The Other Side of Gridlock Policy Stability and Supermajoritarianism in U.S. Lawmaking
Manabu Saeki
..
n n n n n n n ...
40 downloads
432 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
The Other Side of Gridlock Policy Stability and Supermajoritarianism in U.S. Lawmaking
Manabu Saeki
..
n n n n n n n n n n n n n n n n n n n n n n n
...
..
...
... .
...
...
... ..
...
...
............. . . . .. .. .. ... ................ . . . ...... . .. .... ..... ................ . . . . ..... ..... . . . . . . . . . . . . . . . . . . . . . . . . . . . .......... . . . .. . . . . . . . . ... . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .... . . . . . . . . . . ... . . . . ........ . . . . ...... . . . . . . . . . . . . . . . . . . . . . . . . ...................... . . ....... . . . . . . ..... ... ...... .. . . . . ........ ... . . ............................... ... ... . . . . . .... .. . . . .... . . . . . ... . ...... . . . . . . ......... . ....... . . . . . . . . . . . . . . . . . . . . . . . . .............. . .......... .... . . . . . .......... .................
...
.. . . . . . . . . . . . . . . . . . . . .. . . . ......... . . . . . . . . . . . . . . . ......... . . . . .... . . . . . ..... ... ...... .... . . . . . . . . . . . . . . . ....... . . ...................... . . .. . . . . . . .
.....
The Other Side of
Gridlock
This page intentionally left blank.
The Other Side of
Gridlock Policy Stability and Supermajoritarianism in U.S. Lawmaking
Manabu Saeki
Published by State University of New York Press, Albany © 2010 State University of New York All rights reserved Printed in the United States of America No part of this book may be used or reproduced in any manner whatsoever without written permission. No part of this book may be stored in a retrieval system or transmitted in any form or by any means including electronic, electrostatic, magnetic tape, mechanical, photocopying, recording, or otherwise without the prior permission in writing of the publisher. For information, contact State University of New York Press, Albany, NY www.sunypress.edu Production by Ryan Morris Marketing by Anne M. Valentine Book design and typesetting: Jack Donner, BookType Library of Congress Cataloging-in-Publication Data Saeki, Manabu. The other side of gridlock : policy stability and supermajoritarianism in U.S. lawmaking / Manabu Saeki. p. cm. Includes bibliographical references and index. ISBN 978-1-4384-3051-5 (hardcover : alk. paper) 1. Policy sciences. 2. Political planning—United States. 3. United States. Congress—Rules and practice. 4. Parliamentary practice—United States. I. Title. JK468.P64S33 2010 328.73—dc22 2009022694 10 9 8 7 6 5 4 3 2 1
To Steven A. Shull, my mentor
This page intentionally left blank.
contents
List of Illustrations
ix
Preface
xiii
Chapter 1
Introduction
1
Chapter 2
Gridlock and Policy Stability
19
Chapter 3
Pivotal Interval Movement
35
Chapter 4
Empirical Test
61
Chapter 5
Veto Players
81
Chapter 6
Pre-Floor Agenda Block
95
Chapter 7
Conclusion
119
Notes
129
References
135
Index
143
vii
This page intentionally left blank.
illustrations
Tables Table 1.1 Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 2.5
Majority Party Unity Scores in Senate 1991–2000 Correlations among Variables of Legislative Productivity Mean Values and Variations in Legislative Productivity Determinants of Legislative Productivity 1953–2000 Binder Agenda Denominator Correlations between the Divided Government and the Volume of Legislative Initiatives Table 2.6 Mean Values and Variations in Policy Change Table 2.7 Policy Changes and Changes in Majority Party and Presidential Party Table 4.1 Residuum Descriptive Statistics Table 4.2 Residuum Correlations Table 4.3 Senate Policy Change and Bicameral Residuum Table 4.4 Senate Policy Change and Unicameral Residuum Table 4.5 House Policy Change and Bicameral Residuum Table 4.6 House Policy Change and Unicameral Residuum Table 4.7 Weighted Senate Policy Change and Bicameral Residuum Table 4.8 Weighted Senate Policy Change and Unicameral Residuum Table 4.9 Weighted House Policy Change and Bicameral Residuum Table 4.10 Weighted House Policy Change and Unicameral Residuum Table 5.1 Congressional Policy Change Table 5.2 Congressional Weighted Policy Change Table 6.1 Percentage Opposed Bills among Party Votes
ix
12 23 24 25 26 26 33 34 71 71 72 73 74 75 76 77 77 78 92 93 110
x Table 6.2 Table 6.3 Table 6.4 Table 7.1
Illustrations Percentage Agendas in Final-Passage Vote Percentage Bipartisan Support in Final-Passage Vote Opposed Bills and Partisan Roll Theories of Policy Stability and Their Implications
111 115 116 121
Figures Figure 1.1 Figure 1.2 Figure 1.3 Figure 2.1 Figure 2.2 Figure 2.3 Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 3.7 Figure 3.8 Figure 3.9 Figure 3.10 Figure 3.11 Figure 3.12 Figure 3.13 Figure 3.14 Figure 3.15 Figure 3.16 Figure 3.17 Figure 3.18 Figure 3.19 Figure 3.20 Figure 4.1 Figure 4.2 Figure 4.3 Figure 4.4
Party Unity Scores: House 1954–2006 Party Unity Scores: Senate 1954–2006 Percentage Party-Split Vote on Roll Calls 1954–2006 Gridlock in Legislative Productivity, 83rd-106th Congress Policy Change and Weighed Policy Change: House Policy Change and Weighed Policy Change: Senate Macro Model of Policymaking Unidimensional Policy Space Utility Function for Legislators Utility Function of Tax Cut Agenda Status Quo and Agendas of Minimum Wage Twenty Members in a Legislature Voting Decisions of Twenty Legislators Median Voter Theory Filibuster Gridlock Interval Veto Gridlock Pivot Pivotal Gridlock Interval (Conservative President) Pivotal Gridlock Interval (Liberal President) Pivotal Gridlock Interval and Policy Change Movement of Pivotal Gridlock Interval and Policy Change Cartel Gridlock Interval and Policy Change Movement of Cartel Gridlock Interval and Policy Change Bicameral Pivotal Model Bicameral Pivotal Gridlock Interval Bicameral Cartel Model Bicameral Cartel Gridlock Interval Pivotal Bicameral Gridlock Intervals Senate Pivotal Gridlock Intervals House Pivotal Gridlock Intervals Bicameral Cartel Gridlock Intervals
11 11 13 22 31 32 37 40 41 42 42 43 43 44 46 48 48 49 50 52 54 55 57 57 58 58 65 66 67 68
Illustrations Figure 4.5 Figure 4.6 Figure 4.7 Figure 5.1 Figure 5.2 Figure 5.3 Figure 6.1 Figure 6.2 Figure 6.3 Figure 6.4 Figure 6.5 Figure 6.6 Figure 6.7 Figure 6.8 Figure 6.9 Figure 6.10 Figure 6.11
Senate Cartel Gridlock Intervals House Cartel Gridlock Intervals Pivotal and Cartel Bicameral Residuum Circular Indifference Curve of Pivotal Legislator Pivotal Veto Players in the 104th Congress Veto Players’ Indifference Curves 83rd-109th Congresses Vote Alignments Floor Agenda Model Majority Roll 1 Majority Roll 2 Bipartisan Support Cartel Block Model Pivotal Block Model Final-Passage Vote and Legislative Agendas Status Quo 83rd-105th Congresses Majority Party Members Status Quo and Responses of Centrist Members in the Majority Party Figure 6.12 Roll Aversion Hypothesis Figure 7.1 Bicameral Pivotal Gridlock Interval
xi 69 69 70 83 84 87 98 102 104 104 104 105 106 108 108 112 113 113 125
This page intentionally left blank.
p re face
This book is about policy stability, or lack of policy change, in the U.S. government. When I was a graduate student a few years ago, the potential relationship between divided government and the so-called gridlock was a ubiquitous and controversial issue in such courses as public policy, public administration, political parties, presidency, Congress, presidential-congressional relations, and even research methods! I was interested in the issue at that time, and was especially impressed by the works of Mayhew (1991), Krehbiel (1998), and Brady and Volden (1998). However, I was taking many courses and had to deal with other academic issues (and admittedly nonacademic issues) as well. After I became a university professor and started teaching legislative politics, I was fascinated by the theoretical, as well as the empirical, controversies pertaining to divided government and gridlock. When I read Setting the Agenda by Cox and McCubbins (2005), I thought that the partisan model and the nonpartisan supermajority model should be empirically compared, thereby leading me to write this book. As I will detail in this book, since David Mayhew’s Divided We Govern (1991), the politics of gridlock has become one of the most salient and controversial issues in political science. By compiling a list of significant legislative enactments during the period from 1947 through 1990, Mayhew challenged conventional wisdom and argued that divided government did not decrease the amount of important legislation enacted in the United States. Scholars in the post-Mayhewian wave have questioned Mayhew’s measure of the volume of enacted laws; they have alternately explored a ratio of the enacted legislation to the entire legislative agenda, and allegedly found an impact of divided government (Binder 1999, 2003; Coleman 1999; Edwards and Barrett 2000). On the other hand, scholars of the supermajoritarian school have contended that gridlock could occur in unified as well as divided government because the proximity between the status quo and preferences of pivotal legislators could cause gridlock (Krehbiel 1996, 1998; Brady and Volden 1998; Chiou and Rothenberg 2003). In spite of the vast number of studies of gridlock, most studies have focused on various quantifications of legislative productivity as measures of gridlock. xiii
xiv
Preface
In contrast, this book shifts the target of study from the amount of laws to policy output. By following the definition of gridlock as an inability to change policy (Kernell 1991; Krehbiel 1998; Brady and Volden 1998; Chiou and Rothenberg 2003; Tsebelis 2002), this book aims to explain policy change, or inversely policy stability. In order to compare the influences of divided government, the majority party in Congress, and the preferences of supermajoritarian pivotal legislators on policy change, applied models of gridlock interval movement, veto players, and pre-floor agenda block are constructed in this book. These models generate hypotheses pertinent to the influence of predictors, and the hypotheses are empirically tested. I wrote most of the manuscript in summer 2007 and had a great time. While I wrote in my house, I enjoyed the New Orleans’ Rue de la Course Espresso coffee (no, I don’t drink sake in the day time, unlike a few congressional scholars I know) and the music flowing from my Sony speakers. There is no doubt that Miles, Bird, Dolphy, Woody Shaw, Wallace Roney, Geri Allen, and others inspired my work. Friends and colleagues greatly contributed to this project. I would like to acknowledge Roy Bonnette, Keith Krehbiel, Sang Lee, David Mayhew, Margaret Gonzalez-Perez, Keith Poole, and Corina Schulze for their helpful suggestions. Also, I thank Wei Wei Hsing, Jason Husser, and Amritendu Maji for their research assistance; and Sarah Binder, George Edwards, and David Rhode for access to their data.
Chapter 1
Introduction I’ve earned capital in this election—and I’m going to spend it for what I told the people I’d spend it on, which is. . . . Social Security and tax reform, moving this economy forward, education, fighting and winning the war on terror. —President George W. Bush, November 4, 2004
The campaign is over. Democrats are ready to lead, prepared to govern and absolutely willing to work in a bipartisan way. —Nancy Pelosi, November 8, 2006
On November 4, 2004, two days after the GOP triumph both in the presidential and Congressional elections, the reelected president George W. Bush held a press conference and emphasized what he considered to be a mandate emanating from the electoral results. Bush, who had won barely 51 percent of the popular vote, aggressively proclaimed that “the people have spoken and embraced your [his] point of view, and that’s what I [he] intend to tell the Congress.” Bush continued and stated that he would use political capital he had allegedly earned in order to pursue his policy agendas. Ironically, Bush’s optimistic and simplistic view of the political landscape in his relationship with Congress reminded several pundits of the uneasy relationship between the former president Clinton and his first Congress. Clinton’s election in 1992 ended the twelve-year-long divided government, and there was an enormous expectation that the first unified government since 1980 would turn out decisive action and adopt innovative programs. In embracing the public’s expectation, Clinton announced that he anticipated his first hundred days to be the most productive period since Franklin Roosevelt (Fiorina 1996, 159). However, by 1994, the optimism surrounding the unified government was found to be wrong in everyone’s eyes. In the 103rd Congress, Clinton’s health care plan never made it to the floor, and he abandoned his proposal of the energy tax in the 1993 budget reconciliation bill. In addition, Republican filibusters successfully killed many of Clinton’s legislative initiatives, including the economic stimulus package and campaign finance reform bill. 1
2
The Other Side of Gridlock
As for the 109th Congress after the 2004 election, parallel to the 103rd Congress, President Bush suffered the Clintonite legislative quandary in the Republican unified government. Although Bush, in his second term, advocated swift policy changes including the partial privatization of Social Security, the guest worker program, antiterrorism surveillance, and extension of the 2001 and 2003 tax cuts, only the latter two were enacted in a modified version. The indication is that a factor other than divided government leads to legislative stalemate. David Mayhew’s Divided We Govern (1991) examined the amount of significant laws enacted, and remarkably challenged the myth of an adversarial effect of divided government on governmental effectiveness. By extending his analysis of the amount of important laws to the year 2002, Mayhew (2005) has recently reasserted his claim that party control of government has no impact on legislative productivity. Further, scholars of the preference-based school have contended that preferences of individual legislators, rather than party control of the government, influence legislative productivity (Krehbiel 1996, 1998; Brady and Volden 1998, 2006). Specifically, the researchers observe that a passage of legislation needs the support of the supermajority of legislators, otherwise a minority of legislators would block the bill by mounting a filibuster or extracting and supporting the presidential veto. This book extends and tests the assumption of nonpartisan, supermajoritarian lawmaking in the U.S. Congress. Presented herein is the theory that a sizable change from one Congress to the next in the preferences of the minority legislators who are ideologically more extreme decreases the potential of gridlock. Nonetheless, this book does not focus on the amount of enacted laws as a measure of legislative stalemate. If several significant policies are packed in a few omnibus bills, the modest quantity of the omnibus measures underestimates the significance of their policy output. Thus, the ratio or number of enacted bills does not suggest the significance of policy output by Congress. Accordingly, this book embraces a definition of gridlock as an inability to change policy, and attempts to explain policy change, or inversely policy stability, in legislative output. 109th Congress
In his State of the Union speech in January 2005, President Bush revealed the agendas for his second term, including the creation of individual accounts in the Social Security program, the guest worker program, the extension of the tax cuts of 2001 and 2003, and the reauthorization of the antiterrorism surveillance program. Among these agendas, the bills of social security reform and the guest worker program were not enacted because they failed to gain support
Introduction
3
from a supermajority of the members in Congress. The tax cuts extension and reauthorization of the Patriot Act were diluted to win support from moderate members in Congress, and eventually passed. Social Security Reform
President Bush’s Social Security reform plan would introduce individual investment accounts for workers younger than fifty-five years of age. Up to 4 percent of the workers’ wages could be apportioned in the accounts, and the account holders could invest the allocated funds in the stock market. Later, Bush admitted that his plan would also reduce the amount of benefits to retirees. The Democratic minority in Congress immediately expressed its adamant opposition to Bush’s proposal. It became evident that Democrats in the Senate would filibuster any legislation of Social Security overhaul unless a bill was modified to gain support from some Democrats. On March 3, 2005, amidst public opinion polls showing overall opposition to Bush’s plan, the Bush administration launched a “60 stops in 60 days” tour to enhance public support for Bush’s Social Security reform and pressure the Democrats. However, by the end of the tour, it became apparent that all Democrats remained opposed to the presidential proposal, and even some Republicans were against it. On June 23, Republican Sen. Jim Demint (R-SC) introduced a bill (S 1302) to create individual accounts in Social Security program. The bill proposed to use the current Social Security surplus to fund individual accounts. On July 14, several Republican members in the House Ways and Means Committee introduced a similar measure (HR 3304). By August, however, the Republicans in the Senate Finance Committee failed to agree on the sizes of individual accounts and benefit cuts. Also, Democrats and interest groups successfully lobbied moderate Republicans not to support the Social Security bills. In October 2005, sensing political risk for the midterm election the following year, the House GOP leaders, including J. Dennis Hastert (R-IL) and Roy Blunt (R-MO), urged that the Social Security debate be postponed until after the 2006 election.1 Thus, by the end of 2005, the Social Security reform plan was stalled, and no action had been taken on any measures. Guest Worker Program
In the beginning of the 2004 presidential election year, Bush revealed his plan for a temporary guest worker program. On January 7, 2004, Bush asked Congress to consider offering the legal status of “temporary worker” to illegal immigrants residing, and those wishing to find employment, in the
4
The Other Side of Gridlock
United States. Nonetheless, after the 2004 election, the Republican majority in Congress, especially in the House, was cautious toward Bush’s proposal. In December 2005, the Republican majority in the House voted a bill (HR 4437) to enhance border security and increase the severity of penalties for illegal entry into the United States. Under the bill, illegal presence in the country, currently a civil violation, would become a felony, punishable by a year in a prison. The measure did not contain any guest worker provisions. On March 27, 2006, the Senate Judiciary Committee passed its own bipartisan legislation (S 1033), the so-called Kennedy-McCain bill, by a 12-5 margin. In contrast to the House bill, the measure would allow the current illegal immigrants in the United States to temporarily stay in the country and apply for temporary worker visas first and permanent residence visas subsequently. The House and Senate Judiciary bills divided public opinion on the immigration issue. While human rights groups, in conjunction with farming and hotel lobbies, protested against the House bill, numerous groups marched to oppose the Kennedy-McCain plan and advocated a reduction in illegal immigrants. Also, several Republican legislators expressed their concern that the guest worker provision in the Senate Judiciary bill would give amnesty to the illegal immigrants who were currently present in the nation. Soon, President Bush expressed his skepticism toward the conservative wing within his party, who opposed the Senate measure. Bush stressed that “[m]y judgment is, you cannot enforce the border without having a temporary guest worker program—the two go hand in hand.”2 Although several Republican senators maintained their opposition to the Kennedy-McCain bill, Senators Chuck Hagel (R-NE) and Mel Martinez (R-FL) proposed a compromise plan to gain a filibuster-proof support, at least sixty votes, for the Senate Judiciary bill. The Hagel-Martinez plan would strengthen border security, introduce a temporary guest worker program, and allow present illegal immigrants to apply for work and residence visas. On April 6, 2006, it appeared that more than sixty senators supported the Hagel-Martinez plan. Nonetheless, the Senate minority leader, Harry Reid (D-NV), requested majority leader Bill Frist (R-TN) not to consider amendments proposed by the opponents of the Hagel-Martinez measure. Reid also asked Frist to disclose who would represent the Senate in conference committee. In response to Frist’s rebuff of Reid’s requests, Reid filed to invoke cloture to limit the debate on the measure. The next day, the cloture was defeated 38-60. In return, Democrats blocked a GOP effort of cloture on the bill on border security (S 2454) by 36-62. Thus, as of April 7, 2006, the bipartisan effort to enact a guest worker program was broken apart.
Introduction
5
On April 25, 2006, President Bush invited several Democratic and Republican senators, including Frist and Reid, to the White House. President Bush expressed his wish for an end to the legislative impasse and his support for the Hagel-Martinez measure. In May, the Senate held two weeks of debate on the bill, and the bipartisan coalition led by John McCain and Edward M. Kennedy defeated any amendments that were likely to increase opposition to the bill. Also, majority leader Frist informed minority leader Reid who would represent the Senate in conference committee. On May 25, the Senate finally passed the Comprehensive Immigration Reform Act of 2006 (S 2611) by a filibuster-proof margin, 62-36. Overwhelmed by excitement, Senate majority leader Frist could not refrain from expressing his delight, stating, “This is a momentous day for the United States Senate, in large part because we have demonstrated what is the very best about this body.”3 Similarly, Chuck Hagel (R-NE), one of the cosponsors of the measure, stressed the significance of the passed bill, saying that “[t]his bill represents, at least in my brief ten years in the Senate, the most remarkable coalition of leadership I have seen.”4 In sharp contrast to the sense of satisfaction and achievement in the Senate regarding the immigration reform bill, some House Republicans were concerned that the measure would reward current illegal immigrants who had broken the law, thereby attracting more immigrants to seek illegal entrance into the country. A few members even asked the House Speaker, J. Dennis Hastert, not to participate in the conference committee. House Majority Leader John Boehner (R-OH) stated, “I don’t underestimate the difficulty in the House and Senate trying to come together in an agreement.” After Hastert discussed the matter with several committee chairs in the House in August and September, the Speaker decided to postpone the discussion on the Senate measure until after the 2006 midterm election. Hastert stressed that the guest worker program was premature and the border security might be given priority. Haster stated, “We have a border that is bleeding to death. And we have to make sure we can stop that bleeding and get the patient well enough to fix other things.”5 Soon after Hastert’s decision, the House passed the U.S.-Mexican border fence bill (HR 6061), and the Hagel-Martinez bill was stalled until the end of the 109th Congress. Reauthorization of Patriot Act
The House and Senate committee hearings on the reauthorization of sixteen provisions in the 2001 antiterrorism law (PL 107-56), known as the “Patriot Act,” began in early April 2005. These sixteen provisions would expire
6
The Other Side of Gridlock
by the end of that year. The provisions included controversial elements, such as sections allowing federal agencies to install wiretaps and to subpoena corporations, schools, and other organizations for various records and documents. President Bush not only wanted to make all the provisions permanent, but also sought to expand the FBI’s subpoena power to obtain any records without approval from a judge or grand jury. On June 7, the Senate Intelligence Committee approved a bill that granted much of what Bush requested. The Committee bill (S 1266), approved by 11-4 vote, would allow the FBI to issue search warrants in terrorism investigation without prior approval from a judge or grand jury. The bill would also expand the Foreign Intelligence Surveillance Act (FISA) to give the FBI more authority to seize business, medical, and library records. Almost immediately, the Senate Intelligence bill was harshly criticized by Democrats and civil liberty groups, who were concerned about potential abuses and violations of civil rights by the federal government. In the House, Judiciary Committee Chairman F. James Sensenbrenner Jr. (R-WI) introduced the bill (HR 3199), that would make all sixteen provisions permanent. Nonetheless, Sensenbrenner was soon forced by the bipartisan pressure in his committee to support an amendment to impose a ten-year expiration deadline on the provisions granting federal authority in using wiretaps and accessing business records. Subsequently, the Judiciary Committee passed the measure with the amendment on July 13, 2005, by 23-14 vote. On the same day in the Senate, as a compromise measure, the Senate Judiciary Committee chairman, Arlen Specter (R-PA) introduced the cosponsored bill (S 1389) with Sens. Dianne Feinstein (D-CA) and Jon Kyl (R-AZ) to set four-year, rather than ten-year, expiration dealines for the contentious provisions. The Senate Judiciary Committee unanimously passed the bill on July 21. On July 29, the Senate substituted the text of the Senate Judiciary bill into the House Judiciary bill (HR 3199) and passed it by voice vote. On November 16, 2005, the GOP leaders in the House and Senate reached an agreement to place seven-year expiration dates on the two contentious provisions. However, the compromise plan was soon jettisoned when six senators, including three Republican members, threatened a filibuster against the measure. In early December, the GOP leaders in the two houses, joined by Vice President Dick Cheney, agreed with a conference report, imposing a fouryear expiration on the two provisions. Still, the six senators, who requested more restrictions on federal authority to detain records, expressed obstinate opposition. Thus, as Senate Judiciary Committee chairman Specter remarked, a “unique combination of forces from the right and left” objected to the antiterrorism bill.6 On December 16, the Senate failed to invoke cloture on the
Introduction
7
conference report against the filibuster by 52-47. On the cloture vote, four Republicans, including Larry E. Craig (ID), Chuck Hagel (NE), Lisa Murkowski (AK), and John E. Sununu (NH), joined forty-one Democrats in voting against cloture.7 Outraged by the stalemate, President Bush called the filibuster “irresponsible,” and criticized it on the grounds that “it endangers the lives of our citizens.”8 On December 21, the Senate passed a six-month extension of the Patriot Act (S 2167), but the House passed a five-week extension and the Senate concurred. In 2006, legislators faced a February 3 deadline for the reauthorization of the Patriot Act. In January, the White House Office and Justice Department staff met with John E. Sununu, and agreed on the additional changes to the conference report (HR 3199). One change excluded traditional libraries from recipients of “national security letters,” which are requests for subscriber records from phone companies and Internet providers. Other changes included allowing the recipients of business records requests to challenge gag orders in court, and removing the requirement that recipients of national security letters disclose the name of attorneys they consult. On February 1, the House voted to extend the Patriot Act again until March 10. The Senate voted to approve the same extension the next day. On February 9, Republican senators Craig, Hagel, and Murkowski expressed their support for the conference report with the changes offered by the White House. The changes were included in a separate measure (S 2271), and the Senate passed the bill, 95-4, on March 1. Subsequently, the Senate passed HR 3199 on March 2, and the House passed S 2271, by 280-138, on March 7. President Bush signed both the bills on March 9, a day before the expiration of the Patriot Act. Tax Cuts Extension
In his 2006 fiscal year budget proposal, President Bush requested a 2010 extension for all of the 2001 and 2003 tax cuts. They included a two-year extension of the reduced 15 percent capital gains and dividends rate, which was set to expire in 2008, and a permanent extension of the “business tax break” including research and development credits. Because of the growing deficit, the GOP leaders found the permanent extension of the R&D and other credits difficult. They focused on the extension of the capital gains and dividends rate, and decided to pass the legislation within the “budget reconciliation process,” which prohibits a filibuster. Based upon the Congressional Budget Act of 1974, a budget resolution could specify reconciliation instructions to write bills that increase or decrease revenue or spending. The procedure was originally intended to reduce budget deficits.
8
The Other Side of Gridlock
On March 17, 2005, the House narrowly adopted its version (H Con Res 95) of the fiscal 2006 budget resolution by a 218-214 vote. The Senate also barely adopted its version (S Con Res 18) of the budget resolution by 51-49 on the same day. The final budget resolution (H Con Res 95) adopted by conference committee in April instructed the House Ways and Means and the Senate Finance Committees to identify $70 billion in tax cuts through 2010 in the reconciliation procedure. On November 15, the House Ways and Means Committee approved, by a 24-15 vote, its $56 billion tax cut reconciliation bill (HR 4297), which included a two-year extension of the tax cut on the capital gains and dividends. The House passed the committee’s bill 234-197 on December 8. In the Senate, Finance Committee Chairman Charles E Grassley (R-IW) had proposed a five-year, $68.8 billion tax cut measure, including the extension of the capital gains and dividends rate through 2009, a year short of 2010, in order to win support from moderate Republicans. Nonetheless, Grassley faced strong opposition to the bill from fellow Republican Olympia J. Snowe (R-ME) within the committee. After unsuccessful negotiations among the GOP leaders and Snowe, the provision for the capital gains and dividends was eventually removed from the bill, and the Senate passed the $57 billion tax cut measure (S 2020) with support from a supermajority of the members on November18, 2005, by a 64-33 vote. Thus, in contrast to the House bill (HR 4297), the Senate measure did not include the extension of tax cuts on capital gains and dividends. Also, unlike the House package, the Senate bill would reduce the effect of the alternative minimum tax (AMT) on middle-class families. On February 2, 2006, notwithstanding adamant objection from all the Democrats and a few Republicans, the Senate passed the House bill (HR 4297) by a 66-31 vote, but only after amending the measure to assimilate most of the text of the Senate bill (S 2020). After months of negotiations over the conference report, Republican leaders offered the provision to raise $3 billion in revenue over five years and managed to gain support from a few Democrats. On May 10, the House approved the conference report (HR 4297), which included the extensions of both the investment income rate and the patch on the AMT, by a 244-185 vote. The next day, the Senate cleared the measure on a 54-44 vote. Three GOP senators, including Olympia J. Snowe (ME), Lincoln Chafee (RI), and George V. Voinovich (OH), joined Democrats in voting against the conference report, while three Democrats voted for the measure. Divided Government-Gridlock Hypothesis
As the 103rd and 109th Congresses exemplify, unified government, as well as divided government, can result in low efficacy in lawmaking—the problem
Introduction
9
popularly known as gridlock, impasse, or stalemate. Nonetheless, most scholars before the 1990s focused on the different partisan control of the executive and legislative branches as a major cause of gridlock. Woodrow Wilson observed that the Madisonian separation of powers could cause political conflict and inconsistency. Wilson advocated unified government, stating that “harmonious, consistent, responsible party government” connecting the “President as closely as may be with his party in Congress” is necessary for a well-functioning governance (Ranney 1954, 31–32). Wilson’s view was later embraced by the doctrine, if not positive theory, of “responsible parties.” Schattschneider (1942) advocated American democracy based upon political parties. He emphasized that political parties were the most legitimate and effective bodies to represent majorities. Key (1964) argued that the political parties linked and united the two branches. Key contended that “common partisan control of executive and legislature does not assure energetic government, but division of party control precludes it” (1964, 688). Ripley agreed, stating that “the President and a majority of both houses must be from the same party” for governmental productivity (1969, 168). In the 1980s, with divided government becoming increasingly common, scholars viewed it as a major cause of gridlock (Sundquist 1980, 1988, 1992; Cutler 1988; Fiorina 1996, 2003). These scholars observed that divided government provided electoral incentives for interpartisan conflict between the two branches. Fiorina, for example, argued that a majority party in Congress “has every incentive to reject presidential initiatives; to accept them is to acknowledge the president’s competence and sagacity, hence, to support his reelection” (2003, 86). However, by compiling a list of significant legislative enactments during the period from 1947 through 1990, Mayhew (1991) challenged conventional wisdom and revealed that divided government did not decrease the amount of important legislation enacted. Since Mayhew’s groundbreaking contribution, an increasing number of scholars have reexamined the theoretical and empirical grounds of the divided government–gridlock nexus. Several scholars have questioned Mayhew’s measure of the volume of enacted laws; they have alternately explored a ratio of the enacted legislation to the entire legislative agenda, and proclaim an impact of divided government (Binder 1999, 2003; Coleman 1999; Edwards and Barrett 2000). The divided government–gridlock theorem entails two central premises: party unity and majority rule. As for party unity, the majority party in divided government, which is the president’s opposition party, could reject the presidential legislative initiatives if and only if its members are united to maintain sufficient votes to reject the presidential proposals. Several scholars argue that
10
The Other Side of Gridlock
legislators are united along with the party in order to solve collective action and coordination problems (Cox and McCubbins 1993, 2005; Aldrich 1995; Aldrich and Rohde 2000). Aldrich (1995) contends that since the transaction costs of bargaining and negotiating are lower within rather than across parties, legislators are more likely to be united within their parties. In their conditional party government theory, Aldrich and Rohde argue that parties become cohesive and disciplined when party members have homogeneous preferences within parties but heterogeneous preferences across parties (Rohde 1991; Aldrich and Rohde 2000). Also, Kiewiet and McCubbins (1991) assert that the legislators have incentives to engage in “common investment” in a party since the party label is a brand name for them. Consequently, Cox and McCubbins’s (1993) Legislative Leviathan hypothesis maintains that majority parties function as cartels to serve the party and make legislative decisions solely for the electoral interest of the party. Apropos of majority rule, a majority party can block the presidential initiatives if and only if the majority of votes on the floor (and in committees) could reject bills. Aldrich and Rohde (2000) stress that a majority party enjoys substantial advantages in committee assignments, committee leadership, and the majority floor leader’s power over the floor agenda through rigid control of the Rules Committee. Similarly, Cox and McCubbins (2005) contend that the majority party leaders are able to block unfavorable legislation from reaching the floor with the support from committee chairs, the speaker, and members of the Rules Committee. Supermajoritarian and Nonpartisan Lawmaking
While the divided government–gridlock hypothesis is predicated upon party unity and majority rule, there are some theoretical and empirical questions about these assumptions. When voting behavior is examined, some scholars observe that the partisanship of the two parties has been increasing since the 1980s (Rode 1991; Bond and Fleisher 2000; LeLoup and Shull 2003; Jacobson 2004). Many of these studies are based upon party unity scores, which are the percentage of members voting with a majority of their party on party-split votes, if not on all roll call votes. As Figure 1.1 and Figure 1.2 illustrate, party unity scores in the House and Senate have been increasing since the 1980s. Researchers contend that the Southern realignment followed by the civil rights revolution, as well as the Reaganite conservative movement, enhanced the ideological coherence within the two parties and increased the ideological distance between them. In particular, Figures 1.1 and 1.2 show that the party unity scores since the 1990s have reached a pinnacle, ranging from 85 to 95.
1 2
3 4
5 6
7
8
10
9 11
$EMOCRATIC 2EPUBLICAN
13 14
Figure 1.1 Party Unity Scores: House 1954–2006
16
Adapted from Norman J. Ornstein et al, Vital Statistics on Congress (2008, Brookings Institution Press).
17
12
15
18 19 20
21 22
23
24 25
26 27
28
29 30
$EMOCRATIC 2EPUBLICAN
31 32 33 34 35
Figure 1.2 Party Unity Scores: Senate 1954–2006 Adapted from Norman J. Ornstein et al., Vital Statistics on Congress (2008, Brookings Institution Press).
36 37 38 39 40 41
11
12
The Other Side of Gridlock
Nevertheless, since the majority party almost always holds a narrow majority, even the seemingly ultra-high unity scores of the majority party do not indicate an ability to maintain majority votes on the floor. For instance, Table 1.1 shows the party unity scores of the majority party in the Senate from the 102nd to 109th Congresses (1991–2006). In the Senate, the majority party held fewer than sixty out of the one hundred seats. In the table, Loyal Member is calculated from the number of the majority party members multiplied by the party unity scores. Thus, Loyal Member indicates an average number of the majority party members voting with the party. Significantly, in all the examined Congresses, the majority party has maintained fewer than fifty loyal members, which indicates minority votes in the Senate. Of course, a few members of the minority party may have voted with the majority party on some bills. Nevertheless, the indication is that the high party unity scores of the majority party since the 1990s do not guarantee the majority party’s ability to maintain majority votes. When there are few or no defections of minority party members, the majority party is unable to sustain a majority in voting on the floor. For instance, in the 109th Congress, the Social Security reform legislation was stalled because the measure was opposed by a few moderate Republicans and all Democrats. When the majority party needs votes from the minority party members, as it usually does, the majority party will have to modify the legislation. For example, as explained before, the Senate in the 109th Congress barely passed the tax cuts extension measure because the GOP leaders won support from a few Democrats by modifying the legislation. Next, the party unity scores are the percentage of members voting with their party on party-split votes, but not on all roll call votes. If more than 50 percent of the majority party members, say 70 percent of the majority party members, vote with the minority party on a bill, the vote is considered to be a bipartisan
Table 1.1 Majority Party Unity Scores in Senate 1991–2000 Congress
Majority Party
Majority Party Members
Majority Party Unity Score
Loyal Members
102 103 104 105 106 107 108 109
D D R R R D R R
56 57 52 55 55 50 51 55
82.5 86.5 91 88 90.5 87.5 94 88
46 49 47 48 49 43 47 48
13
Introduction
vote and thereby the low party unity on the measure (70) is excluded from the calculation. This suggests that the party-split votes, which is the denominator in the measurement, could have inflated the party unity scores. Figure 1.3 shows the percentage of party-split vote, which is calculated from the percentage of roll calls on which a majority of voting Democrats opposed a majority of voting Republicans. The figure illustrates that party-split votes have increased in the early 1990s. Nonetheless, the average percentage party-split vote from 1954 to 2006 is 47 percent both in the House and Senate. The same figure for the last decade (1996–2006) includes 49 percent and 56 percent in the House and Senate, respectively. Thus, overall, the party-split votes account for approximately one-half of all recorded votes. Mayhew (1991, 2005) stresses the presence of cross-cutting issue cleavages, which are the issues that divide legislators within parties. Mayhew argues that civil rights, labor-management relations, anti-Communist policy (2005, 140–41), and recently, foreign trade (2005, 215) are examples of issues that have divided legislators’ voting behavior in a nonpartisan manner. As for majority rule, the assumption largely ignores the various supermajoritarian procedures in Congress. In accordance with Senate Rule XXII, senators maintain the right to engage in extended debate, the so-called filibuster, subject to a cloture vote, which can be successfully invoked by sixty or more senators. Thus, a minority (n-59) of leftist senators or rightist senators can block
(OUSE
3ENATE
9E A R
Figure 1.3 Percentage Party-Split Vote on Roll Calls 1954–2006 Adapted from Norman J. Ornstein et al., Vital Statistics on Congress (2008, Brookings Institution Press).
14
The Other Side of Gridlock
bills they dislike. Today, scholars observe that threats of filibuster are more common than actual filibusters and the former is as effective as the latter for blocking legislation (Sinclair 2007, 67–72; Oleszek 2004, 239–46). According to Barbara Sinclair (2002, 252–53; 2006, 211–24), approximately one-half of major legislation in the 1990s and early 2000s was filibustered or threatened with prospective filibuster. Also, Article I, Section 7 of the United States Constitution, grants the president the authority to veto legislation, which in turn is subject to the ability of Congress to override the veto by a two-thirds vote of the members present and voting in the two chambers. This indicates that one-third plus one members whose preferences are close to the president’s preference in either congressional chamber could reject unfavorable legislation. However, the mere number of exercised (and sustained) vetoes might not indicate the overall effect of presidential veto power since the threat of veto against a bill could sufficiently discourage the legislators from adopting it ex ante (Hinckley 1985; Kiewiet and McCubbins 1988; Edwards, Barrett, and Peak 1997; Sinclair 2007; Cameron 2000; Deen and Arnold 2002). Scholars of the preference-based school contend that the minority party could challenge a majority with filibusters and presidential vetoes, and thereby gridlock could occur in unified government (Krehbiel 1996, 1998; Brady and Volden 1998, 2006). Jones (2001) finds that legislation is more likely to fail when the ideological positions of the two parties are highly polarized, but the impact of party polarization decreases as the majority party approaches the size of a supermajority. More precisely, Mayhew argues that many pieces of legislation are rejected because they fail to gain support from the necessary broad majorities (2005, 216). Consistent with Mayhew’s observation, the bills regarding Social Security reform and the guest worker program failed in the 109th Congress because they did not receive bipartisan support. The extension of the Patriot Act was passed with bipartisan support after the bill was modified. The 109th Congress barely passed the tax cuts extension package without bipartisan support, but the measure was in the “reconciliation procedure,” which prohibited a filibuster. Thus, the assumptions of robust party unity of the majority party and of majoritarian procedure are in a precarious state. The theoretical foundation, as well as the empirical findings, of the connection between divided government and gridlock are still fragile and need cautious examination. Policy Stability
David Mayhew’s Divided We Govern (1991, 2005) sparked a wave of scholarship on the politics of gridlock. The post-Mayhewian studies have focused on
Introduction
15
various quantifications of legislative productivity as a measure of gridlock. In contrast, by following the definition of gridlock as an inability to change policy (Kernell 1991; Krehbiel 1998; Brady and Volden 1998; Chiou and Rothenberg 2003; Tsebelis 2002), this book essentially focuses on policy change as an indicator of gridlock. Mayhew defines innovative legislation as that which newspaper articles contemporaneously evaluated as important legislation or policy analysts retrospectively assessed as legislation with consequential impact (2005, 37–49). However, innovative laws must result in innovative policy. In the policy context, the innovativeness of laws indicates a new direction away from the status quo policy. Accordingly, several scholars define gridlock as an inability to change the policy of the status quo. Krehbiel defines gridlock as “the absence of policy change in equilibrium in spite of the existence of a legislative majority that favors change” (1998, 26). Thus, as Krehbiel acknowledges (1998, 5), gridlock is equivalent to what some scholars call policy stability (Hammond and Miller 1987; Riker 1992; Tsebelis and Money 1997; Tsebelis 2002). Tsebelis defines policy stability as “the impossibility of significantly changing the status quo” (2002, 6). Tsebelis stresses the importance of studying policy stability as an indicator of governmental decisiveness, but also as an explanatory variable for governmental stability and judicial independence. Plan of the Book
This book explores the assumption of nonpartisan, supermajoritarian lawmaking in the U.S. Congress. The book proposes various explanatory models and conducts empirical tests pertinent to the influence of the preference of the minority legislators on policy change. The next chapter examines the various measurements of legislative productivity by Mayhew and other scholars, followed by the discussion of whether or not legislative productivity reflects the innovativeness of policy output by Congress. Subsequently, the chapter proposes policy stability as an indicator of gridlock in the policy context. The chapter measures policy stability, or inversely policy change, based upon the ADA scores, and examines the variation in the indicator. Chapter 3 studies the supermajoritarianism of the U.S. Congress. Accordingly, the chapter proposes the pivotal interval movement model, based upon the Brady-Kehbiel-Volden school of preferential, nonpartisan theory of lawmaking. The pivotal interval movement model presumes that a gridlock interval between the filibuster and override pivots in the previous Congress is transformed to a status quo interval for the new Congress. The model introduces a concept, residuum, which is a portion of the gridlock interval of the previous Congress
16
The Other Side of Gridlock
not overlapped by the gridlock interval of the new Congress. This suggests that when the residuum is wide, the policy output of legislation is more likely to change. However, in contrast to the nonpartisan basis of the pivotal interval movement model, the cartel agenda theory by Cox and McCubbins (2002, 2005) maintains that the majority party successfully blocks measures it dislikes. In order to compare the cartel assumption with the pivotal interval movement model, chapter 3 constructs the cartel interval movement model predicated upon the partisan assumptions of the cartel agenda model. Consequently, two contending hypotheses are generated from the two models of the interval movement. In chapter 4, empirical models are employed to analyze the influence of the width of the residuum of pivotal and cartel gridlock intervals, separately, on policy change. The models focus on the nominal and weighted ADA policy change scores in twenty-seven Congresses from 1953 to 2006. In addition to the various measurements of the residuum, the independent variables of divided government, budgetary situation, public liberal mood, and start of the presidential term are included in the models. Chapter 5 presents the veto players model to explain policy change. Parallel to the pivotal interval movement model, the veto players model is predicated upon the assumption of nonpartisan, supermajoritarian lawmaking. In contrast to the pivotal interval movement model, however, the veto players model is based upon a two-dimensional policy space. The chapter plots the preferences of veto players, including the filibuster, veto, and House median pivots, in two-dimensional Cartesian space. These veto players are likely to support legislation within their unique indifference curves, which circle their respective ideal points and pass through the status quo point. On the other hand, the veto players are likely to reject bills outside their indifference curves. Since all the veto players must unanimously support a bill for passage to occur, only legislation in the intersection of all the four indifference curves, so-called winset, are likely to pass. The empirical analysis examines the influence of the area of the winset on policy change. Chapter 6 studies the blockage of agendas from reaching floor consideration. A party is rolled when a majority of its members vote against a measure, a majority of the opposite party’s members vote for the measure, and the motion passes on a final-passage vote. Cox and McCubbins (2002, 2005) find low roll rates of the majority party, and argue that the majority party’s negative agenda power is unconditional. The chapter first examines the validity of roll rates as a measure for partisan negative agenda power. Subsequently, the empirical study analyzes: (1) the floor bills a party has opposed, and (2) influence of the
Introduction
17
distance between the status quo and the floor median points on the percentage legislative agendas that reach the floor. The book concludes in chapter 7 by summarizing and comparing the findings on policy change. The chapter discusses influences of party and supermajoritarianism. The chapter also indicates potential reforms to alleviate gridlock in the American government.
This page intentionally left blank.
Chapter 2
Gridlock and Policy Stability The American people voted for a new direction to restore civility and bipartisanship in Washington D.C.
—Nancy Pelosi, November 8, 2006
David Mayhew’s Divided We Govern (1991, 2005) broke new ground when it challenged the conventional wisdom of an adversarial effect of divided government on governmental effectiveness. Recently, Mayhew (2005) extended his study of the amount of important legislation to the year 2002, and reasserted that divided party control of the government has no impact on legislative productivity. Since Mayhew’s contribution, several scholars have questioned Mayhew’s measurement of significant laws (Kelly 1993; Edwards, Barrett, and Peake 1997; Howell, Adler, Cameron, and Riemann 2000). Some researchers have studied the ratio of enacted legislation to the entire legislative agenda finding an impact of divided government (Binder 1999, 2003; Coleman 1999; Edwards, Barrett, and Peake 1997). First, this chapter reviews the various measurements of legislative productivity in the post-Mayhewian literature and examines the impact of party control of the government on legislative productivity. Secondly, the chapter explores policy stability, or policy change, as a measurement of gridlock in “content terms” (Mayhew 2005, 221). Policy change is likely to occur when Congress produces legislative output significantly deviant from status quo policy. In contrast, when legislative output is similar to the status quo, the output indicates policy stability. In this chapter, policy stability is measured based on the passed legislation selected by the Americans for Democratic Action.
19
20
The Other Side of Gridlock
Legislative Productivity: Numerators and Denominators
In order to assess legislative productivity, Mayhew’s empirical study (1991, 2005) focuses on the enactment of landmark laws. In Sweep One, Mayhew examines the end-of-session wrap-up articles from the New York Times and Washington Post to identify those laws with contemporaneous significance. As for the laws with retrospective significance, policy specialists analyze the consequence of the laws and identify fifty-six laws in Sweep Two. Mayhew studies the volume of landmark laws in Congresses, which includes the laws classified by either Sweep One or Sweep Two, and states that divided government has no significant impact on the passage of innovative legislation. Among the immediate criticisms of the Mayhewian repudiation of party government theory, Kelly (1993) questions the validity of Mayhew’s “either” method in the selection of landmark laws. Kelly stresses that innovative policy is “timely and enduring” and therefore the study should focus on the “intersection of promise and performance,” rather than on only one of them (1994, 478). Kelly argues that laws once considered significant, but later insignificant, should be excluded from the category of landmark laws. Accordingly, Kelly examines the volume of laws that are applicable to Sweep One and Sweep Two finding that divided government can result in a reduction in the number of landmark laws enacted. Nonetheless, Howell, Adler, Cameron, and Riemann (2000) observe that the selection of significant laws in Sweep Two is biased. The authors find that the number of laws with only retrospective significance increases in divided government. Howell, Adler, Cameron, and Riemann argue that the significant laws in Sweep Two are overrepresented in the more recent past, when divided government was common. When the trend is controlled, Howell, Adler, Cameron, and Riemann observe that the volume of laws classified in Sweep One decreases in divided government. Fiorina (1996, 2003), another skeptic of Mayhew’s finding, notes that Mayhew pays little attention to demand for legislation, a denominator that should be considered. Fiorina contends that Mayhew “has studied the supply of federal laws.…But we have no information about the demand for legislation” (2003, 89). Mayhew is indubitably cognizant of this problem when he defines legislative gridlock as “some actually-did-pass numerator over some all-thatwere-possibilities-for-passage denominator” (1991, 34), but he concludes that it is “very difficult to see what a denominator for a Congress might be” (1991, 36). Further, in his second edition, Mayhew contends that since legislators always make legislative proposals, only the legislation with a significant policy impact should be counted. Mayhew writes:
Gridlock and Policy Stability
21
Why should we pay so much attention to policy proposals? They can be a kind of vapor that surrounds real legislative action. Politicians are always making proposals. Parties are always making proposals.—But such offerings are often vague, shifting, infeasible, or otherwise insubstantial. In general, a sensible public will discount all of them at a very high rate. Actual governmental production is arguably the news worth paying atte tion to. (2005, 201–202)
On the other hand, Edwards, Barrett, and Peake (1997) argue that since the concern is whether or not divided government “obstructs the passage of legislation” (1997, 825), the proposed legislation that failed should be studied. Edwards et al. study the amount of important bills that failed and observe that the president of divided government is more likely to oppose significant legislation and that more important legislation fails to pass under divided government. Later, Coleman (1999) explores the percentage of important legislation enacted into law by dividing Mayhew’s measure by the sum of the measures of Mayhew and Edwards et al. Coleman finds that the percentage of the enacted significant laws declines in divided government. Binder (1999, 2003) acknowledges what Cobb and Elder (1983) call the systemic agenda as a demand for legislation, which should be a denominator to legislative enactments. The systemic agenda consists of “all issues that are commonly perceived by members of the political community as meriting public attention” (Cobb and Elder 1983, 85). In order to measure the systemic agenda, Binder tallies the legislative agendas appearing in daily editorials in the New York Times and examines the percentage of those that failed to be enacted into law by the close of Congress. Binder finds that legislative agendas are more likely to fail in divided government. Edwards and Barrett (2000) identify major legislative proposals by the president and Congress, and calculate the percentage of enacted proposals. Edwards and Barrett find a lower rate of presidential initiatives being enacted in divided government. However, the majority party in a divided government introduces and passes many congressional initiatives to challenge the president. This, then, results in a higher percentage of the passed congressional initiatives in divided government. Thus, several studies proposed alternative measurements of significant legislation, and some of them augmented various denominators to the number of important laws. Figure 2.1 shows the values of the measures developed by Mayhew (2005), Kelly (1993), Binder (2003), and Edwards and Barrett (2000). The former two measures are numerator-based variables, which focus on the number of important bills enacted. The latter two variables are ratio variables,
22
The Other Side of Gridlock
which indicate the percentage of failed agendas among all the agendas. Figure 2.1 shows the rescaled and reversed figure1 of the amount of Mayhewian legislation from the 83rd to 106th Congresses.2 The low value indicates a larger quantity of enacted legislation. Kelly’s variable is calculated from the volume of enacted laws with contemporaneous and retrospective significance based upon Mayhew’s data. In the figure, Kelly’s variable is also reversed. Binder counts the amount of legislative agendas mentioned by the editorials in the New York Times and calculates the percentage of failed agendas (2003, 37–44). Edwards and Barrett (2000) study major legislative proposals of the president and Congress. Figure 2.1 illustrates the percentage of failed proposals. Figure 2.1 shows that although the four measures of gridlock are related, there is substantial variation in the degree of gridlock measured by these variables. In contrast to other indicators, Mayhew’s rescaled variable shows a larger fluctuation in legislative productivity between the 89th and 93rd Congresses. Nevertheless, as the correlation matrix in Table 2.1 suggests, Mayhew’s variable of important laws is related to all other variables. The variables of Kelly and of Edwards and Barrett are related to all the variables except Binder’s. Thus, surprisingly, the variable used by Edwards and Barrett, which covers the percentage figures based upon a small amount of denominating legislative initiatives, is significantly related to the numerator-based variables of Mayhew
Edwards
Binder
Mayhew (Rescaled)
Legislative Productivity (Reversed)
Kelly (rescaled)
#ONGRESS
Figure 2.1 Gridlock in Legislative Productivity, 83rd-106th Congress
23
Gridlock and Policy Stability
and Kelly. While Binder’s variable is moderately related to Mayhew’s variable, it is not related to the other variables.3 This is not surprising given the uniqueness of Binder’s measurement, especially the measurement of the agenda denominator. In contrast to other measures of gridlock, Binder covers a vast amount of legislative agendas in the denominator. Next, the mean values of the various measures of legislative productivity and t-test results comparing periods of divided and unified government are examined. Results appear in Table 2.2. According to the numerator-based measures by Mayhew and by Kelly, divided government produces a slightly smaller amount of important laws. However, the difference in legislative productivity between unified and divided government is not statistically significant. As for Binder’s measures, I categorize the agendas into those either appearing one or two times in the editorials (Salience 1–2) or three to five times (Salience 3–5). In Table 2.2, Binder’s legislative agendas, especially those with lower salience, are more likely to fail in divided government. For failed legislative initiatives, as measured by Edwards and Barrett, presidential initiatives are more likely to fail in divided government. On the other hand, congressional initiatives are more likely to fail in unified government, as Edwards and Barrett have previously reported (2000). Nonetheless, the difference between unified and divided governments is not statistically significant. As for the variation within the measures, it is not surprising to see that the numerator-based variables have a relatively smaller variation than the ratio-based measures. Next, the influence of the various determinants on legislative productivity is examined in the OLS and Glogit4 models. Table 2.3 shows the results. As for Mayhew’s measure of the enacted laws with contemporaneous or retrospective importance, none of the independent variables is significant. Overall, the model performs poorly as illustrated by the marginal value of the Adjusted R2. Thus, under the Mayhewian measurement, divided government does not matter as Mayhew has asserted (2005, 207–15). Kelly’s model measuring the volume of important laws performs better, explaining 28 percent of the variance in Table 2.1 Correlations among Variables of Legislative Productivity Mayhew Mayhew Kelly
– 0.78***
Kelly
Binder
Edwards and Barrett
–
–
–
–
–
–
Binder
–0.45*
–0.48
–
–
Edwards and Barrett
–0.65**
–0.60*
0.31
–
*p<.05 **p<.01 ***p<.001 (2-tailed)
24
The Other Side of Gridlock Table 2.2 Mean Values and Variations in Legislative Productivity Mayhew
Kelly
Important Important Laws Laws
Edwards and Barrett Failed Initiatives (%)
Binder Gridlock (%) Salience 1–2
Salience 3–5
All Agendas
Presidential Congressional Initiatives Initiatives
All Initiatives
Mean Unified Government
13.38
9.63
45.19
38.43
42.81
46.57
81.76
62.82
Mean Divided Government
11.25
6.31
53.02*
46.71
51.03*
69.38*
75.27
74.37*
5.00
4.16
8.27
11.59
8.92
21.35
12.18
10.60
High
22.00
16.00
64.15
75.00
66.50
100.00
100.00
94.29
Low
5.00
0.00
33.83
30.00
33.48
25.00
56.41
51.35
N
24
24
24
24
24
21
22
22
Standard Deviation
*p < .05 a 83rd – 99th Congress b 83rd – 104th Congress
gridlock in legislative output. The moderate significance of the public mood variable suggests that the government is more likely to enact important laws when public opinion is more supportive of an active government. In contrast, the variable of divided government and other variables are not significant. Thus, divided government has no impact on the quantity of enacted laws with contemporaneous and retrospective significance.5 For Binder’s measurement, the specification for the failed agendas with low salience shows some explanatory power (Adjusted R2 = .26). In this model, start of term is significant, suggesting that legislative agendas are more likely to fail in the later two years of presidential terms. However, the variable becomes insignificant in the models of agendas with higher salience. Also, divided government is significant, suggesting that less salient legislative agendas are more likely to fail in divided government than unified government. In the model of all the agendas, divided government is still significant but the model shows modest explanatory power. 6 In the model of the failed legislative agendas with higher saliency, none of the variables are significant. Regarding Edwards and Barrett’s measures, only the variable of divided government is significant in the model of presidential initiatives. This implies that the legislative agendas proposed by the president are more likely to be rejected in divided government. In contrast, none of the variables are significant in the models of the congressional initiatives and of the all initiatives.
25
Gridlock and Policy Stability Table 2.3 Determinants of Legislative Productivity 1953–2000 Mayhew
Kelly
Important Important Laws Laws
Edwards and Barrett Failed Initiatives (%)
Binder Gridlock (%) Salience 1–2
Salience 3–5
All Agendas
Presidential Congressional Initiatives Initiatives
All Initiatives
Divided Government
–1.258 (2.124)
–2.147 (1.531)
0.289* (0.126)
0.296 (0.186)
0.299* (0.139)
0.934* (0.299)
–0.441 (0.346)
0.328 (1.910)
Start of Term
2.342 (2.003)
0.75 (1.518)
–.261* (0.118)
–.207 (0.182)
–.241* (0.132)
–.049 (0.303)
–.372 (0.293)
–.258 (0.189)
Public Mood
0.368 (0.207)
0.307^ (0.155)
–.005 (0.013)
–.005 (0.019)
–.004 (0.014)
–.013 (0.029)
–.029 (0.033)
–.023 (0.019)
Budgetary Situation
0.047 (0.120)
0.077 (0.110)
.001 (0.007)
.006 (0.011)
.002 (0.008)
–.016 (0.025)
–.028 (0.020)
–.028 (0.014)
.246
–.001
–10.227
–9.139
3.13
1.933
24
17
24
24
24
22
22
22
Adjusted R2
0.08
0.283
0.257
0.021
0.189
0.318
0.089
0.382
F
1.49
2.58
2.99
1.13
2.34
3.1
1.49
4.25
Constant N
.134
.576
Entries for the models of Mayhew and Kelly are unstandardized coefficients from OLS regressions. Entries for the models of Binder and Edwards et al. are coefficients from weighted least squares logistic regressions for grouped data. Standard errors are in parentheses. ^p < .10; *p < .05
Agenda Expansion?
In the study above, only on some of the ratio variables used by Binder and by Edwards and Barrett does the variable of divided government show a significant impact. Mayhew (2005), doubtful about using the volume of legislative agendas as a denominator, argues that there is a higher volume of legislative agendas and proposals in divided government than in unified government (2005, 201–202). However, Binder reports that the amount of legislative agendas she has measured does not increase in divided government (2003, 38). Table 2.4 shows the average number of agendas measured by Binder in unified and divided government, separately. For all the agendas and the agendas with salience level 1–3, there are slightly more agendas in divided government than in unified government. However, the difference is not statistically significant. Pertaining to the agendas with the salience level 4–5, there are slightly fewer agendas in divided government than in unified government, but the difference is not statistically significant. Thus, as for the agendas measured by Binder, there is no agenda expansion or contraction in divided government. Apropos of the amount of the presidential and congressional proposals measured by Edwards and Barrett, Table 2.5 shows the correlations between
26
The Other Side of Gridlock Table 2.4 Binder Agenda Denominator Salience 1–3
Salience 4–5
All Agendas
Mean Unified Government
147.9
83.3
231.2
Mean Divided Government
164.0
76.5
240.5
divided government and the volume of legislative initiatives. Although the results in Table 2.5 show an expansion of congressional initiatives in divided government, Table 2.3 demonstrates that divided government wields no influence on congressional initiatives. The indication is that the increase in the amount of congressional initiatives in divided government has no impact on the percentage of their passage. Also, whereas Table 2.3 shows an impact of divided government on all legislative initiatives, there is no expansion of legislative initiatives in divided government. Furthermore, Table 2.5 shows a minimization, rather than expansion, of presidential initiatives in divided government. Therefore, the effect of divided government on the percentage of failed presidential initiatives in Table 2.3 could be an underestimated, rather than overestimated, figure. Even though the president fears partisan conflict and proposes fewer legislative agendas in divided government, his or her legislative initiatives are more likely to be rejected than in unified government. As for the ratio measures by Binder and by Edwards and Barrett, the negative impact of divided government is not caused by agenda expansion. Gridlock: Legislative Productivity or Policy Stability?
In spite of the vast and sophisticated reexamination of Mayhew’s null finding, the central focus in the literature of post-Mayhew scholarship has converged primarily on legislative productivity using various quantification methods. However, as Mayhew asserts (2005 220–21), the content of the legislative output should be also examined in the study of gridlock. Binder specifically emphasizes that “how we define gridlock largely shapes how we measure it” (2003, 35). The scholarly definition of gridlock encompasses two schools of thought: gridlock as a legislative inability to respond to public agendas (Binder Table 2.5 Correlations between the Divided Government and the Volume of Legislative Initiatives
Divided Government **p < .01 (2-tailed)
Presidential Initiatives
Congressional Initiatives
All Initiatives
–.652**
.601**
.039
Gridlock and Policy Stability
27
1999, 2003; Bond and Fleisher 2000; Jones 2001) and gridlock as an inability to change policy (Kernell 1991; Krehbiel 1998, 26; Brady and Volden 1998, 43; Chiou and Rothenberg 2003, 504; Tsebelis 2002, 165–67). Notably, neither definition of gridlock implies a quantification of legislation enacted into law. Bond and Fleisher define gridlock as “a lack of movement toward solving the nation’s problem” (2000, 188). This view is reinforced by Binder’s more specific definition as “the share of salient issues on the nation’s agenda that is left in limbo at the close of a Congress” (2003, 35). As for this definition of gridlock, it is rather ambiguous as to how the volume or the ratio of passed legislation could measure the degree of legislative responsiveness to the public agenda. Mayhew focuses on landmark laws. However, some so-called significant legislation may only be modestly responsive to public problems. Mayhew’s Sweep Two identifies the legislation that has been proven to have a substantial impact on the nation. Yet, this retrospective selection was conducted by policy analysts, not by objective, impersonal methods. Consequently, its selection method is hardly free of bias (Howell, Adler, Cameron, and Riemann, 2000). Besides, the selection of important laws is based upon a dichotomy whether they are important laws or not. Therefore, all selected laws are presumed to share the same degree of importance. If two significant legislative measures are merged into a single law, governmental responsiveness decreases by half under this measurement. Mayhew is aware of this problem and states that omnibus measures “dwarf most everything else and that render any simple counting up of laws precarious” (2005, 202). Mayhew defines innovative legislation as identified by newspaper articles as being important or as laws that policy analysts retrospectively assessed as legislation with consequential impact (2005, 37–49). Nevertheless, terms such as “importance” and “consequence” have volatile and often impressionistic characteristics, therefore these qualities are less likely to be measurable on a quantitative scale. Rather, innovative laws are laws that result in innovative policy. In the policy context, the innovativeness of laws is tantamount to a new direction away from the status quo policy. This suggests that innovative legislative output is output that is distant from the status quo. Several scholars define gridlock as an inability to change the current policy status quo. Krehbiel defines gridlock as “the absence of policy change in equilibrium in spite of the existence of a legislative majority that favors change” (1998, 26). Thus, as Krehbiel acknowledges (1998, 5), gridlock is equivalent to what some scholars call policy stability (Hammond and Miller 1987; Riker 1992; Tsebelis and Money 1997; Tsebelis 2002). Tsebelis defines policy stability as “the impossibility of significantly changing the status quo” (2002, 6). Tsebelis stresses the importance of studying policy stability as an indicator of governmental decisiveness,
28
The Other Side of Gridlock
but also as an explanatory variable for factors such as governmental stability as well as judicial and bureaucratic independence. Inasmuch as gridlock is defined as policy stability, the number of enacted laws is hardly relevant to the size of policy change. Mayhew notes that his seminal work is not an examination of the content of legislative output: The content of legislation is of course a feature all by itself. In this book I have not made any systematic effort to code enactments by, for example, whether they lean to the left or the right. (Mayhew 2005, 220; italics original)
Schattschneider’s (1942) classic work, which is frequently cited as a theoretical embryo of a divided government/gridlock relationship, in fact draws a distinction between the volume of laws and the content of laws suggesting that divided government does not matter to the former but it does matter to the latter. In his observation about divided government, Schattschneider (1942, 89–90) states: Congress enacts substantially the same volume of legislation when party control of Congress is divided as it does when one party is in control of both houses. It is not contended here that the volume of legislation is an accurate measure of the legislative output of Congress. No doubt some major controversial bills in each case were obstructed by divided party control or had to be emasculated to be passed whenever the consent of both parties has been required for enactment. What is clear is, however, that a great mass of business can be transacted across party lines when the necessity arises.
Schattschneider suggests that under divided government, some major controversial bills are watered down to become legislation with only modest policy impact. Similarly, Kernell (1991) argues that divided government tends to enhance a massive bargaining process, which results in diluted policy output. Sundquist offers the criticism that Mayhew overlooks the impact of divided government on delays in legislative passage and on the dilution of measures (1992, 103–104). Coleman speculates that less divisive legislation may pass under divided governments while more difficult bills could pass under unified government (1999, 827). Quirk and Nesmith (1994, 1998) contend that divided government matters on highly ideological issues. Nevertheless, Mayhew (2005) asserts that the conditions of party control have a marginal impact on the legislative content. While acknowledging that his study is not a systematic examination of the legislative content, Mayhew observes modest policy changes in enacted laws,7 regardless of whether the government is unified or divided
Gridlock and Policy Stability
29
(2005, 220–21). In order to study gridlock, therefore, it is highly important to examine the policy content, especially policy stability, in legislative output. Policy Change
Suppose there is a homemade ice cream shop in San Francisco. Since the owner, Sarah, opened the shop in the 1960s, she has been aware of the changes in people’s preference for the level of sweetness in ice cream, and has changed the amount of sugar used every year. In the 1960s and ’70s, Sarah used five ounces of sugar in a gallon of ice cream. Then, she used six and four ounces per gallon in the 1980s and ’90s, respectively. This suggests that in the 1960s and ’70s, Sarah’s sugar policy remained the same—five ounces per gallon. Thus, there was no policy change in the 1970s. In the ’80s, the amount of sugar in ice cream changed from five ounces to six ounces per gallon. Thus, there was a policy change of one ounce per gallon in the 1980s. In the 1990s, there was a policy change from six to four ounces per gallon. This example demonstrates that the amount of sugar Sarah used for a gallon of ice cream indicates policy output in her sugar policy, and the change of the policy output from the previous decade to the examined decade shows policy change. Similarly, this chapter measures policy change based on the change in the policy output from the previous Congress to the examined Congress. Policy output by Congress is calculated from the ADA scores. In order to compare the legislators’ ideology, political scientists have frequently examined interest group ratings of congressional members (Poole and Rosenthal 1984, 1997). Every year, interest groups, such as the Americans for Democratic Actions (ADA) and the American Conservative Union (ACU), select important roll call votes, and rate a legislator based upon the percentage of his or her votes in accordance with the groups’ positions. Scholars have found that the interest group ratings are very similar across groups (Poole and Rosenthal 1997; Brady and Volden 2006). Brady and Volden (2006, 116–20) calculate the average scores of eighteen interest group ratings for the members of the 103rd House. The average ratings are greatly correlated with the sole ADA ratings, yielding an unstandardized coefficient of 0.98. Poole and Rosenthal (1997) examine 96th Congress and find that their D-NOMINATE scores, which are measured based upon the members’ votes on roll calls, are highly related to the interest group ratings, such as ADA, ACU, and ACA, with correlation coefficients approximating 0.95. Since individual legislators’ ideologies can be measured by examining the concurrence of their votes with interest group preferences, voting outcomes in Congress at the aggregate level should likewise indicate the ideological point
30
The Other Side of Gridlock
of policy output by Congress as a whole. In other words, Congress can be examined in a manner similar to how an individual legislator can be examined, and a voting outcome in Congress can be measured as if a vote were cast by an individual legislator. The Americans for Democratic Action (ADA)8 selects forty bills per chamber it considers the most important during a Congress, and takes a position that either supports or opposes each measure. The ADA rates each legislator from zero to one hundred, depending upon the percentage of his or her votes in accordance with the ADA’s preference. For instance, if a legislator voted nay against a bill that the ADA opposed, or voted for a measure that the ADA supported, she would be in accordance with the ADA’s position. Should she vote nay against a bill that the ADA supported, or vote for a measure that the ADA opposed, her votes are at odds with the ADA’s preference. If 20 percent of her votes are in accordance with the ADA’s positions, the legislator receives twenty points for the ADA score. In a similar manner, I code passed legislation that the ADA opposed as “1” and passed bills that the ADA supported as “–1” in the House and Senate. I calculate the mean value of the measured legislation during a Congress. Thus, voting outcomes in Congress are measured in a manner similar to that of votes cast by an individual legislator. The measurement indicates the overall conservatism of the passed bills among the important legislation selected by the ADA: Policy Output () = {(1 x Po) + (–1 x Ps)} / (Po + Ps), where Po = number of passed bills ADA opposed and Ps = number of passed bills ADA supported.
The policy change is measured by an absolute value of the difference between the points of the policy output in an examined Congress and in the previous Congress: Policy change t = | t – t - 1 |, where = policy output.
The measurement above is likely to suggest the policy change in the realm of passed legislation. However, even if the passed bills show a significant policy change, there may be only a few passed bills, but many failed measures, among important legislation. In the measurement of weighted policy change, failed bills are considered responsible for the lack of policy change in legislative output by Congress. In this measurement, the percentage of passed bills among
31
Gridlock and Policy Stability
the legislation that the ADA has selected for a chamber is calculated. Then, the policy change, which is measured by the method explained above, is multiplied by the percentage of legislation passed: Weighted policy change = policy change x {(Po + Ps) / A} x 100, where A = number of all bills ADA selected.
Thus, the variable of the weighted policy change indicates a policy movement that is based upon all legislation, including failed measures, that the ADA has selected. Under the calculation of the weighted policy change, the measurement assumes that failed bills have caused policy stability (or no policy change). Figure 2.2 exhibits the policy change and weighted policy change in the House from the 83rd to 109th Congress. As shown in the figure, the two measures of policy change are highly correlated (r = .95). There are significant policy changes in the 83rd, 86th, 104th, and 107th Congresses. Except for the 83rd Congress, all these Congresses were under divided government. There were changes in the majority party in the House and in the presidential party in the 83rd Congress. The GOP became the majority in the House and Eisenhower became the president in the 83rd Congress. However, there was no change in the majority party or in the presidential party in the 86th Congress. In the 86th Congress, Democrats retained the majority they had held in the 85th Congress,
r = .947
Policy Change
Weighted Policy Change
Figure 2.2 Policy Change and Weighted Policy Change: House
32
The Other Side of Gridlock
and Eisenhower remained president from the 83rd to the 86th Congresses. After the landslide victory in the 1994 midterm election, the GOP, which had been in the minority for forty years, became the majority party in the 104th Congress. In the 107th Congress, the presidential party changed from Democratic (Clinton) to Republican (Bush), and the Republican Party remained as the majority party in the House. Figure 2.3 shows the policy change and weighted policy change in the Senate. Again, the two measures of policy change are highly correlated (r = .97). As shown in the figure, there are significant policy changes in the 83rd, 92nd, 93rd, and 104th Congresses. Except for the 83rd Congress, all these Congresses were in divided government. Among the four Congresses in question, there were changes in the majority party in the Senate in the 83rd and 104th Congresses. By increasing to forty-eight seats from the previous forty-seven, the Republican Party became the majority party in the 83rd Congress. In the 1994 midterm election, the GOP increased its seats from forty-three to fifty-two, and became the majority party in the Senate in the 104th Congress. In contrast, the 92nd and 93rd Congresses experienced no change in the majority party or in the presidency. In the 92nd Congress and the 93rd Congress, the Democratic Party remained the majority party, and Nixon remained president. The mean values of the measures of policy change and t-test results comparing periods of divided and unified government are examined. The results are provided in Table 2.6. For both the Senate and the House, there are slightly more policy changes and more weighted policy changes in divided
r = .97
Policy Change
Weighted Policy Change
Figure 2.3 Policy Change and Weighted Policy Change: Senate
33
Gridlock and Policy Stability
government than unified government. However, none of the differences are statistically significant. The mean value of the weighted policy change is smaller than that of the policy change in both the Senate and House. This discrepancy is due to the measurement of weighted policy change. Under the measurement of weighted policy change, the size of policy change is reduced by the percentage of failed bills. For the same reason, the standard deviations in the table show smaller variations in the weighted policy change variable than in the policy change variable. The results above suggest that policy change is not related to divided government, change in the majority party, or change in the presidential party. Table 2.7 summarizes the partisan condition of government in the eight chamberCongresses that resulted in significant policy changes. Among the eight cases, six of them occurred in divided government. Also, there were changes in the majority party in only four chamber-Congresses, and changes in the presidential party in only three chamber-Congresses. Thus, a factor other than party in government appears to cause policy stability. Conclusions
The iconoclasm of David Mayhew has generated intensive study concerning the impact of divided government on lawmaking. Some scholars examined legislative productivity using ratios with various denominators, such as presidential initiatives (Edwards, Barrett, and Peake 1997; Edwards and Barrett 2000), and policy agendas discussed in the New York Times editorials (Binder 1999, 2003). This chapter has revealed the negative, if modest, influence of divided government on the passage of presidential initiatives and public agendas. Mayhew (2005) asserts that there is evidence of agenda expansion in
Table 2.6 Mean Values and Variations in Policy Change House
Senate
Change
Weighted Change
Change
Weighted Change
Mean Unified Government
0.316
0.186
0.304
0.120
Mean Divided Government
0.475
0.248
0.424
0.207
Standard Deviation
0.355
0.196
0.315
0.147
High
1.167
0.778
1.317
0.593
Low
0.007
0.004
0.000
0.000
27a
27
27
27
N a
83rd – 109th Congresses
34
The Other Side of Gridlock Table 2.7 Policy Changes and Changes in Majority Party and Presidential Party House
Congress
Senate
83rd
86th
104th
107th
83rd
92nd
93rd
104th
Government
Unified
Divided
Divided
Divided
Unified
Divided
Divided
Divided
Change in Majority Party
Change
No Change
Change
No Change
Change
No Change
No Change
Change
Change in Presidential Party
Change
No Change
No Change
Change
Change
No Change
No Change
No Change
divided government, resulting in an endogeneity problem. This chapter has not found an agenda expansion. Nonetheless, it remains questionable whether these public and presidential agendas are plausible denominators. The number of agendas could be contracted or expanded by political upheavals, elections, public mood, media bias, and measurement method. Denominating the amount of passed legislation with such an influx of agendas would drastically change the value of the ratio indicator. The bottom line is that the ratio approach could show the proportion of failed agendas, if agendas are accurately measured at all. Besides, such a ratio measure merely indicates a conflict in legislative process, ignoring the policy content of legislative output. In this chapter, innovative policy output was defined as policy output deviant from the status quo policy. Accordingly, in order to study innovativeness of legislative output, this chapter explored policy change. Policy change, inversely viewed as policy stability, was measured by the difference in policy output between current and previous Congresses. This chapter found that policy change was not influenced by divided government, change in majority party, or change in presidential party. The next chapter proposes a model illustrating that the change in the preferences of the minority members in Congress, who are ideologically more extreme, has an impact on policy stability.
Chapter 3
Pivotal Interval Movement
A scholar once declared, the merit of formal modeling is the “rigor and precision of argument that it requires” (Morrow 1994, 6). Although some political scientists equate formal theories to mathematical models, the raison d’être of formal theory should be centered on logical and deductive construction of causal theories. In fact, Niou and Ordeshook contend that the formal theories are “reactions to a discipline mired in imprecision, vagueness, obscure logic, ill-defined constructs, non-testable hypotheses, and ad hoc argument” (1999, 87). Nonetheless, several scholars complain about the seeming lack of empirical testing of formal theories (Palfrey 1991; Enelow and Morton 1993; Walt 1999; Morton 1999). Walt (1999) reminds students that the “only way to determine if a theory is truly useful is to compare its predictions against an appropriate body of evidence,” and he maintains that “a logically consistent but empirically false theory is of little value” (1999, 13). If the lack of empiricism in formal theories is substantial, what political science needs today are applied models that are easily connected to empirical models. Such applied models should logically and succinctly explain phenomena in the real world of politics and must generate hypotheses that are empirically testable. This chapter develops an applied model to explicate policy stability in lawmaking. The pivotal interval movement model is predicated upon the assumptions of supermajoritarian mechanism of Congress and nonpartisan voting behavior. Senate Rule XXII confers on senators the right to engage in extended debate, the so-called filibuster, subject to a cloture vote, which may 35
36
The Other Side of Gridlock
be successfully invoked by sixty senators. This suggests that a minority (n–59) of leftist senators or rightist senators can block bills they dislike. Also, Article I, Section 7 of the Constitution grants the president the authority to veto legislation, which is subject to the congressional override by a two-thirds vote of the members present and voting in both the chambers. The predication is that one-third plus one members, whose preferences are close to the presidential preference, in either congressional chamber could reject unfavorable legislation. Thus, when the status quo policy is present within the so-called pivotal gridlock interval, which exists between the preferences of the 41st percentile senator (filibuster pivot) and 67th percentile legislator (veto pivot with an assumption of a conservative president), those bills that move the status quo leftward and rightward, respectively, will be blocked by 34 percent of legislators, who are conservative, and 41 percent of senators, who are liberal. The explanation above suggests that policies inside of the gridlock interval do not change (Krehbiel 1996, 1998; Brady and Volden 1998, 2006). More specifically, in this chapter, the pivotal interval movement model explains that the movement, rather than the width (Heitshusen and Young 2006; McCarty, Poole, and Rosenthal 2006, 177–84) or the change in width (Krehbiel 1998), of the gridlock interval would have an impact on policy stability. The model postulates that the gridlock interval of a current Congress becomes the status quo interval for the next Congress. In addition, the model introduces a concept, residuum, which is the portion of the gridlock interval of the previous Congress not overlapped by the gridlock interval of the new Congress. In this chapter, several core assumptions for the pivotal interval movement model are introduced. Second, the model is detailed and its hypothesis pertinent to policy stability is proposed. Third, the cartel agenda model (Cox and McCubbins 2002, 2005), as an alternative model, is explored. Last, the applied model of cartel interval movement is constructed, and its hypothesis is generated. Premises
The model of pivotal interval movement is based upon several premises of congressional lawmaking. The central premises include exogenous shocks, policy space, preference, and supermajoritarian procedure. Exogenous Shocks
The model of pivotal interval movement presumes that the positioning of the preferences of legislators, especially those of the veto and filibuster pivots, is
37
Pivotal Interval Movement
vital to lawmaking. As Figure 3.1 illustrates, the distribution of legislators’ preferences could be determined by various exogenous shocks. These exogenous shocks include elections, influence of constituency on the incumbent legislators, and extreme events. First, elections change incumbent legislators, thereby altering the configuration of the member preferences in Congress. One of the most salient characteristics of congressional elections is the candidate-centered campaign (Herrnson 2004). In contrast to the party-centered campaign in most democracies, candidates, not political parties, are the center of the U.S. congressional campaigns. Congressional candidates are self-selected rather than recruited by the major parties (Kazee 1994; Maisel, Fowler, Jones, and Stone, 1994). In addition, candidates, not parties, must raise campaign funds, assemble their organizations, formulate strategies, and assume all the responsibilities for their campaigns. In contrast to presidential elections, most congressional (especially House) election campaigns tend to be low-key events and voters in congressional elections are poorly informed about the candidates and the issues (Campbell, Converse, Miller, and Stokes 1965; Abramowitz 1980; Wright and Berkman 1986; Abramowitz and Segal 1992; Squire 1992; Carpini and Ketter 1996; Herrnson 2004). In order to make voting decisions in congressional elections, these uninformed voters are likely to use so-called voting cues, which are shortcut factors determining voting choice, including the incumbent and
Policy Output
Lawmaking
Preferences of Legislators/President
Social Events and Conditions
Preference of Constitutencies Exogenous Shocks
Figure 3.1 Macro Model of Policymaking
Election
38
The Other Side of Gridlock
party cues. The incumbency cue is one of the most influential cues for voting decisions. Incumbents running for reelection enjoy wider name recognition among the public than challengers do (Tufte 1975; Fiorina 1981; Campbell 1985). Unless an incumbent is tainted with scandal, her greater name recognition and previous constituency services are likely to engender sizable support from her constituency (Mayhew 1974; Fiorina 1978; Cain, Ferejohn, and Fiorina 1987; Jacobson 2004; Serra and Cover 1992). The party cue facilitates voters to estimate candidates’ ideological orientation and issue positions (Abramowitz and Saunders 1998). Republican candidates are generally assumed to support lower taxes, a deregulated economy, larger military spending, and socially conservative agendas. Democrats are viewed as being more supportive of social equity, welfare programs, and the interests of minorities and women. Many voters are likely to evaluate candidates based upon the party cue in their voting decisions. In addition to voting cues, other factors, such as the presidency, scandal, and midterm election, could influence the results in congressional elections. When voters are satisfied with the social and economic conditions of the nation, usually accompanied by a higher presidential popularity, they are more likely to vote for the incumbents or congressional candidates of the presidential party. In contrast, when a scandal is present or any national policy failure is salient, voters may support challengers (nonincumbents), candidates of the opposition party to the presidential party, and/or the candidates of the minority party in Congress (Tufte 1975; Fiorina 1981; Uslaner and Conway 1985; Lewis-Beck and Rice 1992). Also, researchers have noticed that the presidential party nearly always loses seats in midterm elections.1 Several scholars argue that the electorate’s return to normal partisan equilibrium in midterm elections, after the less partisan voting in the presidential election, causes the midterm losses of the presidential party (Tufte 1975; Abramowitz 1985; Abramowitz, Cover, and Norpoth 1986; Oppenheimer, Stimson, and Waterman 1986; Jacobson 1990; Campbell 1991). Although midterm elections, scandals, and poor economy increase elite turnover in congressional elections, overall, surprises are few and victory is assured for most incumbents. In the 1990s and 2000s, less than 10 percent of incumbent House members and less than 20 percent of incumbent senators were defeated when they sought reelection (Herrnson 2004; Jacobson 2004). This dominance of incumbent legislators is likely to cause a tenacious immobility in the configuration of the decision makers’ preferences in Congress. Secondly, preferences of voters in districts could have an ample impact on the preferences of incumbent legislators. As Herrnson (2004) argues, members of Congress are in a state of permanent campaign. Campaigns for reelection
Pivotal Interval Movement
39
are likely to start on the day the members of Congress win office. Mayhew (1974) observes that the members of Congress, being single-minded reelection seekers, expend tremendous amounts of resources to expand name recognition, claim credit for popular governmental actions, and take issue positions congruent with the constituents’ preferences. As for the legislators’ preferences, Miller and Stokes’s classic explanation in their diamond model (1963) postulates that constituency opinions shape the legislators’ values and their perception of constituency attitudes, both of which influence the legislators’ voting decision. However, electorates are multifaceted entities consisting of multiple constituencies (Fenno 1978). In addition to the whole constituency, which he calls the “geographic” constituency, Fenno observes that there are also “reelection,” “primary,” and “personal” constituencies. The reliable supporters and the fellow partisans form the reelection constituencies, and the volunteers, financial contributors, and local party elites make up the primary constituencies. The personal constituencies consist of the legislators’ closest friends. Some studies show that legislators may respond better to their reelection constituencies than to the geographic constituencies (Bender 1994; Bullock and Brady 1983; Langbein 1993; Powell 1982; Shapiro et al. 1990; Wright 1993). Nonetheless, these findings are not indicative of an absence of legislative responsiveness to the geographic constituencies. Still, several scholars find a strong relationship between the legislators’ ideology and the ideology of geographic constituencies (viz. Erikson and Wright 2005). Since the 1980s, some researchers have studied legislative shirking, which are voting decisions which deviate from the constituencies’ preferences. Largely based upon the difference between legislators’ voting decisions and their constituencies’ ideologies,2 the analysts measure the legislators’ own personal ideologies. They observe that legislators often shirk because constituencies have limited ability to monitor their trustees’ decisions (Kalt and Zupan 1984, 1990; Carson and Oppenheimer 1984; Davis and Porter 1989; Coates and Munger 1995; Uslaner 1999). Nevertheless, most of the shirking study still finds a solid congruence between constituency ideology and legislative voting decisions.3 In addition, constituency characteristics, other than their ideology, could influence legislators’ preferences. A recent study conducted by McCarty, Poole, and Rosenthal (2006) finds that legislators’ preferences are significantly influenced by the income level and the percentage of ethnic minority voters in their districts. Thus, various socioeconomic and political characteristics of constituencies are likely to influence the preferences of members of Congress. Lastly, extreme events may change legislators’ preferences through an emergence of new issues or drastic change in the nature of an old issue. Although
40
The Other Side of Gridlock
policymaking of many policy programs involve incremental decision makings to alleviate familiar and routine problems, the nation sometimes faces a natural disaster, international crisis, economic catastrophe, new disease, etc. When calamitous events and crises occur, legislators perceive that the status quo policy is undesirable and seek a significant policy change. For example, after the terrorist attacks of September 11, 2001, and the subsequent anthrax threats, the great majority of legislators found the status quo of homeland security policy unacceptable. In a few weeks, Congress passed the Patriot Act and authorized the law enforcement agencies to wiretap, intercept e-mails, and detain aliens for up to seven days without a formal charge.4 Thus, elections, constituencies, and extreme events could influence the configuration of legislators’ preferences. An increasing amount of research suggests that the preferences of members of Congress have grown increasingly polarized (Rohde 1991; Aldrich 1995; Poole and Rosenthal 1997; Jacobson 2000, 2004; Bond and Fleisher 2000; Hetherington 2001; McCarty, Poole, and Rosenthal 2006; Sinclair 2006). These studies reveal that the preferences of legislators have recently become homogeneous within the parties but have grown to be more heterogeneous between the parties. McCarty, Poole, and Rosenthal (2006) contend that the recent polarization in Congress has caused policy stability. In contrast, this chapter posits that a lack of change in the preferences of the ideologues, who include the strongly conservative and liberal members in Congress, leads to policy stagnation. Policy Space
The ideal point of a legislative agenda, the status quo point of a policy, and the preferences of legislators can be placed on a continuum of liberal-conservative dimension. Thus, as Figure 3.2 exemplifies, policy space is assumed to be a unidimensional, Euclidean space. In addition, the policy space is presumed to be a finite interval in that the policy continuum has two endpoints, ranging from extremely liberal to extremely conservative. For various purposes, many political scientists have traditionally examined the interest groups’ ratings of legislators, which usually range from zero to one hundred. Interest groups, such as the Americans for Democratic Action (ADA), the U.S. Chamber of Commerce, and the American Conservative Union (ACU), I—————————— I ————————I Most Liberal
Moderate
Most Conservative
Figure 3.2 Unidimensional Policy Space
Pivotal Interval Movement
41
choose those roll call votes most important to them, and rate a legislator based upon the percentage of her votes that are in accordance with the groups’ positions. On the other hand, Poole and Rosenthal (1997) measure legislators’ ideal points based upon the members’ votes on all nontrivial roll calls. Legislators’ preferences could be metrically scaled so as to minimize the prediction error for the members’ voting decisions on roll call votes. Significantly, the interest group ratings are very similar across groups (Brady and Volden 2006), and highly related with the roll call ratings (Poole and Rosenthal 1997).5 This indicates that these indices are based on the same unidimensional policy space and their measurements maintain substantial methodological validity. Preference
Utility (%)
Legislators are assumed to have single-peaked preferences in the policy space. As Figure 3.3 shows, each lawmaker maintains an ideal point for a particular policy. As the position of the legislative agenda moves away from the member i’s ideal point, Xi, the utility of the legislation for i decreases pro portione. Secondly, the legislators’ preferences are symmetrical. This indicates that the distance of a legislative agenda from a legislator’s ideal point equally decreases the utility of the legislation regardless of whether the agenda is too liberal or too conservative to her ideal point. For example, Figure 3.4 shows a symmetrical utility function of tax cut proposals for a legislator. The figure illustrates that this legislator’s preference is a $500 billion tax cut. Given that her preference is symmetrical, a 20 percent larger tax cut ($600 billion cut) and a 20 percent smaller tax cut ($400 billion cut) are likely to give her an identical utility (80 percent). Thus, she is assumed to be indifferent between these two tax cut proposals.
Xi Policy Space
Figure 3.3 Utility Function for Legislators
42
The Other Side of Gridlock ($500, 100%)
Utility (%)
($600, 80%)
($400, 80%)
($700, 60%)
($300, 60%)
($200, 40%)
($800, 40%)
($900, 20%)
($100, 20%)
Tax Cut ($ billion) Figure 3.4 Utility Function of Tax Cut Agendas
Thirdly, legislators are assumed to be rational, utility maximizers. Since legislators’ preferences are single-peaked and symmetrical, the lawmakers are likely to support an agenda that is closest to their ideal points, offering them the highest utility. For instance, Figure 3.5 exemplifies that the legislator i’s ideal point for minimum wage policy, Xi, is $5 per hour while the status quo of the policy, Q, is $4 per hour. Suppose a member proposes the legislation, _, to change the minimum wage to $5.5 per hour. While the $5.5 minimum wage is 10 percent different from i’s preference, the status quo of the minimum wage, $4, is 20 percent deviant from her ideal point. Assuming that i’s preference is symmetrical, she is likely to prefer _ to Q. On the other hand, if another member introduced a bill, `, to change the minimum wage to $6.5 per hour, it is 30 percent deviant from i’s ideal point. Therefore, she is likely to prefer Q to `, and thereby oppose the legislation `. Lastly, all members of Congress maintain constant preferences among agendas, and their voting decisions on the floor are likely to be polarized based
$4
$5
$5.5
$6.5
—————————I————————I————I————————— I ———————— Q
Xi
_
`
Figure 3.5 Status Quo and Agendas of Mininum Wage
43
Pivotal Interval Movement
upon their conservatism (or liberalism). In other words, extremely conservative members remain extremely conservative and moderately liberal legislators remain moderately liberal on almost all bills. Figure 3.6 hypothesizes that twenty members, each numbered from 1 to 20, are juxtaposed on the policy space based upon their conservatism. Assume that a bill is proposed to move the status quo of present policy, Q, to the new point, Q’. For the legislators 14 to 20, Q is closer than Q’ to their ideal points. In contrast, Q’ is closer than Q to the ideal points of the members 1 to 13. Thus, the legislators 1 to 13 would vote for the bill whereas members 14 to 20 are likely to cast a vote of “nay.”
1—— 2—— 3 —— 4 —5——— 6 — 7 ——— 8 9—— 10 12 — 13 14—— 15——16 17—— 18 19—— 20 — ——11 ———— ——— ——— ——— — Liberal
Q’
Q
Conservative
Figure 3.6 Twenty Members in a Legislature
The example suggests that for all bills, there should be a cutting point dividing votes on the midpoint between the status quo point and the ideal point of a revisionist measure. In Figure 3.7, the cutting point, a, is on the midpoint between Q’ and Q. In the figure, the members 1 to 13, who are to the left of the cutting point, a, vote for the bill to move Q to Q’. On the other hand, the legislators 14 to 20, who are to the right of the cutting point, vote against the measure. Cutting Point ( a) = ½ (Q + Q’), where Q = status quo, and Q’ = proposed policy output.6
Poole and Rosenthal’s D-NOMINATE scores, which scale the legislators’ ideal points, predict 80 percent or higher portion of the members’ voting decisions in the Congresses between 1789 and 1989 (Poole and Rosenthal 1997, 31–34). Thus, legislators’ decisions on numerous votes are highly consistent with their ideology. y y y y y y y y y y y y y n n n n n n n 1 2 3 4 5 6 7 8 9 10 11 12 13 14—— 15——16 17—— 18 19—— 20 ———————————————————————— ——— ——— ——— — Liberal
Q’
Q Cutting Point, a
Figure 3.7 Voting Decisions of Twenty Legislators
Conservative
44
The Other Side of Gridlock
Supermajoritarianism
Essentially, in the subcommittees, standing committees, chambers, and conference committees, most legislation is passed if and only if a majority of the members support a bill. Thus, the façade of congressional procedure maintains the majority rule. Duncan Black’s median voter theory (1958) explains such majority rules and their policy consequences. Black observes that when a bill is introduced in a majoritarian legislature, the policy point of the legislation is amended and moved to the ideal point of the median voter. The logic of the theory is illustrated in Figure 3.8. In the figure, when the status quo of a policy, Q, is relatively liberal, a majority of the legislators, who have more conservative preferences, introduce and successfully pass a bill to move Q to the chamber median, M. This indicates that by every legislative session, policy outputs move to its new chamber median. Nonetheless, as for the U.S. Congress in the real world, students of lawmaking have acknowledged the institutional features of supermajority rules—ubiquitously, Senate filibusters and the presidential vetoes. Some scholars observe that support from a supermajority of the legislators is sine qua non for the passage of legislation since a minority of legislators can block bills through filibusters and presidential vetoes (Mayhew 1991, 2005; Brady and Volden 1998, 2006; Krehbiel 1998, 2006). For instance, Mayhew finds that approximately 70 percent of the important laws enacted from 1947 to 1990 won support from two-thirds majorities in both chambers and majorities of Democratic and Republican members in both houses (2005, 119–25). Senate Filibuster
The Senate is well known for its slow and often cumbersome decision-making process. With few stipulations, senators can hold the floor indefinitely by engaging in extended debate (filibuster). Senate Rule XXII, adopted in 1917,7 allows for a cloture vote to end such extended debates. In order to stop a filibuster, sixteen senators must sign a petition for a cloture, and at least sixty senators8 must vote to invoke cloture two days later. This suggests that a minority of senators, n–59 members, could take dilatory actions to delay, modify, or block any legislative agendas.
M
I Liberal
Q
I
I
Q’
Conservative
Figure 3.8 Median Voter Theory
Pivotal Interval Movement
45
In the past, some filibusters were wielded so dramatically that they became legends in American politics. In 1935, Huey Long (D-LA), a proponent of the Senate confirmation process for the National Recovery Administration’s senior employees, engaged in a filibuster of 15 hours 30 minutes by reading from the Constitution and the Bible. In 1957, Strom Thurmond (D-SC) exercised the longest filibuster in Senate history, 24 hours 18 minutes, against civil rights legislation. More recently, numerous filibusters were exercised in the Democratic-majority 103rd Congress, to block many of President Clinton’s legislative initiatives. In the 103rd Congress, Republican filibusters successfully killed many pieces of major legislation, including the economic stimulus package and campaign finance reform bill. As for the Congresses in the 1990s, overall, Sinclair reports that the frequency of filibusters averaged twenty-eight per Congress (2002, 242–43). Today, experts analyze that threats of filibuster are more common than actual filibusters and the former are as effective as the latter for blocking legislation (Sinclair 2007, 67–72; 2006, 208–14; Oleszek 2004, 239–46). Once, Senator Robert Byrd stated, “It’s the threat of a filibuster that keeps a bill from coming up” (Oleszek 2004, 239). Credible threats of filibuster deter committees from reporting a measure to the Speaker of the House or to the Senate president and discourage a motion to take up the legislation for the floor consideration. For instance, in the first Congress of the Clinton presidency, the threats of filibuster by the Republican minority successfully blocked bills of the Superfund program, clean water regulations, telecommunication reforms, and other important legislation (Sinclair 2007). Threats of filibuster could be implicit as well as explicit. Senators can exert an implicit threat of filibuster when they send a notification called a hold to their party leaders in order to express their objection to a floor consideration of a bill. In the Senate today, threats of filibuster are conspicuous. The former Senate majority leader George Mitchell observes that “the threat of filibuster is now a regular event in the Senate, weekly at least, sometimes daily” (Joint Committee on the Organization of Congress, 103rd Congress 1993). In the 1990s and early 2000s, approximately one-half of major legislation was filibustered or threatened to be filibustered (Sinclair 2002, 252–53; 2006, 211–24). Since the Senate (and the House) is always packed with much legislation for consideration and is deadline-driven, Senate leaders usually do not wish to risk a filibuster by scheduling a measure any member has threatened to filibuster. In order to stop a filibuster, or dissuade it ex ante, at least sixty senators must support a bill. This suggests that by threatening or exercising a filibuster, a minority (n–59) of senators can block legislation to which they object. Figure 3.9 exhibits bills that minority blocs of senators are likely to obstruct.
46
The Other Side of Gridlock 1st
41st
60th
100th
I
I
I
I
Liberal
Q2
Q1
Conservative
Figure 3.9 Filibuster Gridlock Interval
In Figure 3.9, one hundred senators are juxtaposed based upon their preferences, say conservatism, and the status quo of a policy, Q1, is on the ideal point of the sixtieth member. Legislation to move Q1 leftward is likely to make the policy more distant from, hence worse for, the preferences of the sixtieth to one hundredth legislators. Thus, the forty-one senators, who include the sixtieth to one hundredth members en bloc, are likely to impede the legislation by threatening or exercising a filibuster. Next, suppose the status quo of another policy, Q2, is on the ideal point of the forty-first senator. A bill to move Q2 rightward is likely to make the policy more distant from the preferences of the first to forty-first legislators. Consequently, these forty-one senators are likely to block the legislation. Thus, when the status quo is present in the interval between the forty-first and sixtieth senators, which is the so-called filibuster gridlock interval, any legislation to move the status quo is likely to be blocked. Presidential Veto
Once, Woodrow Wilson wrote that the “president is no greater than his prerogative of veto makes him; he is, in other words, powerful rather as a branch of the legislature than as the titular head of the Executive” (1885, 260). Thus, Wilson observed that the veto power of the president gave him enormous command in the legislative branch. The Constitution, Article I, Section 7, grants the president the authority to dismiss legislation passed by Congress. After a bill is sent to the president, she or he has ten days, excluding Sundays, to sign it, veto it, or take no action. When the president vetoes a bill, her or his disapproval of the legislation is subject to congressional override by a two-thirds vote of the members present and voting in the two chambers. For decades, the study of the presidential veto has centered on its propensity (Lee 1975; Copeland 1983; Rohde and Simon 1985; Simonton 1987; Wooley 1991; Shull and Shaw 1999; Gilmour 2002) and its sustainability (Hoff 1992; Cameron 2000; Saeki 2004). Several studies find a higher veto propensity when a president is less popular (Lee 1975; Rohde and Simon 1985; Wooley 1991; Shull and Shaw 1999), during reelection years (Shull and Shaw 1999), and in divided government (Lee 1975; Copeland 1983; Rohde and Simon 1985; Simonton 1987; Shull and Shaw 1999). On the other hand, a few scholars have studied veto sustainability by examining successful congressional overrides.
Pivotal Interval Movement
47
These studies reveal that divided government, the significance of legislation (Cameron 2000), and a larger margin of the president’s opposition party in Congress (Saeki 2004) increase override attempts by Congress, but fail to explain the rate of successful veto overrides. Nonetheless, the quantity of the sustained vetoes may not indicate the overall effect of the presidential veto power in the legislative arena since a threat of veto could sufficiently discourage legislative agendas (Hinckley 1985; Kiewiet and McCubbins 1988; Edwards, Barrett, and Peak 1997; Sinclair 2007; Cameron 2000; Deen and Arnold 2002). Cameron (2000, 18–20) perspicaciously calls the presidential veto power the second face of power. The second face of power is based upon anticipation of action (Bachrach and Baratz 1962; Nagal 1975). When _ (i.e., Congress) anticipates that _’s action (i.e., Passage of legislation) will result in an undesirable response from ` (i.e., Presidential veto), _ restricts herself (i.e., Modification of bill) so as to head off `’s response. Thus, the possibility of exerting a veto can shape the content of legislation even if the veto is rarely wielded. In fact, Sinclair observes that veto threats are more frequent than actual vetoes, and the threats of veto are ubiquitous in divided government (2000, 145–46). Sinclair finds that none of the major legislations were subject to veto threats and only 4 percent were subject to actual vetoes in the 103rd Congress of unified Democratic government. However, in the 104th and 105th Congresses of divided government, 60 percent and 69 percent, respectively, of the major legislation were exposed to veto threats. In those same Congresses, 10 percent and 5 percent of the major bills were subject to actual vetoes (Sinclair 2000, 145–46). Most of the previous studies on veto propensity have been contingent upon the simplistic assumption that the president vetoes legislation because Congress passes bills she or he does not want to approve. Nonetheless, the contradiction is that the president usually informs Congress about the bills she or he is likely to veto before the legislation is sent to the floor (Cameron 2000; Sinclair 2007; Gilmour 2007; Deen and Arnold 2002). This begs the question why Congress passes bills that the president has indicated ex ante that she or he would veto. Cameron’s sequential veto bargaining model (2000) presumes that vetoes (and overrides) occur because of incomplete information and resulting miscalculations about the opponents’ behavior during the bargaining process between the president and Congress. In contrast, Gilmour (2007) contends that Congress purposefully passes bills that the members know the president will veto, in order to crystallize their position differences and provoke the president in public.9 While the dispute between Cameron and Gilmour is subject to further empirical tests, both arguments underscore that the threat of presidential veto effectively forces Congress to modify or relinquish legislation.
48
The Other Side of Gridlock 1st
67th percentile
I
I
Liberal
Q1
100th percentile
I p
Conservative
Figure 3.10 Veto Gridlock Pivot
Formally, Figure 3.10 illustrates when the president is likely to threaten and/or make use of her or his veto power. In Figure 3.10, legislators are placed based upon their conservatism. The president, p, is conservative (Republican), and the status quo of a policy, Q1, happens to be on the ideal point of the 67th percentile member. In order to override a presidential veto ex post or discourage it ex ante, two-thirds of the legislators must support a bill in both chambers. This suggests that with a threat or exercise of presidential veto, a minority (n/3 + 1) of legislators whose preferences are close to the president’s preference in either chamber can block legislation they dislike. In Figure 3.10, legislation to move Q1 leftward is likely to make the policy more distant from, hence worse for, the preferences of the 67th percentile to the 100th percentile legislators. Thus, these legislators, who comprise the one-third plus one members of the chamber, could impede the legislation with a threat or exercise of the presidential veto. Pivotal Gridlock Interval
While the model of filibuster interval suggests that 41 percent of senators could block bills, the model of veto gridlock pivot suggests that an even smaller amount, 34 percent of legislators, can inhibit legislation. Thus, in Congress overall, there is an interval, the so-called pivotal gridlock interval, between the filibuster pivot and the veto pivot (Krehbiel 1996, 1998; Brady and Volden 1998, 2006). Figure 3.11 assumes a unicameral legislature for the benefit of simplification. In Figure 3.11, when the president is conservative, the filibuster pivot is the 41st percentile member and the veto pivot is the 67th percentile member. In contrast, as Figure 3.12 shows, when the president is liberal, the 34th percentile member and the 60th percentile member are, respectively, the
I Liberal
I
Gridlock Interval
I
I
41st percentile
67th percentile
filibuster pivot
veto pivot
p
Figure 3.11 Pivotal Gridlock Interval (Conservative President)
Conservative
49
Pivotal Interval Movement I Liberal p
I 34th percentile veto pivot
Gridlock Interval
I 60th percentile
I Conservative
filibuster pivot
Figure 3.12 Pivotal Gridlock Interval (Liberal President)
veto and filibuster pivots. When the status quo is present in the pivotal gridlock interval, any legislation to move the status quo is likely to be blocked. Krehbiel (1996, 1998) and Brady and Volden (1998, 2006) have developed the theory of pivotal gridlock interval, contending that gridlock occurs not because of divided government, but because the status quo of many policies exists within the pivotal gridlock interval. Apropos of empirical test of the impact of the pivotal gridlock interval, Krehbiel computes the change in the width of gridlock interval based upon interelection swings (1998). Krehbiel estimates expansion and contraction of the gridlock interval in every Congress predicated upon the presidential party’s gains in two congressional chambers and change in the president’s party (1998, 58–61). When the presidential party does not change and the presidential party gains seats, the gridlock interval is minimized. If the presidential party loses seats, the gridlock interval is extended. When the presidential party changes and the (new) president’s party gains seats, the gridlock interval is minimized. If the presidential party changes and the president’s opposition party gains seats, the gridlock interval is assumed to remain the same. Krehbiel finds a negative impact of the change in the width of the pivotal gridlock interval on the change in the amount of Mayhew’s important laws (1998). On the other hand, Hetshusen and Young (2006) measure the distance in the D-Nominate Score between the senators who are present on the one-third point and on the two-third point.10 By examining the change in the number of U.S. Codes, the authors’ ARIMA models exhibit the negative influence of the width of the gridlock interval measured based on the D-Nominate Score and the negative impact of Krehbiel’s change in the width of the gridlock interval. In contrast to the approaches by Krehbiel and by Hetshusen and Young, the next section explains that the movement, rather than the width, of the gridlock interval should influence policy stability. In particular, the applied model of the pivotal interval movement focuses on the portion of the previous gridlock interval that the new gridlock interval does not overlap.
50
The Other Side of Gridlock
Pivotal Interval Movement Model
As for policy stability, the supermajoritarian school (Krehbiel 1996, 1998, 2006; Brady and Volden 1998, 2006) indicates that policy will not change when the status quo point exists inside of the gridlock interval between the filibuster and veto pivots. Inversely, policy is more likely to change when its status quo exists outside the gridlock interval. Figure 3.13 presumes that legislators’ ideal points are spread out over a finite interval. Let F, V and p respectively denote the ideal points of the filibuster pivot (41st percentile senator), veto pivot (67th percentile legislator) and a (conservative) president. When the status quo, q, exists within the gridlock interval between F and V, measures to move the status quo rightward and leftward would generate new policy outputs, which are more distant from, hence unfavorable for, F and V, respectively.11 Thus, these measures, respectively, are likely to be blocked by a filibuster and presidential veto, and the status quo is likely to remain unchanged. This proposition shall be summarized as follows: q’ = q, if F q V, where q is the status quo, and q’ is a new policy output.
Next, in Figure 3.13, let Mf and Mv index the preferential reflection points of the chamber median, M, through F and V, respectively. This indicates that F is the midpoint between Mf and M, and V is equidistant to M and Mv. Thus, the distance between F and M and the distance between Mf and F are equal, and the distances between M and V and between V and Mv are the same. As the figure shows, when the status quo, q, is present in the interval between Mf and F, the majority of the chamber who lies to the right of M, would like to move q to M. However, in this situation, F is likely to support a bill to move q
Gridlock Interval 3/10
Mf 1/10
F ) q ) V Mf ) q < F; V < q ) Mv
q q
q < Mf ; q > Mv New Policy q’
M
F 1/10
M
1/6
q
q
q’
q’
V
1/6
Mv
1/6
q q
q’q’ 2F-q
p
q
2V-q
Figure 3.13 Pivotal Gridlock Interval and Policy Change
M
Pivotal Interval Movement
51
rightward insofar as a new policy q’ is closer than q to her ideal point. Otherwise, F and the members who are to the left of F would filibuster the measure. Therefore, a policy is likely to shift to the reflection point of q through F, which is on 2F–q.12 q’ = 2F–q, if Mf q < F.
When the status quo exists in the interval between V and Mv, a diametrical movement should occur, and a policy is likely to change to the reflection point of q through V, which is on 2V–q.13 q’ = 2V–q, if V < q Mv.
Next, when the status quo exists to the left of Mf, a new policy q’ on M would be closer than q to the preference of F. Thus, F and the members who are to the right of F support legislation to move q to M, and the resulting new policy q’ will be on the chamber median, M. Although the extremely liberal members prefer q to q’(M), they do not have sufficient votes to filibuster the measure. q· = M, if q < Mf.
Similarly, when the status quo exists to the right of Mv, V and the members who are to the left of V prefer a new policy q’ on M to q, thereby supporting any measures to move q to the chamber median, M. Therefore, the policy is likely to move to the chamber median. q’ = M, if q > Mv.
Thus, contingent upon the position of the status quo, the location of the new policy, q’, would be either q, 2F–q, 2V–q, or M. ż (q’) = {q, 2F–q, 2V–q, M}.
Notice that all these points are present within the gridlock interval between F and V. Therefore, new policy outputs are likely to converge within the gridlock interval regardless of the position of the status quo: q’ D [F, V].
52
The Other Side of Gridlock Ft–1
Vt–1 Residuum
Congress t–1
__________________________ Ft
q
Vt
Congress t
q’’
Figure 3.14 Movement of Pivotal Gridlock Interval and Policy Change
The pivotal interval movement model assumes that the preferences of pivotal legislators are exogenously positioned for every Congress. Figure 3.14 illustrates a hypothetical movement of the pivotal gridlock interval from Congress t–1 to Congress t. For Congress t, the status quo points of various policies are likely to be present between Ft–1 and Vt–1, because the status quo points for Congress t are the policy outputs by the previous Congress. The policy outputs by Congress t–1 converge within its gridlock interval between Ft–1 and Vt–1: q D [Ft–1, Vt – 1].
When the status quo point is in the gridlock interval between Ft and Vt, it is unlikely to be changed. Thus, when the status quo point, q, exists within the portion of the previous gridlock interval of Congress t–1 that the current gridlock interval of Congress t overlaps, it is likely to maintain the same position. q’ = q, if Ft q Vt – 1.
In contrast, as Figure 3.14 shows, when the status quo point, q, is present in the portion of the previous gridlock interval of Congress t–1 that the current gridlock interval of Congress t does not overlap, the policy is likely to move into the gridlock interval of Congress t. q’ D [Ft, Vt], if Ft – 1 q < Ft.
This exemplification suggests that when the residuum, the portion of the gridlock interval of the previous Congress that the gridlock interval of the new Congress does not overlap, is large, status quo points of more policies are likely to change. Consequently, the model of pivotal interval movement generates the hypothesis below. Hypothesis 1: When the residuum in the pivotal gridock interval, q (p), is larger, more change in policy is likely to occur, where q t = (Gridlock Interval t – 1) \ (Gridlock Interval t).
Pivotal Interval Movement
53
Alternative Theory: Cartel Agenda Model
In stark contrast to the nonpartisan, supermajoritarian assumptions of the pivotal interval movement model, the cartel agenda model (Cox and McCubbins 2002, 2005) postulates a robust influence of political party on lawmaking. Therefore, it is important to analyze the cartel agenda model, and compare its assumptions about policy stability with the indications by the pivotal interval movement model. Cox and McCubbins (1993, 2002, 2005) observe that a political party functions to solve collective action problems related to externalities and coordination. According to Cox and McCubbins, the legislators’ goals include reelection, internal advancement, and majority status. These goals are interrelated since the majority party status gives the party members a larger share of pork barrel projects, a significant fundraising advantage, and the speakership and committee chairships, all of which are likely to enhance the members’ probability of reelection. Therefore, the reputation of a member’s party, as a brand name, is critically important because it affects the members’ probability of reelection and the party’s probability of securing majority. Thus, a political party’s reputation is a public good to the candidates sharing the party label. Cox and McCubbins (2005, 21) state that if “a party’s reputation improves or worsens, all members benefit or suffer together, regardless of whether contributed to the improvement or worsening.” Cox and McCubbins contend that the reputation of a party is largely contingent upon its record of legislative accomplishment. However, in the legislative process, multifarious national, regional, and district interests could lead the majority party members to confront each other. The implication is that there is a free rider problem. Legislators may defect from the party position in order to please their constituencies, while presuming that they can still benefit from the collective reputation of the party, assuming that other members remain loyal to the party. In order to resolve such collective action and coordination problems, the members of the majority party will favor compliance with party leaders and collectively monopolize the agenda in order to increase the party’s legislative success, enhance its reputation, and maintain majority status. Cox and McCubbins stress that the majority party holds the negative agenda power, which is the ability to block legislation in the legislative process. The majority party secures all committee chairs and speakership. It also maintains a disproportionately large share of seats on the Rules Committee (Cox and McCubbins 1993, 2006; Aldrich and Rohde 2000). These “senior partners” of the majority party, including committee chairs, speaker, and members of the Rules Committee, could block bills from reaching the floor.
54
The Other Side of Gridlock
Cox and McCubbins (2005, 26–29) maintain that the senior partners of the majority party exercise their agenda power for the interest of their party, rather than their personal goals, in order to retain loyalty within the party and party leadership. Therefore, the senior partners are “expected never to push bills that would pass despite the opposition of a majority of their party” (Cox and McCubbins 2005, 27). The Cartel Interval Movement Model
The cartel agenda model predicates that the senior partners of the majority party, including committee chairs, the preponderant members in Rules Committee, and the Speaker of the House, are strongly responsive to the majority of the majority party members (Cox and McCubbins; 2002, 2005). The model assumes that when the majority of its party membership dislikes a bill, the majority party leadership will block the legislation from reaching the floor. In Figure 3.15, let J, M, and N denote the ideal points of the median of the majority party, the median of the floor, and the median of the minority party, respectively. Mj is the reflection point of M through J, in that the distances between J and M and between Mj and J are equal. In the figure, any measures, if reported to the floor, will be amended to meet the preference of the chamber median, M, and will be passed.14 However, when the status quo point, q, is present between Mj and M, q is closer than M to J. This indicates that the majority of the majority party members who lie to the left of J, prefer q to a new policy output on M. Therefore, in this situation, the majority party leadership blocks any revisionist agenda to move q from reaching the floor. Consequently, when q exists within the cartel gridlock interval between Mj and M, its policy is unlikely to change. q’ = q, if Mj q M.
Gridlock Interval Mj
Mj ) q ) M q < Mj;q > M New Policy q’
J
q
M
q q’
q M
N
q
q
Figure 3.15 Cartel Gridlock Interval and Policy Change
M
55
Pivotal Interval Movement
In contrast, when q is to the left of Mj, q’ on M would be closer than q to J and the members who are present to the right of J, which include the majority of the majority party. Thus, the measure to move q to q’ is likely to be reported to the floor and supported by the majority of the chamber. Similarly, when q is to the right of M, q’ on M would be closer than q to the members who are to the left of M, which include the majority of the majority party. Consequently, the agenda to move q to q’ is likely to be reported to the floor and supported by the majority of the chamber. Therefore, when the status quo point of a policy is present outside the gridlock interval, both the majority in the chamber and the majority of the majority party prefer a new status quo, q’, on M to q. The bill is likely to be supported by the majority of the chamber, and the status quo policy would move to M. q’ = M, if q < Mj or q > M.
Thus, depending upon the position of the status quo, the location of the new policy, q’, would be either q or M. ż (q’) = {q, M}.
These points are only present within the gridlock interval between Mj and M. Thus, new policy outputs are likely to converge within the cartel gridlock interval between Mj and M, regardless of the status quo position. q· D [Mj, M].
Next, Figure 3.16 illustrates a hypothetical movement of the cartel gridlock interval from Congress t–1 to Congress t. For Congress t, status quo points of the various policies exist between Mj t–1 and M t–1. q D [Mj t–1, Mt–1].
Mj t - 1 Residuum
M t-1
______________________________ Mj t
q
Congress t-1 Mt
Congress t
q‘
Figure 3.16 Movement of Cartel Gridlock Interval and Policy Change
56
The Other Side of Gridlock
For Congress t, the status quo point in the cartel gridlock interval between Mj t and M t is unlikely to be changed. This indicates that when the status quo point, q, exists within the portion of the previous cartel gridlock interval of Congress t–1 that the current gridlock interval of Congress t overlaps, it is likely to maintain the same position. q’ = q, if Mj t q M t–1.
In contrast, when the status quo point, q, exists in the portion of the gridlock interval of Congress t–1 that the cartel gridlock interval of Congress t does not overlap, the status quo is likely to move into the gridlock interval of Congress t. q’ D [Mj t, M t], if Mj t – 1 q < Mj t.
Thus, when the residuum, q (c), the portion of the cartel gridlock interval of the previous Congress that the gridlock interval of the new Congress does not overlap, is large, status quo points of more policy programs are likely to change. Accordingly, the hypothesis below is tested in the next chapter. Hypothesis 2: When the residuum in the cartel gridock interval, q (c), is larger, more change in policy is likely to occur where q t = (Gridlock Interval t–1) \ (Gridlock Interval t).
Gridlock Intervals in Bicameral Congress
While the United States Congress is a bicameral legislature both in de jure and de facto terms, the previous sections have detailed the pivotal interval movement model and cartel interval movement model in a unicameral context. Essentially, the cartel agenda model is a unicameral theory. The cartel agenda model postulates that the majority of the majority party in a chamber maintains the ability to block measures within that chamber. Nonetheless, Cox and McCubbins imply bicameral interactions when they define and argue for what they call “inconsequential rolls” (2005, 106–23). In the postbellum House, Cox and McCubbins find that forty-nine bills were passed in spite of opposition from the majority of the majority party. The Cartelists maintain that approximately one-half of these passed bills, which are the so-called majority rolls, are measures that the House majority leaders knew ex ante that the Senate would eventually reject or the president would veto.
57
Pivotal Interval Movement
Cox and McCubbins’s argument above suggests that a congressional chamber makes a rational expectation about the other chamber’s behavior. However, their implication raises the question as to which chamber’s majority party maintains the final, that is, decisive, agenda-blocking power. Similarly, for the pivotal interval movement model, a question is likely to arise as to which Senate or House veto pivot would assume a determinative pivotal role. Figure 3.17 shows hypothetical ideal points of bicameral pivotal players on a finite unidimensional continuum. In this situation, the liberal president, p, and the veto pivot in the House, Vh, are likely to oppose legislation that would move the status quo, q, to q’, but the veto pivot in the Senate, Vs, would support the motion. Given that a successful veto override requires the support of override in both the chambers of Congress, a presidential veto of the bill would be sustained even with the Senate override of the veto (and the House support of the veto). Thus, between the veto pivots in the two chambers, the one with a more extreme ideal point assumes the pivotal role. The other veto pivot with a relatively moderate preference would not exert pivotal power.15 Therefore, in this situation, the pivotal gridlock interval extends between the House veto pivot and Senate filibuster pivot, as Figure 3.18 shows. Corollary 1: Pivotal Bicameral Gridlock V congress = V chamber 1, if | V chamber 1 |
>
| V chamber 2 |.
The cartel agenda model is likely to maintain similar assumptions. Figure 3.19 illustrates ideal points of the Senate and House medians (Ms and Mh) and their respective reflection points through the preferences of the majority of the majority parties (Mj s and Mj h). In this situation, the majority of the majority party in the House is likely to support a measure to move the status
p
Vh q
Vs q’
F
Figure 3.17 Bicameral Pivotal Model
p
Vs
F
Senate Pivotal Gridlock Interval
Vh
Figure 3.18 Bicameral Pivotal Gridlock Interval
House
58
The Other Side of Gridlock Mj s q1
Mj h q1’
Ms q2’
Mh q2
Figure 3.19 Bicameral Cartel Model
quo q1 to q1’, but the majority of the majority party in the Senate will likely oppose the bill. Also, the majority of the Senate is likely to support a bill to move the status quo q2 to q2’, but the majority of the House will likely oppose the bill. Given that an enactment of legislation requires the passage of a bill in both chambers of Congress, neither the legislation to move the status quo q1 to q1’, nor the bill to move the status quo q2 to q2’, will be passed by Congress. Thus, among the cameral medians in the two chambers and their reflection points through the preferences of the majority of the majority parties, the two diametrically extreme points constitute a cartel gridlock interval. Consequently, in Figure 3.20, the cartel gridlock interval exists between the reflection point of the Senate median and the ideal point of the House median. Corollary 2: Cartel Bicameral Gridlock Cartel Bicameral Gridlock Interval = [C1, C4], if (C1 < C2 < C3 < C4) where 1 ( Ci) = {Mj s, Mj h, Ms, Mh}.
Conclusions
This chapter has modeled a theory of how policy stability is enhanced in lawmaking. The pivotal interval movement model is constructed based upon a few discriminating assumptions. First, the model assumes no partisanship. The model postulates that legislators engage in individualistic voting based upon the policy proximity among the ideal points of their preferences, the status quo, and legislation. Legislators are assumed to be rational decision makers in that they seek a maximum utility, which is equivalent to the propinquity between the ideal points of the members and the policy output. Secondly, the model Mj s
Ms
Senate Cartel Gridlock Interval
Mj h
Mh
Figure 3.20 Bicameral Cartel Gridlock Interval
House
Pivotal Interval Movement
59
of pivotal interval movement model is contingent upon the supermajoritarian mechanisms in Congress. The model assumes that one of two minority coalitions could block legislation. Because of the senators’ prerogative of extended debate subject to a cloture, n–59 members of the Senate could impede unfavorable measures. Also, the presidential veto power subject to congressional override enables n/3+1 members in a chamber to block legislation. Among the core results of the pivotal interval movement model, is the indication that the new policy output converges within the new gridlock interval between the 41st percentile and 67th percentile members on the policy continuum. As the exogenous shocks move the gridlock interval, the status quo of policies in the portion of the old gridlock interval that the new gridlock interval does not overlap, move into the new gridlock interval. Unequivocally, some scholars would be perturbed by a lawmaking theory with no partisan effect at all. Consequently, this chapter has analyzed the cartel agenda model by Cox and McCubbins (2002, 2005) and developed its own gridlock interval movement model. In sharp contrast to the pivotal interval movement model, the cartel agenda model assumes majoritarian rules and strong party discipline. The model postulates that party members are united for their collective benefit, and that the senior partners of the majority party will block legislation the majority of its party membership dislikes (Cox and McCubbins 2002, 2005). Accordingly, this chapter has constructed the model of cartel gridlock interval movement, and explained that the movement of the interval between the chamber median and its reflection point through the majority party median could influence policy stability. In order to test the two contrasting interval movement models, the next chapter tests the influence of the residuum, which is the portion of the previous gridlock interval not overlapped by the new gridlock interval, on policy change.
This page intentionally left blank.
Chapter 4
Empirical Test
The kaleidoscope of legislative activities in the U.S. Congress beguiles scholars, who observe varied patterns in congressional lawmaking and their policy consequences. For decades, the party government school has maintained that party control of the government has an impact on governmental productivity. Scholars contend that the fierce partisan rivalry between the executive and legislative branches in divided government causes gridlock, which is an otherwise opaque term, and reduces efficacy in government (Sundquist 1980, 1988, 1992; Cutler 1988; Fiorina 1996, 2003). Even after Mayhew’s seemingly effectual repudiation of the party government theory (1991, 2005) and the subsequent emergence of the supermajoritarian models in the 1990s (Krehbiel 1996, 1998; Brady and Volden 1998, 2006), a great number of scholars today still contend that party control of the government substantially influences governmental effectiveness, specifically legislative productivity (Kelly 1993; Edwards, Barrett, and Peake 1997; Binder 1999, 2003; Coleman 1999; Howell, Adler, Cameron, and Riemann 2000; Edwards and Barrett 2000). In its inverse term, the hypothesis of divided government–gridlock nexus suggests that unified government results in higher governmental productivity than divided government. Nevertheless, analysts and pundits are not oblivious to examples of governmental, and perhaps policy, stalemates in prior unified governments. In the first Congress of the Clinton presidency, with
61
62
The Other Side of Gridlock
Democratic majorities in both the chambers, Clinton’s economic stimulus package, campaign finance reform, clean water, and other significant bills were killed by Republican filibusters. More recently, after the GOP triumph in the 2004 election, which resulted in the unified Republican government, President Bush declared that he would use his allegedly enhanced political capital to pursue his policy agendas. Nevertheless, Bush’s main legislative initiatives, such as the partial privatization of Social Security, guest worker program, antiterrorism surveillance, and attempts to make permanent the tax cuts of 2001 and 2003, fell short amid staunch opposition in Congress, and the 109th Congress was eventually dubbed as a “do nothing Congress.” The implication is that a factor other than party in the government might account for legislative gridlock. The previous chapter developed the applied model of the pivotal interval movement based upon the Brady-Krehbiel-Volden school’s supermajoritarian, nonpartisan theory of lawmaking. Nonetheless, the explanandum in the model does not focalize on legislative productivity. As argued in chapter 2, the ratio or number of enacted bills do not reflect the significance of legislative output in the policy context. If several significant policies are packed in a few omnibus bills, the modest number of omnibus measures underestimates the significance of their policy impact. Accordingly, the pivotal interval movement model aims to exposit policy change, or inversely policy stability, in legislative output. The previous chapter explained that a status quo interval for a given Congress is a gridlock interval of the previous Congress. The model introduced a concept, residuum, that was a portion of the gridlock interval of the previous Congress not overlapped by the gridlock interval of the new Congress. The indication is that when the residuum is wide, the policy output of legislation is more likely to change. On the other hand, Cox and McCubbins (2002, 2005) recently presented an innovative formal model of partisan agenda setting in Congress. In contrast to the nonpartisan basis of the pivotal interval movement model, the cartel agenda model explains that the majority party successfully blocks measures it dislikes. In order to make comparison with the pivotal interval movement model, the previous chapter constructed the cartel interval movement model predicated upon the partisan assumptions of the cartel agenda model. In this chapter, empirical models are examined to analyze the influence of the width of residuum of pivotal and cartel gridlock intervals, separately, on policy changes measured by the ADA scores in twenty-seven Congresses from 1953 to 2006. Such a comparative test is vital in light of the sharply contrasting presumptions and predictions by the two applied models.
Empirical Test
63
Hypotheses and Method
The two contrasting hypotheses below are tested in this chapter. И 1: When the residuum, q (p), in pivotal gridock interval, Z(p), is wider, more change in policy is likely to occur, where q (p) = Z(p) t–1 \ Z(p) t. И 2: When the residuum, q (c), in cartel gridock interval, Z(c), is wider, more change in policy is likely to occur, where q (c) = Z(c) t–1 \ Z (c) t.
In order to test the hypotheses, this chapter conducts empirical analysis to examine the influence of the width of residuum on policy change. The empirical models cover the 83rd to 109th Congresses (twenty-seven biennia). The dependent variables are focalized on the weighted and nonweighted policy changes calculated from the important legislation that the Americans for Democratic Action (ADA) selected. Apropos of the independent variables, in addition to the various measurements of residuum, the variables of divided government, budgetary situation, public liberal mood, and start of the presidential term are included. Dependent Variables Policy Change
As detailed in chapter 2, the change in policy output by Congress is calculated based upon the ADA scores. The Americans for Democratic Action1 selects several (usually forty per chamber) bills it considers most important during a Congress, and takes a position in opposition or in support of the measure. The ADA scores each legislator from zero to one hundred, depending upon the percentage of his or her votes in accordance with the ADA’s preference. For instance, if a legislator voted nay against a bill that the ADA opposed, or voted for a measure that the ADA supported, she would be in accordance with the ADA’s position. Should she vote otherwise, her votes are at odds with the ADA’s preference. In a similar method to the ADA’s calculation, passed legislation that the ADA opposed is coded as “1” and passed legislation that the ADA supported is coded as “–1.” Subsequently, the mean value of the examined legislation during a Congress is calculated for each chamber separately. Thus, the measurement indicates the overall conservatism of the passed bills among the important legislation selected by the ADA. The policy change is measured by an absolute
64
The Other Side of Gridlock
value of the difference between the points of the ADA policy output in an examined Congress and in the previous Congress: Policy change = | Policy output t – Policy output t–1 |.
Weighted Policy Change
It is important to note that the discussed operationalization above measures the policy change in the realm of passed legislation. As for the measurement of weighted policy change, failed bills are interpreted as causing the lack of policy change, hence increasing policy stability. In the measurement, the percentage of passed bills among the legislation that the ADA has selected for a chamber is calculated. Then, the policy change, which is measured by the method explained above, is multiplied by the percentage of the passed legislation. Thus, the variable of the weighted policy change indicates a policy movement based upon all the legislation, including failed measures, that the ADA selects. In contrast, the prior variable of the policy change indicates a policy change in passed legislation among the bills that the ADA chooses. Nonetheless, as shown in chapter 2, the two measurements of policy change are numerically highly related. The two variables are related with the values of r, .97, for the Senate and .95 for the House. For both policy change and weighted policy change in the House and Senate, the augmented Dickey-Fuller test and the Phillips-Perron test reject the hypothesis of unit root. Independent Variables Pivotal Bicameral Residuum
For the ideal points of the veto and filibuster pivots under a Republican presidency, I measure the ideal points of the 67th and 41st2 percentile legislators, respectively, based on Poole’s (2003, 2005) Optimal Classification scores. Optimal Classification (OC) is a non-parametric scaling method with an assumption of legislators with single-peaked preferences in Euclidean space. The OC scale, which ranges from –1 to 1, produces ideal points and cutting planes for roll call votes that maximize the number of correctly classified voting decisions (Poole 2005, 46–86). In contrast to the DW-NOMINATE scores, the OC scores are comparable across the chambers (Poole 2003, 2005).3 As for Congresses under a Democratic presidency, the ideal points of the 34th and 60th percentile legislators are measured for the veto and filibuster pivots, respectively. As detailed in the previous chapter, in order to measure the position and the width of residuum, the bicameral gridlock interval between the filibuster
65
Empirical Test
pivot and one of the two veto pivots, which is ideologically more extreme, is identified. Z (p) = [f, max(vh, vs)].
The pivotal bicameral gridlock intervals in the examined Congresses are plotted in Figure 4.1. As the figure shows, the overall movement of the interval is moderate. The only exception is the 97th Congress from 1981 to 1983, which replaced the four-year long unified Democratic government with a divided government consisting of a new Republican majority in the Senate and a Republican president. For the width of residuum, the portion of the gridlock interval of the previous Congress that the gridlock interval of the examined Congress does not overlap is computed. q (p) = Z (p) t–1 \ Z (p) t.
Congress
The moderate movements of the pivotal gridlock interval portend small widths of the residuum. The widths of residuum in the examined Congresses are reported and analyzed afterward. 82 – 83 – 84 – 85 – 86 – 87 – 88 – 89 – 90 – 91 – 92 – 93 – 94 – 95 – 96 – 97 – 98 – 99 – 100 – 101 – 102 – 103 – 104 – 105 – 106 – 107 – 108 – 109 – –1.00
–0.50
0.00
0.50
OC Figure 4.1 Pivotal Bicameral Gridlock Intervals
1.00
66
The Other Side of Gridlock
Pivotal Unicameral Residuum
As a comparative measure, I also identify the positions of the residuum of the unicameral gridlock intervals in the House and Senate, separately. I measure Senate gridlock interval between the filibuster and Senate veto pivots (Figure 4.2), and a House gridlock interval4 between the filibuster and House veto pivots (Figure 4.3). Z(ps) = [f, vs]. Z(ph) = [f, vh].
Cartel Bicameral Residuum
The cartel gridlock interval extends from the ideal point of the chamber median to its reflection point through the ideal point of the majority party median. For the ideal points of the chamber median, M, and the majority party median, J, the ideal points of the fiftieth percentile legislators in the chamber and within the majority party, respectively, are measured. For the reflection point of M through the majority party median point, Mj, I compute the ideal point of the legislator at 2J–M.5
Congress
82 – 83 – 84 – 85 – 86 – 87 – 88 – 89 – 90 – 91 – 92 – 93 – 94 – 95 – 96 – 97 – 98 – 99 – 100 – 101 – 102 – 103 – 104 – 105 – 106 – 107 – 108 – 109 – –1.00
–0.50
0.00
0.50
OC Figure 4.2 Senate Pivotal Gridlock Intervals
1.00
67
Empirical Test
Congress
82 – 83 – 84 – 85 – 86 – 87 – 88 – 89 – 90 – 91 – 92 – 93 – 94 – 95 – 96 – 97 – 98 – 99 – 100 – 101 – 102 – 103 – 104 – 105 – 106 – 107 – 108 – 109 – –1.00
–0.50
0.00
0.50
1.00
OC Figure 4.3 House Pivotal Gridlock Intervals
In order to identify the location of the residuum in bicameral gridlock intervals, I measure an interval between the most liberal and most conservative points among the House and Senate medians and their reflection points. Z(c) = [C1, C4], where (C1 < C2 < C3 < C4), and 1(Ci) = {Mj s, Mj h, Ms, Mh}.
The cartel bicameral residuum is the portion of the cartel gridlock interval of the previous Congress that the gridlock interval of the examined Congress does not overlap (see Figure 4.7). 1(c) = Z(c) t–1 \ Z(c) t.
As shown in Figure 4.4, the cartel bicameral intervals exhibit some recognizable changes. In the figure, the intervals in the 83rd, 84th, 97th, 100th, and 104th Congresses demonstrate sizable movements. As the leitmotif of the cartel agenda model implicates, these moves coincide with the changes in the majority party status. Whereas the Republican majorities replaced the
68
The Other Side of Gridlock
Congress
82 – 83 – 84 – 85 – 86 – 87 – 88 – 89 – 90 – 91 – 92 – 93 – 94 – 95 – 96 – 97 – 98 – 99 – 100 – 101 – 102 – 103 – 104 – 105 – 106 – 107 – 108 – 109 – –1.00
–0.50
0.00
0.50
1.00
OC Figure 4.4 Bicameral Cartel Gridlock Intervals
Democratic majorities in both chambers in the 83rd Congress, Democrats became the majority again in the two chambers in the 84th Congress. Also, the Republican Party replaced the Democratic Party as the majority party in the 97th Senate (1981–83), but the Democrats gained the majority again in the 100th Senate (1987–1990). In the 104th Congress, after the historic GOP victory in the 1994 midterm election, the Republicans gained majority status in both chambers. Cartel Unicameral Residuum
I also measure the widths of the residuum of the unicameral cartel gridlock intervals in the House and Senate, separately. I compute a Senate gridlock interval between the Senate median point and its reflection point through the Senate majority party median (Figure 4.5). The House unicameral interval is measured based on the interval between the House median point and its reflection point through the House majority party median (Figure 4.6). Z (cs) = [Ms, Mj s]. Z (ch) = [Mh, Mj h].
Congress
82 – 83 – 84 – 85 – 86 – 87 – 88 – 89 – 90 – 91 – 92 – 93 – 94 – 95 – 96 – 97 – 98 – 99 – 100 – 101 – 102 – 103 – 104 – 105 – 106 – 107 – 108 – 109 – –1.00
–0.50
0.00
0.50
1.00
OC Figure 4.5 Senate Cartel Gridlock Intervals
Congress
82 – 83 – 84 – 85 – 86 – 87 – 88 – 89 – 90 – 91 – 92 – 93 – 94 – 95 – 96 – 97 – 98 – 99 – 100 – 101 – 102 – 103 – 104 – 105 – 106 – 107 – 108 – 109 – –1.00
–0.50
0.00
0.50
OC Figure 4.6 House Cartel Gridlock Intervals
69
1.00
70
The Other Side of Gridlock
Divided Government
For the divided government variable, Congresses in divided government and unified government are coded as 1 and 0, respectively. Of the 83rd to the 109th Congresses, the ten Congresses in unified government include the 83rd, 87th, 88th, 89th, 90th, 95th, 96th, 103rd, 108th, and 109th Congresses.6 As auxiliary measures, I also code whether the president’s opposition party holds a majority (= 1) or not (= 0) in the Senate and House, separately. Variations in Pivotal Residuum and Cartel Residuum
The widths of the residuum of the pivotal bicameral intervals and the cartel bicameral intervals are plotted in Figure 4.7. The figure displays three parameters. First, the residuum of the pivotal interval are relatively narrower than the cartel interval residuum. Second, the widths of the residuum of the pivotal bicameral interval maintain lower variation than the residuum of the cartel bicameral interval. Third, the widths of residuum of the pivotal and cartel gridlock intervals are unrelated. As the modest swing of the pivotal bicameral gridlock interval and the sizable movement of the cartel counterpart in Figure 4.1 and Figure 4.4 have already implied, Figure 4.7 shows the wider residuum of the cartel intervals than the residuum of the pivotal intervals. In light of the contrasting assump-
Unified Gov = 0.9
Divided Gov = 0.8
Cartel Bicameral Residuum
Pivotal Bicameral Residuum
Figure 4.7 Pivotal and Cartel Bicameral Residuum
71
Empirical Test
tions of the pivotal and cartel interval movement models, this asymmetry is not surprising. The previous chapter has illustrated that the ideal point of the median member of the majority party centers on the cartel gridlock interval. Accordingly, the change of the majority party drastically alters the location of the cartel interval. In contrast, the pivotal gridlock interval lies between the veto and filibuster pivots, which are only moderately influenced by the change in the majority party status. Therefore, the cartel gridlock interval moves across a larger distance than the pivotal gridlock interval does, hence resulting in the wider residuum and the higher variation in the widths of residuum. In fact, Table 4.1 reports higher standard deviations of the various cartel intervals than those of the pivotal intervals. The disparity in the standard deviations provides evidences of more recurrent changes in the width of the cartel interval residuum than the pivotal interval residuum. Figure 4.7 illustrates that except for the 104th Congress, the bulges in the residuum of the pivotal and cartel intervals are not parallel. More specifically, Table 4.2 recounts that the two residuum are unrelated to each other, and they are unrelated to the variable of divided government. Turning back to Figure 4.7, the residuum of the pivotal bicameral interval widens in the 83rd, 97th, and 104th Congresses while the cartel equivalent is enlarged in the 84th, 100th, 104th, and 108th Congresses. Whereas the latter Congresses altered the majority party, the 83rd (Eisenhower) and 97th (Carter) Congresses experienced the change
Table 4.1 Residuum Descriptive Statistics N
Minimum
Maximum
Mean
Std. Deviation
Pivotal Bicameral Residuum
27
0
0.32
0.063
0.088
Pivotal Senate Residuum
27
0
0.2
0.033
0.048
Pivotal House Residuum
27
0
0.032
0.064
0.076
Cartel Bicameral Residuum
27
0
0.75
0.176
0.237
Cartel Senate Residuum
27
0
0.65
0.214
0.244
Cartel House Residuum
27
0
0.46
0.094
0.142
Table 4.2 Residuum Correlations Pivotal Bicameral Residuum
Cartel Bicameral Residuum
Pivotal Bicameral Residuum
1
Cartel Bicameral Residuum
0.151
1
–0.106
0.038
Divided Government Entries are Pearson Correlations
Divided Government
1
72
The Other Side of Gridlock
in the president’s party. Since the change in the presidential party reverses the polarity of the veto and filibuster pivots, it is likely to move the pivotal gridlock interval and widen its residuum. Others
As for other independent variables, similar to the existing literature (Binder 1999, 2002; Krehbiel 1998; Coleman 1999; Chiou and Rosenberg 2003), I examine the influences of public mood, budgetary situation, and start of term. For the public mood variable, I use the lagged value of Stimson’s (1991) public liberalism index.7 For the budgetary situation, I measure the budget deficit as a percentage of federal outlays. For the start of term variable, I code 1 for the first two years of the presidential terms and 0 otherwise. Findings
This section first presents the results for the non-weighted policy change, and later reports the findings for the weighted policy change. Table 4.3 reports the estimates of OLS models in the seven specifications for the non-weighted policy change in the Senate. Among the three bivariate models in equations 1 Table 4.3 Senate Policy Change and Bicameral Residuum 1 Divided Government
2
3
0.12 (.12)
Pivotal Bicameral Residuum
1.69* (.63)
Cartel Bicameral Residuum
0.05 (.27)
Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
5
0.31** (.10) –0.003 0.92 1.8 27
6
7
0.01 (.30)
0.16 (.12) 2.05** (.72) –0.07 (.26)
1.89* (.70)
Budget
Constant
4 0.11 (.14)
0.27** (.07) 0.20 7.16 1.53 27
0.37 (.08) –0.04 0.03 1.8 27
0.02 (.01) –0.01 (.01) –0.02 (.13)
0.01 (.01) 0.01 (.01) –0.05 (.12)
0.01 (.01) –0.01 (.01) –0.03 (.14)
0.01 (.01) 0.01 (.01) –0.03 (.13)
0.54 (.89)
–0.07 (.82)
0.72 (.92)
–0.35 (.88)
–0.13 0.26 1.82 27
0.13 1.94 1.56^ 27
–0.16 0.1 1.83 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
0.12 1.59 1.61^ 27
73
Empirical Test
through 3, divided government and cartel bicameral residuum are insignificant. In contrast, pivotal bicameral residuum in equation 2 is significant. Again, this variable indicates the width of residuum in the gridlock interval between the filibuster pivot and one of the two veto pivots with a more extreme ideal point. The positive coefficient of the variable suggests that when the residuum in the pivotal gridlock interval is wide, the Senate’s policy output is more likely to change. The adjusted R2 in equation 2 suggests that the width of residuum explains 20 percent of the variation in Senate policy change. For equations 4 through 6 with the control variables, the results look similar. While none of the control variables are significant, only pivotal bicameral residuum in equation 5 remains significant. In equation 7, which includes all of the variables, pivotal residuum is robust whereas other variables remain insignificant. In equations 5 and 7, the Durbin-Watson d does not entirely reject the hypothesis of serial correlation at the 0.05 level of significance. Nonetheless, the Durbin-Watson d is insignificant at the 0.01 level of significance in the same equations. Next, the residuum and divided government variables, in a bicameral context, are replaced with equivalent variables with a unicameral assumption. Table 4.4 shows the results of the equations for the Senate policy change. In contrast to the significance of the variable of the pivotal bicameral residuum, Table 4.4 Senate Policy Change and Unicameral Residuum 1 Opposition Party Majority
2
3
0.17 (.12)
Pivotal Senate Residuum
1.01 (1.29)
Cartel Senate Residuum
0.15 (.26)
Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
5
0.29** (.09) 0.04 1.97 1.9 27
6
7
0.11 (.29)
0.18 (.15) 1.34 (1.51) 0.01 (.29)
1.31 (1.48)
Budget
Constant
4 0.18 (.14)
0.35** (.07) –0.01 0.62 1.76 27
0.35** (.08) –0.03 0.35 1.7 27
0.01 (–.01) –0.01 (.01) –0.02 (.13)
0.01 (.01) –0.01 (.01) –0.04 (.13)
0.01 (.01) –0.01 (.01) –0.02 (.13)
0.01 (.01) 0.01 (.01) 0.01 (.13)
0.68 (.85)
0.48 (.90)
0.61 (.92)
0.41 (.95)
–0.08 0.5 1.89 27
–0.12 0.3 1.8 27
–0.15 0.14 1.74 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
–0.15 0.45 1.85^ 27
74
The Other Side of Gridlock
pivotal unicameral residuum remains insignificant throughout the bivariate and multivariate models. Also, other variables remain insignificant. Turning to policy change in the House, Table 4.5 reports the results of the empirical models for change in ADA scores. Among the three bivariate models, divided government is insignificant. However, pivotal bicameral residuum is an important predictor of policy change. Again, its positive coefficient suggests that when the residuum in the pivotal bicameral interval is wide, policy output in the House is more likely to change. As the adjusted R2 in equation 2 shows, the width of the residuum in the pivotal bicameral interval explains 14 percent of the variation in the House policy change. In equation 3, cartel bicameral residuum appears significant with a positive coefficient. This indicates that when the residuum in cartel bicameral interval is wide, the House policy output will change sizably. For the three multivariate models with the control variables, pivotal bicameral residuum remains significant in equation 5. In contrast, the significance of the variable of the cartel residuum disappears in equation 6. In the final equation including all variables, cartel bicameral residuum remains insignificant and pivotal residuum is robust. Table 4.6 shows the results of the models for the House policy change and the unicameral residuum. In contrast to the significant impact of pivotal
Table 4.5 House Policy Change and Bicameral Residuum 1 Divided Government
2
3
0.12 (.14)
Pivotal Bicameral Residuum
1.67* (.74)
Cartel Bicameral Residuum
6
0.57* (.28)
Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
5
1.67* (.78)
Budget
Constant
4 0.07 (.15)
0.35** (.11) –0.01 0.77 2.09 27
0.33** (.07)
0.33** (.08)
0.14 5.11 1.96 27
0.11 4.29 2.15 27
0.46 (.30)
7 0.12 (.13) 1.65** (.78) 0.40 (.29)
0.01 (.01) –0.01 (.01) –0.15 (.14)
0.01 (.01) –0.01 (.01) –0.18 (.12)
0.01 (.01) –0.01 (.01) –0.09 (.14)
0.01 (.01) –0.01 (.01) –0.11 (.13)
1.57 (.96)
0.99 (.91)
1.3 (.92)
0.46 (.96)
0.16 2.21 1.97 27
0.08 1.55 2.3 27
0.18 1.97 2.04 27
–0.01 0.94 2.22 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
75
Empirical Test
bicameral residuum in the previous models, pivotal unicameral residuum is insignificant in bivariate and multivariate equations. Other variables also remain insignificant in all equations. Turning to the findings in the models of weighted policy change, Table 4.7 reports the results of the models for the Senate. In equations 1 through 3 for the bivariate models, only pivotal bicameral residuum in equation 2 is significant. The positive coefficient of the variable indicates that the wider residuum in the pivotal bicameral interval is echoed by a larger policy change in the Senate. The adjusted R2 in equation 2 suggests that the width of residuum explains 17 percent of the variation in the Senate policy change. In contrast, divided government in equation 1 and cartel bicameral residuum in equation 3 are insignificant. Equations 4 through 6, including the control variables, demonstrate the results parallel to the results of bivariate models. While none of the control variables is significant, pivotal bicameral residuum in equation 5 is significant. In equation 7, the same variable remains robust. In contrast, cartel bicameral residuum is insignificant in equation 7, as well as in other specifications. In equation 7, divided government shows a moderate effect, albeit in the wrong direction. The positive coefficient for the variable paradoxically implies that divided government increases, rather than decreases, policy change in
Table 4.6 House Policy Change and Unicameral Residuum 1 Opposition Party Majority
2
3
0.06 (.14)
Pivotal House Residuum
1.01 (.92)
Cartel House Residuum
0.85 (.47)
Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
5
0.39** (.11) –0.03 0.17 2.09 27
6
7
0.58 (.55)
0.06 (.15) 1.03 (1.13) 0.45 (.57)
1.04 (1.01)
Budget
Constant
4 0.01 (.14)
0.37** (.09)
0.35** (.08)
0.01 1.21 2.15 27
0.08 3.24 2.14 27
0.01 (.01) –0.01 (.01) –0.16 (.14)
0.01 (.01) –0.01 (.01) –0.19 (.14)
0.01 (.01) –0.01 (.01) –0.09 (.15)
0.01 (.01) –0.01 (.01) –0.12 (.16)
1.67 (.96)
1.27 (.99)
1.32 (.97)
0.88 (1.13)
0.03 1.18 2.19 27
0.03 1.19 2.25 27
–0.03 0.89 2.18 27
–0.02 0.87 2.22 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
76
The Other Side of Gridlock Table 4.7 Weighted Senate Policy Change and Bicameral Residuum 1 Divided Government
2
3
0.09 (.06)
Pivotal Bicameral Residuum
0.76* (.30)
Cartel Bicameral Residuum
0.02 (.12)
Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
5
6
7
0.01 (.14)
0.11^ (.06) 0.92* (0.33) –0.03 (.12)
0.81* (.34)
Budget
Constant
4 0.08 (.06)
0.12* (.05)
0.13** (.03)
0.05 2.29 1.83 27
0.17 6.36 1.53 27
0.17** (.04) –0.04 0.04 1.79 27
0.01 (.01) –0.01 (.01) 0.01 (.06)
0.01 (.01) 0.01 (.01) –0.01 (.06)
0.01 (.01) –0.01 (.01) –0.01 (.06)
0.01 (.01) 0.01 (.01) –0.01 (.06)
0.27 (.41)
0.06 (.39)
0.40 (.43)
–0.14 (.40)
–0.07 0.57 1.85 27
0.09 1.64 1.52^ 27
–0.16 0.12 1.8 27
0.15 1.77 1.64^ 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
the Senate ceteris paribus. Nonetheless, the variable of divided government remains insignificant in other equations, and this modest finding entails a reservation concerning its robustness. In equations 5 through 7, the Durbin-Watson d shows an imperfect rejection of the hypothesis of serial correlation. However, the Durbin-Watson d is insignificant at the 0.01 level of significance. Next, the variables of the residuum and divided government with a bicameral assumption are replaced with the variables of unicameral residuum. Table 4.8 shows the results of equations for the weighted Senate policy change. Parallel to the models of non-weighted policy change in the Senate, the variable of pivotal unicameral residuum, as well as other variables, remains insignificant throughout the bivariate and multiple models. Apropos of weighted House policy change, reported in Table 4.9, pivotal bicameral residuum in equation 2 is a significant predictor of policy change and possesses a positive coefficient. The adjusted R2 in equation 2 is 0.10. The variable of pivotal bicameral residuum remains robust in equations 5 and 7 with the control variables. In contrast, divided government and cartel bicameral residuum remain insignificant in all equations. Table 4.10 shows the estimates for the weighted policy change in the House and the unicameral residuum. In contrast to the significance of the pivotal
Table 4.8 Weighted Senate Policy Change and Unicameral Residuum 1 Opposition Party Majority
2
3
0.1^ (.06)
Pivotal Senate Residuum
0.58 (.59) 0.12 (.12)
Budget Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
5
0.12** (.04) 0.08 3.19 1.95 27
6
7
0.1 (.13)
0.11 (.07) 0.68 (.68) 0.1 (.13)
0.69 (.69)
Cartel Senate Residuum
Constant
4 0.11 (.07)
0.16** (.03) –0.01 0.93 1.75 27
0.15 (.04) –0.01 0.94 1.64 27
–0.01 (.01) –0.01 (.01) 0.02 (.06)
0.01 (.01) –0.01 (.01) –0.02 (.06)
0.01 (.01) –0.01 (.01) –0.01 (.06)
–0.01 (.01) –0.01 (.01) 0.01 (.06)
0.37 (.39)
0.28 (.42)
0.3 (.13)
0.2 (.43)
–0.11 0.38 1.774 27
–0.13 0.26 1.66^ 27
–0.06 0.77 1.88^ 27
–0.02 0.87 1.94 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
Table 4.9 Weighted House Policy Change and Bicameral Residuum 1 Divided Government
2
3
0.06 (.08)
4
(.42)
Budget Public Mood Start of Term
Adjusted R2 F Durbin Watson d N
–0.02 0.62 2.18 27
0.19 (.17)
0.06 (.08) 0.85^ (.45) 0.16 (.17)
(.44) 0.26 (.16)
0.19** (.06)
7
0.85^
Cartel Bicameral Residuum
Constant
6
0.03 (.08) 0.82^
Pivotal Bicameral Residuum
5
0.17** (.04)
0.18** (.05)
0.10 3.9 2.04 27
0.06 2.75 2.24 27
0.01 (.01) –0.01 (.01) –0.1 (.08)
0.01 (.01) –0.01 (.01) –0.1 (.07)
0.01 (.01) –0.01 (.01) –0.07 (.08)
0.01 (.01) –0.01 (.01) –0.08 (.07)
0.73 (.53)
0.43 (.51)
0.62 (.52)
0.19 (.55)
0.12 1.86 1.98 27
0.02 1.14 2.31 27
0.1 1.45 2 27
–0.03 0.83 2.24 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
77
78
The Other Side of Gridlock
bicameral residuum variable in the previous models, pivotal unicameral residuum is insignificant in bivariate and multivariate equations. Equation 2 shows a modest impact of the cartel unicameral residuum. However, the same variable becomes insignificant in multivariate equations. Overall, the results of the models of nominal policy change are very similar to the results of the models of weighted policy change. The findings attest to the substantive impact of the pivotal residuum, but not the cartel residuum, both on weighted and non-weighted policy change. However, the discriminating finding is that only the variable of the pivotal bicameral residuum is robust whereas the residuum variables based on the unicameral gridlock intervals remain insignificant. Intuitively, the influence of the pivotal bicameral residuum could be stronger for Senate policy change than House policy change because the exercise of filibuster is only possible in the Senate. Nonetheless, such postulation appears to be unsupported. In bivariate terms, the variable of the pivotal bicameral residuum in the models of the Senate and House policy change results in the adjusted R2 of .20 and .14, respectively. As for the standardized coefficients, the same variable for the Senate and House policy change possess .47 and .41, respectively. Thus, there is only a marginal difference in the influence of the pivotal bicameral residuum between the models of the House and Senate policy change.
Table 4.10 Weighted House Policy Change and Unicameral Residuum 1 Opposition Party Majority
2
3
0.07 (.09)
Pivotal Bicameral Residuum
0.59 (.50) 0.46^ (.26)
Public Mood Start of Term
0.21** (.06) –0.04 0.07 2.19 27
6
7
0.31 (.31)
0.07 (.08) 0.67 (.62) 0.24 (.32)
0.68 (.55)
Budget
Adjusted R2 F Durbin Watson d N
5
0.03 (.08)
Cartel Bicameral Residuum
Constant
4
0.19* (.05) –0.01 1.36 2.24 27
0.18 (.04) 0.08 3.13 2.11 27
0.01 (.01) –0.01 (.01) –0.1 (.08)
0.01 (.01) –0.01 (.01) –0.12 (.08)
0.01 (.01) –0.01 (.01) –0.07 (.08)
0.01 (.01) –0.01 (.01) –0.08 (.08)
0.8 (.53)
0.51 (.55)
0.59 (.54)
0.32 (.62)
0.03 1.21 2.2 27
0.01 1.08 2.19 27
–0.04 0.78 2.26 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
–0.03 0.89 2.1 27
Empirical Test
79
This implies that the possibility and threat of filibuster are effective across the chambers. The legislators’ decisions and cameral behaviors could be largely constrained by their rational expectations of the other chamber’s behaviors. Conclusions
Applied models present succinct explanations of causal mechanisms in real-world politics. Such models are productive for generating theoretical assumptions and testable hypotheses. The two models of lawmaking, the pivotal interval movement model and cartel interval movement model, offer contrasting expositions as to when a significant policy change occurs. Significantly, the two models are constructed based upon similar analytical primitives, such as status quo, a single-peak preference, exogenous shocks, and Euclidean policy space. Also, both models assume that policy output often could be repositioned at the preference of the chamber median voter. The essential difference between the two models lies in the assumptions of de facto procedures—specifically, the exertion of the negative power. The cartel interval movement model presumes a partisan block on agenda setting, with an assumption that the majority party leaders block the legislation that the majority of the majority party dislikes. In contrast, the pivotal interval movement model predicates that the coalitions of minority legislators who could secure filibuster and the presidential veto, separately, impede unfavorable measures. The empirical results in this chapter reveal that the width of the residuum of the pivotal, but not the cartel, interval has an impact on policy change. This suggests that two minority coalitions of legislators, one of leftists and the other of rightists, maintain a substantial ability to halt unfavorable bills. This finding stands in contrast to the assumption of the cartel interval movement model. The cartel interval movement model predicts that only one minority coalition, which consists of the majority members of the majority party, holds power to block legislation. Thus, the findings in this chapter implicate a more modest estimate of the institutional ability for policy change in American government than the cartel agenda model depicts. The cartel agenda model predicts relatively sizable policy changes because, overall, policy is presumed to move toward the preference of the majority party (Cox and McCubbins 2005, 65–76). Nonetheless, the empirical support of the pivotal interval movement model, in this chapter, unveils more restrictions on policy movement. The indication is that one of two minority coalitions could block the legislation, and thereby policy output moderately moves toward the preference of the chamber median if it moves at all.
80
The Other Side of Gridlock
Perhaps it is important to consider, albeit carefully, the normative implications as to whether or not the current institutional bias for policy stability in the American political system has a positive impact on governance. Such discussions will also bedevil scholars with the germane question as to whether or not various policy problems and solutions effectually change the configuration of the decision makers’ preferences through potent exogenous shocks.
Chapter 5
Veto Players Enacting coalitions are hard to assemble even if party control is unified. —David Mayhew
Applied models are designed to be empirically applicable to the real world. Constructed upon the assumptions provided by pure theoretical models, applied models offer testable explanations of concrete political phenomena. Chapter 3 presented the pivotal interval movement model, which is predicated upon the assumptions of supermajoritarian procedures and individualistic, nonpartisan voting. However, scholars emphasize the need for testing alternative models. Rebecca B. Morton contends: Researchers know that a given formal model is not the only possible one that can be used to represent the real-world situation under study. Hence an empirical model may be devised to compare the predictions of one formal model against alternative models in explaining real-world outcomes. (1999, 55)
Accordingly, chapter 3 constructed the applied model of cartel interval movement model based upon an alternative theory that assumes a solid partisanship and majoritarian procedure. Chapter 4 examined the influence of the widths of pivotal interval residuum and cartel interval residuum. The findings suggested that the pivotal interval residuum had significant impact on policy stability. Chapter 4 tested the cartel interval movement model as an alternative model to the pivotal interval movement model. However, the cartel interval 81
82
The Other Side of Gridlock
movement model was formulated to generate its unique gridlock interval and residuum in a symmetrical method to the pivotal interval movement model. Thus, the alternative model in the previous chapters was a model of alternative theory but with a similar application. In this chapter, an alternative application of the supermajoritarian thesis is devised and tested. Such a test of the alternative applied model is likely to enhance robustness in an empirical study. When an empirical test of an applied model reveals support for a theory, there is a danger that the results are due to the way a model is constructed. In order to minimize the potential bias from the application method of the interval movement model, this chapter tests the assumption of supermajoritarian, nonpartisan mechanism through a different applied model. Overall, scholars examine the empirical validity of applied models in order to test theories, rather than testing application methods. Veto Players Model
This chapter plots the preferences of pivotal legislators, or veto players, in the two-dimensional, Cartesian plane. The influence of the winset’s area, which is the intersection overlapped by the veto players’ indifference curves, on policy stability is examined. In the previous chapters, the interval movement model assumed players with single-peaked preferences on a unidimensional policy continuum. The pivotal interval movement model focused on two players who play pivotal roles because of the supermajoritarian procedures in Congress: the filibuster pivot and the veto pivot. The model assumed that when the status quo is present within the pivotal gridlock interval between the ideal points of filibuster and veto pivots, the status quo is unlikely to be changed. This indicates that “pivots” are largely parallel to what other scholars call veto players, without whose unanimous agreement the status quo cannot be changed (Tsebelis 2002, 19). In contrast to the unidimensional model, this chapter plots the preferences of veto players in a two-dimensional policy space. Much of the legislation in the period examined is pertinent to social agendas, and divisions within parties are more salient in the social issue dimension than in the economic issue dimension. While Poole acknowledges that the dimension of social issues has slowly declined in importance since the late 1960s, he observes that two dimensions are required to account for the great bulk of voting since World War II period (2003, 7). Civil rights issues, in particular, split the Democratic Party and created the three party system, making the social issue dimension significant (Poole and Rosenthal 1997, 48; Poole 2005, 14–15). Burns (1963) contends that the division within the Democratic Party has an adversarial impact on legisla-
83
Veto Players
tive productivity. Frymer (1994) notices that the conservative contingent in the Democratic Party complicates policymaking under a unified Democratic government. In contrast, Frymer observes, Reagan enjoyed a powerful and unified coalition with conservative Democrats in spite of divided government. Since the Democratic Party was divided on various social issues until the early 1980s, and this period accounts for one-half of the Congresses examined, the influence of the legislators’ preferences should be examined in the social, as well as the economic, dimension. Secondly, the two-dimensional space could encompass the pivotal legislators of different chambers. The previous chapters assumed that the one of two veto pivots who is ideologically more extreme maintains a pivotal role. Accordingly, the gridlock interval was measured based upon the ideal points of the ideologue veto pivot and filibuster pivot. In contrast, the Cartesian plane places the preferences of all the House and Senate pivots on the same space. Figure 5.1 shows veto players with single-peaked preferences in the Cartesian plane. Compatible with Poole’s works (1997, 2003, 2005), the model is based on a dimension of economic issues and a dimension of social issues. In Figure 5.1, the x-coordinate represents the level of economic conservatism of legislators (and policy) while the y-coordinate indicates the level of their social conservatism. The policy space is a finite, two-dimensional plane, ranging from –1 to +1. For instance, in the figure, a pivotal legislator P’s ideal point is at (0.25, –0.01), indicating that she is slightly conservative on economic issues and moderate on social issues. Also, in the figure, the ideal point of the status
Y
Social Conservatism
1
1 n 0 n
n
n
X
1
Economic Conservatism
Figure 5.1: Circular Indifference Curve of Pivotal Legislator
84
The Other Side of Gridlock
quo, Q, is at (–0.22, 0.20). Legislator P would prefer a new policy to the status quo, Q, if the new policy point is closer than Q to her ideal point. By plotting a circle, the so-called indifference curve, centered on her ideal point that passes through the status quo point (Q), it could be assumed that any new policy in the circle is closer than Q to her ideal point. The indication is that legislator P would prefer any new policy in her indifference curve to the status quo, and she prefers the status quo to new policy proposals outside her indifference curve. Thus, P is likely to support any proposed bills, including Q1, inside of her indifference curve, and she would reject legislation, including Q2, outside her indifference curve. The pivotal interval movement model in the previous chapters presumed that the filibuster and veto pivots maintain ability to reject legislation. Accordingly, the veto players model plots the ideal points of the Senate filibuster pivot and the Senate veto pivot as well as the House veto and the House median pivots within the Cartesian plane. For example, Figure 5.2 shows the position of the status quo (Q), and the ideal points and indifference curves of the Senate filibuster pivot (F), the Senate veto pivot (SV), the House veto pivot (HV), and the House median pivot (HM) in the 104th Congress. These filibuster, veto, and House median pivots could reject bills with the support of two-fifths, one-third, and a majority of legislators, respectively. These veto players would support bills within their respective indifference curves and reject legislation outside their indifference curves. Since all veto players must unanimously support a bill for passage to occur, only legislation in the intersection of all the four indifference circles, the Y
4( #/.'2%33
Social Conservatism
& (-
1 X n
n
(6
n
WINSET
36 n
Economic Conservatism
Figure 5.2 Pivotal Veto Players in the 104th Congress
Veto Players
85
so-called winset, are likely to pass. Inversely, bills outside the winset are likely to be rejected by one or more pivots. This suggests that when the area of winset is large, policy is more likely to change. Thus, this chapter tests the following hypothesis. Hypothesis: When the winset, q (v), is larger, more change in policy is likely to occur, where q (v) = a (hv) E a (hm) E a (sv) E a (f ), and a (i ) = indifference curve of i.
Data and Method Dependent Variable
This chapter focuses on policy stability in overall congressional legislative output. The ideal point of congressional legislative output is measured based on the calculation of the ADA legislation, which was explained in chapter 2. Passed legislation that the ADA opposed is coded as “1” and passed bills the ADA supported are coded as “–1.” I calculate the mean value of the measured legislation in Congress in order to examine the ideal point of legislative output. The legislative output ranges from –1 to 1, and the policy change is measured by an absolute value of the distance between the policy output in an examined Congress and in a previous Congress. As for weighted policy change, the value of nominal policy change is multiplied by the percentage of passed bills among all the legislation selected by the ADA. Winset
In order to identify the winset in every Congress, I measure the status quo point (Q) by the MWC scores (Poole and Rosenthal 1997, 59–70) of policy output of the previous Congress in Dimension 1 (Economic Policy Dimension) and Dimension 2 (Social Policy Dimension). The Mean Winning Coordinate (MWC) ranges from –1 to +1 based on policy conservatism, and covers a mean value of the DW-NOMINATE scores of winning outcomes on roll calls. For the ideal points of the veto, filibuster, and House median pivots under a Republican presidency, I measure the ideal positions of the sixty-seventh, forty-first,1 and fiftieth percentile legislators, respectively, based on Poole’s (2003, 2005) Optimal Classification scores2 in Dimension 1 and Dimension 2. As for Congresses under Democratic administration, I measure the positions of the thirty-fourth, sixtieth, and fiftieth percentile legislators. As I elaborated in the previous section, I identify the four pivots’ circular indifference curves that pass through the status quo. In order to calculate the area of winset, which
86
The Other Side of Gridlock
is an intersection overlapped by all four indifference circles, I plot the indifference curves by using the Auto CAD program3 (see Figure 5.3 for the plots of veto players). Other Independent Variables
Similar to the empirical models in chapter 4, I incorporate the following variables: public mood, budgetary situation, divided government, and start of term. For public mood, I use the lagged value of Stimson’s (1991, 2005) public liberalism index. As for the budgetary situation, I measure the budget deficit as a percentage of federal outlays. For divided government, Congresses in divided government and unified government are coded as 1 and 0, respectively. As for start of term, I coded 1 for the first two years of the presidential terms and 0 otherwise. Findings
This section examines the influence of the various determinants on policy change. The results for the models of the congressional policy change appear in Table 5.1. Among the bivariate models, divided government in equation 1 is not significant. In contrast, the variable of the winset in equation 2 is significant. Again, the winset indicates the overlapped area of the indifference curves of the veto players. The positive coefficient of the variable suggests that when the winset is wide, the congressional policy output is more likely to change. The adjusted R2 in equation 2 suggests that the area of winset explains 37 percent of the variation of the congressional policy change. For equations 3 through 5, including the control variables, the results look similar. While divided government is not significant in equations 3 and 5, the winset in equations 4 and 5 remains significant. Thus, the legislative output is more likely to be deviant from the status quo when the indifference curves of the pivotal legislators’ preferences largely overlap. When pivotal legislators share a preference for a revisionist policy, policy output is likely to be changed. This suggests that the legislators’ individualistic preferences, rather than partisanship, are more likely to influence the content of legislative output.4 In equation 5, which includes all of the variables, the winset is robust. The adjusted R2 suggests that the explanatory variables explain 40 percent of the variation in the congressional policy change. In equations 3 through 5, public mood is significant, but the coefficients are negative. Oddly, public support for an activist government weakens, rather than strengthens, policy change. This may imply that the activist mood of
87 RD #/.'2%33
Y
TH #/.'2%33
Y
1
36
2
(6
3
36
(6
4 5
X n
n
n
6
X
(-
n
7 8
(&
9
n
n
10
&
11 n TH #/.'2%33
Y
12
n
TH #/.'2%33
Y
13
36
(6
14
(6
15
36
16 17
(n
(-
X
n
n
18
X
n
19 20 21
n
n
22
&
&
23
n TH #/.'2%33
Y
24
n
TH #/.'2%33
Y
25
26 &
27
&
28 29
n
n
X
(-
n
(-
n
(6
30
X
31 32
(6
33 36
n
36
n
34 35
n
n
36 37
Figure 5.3 Veto Players’ Indifference Curves 83rd–109th Congresses
38
(continues)
39 40
87
1
TH #/.'2%33
Y
Y
TH #/.'2%33
2 3
&
4
&
5 6 7
(-
n
X
n
(-
n
X
n
8 (6
9
(6 n
n
10
36
11
36
12
n
n
13
ST #/.'2%33
Y
ND #/.'2%33
Y
14 15
(6
36
17 18 19
(6
16
36
(n
(-
X
n
n
X
n
20 21 n
22
n &
23
&
24
n
25
RD #/.'2%33
Y
n
TH #/.'2%33
Y
26 27
(6
36
29
36
30 31
(6
28
(n
n
X
X n
n
(-
32 &
33 34
n
n &
35 36
n
n
37 38
Figure 5.3 Veto Players’ Indifference Curves 83rd–109th Congresses
39
(continues)
40
88
89 Y
TH #/.'2%33
Y
TH #/.'2%33
1
2 3
4
&
5 (6
(-
n n
&
6
X
X
n
n
(-
7 8
(6
36
9 n
n
10
36
11 n Y
TH #/.'2%33
12
n
Y
TH #/.'2%33
13
14 15
16
(6
17
36 n
n
X n
(6 (&
18
X n
20
(-
36 &
n
19 21
n
22 23
Y
TH #/.'2%33
24
n
n
Y
TH #/.'2%33
25
26 27 (6
28
36
(6
29
36 X n
n
30
X n
n
(-
&
&
n
31 32 33
n
34 35
n
n
36 37
Figure 5.3 Veto Players’ Indifference Curves 83rd–109th Congresses
38
(continues)
39 40
89
1
Y
ST #/.'2%33
ND #/.'2%33
Y
2 3
(6
4 5 6 7
36
n
(-
(&
9
X
n
8
(6
36
n
X
n
&
n
10
n
11 12
n
13
Y
RD #/.'2%33
n
TH #/.'2%33
Y
14 15
16
&
17
(-
18 19
& n
X
n
20
X n
n
(-
21 22
(6
23
36
n (6
n
36
24
n
25
TH #/.'2%33
Y
n
TH #/.'2%33
Y
26 27
28
(6
(6
29 36
30 31
36 &
n
n
32
& X
X n
n
(-
(-
33 34
n
n
35 36
n
n
37 38
Figure 5.3 Veto Players’ Indifference Curves 83rd–109th Congresses
39
(continues)
40
90
TH #/.'2%33
Y
TH #/.'2%33
Y
1
2 3
4
& (6
& 36 n
n
(-
(-
X n
n
5
36
6
X
(6
7 8 9
n
n
10 11 n
n TH #/.'2%33
Y
12 13
14 15
16
& (6
17
36
18
X n
n
(-
19 20 21
n
22 23 n
24 25
Figure 5.3 Veto Players’ Indifference Curves 83rd–109th Congresses
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
91
92
The Other Side of Gridlock Table 5.1 Congressional Policy Change 1
Divided Government
2
0.04 (0.10)
Winset
Public Mood Start of Term
Adjusted R2 F Durbin Watson N
4
5
8.55** (2.02)
–0.04 (0.08) 8.73** (2.09)
–0.01 (0.01) –0.02* (0.01) –0.03 (0.10)
–0.01 (0.01) –0.02* (0.01) 0.01 (0.08)
–0.01 (0.01) –0.02* (0.01) 0.01 (0.08)
8.41** (2.07)
Budget
Constant
3 0.03 (0.11)
0.28** (0.08)
0.26** (0.04)
1.21^ (0.70)
1.27* (0.51)
1.33* (0.53)
–0.03 0.18 2.26 27
0.37 16.53 1.95 27
–0.05 0.67 2.33 27
0.41 5.67 2.19 27
0.4 4.42 2.2 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
public opinion pressures Congress and the president to pass legislation, which impels a modification of the legislative content to a less drastic, pro–status quo policy in the bargaining process, as Schattschneider indicated (1942). On the other hand, budget and start of terms are not significant in any specifications. Turning to the findings in the models of weighted policy change, Table 5.2 reports the results of the bivariate and multivariate specifications. Apropos of equations 1 and 2 for bivariate models, divided government in equation 1 is not significant. In contrast, the winset in equation 2 is significant, and the positive coefficient of the variable indicates that the wider winset is followed by a larger policy change. The adjusted R2 in equation 2 attests that the area of winset explains 32 percent of the variation in the weighted policy change. Equations 3, 4 and 5, containing the control variables, demonstrate results similar to the bivariate models. Whereas the winset remains robust in equations 4 and 5, divided government is insignificant in equations 3 and 5. The adjusted R2 in equation 5 suggests that the specification explains 36 percent of the variation in the weighted policy change. Parallel to the findings in the models of nominal policy change, public mood is significant with negative coefficients in equations 3, 4, and 5. Thus, public support for an activist government reduces, rather than enhances, policy change.
93
Veto Players Table 5.2 Congressional Weighted Policy Change 1 Divided Government
2
0.03 (0.05)
Winset
Public Mood Start of Term
Adjusted R2 F Durbin Watson N
4
5
3.66** (0.96)
–0.01 (0.04) 3.69** (0.99)
–0.01 (0.01) –0.01* (0.01) –0.03 (0.04)
–0.01 (0.01) –0.01* (0.01) –0.01 (0.04)
–0.01 (0.01) –0.01* (0.01) –0.01 (0.04)
3.66** (0.99)
Budget
Constant
3 0.02 (0.05)
0.13** (0.04)
0.13** (0.02)
0.58^ (0.32)
0.62* (0.24)
0.63* (0.25)
–0.02 0.42 2.27 27
0.32 13.42 1.96 27
–0.01 0.98 2.33 27
0.39 5.16 2.21 27
0.36 3.95 2.21 27
Entries are unstandardized coefficient from OLS regressions. Standard errors are in parentheses. ^p < .10 *p < .05 **p < .01 ^dL < DW < dU *DW < dL (.05 significance level)
Conclusions
Chapter 4 tested the pivotal interval movement model with the assumption of supermajoritarian, preferential voting. The findings suggest that when the residuum in the pivotal gridlock intervals is wide, the policy output of legislation is more likely to change. While this chapter has tested the same assumptions in the veto players model, the two models focused on the different indicators. The residuum in the interval movement model represents an area of the status quo that will be replaced by legislation. On the other hand, the winset in the veto players model indicates an area of the ideal points of legislation that will successfully replace the status quo. This chapter examined the veto players’ indifference curves and revealed the impact of the area of winset, but not divided government, on policy change. The findings suggest that legislative output by Congress could significantly change from the status quo when the winset is large. The two elements could influence the area of winset. First, if the status quo position is distant from the veto players’ ideal points, the larger distances will increase the size of veto players’ indifference curves and expand the area of winset. Secondly, if the veto players’ ideal points are proximate, their indifference curves are more likely to overlap and result in a larger winset. The influence of the area of winset indicates that individualistic preferences of legislators, rather than party control of the government, have an impact on
94
The Other Side of Gridlock
policy change. Apparently, this relationship could be nonrecursive; the content of legislation could influence legislators’ voting decisions, and vice versa. Legislators are likely to make voting decisions based upon the propinquity between their preferences and the policy content of legislation. Also, the content of legislation is likely to be modified in accordance with the preferences of legislators. Therefore, inasmuch as the policy content of legislative output is questioned, gridlock is caused not by divided government but by the proximity between the positions of the status quo and of the veto players’ preferences.
Chapter 6
Pre-Floor Agenda Block
In a recent work by Gary Cox and Mathew McCubbins (2005) the authors propose a formal model of legislative agenda setting predicated upon their well-known Legislative Leviathan thesis (1993). In the cartel agenda model, Cox and McCubbins postulate that the majority party monopolizes the cameral floor agendas because the majority party enjoys a negative agenda power, which is the ability to block unfavorable agendas from reaching the floor. Among the various tests of the cartel agenda model, Cox and McCubbins examine the so-called partisan roll rates. A party is rolled when a majority of its members vote against a measure, a majority of the opposite party’s members vote for the measure, and the motion passes on a final passage vote. Specifically, Cox and McCubbins examine the roll rates in the 83rd–105th Congresses and find that the House majority party was rolled on 1.7 percent of all final passage votes on joint resolutions while the House minority party was rolled on 25.9 percent (2002, 119; 2005, 92–93). Cox and McCubbins conclude that the difference in the “roll rates for the majority and minority party meet our expectations very well” (2002, 119). Cox and McCubbins contend that the majority party’s negative power is unconditional, in that it maintains an ability to block agendas regardless of preferential configuration within the party. In contrast, the pivotal interval movement model, studied in chapters 3 and 4, presents a preference-based model of nonpartisan legislators. The pivotal interval movement model underscores the supermajoritarian features of Congress and maintains that a policy 95
96
The Other Side of Gridlock
cannot be changed when its status quo point exists between the ideal points of the veto and filibuster pivots. The implication of the theory is that these two pivots, in combination with legislators who are ideologically more extreme, could block agendas from reaching the floor through implicit or explicit threats of filibusters and presidential vetoes. This chapter first examines the validity of roll rates as a measure for partisan negative agenda power. Secondly, the applied models of the cartel block and the pivotal block are constructed to offer different explanations of agenda setting behavior ex ante floor, and contending hypotheses are generated. Third, the chapter also proposes the Roll Aversion hypothesis, which postulates that when the majority party is likely to be rolled, the party prefers bipartisan support for the legislation to being rolled. Last, the empirical study tests the hypotheses by examining the House during the 83rd-105th Congresses. Roll Rates and Agenda Power
According to Cox and McCubbins, the majority party unconditionally holds negative agenda power, which is defined as the ability to block bills from reaching a final-passage vote on the floor (Cox and McCubbins 2002, 109; 2005, 20). Although the hypothesis tested by the authors is whether or not the majority roll rate is zero, the asymmetry between the roll rates of the minority party and the majority party generates an appearance of negative power seemingly held by the majority party. Cox and McCubbins presume that a party’s roll signifies the failure of exercising its negative power, thereby a low roll rate indicates a successful exertion of negative power. This seesaw-like supposition is noticeable in their argument for viewing roll rates as a measure for the prefloor negative agenda power, as follows: The better control the majority party has over which bills do and do not make it onto the floor agenda, the more it should be able to prevent the appearance of bills that will roll it on final passage; hence—all else equal—the lower its roll rate should be. (Cox and McCubbins; 2005, 247; italics added)
Nevertheless, it should be noted that roll rates are calculated based upon the amount of passed legislation on the floor. Thus, the roll rates indicate the opposite party’s positive agenda power, which is the ability to have legislation passed. This insinuates that a party’s high roll rate suggests its lack of negative power on the floor, rather than its lack of negative power before the floor consideration. Cox and McCubbins argue that the cartel agenda model postulates a zero majority roll rate (2005, 89 and 247). Let jJ and jN respectively denote
Pre-Floor Agenda Block
97
a majority of the majority party’s members and a majority of the minority party’s members. The amount of majority rolls is the amount of passed bills opposed by the majority of the majority party members and supported by the majority of the minority party members, defined as: N (J roll) = N (passed bills E jJ oppose I jN support).
The cartel agenda model presumes: N (jJ oppose) = 0, thereby N (J roll) = 0.
Thus, Cox and McCubbins assume a zero majority roll rate not because the bills opposed by the majority of the majority party members will fail, but because there will be no floor agendas opposed by the majority of the majority party. Therefore, the number or ratio of the bills opposed by the majority of the majority party on the floor would be a better measurement of the majority party’s negative power before the floor consideration. Notice that the proportion of the passage among the bills opposed by the majority of the majority party members and supported by the majority of the minority party should be an equivalent measure to the majority rolls: J Roll Ratio = N (passed bills) / N (jJ oppose E jN support).
The cartel agenda model predicts that the ratio in question would be undefined (Ø, or impossible), because the denominator is expected to be zero. While Cox and McCubbins maintain that the roll rate is not a continuous measure of agenda power (2005, 247), they presume that the low value of the majority roll rate is close enough to zero. Their argument is hardly convincing since such an assessment is parallel to an assumption that the ratio above, whatever the measured value would be, is close to an undefined value. Figure 6.1 further details the calculation of roll rate. The figure shows the potential six vote alignments, which are mutually exclusive and jointly exhaustive in the final-passage vote. Again, jJ and jN respectively denote a majority of the majority party’s members and a majority of the minority party’s members. In addition, a party is defined to be unrolled when a majority of its members vote against a measure, a majority of the opposite party’s members vote for the measure, and the motion fails on a final-passage vote. Thus, party unrolls indicate the party’s negative power on the floor, whereas the party roll suggests the opposite party’s positive power on the floor.
98
The Other Side of Gridlock Majority party (J) jJ oppose
jJ support Minority roll
jN oppose
Bipartisan opposition Minority unroll
Minority party (N)
Party vote
Majority roll Bipartisan support
jN support
Bipartisan vote
Majority unroll
Figure 6.1 Vote Alignments
As Figure 6.1 shows, when votes are split along the parties (party vote), some bills could pass (roll) or fail (unroll). Also, many votes could be bipartisan in support or opposition of measures. Thus, the set of the vote alignments, 1(v), is defined by: 1(v) = {majority roll, majority unroll, minority roll, minority unroll, bipartisan opposition, bipartisan support}.
Again, any votes should result in exclusively one of the vote alignments in Figure 1, defined as: i) prob (1) = 1. ii) prob (Ø) = 0.
Since the denominator of roll rates encompasses all the final-passage votes, including failed bills and bipartisan votes, a low roll rate of a party, such as, for example, the majority party, could indicate a higher frequency in the five other alignments in Figure 6.1. Nevertheless, none of these five indicators exhibit the majority party’s ability to block agendas from floor consideration. It is apparent that bipartisan support, bipartisan opposition, minority rolls, and minority unrolls are irrelevant to the majority party’s negative power. As Cox and McCubbins presume, a low roll rate of the majority party may indicate a higher unroll rate of the party. However, the majority party’s unroll rate indi-
Pre-Floor Agenda Block
99
cates the party’s ability to block legislation on the floor, rather than before the floor consideration. Thus, both the majority roll rates and majority unroll rates indicate the majority party’s negative power on the floor. A low frequency in the partisan roll rates does not indicate a prevalence of negative agenda power ex ante floor. Thus, the negative agenda power before floor consideration is better measured by the number of bills opposed by the majority of the majority party, or the ratio of the same to the number of all voted bills. If the majority party retains negative agenda power to reject legislation it does not like before floor consideration, the majority party will not oppose any agendas on the floor. Cox and McCubbins maintain that “all bills reported to the floor will pass with a majority of the majority party voting in favor” (2005, 43). Unconditional Gatekeeping?
Cox and McCubbins stress that “officeholders are expected never to push bills that would pass despite the opposition of a majority of their party” (2005, 27). Thus, the cartel agenda model postulates that the senior partners of the majority party invariably block the centrist agendas that could cause the defection of moderate members from the majority party on the floor. Such “unconditional” (Cox and McCubbins 2005, 37) negative power by the majority leadership is less realistic. Blockages of the centrist agendas result in unequal cost sharing among the majority members with a disproportionately large cost levied at the moderate members.1 A persistent imposition of such unequal duties has negative effects on the members’ morale and loyalty in the long term, hence raising the cost for vote buying within the party, enhancing the possibility of a successful discharge petition, and even increasing the chance of members switching parties. When legislators make voting decisions, they are cross-pressured. The legislators take account of their positions announced to their districts, the preference of the voters in their districts, and the views of their financial supporters (Fiorina 1974; Covington 1988; Edwards 1989; Kingdon 1989; Bond and Fleisher 1990; King and Zeckhauser 2003). Persistent pigeonholes of centrist measures supported by moderate members of the majority party would have a detrimental impact on the electoral standings of not only the centrist members, but also of the majority party as a whole in centrist districts. Scholars emphasize the intrapartisan mechanisms of “rewards and punishments” (Cox and McCubbins 1993, 2005; Snyder and Groseclose 2000; Sinclair 2002). However, rewards and punishments could be utilized not only by the party leaders toward the members, but also by the party members toward
100
The Other Side of Gridlock
their party leaders. When the majority party leaders face pressures, threats, and bargaining offers not to block particular measures from sizable amount of members within the party, they may opt to release the bills to the floor. Unequivocally, a speaker’s major goal is the passage of measures supported by the majority of the majority party. When a speaker undertakes coalition building for highly prioritized bills, the centrist members of the majority party may agree to sell their votes or “if you need me” vote options (King and Zeckhauser 2003) quid pro quo in exchange for granting a special rule (order) to the measures they support but the majority of the majority party members oppose (Oleszek 2004, 110). Cox and McCubbins contend that discharge petitions are “extremely difficult to use successfully” and thereby “not a thoroughfare of legislation” (2005, 86). Correspondingly, Beth (1998), Sinclair (2007, 17–20), and Oleszek (2004, 143) report a relatively modest quantity of successful discharge petitions. However, a small volume of discharged measures does not indicate a large amount of unsuccessful petitions. Since a successful discharge petition replaces the majority leader with the petition leader for the floor leadership, the threat of a discharge petition is likely to pressure the majority party leadership into releasing the legislation before it is successfully discharged. A few studies found cases of a bill reported to the floor because it was clear ex ante that a discharge petition would reach 218 signatures (Krehbiel 1987; Groseclose and King, 2004). Inversely, this could imply that there have been few discharge attempts because there were few floor majority-supported agendas blocked from floor consideration. Furthermore, even when a discharge petition is unlikely to be successful, the petition draws public attention to the key agendas blocked by the majority party, hence pressuring the majority leaders to release them (Oleszek 2004, 142–43). Lastly, the assumption of majority party leaders’ unconditional negative agenda power disregards presidential agenda power—Neustadt’s classic (1960) focuses on the presidential ability to persuade others to act on her or his agendas. Neustadt maintains that “congressmen need an agenda from outside, something with high status to respond to or react against. What provides it better than the program of the president?” (1960, 6–7). Kingdon agrees that “no other single actor in the political system has quite the capability of the president to set agendas” (1995, 23). The president could use soft and hard tools in order to promote her or his agendas (and congressional agendas she supports). The soft tools include inviting legislators to the White House, writing letters to legislators, and various electoral and legislative side payments to legislators. The hard approaches include the threat or use of vetoes against alternative measures to presidential proposals.
Pre-Floor Agenda Block
101
In divided government, a significant number of the floor agendas include presidential initiatives supported by the minority party and some members of the majority party (Edwards and Barrett 2000; Sinclair 2000). Edwards and Barrett (2000) observe that the president is more successful in obtaining the floor agenda status for his or her proposals in a unified government rather than a divided government. However, they find that even in divided government, the president plays a significant role in congressional agenda setting.2 Cox and McCubbins find that most of majority party rolls have occurred in divided government, wherein the opposition president exerts significant agenda-setting power, and concede that the majority party’s agenda control appears “more nuanced” (2005, 122) than in the idealized model. Therefore, the majority leaders are unlikely to wield unconditional blockage of agendas supported by the moderate members of the majority party, since the cost of doing so is higher than the cost of reporting the centrist agendas to the floor. For instance, in 2007, House Speaker Nancy Pelosi (D-CA) voted against some floor measures, including a bill that would provide funding in Iraq without setting withdrawal deadlines for troops (HR 2206) and a measure that would give U.S. spy agencies expanded power to eavesdrop on foreign suspects without a court order (S 1927).3 Comparative Analysis
Recently, analysis of roll rates became a focal point of scholarly attention and empirical controversy (Cox and McCubbins 2002, 2005; Krehbiel 2006, 2007; Groseclose and Snyder 2003; Sinclair 2002). A party is said to be rolled when a majority of its members vote against a measure, a majority of the opposite party’s members vote for the measure, and the motion passes. Cox and McCubbins test their cartel agenda model, which postulates that the majority party roll rate should be zero. Cox and McCubbins point at the disproportionately low level of majority roll rates as prima facie evidence for the cartel agenda model. Nevertheless, as detailed in the previous section, the roll rate, as the ratio of the rolls to all final-passage votes, does not indicate negative agenda power ex ante floor as a continuous parameter. Thereby, the comparative models entail a careful analysis and deductive hypotheses. In this section, the floor agenda model is explained as a baseline, nonpartisan pure-majoritarian model. Next, the cartel block model and the pivotal block model are constructed based upon the assumptions by the cartel agenda model and the pivotal politics model (Krehbiel 1996, 1998: Brady and Volden 1998, 2006), respectively, and the resulting hypotheses are proposed.
102
The Other Side of Gridlock
Floor Agenda Model
The floor agenda model is based upon the median voter theory (Black 1958), which presumes that a new median legislator’s ideal point becomes a point of the new policy output. The model postulates that the bills to be considered on the floor are determined by majority vote on the floor, hence positioning the agendas at the ideal point of the floor median. The floor agenda model supposes no influence of partisanship in the agenda-setting process. This implies that the majority of the majority party members does not have the ability to block unfavorable agendas from reaching the floor, and that when they (unsuccessfully) oppose such unfavorable agendas on the floor, the measure will pass with support from the majority of the floor (majority rolls). The logic of the floor agenda model is illustrated in Figure 6.2. The figure assumes that legislators’ ideal points4 are spread out over a finite interval. Let J, M, and N respectively denote the ideal points of the median of the majority party, the median of the floor, and the median of the minority party. This insinuates that half of the majority party, who is more liberal, exists to the left of J, and the other half (centrists) is present to the right of J. On the other hand, N splits the members of the minority party into halves in the diametric version. Mj and Mn are the reflection points of M through J and N, respectively, in that the distances between J and M and between Mj and J are equal, as are the distances between M and N and between N and Mn: | M–J | = | J–Mj |. | N–M | = | Mn–N |.
|
Q1 Qm nnnnnnnnnnnn yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy cutpoint 1
|
Q2 Qm nnnnnnnnnnnnnnnnn yyyyyyyyyyyyyyyyyyyyyyyyyyyy cutpoint 2
jJ jN Vote
Mj J M I I I Support Oppose Oppose Support Support Support Bi-sup J roll J roll
N I Support Oppose N roll
Status Quo: Q1
Status Quo: Q2
Mn . I Support Support Oppose Support N roll Bi-sup
Figure 6.2 Floor Agenda Model
Pre-Floor Agenda Block
103
Below the preference continuum, assumed vote outcomes contingent upon the location of the status quo are shown. Also, let jJ and jN denote the majority of the majority party members and the majority of the minority members, respectively. The floor agenda model postulates that the ideal point of the agendas reported to the floor is equivalent to the ideal point of the floor median, M. In Figure 6.2, the cutpoint is a midpoint between a proposed policy, Qm, and the status quo point. When the status quo is to the left of Mj, for instance, on Q1, a proposed policy on M, Qm, is better than Q1 for the legislators who position to the right of cutpoint 1 and constitute a floor majority. They include half of the majority party members to the right of J. Thus, the majority of the majority party and the majority of the minority party support the agenda Qm, and bipartisan support arises. When the status quo is between Mj and M, for example, on Q2, a proposed policy, Qm, is worse than Q2 for the legislators who position to the left of cutpoint 2. They include half of the majority party members who position to the left of J. However, Qm is better than Q2 for the legislators who are to the right of M and form a floor majority. Therefore, the majority of the majority party members oppose the bill, the majority of legislators on the floor support the measure, and the majority party is rolled. When the status quo is between M and Mn, symmetric actions occur and the minority party is rolled. These suppositions could be summarized as follows: Result F: Majority party, as well as minority party, could be rolled. When the majority of the majority party members oppose bills, their party is likely to be rolled. Likewise, when the majority of the minority party members oppose bills, their party is likely to be rolled.
Cox and McCubbins contend that the floor agenda model must assume the influence of the distance between Mj and M, hence the distance between J and M, on the majority roll rate. Nevertheless, as Krehbiel (2006) criticizes, this presumption is strictly contingent upon an unrealistic postulation of uniformly distributed positions of the status quo on agendas. Krehbiel contends that such an assumption of “fixed, exogenous, constant, and uniformly distributed status quo points for each Congress” defies history (2006, 12). As Krehbiel argues, the varied locations of status quo points of unique agendas should influence the majority roll rate under the floor agenda model (and other models). For example, both in Figure 6.3 and Figure 6.4, status quo Q1 remains at an identical point while the cutpoint is a midpoint between Q1 and the proposed policy Qm. The legislators to the left of the
104
The Other Side of Gridlock J1 I
I
M I
Q1
Qm
N I
.
nnnnnnnnnnnnnnnnn yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy cutpoint
Figure 6.3 Majority Roll 1
J2 I
I
M I
Q1
Qm
N I
.
nnnnnnnnnnnnnnnnn yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy cutpoint
Figure 6.4 Majority Roll 2
I
Mj I
J3 I
Q2
M I
N I
.
Qm
nnnnnnnnnnnnn yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy cutpoint
Figure 6.5 Bipartisan Support
cutpoint prefer Q1 to Qm, whereas the preference of the legislators to the right of the cutpoint is the opposite. Thus, a majority roll occurs regardless of the distance between J and M in so far as the status quo remains between J and M. In contrast, as Figure 6.5 shows, when the status quo exists outside of the reflection point, the majority of the majority party members and the majority of the minority members prefer Qm to the status quo Q2, thereby prompting bipartisan support. Thus, in the floor agenda model and other models, the location of the status quo point, rather than the distance between the floor median and partisan median ideal points, influences partisan rolls. Assumption 1: The lengths of intervals between pivotal ideal points do not indicate a location of the status quo points, hence they do not influence partisan rolls.
105
Pre-Floor Agenda Block Cartel Block Model
The cartel agenda model (Cox and McCubbins 2002, 2005) supposes that the senior partners of the majority party, including the committee chairs, the preponderant members in Rules Committee, and the speaker, are strongly responsive to the majority of the majority party members. The model thereby presumes that the majority party leadership blocks those agendas that the majority of its party membership dislikes from reaching the floor. Figure 6.6 illustrates the pre-floor agenda setting mechanism predicted by the applied model of cartel block model, which is based upon the assumptions of the cartel agenda model. In the figure, any agendas, if reported to the floor, will be amended to meet the preference of the chamber median, M, with the support of the floor majority, and will be passed. Therefore, when the status quo points of any agendas are present between Mj and M, the majority party leadership blocks the agendas from reaching the floor since amended and passed agendas on M are worse than the status quo points for the majority of the majority party members who position to the left of J. Consequently, when the status quo points of agendas locate between Mj and M, there will be no floor measures for the agendas. The cartel block model supposes the following: Result C1: None of the bills reported to the floor are opposed by the majority of the majority party members. Only the majority of the minority party members oppose bills. Thus, all party votes are the measures opposed by the majority of the minority party. When the majority of the minority party members oppose the bills, they are rolled. When they support the bills, there is bipartisan support. Hypothesis C1: The bills opposed by the majority of the minority party account for 100 percent of the party votes, while the bills opposed by the majority of the majority party account for 0 percent.
jJ jN Vote
Support Support Bi-sup
Mj
J
M
N
I
I
I
I Support Oppose N roll
Mn . I Support Oppose N roll
Figure 6.6 Cartel Block Model
Support Support Bi-sup
106
The Other Side of Gridlock
Result C2: Exogenous shocks shift policy output to M except when the status quo of a policy exists within the cartel gridlock interval (Mj–M). Hypothesis C2: When the status quo of an overall congressional policy is close to J, fewer legislative agendas reach the floor.
Pivotal Block Model
As explained in chapter 3, the pivotal interval movement model presumes that the veto pivot and the filibuster pivot maintain negative agenda power on legislation. Figure 6.7 shows the pre-floor agenda setting explained by the pivotal block model, which is based upon the assumptions of the pivotal interval movement model. In the figure, let f and v respectively denote the ideal points of the filibuster and veto pivots (40th percentile and 67th percentile legislators). When the status quo exists in the f –v (gridlock) region, any measures to move the status quo rightward and leftward are successfully rejected by a filibuster and the presidential veto, respectively. This suggests that the negative power held by pivotal legislators derives not only from the exercise of filibuster and veto, but also the implicit and explicit threat of such actions. When the status quo of a policy locates outside of the gridlock interval, the legislators are optimistic about, and thereby strive to build, a bipartisan coalition to change the policy (Edwards, Barrett, and Peake 1997; Krehbiel 1998, 82–92; Mayhew 2005, 216).5 On the other hand, when the status quo point exists within the gridlock interval, the legislators, who anticipate that any revisionist measure will be eventually rejected by a filibuster or presidential veto, would not be likely to expend resources to alter the policy. Thus, a bloc of legislators who are present outside the pivotal interval maintains the agenda block power. Result P1: Exogenous shocks shift policy output into the pivotal gridlock interval between f and v except when the status quo of a policy exists within the interval.
jJ jN Vote
Support Support Bi-sup
Mj
f
J
M
N
v
Mn
I
I
I
I
I
I
I
Opp Sup J roll
Figure 6.7 Pivotal Block Model
p.
Supp Support Opp Support N roll Bi-sup
Pre-Floor Agenda Block
107
Hypothesis P1: When the status quo of an overall congressional policy is close to a chamber median, M, fewer legislative agendas reach the floor.
Next, the pivotal block model in Figure 6.7 presumes that when a centrist agenda arises to move the status quo in the interval between Mj and f, the majority of the majority party members would oppose the bill, but the centrist members of the same party would support the measure with the members of the minority party. Result P2: When the status quo points of agendas locate between Mj and f, the majority of the majority party is likely to oppose the agendas on the floor. Hypothesis P2: Some party votes are measures opposed by the majority of the majority party.
Data and Method
In order to test the contending hypotheses generated from the cartel block model and pivotal block model, the empirical study focuses on the floor agendas opposed by the majority and minority parties, and the legislative agendas that reached the floor in the U.S. House during the 83rd-105th Congresses (twentythree biennia). Also, the number of floor measures opposed by the two parties among all joint resolutions reaching a final-passage vote6 is calculated from the dataset compiled by David Rohde.7 Dependent Variable
As for the size of the agendas that have achieved the floor agenda status, the number of the legislative agendas extracted by Binder (1999, 2003) is utilized as the denominator for the amount of the final-passage votes. Binder acknowledges what Cobb and Elder (1983) call the systemic agenda as a demand for legislation, hence a denominator for potential legislative enactments. The systemic agenda consists of “all issues that are commonly perceived by members of the political community as meriting public attention” (Cobb and Elder 1983, 85). To measure the systemic agenda, Binder tallies the agendas appearing in the editorials with legislative content in the New York Times.8 The empirical test in this study examines the ratio of the final-passage votes to the legislative agendas (see Figure 6.8).9
1
2 3
Binder Legislative Agendas
4 5
6 7
8 9
10 11
Final Passage Votes
12 13
14 15
16
Congress
17
Figure 6.8 Final-Passage Vote and Legislative Agendas: 83rd–105th Congresses
18 19 20 21 22 23
U
24 25 26
U
U
27 28 29
30 31 32
U U
U
U
U
U
U
n
U
U
U
U
U
U
U U
U
U
U U
U
33 34 35 36 37 38
n
Congress
Figure 6.9 Status Quo (Mean Winning Coordinate Dimension 1): 83rd–105th Congresses
39 40
108
Pre-Floor Agenda Block
109
Independent Variables
In the model of floor-reached agendas, the influence of the distance of the status quo from the floor median (Q–M) and from the majority party median (Q–J) are compared. All these ideal points are measured by the DW-NOMINATE scores in Dimension 1, and the distance is measured by an absolute value of the difference between the two points. The status quo, Q, is measured by the overall policy output by the previous Congress based upon the scores of the Mean Winning Coordinate (Poole and Rosenthal 1997, 59–70) in Dimension 1 (see Figure 6.9). The Mean Winning Coordinate (MWC) covers a mean value of the DW-NOMINATE scores of winning outcomes on roll calls. This figure approximates a central ideal point of the supporters of passed bills. Also, in the model, control variables of budget situation, divided government, and (lagged) public mood are added. For the variable of public mood, I use the lagged value of Stimson’s (1991, 2006) public liberalism index. As for the budgetary situation, I measure the budget deficit as a percentage of federal outlays. For the variable of divided government, the House in which the presidential party holds minority is coded 1, and 0 otherwise. Findings
Table 6.1 shows the proportion of the floor agendas opposed by the majority of the majority party and by the majority of the minority party in party votes. The cartel agenda model postulates that all the party votes are measures opposed by the minority party. Nevertheless, in Table 6.1, the bills opposed by the minority party amount to 90 percent, not 100 percent, of the party votes during the twenty-three Congresses. Cox and McCubbins maintain that the cartel agenda model is a baseline model, and in the real world, the party leaders could make errors (2005, 89). Nonetheless, the motions the majority party has opposed account for 10 percent, which seems to be beyond the error parameter. Thus, although there is an apparent asymmetry in the capability of negative agenda power between the majority and minority parties, the negative agenda power of the majority party is not unconditional. Next, the influence of the distances of the status quo from the floor median (Q–M) and from the majority party median (Q–J), on the percentage of the legislative agendas reached on final-passage vote is examined in the Glogit10 model. The results appear in Table 6.2. While the variable of Q–M Distance exhibits a recognizable impact in all the equations,11 the explanatory power of the multivariate equations suffers from the low degrees of freedom. The positive coefficient of the variable of Q–M Distance implies that the proximity
110
The Other Side of Gridlock Table 6.1 Percentage Opposed Bills among Party Votes Majority Party
Minority Party
Congress
Party Votes
Opposed Billsa
83
14
2
14.3
12
85.7
84
11
2
18.2
9
81.8
85
16
6
37.5
10
62.5
86
27
3
11.1
24
88.9
87
34
1
2.9
33
97.1
88
41
3
7.3
38
92.7
89
47
0
0.0
47
100.0
90
34
3
8.8
31
91.2
91
19
5
26.3
14
73.7
92
17
5
29.4
12
70.6
93
32
2
6.3
30
93.8
94
67
7
10.4
60
89.6
95
58
4
6.9
54
93.1
96
51
0
0.0
51
100.0
97
34
4
11.8
30
88.2
98
58
6
10.3
52
89.7
99
52
5
9.6
47
90.4
100
48
1
2.1
47
97.9
101
42
1
2.4
41
97.6
102
53
5
9.4
48
90.6
103
63
1
1.6
62
98.4
104
73
1
1.4
72
98.6
105
66
4
6.1
62
93.9
Total
957
71
–
886
–
Standard Deviation
18.55
2.07
Average
41.6
3.1
a Excludes b Opposed
% Opposed Billsb
9.66 10.2
Opposed Bills
18.72 38.5
% Opposed Bills
9.66 89.8
bipartisan opposition bills / Party votes *100
between the status quo and the floor median decreases the proportion of agendas eventually reaching the floor. On the other hand, the variable of Q–J Distance is insignificant in any equation. Thus, when the status quo is closer to the floor median, rather than majority party median, more agendas are blocked from reaching the floor. Roll-Aversion Hypothesis
The findings above substantially question the unconditional negative power of the majority party before floor debate, postulated by the cartel agenda model.
111
Pre-Floor Agenda Block Table 6.2 Perecentage Agendas in Final-Passage Vote 1 Q-M Distance
2
4.08+ (2.70)
Q-J Distance
3 4.52+ (2.91)
0.05 (1.59)
4
5
4.52+ (3.09)
–0.77 (1.63)
6 4.95 (3.44)
0.68 (1.96)
–0.46 (2.06)
Budget
–0.02 (0.02)
–0.01 (0.02)
–0.02 (0.02)
Divided Government
–0.24 (0.32)
–0.23 (0.39)
–0.19 (0.37)
Public Mood
0.01 (0.03)
0.03 (0.03)
0.01 (0.03)
Constant Adjusted R2 F Root MSE N
–0.09 (0.21)
0.15 (0.40)
0.07 (0.39)
–0.9 (1.96)
–1.49 (2.11)
–0.77 (2.10)
0.06 2.28 0.67 23
–0.04 0 0.71 23
0.02 1.21 0.68 23
–0.04 0.81 0.7 23
–0.09 0.55 0.8 23
–0.09 0.62 0.72 23
Entries are coefficients from weighted least squares logistic regressions for grouped data (G logit). +p < .10. Standard errors are in parenthesis.
The results show that 10 percent of party votes are the measures the majority of the majority party opposed. On the other hand, the findings did not strongly support the pivotal block model either. Among the party votes, the measures the majority of the majority party has opposed account for 10 percent, while the equivalent rate for the minority party is 90 percent. Although the pivotal block model assumes that the number of opposed bills should be influenced by the location of status quo, the disparity in the number of the opposed measures generates an appearance of the preponderant, if not unconditional, pre-floor negative power held by the majority party. Nevertheless, what if the majority party opts to support, rather than oppose, the unfavorable measures to the majority of its members? Occasionally, such an option could be economic and optimal for the majority party, especially when the party members know ex ante that its centrist members are likely to vote with the minority party for a bill and that a blockage of the measure would be costly and/or result in a discharge petition. As scholars observe (Key 1966; Jacobson 1996; Gerber and Green 2000; Snyder and Groseclose 2000; Cox and McCubbins 2005), political parties, especially the majority party, are likely to pursue both substantive and ornamental legislative achievement. The majority party’s legislative accomplishment is
112
The Other Side of Gridlock
signified by success in the passage of agendas the majority party proposes—and by the effective rejection of the agendas proposed by minority party. These two goals indicate the legislative defeat the majority party wants to avoid: (1) failure in the passage of the agendas proposed by the majority party, and (2) successful passage of the alternative agendas proposed by the minority party. Which would look worse and have more adverse effect on the majority party’s reputation and prestige? Obviously, the answer is the latter. While the former reveals the weakness of the majority party’s legislative leadership, the latter epitomizes both that and the minority party’s effective leadership. For example, in Figure 6.10, legislators’ ideal points are spread out over a finite interval [0, 10]. Assume a Republican majority floor, and let M, Q, and J respectively denote the ideal points of the floor median, status quo, and the majority median. For the current status quo, policy utility for the majority median, J, is –1, in that J’s preference is distant from Q by one degree (7 minus 8). Thus, zero is assumed to be a highest possible utility, and the majority party members at M retain –2 policy utility (5 minus 7). Suppose that a minority party member has introduced an agenda to move Q to M, which would increase the utility of the majority party members at M from –2 to 0. The majority leaders would use carrots and sticks (Cox and McCubbins 2005, 23) to have the members at M not support the minority agenda, which would decrease the utility of J by 2 (from –1 to –3). Imagine, however, that it has become clear that the majority leaders’ pressure is ineffective, and the members at M would support a discharge petition if the agenda were blocked. In Figure 6.11, the first and second entries in parentheses indicate the utilities of J and M, respectively. Let m denote the majority party members whose ideal point is on M. The entries in brackets denote the positive (+) and negative (–) significance of legislative achievement by the majority party. The situation above suggests that in Figure 6.11, the majority party members accurately anticipate that the _–r option cannot be maintained. In this circumstance, an obstinate effort to maintain Q(7) by the majority party would merely lead to the _–b path, the worst outcome, which would result in the lowest utility for J (–3) and a significant legislative failure, [ - - ], for the majority party. In order to avoid a legislative failure, the majority leaders may offer their members on M a compromise plan, ¡, to change Q (7) to Q’ (6). Imagine that
0
M
Q’
Q
J
5
6
7
8
Figure 6.10 Majority Party Members
10
113
Pre-Floor Agenda Block J unroll (–1, –2) [++]
r.
loyal
b.
defect
¡.
Q’ (6)
q.
Q’’ (5)
m
_.
maintain
J roll (–3, 0) [– –]
Q (7)
`.
change
Bipartisan Support (–1, –2) [+]
m
Bipartisan Support (–3, 0) [+]
Figure 6.11 Status Quo and Responses of Centrist Members in the Majority Party
a few members are still likely to defect. Then, the majority leadership would allow its moderate members in a committee (or on the floor or in conference committee) to propose a majority agenda to move the Q (7) to Q” (5), which is not different from the minority agenda. Still, such a choice is rational for the majority leadership since the `–q path would lead to [+], a moderate legislative accomplishment, while the _–b path would cause a legislative disaster for the majority party. Thus, when centrist members of the majority party are likely to defect and the majority leadership considers that a blockage of the measure is impossible, the party may strive to build intraparty support for the bill to form bipartisan support in an effort to avert the political mishap of a majority roll. Therefore, the bipartisan centrists could invisibly influence the majority party to build a bipartisan support of legislation through the second face of power—power through anticipation (Bachrach and Baratz 1970; Lukes 1974). Figure 6.12 models the Roll Aversion Hypothesis. When a centrist agenda arises to move the status quo between Mj and f toward the center, the centrist members of the majority party would support the measure. The majority
jJ jN Vote
Mj f I I Support Support or Oppose Support Support Bi-sup Bi-support or J roll
J I
M I
Figure 6.12 Roll Aversion Hypothesis
114
The Other Side of Gridlock
leadership may strive to gain intraparty support for the bill to craft bipartisan support in an effort to avert a majority roll. If the majority party leaders’ effort to gain support from the majority of the party is unsuccessful and the majority of the majority party members still oppose the measure, a majority roll occurs.12 Result R1: When the status quo points of agendas locate between Mj and f, there will be bipartisan support or majority roll. When the majority of the majority party opposes bills, the majority party is likely to be rolled. Hypothesis R1: The proximity between the status quo of an overall Congressional policy and Mj generates more bipartisan support than does the proximity between the status quo and Mn. Hypothesis R2: When the majority of the majority party opposes a floor agenda, the majority party is likely to be rolled.
Roll Aversion: Findings
In order to test Hypothesis R1, the influences of the distances of the status quo from the majority party’s reflection point (Q–Mj) and from the minority party’s reflection point (Q–Mn), on the percentage of bipartisan support in all final-passage vote are examined in the Glogit13 model. The results appear in Table 6.3. Both the variables of Q–Mj Distance and of Q–Mn Distance are significant with negative coefficients in their bivariate models. The negative coefficients suggest that when the status quo exists closer to the reflection points, bipartisan support is more likely to occur, as all three models predict.14 However, the larger value of the coefficient of the variable of Q–Mj Distance indicates that the majority party is more likely than the minority party to support measures as the status quo point approaches the party’s reflection point. This suggests that the majority party is less likely to oppose unfavorable legislation than the minority party is, as the Roll Aversion Hypothesis postulates. In the multiple models, the influence of the variable of Q–Mj Distance is robust while the counter variable of Q–Mn Distance becomes insignificant. Among the control variables, budget situation is significant, and divided government and public mood exhibit a recognizable impact. The significance of budget implies that the increasing Federal budget deficit has encouraged bipartisanship. The positive coefficient of divided government suggests a lower
115
Pre-Floor Agenda Block Table 6.3 Perecentage Bipartisan Support in Final-Passage Vote 1 Q-Mn Distance
2
3
–1.64+
–1.05+
(0.88)
(0.90)
Q-Mj Distance
–2.14* (0.90)
4
5
–0.93 (1.08)
–1.74+ (0.96)
6 –0.85 (1.25)
–2.96** (1.01)
–3.86** (1.30)
Budget
0.03 (0.02)
0.04* (0.02)
0.05* (0.02)
Divided Government
–0.1 (0.30)
0.49 (0.31)
0.72+ (0.37)
Public Mood
–0.04 (0.03)
–0.04 (0.03)
–0.06+ (0.03)
Constant Adjusted R2 F Root MSE N
1.91* (0.67)
1.74** (0.46)
2.32** (0.68)
3.99* (1.98)
4.9** (1.68)
5.29** (1.71)
0.1 3.45 0.65 23
0.17 5.64 0.62 23
0.18 3.54 0.19 23
0.06 1.35 0.66 23
0.33 3.79 0.56 23
0.34 3.3 0.66 23
Entries are coefficients from weighted least squares logistic regressions for grouped data (G logit). +p < .10. *p < .05 **p < .01 Standard errors are in parenthesis.
tendency toward bipartisan support for legislation in unified government. This is likely to indicate that the opposition (minority) party in unified government is more likely to challenge the majority party’s agendas, many of which are presidential agendas. In contrast, the opposition (majority) party in divided government may anticipate presidential vetoes and propose agendas that are more likely to be supported by the presidential (minority) party. The negative coefficient of public mood shows that the public liberal mood decreases, not increases, bipartisanship. Except for the 83rd and 105th Congresses, the Democratic Party always has been the majority party in the examined Congresses. In the Democratic majority Congresses, a low level of liberal mood in the public may have fostered the Democratic Party to propose and enact centrist agendas supported by the GOP. Next, in order to test Hypothesis R2, the quantity of rolls among the measures opposed by the majority of the majority party is examined. Table 6.4 shows the overall number of rolls, opposition against measures, and the ratio of the former to the latter, for the majority and minority parties during the 83rd-105th Congresses. The table shows a stark contrast between the majority and minority parties in the amount of partisan rolls. While the majority party was rolled fewer than three times per Congress, the minority party was rolled
116
The Other Side of Gridlock Table 6.4 Opposed Bills and Partisan Roll Majority Party Opposed
Total Average Standard Deviation N aExcludes
Minority Party Opposed
Rolls
Billsa
57
71
–
822
886
–
2.6
3.2
80.3
35.7
38.5
92.7
1.72
2.92
27.38
18.34
19.12
21
21
21
23
23
% Rolls
Rolls
Billsa
% Rolls
9.69 23
bipartisan opposition
thirty-five times per Congress. Nonetheless, 80 percent of the motions opposed by the majority party have resulted in majority party rolls, whereas the corresponding ratio for the minority party is 93 percent. The conspicuous quantity of rolls among the measures opposed by the majority party begs the question of why the senior partners of the majority party have not blocked these agendas and prevented the defection of its centrist members as predicted by the cartel agenda model.15 Conclusions
The cartel agenda model, the extension of the majority party cartel theory, entails the presumption of an insuperable intraparty unity that monopolizes floor agendas without a defection of its centrist members. However, because of the substantial amount of centrist districts, candidate-centered campaign machines, and discharge petitions, the party cartel assumption is likely to be often violated in actuality. The findings in this chapter mainly support a disproportionate, if not unconditional, majority party’s negative agenda power ex ante floor. Still, 10 percent of the party votes in the U.S. House during the 83rd-105th Congresses were measures opposed by the majority of the majority party members. In addition, this asymmetry may be partially exaggerated since the results also reveal that the majority party is more likely than the minority party to support unfavorable legislation ceteris paribus. The cartel block model entails an influence of the distance between the status quo and the majority party median ideal point on the agenda mobilization. Nonetheless, the findings in this chapter do not support the presumption. Rather, the results for the agenda mobilization largely support the inference based upon the pivotal block model. The findings indicate that the proximity between the status quo point and the floor median ideal point impedes the mobilization of agendas to the floor. Thus, the ideal point of the floor median,
Pre-Floor Agenda Block
117
not the majority party median, appears to be an approximate center of the pre-floor agenda block interval. Whereas the results in this chapter question the assumptions of the cartel block model, the study does not reveal when unfavorable agendas for the majority of the majority party members reach the floor. Perhaps the influence of an ideological configuration of the dissidents within the majority party, the policy context of the measures considered, and the centrist members’ electoral advantage in their districts deserve careful study in the future.
This page intentionally left blank.
Chapter 7
Conclusion You talk about discipline! Here, he’s elected by the caucus, and then he’s elected by the House. Here, you got conservatives and moderates and liberals! You can’t discipline that! I heard on the radio saying we ought to discipline. There’d be five parties here if we tried to do that. —Speaker “Tip” O’Neill, 1981
Political scientists once envisaged an American government founded on active and effective political parties. The Report of the Committee on Political Parties of the American Political Science Association (1950) legendarily proposed an intermingled system of representative democracy and active political parties in the so-called responsible party government doctrine: An effective party system requires, first, that parties are able to bring forth programs to which they commit themselves and, second, that the parties possess sufficient internal cohesion to carry out these programs. (Italics added)
Nonetheless, the American party system has been historically characterized by fragmentation within the parties and prevalent individualism of party members. In the beginning of the twentieth century, A. Lawrence Lowell compared the partisanship in the British House of Commons and in the U.S. House of Representatives (1901). Lowell focused on party voting, which was a vote in which 90 percent of the voting membership of one party opposed 90 percent of the voting membership of the other party. He found that the percentage of party voting in the House of Commons was higher than 90 percent whereas the same in the U.S. House was lower than 30 percent. After the revolt against Speaker Joe “Czar” Cannon in 1910, party leadership in the U.S. Congress was further diffused. Party members retained 119
120
The Other Side of Gridlock
considerable freedom in their voting decisions, and successful party leaders were skilled negotiators with the party members. In the prewar period, E. E. Schattschneider observed: On difficult questions, usually the most important questions, party lines are apt to break badly, and a straight party vote, aligning one party against the other, is the exception rather than the rule—when all is said, it remains true that the roll calls themselves demonstrate that the parties are unable to hold their lines in a controversial public issue when the pressure is on. (1942, 130–31; italics in original)
In the postwar era, scholars still recognized the decentralized nature of the parties and autonomous party members. Julius Turner (1951) and Julius Turner and Edward V. Schneier (1970) reexamined the saliency of party voting in the House, and found a significant and continuous decline in the percentage of party voting. Cooper and Brady argued that the Speaker and other party leaders “function less as the commanders of a stable party majority and more as brokers trying to assemble particular majorities behind particular bills” (1981, 417). In parallel, Barbara Sinclair (1989) contends: Members of Congress are elected on their own: they build their own organizations, they raise their own money and, to a considerable extent they manage to cultivate their constituencies to insulate themselves from national tides. Consequently, party leaders cannot influence whether members attain their reelection, power or policy goals to the extent that the pre-1910 party leadership could. To be sure, current leaders do have resources that can be employed to influence their members, but, for most members, most of the time, what the leadership can do for them or to them is not critical for their goal attainment. Current leaders must rely upon persuasion-based strategies; they do not possess the resources necessary to command.
However, in the 1990s, several scholars began to contend that politics in Congress and the presidency had become more partisan in the 1980s (Aldrich 1995; Fleisher and Bond 1996; Rohde 1991; Sinclair 1995). These researchers attributed their argument to the surge in the percentage of party votes—votes in which a majority of one party vote against a majority of the other party. For instance, Rohde (1991) contended that the enhanced equivalence between the northern and southern Democrats after the entrance of African American voters in the South had increased partisanship.
121
Conclusion Phantom of Party Government
The increasing party votes since the 1980s, and thereby the façade of heightened partisanship, have rejuvenated party-centered theories, which Schattschneider and other prewar scholars jettisoned forty years ago. For instance, Cox and McCubbins’s (1993) Legislative Leviathan hypothesis maintains that majority parties function as cartels to serve the party, not the chamber, and make legislative decisions solely for the electoral interests of the majority party. As summarized in Table 7.1, one of the party-centered theories includes the party control of government theory. Several researchers contend that divided government is a major cause of stalemate in presidential-congressional relations (Sundquist 1980, 1988, 1992; Cutler 1988; Fiorina 1996, 2003). These scholars observe that in divided government, a majority party in Congress “has every incentive to reject presidential initiatives; to accept them is to acknowledge the president’s competence and sagacity, hence, to support his reelection” (Fiorina 2003, 86). The party control of government theory presumes majoritarian rule combined with solid partisan unity. In divided government, the opposition (majority) party in Congress is assumed to be united and maintain majority votes in a chamber. The majority votes held by the majority party are presumed to effectively reject the presidential agendas. Vice versa, the president supposedly rejects the legislative agendas, which are assumed to be the majority party’s agendas. The indication in the party control of government theory for policy stability is that divided government stalls lawmaking and causes high policy stability. On the other hand, Cox and McCubbins (2002, 2005) propose a formal model, which is theoretically more rigorous than the party control of governTable 7.1 Theories of Policy Stability and Their Implications Causes of Policy Stability
Theories Party Control of Government
Stability in Preference of Majority Party Median
⫻
Cartel Agenda Model Pivotal Interval Movement Model
Divided Government
Split Government
⫻
⫻
Veto Coalition Stability in Preference of Supermajority Pivots
⫻
Majority Party
⫻
Majority Members of Majority Party 1/3+1 or 4/10+1 Ideologically Extreme Legislators
122
The Other Side of Gridlock
ment theory. The cartel agenda model posits that the senior partners of the majority party, including committee chairs, the preponderant members in the Rules Committee, and the Speaker of the House, are strongly responsive to the majority of the majority party members. When the majority of the majority party membership dislikes a bill, the majority party leadership will block the legislation from reaching the floor ex ante. Accordingly, as shown in Table 7.1, the majority of the majority party, with support from majority party leaders, maintains the ability to impede legislation. Thus, overall, policy output is assumed to move toward the median preference of the majority party in every Congress. Implicit in the theory is that high policy change occurs when the majority party changes. Although the cartel agenda model does not explicitly offer an explanation about the impact of divided government on policy stability, its assumption of the majority party’s dominance within chambers indicates that split government, with different parties holding majorities in the two chambers, is likely to cause policy stability. Do parties matter as party-centered scholars argue? Do parties in government influence policy output? The answer appears to be “no,” or, at the very least, “not that much.” This book has utilized the applied models of gridlock interval movement and veto players to examine the influences of the party control of government and of the median preference of the majority party. The findings show that divided government and the change in the median preference of the majority party have no impact on policy stability. Rather, the findings suggest that the ideal points of the supermajority pivots have significant impact on policy stability. The test of the pivotal interval movement model in chapter 4 reveals that the width of the residuum, which is the portion of the previous pivotal interval not overlapped by the new interval, has a positive influence on policy change. This means that the ideological change in the endpoints of pivotal interval, which are the ideal points of the 34th percentile and 60th percentile members, increases policy change. Similarly, the veto players model in chapter 5 demonstrates that the 34th percentile and 60th percentile members in the two chambers maintain the ability to singlehandedly reject legislation. Also, chapter 6 reveals that the proximity between the previous policy output and the floor median ideal point decreases the percentage of legislative agendas reaching the floor. These findings suggest that the filibuster pivot or veto pivot, in juxtaposition with the legislators who are more ideologically extreme, can reject legislation she dislikes. Thus, 33 percent plus one legislators and forty-one senators who are ideologically extreme, maintain separate coalitions to halt legislative measures. The corollary, that the majority party status and unified government have no impact on policy output, may perturb some students. Nonetheless, a passage
Conclusion
123
of legislation needs the support of a supermajority of legislators to preempt or overturn filibusters and presidential vetoes. Beyond Positive Theories
This book has demonstrated that a minority of legislators inhibits revisionist measures that would change policy significantly. The indication is that American government is unable to change policy extensively regardless of the change in the majority party status. Thus, the Madisonian system maintains a proclivity for high policy stability. Is this good for American representative democracy? Some scholars (viz. Jones 1995; Krehbiel 1998) tacitly imply their affirmative views about the current system. For example, Krehbiel (1998, 231) argues: Supermajoritarianism seems to manage quite well the inherent tensions between policy stability and policy responsiveness. The resulting mix of incremental changes, moderate policies, and gridlock is not such a bad thing.
On the other hand, Brady and Volden (2006, 209) are more explicit and more elitist: America has moved further toward solving these common problems than have other countries, perhaps due to the policy gridlock of the past, which resisted public preferences for policies that were unsustainable in the long run: dramatic increase in entitlements, changes in the tax codes, and increased government regulation.
Nevertheless, whether or not the policy-stable American system is propitious depends upon the desirability of the status quo in the United States today. Tsebelis (2002, 7–8) contends: It is reasonable to assume that those who dislike the status quo will prefer a political system with capacity to make changes quickly, while advocates of the status quo will prefer a system that produces policy stability.
Before the collapse of the Soviet Union, cold war–related issues dominated the agenda in American politics. Today, in contrast, the United States is preoccupied with diverse, catalytic problems. Currently, the federal governmental debt accounts for 400 percent of the annual budget size. The interest
124
The Other Side of Gridlock
payment for the debt exceeds 15 percent of annual federal spending. The public and pundits believe that there will be another terrorist attack by Al Qaeda soon. Yet, the majority of the public believes that war in Iraq weakens the national effort to combat terrorism.1 Social Security funds are projected to be exhausted within thirty years. Both total medical costs in the nation and the number of Americans who are having health problems have skyrocketed in the last decade. Economic inequality between the rich and the poor is high, as divulged in the recent report of the GINI index by the U.S. Census Bureau. None of these problems can be solved by a marginal change to the status quo policy. In fact, the great majority of the public today expresses the opinion that the nation needs a new direction.2 However, in the American system today, significant policy change, to solve any national problem, can hardly be adopted because it requires support from two-thirds of the members of Congress. In addition, the judicial branches at the state and national level may reject a new policy, and government agencies at various levels may stall the implementation of new programs. Therefore, it might be worth considering several reforms to shorten the supermajority gridlock interval and make the Madisonian system more capable of policy innovation. Notice that these reforms would not generate policy instability when the public desire stability. When constituents do not wish policy change, they can vote for incumbent legislators and the incumbent president. The suggestion in this concluding chapter is that the United States may need a flexible mechanism to produce innovative policy at the time that the majority of the public and decision makers seek substantial policy change. In order to enhance governmental responsiveness to national problems, radical reformers may advocate the entire abolition of the supermajoritarian procedures such as senatorial filibusters and presidential vetoes. This chapter considers two less drastic reforms, including a unicameral override of presidential vetoes and a change in the Senate cloture rule. The chapter also examines a more drastic reform to introduce quadrennial elections of all the Senate seats. The first two plans are likely to minimize the gridlock interval whereas the last plan is likely to enhance the movement of the gridlock interval. All three plans would expand the width of residuum. Unicameral Override
The Constitution, Article I, Section 7, grants the president the authority to dismiss legislation passed by Congress. After a bill is sent to the president, she or he has ten days, excluding Sundays, to sign it, veto it, or take no action. When the president vetoes a bill, her or his veto is subject to the ability
125
Conclusion
of Congress to override the veto by a two-thirds vote of the members present and voting in the two chambers. Thus, with a threat or exercise of presidential veto, a minority (n/3 + 1) of legislators whose preferences are close to the president’s preference in a single chamber can block legislation they dislike. As explained in chapter 3, since a minority of legislators in only one chamber can sustain presidential vetoes, that veto pivot in the two chambers which is more ideologically extreme is likely to maintain the pivotal role. For example, in Figure 7.1, the pivotal interval is likely to extend between Vh and F as Pivotal Interval 1 exemplifies. Hypothetically, if Congress had the ability to override the veto by a twothirds vote of the members present and voting in one chamber, one-third plus one of the legislators in both chambers must oppose a bill in order to sustain a presidential veto. In this situation, that veto pivot in the two chambers which is ideologically more moderate is likely to maintain a pivotal role. Therefore, in Figure 7.1, the pivotal interval is likely to be present between Vs and F, as Pivotal Interval 2 exemplifies. Thus, unicameral override is likely to shorten the gridlock interval and widen the residuum. Cloture Vote
Senate Rule XXII was adopted in 1917 and allows for a cloture vote to end extended debates, so-called filibusters, in the Senate. Before 1975, twothirds of senators present on the floor were required to vote for a cloture. In contrast, since 1975, at least three-fifths of the senate membership must vote to invoke cloture. Thus, under the current rule, regardless of the number of senators present on the floor, a cloture vote requires sixty votes. This indicates that when the number of present senators on the floor (n) is relatively small, an advantage to the supporters of filibusters is given. A minority of senators, n–59 members, on the Senate floor could take dilatory action to block legislative agendas. If cloture requires votes by three-fifths of the senators present, it is likely to shorten the gridlock interval. For example, if ninety senators are present on the p
Vs
F
Senate
|––––––––––––––––––––––| Pivotal Interval 1 |–––––––––––––––| Pivotal Interval 2 Vh
House
Figure 7.1 Bicameral Pivotal Gridlock Interval
126
The Other Side of Gridlock
floor, cloture will require fifty-four votes. If ninety-five senators are present, fifty-seven votes are required to end filibusters. Thus, replacing the membership denominator with the number of present senators is likely to contract the gridlock interval and extend the width of residuum. Quadrennial Election of Senators
Today, senators are elected for staggered six-year terms, with one-third of the Senate’s seats available for contest every two years. Thus, approximately thirty-three out of one hundred seats are competed for every second year. This indicates a sluggish rate of change in the ideological configuration of members in the Senate. For instance, suppose all thirty-three senators up for election sought reelection and approximately one-half of them (sixteen senators) lost. Only the ideal points of 16 percent of all senators would change. In parallel, Table 4.1 in chapter 4 reports that pivotal Senate residuum are significantly shorter than the pivotal House residuum. The implication is that the change in the ideal point of the Senate veto pivot is considerably smaller than the equivalent of the House veto pivot. If all the Senate seats are up for a vote every four years, especially in presidential election years, electoral outcomes are more likely to change the ideal points of the veto and filibuster pivots, thereby widening the residuum. Gridlock Continues: The 110th Congress and the War in Iraq
Soon after the Democratic landslide in the 2006 congressional elections, the soon-to-be Speaker of the House Nancy Pelosi (D-CA) indicated her desire to change the policy on the war in Iraq. Pelosi stressed that “nowhere was the call for new direction more clear from the American people than in the war in Iraq.”3 Nonetheless, on January 10, 2007, President Bush announced that he was sending an additional 21,500 troops to Iraq. Within a few weeks, Senators Joseph R. Biden (D-DL), chairman of the Foreign Relations Committee, Carl Levin (D-MI), chairman of the Armed Services Committee, and Chuck Hagel (R-NE) introduced a bipartisan, nonbinding resolution (S Con Res 2) to oppose widening the military’s involvement in Iraq. The Biden-Levin-Hagel resolution stated that the responsibility for providing internal security should be transferred to the Iraqi government under an expedited timeline. On the other hand, John W. Warner (R-VA) introduced another resolution, which expressed a disagreement with Bush’s decision to deploy additional troops to Iraq but also advised against cutting funds as a way to express opposition to the conduct of war in Iraq. In February, by incorporating Warner’s provi-
Conclusion
127
sion, Levin and Warner introduced a compromised resolution (S Con Res 7). On February 5, although Republicans requested consideration of alternative resolutions on the Senate floor, majority leader Harry Reid (D-NV) declined to discuss measures other than the Levin-Warner bill (S 470, originally S Con Res 7). As a result, the GOP members refused to begin debate on the LevinWarner Resolution, and the bill was stalled by rejection of a motion to invoke cloture, 49–47. On February 5, 2007, President Bush sought $103 billion supplementary spending for the war in Iraq and hurricane relief efforts. Democratic leaders planned to use the supplementary spending bill as the next stage to oppose the escalation of the military involvement in Iraq. However, Democrats were divided on this political tactic to link war funding and war strategy. Some moderate Democrats did not want to appear to be unsupportive of the U.S. troops in Iraq. For instance, Senator Sanford D. Bishop Jr. (D-GA) said, “Every dollar, every dime and every penny that the president’s asking for Iraq is going to be provided.”4 On the other hand, Dennis J. Kucinich (D-OH) maintained that “if you’re opposed to the war, you don’t vote to fund it.”5 Also, Senator Russ Feingold (D-WI) stated, “I will not vote for anything that the president could read as an authorization for continuing with a large military campaign in Iraq.”6 On March 22, 2007, the Senate Appropriations Committee approved a $121.7 billion supplementary spending bill by voice vote. The measure (HR 1591) would require reports by the president on the war’s progress and would set a nonbinding goal for completing most of the troop withdrawal by March 31, 2008. On March 26, by a 218–212 vote, the House passed its version of a $124.3 billion supplemental spending bill (HR 1591). In contrast to the Senate version, the House version set a binding withdrawal date. Under the bill, by July 1, 2007, the president must certify that various indicators in Iraq are meeting political and military benchmarks, and by October 1, 2007, he must do the same on a second set. If benchmarks are not met, a redeployment from Iraq must begin immediately and be completed by the end of December 2007. If benchmarks are met, redeployment must begin by March 2008 and be completed by the end of August 2008. Although both the Senate and House versions exceeded the president’s $103 billion request, President Bush immediately threatened to veto either measure because of the troop withdrawal provision and of the spending he did not request. In spite of President Bush’s repeated threats to veto, the Conference Committee adopted a compromised version of HR 1591 with a nonbinding goal of withdrawal. Under the bill, if the president did not certify to Congress by July 1 that various benchmarks were being met, withdrawal of most troops
128
The Other Side of Gridlock
would begin immediately with a goal of completing it by year’s end. The House adopted the conference report, 218–208, on April 25, and the Senate cleared the report, 51–46, the next day. In the House, thirteen Democrats voted against the bill while two Republicans voted for the measure. In the Senate, all of the present Democrats, one Independent, and two Republicans, Gordon H. Smith (R-OR) and Chuck Hagel (R-NE), voted for the measure. The Republican senators did not exercise a filibuster because the president had stated ex ante that he would veto the legislation. On May 1, the fourth anniversary of President Bush’s “Mission Accomplished” speech, Congress presented the supplementary bill to the president. The president promptly vetoed the supplementary bill with a pen given to him by the father of a Marine killed in Iraq. The president maintained that the bill “would set an arbitrary date for beginning the withdrawal of American troops without regard to conditions on the ground.” Also, the president asserted that the supplementary bill was “unconstitutional because it purports to direct the conduct of the operations of the war in a way that infringes upon the powers vested in the presidency by the Constitution, including as commander in chief of the armed forces.”7 On May 24, Congress eventually passed the $100 billion war funds bill without benchmarks. Thus, the impasse in the 110th Congress over policy change in the war in Iraq was mainly caused by supermajoritarian procedures. The bill (S 470) opposing the troop surge was filibustered in spite of the support from a majority of senators. The supplementary spending bill, with the goal of withdrawal from Iraq (HR 1591), was vetoed by the president without a supermajority support to override the veto. Would it have been different if the 110th Congress were in a Democratic unified government with a Democratic president? Probably not. Even if the president and the cameral majorities were Democrat, the Republican minority in the Senate would have filibustered bills they disliked. Thus, regardless of unified or divided government and regardless of the majority party in Congress, a change in the preferences of the forty-one senators who are relatively most supportive of the war in Iraq is likely to be influential on a potential change in Iraq war policy.
notes
Chapter 1: Introduction
1. 2. 3. 4. 5. 6. 7.
CQ Weekly, October 31, 2005. CQ Weekly, April 3, 2006. CQ Weekly, May 29, 2006. Ibid. CQ Weekly, September 18, 2006. CQ Weekly, December 12, 2005. Senate majority leader Bill Frist also voted against the cloture to merely preserve some procedural options. 8. CQ Weekly, December 26, 2005.
Chapter 2: Gridlock and Policy Stability 1. Rescaled Score = 120 – (Original Score x 5). 2. Mayhew (2005) updates his list of significant legislation until 2002 (2005, 208–13). 3. Binder notes a similar finding (2003, 42). 4. For the variables of Binder’s and Edwards’s measurements, which are ratio variables constructed from grouped data, I model the variation using weighted least squares logistic regression for grouped data (Glogit). The Glogit function accounts for the variation in the denominator for the ratio variables. 5. By examining Congresses from 1946 to 1986, Kelly (1993) fi nds a moderate impact of divided government. 6. In contrast, Binder finds an impact of divided government regardless of the salience level of legislative agendas (2003, 68–73). 7. Except for laws in the tax policy area. 8. A nonprofit political organization advocating liberal and democratic values. http://www.adaction.org/voting.htm.
129
130
Notes
Chapter 3: Pivotal Interval Movement 1. Exceptions in postwar midterm elections include the 1998 and 2002 elections. In 1998, Democrats, helped by a superb economy, gained five seats in the House and maintained the same amount of senatorial seats. In 2002, Republicans gained seats under President Bush’s post- 9/11 popularity. 2. Measured by the residuals from the equation, Voting decision = c + b (Electorate ideology) + e. 3. Vice versa, the legislators’ preferences could influence the constituency ideology. The “issue evolution” study reveals that political elites’ opinion often drives mass opinion changes (Carmines, Renten, and Stimson 1984; Carmines and Stimson 1989; Page and Shapiro 1992; Zaller 1992; Hetherington 2001; Layman and Carsey 2002; Brewer 2005). 4. Mayhew discusses “external events”, which had direct impact on the enactment of important laws (2005; 136–39, 217–19). According to Mayhew, examples of external events include Sputnik, the assassinations of Martin Luther King Jr. and Robert Kennedy, recession and energy crisis in the early 1970s, terrorist attack of September 11, collapse of Enron, etc. 5. Brady and Volden (2006, 116–20) compute the average scores of eighteen interest group ratings for the members of the 103rd House. The average ratings are highly correlated with the ratings solely by the ADA with the unstandardized coefficient 0.98. Poole and Rosenthal (1997, chapter 8) examine the 96th Congress and find that the D-NOMINATE scores are greatly related to the most commonly used interest group ratings, such as ADA, ACU, and ACA, with correlation coefficients near 0.95. 6. a = Q–{1/2(Q-Q’)} when Q > Q’, and a = Q’–{1/2(Q’–Q)} when Q’ > Q. 7. During World War I, the Senate fi libustered a bill to arm commercial ships against attacks by German submarines. Displeased, President Woodrow Wilson called a special session of the Senate, which eventually adopted Rule XXII. 8. Before 1975, cloture required two-thirds of senators present. 9. Oleszek (2004, 283–84) offers similar postulation. 10. For the pre-cloture period, the distance between the most conservative and the most liberal members is measured. 11. More specifically, the bills in question are unfavorable for F and the members who are to the left of F, and for V and the members who are to the right of V. 12. The reflection point of q through F is F + (F–q). 13. The reflection point of q through V is V–(q–V). 14. The cartel agenda model assumes open rules on the floor. 15. Brady and Volden maintain a similar assumption (2006, 20–21). Chapter 4: Empirical Test 1. A nonprofit political organization advocating liberal and democratic values. http://www.adaction.org/voting.htm.
Notes
131
2. Before 1975, the fi libuster pivots under Republican and Democratic presidency are the 34th and 67th percentile legislators, respectively. 3. Poole places the House and Senate in the same space based upon the ideal points of the 615 legislators who served in both the House and Senate (2005, 192–95). Legislators who switched parties are treated as two individuals. Binder questions Poole’s assumption of a stable preference of a member across the two chambers (2003, 142–43). Nonetheless, Poole’s study shows that legislators’ voting behavior changes marginally across the two chambers. While 86.4 percent of the voting behavior of the legislators who served in the two chambers is correctly classified based upon the rank ordering within separate chambers, an almost identical portion (86.1 percent) of the same members’ voting patterns is correctly classified based on the rank ordering in a joint configuration of the two chambers (2003, 19–20). Overall, the Pearson correlations between the joint OC scaling and separate House and Senate scaling is .98 (2003, 21–23). 4. Due to the inclusion of a fi libuster pivot, which exists in the Senate, this is not a unicameral interval in a stringent term. 5. Mj = J – (M–J) for the Democratic majority Congress, and Mj = J + (J–M) for the Republican majority Congress. 6. The 107th Congress (2001–03) is computed as divided government with the Democratic majority Senate. From January 3 to January 20, 2001, with the Senate divided evenly between the two parties, the Democrats held the majority due to the deciding vote of the outgoing Democratic Vice President Al Gore. Between January 20, 2001, and June 6, 2001, Republican Vice President Richard Cheney held the deciding vote, giving the majority to the Republicans. On May 24, 2001, Vermont Senator James Jeffords announced his switch from Republican to Independent status, effective June 6, 2001. Jeffords revealed that he would caucus with the Democrats, giving the Democrats a one-seat advantage, thereby changing control of the Senate from the Republicans back to the Democrats. Thus, the Democratic Party again became the majority on June 6, 2001. The November 5, 2002, election brought to office elected Senator James Talent (R-MO), replacing appointed Senator Jean Carnahan (D-MO), shifting the balance once again to the Republicans—but no reorganization was completed at that time since the Senate was out of session. 7. James A Stimson. http://www.unc.edu/~jstimson/time.html. Chapter 5: Veto Players 1. As for the fi libuster pivots before 1975, the thirty-fourth and sixty-seventh percentile legislator under Republican and Democratic presidency, respectively, are measured. 2. The OC scores are comparable between chambers. The OC scale ranges from –1 to +1, and produces ideal points and cutting planes for the roll call votes
132
Notes
that maximize the number of correctly classified voting decisions (Poole 2005, 46–86). 3. I used AutoCAD® by Autodesk. AutoCAD® is a design and drafting software used for industrial engineering. 4. The variables of winset and divided government are not related (r = .17). Chapter 6: Pre-Floor Agenda Block 1. Interestingly, there are a few moderate empirical fi ndings about rewards and compensations toward senior members of the majority party (viz. Leviott and Poterba 1999), but not to the centrist members of the majority party (Schickler and Rich 1997). 2. Edwards and Barrett find that during the 83rd-104th Congresses, presidential initiatives in unified government accounted for 57percent of the floor agendas while the same in divided government accounted for 24percent. In contrast, Sinclair (2000) observes that a significant portion of the floor agendas are presidential agendas regardless of which party controls Congress. Whereas Edwards and Barrett identify the presidential initiatives through the CQ Almanac, New York Times, and Washington Post, Sinclair measures the presidential agendas mainly from State of the Union addresses and other significant presidential messages. 3. Washington Post, http://projects.washingtonpost.com/congress/members/ p000197/key-votes/. 4. A symmetric, single-peaked preference is assumed. 5. Edwards, Barrett, and Peake (1997) find that a substantial amount of legislation has failed because of threats of vetoes. Mayhew observes that many bills die because they fail to gain support from the necessary broad majorities (2005, 216). 6. Including HR, S, HJ Res., and SJ Res. 7. Political Institutions and Public Choice Program, Michigan State University (rohde@ msu.edu). 8. Binder (2003, 37) explains that “Editorials, in short, capture issues at the ‘much talked about stage’ and issues that might be considered the ‘agenda of potential enactments.’” Binder codes the legislative content of each editorial that mentions Congress, the House, or the Senate and then uses the issues mentioned in those editorials to compile a list of agenda items for each Congress (2003, 37). 9. The augmented Dickey-Fuller test and Phillips-Perron test reject a hypothesis of unit root. 10. For the percentage of the legislative agendas that reach the floor, which is a ratio variable constructed from grouped data, I model variation with the weighted least squares logistic regression for grouped data (Glogit). The Glogit function accounts for the variation in the denominator for ratio variables.
Notes
133
11. In the last equation, the variable of Q–M is moderately significant with p = 1.3. 12. A majority roll in this situation would still be better than a majority roll after a discharge petition for the majority party. 13. For the dependent variables of the percentage of bipartisan vote in all final-passage votes and of the legislative agendas that reach the floor, which are ratio varia-bles constructed from grouped data, I model variation with the weighted least squares logistic regression for grouped data (Glogit). The Glogit function accounts for the variation in the denominator for ratio variables. The OLS results are in Appendix C. 14. During the examined Congresses, the measured status quo has always been less extreme than the reflection points. 15. Cox and McCubbins maintain that approximately half of the majority party rolls are “inconsequential rolls” of the measures, which the House majority leaders know ex ante that the Senate will eventually reject or the president will veto (2005, 106–23). However, this assumption is predicated upon a questionable and untested premise that the majority party contingent in the Senate and the president are more willing to pay the cost of rejecting legislation than the majority party in the House. Chapter 7: Conclusion 1. For example, the Gallup Opinion Poll conducted on October 20–22, 2006 shows that 52 percent of respondents think that the war with Iraq has made the United States less safe while 37 percent think the war has made the United States safer. Also, the Gallup Opinion Poll conducted in August 2007 reveals that 52 percent of respondents favor moving U.S. troops from Iraq to Afghanistan in order to fight Al Qaeda and Taliban terrorist operations, while 37 percent oppose it. 2. The Gallup Opinion Poll (June 20, 2007) reports that 24 percent of respondents say they are satisfied with the way things are going in this country while 74 percent are dissatisfied. 3. CNN.com. November 8, 2006. 4. CQ Weekly, March 5, 2007. 5. Ibid. 6. Ibid. 7. CQ Weekly, May 7, 2007.
This page intentionally left blank.
r e f e re n ce s
Abramowitz, Alan I. 1980. A comparison of voting for U.S. Senator and Representative in 1978. American Political Science Review 74: 633–40. ———. 1985. Economic conditions, presidential popularity, and voting behavior in midterm congressional elections. Journal of Politics 57: 31–43. ———, Albert D. Cover, and Helmut Norpoth. 1986. The president’s party in midterm elections. American Journal of Political Science 30: 562–76. ———, and Jeffery A. Segal. 1992. Senate elections. Ann Arbor: University of Michigan Press. ———, and Kyle L. Saunders. Ideological realignment in the U.S. electorate. Journal of Politics 61: 634–52. Aldrich, John H. 1995. Why parties? The origin and transformation of party politics in America. Chicago: University of Chicago Press. ———, and David W. Rohde. 2000. The consequence of party organization in the House: The role of the majority and minority parties in conditional party government. In Polarized politics: Congress and the president in a partisan era, ed. Jon R. Bond and Richard Fleisher, 31–72. Washington, DC: CQ Press. Americans For Democratic Action. 2007. http://www.adaction.org/votingrecords. htm. American Political Science Association. 1950. Toward a more responsible party system, a report. New York: Rinehart. Bachrach, Peter, and Morton S. Baratz. 1970. Power and poverty. New York: Oxford University Press. Bender, Bruce. 1994. A reexamination of the principal-agent relationship in politics. Journal of Public Economics 53: 149–63. Beth, Richard. 1998. The discharge rule in the House. Congressional Research Service. Report no. 97-856. Binder, Sarah A. 1999. The dynamics of legislative gridlock, 1947–1996. American Political Science Review 93 (September): 519–33. ———. 2003. Stalemate: Causes and consequences of legislative gridlock. Washington, DC: Brookings Institution. Black, Duncan. 1958. The theory of committee and elections. Cambridge: Cambridge University Press. Bond, Jon R., and Richard Fleisher. 2000. Polarized politics: Congress and the president in a partisan era. Washington, DC: CQ Press. Brady, David, and Craig Volden. 1998. Revolving gridlock. Boulder: Westview.
135
136
References
———. 2006. Revolving gridlock. 2nd ed. Boulder: Westview. Brewer, Mark D. 2005. The rise of partisanship and the expansion of partisan conflict within the American electorate. Political Research Quarterly 58: 219–29. Bullock, Charles S. III, and David W. Brady. 1983. Party, constituency, and roll-call voting in the U.S. Senate. Legislative Studies Quarterly 8: 29–44. Cain, Bruce, John Ferejohn, and Morris Fiorina. 1987. The personal vote. Cambridge: Harvard University Press. Cameron, Charles M. 2000. Veto bargaining. New York: Cambridge University Press. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American voter. New York: John Wiley and Sons. Campbell, James E. 1985. Explaining presidential losses in midterm congressional elections. Journal of Politics 47: 1140–57. ———. 1991. The presidential surge and its midterm decline in congressional elections. Journal of Politics 53: 477–87. Carmines, Edward G., Steven H. Renten, and James A. Stimson. 1984. Events and alignments: The party image link. In Controversies in voting behavior, 2nd ed., ed. Richard G. Niemi and Herbert F. Weisberg, 545–60. Washington, DC: CQ Press. Carmines, Edward G., and James A. Stimson. 1989. Issue evolution: Race and the transformation of American politics. Princeton: Princeton University Press. Carpini, Delli, and Scott Ketter. 1996. What Americans know about politics and why it matters. New Haven: Yale University Press. Carson, Richard D., and Joe A. Oppenheimer. 1984. A method of estimating the personal ideology of political representatives. American Political Science Review 78: 163–78. Chiou, Fang-Yi, and Lawrence S. Rothenberg. 2003. When pivotal politics meets partisan politics. American Journal of Political Science 47 (July): 503–22. Coates, Dennis, and Michael Munger. 1995. Legislative voting and the e$conomic theory of politics. Southern Economic Journal 61: 861–72. Cobb, Roger W., and Charles D. Elder. 1983. Participation in American politics: The dynamics of agenda-building. 2nd ed. Baltimore: John Hopkins University Press. Coleman, John J. 1999. Unified government, divided government, and party responsiveness. American Political Science Review 93 (4): 821–35. Cooper, Joseph, and David W. Brady. 1981. Institutional context and leadership style: The House from Cannon to Rayburn. American Political Science Review 75: 411–25. Copeland, Gary W. 1983. When congress and the president collide: Why presidents veto legislation. Journal of Politics 45: 696–710. Covington, Cary. 1988. Building presidential coalitions among cross-pressured members of Congress.” Western Political Quarterly 41: 47–62. Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan. Berkeley: University of California Press.
References
137
———. 2002. Agenda power in the U.S. House of Representatives, 1877 to 1986. In Party, process, and political change, ed. David Brady and Mathew McCubbins. Stanford: Stanford University Press. ———. 2005. Setting the Agenda. New York: Cambridge University Press. Cutler, Lloyd. 1988. Some reflections about divided government. Presidential Studies Quarterly 17 (Summer): 485–92. Davis, Michael L. and Philip K. Porter. 1989. A test for pure or apparent ideology in congressional voting. Public Choice 60: 101–11. Deen, Rebecca E. and Laura W. Arnold. 2002. Veto threats as a policy tool: When to threaten? Presidential Studies Quarterly 32: 30–45. Edwards, George C. III, Andrew Barrett, and Jeffrey Peake. 1997. The legislative impact of divided government. American Journal of Political Science 41 (April): 545–63. Edwards, George C. III, and Andrew Barrett. 2000. Presidential agenda setting in Congress. In Polarized politics: Congress and the president in a partisan era, ed. Jon R. Bond and Richard Fleisher, 109–33. Washington, DC: CQ Press. Enelow, James M., and Rebecca B. Morton. 1993. Promising directions in public choice. Public Choice 77: 85–93. Erikson, Robert S., and Gerald C. Wright. 2005. Voters, candidates, and issues in congressional elections. In Congress reconsidered, ed. Bruce I. Oppenheimer and Lawrence C. Dodd. Washington, DC: CQ Press. Fenno, Richard F. 1978. Home style. Boston: Little, Brown. Fiorina, Morris P. 1978. Congress: Keystone of the Washington establishment. New Haven: Yale University Press. ———. 1981. Retrospective voting in American national elections. New Haven: Yale University Press ———. 1996. Divided government. 2d ed. Boston: Allyn and Bacon. Frymer, Paul. 1994. Ideological consensus within divided government. Political Science Quarterly 109 (Summer): 287–311. Gerber, Alan, and Donald Green. 2000. The effects of canvassing, telephone calls, and direct mail on voter turnout. American Political Science Review 94: 653–63. Groseclose, Tim, and James M. Snyder. 2003. Interpreting the coefficient of party influence: Comment on Krehbiel. Political Analysis 11: 104–107. Groseclose and King. 2004. Committee theories reconsidered. In Congress reconsidered, ed. Lawrence C. Dodd and Bruce I. Oppenheimer, 191–216. Washington, DC: CQ Press. Gilmour, John B. 2002. Institutional and individual influences on the president’s veto. Journal of Politics 64: 198–218. ———. 2007. Sequential veto bargaining and blame game politics as explanations of presidential vetoes. Manuscript. Hammond, Thomas H., and Gary J. Miller. 1987. The core of the Constitution. American Political Science Review 81 (December): 1115–74.
138
References
Heitshusen, Valerie, and Garry Young. 2006. Macropolitics and changes in the U.S.Code. In The macropolitics of Congress, ed. E. Scott Adler and John S. Lapinski, 129–50. Princeton: Princeton University Press. Herrnson, Paul S. 2004. Congressional elections. Washington, DC: CQ Press. Hetherington, Marc J. 2001. Resurgent mass partisanship: The role of elite polarization. American Political Science Review 95: 619–31. Hinckley, Barbara. 1985. Problems of the presidency. Glenview: Scott, Foresman. Hoff, Samuel B. 1992. Presidential support and veto overrides. Midsouth Journal of Political Science 13: 173–90. Howell, William, Scott Adler, Charles Cameron, and Charles Riemann. 2000. Divided government and the legislative productivity of Congress, 1945–1994. Legislative Studies Quarterly 25, no. 2: 285–312. Jacobson, Gary C. 1990. The electoral origins of divided government. Boulder: Westview. ———. 2000. Party polarization in national politics: The electoral connection. In Polarized politics: Congress and the president in a partisan era, ed. Jon R. Bond and Richard Fleisher, 9–30. Washington, DC: CQ Press. ———. 2004. The politics of congressional election, 6th ed. New York: Longman. Joint Committee on the Organization of Congress, 103rd Congress 1st Session, January 26, 1993, 50. Jones, Charles O. 1995. It ain’t broke. In Back to gridlock? ed. James L. Sundquist. Washington, DC: Brookings Institution Press. Jones, David R. 2001. Party polarization and legislative gridlock. Political Research Quarterly 54 (March): 125–41. Kalt, Joseph P., and Mark Zupan 1984. Capture and ideology in the economic theory of politics. American Economic Review 74: 279–300. ———. 1990. The apparent ideological behavior of legislators. Journal of Laws and Economics 33: 103–31. Kazee, Thomas A. 1994. The emergence of congressional candidates. In Who runs for Congress, ed. Thomas A. Kazee. Washington, DC: CQ Press. Kelly, Sean Q. 1993. Divided we govern? A reassessment. Polity 25 (Spring): 475– 84. Kernell, Samuel. 1991. Facing an opposition Congress: The president’s strategic circumstance. In The politics of divided government, ed. Gary W. Cox and Samuel Kernell, 87–112. Boulder,: Westview. Key. V.O. Jr. 1942. Politics, parties, and pressure groups. New York: Crowell. Key, V. O. 1966. The responsible electorate. Cambridge: Harvard University Press. King, David C., and Richard Zeckhauser. 2003. Congressional vote options. Legislative Studies Quarterly 28: 387–411. Kingdon, John W. 1989. Congressmen’s voting decisions, 3rd ed. Ann Arbor: University of Michigan Press. ———. 1995. Agendas, alternatives, and public policies, 2nd ed. New York: Harper Collins.
References
139
Kiwiet, D. Roderick, and Mathew D. McCubbins. 1988. Presidential influence on congressional appropriations decisions. American Journal of Political Science 32: 713–36. Krehbiel, Keith. 1996. Institutional and partisan sources of gridlock: A theory of divided and unified government. Journal of Theoretical Politics 8 (January): 7–40. ———. 1998. Pivotal politics. Chicago: University of Chicago Press. ———. 2006. Pivots. In Handbook of political economy, ed. Barry R. Weigast and Donald Wittman. New York: Oxford University Press. ———. 2007. Partisan roll rates in a nonpartisan legislature. Journal of Law, Economics and Organization 23, no.1: 1–23. Lowell, A. Lawrence. 1901. The influence of party upon legislation in England and America. Report of the American Historical Association, 319–542. Lukes, Steven. 1974. Power: A radical view. New York: Macmillan. Langbein, Laura I. 1993. Shirking and ideology: Defense in the Senate. American University. Mimeo. Layman, Geoffrey C., and Thomas M. Carsey. 2002. Party polarization and ‘conflict extension’ in the American electorate. American Journal of Political Science 46: 786–802. Lee, Jong R. 1975. Presidential vetoes from Washington to Nixon. Journal of Politics 37: 523–46. Lewis-Beck, Michael S., and Tom W. Rice. 1992. Forecasting elections. Washington, DC: CQ Press. Maisel, L. Sandy, Linda L. Fowler, Ruth S. Jones, and Walter J. Stone. 1994. Nomination politics: The roles of institutional, contextual, and personal variables. In The parties respond: Changes in the American party system, 2nd ed., ed. L. Sandy Maisel. Boulder: Westview. Mayhew, David R. 1974. Congress: The electoral connection. New Haven: Yale University Press. ———. 1991. Divided we govern: Party control, lawmaking, and investigations, 1946–1990. New Haven: Yale University Press. ———. 2005. Divided we govern: Party control, lawmaking, and investigations, 1946–2002. 2nd ed. New Haven: Yale University Press. McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2006. Polarized America. Cambridge: The MIT Press. Miller, Warren E., and Donald E. Stokes. 1963. Constituency influence in Congress. American Political Science Review 57: 45–56. Morrow, James D. 1994. Game theory for political scientists. Princeton: Princeton University Press. Morton, Rebbeca B. 1999. Methods and models. New York: Cambridge University Press. Nagal, Jack. 1975. The descriptive analysis of power. New Haven: Yale University Press. Neustadt, Richard. 1960. Presidential power. New York: John Wiley.
140
References
Niou, Emerson M. S., and Peter C. Ordeshook. 1999. Return of Luddites. International Security 24, no.2: 84–96. Oleszek, Walter J. 2004. Congressional procedures and the policy process. Washington, DC: CQ Press. Oppenheimer, Bruce I., James A. Stimson, and Richard W. Waterman. 1986. Interpreting U.S. congressional elections: The exposure thesis. Legislative Studies Quarterly 11: 227–47. Ornstein, Norman J., Thomas E. Mann, and Michael J. Malbin. 200*. Vital Statistics on Congress. Washington, DC: Brookings Institution Press. Page, Benjamin I., and Robert Y. Shapiro. 1992. Rational public. Chicago: University of Chicago Press. Palfrey, Thomas R. 1991. Introduction. In Laboratory research in political economy, ed. Thomas R. Palfrey. Ann Arbor: University of Michigan Press. Poole, Keith T. 2003. Changing minds? Not in Congress. http://pooleandrosenthal. com/chminds.pdf. ———. 2005. Spatial models of parliamentary voting. New York: Cambridge University Press. ———, and Howard Rosenthal. 1997. Congress: A political economic history of roll call voting. New York: Oxford University Press. ———. 1984. The polarization of American politics. Journal of Politics 46:1061– 79. Powell, Lynda W. 1982. Issue representation in Congress. Journal of Politics 44: 658–78. Quirk, Paul J., and Bruce Nesmith. 1994. Explaining deadlock: Domestic policy making under the Bush presidency. In New perspectives on American politics, ed. Lawrence C. Dodd and Calvin Jillson, 191–211. Washington, DC: CQ Press. ———. 1998. Divided government and policy making: Negotiating the laws. In The Presidency and the Political System, ed. Michael Nelson, 565–88. Washington, DC: CQ Press. Riker, William H. 1992. The justification of bicameralism. International Political Science Review 13, no. 1: 101–16. Ripley, Randall. 1969. Majority party leadership in Congress. Boston: Little Brown. Rohde, David W. 1991. Parties and leaders in the postreform House. Chicago: University of Chicago Press. ———, and Dennis M. Simon. 1985. Presidential vetoes and congressional response: A study of institutional conflict. American Journal of Political Science 29: 397–427. Saeki, Manabu. 2004. Override propensity in the U.S. Congress: Veto challenges and override vote by the two chambers. Journal of Legislative Studies, 10, no. 4: 70–83. Schattschneider, E. E. 1942. Party government. New York: Farrar and Rinehart. Serra, George, and Albert Cover. 1992. The electoral consequences of perquisite use: The CaseWork case. Legislative Studies Quarterly 17: 233–46.
References
141
Shapiro, Catherine R., David W. Brady, Richard A. Brody, and John A. Ferejohn. 1990. Linking constituency opinion and Senate voting scores: A hybrid explanation. Legislative Studies Quarterly 15: 559–621. Shull, Steven A., and Thomas C. Shaw. 1999. Explaining congressional-presidential relations: A multiple perspectives approach. Albany: State University of New York Press. Simonton, Dean Keith. 1987. Presidential inflexibility and veto behavior. Journal of Personality 55: 1–18. Sinclair, Barbara. 1989. Leadership strategies in the modern Congress. In Congressional politics, ed. Christopher J. Deering. Chicago: Dorsey Press. ———. 2000. Hostile partners. In Polarized politics: Congress and the president in a partisan era, ed. Jon R. Bond and Richard Fleisher, 134–53. Washington, DC: CQ Press. ———. 2002. Do parties matter? In Party, process, and political change, ed. David Brady and Mathew McCubbins. Stanford: Stanford University Press. ———. 2006. Party wars. Norman: University of Oklahoma Press. ———. 2007. Unorthodox lawmaking. Washington, DC: CQ Press. Snyder, James M., and Tim Groseclose. 2000. Estimating party influence in congressional roll call voting. American Journal of Political Science 44, no. 2: 193–211. Stimson, James A. 1991. Public opinion in America: Moods, cycles, and swings. Boulder: Westview. ———. 2007. http://www.unc.edu/~jstimson/time.html. Sundquist, James. 1980. The crisis of competence in our national government. Political Science Quarterly 95, no. 2: 183–208. ———. 1988. Needed: A political theory for the new era of coalition government in the United States. Political Science Quarterly 103 (Winter): 613–35. ———. 1992. Constitutional reform and effective government. Washington, DC: Brookings Institution. Squire, Peverill. 1992. Challenger quality and voting behavior. Legislative Studies Quarterly 17: 247–63. Tsebelis, George. 2002. Veto players. Princeton: Princeton University Press. ———, and Jeanette Money. 1997. Bicameralism. New York: Cambridge University Press. Tufte, Edward R. 1975. Determinants of the outcomes of midterm congressional elections. American Political Science Review 69: 812–26. Turner, Julius. 1951. Party and constituency: Pressure on Congress. Baltimore: Johns Hopkins University Press. ———, and Edward V. Schneier. 1970. Party and constituency: Pressure on Congress, rev. ed. Baltimore: Johns Hopkins University Press. Uslaner Eric M. 1999. The movers and the shirkers: Representatives and ideologues in the Senate. Ann Arbor: University of Michigan Press. ———, and M. Margaret Conway. 1985. The responsible electorate: Watergate, the economy, and vote choice in 1974. American Political Science Review 79: 788–803.
142
References
Walt, Stephen M. 1999. Rigor or Rigor Mortis? Rational choice and security studies. International Security 23, no.4: 5–48. Wilson, Woodrow. 1885. Congressional government. Boston: Houghton Mifflin. Woolley, John T. 1991. Institutions, the election cycle, and the presidential veto. American Journal of Political Science 35: 279–304. Wright, Gerald C., and Michael B. Berkman. 1986. Candidates and policy in Unites States Senate elections. American Political Science Review 80: 567–88. Wright, Matthew. 1993. Shirking and political support in the U.S. Senate, 1964– 1984. Public Choice 76: 103–23. Zaller, John R. 1992. The nature and origins of mass opinion. New York: Cambridge University Press.
index
ADA scores, 30, 63 Adler, Scott, 2, 20, 27, 61 Aldrich, John H., 10, 40, 20 American Political Science Association, 119 Arnold, Laura W., 14, 47 Bachrach, Peter, 47, 113 Baratz, Morton S., 47, 113 Barrett, Andrew, 9, 14, 21–26, 33, 101 Bender, Bruce, 39 Berkman, Michael B., 37 Beth, Richard, 100 Binder, Sarah A., 22–26, 107–108 Black, Duncan, 44, 102 Bond, Jon R., 10, 27, 40, 99, 120 Brady, David, 14–15, 27, 36, 39, 48–50, 120, 123 Brewer, Mark D., 130 Brody, Richard A., 39 Bullock, Charles S., III, 39 Bush, George W., 1–8, 62, 126–28 Cain, Bruce, 38 Cameron, Charles M., 2, 14, 20, 27, 46–47, 61 Campbell, Angus, 37, 38 Carmines, Edward G., 130 Carpini, Delli, 37 Carsey, Thomas M., 130 Carson, Richard D., 39 cartel agenda model, 16, 53–54, 56–57 cartel block model, 105–106 cartel interval movement model, 54–56 Chiou, Fang-Yi, 15, 27, 72 Clinton, Bill, 1–2, 45, 61–62 Cloture, 35–36, 44, 125–26 Coates, Dennis, 39
Cobb, Roger W., 21, 107 Coleman, John J., 9, 19, 21, 28, 61, 72 conditional party government theory, 10 Congress, 109th, 2–7 Converse, Philip E., 37, 38 Conway, M. Margaret, 38 Cooper, Joseph, 120 Copeland, Gary W., 46 Cover, Albert D, 38 Covington, Cary., 99 Cox, Gary W., 10, 14, 16, 36, 47, 53–59, 62, 79, 95–109, 11–12, 121, 123 Cutler, Lloyd., 9, 61, 121 Davis, Michael L., 39 Deen, Rebecca E., 14, 47 divided government, 2, 8–10, 19–20, 23–26, 31–32, 70, 72–77, 91–93, 110–11, 114–15, 121 Edwards, George C., III, 9, 14, 21–26, 33, 101 Elder, Charles D., 21, 107 Enelow, James M., 35 Erikson, Robert S., 39 Fenno, Richard F., 39 Ferejohn, John, 38 filibuster, 44–46 Fiorina, Morris P., 1, 9, 20, 38, 61, 99, 121 Fleisher, Richard, 10, 27, 40, 99, 120 Fowler, Linda L., 37 Frymer, Paul, 83 Gerber, Alan, 111 Gilmour, John B., 46–47 Green, Donald, 111
143
144
Index
Groseclose, Tim, 99–101, 111 guest worker program, 3–5
Morton, Rebbeca B., 35, 81 Munger, Michael, 39
Hammond, Thomas H., 15, 27 Heitshusen, Valerie, 36 Herrnson, Paul S., 37–38 Hetherington, Marc J., 40 Hinckley, Barbara. 14–47 Hoff, Samuel B., 46 Howell, William, 2, 20, 27, 61
Nagal, Jack, 47 Nesmith, Bruce, 28 Neustadt, Richard, 100 Niou, Emerson M. S., 35 Norpoth, Helmut, 38
Jacobson, Gary C., 10, 38, 40, 111 Joint Committee on the Organization of Congress, 103rd Congress 1st Session, 45 Jones, Charles O., 14, 123 Jones, David R., 27 Jones, Ruth S., 37 Kalt, Joseph P., 39 Kazee, Thomas A., 37 Kelly, Sean Q., 19–25, 61 Kernell, Samuel, 15, 27, 28 Ketter, Scott, 37 Key, V. O., Jr., 9, 111 King, David C., 99, 100 Kingdon, John W., 99–100 Krehbiel, Keith, 15, 27, 36, 48–49, 103, 123 Langbein, Laura I., 39 Layman, Geoffrey C., 130 Lee, Jong R., 46 legislative productivity, 20–25 Lewis-Beck, Michael S., 38 Lowell, A. Lawrence, 119 Lukes, Steven, 113 Maisel, L. Sandy, 37 Malbin, Michael J., 11–13 Mann, Thomas E., 11–13 Mayhew, David R., 9, 13–15, 19–28, 39, 44 McCarty, Nolan, 36, 39, 40 McCubbins, Mathew D., 10, 14, 16, 36, 47, 53–59, 62, 79, 95–109, 11–12, 121 Median voter theory, 44 Miller, Gary J., 15, 27 Miller, Warren E., 37, 38 Money, Jeanette, 15, 27 Morrow, James D., 35
Oleszek, Walter J., 14, 45, 100 Oppenheimer, Bruce I., 38 Oppenheimer, Joe A., 39 Ordeshook, Peter C., 35 Ornstein, Norman J., 11–13 override, veto, 44–48, 124–25 Page, Benjamin I., 130 Palfrey, Thomas R., 35 party unity scores, 10–12 Patriot Act, reauthorization of, 5–7 Peake, Jeffrey, 19, 21, 33, 47, 61, 106 Pelosi, Nancy, 1, 19, 101, 126 pivotal block model, 106–107 pivotal interval movement model, 50–52 policy making, macro model of, 37 policy stability, and policy change, 26–34, 63–65, 85, 122–28 Poole, Keith T., 39–41, 43, 64, 82, 85–89 Porter, Philip K., 39 Powell, Lynda W., 39 presidential power, 100–101 Quirk, Paul J., 28 Renten, Steven H., 130 Residuum, 52, 55–56, 62–79 Rice, Tom W., 38 Riemann, Charles, 2, 20, 27, 61 Riker, William H., 15, 27 Ripley, Randall, 9 Rohde, David W., 14, 40, 46, 120 roll rates, 96–99 Rosenthal, Howard, 39–41, 43, 82, 85–89 Rothenberg, Lawrence S., 15, 27, 72 Saunders, Kyle L., 38 Schattschneider, E. E., 9, 28, 92, 120, 121 Schneier, Edward V., 120 second face of power, 47, 117
Index Segal, Jeffery A., 38 sequential veto bargaining model, 47 Serra, George, 38 Shapiro, Catherine R., 39 Shapiro, Robert Y., 130 Shaw, Thomas C., 46 Shull, Steven A., 46 Simon, Dennis M., 46 Simonton, Dean Keith, 46 Sinclair, Barbara, 14, 40, 45, 47, 99–101, 120, 132 Snyder, James M., 99–101, 111 social security reform, 3–5 Squire, Peverill, 37 Stimson, James, 38, 72, 86, 109, 130 Stokes, Donald E., 37, 38 Stokes, Miller, 39 Stone, Walter J., 37 Sundquist, James, 9, 28, 61, 121 supermajoritarianism, 44–48 tax cuts of 2001 and 2003, extension of, 7–8
145
Tsebelis, George, 15, 27, 82, 123 Tufte, Edward R., 38 Turner, Julius, 120 Uslaner, Eric M., 38, 39 veto players model, 82–85 veto threats, 47 veto, presidential, 46–48 Volden, Craig, 14–15, 27, 36, 48–50, 123 Walt, Stephen M., 35 Waterman, Richard W., 38 Wilson, Woodrow, 9, 46 winset, 84–86 Wright, Gerald C, 37, 39 Wright, Matthew, 39 Young, Garry, 36 Zaller, John R., 130 Zeckhauser, Richard, 99, 100 Zupan, Mark, 39
POLITICAL SCIENCE
“This work is a smart and interesting contribution to the study of congressional policymaking.” — David R. Mayhew,Yale University
. State University of New Yor k Press www.sunypress.edu