its merchant investments using mark-to-market accounting procedures whereby it recorded as income changes in the value of these investments that occurred during the period. Thus, while the values of these investments was rising Enron got to record these increases in value of its investments as earnings for the period. When the values of some significant investments were threatened, Enron’s management engaged in a series of transactions with Special Purpose Entities that it created to hedge these accounting risks. For a discussion of the specifics of the transactions see Powers (2001).

144

JOHN D. MARTIN AND AKIN SAYRAK

20. Schipper (1989) defines earnings management as follows: “The purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain (as opposed to say, merely facilitating the neutral operation of the process).” 21. All equity analysts are not equally optimistic, however. Dugar and Nathan (1995) find that the financial analysts of brokerage firms that provide investment banking services to a company are more optimistic, relative to other analysts in terms of their earnings forecasts and investment recommendations. Lin and McNichols (1998) find no difference in the earnings forecasts of the affiliated and unaffiliated analysts, but do find that the longterm growth forecasts and investment recommendations of the affiliated analysts are more optimistic. 22. As reported in Bethany McLean, Is Enron overpriced? Fortune Magazine (March 5, 2001). 23. See David Barboza and Barnaby J. Feder, Enron’s swap with Qwest is questioned, The New York Times (March 29, 2002). 24. The transaction called for Qwest to pay Enron $308 million for fiber optic assets (known as dark fiber since these idle network strands would require added investments before they could be put into service). Enron, in exchange, agreed to pay Qwest $195.5 million for active optic cable services over a 25 year period. In addition, each company exchanged checks for about $112 million around the close of the deal (see Barboza & Feder, 2002). 25. David Barboza, Former officials say Enron hid gains during the crisis in California, The New York Times (June 23, 2002).

REFERENCES Barboza, D., & Feder, B. J. (2002). Enron’s swap with Qwest is questioned. New York Times (March 29). Black, F. (1986). Noise. Journal of Finance, 41, 529–543. Bratton, W. (2002). Enron and the dark side of shareholder value. Tulane Law Review (forthcoming). Chan, L. K. C., Karceski, J., & Lakonishok, J. (2001). The level and persistence of growth rates, National Bureau of Economic Research, Working Paper 8282 (May). Dugar, A., & Nathan, S. (1995). The effect of investment banking relationships on financial analysts’ earnings forecasts and investment recommendations. Contemporary Accounting Research, 131–160. French, K. R., & Poterba, J. M. (1991). Were Japanese stock prices too high? Journal of Financial Economics, 29, 337–363. Fuller, J., & Jensen, M. C. (2002). Just say no to Wall Street: Courageous CEOs are putting a stop to the earnings game and we’ll all be better off for it, negotiation, organization and markets unit, Harvard Business School, Working Paper No. 02-01, forthcoming in the Journal of Applied Corporate Finance, Spring 2002. Jenkins, H. W., Jr. (2002). Business world: Enron for beginners. Wall Street Journal (January 23). New York, NY. Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80, 52–65.

Did Earnings Management Contribute to the Overvaluation of Enron’s Stock?

145

Lamont, O. A., & Thaler, R. H. (2001). Can the market add and subtract? Mispricing in tech stock carve-outs, The Center for Research in Security Pries Working Paper No. 528, University of Chicago (May). Lin, H.-W., & McNichols, M. F. (1998). Underwriting relationships, analysts’ earnings forecasts and investment recommendations. Journal of Accounting and Economics, 25, 101–127. Miller, M., & Modigliani, F. (1961). Dividend policy, growth, and the valuation of shares. Journal of Business, 34, 411–433. O’Brien, P. C. (1988). Analysts’ forecast as earnings forecast. Journal of Accounting and Economics, 10, 53–83. Ofek, E., & Richardson, M. (2001). Dotcom mania: The rise and fall of Internet stock prices, NBER Working Paper 8630 (December). Pearlstein, S. (2002). Executive privilege? Here’s the new take on stock options: They reward corporate leaders for all the wrong reasons. Washington Post (March 24), H01. Powers, W. C., Jr., Troub, R., & Winokur H. S., Jr. (2002). Report of investigation by the Special Investigative Committee of the Board of Directors of Enron Corp. (February 1). Rangan, S. (1998). Earnings management and the performance of seasoned equity offerings. Journal of Financial Economics, 50, 101–122. Schipper, K. (1989). Commentary on earnings management. Accounting Horizons, 3, 91–102. Sharpe, S. A. (2002). How does the market interpret analysts’ long-term growth forecasts? Unpublished paper, Division of Research and Statistics, Federal Reserve Board, Washington, DC (January 12). Shleifer, A. (2000). Inefficient markets: An introduction to behavioral finance. Oxford, UK: Oxford University Press. Teoh, S. H., Welch, I., & Wong, T. J. (1998). Earnings management and the underperformance of seasoned equity offerings. Journal of Financial Economics, 50, 63–99.

OPERATIONAL RISK IN FINANCIAL SERVICE PROVIDERS AND THE PROPOSED BASEL CAPITAL ACCORD: AN OVERVIEW Jeffry M. Netter and Annette B. Poulsen ABSTRACT The 1988 Basel Accord and the proposed revisions to the Accord represent some of the most significant international regulations impacting the financial decisions of firms, in this case, financial services firms, in recent years. The revisions to the Accord incorporate operational risk into the capital, supervisory and market requirements. In our review of the issues in this area, we provide insight into the workings of an important international regulation. We also present suggestions for further research in this area that will become feasible when data on the impact of the new regulations become available after the proposed implementation in 2006.

1. OVERVIEW One of the key developments in recent managerial practices is the recognition of the importance of enterprise-wide risk management in the strategic analysis of corporate activities. Firms are recognizing the need to more fully understand the risks of the firm’s various actions, their interrelations within the firm and their

Corporate Governance and Finance Advances in Financial Economics, Volume 8, 147–171 © 2003 Published by Elsevier Science Ltd. ISSN: 1569-3732/PII: S1569373203080071

147

148

JEFFRY M. NETTER AND ANNETTE B. POULSEN

connectedness to the rest of the economy. While traditional risk management emphasizes insurance products for hazard risks and, to some degree, financial hedging for market risks, enterprise-wide (also called holistic or integrated) risk management demands that managers reconceptualize risk within the firm, identifying, measuring and managing risks as disparate as accounting fraud, soft consumer demand, and environmental requirements. These non-traditional risks are frequently labeled “operational risk.” Partly because of its highly regulated nature as well as its dependence on the capital markets with their well-defined prices for products, this holistic approach to risk management has been more fully analyzed and implemented in the financial services industry as compared to any other. Financial firms have for years actively measured and managed credit risk, related to the probability of defaults, and market risk, related to potential losses in trading accounts. The level of sophistication of the mathematical models for the analysis of these risks is remarkable. Any review of recent publications in this area illustrates the technical expertise in the quantification of risk measurement and management. The framers of the proposed Basel Capital Accord argue that the new Accord would again make the financial services industry a leader in risk management. Operational risk, in addition to credit and market risk, would be a determinant of minimum capital requirements. While firms in general are beginning to more explicitly discuss the importance of operational risk, the new Basel Capital Accord explicitly requires the financial services industry to manage that risk. The Basel Committee on Banking Supervision (Basel Committee) defines operational risk as “the risk of direct or indirect loss resulting from inadequate or failed internal processes, people and systems or from external events.”1 In surveys, the Basel Committee found the current measurement of operational risk by banks to be relatively undefined and qualitative in nature. In its proposal, the Basel Committee suggests three alternative means to proxy operational risk, all depending to some extent on the mapping of business lines and allocating capital requirements as fixed percentages of those varying lines (see, e.g. Basel Committee, 2001c). In this paper we discuss operational risk and its applications to financial services firms. Our main focus is a review of the literature and the issues in this critical area in international corporate finance. The Basel Accord and the proposed revisions to the Accord represent some of the most significant international regulations impacting the financial decisions of firms, in this case, financial services firms, in recent years. In our review of the issues in this area, we believe we can provide insight into the workings of an important international regulation. We also present suggestions for further research in this area that will become feasible when data on the impact of the new regulations become available.

Operational Risk in Financial Service Providers

149

Note that the application of the new Accord is very much an evolving process. The Basel Committee intends that there should be a common implementation of the revised Accord by 2006. Both regulators and practitioners are debating the definition of operational risk, how to measure it and how to determine appropriate capital requirements. Therefore, in our review of the knowledge in this area, we refer to very recent presentations and proposals. The empirical analysis is limited generally to surveys and small samples. Thus, while there is limited academic evidence on operational risk management now, we believe that it will be a fruitful area for academic research in coming years.

2. THE 1988 BASEL ACCORD The Bank for International Settlements (BIS), headquartered in Basel, Switzerland, is the world’s oldest international financial institution (as noted in BIS, 2002). The stated mission of the BIS is to encourage and facilitate cooperation among central banks. The BIS was originally established after 1930 to take over the supervision of reparation payments imposed on Germany by the Treaty of Versailles. In addition, the BIS promotes cooperation among central banks and other agencies to provide for monetary and financial stability. The role of monitoring reparations ended, but the BIS has become an important international regulatory body for banks. Today (and historically), the BIS has several roles. The BIS has regular meetings of officials of member central banks, is a forum for information exchange among central banks, conducts research and produces data on international finance, performs traditional banking functions such as reserve management for central banks, organizes emergency financing in financial crises, and, through committees of international experts, makes recommendations to the financial community with the stated goal of strengthening the international financial community (see BIS, 2002, for further information on the BIS). The Basel Accord of 1988 resulted from the role of the BIS in making recommendations to the financial community. The Basel Accord defined a set of risk-based capital requirements for financial services firms. The Accord was implemented in the U.S. in the 1991 Federal Deposit Insurance Corporation Improvement Act. The underlying premise of the 1988 Accord is to link banks’ capital requirements to the financial assets in their portfolio. The assets of banks are divided into four risk categories – from the risk-free Category I (for example, cash and reserves), to Category IV (for example, business and household loans – although as Kirstein (2002, p. 394) notes the requirements do not distinguish between loans to firms with differing risk levels, which can lead to adverse incentive effects). The capital requirements of the banks are then linked to the amount of assets in each of the four

150

JEFFRY M. NETTER AND ANNETTE B. POULSEN

categories. When banks shift the composition of their portfolio into more risky assets, they must increase their reserves and therefore decrease their financial leverage. Krainer (2002, p. 425) notes that since banks generally decrease the risk of their portfolios in recessions (the so-called flight to quality), resulting in increased financial leverage of banks, these capital requirements will be countercyclical. Krainer (2002) argues that the Basel Accord of 1988 is a regulatory solution to the agency problem that arises because depositors and bondholders are more risk adverse than stockholders. He suggests that a solution to this agency problem is a corporate governance system that requires managers to make portfolio/investment decisions in the interest of shareholders and financing decisions based on the interests of depositors/bondholders. Then, depositor/bondholders can reduce the agency problem that arises when risk shifting portfolio investment decision are made by shareholders. The Basel Accord, with risk-based capital requirements on depository institutions, is a regulatory framework that attempts to solve this agency problem. Krainer (2002) notes that the Basel Accord in 1988, and its revisions in the 1990s, were responses to the major problems banks had faced worldwide in the 1980s and 1990s. Regulators and commentators believed that many of the banking crises that had occurred were indirectly a result of non-financial firms increasingly using the capital markets (commercial paper and long-term bonds) as a replacement for bank financing. Banks as a response turned to more risky loans and became overleveraged given the riskiness of their portfolios. Kane (2001) argues that international regulatory standards (such as the 1988 Basel Accord) are inferior to competition among national regulatory systems, especially in strengthening the banking systems in developing countries. Kane suggests that even when policy makers act in the public interest and the world is static enough to identify optimal standards, it is unlikely that regulators from high-income countries will agree on standards that are appropriate for emerging economies. It is also unlikely that the high-income countries would agree to compensate emerging economies to adopt and enforce the non-optimal standards. Kane notes that the problem is worse under a public choice model, where policy makers pursue goals other than the public interest. This is especially true in the dynamic complex world of international finance. He believes that a best outcome is much more likely when national regulatory agencies compete in promulgating regulations – those countries that develop the best regulations will attract capital. Kane (2001) illustrates his point with the 1988 Basel Accord. While Kane agrees that risk-based capital requirements can be an effective way of reducing risk-shifting and reducing the likelihood of crises, he does not believe that the Basel Accord has worked well, noting numerous failings. First, methods of measuring capital and risk identified by the Basel Accord are not closely related to the measures that would be used by the market to consider the same factors. Second,

Operational Risk in Financial Service Providers

151

the standards emphasize crisis prevention rather than crisis management. Third, the standards are generally not appropriate for the policy making environments of developing countries. In his comment on Kane’s paper, Goldstein (2001) is much more positive on the potential for value-increasing international regulatory standards, especially if flexibility is built into the standards and if the international standards do not reach down into all aspects of the financial system. Goldstein is also less willing to believe that competition between national regulatory bodies can effectively manage the potential for financial crises on a timely basis. Barth, Caprio and Levine (2001a, b), however, empirically examine the relation between regulatory restrictions and bank performance. They find that greater restrictions are associated with higher probability of a major banking crises, lower bank efficiency and no countervailing positive effects. There have been numerous other criticisms of the 1988 Basel Accord. One of the primary criticisms has focused on the fact that there is a much greater variation in the quality of assets than those defined by the four categories named under the Accord. Krainer (2002, p. 425) points out that the categories are based on legal classifications of assets, which are only remotely related to the investment quality of the assets. In addition, the Accord is only advisory; its capital requirements are actually determined by local bank regulators, who respond to local political considerations. For example, Krainer (2002, p. 426) reports that if an economically important country, such as Japan or Germany, relaxes its capital adequacy requirements, the BIS tends to revise the Accord to accommodate that country. Petrou (2002) outlines another problem with the 1988 Accord. She notes that regulatory actions such as the Basel Accord can make economic cycles more volatile. For example, since the Basel Accord’s requirements only apply to banks, it can have the effect of driving risk-taking to non-regulated entities, which are not subject to the ongoing supervision that banks face from regulators. In addition, Petrou notes that in the U.S., financial service firms have an incentive to abandon their charters and the regulation of the Federal Reserve Board to remain competitive.

3. THE NEW BASEL ACCORD: OVERVIEW In response to the many problems with the 1988 Basel Accord, the Basel Committee is revising the Accord. The changes we discuss in this paper are part of the comprehensive revision of the 1988 Basel Accord developed by the Basel Committee (see Basel Committee, 2001a, b, c, d, 2002a, b) for a much more thorough discussion of the many issues related to the proposed Accord). We do not discuss

152

JEFFRY M. NETTER AND ANNETTE B. POULSEN

all the changes in the new Accord in detail. Our focus is on operational risk, which was not recognized explicitly by the Basel Committee in the 1988 Accord. While factors that are classified as operational risk were recognized in the past, it is only recently that a formal framework for recognizing and dealing with operational risk has developed. The Basel Committee has become one of the first policy making groups to explicitly recognize this risk and attempt to incorporate effects of operational risk in its regulations. In the words of the Basel Committee, “the Basel Committee believes operational risk is an important factor facing banks, and that banks need to protect against the potential loss from it” (Basel Committee, 2002b, p. 1). (Note that the Basel Committee is not the only regulatory institution that has incorporated concepts of operational risk in their regulation of banks and other type firms – see King (2001, pp. 35–43) for other examples.) In broad terms, the Basel Committee has developed a “more comprehensive framework for capital regulation based on three pillars – minimum capital requirements, supervisory review and market discipline” (Basel Committee, 2002b). While we will concentrate on the operational risk component of the new Capital Accord, note that it is only part of the proposed Accord. More broadly, the Basel Committee intends to emphasize risk management of all types of risks in the new Accord and, especially, improvements in banks’ assessments of risk. As opposed to the 1988 Basel Accord, the new Accord has many more options for banks to choose from in calculating their minimum capital requirements. For example, in the calculation of credit risk, the Basel Committee has suggested that a common risk classification for all businesses, large and small, ignores certain risks of lending to small- and medium-size firms. Thus, lending to small- and medium-size firms will be treated differently in the new capital requirements as compared to lending to large firms. Another example is that in calculating the risk to certain types of loans (such as residential mortgages), the banks will perform internal assessments of the risk, which will be used in calculating the capital requirements. In addition, banks can choose from multiple approaches to calculate their minimum capital requirements. In sum, the new Accord is much less “one size fits all” than the 1988 Basel Accord. One of the most fundamental changes proposed in the new Accord, however, is the requirement that the banking industry incorporate operational risk into overall risk analysis for the purpose of setting capital requirements. The Basel Committee notes that several major international banks have been leaders in analyzing and assessing operational risk. We discuss the Basel Committee’s view of operational risk in detail in Section 6. In sum, the proposals for the new Basel Accord make changes to areas that were already included in the Accord and add another important dimension to regulatory

Operational Risk in Financial Service Providers

153

capital requirements – operational risk. Thus, there are now three areas of risk that are related to the minimum capital requirement in: (1), credit risk (which was the focus of the original 1988 Accord); (2), market risk of trading activities (which was introduced in a 1996 amendment to the Accord); and (3) operational risk. Implementation and development of the exact form of the new Accords will take some time. After survey analysis of the impact of the new Accord and any necessary revisions, the Basel Committee anticipates the implementation of the new Accord by the end of 2006.

4. OVERVIEW OF OPERATIONAL RISK 4.1. Definition The widely used definition of operational risk – “the risk of loss resulting from inadequate or failed internal processes, people, and systems or from external events” – resulted from an industry study by the British Bankers’ Association, the International Swaps and Derivatives Association, RMA and Pricewaterhouse Coopers (BBA, ISDA, PwC, RMA, 1999). The Basel Committee adopted this definition early in their documentation of operational risk (see, e.g. Basel, 2001a). King (2001, p. 3) defines operational risk as the risk “not related to the way a firm finances its business, but rather to the way a firm operates its business. He offers an alternative definition (King, 2001, p. 7) – operational risk is “a measure of the link between a firm’s business activities and the variation in its business results.” Culp (2001, p. 432) notes that the data entry form for the British Banker’s Association Operational Risk and Loss Database suggests that examples of losses from operational risk include “failed securities trades, settlements errors in funds transfers, stolen or damaged physical assets, damages awarded in court proceedings, penalties and fines assessed by member associations or regulators, irrecoverable or erroneous funds and asset transfers, unbudgeted personnel costs, and negligence or fraud.” Note these examples include legal risk but exclude reputational and strategic risk (Harris, 2002a). Another approach to a definition of operational risk starts with the observation that all risk faced by banks is financial or non-financial. Kuritzkes (Wharton, 2002) suggests that operational risk is a non-financial risk that has three sources. The first is internal risk such as risk of rogue traders. The second is external event risk, which is an uncontrollable external event such as a terrorist attack or weather destruction. The third is business event risk, which captures many things such as price wars or stock market downturn. Kuritzkes argues that business risk is the most important but is ignored in the proposed Basel Accord.

154

JEFFRY M. NETTER AND ANNETTE B. POULSEN

While operational risk has existed for as long as financial institutions (and all firms) have existed, it is only recently that a discipline of operational risk management has begun to develop. It is most developed among financial services firms who are attempting to quantify its analysis, just as credit and market risk are quantified.2 If and when the new Basel Accord is implemented as regulations in individual countries, there will be a significant movement toward the analysis of operational risk within financial services firms. We will look at several issues in operational risk management including the proposed regulations of the new Basel Accord. Culp (2001) notes that operational risk is so broad a concept that it can be found anywhere. Therefore an appropriate strategy is not in identifying all operational risks, but rather in identifying important operational risks that can affect the value of a firm. Culp (2001, p. 433) discusses several difficulties in identifying operational risk. First, the distinction between operational risk and business risk at any firm is a function of the strategies of the firm. For example, Culp notes a commodityexporting firm has a number of risks such as crop spoilage, truck breakdowns, and worker strikes that the firm’s managers probably consider as normal business risks it expects to deal with. Those same factors might be considered operational risk to a bank. Another difficulty in identifying operational risk is that it is often linked to credit risk or market risk. For example, Culp discusses the often-mentioned example of Baring’s – the investment bank that failed as a result of the losses from a rogue trader. While a rogue trader is generally considered an operational risk, in the end failure occurred because the rogue trader’s positions lost huge amounts of money resulting from a market downturn, i.e. market risk. Culp also argues that it is difficult to identify operational risk because different firms use different organizational processes to deal with operational risk. He notes that financial firms have only recently begun to address operational risk in an organized way while nonfinancial firms have often long paid attention to risks from product management and product liability. Conversely, non-financial firms have only recently begun to address credit and market risk, which financial firms have long considered it.

4.2. Why Do We Care? Hiwatashi (2002) argues that banks have already begun to consider operational risk because of advances in information technology, deregulation, and increased international competition. Similarly, the increase in the number of large mergers and acquisitions, where the combined firm must integrate the systems of the merged firms, can lead to increased operational risk. New financing techniques have reduced credit and market risk but have led to more operational risk within

Operational Risk in Financial Service Providers

155

firms, increasing the benefits of managing this risk (Wharton, 2002). Hiwatashi argues that the increase in the complexity of bank operations as well as increased ability to consider risk have made traditional qualitative risk management inadequate. He reports several reasons why banks try to measure operational risk (he also notes that it is mainly large banks that try to measure the risk). First, by measuring operational risk banks are able to develop objective criteria in analyzing the adequacy of internal risk control measures. Second, banks have long established risk management systems to analyze whether they have adequate economic capital to deal with market and credit risk. Banks have been able to do this because of well-established methods of measuring market and credit risk. Thus, as methods of calculating operational risk have become more feasible, banks have begun to allocate economic capital to operational risk. Harris (2002a) classifies the benefits from operational risk management in a slightly different way. He notes that proper management can improve bottom line earnings by reducing exposure to low frequency but high impact losses. In addition, proper risk management can reduce insurance premiums and lower capital requirements, especially if and when the new Basel Capital Accord is implemented. In fact, large banks that have implemented operational risk management generally already allocate economic capital to operational risk (Hiwatashi, 2002, p. 2). Further, he notes that operational risk management can develop information useful to senior management. Donnelly (2001) takes this final point further and argues that proper operational risk management needs to provide audit committee members with information on the methodology used in risk assessment, identification of issues, and resolution and tracking mechanisms. In the present post-Enron climate, Donnelly’s argument of the importance of the audit committee in providing independent information for senior management is especially relevant. Rosengren (2002) also argues that financial organizations should manage operational risk because of the significant potential costs of operational risk losses. He reports examples of operational risk that have imposed significant costs on firms. First, damage to physical assets and disruption of the business are important considerations, including the $27 billion publicly announced insurance exposure to the 9/11 attack on the World Trade Center. Second, internal fraud and criminal behavior also impose costs, such as the losses to Allied Irish banks of $690 million in rogue trading.3 Third, losses that result from dealings with clients, products, and businesses must also be considered. For examples, he cites the $2 billion settlement of the class action lawsuit by Prudential Insurance caused by its improper sales practices and the $400 million paid by Providian Financial for its unfair sales and collection practices. A much more comprehensive analysis of the overall operational risk loss experience in financial institutions was conducted by the Risk Management

156

JEFFRY M. NETTER AND ANNETTE B. POULSEN

Group (RMG) of the Basel Committee (Risk Management Group, 2002). The RMG performed two studies – QIS2 – Tranche 1 focused on internal capital allocation data for operational risk and information about other exposure indicators. In QIS2 – Tranche 2, the RMG gathered data on individual operational risk loss examples. The data were collected from 30 banks in 11 countries. The studies are really trial runs and one should not read too much into the results. For example, the dataset spans only three years, which limits its usefulness in identifying the rare but extreme events that characterize operational risk. In addition, different banks had different levels of completeness in reporting. In fact, the difficulty of compiling these data illustrates the enormous difficulty in performing one key aspect of operational risk management – collecting data on relatively infrequent and unpredictable events. However, there are still some suggestive results from the study, at least in indicating types of operational risk losses. The RMG collected the number of loss events and gross loss amounts for eight business lines: (1) (2) (3) (4) (5) (6) (7) (8)

corporate finance; trading and sales; retail banking; commercial banking; payment and settlement; agency and custody services; asset management; and retail brokerage.

They considered seven different types of loss events: (1) (2) (3) (4) (5) (6) (7)

internal fraud; external fraud; employment practices and workplace safety; clients; products, and business services; damages to physical assets; business disruption and system failures; execution delivery.

There were 27,371 loss events with a total value of 2.6 billion Euros. Most of the events and the largest Euro value of the losses were in retail banking (67% and 39% of all events and losses respectively) and commercial banking (13%, and 23% respectively), which may reflect where the sample firms do most of their business. Across all business lines, the breakdown of the percent of all events and percent of the value of the losses by event type is as follows: Internal fraud – 2.72%, 10.66%, external fraud – 36.39%, 20.32%, employment practices

Operational Risk in Financial Service Providers

157

and workplace safety – 2.71%, 2.92%, clients, products, and business services – 6.39%, 27.51%, damage to physical assets – 4.48%, 3.02%, business disruption and system failure – 5.32%, 0.82%, execution, delivery, and process management – 41.99%, 34.76%). The most likely and the most costly events were in external fraud and in execution, delivery and process management. The RMG report also compiled results on the distribution of the size of the loss events. Most of the loss events were relatively small – only one percent of the sample were events with losses of one million Euros or more. However, the large loss events dominated the total value of the losses. Events with losses over one million Euros accounted for almost three-fourths of the total losses.4 Finally, one of the most important reasons banks care about operational risk is in preparation for the proposed Basel Capital Accord. We will first discuss the present ways firms have in dealing with operational risk and then review the proposed requirements of the Basel II Accord.

5. METHODS OF DEALING WITH OPERATIONAL RISK 5.1. Approaches Since operational risk is a broad concept that is not well-defined, it is not surprising that there is no one methodology for dealing with operational risk. We will briefly review some approaches and measures that have been used. We then turn to the proposed guidelines of the new Capital Accord. Culp (2001, p. 434) reports (based on survey results) that companies often go through five stages of operational risk implementation. It is useful to review the stages because operational risk management is such a new, ongoing process and the pattern of implementation tells us something. First is the traditional baseline stage. The firm has no processes or personnel to deal with operational risk. Each case is dealt with in an ad hoc reactive manner. In the second stage, firms recognize that operational risk must be considered. Personnel are appointed with operational risk responsibilities and senior managers draft policies for identification, measurement, and control of operational risk. The third stage is the monitoring stage, where “explicit and formal risk tolerances for op risk first begin to appear, albeit on a largely qualitative basis.” The business starts to analyze and track risk measures and adjust performance for operational risk. Senior managers define tolerances for operational risk based on the business and risk management strategies of the firm. In the next stage, the firm develops an operational risk management function, under the Chief Risk Officer or in an Enterprise Wide Risk Management Unit. The firm will have a formal measurement of operational risk on a business unit basis,

158

JEFFRY M. NETTER AND ANNETTE B. POULSEN

and senior managers express their tolerance for risk that are compared to actual measures of exposure to operational risk. In the final stage (to which few firms have moved), operational risk is fully integrated into an Enterprise Wide Risk Management program. The firm integrates quantitative measures of operational risk with the general risk tolerances of the firm, “for example, VaR inclusive of integrated market, credit, liquidty, and operational risk sources” (Culp, 2002, p. 435).5 Harris (2002a) provides a basic overview of what advanced financial organizations are doing to address operational risk that summarizes the implementation of operational risk management. He identifies this pattern: recognizing operational risk as a separate discipline, restructuring the organizational hierarchy, defining a management process, creating measurement tools, developing monitoring systems.6 Hiwatashi (2002) outlines several approaches to operational risk management in banks. He notes first that banks traditionally controlled operational risk based on qualitative risk management checklists and guidelines. This has become inadequate due to the increased complexity and speed of bank operations. Now, banks must first try to measure operational risk so that senior managers can establish objectives in “prioritizing risk control among different business lines and risk categories, in order to supplement internal control in a more robust way” (p. 1). Measurement is also necessary for the management to determine whether the banks have appropriate capital for their level of operational risk. In addition, measurement also enables the bank to tie performance to employees’ risk management effectiveness.

5.2. Measurement Hiwatashi (2002, p. 2) discusses several ways operational risk is measured. The methods are generally categorized as “top-down” or “bottom-up” methods. In top-down methods, risk is estimated based on macro data without identifying the individual events or the causes of losses. One top-method method uses the indicator approach, where some variable, perhaps gross income or cost, is a proxy for firm performance and a certain percentage of the variation in that variable is considered risk. Another approach relies on the CAPM, where total risk is estimated based using the CAPM model. Then, market risk and credit risk are subtracted and what is left is considered operational risk. In the volatility approach, the volatility of some variable, say non-interest income, is treated as the operational risk. Hiwatashi (2002, p. 2) also provides illustrations of the bottom-up methods of measuring operational risk, which uses individual events to determine the source and amount of operational risk. (King, 2001, also provides a complete description

Operational Risk in Financial Service Providers

159

of many of these methods.) These methods include the statistical measurement approach, where operational risk is measured using data from individual events with frequency based on a Monte Carlo simulation or an analytical solution. Another approach is scenario analysis, where losses are estimated based on scenarios derived from other banks and events. A third approach is factor analysis, where factors related to losses are identified and used to calculate risk. For example (King, 2001, p. 73) discusses the Delta Methodology, which is based on error propagation. Under the Delta Methodology, the uncertainty of risk factors is used to calculate the uncertainty in earnings based on sensitivities. The sensitivities represent the correlations of the changes in earnings with changes in risk factors. The Delta Methodology allows losses to be predicted when there is no comprehensive loss data. In addition, it is linked to business activities through the sensitivities. A final method of operational risk measurement considered by King uses Bayesian Network Models to model the causes and effects of operational risk. Culp (2001, p. 435) also discusses methods of measuring operational risk based on four regimes suggested by the ISDA (International Swaps and Derivatives Association). The first is the “basic indicator” regime. Measurement is based on a few roughly defined risk indicators. The firm usually uses an ad hoc control process that depends on existing controls (such as audits) for operational risk management. Often firms use industry or regulatory measures of operational risk. The second is the “standard lines of business” regime. Here the risk management process is less ad hoc and operational risk is measured at the business unit level, often using survey data. The “internal ratings” regime uses subjectively determined quantitative ratings for specific operational risk factors in individual business units. Fourth, the “internal models” regime uses institution-specific loss data and then uses structural econometric models (like credit scoring models) or analytical-and simulation-based VaR-like constructs to determine operational risk. Critical to all the above are the data. Harris (2002a) notes several reasons why an operational risk database is so important. It is necessary for the effectiveness of “sound risk management, realistic evaluation of performance and risk indicators, determining profitability by business unit, locating problem areas within a business unit, senior management and board oversight, and maximizing profits.” At a minimum, Harris argues that a database must contain in which business unit the loss was recognized, which business line recognized the loss, and in what function the loss occurred. He also notes that accuracy depends on having those who collect the data being knowledgeable in the area, on having reasonable reporting thresholds, and on verifying the accuracy and the completeness of the data through audits.

160

JEFFRY M. NETTER AND ANNETTE B. POULSEN

5.3. Challenges and Implementaion A major difficulty in operational risk management is the development of meaningful data. The definition of operational risk is still nebulous as are the data collection methods, especially for low-frequency and high-severity events that represent perhaps the greatest source of operational risk. The most obvious example is the 9/11 terrorist attack. How many firms seriously contemplated a terrorist attack that affected them directly, or even indirectly? How would they have modeled it? The widespread impact of 9/11 is still being felt through a worldwide recession. Hiwatashi (2002, p. 3) lists several areas that pose challenges to banks implementing operational risk. Firms must develop a robust database, which probably needs to include difficult to collect data. A challenging area is determining how much data to collect on indirect losses. The firm must supplement its own internal data with external data and develop a meaningful way to link the two. He also recommends banks establish qualitative data objectively (noting situations where individual managers’ incentives and firm value maximization are not aligned), and keep abreast of rapidly developing risk-transfer methods. Once operational risk is identified (to the extent it can be), the firm must then perform risk control and capital management. One response is insurance. The insurance industry is developing new tools to insure some of the risks we have discussed but many of the risks in operational risk are uninsurable (BBA, ISDA, PwC and RMA, 1999, p. 77). Top management must be informed of the operational risk analysis to consider its implications on the strategy of the firm. The bank must also develop an approach to “quantifying economic-capital for operational risk” (BBA, ISDA, PwC, and RMA, 1999, p. 87). There is no well-developed view of how to do this, however.

5.4. Bank Examiner’s Treatment of Operational Risk Before we turn to the proposed Basel Accords, it is important to note that the concept of operational risk is actually very old. However, it has only recently begun to be considered directly by regulators. Kvistad and Donnelly (2001) note that since the mid-1990s, the Office of the Comptroller of the Currency and the Federal Reserve System have attempted to identify operational risk (Fed) or transaction risk (OCC) within their CAMELS rating and have assigned a charge to it. Kvistad and Donnelly note that the Fed examiners evaluate operational risks in two ways – inherent level of risk in an activity (high, moderate, low) and quality of risk management (strong, acceptable, weak). The Federal Reserve Bank of Chicago has developed eight components for examiners to use in evaluating operational

Operational Risk in Financial Service Providers

161

risk. Each of these indicates potential for operational risk losses. The first is growth and consolidation – a rapidly growing bank, or recently consolidated banks have greater potential for operational risk losses. Second is the quality of the information systems. Third are the quality, training, and morale of the employees. Fourth is the transaction volume and complexity of the transactions of the bank. Fifth is the bank offering new products and services. Next are ripple effects – what would be the indirect effect of an operational disruption. Seventh are the facilities and geographical dispersion of the bank. The final component of operational risk is electronic delivery, with its complexity and security challenges.

6. OPERATIONAL RISK AND THE NEW BASEL ACCORD While the concept of operational risk management has been well-known to financial firms for many years, the response of firms to operational risk had not developed in a systematic way due to the complexity of the concept and the difficulty of developing meaningful data. The proposed Basel Accord offers more standardization in the various approaches to operational risk management. However, as we discuss in the next section, not all share the view that the proposed Accord is the proper regulatory approach. Over the past several years, the Basel Committee has been revising the Capital Accord to better integrate measures of risk, including operational risk, into capital requirements. The revised Accord is based on three pillars – minimum capital requirements, supervisory review, and market discipline (see Basel Committee, 2002a, b). The Basel Committee says its goal is to “foster a strong emphasis on risk management and to encourage ongoing improvements in banks’ risk assessment capabilities. The Committee believes that minimum capital requirements can and should be aligned with prevailing strong risk management practices” (Basel Committee, 2002b, p. 1). As we discussed above, there are numerous changes in the new Accord. We concentrate here on the inclusion of operational risk into the three pillars. In July 2002, the Basel Committee (2002a) published “Sound Practices for Management and Supervision of Operational Risk,” a paper that explains in more detail the concepts behind the inclusion of operational risk in determining capital requirements, the proposals for measuring operational risk and the results of surveys and data collection already undertaken. This document provides an excellent overview of the Basel Committee’s work. We will summarize some of the major points of the report. The Basel Committee (2002a, p. 2) begins with a discussion of the causes and impacts of operational risk. They suggest that operational risk is growing as a result

162

JEFFRY M. NETTER AND ANNETTE B. POULSEN

of several factors, including increased use of automated technology, e-commerce, large scale mergers, banks acting as high volume service providers, the growing use of risk mitigation to optimize exposure to market and credit risk that may actually increase the banks’ exposure to operational risk, and the growing use of outsourcing. The Basel Committee (2002a, pp. 2–3) goes on to note that substantial losses can occur from internal fraud, external fraud, employment practices, clients, products, and business practices, damage to physical assets, business disruption, execution, delivery and process management (note these are the same items in the Committee’s survey of banks’ losses from operational risk). The Basel Committee (2002a) then outlines what they believe are sound practices for operational risk management based on their work on credit risk and market risk management. The Committee notes that operational risk is fundamentally different from other sources of risk since it is not directly taken as a tradeoff for expected return. Thus, the sound practices report first discusses that banks must develop an appropriate risk management environment, adopting “policies, processes, and procedures to control or mitigate material operational risk.” The first part of the report offers suggestions on how to set up an appropriate risk management environment including the role of management in setting up the framework, the principles for the definition of operational risk, and how operational risk is to be “identified, assessed, monitored, and controlled/mitigated” (Basel Committee, 2002a, p. 6). The report also covers the role of supervisors in monitoring management. The Basel Committee also advocates the release of additional information to the market, stating that banks “should make sufficient public disclosure to allow market participants to assess their approach to operational risk management” Basel Committee (2002a, p. 5). This is the third pillar of the proposed Basel Accord – providing information that allows the market to monitor banks’ policies for operational risk management. The report also emphasizes the need for the mitigation of operational risk through business practices. “Banks should have policies, processes and procedures to control or mitigate material operational risks. Banks should assess the feasibility of alternative risk limitation and control strategies and should adjust their operational risk profile using appropriate strategies, in the light of their overall risk appetite and profile” (Basel Committee, 2002a, p. 10). The report provides suggested guidelines for risk mitigation, but many of them actually deal with what to watch out for as a source of operational risk rather than true methods to minimize risk. In particular, there are still no well-developed methodologies for quantifying low probability but large impact events in an overall measure of operational risk. The Basel Committee does make one specific recommendation, however. The report states, “Banks should have in place contingency and business

Operational Risk in Financial Service Providers

163

continuity plans to ensure their ability to operate as going concerns and minimize losses in the event of severe business disruption” (Basel Committee, 2002a, p. 12). Perhaps the best discussion of the basic outlines of the Proposed Accord with respect to operational risk come from the Basel Committee itself (Basel Committee, 2002b, pp. 7–8): 42. The Basel Committee believes that operational risk is an important risk facing banks and that banks need to hold capital to protect against potential losses from it. This view is shared by a number of globally-active financial institutions, which have been at the forefront of analyzing and assessing operational risk. The Committee proposes to define operational risk as the risk of loss resulting from inadequate or failed internal processes, people and systems, or external events, and therefore includes legal risk. 43. Approaches to the measurement of operational risk continue to evolve rapidly, but are unlikely to attain the precision with which market and credit risk can be quantified. This poses obvious challenges to the incorporation of a measure of operational risk within the minimum capital requirement. Nevertheless, the Committee believes that such inclusion is essential to ensure that strong incentives exist for the continued development of approaches to operational risk measurement and management and to ensure that banks are holding sufficient capital buffers for operational risk. It is clear that a failure to treat operational risk within the minimum capital requirement (pillar one) would reduce these incentives and result in a reduction of industry resources devoted to operational risk issues. 44. On the other hand, the Committee is prepared to allow an unprecedented amount of flexibility to banks in choosing how to measure operational risk and the resulting capital requirement. Under the advanced measurement approaches (AMA), banks will be permitted to choose their own methodology for assessing operational risk, so long as it is sufficiently comprehensive and systematic. The extent of detailed standards and criteria for use of the AMA are minimal in an effort to spur the development of innovative approaches, although the Committee intends to review progress in regard to operational risk approaches on an ongoing basis.

Thus, an important part of the proposed Accord is the minimum capital requirement of Pillar I, where the Committee will allow “an unprecedented amount of flexibility to banks in choosing how to measure operational risk and the resulting capital requirement.” The Basel Committee has suggested several alternative means to determine capital requirements for capital risk. The Basic Indicator Approach is the simplest measure and would be easiest to implement for many smaller banks. The Basel Committee anticipates that the Basic Approach could be used by any bank regardless of its complexity or sophistication, though it does not expect the largest or most active banks to choose this approach. The Basic Approach states that required capital for operational risk would be equal to a fixed percentage (denoted as ␣) times a single indicator such as the level of gross income. In the Standardized Approach, banks’ activities are classified into eight categories. Each firm would determine the relative weights of each line of

164

JEFFRY M. NETTER AND ANNETTE B. POULSEN

business in their overall activities through an indicator measure such as that line’s gross income. Required capital would be equal to the summation of the indicator for each line multiplied by a factor (␤) that would reflect the riskiness of that line. The Basel Committee would set the ␤ so it would reflect the operational risk of that business line. Banks would have to meet certain standards to be eligible to use this approach, basically requiring well-developed processes and controls for the measurement and management of operational risk. Both the Basic Indicator Approach and the Standardized Approach are relatively simple. However, they are not sensitive to the specific activities of each bank nor do they take into account the bank’s ability to manage its operational risk. In response to industry critiques of these two approaches, the Basel Committee has turned its emphasis to the Advanced Measurement Approach (AMA). Under the AMA, there is a firm-specific calculation of the capital requirement based on the firm’s internal risk measurement models. These models use the bank’s own metrics to measure operational risk including internal and external loss data, scenario analysis, and risk mitigation techniques to then set its capital requirements. There are essentially three components: (1) operational loss data; (2) quantification methodologies (which can be very complex); and (3) qualitative risk assessment and risk indicators (like scorecards and early warning systems). Cumming and Hirtle (2001) offer some preliminary guidelines on the development of consolidated risk management and the economic rationale for doing so. Rosengren (2002) argues that this approach is risk sensitive since the capital requirements are based on the operational risk exposure of the bank and flexible since banks choose the methodologies. He also notes that this approach provides a return to banks for investing in controls for operational risk losses. The Basel Committee has provided extensive guidelines of the requirements for banks to be eligible to use AMA. Basically, the bank must have a well-developed and comprehensive risk management plan that carefully analyzes all risk components and includes controls to insure that risks are managed correctly. Pillar 2 (the supervisory review process) and Pillar 3 (the enhancement of market discipline through disclosure) also play important roles in the proposed Capital Accord. Under Pillar 2, banks will be required to establish efficient means to measure and manage operational risks and supervisors will be given guidelines to assess capital adequacy and risk management techniques in the firm. The Committee anticipates including disclosure requirements with respect to operational risk to support Pillar 3 in the new Accord. These disclosures would include the approach used to determine capital requirements, the operational risk objectives and policies of the bank, the internal systems for control of operational risk and the success of the bank in its operational risk management.

Operational Risk in Financial Service Providers

165

7. COMMENTARY ON BASEL CAPITAL ACCORD II Several commentators have expressed strong criticisms of the proposed Basel Capital Accord, especially in the initial stages of its development. Petrou (2002) agrees that there is a need to modify the regulatory capital requirements applying to credit and market risk, but she is critical of the very idea of applying capital requirements to operational risk and also of the method of implementation. She notes that regulatory standards can distort economic incentives and create perverse incentives, perhaps leading to the holding of excessive operational risk. Even with the AMA approach, which attempts to benefit those firms that actively manage their risks, she questions why the Basel Committee believes it can define capital requirements for operational risk when regulators have long agreed to allow supervisory agencies to consider interest-rate risk on a more qualitative basis. Since interest-rate risk is far more quantifiable than operational risk, she argues that the Basel Committee is attempting something that cannot be done on a methodological basis and should be left to the Pillar 2 and Pillar 3 guidelines. She argues that “regulators should devote the resources now distracted by the complex quantitative exercises necessary to develop the operational risk-based capital rules to improving supervisory standards and bank operational risk management and mitigation.” In sum, Petrou (2002) notes that while there are three pillars to the proposed Accord, almost all the focus has been on the capital requirements. She argues that relatively much more attention should be paid to supervision, which may be much more effective than capital requirements. Several participants at a Wharton Conference on Operational Risk (Wharton School, 2002) were also critical of the capital requirements tied to operational risk. Kuritzkes said, “I do not think that BIS or any other regulatory authority can come up with any rules for how much capital banks can hold against operational risk. The first line of defense for such risk is internal controls.” He was especially critical of the AMA method of calculating regulatory capital, stating “The AMA not only tries to specify a rule-based approach but tries to do it in a highly structured and sophisticated way that I think stretches the bounds of what operational risk can deliver. It seems not worth a candle. There is a much better payback for BIS to concentrate on other components of the Risk Framework like the credit risk.” Muermann (Wharton School, 2002) noted that operational risk is bank specific and thus regulatory capital requirements are not appropriate. He asked an insightful question about capital requirements, “Could capital requirements have avoided major risk events such as the collapse of Barings?” ISDA (2000) also presents strong criticisms of the proposed capital requirements, arguing that the capital requirements are unworkable and can lead to distorted actions, such as attempts to avoid control rather than mitigate risk. The

166

JEFFRY M. NETTER AND ANNETTE B. POULSEN

ISDA report argues instead for the importance of strong supervision and market discipline (Pillars II & III). They note that any attempt to set capital requirements based on system-wide operational risks would: “reintroduce the sort of failings that undermined the 1988 Capital Accord, by making it insensitive to the true risks being run by any one bank (ISDA, 2000, p. 43).” Further, they discuss other types of measures of risks for setting capital requirements such as activity-based or cost-based measures. The ISDA paper states that, with respect to the various methods of determining capital charges, “Each remains experimental, with no single method or combination of methods currently deemed satisfactory by the institutions that have reviewed them. Each has serious and readily apparent drawbacks, consistent with the general philosophical problem of providing capital against operational risk (ISDA, 2000, p. 43).” Finally, while the academic analysis in this area is still developing, Barth, Caprio and Levine (2002) analyze the relationship between different regulatory and supervisory practices and banking-sector development, efficiency and fragility. The authors use a cross-sectional database on banks in 107 countries. While this type of research is difficult to do and must be interpreted with skepticism (see Megginson & Netter, 2001, for a discussion of the difficulties of performing international cross-sectional studies of firm performance), their results have implications for the effectiveness of the proposed Basel Accord. For example, they find that stringent capital requirements are not associated with “bank development, performance or stability when controlling for other features of bank regulation and supervision” (Barth, Caprio & Levine, 2002, p. 31). They also find that there is little relation between supervision of banks and performance and stability. They do find that policies that force information disclosure and encourage private-sector control of banks do improve performance and stability. Barth, Caprio and Levine note that their results are consistent with other recent studies that show harmful effects of capital requirements imposed on banks. Thus, these results suggest that the first two pillars of the proposed Basel Accord (capital requirements and supervision) may not be beneficial while the third (market discipline through increased disclosure by banks) may be useful.

8. CONCLUSION Throughout the world, the governance of financial service firms is perhaps the most regulated area of firm behavior due to the importance of the banking system to national and international economies. The Bank of International Settlements (BIS) makes recommendations for regulatory strategies that are adopted by national

Operational Risk in Financial Service Providers

167

regulatory bodies worldwide. As such, the regulatory standards adopted by the BIS are an important example of external corporate governance mechanisms. In 1988, the BIS adopted the Basel Accord, which defined a set of risk-based capital requirements for financial firms that were adopted by many governments. The Basel Committee is presently proposing revised Accords, which are based on three pillars – minimum capital requirements, supervisory review, and market discipline. For the first time, the requirements will apply to operational risk. In this paper we discuss operational risk and its applications to financial service firms, through a review of the issues and existing literature in this critical area of international standards of corporate governance. We believe we provide a timely review of a developing managerial area and an important international corporate governance standard. While the specific requirements of the new Basel Capital Accord are still under discussion, and are not expected to be fully implemented until 2006, it is essential that financial services firms begin to understand the definition, measurement, analysis and management of operational risk. Overall, whether as a result of the new Accord, or in recognition of the increasing level of operational risk in financial service firms, banks will have to continue to develop sophisticated ways to measure operational risk, integrating market and credit risk into its analysis in the next few years. There are numerous research issues to be pursued as operational risk management is integrated into the overall risk profile of financial institutions. Several key questions include:
What are the key challenges in quantifying operational risks in banks. Is it possible to truly measure these risks or will banks end up relying on ad hoc schemes to satisfy regulators.
What will be the requirements for banks to qualify for the Advanced Measurement Approach in determining capital requirements? This approach places emphasis on developing appropriate risk control measures, in addition to simply quantifying risks. As such, it seems to offer the best approach to operational risk management.
To what extent should firms differentiate between regulatory capital requirements versus economic capital needs?
To what extent can firms integrate the three areas of risk – credit, market and operational? By identifying each in isolation, financial services firms can allocate risk within the firm, but in so doing may ignore correlations between the risks that actually lower the overall risk profile of the firm. As the sophistication in the measurement of operational risk increases, can these correlations be used to reduce capital requirements?

168

JEFFRY M. NETTER AND ANNETTE B. POULSEN


It is important for the Basel Accord and national regulatory bodies to recognize the differences between firms. Are there differing bank characteristics that suggest different approaches to risk management? How does a firm determine its “best practices” relative to its competitors? We look forward to answers to these questions and more as the financial services industry reacts to new Basel Capital Accord and as new data become available on the integration of operational risk measurement and management in financial firms.

NOTES 1. The Basel Committee adopted this definition from a comprehensive analysis of operational risk by the British Bankers’ Association, ISDA (International Swaps and Derivatives Association), RMA and PricewaterhouseCoopers, published as Operational Risk: The Next Frontier (BBA, ISDA, RMA and PwC, 1999). 2. See, e.g., Cumming and Hirtle (2001) for a discussion of the analysis of comprehensive risk management in financial institutions. A recent conference at the Federal Reserve Bank of Boston (2001) provides several examples of the differing approaches to operational risk management that have been suggested by various banks and consulting firms. Cagan (2001) and Nash, Nakada and Johnston (2002) offer specific suggestions for institutions preparing for the 2006 implementation of the new Accord. 3. In addition, to the widely publicized incidents of fraud in the U.S. recently, there are many other examples of large losses often due to fraudulent behavior in large firms. King (2001, pp. 21–34) discusses some examples (and provides data on 89 cases) where financial service firms suffered large losses because managers did not monitor and control the risk of operational processes. Two other papers provide case studies of major losses by financial services firms. Buchanan, Arnold and Nail (2003) analyze the corporate governance failures that led to the collapse of Australia’s second largest insurer – HIH Insurance in March 2001. Buchanan and Netter (2002) analyze the money laundering scandal of the Bank of New York, the 16th largest bank in the U.S. Improper oversight by supervisors at the bank allowed others in the bank to launder $10 billion for Russian organized crime. 4. The RMG report also reports recovery rates and percent of losses that were recovered, where recovery comes from insurance and other sources. The RMG reports that there are significant problems with these data but they do show recovery in 12.2% of all events (36.1% of the greatest magnitude loss events), with recovery when it occurs averaging 81.6% of the loss. 5. Harris (2002b) outlines the pattern of adoption of enterprise risk management that is very similar. He notes the common approach relies on “siloed line of business risk management, business line manager owns the risk, ad-hoc assessment, many performance indicators, inconsistent reporting, internal data only, crude capital indicators, or 8% minimum regulatory capital.” In contrast, the emerging practices are “integrated risk management, corporate risk management, standardized firm-wide risk self assessment,

Operational Risk in Financial Service Providers

169

core indicators/triggers, uniform reporting, economic loss, internal and external data, and risk sensitive capital.” 6. Harris (2002b) also outlines the role of an operational risk manager in a firm. The roles include: document risk management policy, ensure senior management buys into policy, establish reporting and metrics, promote capital data management systems, develop loss tracking methods, and map to business line by proxy (such as net income).

ACKNOWLEDGMENTS We thank the Financial Services Exchange for their generous support of this project through a grant program entitled “What would the most highly evolved financial service provider look like in the year 2010.” We also thank the Center for Strategic Risk Management in the Terry College of Business at the University of Georgia for its support.

REFERENCES Bank for International Settlements (2002). About the Bank for International Settlements. Available at: http://www.bis.org/index.htm Barth, J. R., Caprio, G., Jr., & Levine, R. (2001a). Banking systems around the globe: Do regulations and ownership affect performance and stability. In: F. Mishkin (Ed.), Prudential Supervision: What Works and What Doesn’t. Chicago: University of Chicago Press. Barth, J. R., Caprio, G., Jr., & Levine, R. (2001b). The regulation and supervision of banks around the world: A new database. In: R. E. Litan & R. Herring (Eds), Integrating Emerging Market Countries into the Global Financial System. Brookings-Wharton papers on financial services. Washington DC: Brookings Institution Press. Barth, J. R., Caprio, G., Jr., & Levine, R. (2002). Bank regulation and supervision: What works best? NBER working paper 9323. Basel Committee, on banking supervision (2001a). The new Basel capital accord: An explanatory note. Report to the bank for international settlements (January). Basel Committee, on banking supervision (2001b). Consultative document: Operational risk. Report to the bank for international settlements (December). Basel Committee, on banking supervision (2001c). Working paper on the regulatory treatment of operational risk. Report to the bank for international settlements (September). Basel Committee, on banking supervision (2001d). Sound practices for the management and supervision of operational risk. Report to the bank for international settlements (December). Basel Committee, on banking supervision (2002a). Sound practices for the management and supervision of operational risk. Report to the bank for international settlements (July). Basel Committee, on banking supervision (2002b). Overview Paper for Impact Study. Report to the bank for international settlements (October). British Bankers’ Association, ISDA, Pricewaterhouse Coopers, and RMA (1999). Operational risk: The next frontier. Philadephia: RMA.

170

JEFFRY M. NETTER AND ANNETTE B. POULSEN

Buchanan, B., Arnold, T., & Nail, L. (2003). Beware the ides of March: The collapse of HIH insurance. Research in International Business and Finance (forthcoming). Buchanan, B., & Netter, J. (2002). Money laundering and the bank of New York. Terry College of Business, University of Georgia working paper. Cagan, P. (2001). Standard operating procedures. Erisk.com (March). Culp, C. (2001). The risk management process: Business strategy and tactics. New York: John Wiley & Sons. Cumming, C., & Hirtle, B. (2001). The challenges of risk management in diversified financial companies. FRBNY Economic Policy Review (March), 1–17. Donnelly, P. (2001). What every audit committee member should know. The RMA Journal (September), 48–51. Federal Reserve Bank of Boston (2001). Conference on capital allocation for operational risk, (November). Available at: http://www.bos.frb.org/bankinfo/conevent/oprisk/ Goldstein, M. (2001). Comment and discussion on “Relevance and the need for international regulatory standards.” In: R. E. Litan & R. Herring (Eds), Brookings-Wharton Papers on Financial Services (pp. 116–126). Washington DC: Brookings Institution Press. Harris, R. (2002a). Emerging practices in operational risk management. Federal reserve bank of Chicago, June 24, (2002). Available at: http://www.chicagofed.org/bankinforeg/bank regulation/opsrisk/june2402ny.pdf Harris, R. (2002b). A domestic regulatory approach to operational risk. Federal reserve bank of Chicago, December 2, 2002. Available at: http://www.chicagofed.org/bankinforeg/bank regulation/opsrisk/dec2002/.pdf Hiwatashi, J. (2002). Solutions on measuring operational risk. Capital markets news, the federal reserve bank of Chicago (September), 1–4. ISDA (2000). A new capital adequacy framework: Comments on a consultative paper issued by the Basel Committee on banking supervision in June 1999. International swaps and derivatives association, Inc: Working paper (February). Kane, E. J. (2001). Relevance and the need for international regulatory standards. In: R. E. Litan & R. Herring (Eds), Brookings-Wharton Papers on Financial Services (pp. 87–115). Washington DC: Brookings Institution Press. King, J. L. (2001). Operational risk. New York: John Wiley & Sons. Kirstein, R. (2002). The new Basel accord, internal ratings, and the incentives of banks. International Review of Law and Economics, 21, 393–412. Krainer, R. (2002). Banking in a theory of the business cycle: A model and critque of the basel accord on risk-based capital requirements for banks. International Review of Law and Economics, 21, 413–433. Kvistad, J., & Donnelly, P. (2001). An examiner’s view of operational risk. Bank news (June). Megginson, W., & Netter, J. (2001). From state to market: A survey of empirical studies on privatization. Journal of Economic Literature, 39, 321–389. Nash, M., Nakada, P., & Johnston, B. (2002). Start today for enterprise-wide risk management in 2006. The RMA Journal (November), 56–61. Petrou, K. S. (2002). Taking care with capital rules: Why getting them right matters so much. Working paper: Federal financial analytics, Inc. Available at: http://www.chicagofed.org/ newsandevents/bankstructureconference/2002/pdffiles/shaw-petrou2002.pdf Risk Management Group (2002). The quantitative impact study for operational risk: Overview of individual loss data and lessons learned. Report to basel committee on banking supervision, bank for international settlements (January).

Operational Risk in Financial Service Providers

171

Rosengren, E. (2002). Quantification of operational risk. Presentation at FRB of Chicago bank structure conference, (May 9). Available at: http://www.chicagofed.org/newsandevents/ bankstructureconference/2002/doc/Presentation.ppt. Wharton School (2002). Operational risk poses challenges to financial institutions and regulators. Summary of 2002 risk roundtable series organized by Wharton financial institutions centre (July 3). Available at: http://knowledge.wharton.upenn.edu/articles.cfm?catid=1&articleid=582

AUDITOR RESIGNATIONS, LITIGATION RISK AND LITIGATION EXPERIENCE Susan Scholz ABSTRACT Accounting firms claim that the risk of costly litigation leads to resignations from high-risk clients, and that these resignations represent an economic inefficiency. This study examines the association between resignations, dismissals and litigation in the computer industry from 1988–1995. Resignations and dismissals appear to be similar, suggesting some dismissals are implicit resignations. Results support a relationship between risk and resignations. Since some characteristics of auditor litigation risk are also characteristic of unprofitable audit engagements, the analysis incorporates the actual litigation experience of sample companies to provide insights into claims of inefficiencies surrounding the switches.

1. INTRODUCTION This study investigates the association between auditor litigation risk and auditor resignations. Accounting firms have emphasized that the risk of costly litigation prevents them from providing audit services for certain clients. Specifically, auditors claim that they resign from clients that seem likely to involve them in litigation. For example, the largest accounting firms (now the Big 4) stated: “Accountants are practicing risk reduction. [Large firms] are attempting to reduce the threat of litigation by avoiding what are considered high-risk audit clients

Corporate Governance and Finance Advances in Financial Economics, Volume 8, 173–193 © 2003 Published by Elsevier Science Ltd. ISSN: 1569-3732/PII: S1569373203080083

173

174

SUSAN SCHOLZ

and even entire industries” (Andersen et al., 1992). Auditors used the claim of an association between auditor litigation and resignations to successfully lobby for relief in the Securities Litigation Reform Act of 1995. Resignations attributable to assessments of increased auditor litigation risk represent an economic inefficiency when laws intended to deter undesirable behavior (sub-standard audits) also deter desirable behavior (adequate audits.) The cost of these inefficiencies is borne by both auditors and clients. Auditors forfeit revenues, while client costs include transactions costs of hiring a new auditor and agency costs (i.e. negative stock price reactions) arising from information asymmetries regarding the cause of the resignation (DeAngelo, 1981; DeFond et al., 1997). There are several factors that mediate the relationship between perceived litigation risk and resignation. First, some client characteristics associated with litigation risk are also associated with poor financial health, making it unclear whether a resignation is attributable to heightened litigation risk or audit profitability concerns. The former suggests possible inefficiencies, while the latter does not. Second, the line between auditor dismissals and resignations is not always clear. Auditors have political incentives to overstate the incidence and costs of resignations, but in certain circumstances may choose to force a dismissal by raising fees or demanding very conservative accounting (Krishnan, 1994; Simunic & Stein, 1995). These implicit resignations may be to preserve a relationship with the client (either non-audit services or future audit work) or to avoid a reputation for being overly harsh. This study analyzes auditor resignations in the computer industry from 1988 to 1995.1 The resignations are compared to dismissals and all the switching companies are compared to a matched sample of companies that did not switch. Auditors perceive the computer industry to have a high litigation risk (Andersen et al., 1992), providing a setting that is likely to exhibit the risk resignation phenomenon, while also controlling for industry-specific litigation risk. The study extends existing research by incorporating the actual litigation experience of the sample companies. Both litigation where the auditor is named as defendant (auditor litigation) and financially-related litigation that does not directly involve the auditor (other litigation) are considered. Results support auditors’ contentions that high litigation risk is associated with resignations. Modified audit reports, a proxy for auditors’ perceptions of a client’s prospects, is the most consistently significant variable. Actual litigation and revelations of questionable management integrity also appear to be important resignation factors. However, the analysis also shows that smaller, less profitable clients are disproportionately affected by resignations, suggesting that resignations are

Auditor Resignations, Litigation Risk and Litigation Experience

175

determined by low perceived benefits as well as an increase in expected litigation costs. Thus, support for auditors’ claims that risk resignations result in economic inefficiencies in the audit market is ambiguous.

2. PRIOR RESEARCH Existing evidence supports the contention that resignations are costly for clients. Wells and Loudder (1997) find a negative price reaction at the report of a resignation and DeFond et al. (1997) find that stock price reactions are more negative for resignations than dismissals. Bockus and Gigler (1998) develop an economic rationale for auditor claims that increases in expected liability give rise to an increase in resignations. Their theory explains that an incumbent auditor rationally resigns an engagement because any attempt at risk-adjusted pricing leaves him/her with only unprofitable clients. Krishnan and Krishnan (1997) test an empirical model of the association between resignations and auditor litigation risk, based primarily on Stice’s (1991) risk model. (In addition to using a different model than the current study, they also use a different sample period and sampling method.) Their evidence indicates that resignation companies differ from dismissal companies along dimensions that capture the probability of litigation: financial distress, variance of abnormal returns, auditor independence, tenure and modified opinions. Shu (2000) finds that resignations are related both to litigation risk and auditor adjustments to correct clientele mismatches. Her evidence suggests that resignations are driven by supply-side concerns. Consistent with earlier research, she finds that investors react negatively to resignations, and also that the price drop varies with litigation risk.

3. FRAMEWORK, VARIABLES AND EXPECTATIONS Given an increase in litigation risk for a client, the auditor has incentives to resign in order to minimize the expected costs of litigation. These incentives exist even when the auditor believes he provides an acceptable level of audit quality (i.e. that he performs audits which are in accordance with Generally Accepted Auditing Standards (GAAS)). This is because litigation imposes reputation and legal costs regardless of the eventual outcome (Palmrose, 1991), and because courts sometimes err in assigning liability. These errors arise from uncertainties inherent in determining what constitutes GAAS and whether it has been followed (Shavell, 1987).

176

SUSAN SCHOLZ

Kinney’s (1993) model of auditor liability provides a framework for identifying characteristics which increase the litigation risk for a particular client. The model’s premise is that investor losses are a necessary (but not sufficient) condition of auditor litigation.2 There may be multiple causes of such losses, for which auditors, other parties or both may be responsible. Because responsibility for the loss is often unclear and the liability allocation process uncertain, any factor which increases the risk of user losses also increases the risk of auditor litigation. Assuming that the level of audit quality is at or above that required by GAAS, the model identifies two important factors that affect the possibility of user losses but are not under the direct control of the auditor: the quality of the pre-audit financial statements and other causes of investment value decline, where “other” refers to reasons unrelated to the audited financial statements. Variables used in this analysis are intended to capture these two concepts.

3.1. Pre-audit Financial Statement Quality Variables The quality of the pre-audit financial statements provided by the client is determined in large part by client internal controls and management integrity. The importance auditors place on management integrity is demonstrated by the background checks of potential client management and owners usually conducted during the client acceptance process.3 Internal control is considered to be so important that discovery of a material weakness should be reported to Board of Directors. (AICPA, 1988). Thus, the professional literature reinforces Kinney’s insight regarding the importance of the pre-audit statement quality. However, auditors’ assessments of these client attributes are not publicly available, so actual litigation, auditor tenure and SEC enforcement actions (AAERs) are used as proxies. The measurement and rationale for each is discussed next. Litigation is coded one if there is a lawsuit involving financial reporting or disclosures against the client about the time of the switch, zero otherwise. These lawsuits may or may not name the auditor as a defendant. Even litigation that does not directly name the auditor (other litigation) is likely to increase the auditor’s perception of risk. First, a lawsuit is typically an allegation that management made misleading or fraudulent representations, which could affect the auditor’s perception of management integrity. Second, auditors claim they are often attached to lawsuits merely because they provide “deep pockets” for potential settlements. So, any lawsuit based on allegations of financial wrong-doing increases the likelihood that the auditor will be named as a defendant. The subsets of litigation that do/do not involve the auditor directly are considered in sensitivity analysis (auditor litigation = 1, 0 otherwise and other

Auditor Resignations, Litigation Risk and Litigation Experience

177

litigation = 1, 0 otherwise). The expected relationship between all litigation variables and resignation is positive. Auditor tenure is the number of years the auditor issued an opinion on a client’s financial statements. An association between litigation and relatively new clients has been noted by the profession (AICPA, 1995). Also, a longer relationship suggests that the auditor has experience with the quality of controls and management and considers them acceptable. The expected relationship with resignation is negative. SEC AAER is coded one if the SEC issued an enforcement release (AAER) against the company, zero otherwise. AAERs are issued by the SEC when companies (or auditors) have been found to have engaged in inappropriate accounting (or auditing) activities. Their use in this setting is somewhat limited because they are relatively rare and are often issued long after the accounting violation occurred (Feroz et al., 1991). However, AAERs are the only publicly available indication of potential management integrity issues which are available for both switching and matched companies. The expected relationship with resignation is positive. Information from auditor switch reports (Form 8-K) is used to compare resignation to dismissal companies. Information reported in Form 8-K includes whether or not the auditor had a material disagreement with management, deemed management unreliable, withdrew outstanding audit reports or identified a weakness in the client’s internal control system. Each item is measured as an indicator variable equal to one if it is mentioned in the 8-K, zero otherwise. A summary variable, revealed auditor concerns (RAC), is also constructed from this data. It equals one if any of the auditor concerns about management or controls is revealed in the 8-K, zero otherwise. Each of these variables is expected to be positively associated with resignation.

3.2. Other Causes of Investment Decline Variables Deteriorating client financial condition is the primary “other” cause of investment losses. Pratt and Stice (1994) find evidence that auditors associate declining financial condition with higher litigation risk, and various measures of poor financial condition have been found to be significantly associated with auditor litigation (Lys & Watts, 1994; Stice, 1991). This model uses four variables to capture the auditor’s perception of clients’ financial condition risk: indicator variables for clients reporting net losses or filing for bankruptcy (=1 if so, zero otherwise), the debt/asset ratio and a variable for audit report type (unmodified or consistency modification = 0, uncertainty modification = 1, going concern modification = 2).4 Each of these variables is expected to be positively associated with resignations.

178

SUSAN SCHOLZ

3.3. Control Variables Client size, percentage change in sales, initial public offering (IPO) and auditor type (Big 6/Non-big 6) are included as control variables. Each has been identified as a factor related to either auditor switching or litigation, but their relationship with resignations is unclear. Size is measured as ln (total assets) at the year end prior to the switch. Research shows that larger companies are positively associated with auditor litigation (Carcello & Palmrose, 1994; Lys & Watts, 1994; Stice, 1991). However, it is not clear that larger clients would be associated with resignations, since they generally provide larger fees and greater prestige for their auditors. The percentage change in sales for the year ending prior to switch is a control for client growth. Similar to size, growth has been hypothesized to be associated with litigation risk (e.g. Lys & Watts, 1994; Stice, 1991) but otherwise, growing companies are expected to be attractive clients due to the probability of increasing fees and prestige. IPO equals one if the client had an IPO within three years of the switch, zero if not. Auditors claim that the IPO market is a litigation prone environment (AICPA, 1995; Andersen et al., 1992). However, some academic research suggests that IPOs may not have higher rates of auditor litigation than do other shareholder lawsuits (Palmrose, 1987). Also, the rate of auditor litigation associated with IPOs may have declined dramatically in recent years (Bunsis & Drake, 1995). Furthermore, to avoid liability for IPO related issues, the auditor would have to end the relationship before the IPO. In fact, audit programs (see note 3) indicate that clients contemplating an IPO are given particular scrutiny. Big 6 auditors are generally associated with larger clients, which in turn have a greater likelihood of other litigation (Francis et al., 1994; Kellogg, 1984). They also have more resources (deep pockets) to attract entrepreneurial attorneys. On the other hand, Palmrose (1988) found that non-Big 6 auditors had higher litigation rates than Big 6 auditors. Furthermore, Stice (1991) and Lys and Watts (1994) did not find evidence supporting a distinction between different types of auditors and litigation. Big 6 is coded one if the client has a Big 6 auditor, zero otherwise.

4. RESEARCH DESIGN AND SAMPLE SELECTION The study uses a sample of all resignations and dismissals that could be identified in the computer industry from 1988 to 1995. Each resignation and dismissal is matched (by SIC code and year of switch) to a client who did not experience an auditor change during the sample period.

Auditor Resignations, Litigation Risk and Litigation Experience

179

The industry study design is used primarily to control industry effects. Resignation behavior is likely to be associated with the varying levels of litigation in general (Francis et al., 1994) and audit litigation in particular (Palmrose, 1988; St. Pierre & Anderson, 1984) across industries. Industry membership is also associated with varying perceptions of litigation risk among practitioners (Andersen et al., 1992; Pratt & Stice, 1994). Previous studies have controlled for industry effects by matching control samples based on industry codes. An industry study design provides the same control, and also provides information about the rates and characteristics of dismissals and resignations based on the population of switches within the industry. The computer industry is selected for this study for three reasons. First, anecdotal evidence suggests that auditors perceive it to be an industry in which high litigation risk has caused resignations (e.g. Simonetti & Andrews, 1994). Second, academic evidence supports auditors’ perception of high litigation risk (Francis et al., 1994; Palmrose, 1988). This is important because resignations can be relatively unusual (DeFond et al., 1997; Menon & Williams, 1991). Therefore, it is important to identify an industry which will generate a sufficient sample size within the sample period. Finally, the wide variety of risk characteristics (e.g. size, age, product risk, financial condition) within the client population can provide meaningful tests of the hypotheses. The industry is defined as SIC codes 3570-3577 and 7371-7373. This definition is similar to that used by Francis et al. (1994). The June 1988 starting date is selected because it was about the time disclosure of resignations in Form 8-K was mandated by the SEC (in FRR-31, which became effective May 1988). December 1995 is the cut-off date because legislation passed by the U.S. Congress in December 1995 significantly modified the litigation environment under which auditors operate. Thus, the sample is collected during a period with a consistent legal and disclosure environment.

4.1. Sample Selection and Data Sources Switching companies are identified from searches of the Lexis-Nexis Form 8-K database for companies reporting Item 4 (Changes in Registrant’s Certifying Accountant). Additional switches are identified from a review of Compustat auditor codes. Each switch is matched with a Compustat company that did not switch auditors during the sample period. Table 1 summarizes the sample composition and attrition. A total of 207 dismissals and 46 resignations (253 switches) are identified. Eleven resignations are eliminated because they are attributed to the dissolution of the incumbent auditing firms (7) or because they are related to auditor independence issues (4).

180

SUSAN SCHOLZ

Table 1. Dismissals and Resignations Identified in the Computer Industry from 1988–1995. Computer Industry is SIC codes 3570–3577 and 7371–7373. Dismissals

Resignations

Total

Switches identified in searches Less auditor bankruptcy/independence related resignations Less merger-related dismissals Less switches missing data

207

46 (11)

253 (11)

(8) (39)

(1)

(8) (40)

Switches in sample

160

34

194

DeFond et al. (1997) document that such resignations are significantly different from risk-related resignations. Eight dismissal observations are eliminated because they are attributed to mergers. Mergers force clients to dismiss an auditor, and so are not representative of dismissal decisions in general. Since several clients had multiple dismissals and/or resignations, these 234 switches represent 178 companies. An additional resignation and 39 dismissals are missing data required for the regression model. Therefore, the analysis uses 34 resignations and 160 dismissals plus 194 matched no-switch clients, for an overall sample of 388. Information about the switch (dismissal or resignation, disagreements, etc.) is mainly from copies of 8-K’s obtained from Disclosure, Inc. For unavailable 8Ks, information from Item 9 of Form 10-K (Disagreements with and Changes in Auditors) is used. The primary source of financial, auditor report and tenure data is Compustat, supplemented with information from Lexis-Nexis, compact disclosure, Moody’s databases or financial statement micro-fiche. The measurement date for these variables is the fiscal year end immediately prior to the switch. IPO data is from The Wall Street Journal’s “New Securities” section. Litigation information is from the database described in Palmrose (1994) and includes information from Investor Class Action Monitors and Securities Class Action Alert (April 1988 to December 1995), legal notices in the Wall Street Journal (1985 to 1995) and Dow Jones News Retrieval.

5. RESULTS Summary statistics and univariate results are presented in Table 2. Results of four comparison tests are provided: differences across all three groups, between noswitch observations and dismissals and resignations, and between dismissals and resignations.

Auditor Resignations, Litigation Risk and Litigation Experience

181

Table 2. Summary Statistics and Results of ANOVA, t-Test and Chi-squared Analysis for Litigation Risk and Control Variables for No-Switch (Matched), Resignation and Dismissal Companies. No-switch

Dismiss

Resign

n

194

160

34

Financial condition variables Net loss reporteda,b,c,d Number Percent

89 46%

90 56%

27 79%

Bankruptcya,b,c Number Percent

7 4%

18 11%

6 18%

Debt/asset ratioa,b,c Mean Std. dev. Median

0.62 0.94 0.38

0.73 0.73 0.52

1.07 1.63 0.55

Report typea,b,c,d Unmodified Percent Uncertainty modification Percent Going concern modification Percent

159 82% 28 14% 7 4%

100 63% 23 14% 37 23%

13 38% 7 21% 14 41%

40 21%

28 18%

10 29%

Auditor litigationc Number Percent

5 3%

9 6%

3 9%

Non-auditor litigation Number Percent

35 18%

19 12%

7 21%

Tenurea,b,c Mean Std. dev. Median

6.58 3.64 6.00

4.62 3.16 4.00

4.39 3.42 4.00

SEC AAER Number Percent

4 2%

7 4%

1 3%

Pre-audit financial statement variables Litigation Number Percent

182

SUSAN SCHOLZ

Table 2. (Continued ) No-switch

Dismiss

Resign

Disagreements with management Count n/a Percent

16 10%

5 15%

Unreliable managementd Count Percent

1 1%

5 15%

Financial statements withdrawn/restatedd Count n/a Percent

8 5%

6 18%

Revealed auditor concernsd Count Percent

n/a

26 16%

9 26%

Internal control weaknessd Count Percent

n/a

10 6%

5 15%

3.33 1.94 3.50

1.97 1.57 1.83

1.81 1.71 1.69

% change in sales Mean Std. dev. Median

0.30 1.25 0.13

1.11 6.02 0.10

0.52 2.33 0.08

IPOa,b Count Percent

28 14%

11 7%

2 6%

Big 6 auditora,b,c Count Percent

177 91%

108 68%

26 76%

Control variables Ln(assets)a,b,c Mean Std. dev. Median

a Difference

n/a

across all three groups is significant at 0.10 level or better. between no-switch and dismissal is significant at 0.10 level or better. c Difference between no-switch and resignation is significant at 0.10 level or better. d Difference between dismissal and resignation is significant at 0.10 level or better. b Difference

Auditor Resignations, Litigation Risk and Litigation Experience

183

Many of the variables differ significantly between switch and no-switch groups. Switching companies (dismissals and resignations) report more losses, have more bankruptcies, modified reports and higher debt ratios. They are also smaller and more likely to be represented by a non-Big 6 auditor. They have shorter auditor tenure and fewer recent IPOs. Resignations and dismissals do not show many significant differences. Only net losses and modified reports are significantly higher for resignations than for dismissals. However, the switch-only variables from Form 8-K (management disagreements, withdrawn financial statements, etc.) indicate that revelations of unreliable management, withdrawn financial statements and poor internal controls are more prevalent for resignation than dismissal companies. The actual litigation variables do not differ significantly across groups, except for a higher frequency of auditor litigation for resignation clients vs. no-switch clients. Still, it is interesting to note that while resignation clients have the highest litigation frequency (29%), dismissal clients are more similar to no-switch clients (18% and 21% respectively).

5.1. Matched Sample Logit Analysis Results and Discussion Results for the logit model used to evaluate the association of the litigation risk variables with dismissals and resignations are presented in Table 3. Correlation coefficients for the model variables are in Table 4. Four different partitions of the sample are evaluated:5 Partition 1: Switch companies vs. no-switch companies; Partition 2: Dismissal companies vs. no-switch companies; Partition 3: Resignation companies vs. no-switch companies; Partition 4: Resignation companies vs. dismissal companies. Results for partitions 1–3 are highly significant ( p = 0.00); indicating that the model can effectively distinguish between no-switch observations and both types of switches. All significant coefficients are consistent with expectations. Switches are positively associated with report type and actual litigation. Based on the litigation frequencies noted in Table 2, and the weaker results for the variable in the dismissal partition (regression 2) it appears resignations are mainly responsible for the actual litigation. Tenure is negatively associated with dismissals and has a similar, but weaker relationship with resignations. This differs from Krishnan and Krishnan (1997)

184

SUSAN SCHOLZ

Table 3. Logit Results of Litigation Risk Model for Various Partitions of the Sample. Dependent variable

Partition 1

Partition 2

Partition 3

Partition 4

Switch = 1 v. no switch

Dismissal = 1 v. no switch

Resign = 1 v. no switch

Resign = 1 v. dismissal

Model statistics Sample size Model chi-squared value p-Value Correctly classified Prior probability

388 117.06 0.00** 70.36% 50%

354 97.74 0.00** 71.19% 55%

228 65.67 0.00** 89.91% 85%

194 15.63 0.16 82.99% 82%

Financial condition variables Bankruptcy Net loss Debt/asset Report type

Coefficient +0.7876 ±0.3366 ±0.1976 +0.7668**

Coefficient 0.8737 −0.4005 −0.2529* 0.6685**

Coefficient 1.2060* 0.5566 −0.1842 1.2650**

Coefficient 0.1810 0.6971 0.1725 0.3948

Pre-audit financial statement quality variables Litigation +0.8487** Tenure −0.1313** AAER +0.3896

0.6651* −0.1251** 0.3314

1.9821** −0.1475* 0.8122

0.9054* −0.0483 −0.9405

Control variables Ln(assets) % change in sales IPO Big 6 auditor

−0.3701** 0.0281 −1.1145** −0.7670**

−0.3507** 0.0310 −1.0398** −0.8467**

−0.4572** 0.0253 −1.2678 −0.4348

0.0110 −0.0194 −0.5604 0.3983

Constant

2.2011**

2.1482**

−0.8774

−2.7479**

∗ Coefficient

is significant at 0.10 level or better. is significant at 0.05 level or better.

∗∗ Coefficient

who find resignations to be more strongly associated with shorter tenure than dismissals. Control variables indicate a negative relationship between switching and client size. This is evidence that even though larger companies may be more likely to experience litigation (Lys & Watts, 1994; Stice, 1991), they are less likely to experience either dismissals or resignations. This finding is consistent with Krishnan and Krishnan (1997). IPOs and Big 6 auditors are negatively associated with dismissals, but not related to resignations. Thus, Big 6 auditors appear to be less likely to be dismissed by their clients. However, auditor type does not appear to affect resignation behavior. This also differs from Krishnan and Krishnan (1997) who generally find that Big 6 auditors are positively related to resignations, but not associated with dismissals.

Auditor Resignations, Litigation Risk and Litigation Experience

185

Results from partition 4 show that the model does not distinguish between dismissals and resignations ( p = 0.16). Overall classification accuracy is equal to the prior probability at about 82%. In effect, the model classifies all observations as dismissals. The only variable which is even marginally significant is litigation (positive). Sensitivity analysis is performed to address the conflicting results for the financial condition variables in univariate results (generally significant) and logit results (generally insignificant). Due to the high levels of correlation among this set of variables (see Table 4), the model was tested using each of the financial condition variables in turn. Report type is highly significant when used alone ( p < 0.05) in all partitions. Bankruptcy is significant ( p < 0.05) in partitions 1–3 and net loss was significant ( p < 0.05) in partitions 3–4 (all coefficients positive). The overall results for the other variables are substantially the same. Thus, it appears that some measures of financial condition (i.e. report type and net loss) do distinguish resignations from dismissals. A general observation from this analysis is that dismissal and resignations are quite similar. This result has two possible interpretations: (1) The model does not capture litigation risk, but instead captures other factors associated with dismissals; and resignations are substantially the same. (2) The model does capture litigation risk, but the variables do not have sufficient power to distinguish explicit resignations from the implicit resignations likely to be in the dismissal group. To evaluate the relationship of the model with litigation risk, the model variables (except litigation) were regressed on actual litigation. The result of this regression (not shown) indicate that the model is significantly associated with all measures of actual litigation (any litigation, auditor and non-auditor) at the 0.00 level. Measures representing both poor financial condition and low pre-audit financial statement quality are significant. Furthermore, when a switch/no switch indicator variable is added as an explanatory variable, it is significantly associated with litigation (positive). These results suggest that the model captures actual litigation risk, although actual and perceived litigation risk may differ. The distinction between resignations and dismissals is explored further in the next section.

5.2. Switch Sample Logit Model and Discussion The switch sample provides several additional variables related to issues of management integrity and internal controls. Results from the matched sample model

186

Table 4. Pearson Correlation Coefficients ( p-Values) for Litigation Risk Variables. SEC AAER Litigation Tenure Bankruptcy Net loss Debt/asset Report type IPO Ln(Asset) Sales change

Tenure

Bankruptcy

Net loss

−0.01 (0.83) 0.04 (0.43) −0.05 (0.32) −0.10 (0.05) −0.01 (0.81) 0.14 (0.01) 0.41 (0.00) −0.03 (0.55) 0.19 (0.00)

−0.01 (0.89) −0.12 (0.02) −0.01 (0.83) −0.13 (0.01) −0.27 (0.00) 0.31 (0.00) −0.10 (0.05) 0.25 (0.00)

−0.14 (0.01) 0.07 (0.16) 0.31 (0.00) −0.01 (0.87) −0.11 (0.02) 0.10 (0.05) −0.04 (0.39)

0.22 (0.00) 0.37 (0.00) −0.13 (0.01) −0.34 (0.00) 0.02 (0.69) −0.05 (0.30)

Debt/asset

0.34 (0.00) −0.07 (0.14) −0.36 (0.00) −0.06 (0.23) −0.01 (0.88)

Report type

IPO

−0.05 (0.33) −0.31 (0.00) 0.03 (0.58) −0.12 (0.02)

0.14 (0.01) 0.01 (0.87) 0.15 (0.00)

Ln(assets) Sales change

−0.03 (0.50) 0.35 (0.00)

−0.12 (0.02)

SUSAN SCHOLZ

Big 6 auditor

0.21 (0.00) 0.01 (0.88) 0.17 (0.00) 0.02 (0.71) −0.03 (0.55) 0.05 (0.30) 0.04 (0.49) 0.07 (0.19) 0.17 (0.00) −0.06 (0.24)

Litigation

Auditor Resignations, Litigation Risk and Litigation Experience

187

are used to develop switch models to explore the marginal difference between dismissal and resignation companies. As reported above, the full model is insignificant when applied to the switch sample; and only litigation shows any relationship to resignations. The sensitivity analysis indicates that report type is also a significant factor when it is included as the only financial condition variable.6 Therefore, the switch models begins with these two variables (Model S1). From this point the model is gradually expanded to explore the role of the 8-K information. Model S2 adds the summary variable Revealed Auditor Concerns (RAC) which equals one if any indication of low management integrity or poor controls was revealed in Form 8-K.7 Model S3 separates the component variables of litigation (auditor litigation and other litigation) and RAC (management disagreements, unreliable management, withdrawn/restated financial statements & internal control weakness.) These three models are presented in Table 5.

Table 5. Logit Results of Risk Models for the Switch (Dismissals and Resignations) Sample. Dependent variable

Resign = 1 v. dismissal Model S1

Model S2

Model S3

194 9.307 0.01** 82.47% 82.5%

194 10.258 0.02** 82.47% 82.5%

194 19.83 0.01** 85.05% 82.5%

Explanatory variables Report type Litigation Other litigation Auditor litigation Revealed auditor concern Management disagreement Unreliable management F/S withdrawn/restated Internal control weakness

Coefficient 0.5673** 0.7672*

Coefficient 0.5366** 0.6157

Coefficient 0.4713**

Constant

–2.1875**

Model statistics n Chi-square p-Value Correctly classified Prior probability

∗ Coefficient

0.5249 –1.1980 0.4722 –0.1762 3.194** 0.9566 0.0712

is significant at 0.10 level or better. is significant at 0.05 level or better.

∗∗ Coefficient

–2.2296**

–2.1430**

188

SUSAN SCHOLZ

All models are significant at better than the 0.05 level. Model S3 exhibits the best fit, although it still correctly classifies of only 18% of resignations. Report type and litigation are both significant and positively associated with resignations in Model S1.8 However, when RAC is added to the analysis (Model S2), only report type is significant. When litigation and RAC are split into their components (Model S3), management unreliability is significant along with report type. Both coefficients are positive, as expected. The significance of report type across all models highlights the importance of the auditor’s perception of the client’s financial prospects in the resignation decision. It appears that the significance of litigation (due primarily to other litigation) in Model S1 is transferred to revelations of management unreliability in Model S3. This supports the hypothesis that litigation (significant in all partitions of the full model) is associated with the auditor’s perception of management’s integrity. It is also generally consistent with Krishnan and Krishnan’s (1997) finding that disagreements and other reportable events disclosed in 8-Ks are associated with resignations. Since unreliable management is one of the few variables which distinguishes between resignations and dismissals, it is interesting to note that there is a high correlation between indications of management unreliability and actual auditor litigation (31%, p-value = 0.00) (correlation table not shown). It appears that indications of management unreliability are a reasonable indicator of a high likelihood of auditor litigation. Diagnostic tests of these models identified two influence points for auditor litigation and management disagreements which affect the results of Model S3. If the observations are eliminated, overall model significance is not affected, but management unreliability loses its significance, leaving only report type significant. Thus, although measures of internal controls are never significant and the significance of measures of management integrity (mainly unreliable management and litigation) is somewhat unstable; the statistical significance of report type, representing poor client financial condition is robust. These results provide only limited support for the expectation that low management integrity is a primary factor in distinguishing explicit resignations from dismissals. It is possible that 8-K revelations are weak measures of these concepts. Certainly, auditors possess much detailed information, and any disclosure in Form 8-K is largely discretionary. Another explanation is that modified reports are indeed a more heavily weighted consideration. A negative evaluation of the client’s financial prospects impacts both sides of the cost/benefit equation. It indicates higher expected costs through increased litigation risk (due to the risk of investor losses related to financial failure); plus lower expected benefits, due to the client’s potential inability to pay

Auditor Resignations, Litigation Risk and Litigation Experience

189

fees. On the other hand, low quality pre-audit financial statements are related mainly to higher expected litigation costs, and do not necessarily affect expected client benefits. Additional analyses using the actual litigation data are used to investigate these possibilities.

6. ADDITIONAL ANALYSIS: ACTUAL LITIGATION To evaluate the cost-benefit trade-off between retaining clients with high litigation risk, I use actual litigation as an indication of demonstrably high-risk companies. First, I focus on the no-switch sample, comparing clients that experienced litigation to those that did not. The no-switch/litigation observations provide a setting to evaluate when auditors perceive the benefit of retaining clients to be greater than the expected cost of litigation. The matched (no-switch) sample has 40 (21%) litigation and 154 (79%) no-litigation observations (see Table 2). A comparison of the no-switch companies that did have litigation to those that did not indicates that the litigation companies report significantly fewer net losses (50% vs. 30%) and are significantly larger (mean ln(assets) = 5.1 vs. 2.9). Thus, litigation clients that avoid switches are large and profitable, even compared to no-switch/no-litigation companies. Second, I compare characteristics of the 78 litigation observations across groups (40 no-switch, 28 dismissal, 10 resignation, see Table 2). Again, the comparison confirms that no-switch clients are larger (mean ln(assets) is 5.1 vs. about 3.3) and in better financial condition than switching companies. They report fewer losses (30% vs. 65%) and have fewer bankruptcies (0% vs. 18%). Dismissal and resignation rates are similar. Logit analysis (not shown) primarily emphasizes the significance of size. Smaller companies are more likely to experience both dismissals and resignations. Overall, these results provide additional evidence that the expected benefits of continuing with a client are an important part of the client continuance decision, even when the expected costs of litigation risk are particularly high. They suggest that auditors do not terminate client relationships unless both litigation risk is high and expected benefits of continuing with the client are relatively low.9

7. SUMMARY AND CONCLUSIONS The primary purpose of this study is to test the association between litigation risk and auditor resignations. Auditors claim that resignations are primarily

190

SUSAN SCHOLZ

attributable to a legal environment that over-deters auditor behavior. That is, rather than deterring sub-standard audits, the legal environment causes auditors to discard risky clients, imposing costs on both clients and auditors. This study provides evidence supporting a necessary condition of the auditors’ claims, i.e. that resignations are indeed related to litigation risk. It also generally confirms results reported by Krishnan and Krishnan (1997). In addition, it uses realized litigation risk to provide some insights into whether resignations are attributable to over-deterrence. The litigation risk model consistently distinguishes no-switch clients from switch clients, but usually does not discriminate between resignations and dismissals. This is consistent with the existence of implicit resignations within the dismissal sample. The most robust risk variables are report type, representing financial condition risk; and actual litigation, representing pre-audit financial statement risk. Form 8-K information, available only for switching clients, indicates that (relative to dismissals) resignations are associated with unreliable management, another proxy for poor pre-audit financial statement quality. However, it is not necessarily indicative of over-deterrence, since audits are not expected to consistently detect fraudulent misstatements. Thus, if an auditor suspects that management is unreliable, a resignation would be an appropriate, not inefficient step. The significance of modified reporting can be interpreted as support for auditors’ claims of over-deterrence, since the purpose of laws governing auditor litigation is not to deprive clients in poor financial condition of their auditors. However, the robust significance of this measure is likely attributable to its relationship with both sides of the client cost/benefit calculation. Litigation risk increases with the increased probability of potential user losses while expected benefits decrease when the auditor perceives that a client will not continue to be a profitable engagement. Additional analysis of characteristics of client experiencing litigation, the highest level of risk, also suggests that high litigation risk is not sufficient to cause a resignation. Large, profitable clients are less likely to experience resignations (or dismissals). This evidence can be interpreted to suggest that resignations are due to auditors discarding unprofitable clients. On the other hand, auditors may react more to litigation risk attached to smaller, less profitable clients to avoid being named as a deep pocket defendant. Overall, although the data provide evidence of an association between litigation risk and resignations, it does not unambiguously support auditors’ claims that the legal environment causes auditors to resign from clients they would otherwise serve.

Auditor Resignations, Litigation Risk and Litigation Experience

191

NOTES 1. The computer industry is defined as software and hardware. SIC codes are listed later. 2. Investors are defined to be users of audited financial statements. 3. This information is based on reviews of confidential client acceptance programs obtained from several accounting firms. 4. A consistency modification notes that the client has (appropriately) changed an accounting policy. An uncertainty modification states that there is a possible material cost to the company that has not been recorded in the financial statements due to uncertainty about outcome or amount. A going concern opinion states that the auditor believes the company may not be able to continue as a going concern through the next year. 5. This methodology is similar to that used by Carcello and Palmrose (1994). 6. Net loss is significant in this sensitivity analysis, too, but not as consistently as report type. Report type is also preferred because it is a direct indication of the auditor’s assessment of the client’s prospects. 7. As an initial test, RAC was added to the full model, partition 4. Neither the overall model nor any of the variables were significant. 8. If litigation is split into auditor and other litigation, report type ( p = 0.01) and other litigation ( p = 0.08) are significant, but not auditor litigation. 9. Eight of the 10 resignations with litigation (80%) occurred after litigation was filed, compared to 16 of the 28 dismissals (57%). So, resignation may be a legal tactic rather than a response to increased risk assessments.

ACKNOWLEDGMENTS This study is from my dissertation (written at the University of Southern California), and so was greatly aided and abetted by my committee members: Zoe-Vonna Palmrose (chair), Mark Defond, Ted Mock and Tim Cason. I am particularly indebted to Professor Palmrose who not only provided patient and excellent guidance, but also access to her litigation database. Participants at the 1997 International Audit Research Symposium and University of Kansas research workshop also supplied helpful comments. Finally, Mark Hirschey offered invaluable encouragement and advice. Of course, any errors are my own.

REFERENCES American institute of certified public accountants (1988). Statement on auditing standards No. 60: Communication of internal control structure related matters noted in an Audit. New York: AICPA. American institute of certified public accountants (1995). Audit risk alert-1995/96: General update on economic, accounting and auditing matters. New York: AICPA.

192

SUSAN SCHOLZ

Andersen, A., Coopers & Lybrand, Deloitte & Touche, Ernst & Young, KPMG Peat Marwick and Price Waterhouse, (1992). The liability crisis in the United States: Impact on the accounting profession. Statement of position (August 6). Bockus, K., & Gigler, F. (1998). A theory of auditor resignation. Journal of Accounting Research, 36, 191–208. Bunsis, H. J., & Drake, P. D. (1995). The decline of accountants as defendants in initial public offering lawsuits: An empirical analysis, Southern Methodist University working paper. Carcello, J. V., & Palmrose, Z. V. (1994). Auditor litigation and modified reporting on bankrupt clients. Journal of Accounting Research (Supplement), 1–31. DeAngelo, L. E. (1981). Auditor independence, ‘low-balling’, and disclosure regulation. Journal of Accounting and Economics, 3, 113–127. DeFond, M. L., Ettredge, M., & Smith, D. B. (1997). An investigation of auditor resignations. Research in Accounting Regulation, 11, 25–46. Feroz, E. H., Park, K., & Pastena, V. S. (1991). The financial and market effects of the SEC’s accounting and auditing enforcement releases. Journal of Accounting Research (Supplement), 107–142. Francis, J., Philbrik, D., & Schipper, K. (1994). Shareholder litigation and corporate disclosures. Journal of Accounting Research, 32, 137–164. Kellogg, R. L. (1984). Accounting activities, security prices, and class action lawsuits. Journal of Accounting and Economics, 6, 185–204. Kinney, W. R. (1993). Auditors’ liability: Opportunities for research. Journal of Economics and Management Strategy, 2, 349–360. Krishnan, J. (1994). Auditor switching and conservatism. The Accounting Review, 69, 200–215. Krishnan, J., & Krishnan, J. (1997). Litigation risk and auditor resignations. The Accounting Review, 72, 539–560. Lys, T., & Watts, R. L. (1994). Lawsuits against auditors. Journal of Accounting Research, 32(Supplement), 65–93. Menon, K., & Williams, D. (1991). Auditor credibility and initial public offerings. Accounting Review, XX, 313–333. Palmrose, Z. V. (1987). Litigation and independent auditors, the role of business failures and management fraud. Auditing: A Journal of Practice and Theory, 6, 90–103. Palmrose, Z. V. (1988). An analysis of auditor litigation and audit service quality. The Accounting Review, 63, 55–73. Palmrose, Z. V. (1991). Trials of legal disputes involving independent auditors: Some empirical evidence. Journal of Accounting Research, 29(Supplement), 149–185. Palmrose, Z. V. (1994). The joint and several versus proportionate liability debate: An empirical investigation of audit-related litigation. Stanford Journal of Law, Business & Finance, 53–72. Pratt, J., & Stice, J. D. (1994). The effects of client characteristics on auditor litigation risk judgments. The Accounting Review, 69, 639–656. Shavell, S. (1987). Economic analysis of accident law. Cambridge, MA and London, U.K.: Harvard University Press. Shu, S. Z. (2000). Auditor resignations: Clientele effects and legal liability. Journal of Accounting and Economics, 29, 173–205. Simonetti, G., Jr., & Andrews, A. R. (1994). A profession at risk/a system in jeopardy. Journal of Accountancy, XX, 45–64.

Auditor Resignations, Litigation Risk and Litigation Experience

193

Simunic, D. A., & Stein, M. T. (1995). The impact of litigation risk on audit pricing: A review of the economics and the evidence. Conference on auditor and director and officer liability for misstated financial statements. John M. Olin School of business, Washington University. Stice, J. D. (1991). Using financial and market information to identify pre-engagement factors associated with lawsuits against auditors. The Accounting Review, 66, 516–533. St. Pierre, K., & Anderson, J. (1984). An analysis of the factors associated with lawsuits against public accountants. The Accounting Review, 59, 242–263. Wells, D., & Loudder, M. (1997). The market effects of auditor resignations. Auditing: A Journal of Practice and Theory, 16, 138–144.