Journal of
ISSN 1463-578X
Property Investment & Finance
Volume 22 Number 6 2004
The Gerald Brown memorial issue Edit...
19 downloads
588 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
Journal of
ISSN 1463-578X
Property Investment & Finance
Volume 22 Number 6 2004
The Gerald Brown memorial issue Editor Nick French
Access this journal online __________________________ 451 Editorial advisory board ___________________________ 452 Abstracts and keywords ___________________________ 453 Editorial ___________________________________________________ 457 Equilibrium time on the market (ETOM) for commercial real estate in the UK
Gerald R. Brown and Tien Foo Sing ________________________________
How should unsmoothing affect pension plan asset allocation?
Philip Booth and George Matysiak__________________________________
The uncertainty of valuation
Nick French and Laura Gabrielli ___________________________________
Different risk measures: different portfolio compositions?
Peter Byrne and Stephen Lee ______________________________________
Access this journal electronically The current and past volumes of this journal are available at:
www.emeraldinsight.com/ISSN 1463-578X.htm You can also search over 100 additional Emerald journals in Emerald Fulltext at:
www.emeraldinsight.com/ft
See page following contents for full details of what your access includes.
458
472 484
501
CONTENTS
CONTENTS continued
The level of direct property in Hong Kong property company performance
Graeme Newell, Chau Kwong Wing and Wong Siu Kei_________________
PRACTICE BRIEFING The valuation of specialised property: a review of valuation methods
Nick French ____________________________________________________
512
533
Book review _______________________________________________ 542
JPIF 22,6
452
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 p. 452. # Emerald Group Publishing Limited 1463-578X
EDITORIAL ADVISORY BOARD Professor Alastair Adair University of Ulster, Northern Ireland Dr Andrew Adams The University of Edinburgh, UK Dr Richard Barkham Grosvenor Estate, London, UK Professor Andrew Baum The University of Reading, UK Professor Terry Boyd Queensland University of Technology, Australia Peter Byrne The University of Reading, UK Professor Kwong Wing Chau University of Hong Kong, Hong Kong Professor Neil Crosby The University of Reading, UK Eamonn D'Arcy The University of Reading, UK Professor Julian Diaz III Georgia State University, Atlanta, USA Dr Tim Dixon College of Estate Management, Reading, UK Professor Paul Gallimore Nottingham Trent University, UK Professor Robin Goodchild LaSalle Investment Management, London, UK Dr Liow Kim Hiang National University of Singapore, Singapore Professor Martin E. Hoesli University of Geneva, Switzerland and University of Aberdeen, UK Sandra Jones BH2, London, UK Angelo Karantonis University of Technology, Sydney, Australia Geoff Keogh University of Aberdeen, UK Tony Key Investment Property Databank, London, UK Professor Colin Lizieri The University of Reading, UK Professor Bryan MacGregor University of Aberdeen, UK
George Matysiak CB Hillier Parker, London, UK Professor Norman Miller University of Cincinnati, USA Dr Yu Shi Ming National University of Singapore Dr Seow Eng Ong National University of Singapore Dr David Parker Cap Gemini Ernst & Young, Sydney, Australia Professor John Ratcliffe Dublin Institute of Technology, Eire Dr Tim Richards CoÂras Iompair Eireann, Dublin, Eire Professor Jon Robinson University of Melbourne, Australia Professor Stephen Roulac Roulac Global Places, USA and University of Ulster, Northern Ireland Patrick Rowland Curtin University, Perth, Australia Dr Ed Schuck Frank Russell Company, New Zealand Dr Karen Sieracki KASPAR Associates, Tunbridge Wells, UK Andrew Smith Aberdeen Property Investors, London, UK Professor Dogan Tirtiroglu Concordia University, Quebec, Canada Professor Ko Wang The Chinese University of Hong Kong Professor Charles W.R. Ward The University of Reading, UK Professor James R. Webb Cleveland State University, USA Dr Larry Wofford C&L Systems Corp., Tulsa, USA Professor Elaine Worzala University of San Diego, USA Dr Peter Wyatt University of the West of England, Bristol, UK
Equilibrium time on the market (ETOM) for commercial real estate in the UK Gerald R. Brown and Tien Foo Sing Keywords Real estate, Time to market, Market economy, Asset valuation, Accuracy, United Kingdom Time on the market (TOM) has been widely tested in the US real estate literature using listing and selling data of houses captured in the multiple listing services (MLSs). Unfortunately in the UK there are no MLSs so it is not possible to undertake similar analyses. The approach adopted in this paper differs from traditional TOM analyses in that it focuses on the speed or time the market takes to correct for information differences between open market valuations and traded prices. In this context the paper introduces the concept of equilibrium time on the market (ETOM). The study therefore adopts a different approach to estimating TOM and in addition also examines the phenomenon within the UK commercial real estate sector. Based on a simple present value model, the time taken for the difference between an appraiser’s estimate of open market value and known selling prices define our time on the market under equilibrium market conditions. Using the annualised UK Investment Property Databank all-property total return index for a sample period of 17 years between 1983 and 1999, the average ETOM was estimated to be 8.4 months. This figure, however, varied and depended on market conditions.
How should unsmoothing affect pension plan asset allocation? Philip Booth and George Matysiak Keywords Pension funds management, Assets management, Resource allocation Examines the impact of using “unsmoothing” techniques on real estate data to take pension-plan asset-allocation decisions. It is generally believed that valuation-based real estate indices give rise to returns figures which are “smoothed” versions of the underlying transaction prices. Unsmoothing techniques can be used to develop real estate return data series that are believed to be a
more accurate representation of underlying transaction prices. If this is done, the resulting data reveal greater volatility of real estate returns. When such data are applied to portfolio selection models, they often reveal a reduced allocation to real estate in efficient portfolios. Looks at the impact of unsmoothing data when taking pension-plan asset-allocation decisions. Finds here that the unsmoothed data are more closely correlated with pension plan liabilities. As a result, efficient pension plan portfolios sometimes contain more real estate, rather than less. In general, there is little change in the efficient real estate allocation. These results are very important. They reveal that so-called “valuation smoothing” may distort property investment decisions less than is commonly thought.
The uncertainty of valuation Nick French and Laura Gabrielli Keywords Uncertainty management, Market value, Asset valuation, Property, United Kingdom Valuation is often said to be “an art not a science” but this relates to the techniques employed to calculate value not to the underlying concept itself. Valuation is the process of estimating price in the market place. Yet, such an estimation will be affected by uncertainties. These input uncertainties will translate into an uncertainty with the output figure, the valuation. The degree of the uncertainties will vary according to the level of market activity; the more active a market, the more credence will be given to the input information. In the UK at the moment the Royal Institution of Chartered Surveyors (RICS) is considering ways in which the uncertainty of the valuation can be conveyed to the use of the valuation, but as yet no definitive view has been taken apart from a single Guidance Note (GN5). One of the major problems is that valuation models (in the UK) are based on comparable information and rely on single inputs. They are not probability-based, yet uncertainty is probability driven. This paper discusses the issues underlying uncertainty in valuations
Abstracts and keywords
453
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 Abstracts and keywords q Emerald Group Publishing Limited 1463-578X
JPIF 22,6
454
and suggests a probability-based model (using Crystal Ball) to address the shortcomings of the current model.
Different risk measures: different portfolio compositions? Peter Byrne and Stephen Lee Keywords Risk analysis, Modelling, Portfolio investment Traditionally, the measure of risk used in portfolio optimisation models is the variance. However, alternative measures of risk have many theoretical and practical advantages and it is peculiar therefore that they are not used more frequently. This may be because of the difficulty in deciding which measure of risk is best and any attempt to compare different risk measures may be a futile exercise until a common risk measure can be identified. To overcome this, another approach is considered, comparing the portfolio holdings produced by different risk measures, rather than the risk return trade-off. In this way we can see whether the risk measures used produce asset allocations that are essentially the same or very different. The results indicate that the portfolio compositions produced by different risk measures vary quite markedly from measure to measure. These findings have a practical consequence for the investor or fund manager because they suggest that the choice of model depends very much on the individual’s attitude to risk rather than any theoretical and/or practical advantages of one model over another.
The level of direct property in Hong Kong property company performance Graeme Newell, Chau Kwong Wing and Wong Siu Kei Keywords Management styles, Hong Kong, Property management, Performance appraisal Hong Kong is one of the most dynamic property markets in the world, and now provides the
economic gateway to China. Using style analysis, the level of direct property in Hong Kong property company performance is shown to be approximately 15 per cent over 1984-2000, with the level of direct property increasing to approximately 25 per cent in recent years. The level of direct property in Hong Kong property company performance is significantly below that seen for the USA, Europe and Australia. This highlights a number of key strategic property investment issues over 1984-2000, relating to the level of direct property in Hong Kong property company performance. Also assesses the level of direct property at the individual property company level, as well as the property company sector level, further emphasising the strategic role of Hong Kong property companies in an investment portfolio. This research complements the previous research by Brown and Chau on excess returns in the Hong Kong property market, as well as highlighting the issues and role of both direct and indirect property for inclusion in diversified investment portfolios; these being key areas of Gerald Brown’s extensive property research agenda.
The valuation of specialised property: a review of valuation methods Nick French Keywords Asset valuation, Property marketing Provides a brief overview of the methods that used in real estate valuation with a particular emphasis on the valuation of specialised property. Proposes that the underlying requirement is to estimate market value and that the role of the valuer is to choose the method that is the best model to achieve this objective. Concludes that a valuer must work with the recognised techniques and, in the case of specialised property, these are methods that go back to analysing value from first principles by identifying the value of the property to the business.
The Gerald Brown memorial issue This issue is respectively dedicated to the memory of Gerald Brown (co-editor 1999-2002)
Editorial Dedicated to the memory of Professor Gerald Brown As many of you will know, Gerald Brown died in May 2002 following a heroic, and always good spirited, fight against cancer. Gerald was the co-editor of this journal from 1999 and he was instrumental in steering the editorial policy during his short tenure. It was an honour to work with Gerald. This issue is a small, and probably inadequate, attempt by those who worked with him to pay tribute to a special friend, mentor, supervisor and colleague. Gerald was a very dignified British academic with an eclectic background and blessed vision of what could be achieved with effort and dedication. Some of us knew Gerald from his time at City University in London, some from when he was Professor at Salford University, and some from his time abroad in Auckland and Singapore. No matter the point of first reference, we all share a genuine and sincere love for a man who touched our lives with kindness and humour. Gerald was one of the world’s celebrated property academics. Professor Austin Jaffe described him as “a brilliant and serious thinker”. Gerald’s career as an academic leader peaked when he was appointed Professor of Real Estate Investment and Finance and Director of Real Estate Research at the National University of Singapore. He had by this time undertaken a number of major consultancy projects and published numerous papers in international academic and professional journals. He was a joint editor of the Journal of Property Investment & Finance, and on the editorial boards of many other internationally refereed academic journals. He wrote two books. The first, Property Investment and Capital Markets, was seminal in real estate as it stretched and influenced many academics and practitioners in their thinking about property as an investment. This was followed by a second book, co-authored by George Matysiak, called Real Estate Investment – A Capital Market Approach. This book has become essential reading for all involved in property investment and finance. Professor Charles Ward stated “it is rare for authors who write advanced material in investment to be so considerate to the reader. The text is clear and explanations are comprehensive – a model of pedagogical clarity”. Professor Graeme Newell described the book as “one of the most rigorous and comprehensive texts available in property investment. It is an excellent text that will be in regular use throughout undergraduate, postgraduate and professional careers of future property professionals”. Apart from being such an outstanding scholar, Gerald was also someone who touched all his colleagues with his wit and sense of humour. Over and above all of this Gerald will be remembered by all his colleagues as a genuinely nice person, a great mate, a kind and warm human being and a gentle soul. In this issue a number of his colleagues have chosen to publish their work in his memory. The subjects are all close to Gerald’s heart and we hope that in some small way that this issue is a fitting tribute to the legacy that Gerald has left behind. Nick French The University of Reading, Reading, UK Deborah Levy The University of Auckland, Auckland, New Zealand
Editorial
457
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 p. 457 q Emerald Group Publishing Limited 1463-578X
The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-578X.htm
JPIF 22,6
Equilibrium time on the market (ETOM) for commercial real estate in the UK
458
Gerald R. Brown and Tien Foo Sing
Received July 2001 Accepted February 2003
Department of Real Estate, School of Design and Environment, National University of Singapore, Singapore Keywords Real estate, Time to market, Market economy, Asset valuation, Accuracy, United Kingdom Abstract Time on the market (TOM) has been widely tested in the US real estate literature using listing and selling data of houses captured in the multiple listing services (MLSs). Unfortunately in the UK there are no MLSs so it is not possible to undertake similar analyses. The approach adopted in this paper differs from traditional TOM analyses in that it focuses on the speed or time the market takes to correct for information differences between open market valuations and traded prices. In this context the paper introduces the concept of equilibrium time on the market (ETOM). The study therefore adopts a different approach to estimating TOM and in addition also examines the phenomenon within the UK commercial real estate sector. Based on a simple present value model, the time taken for the difference between an appraiser’s estimate of open market value and known selling prices define our time on the market under equilibrium market conditions. Using the annualised UK Investment Property Databank all-property total return index for a sample period of 17 years between 1983 and 1999, the average ETOM was estimated to be 8.4 months. This figure, however, varied and depended on market conditions.
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 458-471 q Emerald Group Publishing Limited 1463-578X DOI 10.1108/14635780410569452
I. Introduction Time on the market (TOM) studies have generally focused on the period between the time of listing and when transactions take place in the residential market. Many studies have examined the ex-post relationships between TOM and variables such as location, physical attributes, firm size, broker’s efforts and experience, listing and selling prices and seasonal characteristics. The results of these studies are varied and in some cases contradictory, although they do provide useful insights into some of the factors that can affect TOM. To date there have been no similar TOM studies published using UK data. The reason for this is due to the non-availability of residential listing data. By contrast the issue of valuations accuracy and the difference between valuations and transaction prices in the commercial market has attracted greater attention (Brown and Matysiak (2000a, Ch. 7), for a full discussion of the issues involved). We believe, therefore, that the difference between open market valuations and transaction prices should provide useful information on the duration of TOM, though not directly. In the USA, the ratio of selling price over listing price has been used as a determinant in analysing its impact on the TOM (Belkin et al., 1976; Janssen and Jobson, 1980; Kang and Gardner, 1989; This paper is dedicated to the late Professor Brown, who passed away in 2002. He contributed fully to the first draft of this paper.
Yavas and Yang, 1994). If the selling price is regarded as a proxy for the market-clearing price, and the listing price reflects seller’s expectations then the time it takes for a seller to adjust his or her expectations to the market-clearing price can be interpreted as the TOM, ceteris paribus. However, as neither of these figures is based on consensus views concerning expectations the results are likely to be heroic. In this context TOM studies can only tell us what factors are generally likely to improve salability. If, however, a homeowner has unrealistic expectations concerning a sales price then the traditional view of TOM has little meaning. In contrast to US listing services every UK commercial real estate property that is offered for sale is subject to an independent open market value (OMV). Professional appraisals are therefore estimated in accordance with international valuation standards and provide a consistent benchmark of equilibrium clearing market values. As part of this process professional appraisers must ignore any abnormal bids from special purchasers or any unusual circumstances and be prepared to sign off on an appraisal that captures the consensus view of the market at a particular point in time. By contrast the price at which a property is offered and sold may contain information that captures the negotiating strength between the parties involved in the transaction. There is likely to be a difference between open market values and transaction prices that will be time varying. Based on this rationale and using UK commercial real estate data, we argue that the time it takes for the offer prices to clear at their open market values can on average be regarded as a measure of equilibrium time on the market (ETOM). We have, therefore, introduced this term in this paper to distinguish it from traditional TOM studies. This study aims to integrate previous UK studies on appraisals and prices in an equilibrium market framework to estimate the ETOM. The analysis uses a simple present value model to estimate the time it takes for traded prices to reach their open market equilibrium values, discounted at a risk-adjusted rate of return. The ETOM is empirically analysed using the UK annual property indices compiled by the Investment Property Databank (IPD). Because we are using an equilibrium model for appraisals it will be clear that traded prices may either be above or below their respective open market values. This is not unusual and will depend on market conditions, but one logical implication of adopting an equilibrium approach is that our ETOM figures can either be positive or negative. The issue here is one of interpretation. If it is positive it represents the average time it takes for a commercial property to sell. If it is negative it represents search time. Under priced properties are unlikely to exist on the market for long periods so we would not expect persistence of negative ETOMs. This just reflects the way markets operate. Negative ETOMs are most likely to be observed when properties sell before they are listed. They capture the fact that some properties may be undervalued so that they are likely to attract some interest. There is however a cost associated with locating these properties, which would be reflected in search time. There are three motivations for this study. First, by using the IPD data, we are able to overcome the data limitations faced in the estimation of the time lag between listing and sales and are able to impose a formal model of equilibrium appraisal. Second, as the US brokerage multiple listing services (MLSs) are uncommon in the UK, compared with sole agency and private treaty, the typical TOM model is not be able to capture the evidence of sales that are not sold through broker firms or through any listings.
ETOM for commercial real estate 459
JPIF 22,6
460
Third, the heterogeneity and seasonal factors that affect the offer prices are effectively reflected in the IPD valuation index, and this avoids the simultaneity problems encountered in typical TOM models (Yavas and Yang, 1994). This study is organised into five sections. Section I provides an introduction and background review and objectives of our ETOM study. The literature review is included in Section II. Section III provides the conceptual framework for estimating the proposed ETOM based on a present value model. The IPD data are used for the empirical analysis in Section IV and the implications of the results are discussed. Section V concludes the findings. II. Literature review The empirical study of TOM has evolved rapidly into an important part of the real estate search theoretic literature which previously focused mainly on issues relating to listing prices, incentive structure and broker’s effort in the search process[1]. Most of these TOM papers are predominantly found in the USA because data on listing and sale dates are easily available and normally well recorded through the MLSs. In the UK and in other countries markets where brokerage search and for sale by owner (FSBO) arrangements are the more prevalent mode of selling houses, records of TOM are either not traceable, or if they are available, the actual date when a house is put up for sale may not be determinable. There were cases whereby the intent of sale and negotiation may be initiated way before the houses were put up for sale in the market following the failure of the parties to reach a consensus on the price. Therefore, the subsequent listing and engagement of a brokerage service by the owners may not be reflective of the actual commencement of search for the houses. Similarly, for cases where sales were concluded on agreeing on a market price during the initial negotiation stage, search through the public listing may not even be initiated. On the assumption that the TOM was empirically observable via the multiple listing services, the earlier TOM papers in the real estate literature invariably focused on explaining the how TOM varies with respect to house characteristics, brokerage and market factors. Miller (1978) empirically tested the effects of the TOM on the selling prices for houses sold in 1976 in Columbus, Ohio, and found a positive relationship between sale prices and marketing time[2]. Kang and Gardner (1989) reaffirmed Miller’s findings with the qualification that the positive TOM-selling price relationship was only valid in a high mortgage interest rate regime (. 13 per cent). When the mortgage rate was relatively low, sellers would not obtain a better offer if they rejected early offers and put the houses in the market for a longer time. Knight (2002), on the other hand, found that the positive price and TOM relationship was significant only when the revised listing price information prior to the sale was taken into account. By regressing the TOM variable in an opposite direction against the spread between listing and selling prices, Belkin et al. (1976) found that overpriced properties, as a result of misjudgement of the market conditions, stayed longer on the market. In a study by Jud et al. (1995) they find a positive, but statistically insignificant, price spread where TOM was defined as an inverse function of the probability of an accepted offer. The paper identified the relationship between the spread and distribution of information in the market. They used this as a measure of liquidity and argued that as the market becomes illiquid, it takes a longer time to sell a property, i.e. longer TOM.
Kang and Gardner (1989) also confirmed the findings of Belkin et al. (1976) when a percentage discount of the sale price (i.e. (listing price 2 selling price)/selling price) was used to proxy the price spread. The importance of the listing and selling prices in determining the TOM was also supported by the multivariate analysis conducted by Janssen and Jobson (1980), in Knight’s (2002) two-stage least squares (2SLS) model and also in Ong’s (2000) heterogeneity-adjusted duration model. Yang and Yavas (1995) disaggregated the sampled house price data into different price ranges in their 2SLS analysis and found that the overpricing effects on TOM was only significant for the mid-price houses. They showed that the overpricing did not increase TOM for low- and high-price ranged houses. Beside the listing and selling price variables, the TOM model has also been specified as a function of the physical characteristics of houses, seasonal variation, broker’s reward and effort as well as other brokerage firm related variables. Size has a positive effect on the house prices (Sirmans et al., 1991), but it was not a significant determinant of TOM (Kang and Gardner, 1989). The age of a property also has a positive and significant relationship on the TOM (Kang and Gardner, 1989; Ong, 2000). Marketing time is significantly shorter for newer homes. Ong (2000) also found that TOM was also negatively related with the floor level in his study that used sampled resale prices of public houses in Singapore. Flats on higher floors have significantly shorter TOM, and flats located in the eastern region of Singapore also have a significantly shorter TOM. On the broker’s incentive and effort factors, the commission rate was found to have a positive relationship on house prices, which also led to a longer TOM (Sirmans et al., 1991, Yang and Yavas, 1995). Properties that generate higher commissions take longer to sell. However, the commission split between the listing and selling agent in the MLS was found to have no significant effects on TOM (Sirmans et al., 1991; Yang and Yavas, 1995). These findings contradicted the game theoretical model proposed by Miceli (1991), which predicted an increased effort by brokers when the commission was split between listing and finding brokers. Larsen and Park (1989) further argue that the net effect of non-uniform percentage commission on TOM is ambiguous and depends on its influence on both the broker’s search effort. Beside the effort expended by brokers, Ong (2000) also looked at how the experience of a broker helped to improve the TOM. His empirical analysis using Singapore’s public resale house price data showed that experience without effort did not improve time on the market. The results contradicted those of Yang and Yavas (1995) who showed that experienced agents, who sold more houses than the average agent, were more able to reduce TOM. They, however, rejected the hypothesis that listing agents would expend more effort on their own listing, so that TOM of houses listed and sold by the same agent is shorter. The choice of a larger broker firm increases the probability of finding a match between the listing and sale of houses and thus cuts down TOM (Larsen and Park, 1989). Sirmans et al. (1991) further showed that it was not the size of the listing firms that mattered in shortening the TOM but the selling firm’s size that established a negative relationship with TOM. The results of Yang and Yavas (1995) were consistent with those of Sirmans et al. (1991). They reaffirmed that the negative relationship between the selling firm’s size and TOM was significant, but only for houses in the low-price range. Their results failed to reject the hypothesis that listing firm’s size does not affect TOM. This is consistent with the proposition that the MLS system serves to
ETOM for commercial real estate 461
JPIF 22,6
462
foster broker’s cooperation and minimise competition for listings (Miceli, 1991; Sirmans et al., 1991). Yang and Yavas (1995) also showed that an increase in the volume of listings by the listing agent has a corresponding increase in TOM. Seasonal effects on TOM were significant in the empirical results of Larsen and Park (1989) and Yang and Yavas (1995). They found that houses listed in summer would have shorter TOM. In the Singapore study by Ong (2000), he found that public flats could be sold faster in a declining private property market because of the increased propensity to upgrade to private flats. Exploring the TOM question from the demand side analysis, Baryla and Zumpano (1995) showed that both first-time buyers and out-of-town buyers take a longer time in the search process than more experienced local and knowledgeable buyers. They also found that buyers relocated by their employers have shorter TOM in the house search process. The earlier TOM studies that used single structural regression models (Miller, 1978; Belkin et al., 1976; Janssen and Jobson, 1980; Kang and Gardner, 1989) are affected by simultaneity in the relationship between selling or listing prices and TOM. The regression biases in the single equation regression models were explicitly addressed using 2SLS regression (Sirmans et al., 1991; Yang and Yavas, 1995; Knight, 2002) and using the duration model (Yang and Yavas, 1995; Baryla and Zumpano, 1995; Ong, 2000). These empirical TOM studies were also subject to several limitations. The analysis of TOM is assumed to be independent of the seller’s choice of employing a brokerage service. The results of the TOM studies are therefore not representatives of the “for sale by owner (FSBO)” cases (Yavas, 1994). The studies also excluded samples of houses that were listed but not sold, and the effects of the proportion of unsold listed houses on the TOM were not known empirically (Janssen and Jobson, 1980). Removal of sample observations where selling price exceeded listing price from the analysis also rule out the possibility of including search time as part of the TOM results (Knight, 2002). The empirical relationships between TOM and listing price, house features, and market forces were indeed not insignificant, and they have all along been recognised by valuers when determining an open market value for a house. Therefore, in our attempt to examine the adjustment process between the open market value and the market clearing price, we implicitly assume that the two prices do already capture the TOM-dependent house-specific and market factors. The time it takes for the dynamic price process to adjust the price gap towards equilibrium is, therefore in our study, referred to the ETOM. The ETOM, which is a dynamic measure of market clearing time for a house vis-a`-vis the static measure in the earlier TOM, does overcome some of the limitations highlighted earlier. This application of the ETOM concept could also eliminate the need to observe actual listing and selling times, and the study of TOM that were not possible for commercial real estate is now made possible with the ETOM model.
III. Conceptual framework TOM has been extensively studied in the US real estate literature, due partly to the availability of listing and selling price data in the MLS. The same advantages may not be available in other countries where listing and selling data are proprietary in nature.
Equilibrium price correction process Jud et al. (1995) develop an asset liquidity theory that relates the spread between listing and contract prices to the TOM. The larger the spread the longer the time for the listing prices to match the acceptance price of the seller. The search for a right offer is a sequential process in their theoretical model. The search cost increases as the listing period persists. The model also shows that large variations in the price spread, caused by asymmetry in market information between buyers and sellers, would encourage sellers to hold out for a high acceptance price, thus reducing the liquidity of the real estate market. Correction of this information inefficiency, which was not examined in the Jud et al. (1995) framework, will be modelled in an equilibrium market context in the following section. Inefficiencies in market information exist in the proposed equilibrium market framework, as a result of the discrepancies between the transaction costs and appraisal values during the search process. Within the UK context this paper examines the subject of TOM from an equilibrium market perspective. We define ETOM as the time it takes for sales information to be fully reflected into the open market appraisals of commercial property at which the market will clear. The value of each property is subject to an independent appraisal that is governed by internationally recognised valuation standards. The resulting appraisal is intended to represent an unbiased estimate of the prevailing underlying equilibrium market conditions assuming that the property is freely exposed to the market and is not subject to any abnormal bids. With large sample sizes open market appraisals should converge to their equilibrium values. Empirical studies in the UK have shown that this is indeed the case and under equilibrium conditions, open market appraisals appear to be a good proxy for traded prices (Brown and Matysiak, 2000a, b). No significant bias should therefore exist between appraisals and prices when buyers and sellers are assumed to have access to the same set of information. Short-term variations between appraisals and prices may, however, exist from time to time due to asymmetric information and a mismatch between the expectations of buyers and sellers. Over time these mismatches should disappear as a result of a market self-correction process. Appraisers should also continually respond to changes in market conditions and reflect this information in their estimates of open market value. The market should, therefore, always revert back to their equilibrium values in the long run, in accordance with the concepts of equilibrium asset pricing models[3]. The time it takes for the market to clear and for the variations between appraisals and market prices to converge is, in our case, taken to be the time on the market in equilibrium market conditions. Technically, we define the ETOM as an annualised compounding period (n), within which prices (Pt) are discounted by a risk-adjusted rate of return (r) in order to arrive at an equivalent equilibrium open market value (Vt). When the market is in equilibrium open market appraisals should be a good proxy for prices so that V t ¼ P t . However, due to imperfections or asymmetries in market information properties may be offered for sale at prices that do not reflect open market, equilibrium conditions. The internationally accepted definition of open market value represents the “best price at which the sale of an interest in property would have been completed unconditionally for cash consideration on the date of the valuation” (RICS, 1995). This definition is based on the notion of a willing seller; that there has been a reasonable period to market the property; that the state of the market using comparable information was the same as the date of the valuation; bids by a purchaser
ETOM for commercial real estate 463
JPIF 22,6
464
with a special interest were ignored and that both parties to the transaction acted without compulsion. If, on average, appraisers follow this definition they will try to arrive at estimates of value that represent equilibrium market conditions. If open market appraisals and prices are a good proxy for each other then the market should clear and the ETOM should be equal to zero. If, however, sellers have unrealistic expectations of the price they believe they could achieve for a property it is likely to remain on the market for very long periods. No amount of good marketing or presentation will influence the TOM in its traditional sense, as it becomes an issue of equilibrium. There are many anecdotal examples of this occurring in both the residential and commercial property markets particularly after a collapse in the market. For example, in the late 1980s the property market in the UK was very depressed. Some sellers kept their asking prices at pre-collapse levels and found that it took almost two years before a sale was achieved. This is a behavioural issue that is not fully captured in the traditional studies. Estimation of ETOM The estimation of an open market appraisal is not a static process. Appraisers respond to new market information and alter their estimates of open market value on a regular basis. Convergence takes place in the long run at a stage where the change in open market value matches the expected market prices of a property. In abnormal conditions sellers can revise their price expectations so that they are in line with the open market appraisal or they will have to wait until the market changes so that open market appraisals and prices begin to converge. This process can be modelled in a present value framework using a time varying discount rate (rt)[4], where the expected open market value (Vt) can be equated to the discounted expected price (Pt) as follows: E ðV t Þ ¼
E ðP t Þ : ½1 þ E ðr t Þn
ð1Þ
This model is drafted in terms of expectations because open market appraials are expressed in this form and prices represent what sellers expect to achieve. By rearranging the ETOM can be found by solving for “n”. Equilibrium TOM ¼ n ¼
logE ðP Þt logE ðV Þt : log½1 þ E ðr Þ
ð2Þ
For illustration purposes, assume that a property has an open market appraisal at the beginning of the year of £1,000,000. During the year, the market collapses such that its “true” open market value declines to £500,000, which is equivalent to a 50 per cent drop in value. The property is offered for sale but the owner, although acknowledging that the market has fallen, wishes to get as close as possible to the original open market value. The asking price of the property is, therefore, set at £700,000 representing only a 30 per cent drop in value. If the property market is reasonably efficient, it would respond to the difference between the revised open market value and the asking price by altering the period it takes to sell the property. The asking price of £700,000 and the “true” open market value of £500,000 must theoretically be equal in economic terms at some point in the future. Assuming that the appropriate return on the property is 18
per cent, the time it takes for the property to “sell” at the reported asking price of £700,000 can be estimated from equation (2) as follows: log £700; 000 log £500; 000 ¼ 2 years: n¼ logð1:18Þ
ETOM for commercial real estate
As shown above, the proposed ETOM model is able to capture the variation between expected asking prices and open market values, and translate this into an annualised period by which the price gap would be cleared by the market. This ETOM estimate is, however, derived on the assumption that the market is stable and there is zero growth with respect to the market clearing price. If different expected growth rate of the market price, E(g), is allowed in equation (1), such that prices may be expected to increase in a bull market, while they may also be expected to fall in a bear market, equation (1) can then be expanded as follows[5]: n E ðP t Þ 1 þ E ð gÞ : ð3Þ E ðV Þt ¼ ½1 þ E ðr t Þn
465
In a rising bull market where the market price is expected to increase, Equation (3) adjusts the price to align with the expected prices that correspond to the market condition. Equation (2) would otherwise under-estimate the ETOM because the actual clearing price may grow by a rate g during the period n. The ETOM can then be derived as follows: Equilibrium TOM ¼ n ¼
logE ðP Þt logE ðV Þt h i : ðr Þ log 1þE 1E ð g Þ
ð4Þ
Based on the earlier illustration, the expected market clearing price is expected to increase to $735,000 if the asking price of the subject property increases by 5 per cent per annum in a bull market. The property is likely to take a longer than the earlier estimated ETOM of two years to clear given that the open market value remain unchanged. Based on the same assumptions, the ETOM estimated using Equation (4) increases to 2.9 years because of the wider gap between the open market value and the expected clearing price. In a bull market where demand exceeds the supply, there are likely to be more buyers than sellers in the market, which may intensify the competition for properties put up for sale. We may, therefore, expect prices to be over-inflated compared with their open market valuations. Under such market conditions, and assuming that the sellers are rational and have profit maximisation objectives, the property should have a zero or positive ETOM. In bear market conditions quoted selling prices may fall below their equivalent equilibrium value. This may well be the case if a seller is desperate to dispose of a property. In this case the ETOM may be zero or negative. In other words, it could be sold almost instantly or even before the property is placed on the market for sale. No transaction costs, in terms of search time and commission for brokerage, would be incurred, and these savings would be reflected in a negative ETOM within our proposed equilibrium framework. In the context of the typical TOM literature, negative results may be interpreted as the time spent prior to the listing of the property for sale
JPIF 22,6
that is not captured in the MLS. Negative ETOM may also take the form of time involved in the “for sale by owner” stage prior to engaging brokerage services, or the time involved in the pre-listing activities such as negotiation and appraisal. Negative ETOM is, however, unlikely to last for long periods, as they are a clear indication of the presence of under priced properties.
466
IV. Empirical analysis IPD data This paper proposes the concept of ETOM using a present value framework applied to a portfolio of commercial real estate. The IPD, one of the most widely used property market performance benchmarks in the UK, is used as the closest available sample for our ETOM analysis. The IPD All-Property Index is based on a portfolio of over 12,000 commercial properties predominantly owned by pension funds and insurance companies. For the ETOM analysis in this paper, the IPD portfolio-based all-property index is collected for a period of 17 years on an annual basis from 1983 to 1999. This index is used to proxy the market prices of a fully diversified portfolio of investment grade properties (Pt). As a separate exercise IPD has over a number of years undertaken extensive analyses of prices versus valuations using a single factor ordinary least squares (OLS) regression model with open market appraisals (Vt) as the independent variable: P t ¼ at þ bt V t þ ht
ð5Þ
where Vt is the aggregate open market appraisal observed in time t, at is the intercept, bt is the estimated coefficient and ht is the residual. The regression parameters: at and bt, have been computed by IPD on a yearly basis from 1983 using samples of properties that have sold from the index together with their equivalent open market appraisals. The sample sizes in each year varied but the total number of properties involved over the whole period from 1983 to 1999 was 11,819. By taking expectations of equation (5) the residual term can be dropped giving estimates of the expected market price. By assuming that the IPD All Property Index is a good representation of open market appraisals the expected value of equation (5) can be used in each period to estimate expected prices of the index based on the information set prevailing in each period. A comparison of open market values and expected prices for each year are shown in Figure 1. The average sample annualised rates of return of the All Property Valuation Index for the study period was computed at 10.69 per cent. It should be pointed out that in this case the differences between the appraisals and traded prices shown in this illustration have nothing to do with the concept of smoothing. Our analysis is concerned with differences in information sets that exist between valuations and prices at different points in time. In contrast smoothing is concerned with trying to identify fully informed market prices from appraisal data where it is believed that current appraisals are influenced in some way by previous appraisals. However, over the period from 1971 the IPD annual index has exhibited first order serial correlation close to 0.25. This is consistent with the findings of Working (1960) and suggests that the underlying changes in value are close to being random, but are averaged round the date of the valuation. Similar results can be observed with other UK annual indexes. In an additional piece of research on the effect
ETOM for commercial real estate 467
Figure 1. Open market valuations and expected prices (IPD All Property Index 1983 to 1999)
of sticky prices Brown and Matysiak (2000a, b) have shown that the effects of smoothing in index returns are negligible at the annual level but increase as the reporting interval reduces. This is largely due to the effects of serial cross correlation. In our analysis the difference between prices and appraisals using annual data can be accounted for by differences in information sets. It is this which identifies our equilibrium time on the market. Even if it could be argued that smoothing were an issue our analysis also includes standard error estimates for ETOM which would accommodate this phenomenon. Analysis of results Applying the periodic regression results to the IPD annual index provides an opportunity to observe the market view of traded prices whenever an open market valuation is prepared. The difference between the two figures provides the framework using equation (4) to estimate the ETOMs for the IPD All Property Index over the period from 1983 to 1999. Inputs of the discount rate (r) and the growth rate (g) two other variables are also required for the ETOM estimation using equation (4). The discount rate in each period can be represented by the annualised T-Bill rate plus a constant risk premium of 2per cent[6]. For the growth rate variable, as our model is drafted in expectations form, short-term annualised growth figures estimated by inter-temporal changes in the IPD annual indices may not be a good proxy for the equilibrium model. Therefore, we propose to use the initial yield of the sample IPD portfolio, (y), to capture jointly the investment risks and also the long-term growth potential of the sample properties[7]. It is thus not unreasonable in our analysis to substitute the discounting factor in the denominator of equation (4) by the initial yield discounting factor, ½1 þ EðyÞ. The resulting discounting figures, after taking into consideration implicitly the growth potential of property, represent a time varying discount rate that investors expected to earn on a period-by-period basis by investing in property. We also estimated standard errors for the mean ETOM in each period. These figures are summarised in Table I and are shown in Figure 2.
JPIF 22,6
468
Table I. Analysis of results – ETOM and confidence intervals
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Years Months
Estimated mean ETOM in years ETOM
ETOM (+1SE)
3.1812 3.0758 2.0053 0.5951 3.0175 4.1055 2.3830 21.1873 20.4934 20.2921 0.4052 0.9325 0.2218 0.7356 1.4584 0.9466 0.3139 1.2591 15.1092
3.5415 3.4188 2.3377 0.9276 3.3777 4.5062 2.7692 20.8871 20.2387 20.0613 0.6630 1.2063 0.4890 1.0095 1.7545 1.2554 0.6414 1.5712 18.8545
Mean ETOM plus confidence intervals ETOM (2 1SE) ETOM (+2SE) 2.8210 2.7327 1.6728 0.2627 2.6572 3.7049 1.9968 21.4876 20.7481 20.5229 0.1475 0.6587 20.0453 0.4618 1.1623 0.6379 20.0135 0.9470 11.3638
3.9017 3.7618 2.6702 1.2600 3.7380 4.9069 3.1554 2 0.5869 0.0161 0.1694 0.9207 1.4801 0.7561 1.2833 2.0507 1.5641 0.9688 1.8833 22.5998
ETOM (2 2SE) 2.4607 2.3897 1.3403 20.0698 2.2969 3.3042 1.6106 21.7878 21.0029 20.7537 20.1102 0.3849 20.3124 0.1880 0.8662 0.3292 20.3409 0.6349 7.6185
Figure 2. Equilibrium time on the market (ETOM)
Out of the sample of 17 annualised ETOM computations, we observed only three years from 1990 to 1992 when the evidence from this analysis suggested that commercial properties were selling below their open market values and were almost certainly under priced. Without additional information about the terms of the individual property sales it is difficult to draw any firm conclusions about why there was evidence of under pricing. Within an equilibrium framework it is quite logical to experience negative time on the market. The traditional view of time on the market is unable to accept this view but it is nevertheless an important consequence of
equilibrium markets. The UK commercial property market was very depressed during this period so the sample of properties sold from the index may have included a high proportion of “fire sales”. In any event it is clear that the phenomenon only lasted for a short period. The performance of the UK property market in the sample periods 1983-1989 and 1993-1999 were relatively bullish with expected prices consistently outpacing their open market equivalent valuations. During these periods, it may take a longer time for the willing buyer and willing seller to reach an agreement on the market prices due to the divergence in there respective expected prices[8]. On average, the ETOM for the 17-year sample period was estimated to be 15.1 months. In the context of the traditional TOM studies, these results imply that it would take about 15.1 months for the listed price of a commercial property to adjust to its open market clearing price. By also estimating standard errors of the mean ETOM it is possible to show that there is a 67 per cent probability that the mean ETOM will lie within the range of 11.36 months to 18.85 months. At the 95 per cent confidence interval this stretches to 7.62 months and 22.60 months. These bounds are shown in Figure 2. In a bearish market where properties are under priced relative to their equilibrium open market values buyers may have to take a pro-active role in the search process to identify potential property acquisitions. If the market is reasonably efficient it is unlikely that periods of mispricing would last for very long. Figure 2 shows that, within the context of the real estate market, the adjustment period is relatively quick.
V. Conclusion TOM has been widely tested in the US real estate literature due to the accessibility and availability of listing and selling data of individual houses in the MLSs. However, the results of these studies are varied and in some cases contradictory. This may be due to differences in sample data used, the time period analysed or empirical techniques used. Nevertheless, the earlier TOM literature has provided significant insights into factors affecting the tradition view of TOM in the US housing market. To date, there have been no similar TOM studies in the UK real estate literature as multiple listing services are uncommon in the UK market. By contrast, this study has examined TOM from an equilibrium market perspective using the commercial property market by analysing the speed or time the market takes to correct for differences in information between open market appraisals and expected prices. We defined this as the ETOM. In carrying out this study we have drawn on the UK research concerning valuation accuracy and used this as the basis for our proposed ETOM measure for the UK commercial real estate market. Using a time varying discount rate present value model, the time taken for the discrepancy between expected prices and open market appraisals to converge defines time on the market in equilibrium market conditions. The ETOM model is applied to the UK IPD All Property Index for a sample period of 17 years between 1983 and 1999, on an annualised basis. As the construction of the IPD All Property Index is based on the appraisals of a portfolio of investment grade properties in the UK, our ETOM estimates represent a portfolio approach. The empirical results showed that the average ETOM for the 17-year sample period was 8.40 months. Given the differences in information sets this reflects the average time it
ETOM for commercial real estate 469
JPIF 22,6
470
would take for expected prices to converge to their open market appraisals so that a sale would take place[9]. The concept of equilibrium is a pervasive factor in many areas of financial economics. Formal interpretation of traditional TOM studies should therefore be conditional on incorporating the notion of equilibrium markets[10]. Reconciling the factors that improve marketability with the different stages of equilibrium could offer useful marketing strategies for real estate agents in both the residential and commercial real estate markets. Notes 1. See Yavas (1994) for an excellent review of the theoretical issues in the broker’s search process. 2. In the two-stage least squares model of Sirmans et al. (1991), they found an opposite result vis-a`-vis Miller’s. A negative, but statistically insignificant coefficient was estimated in their high price hedonic equation, which reveals that house price falls as TOM increase. 3. This proposed equilibrium TOM is another example of the growing application of equilibrium analysis in real estate research. Hendershott (1997) has reviewed that equilibrium models have been widely used to analyse the risk premium for real estate assets, real house prices, rental adjustment and valuations. 4. Geltner and Mei (1995) have also proposed the use of time-varying discount rates in the present value model to improve the predictability of the property market value. 5. We acknowledge the anonymous referee for his kind suggestion to incorporate the explicit growth rate term into the simple present value model in equation (1). 6. Conventional real estate wisdom in the UK suggests a risk premium of 2 per cent is widely accepted. Brown and Matysiak (2000a, b) have also shown that over very long periods it is difficult to reject this hypothesis. 7. The use of initial yield variable as a long term discount rate in the ETOM, as suggested and commented by the anonymous referee, will simultaneously take into account the long-term growth of the sample portfolio. 8. The positive ETOM in the bull and bear market and the negative ETOM in the bear market simply reflect the time it takes for the open market value to adjust towards the market clearing prices. For example in the case of “fire sale,” a property if identified will probably involve shorter negotiation time for the parties to close the deal. The ETOM does not, however, imply that there will be more properties sold during the bear market period and vice versa. 9. These ETOM results are not an indicator of the market activities during the sample period, they simply suggest that in bull market, the price gap between the expected market clearing price and open market valuation is wider, and it will take a longer time for the market to clear, i.e. the ETOM is positive. 10. The equilibrium framework proposed in this study ignores any market arbitrage opportunity that may exist, i.e. the market may not clear at the equilibrium price and buyers may reap a positive return during the bull market. The ETOM concept is thus not able to capture capital returns of a sale of property in different market condition. This point was highlighted by the anonymous referee and it deserves further examination. References Baryla, E.A. and Zumpano, L.V. (1995), “Buyer search duration in the residential real estate market: the role of the real estate agent”, The Journal of Real Estate Research, Vol. 10 No. 1, pp. 1-13.
Belkin, J., Hempel, D. and McLeavey, D. (1976), “An empirical study of time on the market using multidimensional segmentation of housing markets”, Journal of American Real Estate and Urban Economics Association, Vol. 4 No. 2, pp. 57-75. Brown, G. and Matysiak, G.A. (2000a), Real Estate Investment: A Capital Market Approach, Financial Times Prentice-Hall, Harlow. Brown, G.R. and Matysiak, G.A. (2000b), “Sticky valuations, aggregation effects and property indices”, Journal of Real Estate Finance and Economics, Vol. 20 No. 1, pp. 49-66. Geltner, D. and Mei, J.P. (1995), “The present value model with time-varying discount rates: implications for commercial property valuation and investment decisions”, Journal of Real Estate Finance and Economics, Vol. 11 No. 2, pp. 119-35. Hendershott, P.H. (1997), “Uses of equilibrium models in real estate research”, Journal of Property Research, Vol. 14 No. 1, pp. 1-13. Janssen, C.T.L. and Jobson, J.D. (1980), “On the choice of realtor”, Decision Sciences, Vol. 11, April, pp. 299-311. Jud, G.D., Winkler, D.T. and Kissling, G.E. (1995), “Price spreads and residential housing market liquidity”, Journal of Real Estate Finance and Economics, Vol. 11 No. 3, pp. 251-60. Kang, H.B. and Gardner, M.J. (1989), “Selling price and marketing time in residential real estate market”, The Journal of Real Estate Research, Vol. 4 No. 1, pp. 21-35. Knight, J.R. (2002), “Listing prices, time on market, and ultimate selling prices: causes and effects of listing price changes”, Real Estate Economics, Vol. 30 No. 2, pp. 213-38. Larsen, J.E. and Park, W.J. (1989), “Non-uniform percentage brokerage commissions and real estate market performance”, AREUEA Journal, Vol. 17 No. 4, pp. 422-38. Miceli, T.J. (1991), “The multiple listing service, commission splits, and broker effort”, AREUEA Journal, Vol. 19 No. 4, pp. 548-66. Miller, N.G. (1978), “Time on the market and selling price”, AREUEA Journal, Vol. 6 No. 2, pp. 164-74. Ong, S.E. (2000), “Time-on-market revisited: effort, distributional form and unobserved heterogeneity”, paper presented at the Allied Social Science Association-American Real Estate and Urban Economics Association Conference, Boston, MA. Royal Institute of Chartered Surveyors (RICS) (1995), ““Practice statement 4.2”, RICS Appraisal and Valuation Manual, RICS, London. Sirmans, C.F., Turnbull, G.K. and Benjamin, J.D. (1991), “The markets for housing and real estate broker services”, Journal of Housing Economics, Vol. 1 No. 4, pp. 207-17. Working, H. (1960), “Note on the correlation of first differences of averages in a random chain”, Econometrica, Vol. 28, October, pp. 916-18. Yang, S.X. and Yavas, A. (1995), “Bigger is not better: brokerage and time on the market”, The Journal of Real Estate Research, Vol. 10 No. 1, pp. 23-33. Yavas, A. (1994), “Economics of brokerage: an overview”, Journal of Real Estate Literature, Vol. 2, July, pp. 169-95. Yavas, A. and Yang, S. (1994), “The strategic role of listing price in marketing real estate: theory and evidence”, Real Estate Economics, Vol. 23 No. 3, pp. 347-68. Further reading Haurin, D. (1988), “The duration of marketing time of residential housing”, AREUEA Journal, Vol. 16 No. 4, pp. 396-410.
ETOM for commercial real estate 471
The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-578X.htm
JPIF 22,6
How should unsmoothing affect pension plan asset allocation?
472
Sir John Cass Business School, City of London, London, UK, and
Philip Booth
Received September 2002 Accepted December 2002
George Matysiak CB Hillier Parker and Sir John Cass Business School, City of London, London, UK Keywords Pension funds management, Assets management, Resource allocation Abstract Examines the impact of using “unsmoothing” techniques on real estate data to take pension-plan asset-allocation decisions. It is generally believed that valuation-based real estate indices give rise to returns figures which are “smoothed” versions of the underlying transaction prices. Unsmoothing techniques can be used to develop real estate return data series that are believed to be a more accurate representation of underlying transaction prices. If this is done, the resulting data reveal greater volatility of real estate returns. When such data are applied to portfolio selection models, they often reveal a reduced allocation to real estate in efficient portfolios. Looks at the impact of unsmoothing data when taking pension-plan asset-allocation decisions. Finds here that the unsmoothed data are more closely correlated with pension plan liabilities. As a result, efficient pension plan portfolios sometimes contain more real estate, rather than less. In general, there is little change in the efficient real estate allocation. These results are very important. They reveal that so-called “valuation smoothing” may distort property investment decisions less than is commonly thought.
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 472-483 q Emerald Group Publishing Limited 1463-578X DOI 10.1108/14635780410569461
Introduction This paper offers a new insight into the problem of taking decisions using real estate investment return data. It is generally thought that reported real estate returns, based on valuations, are “smoother” than returns that would be derived from transaction-based real estate indices. It is also commonly believed that the use of “unsmoothed” data had a higher standard deviation of returns and possibly a higher correlation of returns with other asset classes which would give rise to lower proportions of real estate in efficient asset portfolios. We demonstrate that this commonly-held belief is reasonable. Most investment in real estate is held to meet actuarial liabilities (for example, long-term life insurance or pension fund liabilities). In this context, the use of unsmoothed real estate data gives rise to results that are quite different from the results in the asset-only portfolio selection model. The authors have used a model of an immature and mature pension scheme (used also in Booth, 2001). We find that unsmoothed real estate returns are more highly correlated with the liability returns from both the immature and the mature pension scheme model than are real estate returns that are taken directly from valuation-based data. As a result the real estate proportion in efficient pension fund portfolios sometimes falls by much less than in the asset-only case. However, frequently the real estate proportion in efficient portfolios remains the same or actually increases, when we compare the results using “smooth” and unsmoothed data. These results are highly significant.
We begin by discussing the smoothing method and methodology. The use of unsmoothed data in portfolio selection models is then presented. In the second half of the paper, we use the unsmoothed data to find efficient and optimal portfolios in a pension plan portfolio selection context. Unsmoothing real estate data Assets that are freely traded in a competitive market provide a useful reference point for determining the value of other assets. Unfortunately, the competitive market model is unlikely to offer, at all times, an accurate representation of the way real estate markets work in practice. For a variety of reasons such imperfections can create valuations that are “smoothed” versions of their underlying market price. Broadly, smoothing can arise from two distinct processes: one due to valuation at the individual property level and the other due to the effect of aggregating groups of properties into indexes (Brown and Matysiak, 2000). A variety of models have been proposed for transforming a valuation-based series into an unsmoothed derivative version of the underlying price series. The approach we have adopted here is to adjust for smoothing by using a first order, AR(1), autoregressive process (see also Brown and Matysiak, 2000). This enables an unsmoothed total returns series for the Index Property Databank (IPD) all property index to be backed-out, which has subsequently been used in the simulations. The sensitivity of the simulation results to different unsmoothing procedures is part of ongoing work. The unsmoothing procedure was applied to the IPD capital values series to which the income return was added back in order to obtain an unsmoothed total rate of return for each year. The standard deviation of the published IPD total-return series was 11.2 per cent and that for the unsmoothed series 15.7 per cent[1], representing an increase of 40 per cent in variability. The corresponding first order auto-correlation coefficients were 0.27 and 0.028 respectively. The result of this simple approach produced a more volatile return series that exhibits low first order correlation. The estimated smoothing parameter was 0.32 and may be interpreted as follows. Over the period, one-third weight was assigned to previous valuations and two-thirds to new market evidence in arriving at a current valuation. This is an average figure and, depending on market conditions, is likely to exhibit considerable variation (see Brown and Matysiak, 2000). The unsmoothed real estate, total-return data were then used in the optimisation process to find optimal asset portfolios for pension plans with different liability structures. The results were contrasted with those found when using smooth real estate data and those found for “asset-only” optimisation. Analysis of the asset return data The data period is limited to the length of the return series for index-linked gilts. We have started the data series in 1984, by which time index-linked gilts had been in existence for nearly three years and had become reasonably marketable. The matrices for expected returns, standard deviations of returns and covariances of annual returns from assets are estimated from historical data over the period 1984 to 2000 inclusive. Mean returns and standard deviation of returns are shown in Table I. Asset returns were taken from Barclays Capital (2001); exchange rates were taken from the Bank of England Statistical Abstract (2001 and earlier volumes (Bank of England, 2001); real estate returns were taken from IPD (2001). There is a discussion of the characteristics of the raw asset return data in Booth (2001). However, in this paper, the unsmoothed property returns have been used. Here we just draw out those characteristics of the unsmoothed returns that are relevant and
Pension plan asset allocation
473
JPIF 22,6
474
refer the reader to the earlier paper for a discussion of the return characteristics of the other assets. The mean return from real estate, using unsmoothed data, is the same, to one decimal place, as that for the smooth data. The unsmoothing process does not give rise to systematic changes in the mean return, although some change is possible depending on the sample period chosen. The standard deviation of returns from real estate increases from 9.5 to 12.6, as a result of the unsmoothing process: this is an increase of nearly one-third. The standard deviation of returns from the unsmoothed data is exactly the same as that from UK equity returns[2]. The standard deviation of returns from the unsmoothed real estate data is now greater than that from conventional gilts, as would normally be expected. Portfolio decisions are, of course, taken on the basis of ex-ante expectations and not ex-post values of the mean and standard deviation of returns. It is therefore reasonable to ask, before proceeding to determine efficient portfolios, whether the asset return statistics would form a reasonable basis for expectations. In terms of the ordering of mean returns and of standard deviations of returns the data are not unreasonable, given the assumptions underlying the smoothing process, although it could be argued that we would expect bond returns to be lower than real estate returns and conventional bond returns to be much closer to index-linked bond returns. It can also be said that index-linked gilts have produced real returns that may seem unreasonably low, relative to those from equities. The effective equity risk premium over index-linked bonds, at 6.9 per cent, is high (as noted in Booth, 2001). Nevertheless, the magnitude of the risk premium is not so unreasonable that it should not be used as a basis for the simulations in this paper. Chun et al. (2000) have also looked at the role of real estate in pension plan portfolios, using US data. It was noted, in Booth (2001) that they found a higher standard deviation of returns form real estate investment than we found for the smooth UK data. One of the reasons for this was that Chun et al. used US real estate investment trusts (REIT) returns, which are quoted vehicles. As would be expected, the unsmoothing process has brought the standard deviation of returns for real estate, in our data set, much closer to the standard deviation found by Chun et al. (2000). The correlation matrix for the asset classes, using the unsmoothed real estate data is shown in Table II. Again, there is a general discussion of the correlation matrix in Booth (2001). Here we just concentrate on the changes arising from unsmoothing the data. The correlation coefficient between real estate and UK equities is substantially higher. It is similar in magnitude to the correlation coefficient found between the UK equity market and other equity markets outside the EU and the USA in many studies. The correlation coefficient between real estate and both index-linked and conventional gilts is substantially higher using the unsmoothed real estate data. There is little change in the Asset class
Table I. Mean and standard deviation of asset returns
UK equities US equities Real estate Index-linked gilts Cash Conventional gilts
Mean return (%)
Standard deviation of return (%)
15.6 16.9 10.9 8.7 9.1 11.6
12.6 19.8 12.6 7.3 3.1 9.7
correlation coefficients between real estate and both cash and US equities. It would therefore appear that the correlation coefficients between the unsmoothed real estate data and UK securities markets are substantially changed, with little impact on the other correlation coefficients. Unsurprisingly, our correlation figures are closer to those of Chun et al. (2000) than those found by Booth (2001). It is of interest to ask why the correlation between real estate and other asset classes may be higher using unsmoothed data. One reason is that the unsmoothed data should be a better representation of transaction prices than the smooth data. These transaction prices should take more immediate account of the economic factors that affect all investment markets (for example long-term risk-free interest rates). For a similar reason, we might also expect the unsmoothed real estate data to have a higher correlation with pension plan liabilities (see below).
Pension plan asset allocation
475
Optimisation in the asset only framework In this section we determine efficient portfolios, which include all the available asset classes analysed above, using unsmoothed real estate data. We will look at efficient portfolios in the “asset only” context first, before bringing in pension plan liabilities in later sections. We begin by finding the absolute minimum standard deviation of return portfolio and then find the minimum standard deviation of return portfolio at four different levels of expected return. The maximum expected return portfolio is 100 per cent invested in US equities. The efficient portfolios we found are shown in the Table III. The figures in column A show the optimal asset allocation when the smooth real estate data was used (i.e. the Correlation matrix
UK equities
US equities
Real estate
IL gilts
Cash
US equities Real estate Index-linked gilts Cash Conventional gilts
0.81 0.26 0.49 0.01 0.34
20.02 0.47 20.04 0.26
0.26 20.40 0.17
0.00 0.76
2 0.01
Asset class Risk level UK equities US equities Real estate Index-linked gilts Cash Conventional gilts Expected return (%) Standard deviation of return (%)
A B Minimum risk 0 1.7 14.6 0 77.9 5.8 9.66 2.48
0 1.4 11.3 2.2 82.7 2.4 9.50 2.47
Proportion in asset class (%) A B A B Low risk 8.6 3.9 16.6 0 61.9 9.0 10.5 3.10
10.7 3.8 10.6 0 67.1 7.8 10.5 3.26
Mid risk 31.8 2.7 17.3 0 35.6 12.6 12.0 5.48
Notes: A ¼ smoothed real estate data; B ¼ unsmoothed real estate data
37.4 0.5 6.2 0 43.3 12.6 12.0 5.69
A
Table II. Asset correlation matrix
B
High risk 62.7 1.0 18.2 0 0.6 17.5 14.0 9.23
67.9 0 1.6 0 11.7 18.8 14.0 9.41
Table III. Efficient portfolios in the asset only framework
JPIF 22,6
476
results from Booth, 2001). The figures in column B show the optimal asset allocation using the unsmoothed data. The results are intuitive, given the patterns in the data. There are substantial falls in the optimal real estate holdings, in all portfolios. The falls are greater for the higher risk portfolios. There is a small fall in the real estate holding in the minimum risk portfolio (of 3.3 per cent of the portfolio). The standard deviation of return of the minimum risk portfolio has also fallen slightly. This seems paradoxical but will have arisen because of the lower correlation between real estate and cash, using the unsmoothed figures. In the low risk portfolio, the real estate allocation falls by one third, in the mid-risk by two thirds and, in the high-risk portfolio, real estate becomes an insignificant investment. For all these portfolios, it is impossible to obtain as low a standard deviation of return for a given expected return, when the real estate data are unsmoothed as when they are smoothed. These results arise because the return characteristics of real estate become more like those of UK equities when the real estate returns are unsmoothed. Real estate has a standard deviation of returns exactly the same as that for UK equities and is more highly correlated with UK equities. The latter characteristic means that real estate is less effective at diversifying portfolios than is apparent when smooth real estate data are used. It is of interest to investigate the marginal consequences for optimal asset allocation of the change in the correlation coefficient between UK equities and real estate which arises from unsmoothing the data and the change in the standard deviation of return from real estate. If the standard deviation of return from real estate is changed back to its value from smooth data (9.5 per cent) but the covariance structure with other asset classes remains that calculated from the unsmoothed data, the real estate proportion in the optimal mid-risk portfolio increases to 12.0 per cent. If the standard deviation of real estate returns is kept at its value from the unsmoothed data, but the correlation coefficient is returned to its value from the smooth data, the real estate proportion in the optimal mid-risk portfolio increases to 9.2 per cent. Thus both the change in the correlation structure and the change in the standard deviation of returns, which arise from unsmoothing the real estate data, each make a substantial individual contribution to reducing the optimal real estate proportion from 17.3 per cent to 6.2 per cent. There are some general conclusions that can be noted which were also drawn from Booth (2001). The low-risk portfolios are invested predominantly in cash and the high-risk portfolios in UK equities. Because of the high standard deviation of returns of US equities and high correlation of returns between US equities and UK equities, there appear to be few benefits from diversification into overseas equities. Index-linked gilts do not appear in either low or high-risk portfolios. This situation changes significantly when we consider investing to meet long-term pension liabilities. The pension plan liability model The next step was to investigate the impact of unsmoothing real estate data on optimal asset allocations to pension plans. The pension fund liability model used in this paper is exactly the same as that used in Booth (2001). It will be described very briefly here. The liability model is constructed in two parts. The first part of the liability model is for those members who have not yet retired and are still active in the scheme (the immature part of the scheme). The second part is the liability model for the mature part of the scheme (members who have already retired). We ignore members who may have deferred pensions. We assume that the scheme is either 100 per cent mature or 100 per
cent immature. Some schemes will follow one of these structures. In practice, most will fall in between. Nevertheless, it is instructive to look at the impact of the extreme cases on optimal investment policy. Equal numbers of individuals are assumed to be members in each age group. For simplicity, we assume an entry into the scheme every five years. Individuals accrue a pension of 1/60 of final salary for each year of their membership, which is assumed to be between age 25 and their current age. The value of the accrued liabilities is determined by projecting salary increases until retirement and discounting the annuity benefits that would be given, based on service to date. The valuation of these liabilities, in respect of the active, working population is carried out with all variables being expressed in “real” terms. The real rate of interest for discounting liabilities before retirement is the rate of return from long-term UK government index-linked bonds. This means that no inflation projection needs to be carried out as all variables are defined in real terms. After retirement we have assumed that liabilities are fixed in nominal terms; discounting in this period is carried out at the yield from long-dated conventional gilts. The real rate of salary increases for projections is assumed to be 3 per cent[3]. The value of the liabilities can thus change for two reasons. Interest rates can change and actual salary increases can be different from projections. The second part of the liability model represents the liabilities in respect of those members who have already retired. This part of the scheme will also have pensioners at representative five-year age intervals. Pensions in payment are assumed to be fixed nominal, amounts[4]. Specifically, the pensions in payment liabilities are determined as follows: it is assumed that, at age 65, an equal number of people had retired in each of the past 35 years (again we take representative ages of 70, 75, 80 and so on); they each had a pension of £18,579 (the pension accrued by 65 year olds in the active part of the scheme) reduced by an assumed rate of average real earnings growth in the period since retirement for each year of age after 65 (as this represents the period since retirement). It is therefore implicitly assumed that historically, pensions in payment have been uprated to compensate for inflation but that, in the future, there will be no such uprating. Again, this is common in UK schemes, where uprating is discretionary. It may be assumed that such uprating will not occur but, if there are surpluses within the scheme, they may be used to provide price indexation (see Faculty and Institute of Actuaries, 1999). Mortality tables are then used to determine the proportion still living at each age and future payments are then projected using the mortality tables and discounted. The mature liabilities are expressed in nominal terms and discounted at nominal rates of interest from long-term conventional government bonds. Analysis of the asset/liability modelling data The mean rate of growth of the liabilities of the mature scheme is 10.56 per cent. The standard deviation of the rate of growth of the mature liabilities is 5.93 per cent. The mean rate of growth of the liabilities of the immature scheme is 11.42 per cent and the standard deviation of the rate of growth is 11.62 per cent (see Booth, 2001). The correlation structure between the pension plan liabilities and the asset classes, using the unsmoothed data for real estate returns is shown in Table IV. Comparison with Booth (2001) shows that the correlation between real estate returns and the mature liability returns has increased from 0.11 to 0.27; the correlation between the immature liabilities and real estate returns has increased from 0.23 to 0.33. The other characteristics of these data were discussed in Booth (2001). As is demonstrated in the literature on asset/liability modelling (for example, Wilkie, 1985),
Pension plan asset allocation
477
JPIF 22,6
478
the correlation between assets and liabilities is a crucial factor determining optimal asset allocation in a pension fund. Looking at the asset-only data, we could make a strong prima facie case that optimal real estate allocations should be lower because the standard deviation of returns from real estate was higher and the correlation coefficients between real estate and other asset classes were generally higher as a result of unsmoothing the data. The asset/liability data paint a more ambiguous picture. The characteristics of the asset-only data combined with the information on asset/liability correlation coefficients means that we cannot form an a priori view on the implications for optimal asset allocations of basing decisions on unsmoothed rather than the smooth data. Asset/liability modelling with unsmoothed real estate data The discussion of the choice of and the derivation of the objective function, for asset liability modelling is in Booth (2001). The various options proposed by Wilkie (1985) and Sherris (1992), Chun et al. (2000) and Leibowitz et al. (1994) were discussed. On balance it was decided that the best approach was to maximise a function of the rate of return (or rate of increase in) the surplus of the scheme, where the surplus was expressed as a proportion of the asset values. The following notation was used: SR0t
¼ surplus return, standardised for initial asset values, in year t.
At-1
¼ assets invested in the scheme at time t 2 1.
xi
¼ proportion of asset portfolio invested in asset i.
Ri
¼ rate of return on asset i.
Rl
¼ rate of return (rate of increase) in the plan’s liabilities.
Lt-1
¼ scheme liabilities at time t 2 1. SR0t ¼
i¼n X
xi R i R l £
i¼1
Lt1 : At1
We assume that pension plans maximise the objective function: 2 F SR0t ¼ E SR0t a s SR0t : The higher is a, the greater will be the pension plan’s aversion to risk. If a ¼ 0, the pension plan sponsors will be risk neutral and will try to maximise expected surplus. As a ! 1, the plan sponsor will choose the portfolio that will minimise the variance of the plan surplus,
Correlation matrix Table IV. Asset and liability correlation coefficients
Liability mature scheme Liability immature scheme
UK equities US equities Real estate IL gilts 0.27 0.28
0.14 0.25
0.27 0.33
0.72 0.88
Cash
Conventional gilts
0.04 2 0.22
0.97 0.91
where: i¼n X Lt1 E SR0t ¼ xi E ðRl Þ £ A t1 i¼1
with E(Ri) ¼ expected rate of return from asset i; E(Rl) ¼ expected rate of growth (rate of return) of the liability. Also: j¼n i¼n i¼n X 1¼n X X 2 X L2 L2 Lt1 s SR0t ¼ xi s2i þ sij xi xj £ t1 þ s2l £ t1 sil xi £ 2 2 A t1 At1 At1 i¼1 i¼1 i¼1 j¼1
with:
si
¼ standard deviation of return from asset i.
sij
¼ covariance of return between asset i and asset j.
sl
¼ standard deviation of return of the liability.
sil
¼ covariance of return between asset i and the liability.
Optimisation for the mature scheme In this case, we found portfolios which maximised the values of the objective function, defined above, for sample values of a between 0.006 and 1, for the pension scheme with mature liabilities. We also found global minimum variance of surplus return portfolios and the portfolio that maximised expected surplus return (trivially, 100 per cent US equities). The efficient portfolios for the scheme with mature liabilities are shown in Table V. The figures in column A show the optimal asset allocation using the smooth real estate data. The figures in column B show the optimal allocation using the unsmoothed real estate data. There is remarkably little difference between the two sets of optimal allocations. Real estate only forms part of the optimal allocation for the mature scheme, when smooth data are used, in the low and medium risk portfolios. In the low risk portfolios, the optimal real estate allocation drops slightly when unsmoothed data are used. In the medium risk portfolio, the optimal real estate proportion drops significantly when unsmoothed data are used but by far less than the fall in the equivalent asset-only portfolios. The optimal real estate proportion falls by 15 per cent of its value for the low risk portfolio and by one third of its value for the medium risk portfolio. This compares with a fall of over one-third of its value and nearly two-thirds of its value in the two corresponding asset-only cases (see above). In Booth (2001), it was found that setting a ¼ 0:05 would lead to an asset allocation such that the probability of falling below a 95 per cent funding level for the scheme (assuming that it was fully funded to begin with) was about 5 per cent[5]. The optimal asset allocation to achieve such a result, using smooth real estate data was then: 31.6 per cent UK equities, 6.8 per cent real estate and 61.6 per cent conventional gilts. If we
Pension plan asset allocation
479
JPIF 22,6 Asset class .Risk level
480
Table V. Efficient portfolios for the mature pension plan
A B .Minimum risk
A
Proportion in asset class (5) B A B A
.a ¼ 1
.a ¼ 0:15
B
.a ¼ 0:02
A
B
.a ¼ 0:006
UK equities 0 0 0 0 9.6 9.5 69.2 69.2 56.7 56.7 US equities 0 0 0 0 0 0 0 0 43.3 43.3 Real estate 9.9 8.6 10.4 8.8 11.1 7.5 0 0 0 0 Index-linked gilts 0 0 0 0 0 0 0 0 0 0 Cash 29.4 32.6 27.7 31.3 15.5 20.3 0 0 0 0 Conventional gilts 60.7 58.8 61.8 60.0 63.8 62.7 30.8 30.8 0 0 Expected surplus return 0.26 0.19 0.29 0.22 0.97 0.88 3.80 3.8 5.59 5.59 Standard deviation of surplus return 1.20 1.10 1.21 1.11 2.03 1.98 8.48 8.48 14.76 14.76 Notes: A ¼ smoothed real estate data; B ¼ unsmoothed real estate data
maximise expected return on surplus for the same standard deviation of return on surplus (4.47 per cent), using the unsmoothed real estate return data, the optimal asset allocation for the scheme is: 30.1 per cent UK equities, 4.0 per cent real estate and 65.9 per cent conventional gilts. This represents a significant reduction in the real estate allocation but, again, the effect of unsmoothing the data are much less than in the asset-only case. We can conclude from this analysis that using real estate data that are believed to reflect underlying transactions, rather than smoothed valuations, to take pension plan investment decisions for mature schemes, makes little difference to the optimal real estate allocation. The impact of the higher standard deviation of returns and higher correlation with other asset classes is offset by the higher correlation of the unsmoothed real estate data with the actuarial liabilities of the scheme. Optimisation for the immature scheme It is worth noting that real estate did not feature in any of the optimal portfolios for the immature plan, except for the medium risk portfolio, when smooth real estate data were used in Booth (2001). In the low risk portfolios, real estate was “replaced” by index-linked gilts. The optimal portfolios using smoothed and unsmoothed data are shown in Table VI. Once again, we have found the minimum standard deviation of return portfolio and optimal portfolios for values of a from 1 to 0.006. The highest expected return of surplus portfolio is 100 per cent invested in US equities. In Booth (2001), index-linked and conventional bonds dominated the low-risk portfolios. A mixture of equities and conventional bonds dominated the high-risk portfolios. This was a reasonable result. A mix of conventional and index-linked bonds is probably the closest match to the liabilities. As riskier portfolios are chosen, one real asset (equities) is chosen in place of another (index-linked bonds): equities are higher risk but have a higher expected return. When the unsmoothed data are used, the optimal real estate proportion increases in the mid-risk portfolio and appears for the first time in the low risk portfolios (at percentages of 4-5 per cent). This result is of considerable importance. Real estate appears “riskier” when the unsmoothed data are used in asset-only studies because it has a higher standard deviation of returns and higher correlation coefficients with
0 0 0.3 26.0 0 73.7 2 0.56 4.56
Notes: A ¼ smoothed real estate data; B ¼ unsmoothed real estate data
0 0 4.1 25.7 0 70.3 20.58 4.52
0 0 4.8 21.1 0 74.1 20.45 4.53
0 0 0 30.1 0 69.9 2 0.68 4.55
UK equities US equities Real estate Index-linked gilts Cash Conventional gilts Expected surplus return Standard deviation of surplus return
Risk level
Proportion in asset class (%) A B A B Minimum risk a¼1
Asset class B 0 3.9 7.7 0 0 88.5 0.35 4.83
a ¼ 0:15 0 3.7 6.0 0 0 90.2 0.35 4.89
A
B 42.2 11.6 0 0 0 46.2 2.48 9.09
a ¼ 0:02 42.2 11.6 0 0 0 46.2 2.48 9.09
A
45.9 54.1 0 0 0 0 4.87 16.7
45.9 54.1 0 0 0 0 4.87 16.7
A B a ¼ 0:006
Pension plan asset allocation
481
Table VI. Efficient portfolios for the immature pension plan
JPIF 22,6
other asset classes. However, as is clear from Table IV, real estate returns are also more highly correlated with immature pension plan liabilities, thus providing a “better match”. It is therefore not surprising that optimal low-risk portfolios, for immature pension plans, contain a higher proportion of real estate when using unsmoothed real estate data, than when smoothed data are used to take the investment decision.
482
Conclusions It is generally believed that valuation-based indices lead to “smoothed” real estate returns in the sense that returns are less volatile than those which would underlie transaction prices. It is possible to use techniques of “unsmoothing” so that the characteristics of real estate data used in portfolio selection models reflect more closely data which would be obtained from transaction prices. When this is done, the revised data have the same expected return, a higher standard deviation of return and higher correlation coefficients with other asset classes. Applying such unsmoothed real estate data in portfolio selection models leads to significantly reduced optimal real estate holdings in efficient asset portfolios. This is not surprising given the changed characteristics of the data. This conclusion may lead us to believe that real estate is a more risky investment than it has hitherto been considered. However, institutional investors invest in order to meet liabilities and, in the UK, nearly 40 per cent of long-term investments are held by pension funds. When we set up an asset/liability model, we find that the unsmoothed data are more highly correlated with both mature and immature pension fund liabilities and thus may provide a better match for liabilities. In an asset/liability matching context, real estate might provide a less risky investment than has hitherto been assumed. When we determine efficient portfolios for a mature pension fund there is a fall in optimal real estate allocations but the fall is much smaller than occurs in the asset-only optimisation. However, when we determine asset allocation for the immature scheme we find that optimal real estate holdings actually increase. These results are highly significant. When real estate data are unsmoothed, they may actually reveal an asset which is less risky, rather than more risky than has generally been assumed to be the case, if it is seen in the context of institutions investing to meet liabilities. The extent to which the perceived risk changes would depend on the precise nature of the liabilities. It could be asked whether pension fund asset allocation decisions should be based on unsmoothed or smooth real estate data. Pension funds are long-term funds and are often subject to liability valuations that are themselves smoothed and do not react immediately to changing financial conditions. It would not be unreasonable for such funds to use raw index data for determining the structure of real estate returns. However, pension funds in the UK are increasingly using market-value valuation techniques, in line with US practice. In effect, the model scheme in this paper has used that approach. Where market value valuations are used, it is reasonable to use real estate data in asset allocations, which are unsmoothed to reflect underlying market conditions more effectively. Notes 1. These figures come from the data series from 1971 to 2000. In this paper, we have only used real estate returns from 1984 to 2000 (inclusive) because of unavailability of returns from other asset classes during the longer period.
2. This equality is not a necessary result of the unsmoothing process. 3. This is equal to the average rate of real salary growth over the recent past plus an assumed scale rise of 0.5 per cent per annum. 4. In the UK, there is considerable variation in the way in which pensions in payment are varied. Some will be fixed in nominal terms; some will be fixed in nominal terms but trustees will give discretionary uplifts if there is a surplus in the fund; some will be fixed in real terms; and some will be subject to “limited price indexation” so that they change with changes in the price level with a minimum increase of 0 per cent and a maximum of (say) 3 per cent or 5 per cent. We do not model these particular circumstances. However, such modelling may provide an interesting extension of this work. In particular, the limited price indexation liabilities have an option structure which has similarities with (but is still different from) the upward only rent review option structure in the traditional UK institutional lease. 5. In fact, it was 5.26 per cent, assuming that surplus was normally distributed. References Bank of England (2001), Monetary and Financial Statistics, Vol. 5, Bank of England, London. Barclays Capital (2001), Equity Gilt Study, Barclays Capital, London. Booth, P.M. (2001), “Real estate investment in an asset/liability modelling context”, Proceedings of the 8th European Real Estate Society Conference (submitted to the Journal of Real Estate Portfolio Management). Brown, G.R. and Matysiak, G.A. (2000), Real Estate Investment: A Capital Market Approach, Financial Times/Prentice-Hall, London. Chun, G.H., Ciochetti, B.A. and Shilling, J.D. (2000), “Pension plan real estate investment in an asset/liability framework”, Real Estate Economics, Vol. 28 No. 3, pp. 467-92. Faculty and Institute of Actuaries (1999), Retirement Benefit Schemes – Minimum Funding Requirement: Guidance Note 27, Faculty and Institute of Actuaries, London. Investment Property Databank (IPD) (2001), IPD UK Annual Index, IPD, London. Leibowitz, M.L., Kogelman, S. and Bader, L.N. (1994), “Funding ratio return”, Journal of Portfolio Management, Vol. 21, Fall, pp. 39-47. Sherris, M. (1992), “Portfolio selection and matching: a synthesis”, Journal of the Institute of Actuaries, Vol. 119 No. 1, pp. 86-106. Wilkie, A.D. (1985), “Portfolio selection in the presence of fixed liabilities: a comment on the matching of assets to liabilities”, Journal of the Institute of Actuaries, Vol. 112 No. 2, pp. 229-77.
Pension plan asset allocation
483
The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister
JPIF 22,6
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-578X.htm
The uncertainty of valuation Nick French The Department of Real Estate and Planning, The University of Reading Business School, Reading, UK, and
484 Received June 2003 Accepted July 2004
Laura Gabrielli Urban Planning Department, IUAV Venice University of Architecture, Venice, Italy Keywords Uncertainty management, Market value, Asset valuation, Property, United Kingdom Abstract Valuation is often said to be “an art not a science” but this relates to the techniques employed to calculate value not to the underlying concept itself. Valuation is the process of estimating price in the market place. Yet, such an estimation will be affected by uncertainties. These input uncertainties will translate into an uncertainty with the output figure, the valuation. The degree of the uncertainties will vary according to the level of market activity; the more active a market, the more credence will be given to the input information. In the UK at the moment the Royal Institution of Chartered Surveyors (RICS) is considering ways in which the uncertainty of the valuation can be conveyed to the use of the valuation, but as yet no definitive view has been taken apart from a single Guidance Note (GN5). One of the major problems is that valuation models (in the UK) are based on comparable information and rely on single inputs. They are not probability-based, yet uncertainty is probability driven. This paper discusses the issues underlying uncertainty in valuations and suggests a probability-based model (using Crystal Ball) to address the shortcomings of the current model.
In those situations where a single value can be misleading it has been suggested that a range of values might be more meaningful (Brown, 1991, p. 63).
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 484-500 q Emerald Group Publishing Limited 1463-578X DOI 10.1108/14635780410569470
Introduction The thesis of this paper is that uncertainty is a real and universal phenomenon in valuation. The sources of uncertainty are rational and can be identified. They can be described in a practical manner and, above all, the process of identification and description of uncertainty will greatly assist many clients and will improve the content and the credibility of the valuer’s work. The paper concentrates on the practical and the impact of uncertainty in property valuation. Uncertainty impacts on the process in two ways: first, the cash flows from investment are, to varying degrees, uncertain and second, the resultant valuation figure is therefore open to uncertainty. The paper looks at how uncertainty can be accounted for in the valuation and how it can be reported to the client in an effective and meaningful way. This requires a standardised approach and we suggest that the use of a generic forecasting software package, in this case Crystal Ball [1], allows the valuer to work with familiar pricing models set up in Excel or Lotus 123 and to work with a predetermined set of probability distributions.
Literature review – risk and uncertainty Before we can consider uncertainty within the valuation process it is important to define what it is that we mean by uncertainty. Both within the academic literature and, more so, the property profession, the terms risk and uncertainty are often used interchangeably. Risk is seen as a euphemism for uncertainty. However, this colloquial use of the words is unhelpful in identifying the principal issues involved. It is important to define these words more precisely. Definitions and discussion about risk and uncertainty are the cornerstone of a number of papers and books (see for example Byrne, 1995, Hargitay and Yu, 1993; Pellat, 1972; Pyrrh, 1973; Robinson, 1989; Sykes, 1983; Whipple, 1988; Wooford, 1978). The definitions that we are adopting follow the work of Byrne and Cadman (1984): . Uncertainty. This is anything that is not known about the outcome of a venture at the time when the decision is made. . Risk. This is the measurement of a loss identified as a possible outcome of the decision. It is generally agreed that uncertainty is due to the lack of knowledge and poor or imperfect information about all the inputs that can be used in the analysis. In the context of valuation this refers to the input variables; the comparable information. If we are unable to confirm the veracity of the inputs then the resulting outcomes (valuation) are partially uncertain. However, if we are able to assign a probability to the input variables it will allow us to determine the range of possible outcomes. The output is therefore a measure of risk (Byrne, 1995). The outcome of a valuation is only certain if we can accurately predict the future. Given that is not possible, there will always be an element of risk that the “actual” value differs from the predicted estimate. With a single point valuation, a single figure is produced with no understanding of the uncertainty pertaining to the input variables and thus no measure of the resulting risk. An improvement on this method would be to undertake the same valuation a limited number of times, allowing the user to change the input variables and recalculate each time to derive a number of possible outcomes or values. This analysis is a simple sensitivity or scenario analysis, but is restricted to (maybe) only three or four scenarios based on a subjective assessment of how the input variables should be changed. A more robust model would allow the user to simulate a much larger range of possible outcomes. Literature review – simulation Probability theory is a way of measuring uncertainty (Byrne and Cadman, 1996). It allows the user to identify a range of outcomes for the most important variables and to assign probabilities to these variables. Simulation is a further development of probability analysis and Monte Carlo simulation has been an important component of quantitative risk since 1960s (see Hertz, 1964). The underlying premise of Monte Carlo simulation is to carry out the process, in this case valuation, a large number of times. Instead of using a single point estimate for each input variable the user ascribes a probability distribution to each input and the Monte Carlo technique selects random numbers for each variable and produces an answer (valuation) on that basis before selecting another random input (from within the set range) and repeating the exercise.
The uncertainty of valuation
485
JPIF 22,6
486
The model will carry out this process to produce a multiple of possible outcomes that can then be statistically analysed to provide an average outcome, a range, a standard deviation etc. Each output would be the distribution of the possible outcomes and the range of possible valuations figures. The results of simulations are represented in a form of a discrete distribution (histogram) or continuum distribution (normal distribution). Those distributions allow the valuers to know about the range of the outcomes and the probability of the values at each point of the distribution (Evans, 1992). In statistics there are many forms of probability distributions, which describe both the range of the values and the likelihood of their occurrence. The normal distribution (a bell-shape distribution) is the most known and its parameters (the mean and the standard deviation) are the most used. The variability of valuation then depends on the variability of the inputs. Therefore, the process involves the identification of these variables, defining their probability distribution and, if there is any, there correlation (or inter-relationship) and then calculate the output (valuation figure). Byrne (1989), suggesting that all valuers are aware that inputs and output from appraisal and valuations are uncertain, used a technique for risk analysis (and a package called @Risk) in a discounted cash flow (DCF) model to provide a better decision-making model for property investments. This was echoed by Kelliher and Mahoney (2000) who used a Monte Carlo Simulation to model outcomes in the context of a long-term investment decision. This was further developed by Fraser (2004) who also suggested the use of a DCF analysis to generate a number of outcomes via a simulation model. The accuracy of the simulation depends on the quality of the data used in the models. The problem still remains of the ability to specify the real range of the inputs and their probability distributions, especially for the practitioners who have no familiarity with statistical measures such as mean and standard deviation. This is discussed later in the paper. Valuation variance and uncertainty The definition of uncertainty that is the subject of this paper is uncertainty in the individual valuation figure of the individual valuer. It is not the difference of values of the same subject property by different valuers. The observed difference between different valuer’s values is known as variance and is a very different concept to the uncertainty pertaining to the individual valuation. The problem with variance is that information pertaining to it either has to be set up artificially with a number of valuers asked to provide valuation on a common set of properties (see Crosby et al., 1998), or the analysis relates to valuations carried out at different points of time in the market. The outcomes of such studies varies substantially and in essence simply reports that different valuers have different ideas and thus produce different valuation figures. This is a very different concept to the uncertainty pertaining to the individual valuations within the study. The former deals with uncertainty (as expressed as variance) in output, the latter is a reference to the uncertainty pertaining to the inputs that go into the valuation to produce the specific valuation figure reported.
The UK experience In March 1994 the Mallinson Working Party on commercial property valuations produced its report outlining a number of initiatives that the RICS should undertake to help improve the standing of the valuation surveyor within the business world. There were 43 recommendations made by Mallinson, 42 of which have been acted on. The remaining recommendation, recommendation 34 proposes: Mallinson Recommendation 34 Common professional standards and methods should be developed for measuring and expressing valuation uncertainty.
This recommendation was re-addressed by the RICS Carlsberg report in 2002. Similarly, French and Mallinson (2000) proffered the use of normal probability distributions in the process and argued that: “Normal uncertainty”[2] is a universal and an unsurprising fact of property valuation. The open acknowledgement of that fact, and transparent management of its implications, will enhance both the credibility and the reputation of valuers. More than that, and of even greater importance, it will enhance the utility of valuations.
There will always be a degree of uncertainty in any valuation, but it should be incumbent on the valuer to report “abnormal uncertainty”. This arises when some particular condition of the market or the property leads to the valuer being unable to value with the confidence of accuracy that might normally be expected. But this paper is predominantly concerned with “normal uncertainty”, which hereafter we will term only as “uncertainty”. The principal problem as argued by the Mallinson Report is that that all valuations are uncertain. A valuation figure is an individual valuer’s estimate of the exchange price in the market place; it is an expert’s opinion. Despite this, clients and third parties tend to view the valuation figure as fact. Oddly, such a view does not prevail in other areas of asset valuation; all players in the stock market and, indeed the chattels and fine art market, are fully aware that the valuation is only an estimate and may not correspond with the final sale price. Yet, for real estate, there is general belief that valuations are final and exact. There is very little understanding of the uncertainty pertaining to them and that the uncertainty will vary according to market conditions and property type. Historically, the only reference to uncertainty in the RICS’ Red Book (RICS, 1996)[3] is a specific reference to “abnormal uncertainty”. Uncertainty and abnormal uncertainty Abnormal uncertainty was a concept that was included in the 1996 Red Book in PS 7.5.31. (Valuation Reports). It suggested that abnormal uncertainty might occur when there is a significant concern about market conditions such as times of financial turmoil. Alternatively the abnormal uncertainty may be property specific and related to impending litigation (such as major rent review case) or in relation to the property type (maybe the building is of an unusual size). Wherever the valuer considers that the range of uncertainty may be greater than normal then the valuer should refer in report to specific circumstances and/or lack of information, so that the client can judge the significance of the uncertainty in relation to the estimated capital value.
The uncertainty of valuation
487
JPIF 22,6
488
The odd point of this reference is that it alludes to “uncertainty greater than normal”, yet until the new edition of the Red Book (RICS, 2003) there was no reference in any RICS publication (apart from recommendations contained in The Carsberg Report and the Mallinson Report) to normal uncertainty. This matter was revisited in The Carsberg Report in 2002. The Carsberg Report The RICS set up the Carsberg Committee to respond to research carried out by The University of Reading and Nottingham Trent University (2001) on the impact of client influence on (investment) valuations. Although the Reading/Trent research was principally concerned with how valuations influenced the workings of the (property investment) market and specifically in the fund market, Carsberg expanded the brief of his response to encompass all issues that he felt were pertinent to the interpretation and use of valuations in all circumstances. One of the areas that he considered was the reporting of uncertainty in valuation and he made specific recommendations therewith (RICS, 2002): Carsberg Recommendation 15 RICS should commission work to establish an acceptable method by which uncertainty could be expressed in a manner which will be helpful and will not confuse users of the valuation. RICS should also seek to agree with appropriate representative bodies of those commissioning and using third party valuations the circumstances and format in which the valuer would convey uncertainty.
This recommendation follows on directly from Mallinson and embraces the same view that uncertainty should be reported to enhance the decision-making process and aide the valuation users’ understanding of the valuation. It was the view of Carsberg that the RICS should commission work to establish an acceptable method of expressing the inherent level of uncertainty within a valuation. This has been embraced by the Property Valuation Forum (PVF), which has run a number of seminars to consider the market response to such a proposal. One of the outcomes of this consultative process was the introduction of Guidance Note 5 in the 2003 edition of the RICS Appraisal and Valuation Standards. They have considered the ways in which uncertainty can be reported to the client and have identified three possible approaches: (1) Verbal reporting. The valuer articulates the uncertainty pertaining to the valuation in words within the report. (2) Ranking. The valuer allocates a “rank” to the valuation on a prescribed agreed basis. This may be numerical or alphabetical (i.e. 1, 2 or 3; A, B or C). (3) Statistical. The valuer conveys the uncertainty pertaining to the valuation by the use of recognised statistical information such as central tendency and/or standard deviation. This paper will concentrate on the third option noted, but that is not suggesting that either of the other options is not appropriate. Indeed, the current RICS guidance suggests that verbal reporting is the preferred mechanism of alerting the client to the uncertainty within the valuation.
Verbal reporting In the 2003 edition of the RICS Appraisal and Valuation Standards, there is a guidance note relating to uncertainty in valuation. In this guidance (RICS, 2003, GN5) it states that:
The uncertainty of valuation
All valuations are opinions of the price that would be achieved at the valuation date. The degree of certainty will vary significantly. These variations can arise because of the inherent features of the property, the market place or the in the information available to the valuer. Where uncertainty could have a material effect on the valuation, the valuer should draw attention to this, indicating the cause of the uncertainty and the degree to which this is reflected in the reported valuation.
489
Yet, contrary to the recommendations of Mallinson and Carsberg, there is no suggestion of a standard way of reporting this uncertainty to the client. By inference, the guidance note is suggesting that the valuer reports uncertainty in valuation to the client in a form of words within the report but it does not suggest an acceptable form of words nor any prescribed format for the measurement of the said uncertainty. Ranking A further option for reporting risk, as suggested by Mallinson Committee (1994), which was considered by French (1996) and developed by Hutchison et al. (2004) is to provide a simple risk score. The premise in this case is that the valuation could be provided as an indication of the risk of variance (say “1” for a low risk of variation to “4” for high risk of variation). The problem with this approach is that it possibly conveys a perception of “good” and “bad” valuations. When, it is not the veracity of valuation that is in question but the certainty of the specific figure. It may be a tenuous distinction, but one that could lead to significant misinterpretations in the market. If fully understood, this could be a useful and workable solution, particularly as it would be very easy to develop the ranking of individual property scores into a portfolio average. However, again for reasons noted above, this option is not considered further in this paper. Valuation and market sentiment The simple premise is that a valuation is a pricing model that, depending upon the implicit or explicit nature of the module used, identifies market sentiment towards pricing by a number of benchmarks (e.g. The capitalisation rate, the target rate (equated yield, market rent, market growth expectations, exit yields etc). The valuer will use the benchmark figure that he/she feels is most appropriate (most probable?) but he/she will not be 100 per cent confident that each of the figures used is exact. There will be a degree of uncertainty pertaining to each of the inputs. For the purposes of this paper we are seeking to identify the substance and the characteristics of the uncertainty which lies in the valuer’s mind as he or she attempts to assess the hypothetical purchaser’s view of the inputs involved. Thus we need to address the probability and range relating to the inputs not the output. A single valuation figure still needs to be provided, but an understanding of the uncertainty relating to the inputs used in the model will allow the valuer to report the uncertainty related to that specific single valuation figure. As both Carsberg and Mallinson suggest in their respective recommendations (15 and 34 respectively), the aim is to establish an acceptable method by which uncertainty could be expressed in a uniform and useful manner. Mallinson and French (2000)
JPIF 22,6
490
suggest that the solution must lie in the creation of some format description, accepted as a norm, which conveys the essence with simplicity, but is capable of expansion and interpretation. This would need to be presented in a prescribed professional standard, and would always be appended to a valuation figure. In its simplest form this would be the mean expectation of value (based on the varying probability of the inputs) plus the variation pertaining to that value within that one valuation (not variance of value between different valuers). This is effectively the best estimate plus standard deviation. Mallinson and French (2000) argued that there are six items of information that must be conveyed. (1) the single figure valuation; (2) the range of the most likely observation; (3) the probability of the most likely observation; (4) the range of higher probability; (5) the range of 100 per cent probability; and (6) the skewness of probabilities. However, this is a representation of the uncertainty of the output; and the figures generated are dependent on input benchmarks and the uncertainty relating to each of those variables. In both cases the underlying information will be represented by normal of bell distributions, skewed or otherwise. A simpler alternative may be to report the figures as a stated absolute range on a triangular basis; most probable, best and worst. This pragmatic point will be revisited once we have discussed the application of such an approach utilising probability distributions. Probability and valuation As noted above there is a significant difference between the use of probability in looking at the range of possible outcomes between the values produced by different valuers and the range of outcomes that would be produced by an individual valuer due to the uncertainty she or he may have in the benchmarks which are utilised in the valuation model. In this paper we are concerned with the second interpretation of uncertainty – the uncertainty of the inputs. As discussed previously, even the simplest of valuations there are likely to be a number of variables that the valuer must assess. For example, in a vacant possession office valuation, even if the office is similar to others which have been sold recently, the valuer must assess, through the eyes of the hypothetical purchaser, slight differences in location, the time since the last transaction, differences in standards of fitting, and so on. This is normally done through the use of a comparative benchmark; in the case of implicit valuations, the all risk yield (ARY) or the property yield. Through the use of a single yield indicator the valuer will assess the capital value of the property by a multiplier (years purchase or YP) derived from the yield, which is then applied to the market rent. In such a model, there are only two variables: the rent and the yield. However, if we assume that the initial rent has already been agreed, then the capitalisation model relies on only one variable; the yield.
The valuer will have taken a view on the appropriate yield by an analysis of comparables of the sale of similar properties. Assuming that he or she analyses, say, 20 previous transactions they will have a database of 20 observed yields which will form the foundation of the valuer’s judgement of the appropriate yield to be applied to the subject property. This is not a mathematical exercise but a heuristic approach, and the valuer’s judgement of the uncertainty pertaining to his or her final choice of yield will not be a direct correlation to the range of the observed yields. It will, however, be influenced by the perceived robustness of the database. If the market is strong and there are a lot of transactional data available, it is likely that the observations will be closely aligned and that the range of the observed yields will be small. This is because available data are both more comparable to the subject property and because the transaction dates are more likely to be closer to the valuation date. However, as market conditions deteriorate, the number of direct comparables sales falls and the valuer has to rely on observed transactions that are less comparable in terms of location, specification and time. Here the range of observed yields will be greater. In each case the valuer will choose a yield that he or she believes is the most appropriate. It is not directly a mean of the observations, or a mode. Indeed, as the final choice of yield will be influenced by how the valuer believes that the market has moved since the transaction dates of the comparables, the final choice of yield may not be the same as any of the observations. The process is not a science; it is a process of judgement and expert analysis. The valuer will identify the yield that he or she feels is most appropriate and use that figure to derive the rental multiplier for the valuation model. The model will produce a single answer based on the single point estimation of the inputs. Yet, the valuer will not be 100 per cent certain of the input figure. In effect, they will ascribe a degree of uncertainty to their belief in the input variable being “correct”. This is a subjective probability and will vary according to the confidence level that they feel applies for that variable, in this case the ARY. This subjective probability is currently not quantified within the model, although an expression of the valuer’s uncertainty may be articulated in market commentary accompanying the valuation. However, it would be possible to ascribe a probability distribution to this variable in accordance with the valuer’s perception of market conditions. For more complex properties the number of variables will multiply. In order to produce the valuation, the valuer must weigh all the variables, using his or her skill and experience, and decide on the most probable conclusion for each variable that would then feed through to the final valuation figure. A more complex pricing model with multiple variables will be considered in a later paper. The principal valuation model used in the UK for a rack-rented or fully-let property is the income capitalisation model (sometimes referred to as the “traditional method”). This model requires the valuer (for a vacant property) to estimate the best rent readily achievable for that property and the corresponding all risk yield or capitalisation rate. The capitalisation rate is derived by the analysis of other similar sales and is effectively the initial yield (first year’s income divided by price) of the comparable properties duly adjusted to reflect the specific circumstances of the subject property.
The uncertainty of valuation
491
JPIF 22,6
492
The reciprocal of the ARY will indicate the multiplier that converts market rent to capital value. For instance an ARY of 5 per cent will give a multiplier of 20 (called the YP), which suggests that in the market we would expect the property to sell for 20 times the rent. This can be illustrated in Table I, where an office building has just been let for a market rental of £10,000. The valuer’s assessment of the ARY is 5 per cent and, for the explicit model, the corresponding market target rate (or equated yield) is 8 per cent. This produces a capital value of £200,000. Inputs and probability distributions In the previous section, it was suggested that heuristic process that the valuer would follow in the implicit model would be to assess the market for comparable sales and derive a ARY appropriate for the subject property by an intuitive interpretation of the range of yields produced by the comparables. In our example, the valuer felt that a 5 per cent yield was appropriate for the subject property and that this choice of yield would reflect all the risks and growth potential for that property and thus would produce the appropriate value (estimate of price) in the market place at the valuation date. However, as discussed, the valuer will have a “view” on how certain he or she is about that variable and, depending on the state of the market, how likely he or she thinks that the yield might be higher or lower. If the market is relatively stable then the likelihood of the yield being higher than 5 per cent should be equal to the likelihood of the yield be lower that 5 per cent. The degree by which it might deviate from the assumed figure is again dependent on the market conditions. If there is sufficient market evidence, the valuer will feel more certain of the market conditions and thus more confident in the ARY adopted; the corresponding range, above and below the adopted figure, will therefore be proportionally less than the range were there more uncertainty in the adopted figure. In statistical terms this thought process can be represented by a probability distribution. Equal likelihood of the adopted figure being higher or lower would be a symmetrical distribution; an unequal probability would result in a skewed distribution. Each input into the model will be represented by a probability density function (pdf), which allows us to consider a range of values instead of a single figure. The single figure is the most likely value, the uncertainty pertaining to that figure being represented by extent of the range around that figure. Normal probability distribution French and Mallinson suggested that the appropriate probability distribution would be a normal or bell distribution. This is a distribution that is symmetrical around a central tendency; a non-skewed distribution will have the mean, the mode and the median n
Table I. Implicit valuation (implicit (traditional) capitalisation model)
Market rent (£) YP perp @ 5.00 per cent Capital value (£)
10,000 20 200,000
coinciding. In our analysis the most likely figure will be represented by the central figure (the mean) and the uncertainty by the range around that number. There is equal probability that the observed figure will be above or below the central assumed figure. The majority (99.74 per cent) of the possible observations will lie within plus or minus three standard deviations of the mean. The standard deviation is a measure of how widely values are dispersed from the average value (the mean). The exact standard deviation will vary according to the uncertainty pertaining to the average value; the greater the uncertainty the higher the standard deviation. However, there is a small probability that the observed figure will lie outside the three standard deviation range and, as the distribution is open ended, it is possible that the observed value will be in found in the infinite tails of the normal distribution. The distribution is continuous. While the normal distribution is the most readily understood probability distribution in statistical terms, as it can be modelled with reference only to the mean and standard deviation, it does not fit closely with the heuristic processes of the market place. Obviously the valuer is happy to determine the most likely (mean) figure for the ARY, but they would not think about the range either side of the mean as a percentage variation, which is the normal expression of the standard deviation. The valuer is more likely to think in terms of absolute figures either side of the mean. Triangular probability distribution This representation is much more akin to the thought process of the valuer as it requires the user to provide three absolute figures; the most likely, the maximum and the minimum. This is a closed distribution and can be symmetrical or skewed. This is a more useful tool as it information requirements mirror the likely thought process of an expert; in this case the valuer. However, the advantage of the triangular distribution, its simplicity, is also its disadvantage. In reality the observed distributions will tend toward a normal distribution and thus by imposing a definite limit to the range, connected by a straight line relationship, it suggests that the observed values will not be concentrated around the mean and thus the outcomes are likely to have a greater spread. In statistical terms, typically the triangular distribution overestimates the variance. Although there are other probability distributions that may be considered (e.g. lognormal, Beta, etc), the purpose of this paper is to review approaches that might be readily acceptable by the profession. This requires the approach to be easy to implement, pragmatic and readily understood. Applications of probability to the capitalisation model In Table I, we have shown the valuation of an office building at an initial agreed rent of £10,000. The valuation can be carried out implicitly to produce the capital value of £200,000. Within the implicit capitalisation model for a “just let” building, the only uncertain variable is the ARY. By using Crystal Ball, we are able to ascribe a probability distribution to that input and, by using a Monte Carlo analysis, test the veracity of the £200,000 figure.
The uncertainty of valuation
493
JPIF 22,6
494
Monte Carlo simulation is a re-sampling iterative process. In simple terms it changes the input in the calculation by randomly choosing a figure within the defined probability distribution. It then calculates the corresponding value using that chosen input and records that value. It then repeats the process by randomly choosing another input figure. It will continue to do this until the chosen number of iterations, normally several thousand, is complete. The output is expressed as the mean of all the calculated values. It provides a structured approach that allows the user to incorporate uncertainty into the analysis in a relatively simple form. Because each input is defined by the chosen probability density function. If there is more than one variable to be analysed, then it is possible to allow for any interrelationship between the chosen variables. For example in a DCF model, rental growth and exit yield should be negatively correlated. The normal distribution – Crystal Ball The capitalisation model works well when the possible inputs are normally distributed within a tight distribution. In Figure 1, we have a mean (expected ARY) of 5 per cent. Crystal Ball then sets the standard deviation as 0.5 per cent (10 per cent of the mean figure) and runs the Monte Carlo simulation 50,000[4] times. This produces the outcome shown in Figure 2. In numerical terms this can be represented as shown in Table II.
Figure 1. Normal distribution; fixed standard deviation
Figure 2. Output distribution
Forecast: capitalisation Summary: Display range is from £147,472 to £256,657 Entire range is from £137,000 to £335,738 After 50,000 trials, the standard error of the mean is £93 Statistics Trials Mean (£) Median (£) Standard deviation (£) Skewness
The uncertainty of valuation Value 50,000 201,973 199,944 20,871 0.63
495 Table II. Statistical data
Here it can be seen that the expected mean (capital value) of £201,973 is not significantly different from the £200,000 produced by the discreet use of the implicit model. But the advantage of the Monte Carlo simulation (using Crystal Ball) is that provides additional information about the certainty of the result. In this case, the standard deviation (of £20,871) is a representation of the uncertainty. The skewness (of 0.63) represents the degree of asymmetry of the distribution around its mean. Here the positive skewness indicates a distribution with an asymmetric tail extending toward more positive values. Whereas a negative skewness would indicate a distribution with an asymmetric tail extending toward more negative values. However, the normal distribution has the pragmatic drawback that the user is unlikely to express their view of uncertainty as a standard deviation. Instead, as previously discussed, it is likely that they would suggest an absolute range.
The triangular distribution – Crystal Ball Although the capitalisation model works best with the assumption of a normal distribution, it is more pragmatic that the choice of model should be driven by the ease of articulating the uncertainty. A valuer is likely to say that the expected ARY is 5 per cent, although it may be as low as 4.5 per cent or as high as 5.5 per cent. This can be directly inputted into Crystal Ball as a most likely, maximum and minimum and again run for 50,000 simulations (see Figures 3 and 4 and Table III). Again the observed mean of £200,369 is not dissimilar from the original result of £200,000. However the standard deviation is lower at £8,181 as the input range was curtailed between 4.5 per cent and 5.5 per cent.
Figure 3. Triangular distribution; likeliest, maximum and minimum
JPIF 22,6
496
Figure 4. Output distribution
Table III. Statistical data
Forecast: capitalisation Summary: Display range is from £182,611 to £220,262 Entire range is from £182,053 to £222,066 After 50,000 trials, the standard error of the mean is £37 Statistics Trials Mean (£) Median (£) Mode Standard deviation (£) Skewness
Value 50,000 200,369 200,051 – 8,181 0.17
Applications of probability to a second related variable As shown in Table I, the implicit capitalisation model will produce a capital value of £200,000. However, this assumed that the property had just been let and the rent had been fixed. As noted earlier, it is possible when undertaking the valuation of a vacant property that the valuer will have to estimate the market rent in addition to the capitalisation rate. We can therefore extend the model to incorporate the extra variable of rent. However, as noted, these inputs are not independent and thus it is necessary not only to consider the range of uncertainty, but also the inter-relationship of the two variables, rent and ARY. This is illustrated in Figures 5 and 6 and Table IV. Here we have decided to apply the triangular distribution. Although it is recognised that the normal distribution is more statistically robust, the triangular representation is much more akin to the thought process of the valuer as it requires the user to provide three absolute figures: the most likely, the maximum and the minimum for each input. The triangular distribution – Crystal Ball In this example the valuer is required to identify the ARY and the corresponding market rental. In this case, the ARY was estimated to be 5 per cent, a low of 4.5 per cent and a high of 5.5 per cent. Correspondingly the rental was a low of £9,500, a high of
The uncertainty of valuation
497
Figure 5. Triangular distribution; likeliest, maximum and minimum
Figure 6. Output distribution
JPIF 22,6
498 Table IV. Statistical data
Forecast: valuation Summary: Display range is from £170,000 to £230,000 Entire range is from £172,946 to £232,217 After 50,000 trials, the standard error of the mean is £46 Statistics Trials Mean (£) Median (£) Standard deviation (£) Skewness
Value 50,000 200,373 199,990 10,253 0.17
£10,500 and a most likely of £10,000. However as the two variables are interrelated, it has been necessary to add in a correlation factor between the two variables. The relationship between rent and ARY is a negative relationship. As rents increase in the market we would expect the ARY to fall. This is because as rents start increasing, the market perception will be that property is a more attractive asset than it was and thus the multiplier (YP) against rent will increase. As the YP is the reciprocal of the ARY, then the ARY decreases as property becomes more attractive and hence the negative correlation. However, the correlation is not 100 per cent. That is, as the rent increase by (say) 10 per cent, the ARY does fall by the same percentage rate. It would be possible t analyse past data to derive a robust correlation coefficient but indicatively a correlation of 2 0.32 (i.e. the ARY falls by 32 per cent for every 100 per cent increase in rent) is felt to be appropriate. This is applied to the variables. Again, these inputs were fed into Crystal Ball and run for 50,000 simulations. As expected the outcome of this simulation is to increase the output range. There is more uncertainty in the inputs, which will lead to more uncertainty in the outputs. By a comparison of the output range (or standard deviation), a client would be able to realise that the valuation of a property with vacant possession is less certain than the valuation of a corresponding property that has just been let (and hence the rent is already fixed). This is a facet that is currently overlooked in valuations that illustrates quite succinctly the importance in conveying uncertainty to a client in a way in which they can understand. Conclusion As with all models, there are advantages and disadvantages to each of the distributions chosen. Similarly, each could be adjusted to reflect better the market conditions at any point of time by expanding or contracting the range and varying the skewness. There will always be debate about the appropriateness of the distribution chosen. However, for ease of use by the profession, we believe that the triangular approach is the most appropriate given the objectives. The objective of both the Mallinson and Carsberg reports is to establish an acceptable method by which uncertainty could be expressed in a uniform and useful manner. This would require agreement on the expression of the uncertainty of the inputs and agreement on the output information that must be conveyed with each valuation.
More work will be required to agree on these issues, but the use of a Monte Carlo model, we believe is sufficiently easy, robust and accessible for the profession to consider as a possible means of expressing uncertainty in valuation. Notes 1. An alternative would be to use @Risk which is a very similar software package. 2. This paper is predominantly concerned with “normal uncertainty”, which is hereafter we will term only as “uncertainty” 3. Now superseded by the RICS Appraisal and Valuation Standards (RICS, 2003). 4. We chose 50,000 iterations as it is sufficient to allow consistent results between different simulations. References Brown, G. (1991), Property Investment and the Capital Market, E. & F.N. Spon, London. Byrne, P. (1995), “Fuzzy analysis, a vague way of dealing with uncertainty in real estate analysis?”, Journal of Property Valuation and Investment, Vol. 13 No. 3, pp. 22-41. Byrne, P. and Cadman, D. (1984), Risk, Uncertainty and Decision-making in Property Development, 1st ed., E. & F.N. Spon, London. Byrne, P. and Cadman, D. (1996), Risk, Uncertainty and Decision-making in Property Development, 2nd ed., E. & F.N. Spon, London. Crosby, N., Lavers, A. and Murdoch, J. (1998), “Property valuation variance and the ‘margin of error’ in the UK”, Journal of Property Research, Vol. 15 No. 4, pp. 305-30. Evans, A.H. (1992), “Monte Carlo Analysis: a practical application to development appraisal”, Journal of Property Finance, Vol. 3 No. 2, pp. 271-81. Fraser, W.D. (2004), Cash-Flow Appraisal for Property Investment, Palgrave Macmillan, New York, NY. French, N. (1996), “Investment valuations: developments from the Mallinson Report”, Journal of Property Valuation and Investment, Vol. 14 No. 5, pp. 48-58. Hargitay, S. and Yu, S.M. (1993), Propery Investment Decisions: A Quantitative Approach, E. & F.N. Spon, London. Hutchinson, N., Adair, A. and Leheny, I. (2004), “The reporting of risk in real estate appraisal: property risk scoring”, paper presented at the European Real Estate Society (ERES) Annual Conference, Milan. Kelliher, C.F. and Mahoney, L.S. (2000), “Using Monte Carlo simulation to improve long-term investments decisions”, The Appraisal Journal, January, pp. 44-56. Mallinson Committee (1994), Commercial Property Valuations (Mallinson Report), Royal Institution of Chartered Surveyors, London. Mallinson, M. and French, N. (2000), “Uncertainty in property valuation: the nature and relevance of uncertainty and how it might be measured and reported”, Journal of Property Investment and Finance, Vol. 18 No. 1, pp. 13-32. Pellat, P.G.K. (1972), “The analysis of real estate investments under uncertainty”, Journal of Finance, Vol. 27, pp. 459-71. Pyrrh, S.A. (1973), “A computer simulation model to measure risk in real estate investment”, Journal of the American Real Estate and Urban Economics Association (AREUEA), Vol. 1 No. 1, pp. 48-78.
The uncertainty of valuation
499
JPIF 22,6
500
Robinson, J. (1989), Property Valuation and Investment Analysis: A Cash Flow Approach, Law Book Company, Sydney. Royal Institution of Chartered Surveyors (RICS) (1996), RICS Appraisal and Valuation Manual, RICS, London. Royal Institution of Chartered Surveyors (RICS) (2002), The Carsberg Report, RICS, London. Royal Institution of Chartered Surveyors (RICS) (2003), RICS Appraisal and Valuation Standards, RICS, London. Whipple, R.T.M. (1988), “Evaluation development projects”, Journal of Valuation, Vol. 6, No. 3, pp. 253-86. Wooford, L.E. (1978), “A simulation approach to the appraisal of income producing real estate”, Journal of the American Real Estate and Urban Economics Association (AREUEA), Vol. 6 No. 4, pp. 370-93.
Further reading Baum, A. and Crosby, N. (1991), Property Investment Appraisal, Routledge, London. Baum, A., Crosby, N. and MacGregor, B. (1996), “Price formation, mispricing and investment analysis in the property market”, Journal of Property Valuation and Investment, Vol. 14 No. 1, pp. 36-49. Baum, A., Mackmin, D. and Nunnington, N. (1997), The Income Approach to Property Valuation, 4th ed., Thompson, London. Brown, G. and Matysiak, G. (2000), Real Estate Investment. A Capital Market Approach, Prentice-Hall, London. Fisher, J.D. and Martin, R.S. (1994), Income Property Valuation, Dearborn Financial Publishing, Chicago, IL. French, N. (1997), “Market information management for better valuations: concepts and definitions of price and worth”, Journal of Property Valuation and Investment, Vol. 15 No. 5, pp. 403-11. French, N. and Ward, C. (1995), “Valuation and arbitrage”, Journal of Property Research, Vol. 12 No. 1, pp. 1-11. French, N. and Ward, C. (1996), “Applications of the arbitrage method of valuation”, Journal of Property Research, Vol. 13 No. 1, pp. 47-57. MacLeary, A.R. and Nanthakumaran, N. (Eds) (1998), Property Investment Theory, E. & F.N. Spon, London. Peto, R. (1997), “Market information management for better valuations: data availability and application”, Journal of Property Valuation and Investment, Vol. 15 No. 5, pp. 411-22. Peto, R., French, N. and Bowman, G. (1996), “Price and worth: developments in valuation methodology”, Journal of Property Valuation and Investment, Vol. 14 No. 4, pp. 79-100.
The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-578X.htm
Different risk measures: different portfolio compositions?
Different risk measures
Peter Byrne and Stephen Lee Centre for Real Estate Research, The University of Reading Business School, Reading, UK Keywords Risk analysis, Modelling, Portfolio investment
501 Received June 2004 Accepted July 2004
Abstract Traditionally, the measure of risk used in portfolio optimisation models is the variance. However, alternative measures of risk have many theoretical and practical advantages and it is peculiar therefore that they are not used more frequently. This may be because of the difficulty in deciding which measure of risk is best and any attempt to compare different risk measures may be a futile exercise until a common risk measure can be identified. To overcome this, another approach is considered, comparing the portfolio holdings produced by different risk measures, rather than the risk return trade-off. In this way we can see whether the risk measures used produce asset allocations that are essentially the same or very different. The results indicate that the portfolio compositions produced by different risk measures vary quite markedly from measure to measure. These findings have a practical consequence for the investor or fund manager because they suggest that the choice of model depends very much on the individual’s attitude to risk rather than any theoretical and/or practical advantages of one model over another.
1. Introduction Selecting the appropriate portfolio of assets in which to invest is an essential component of real estate fund management. Although a large proportion of portfolio selection decisions are still taken on a qualitative basis, quantitative approaches to selection are increasingly being employed. Markowitz (1952) established a quantitative framework for asset selection into a portfolio that is now well known. In this it is assumed that asset returns follow a multivariate normal distribution or that investors have a quadratic utility function. This approach shows that characteristics of a portfolio of assets can be completely described by the mean and the variance (risk) and so is described as the mean-variance (MV) portfolio model. For a particular universe of assets, the set of portfolios of assets that offer the minimum risk for a given level of return form the efficient frontier. The portfolios on the efficient frontier can be found by quadratic programming and such problems can now be solved easily in spreadsheet programs (see Byrne and Lee, 1994a, b). The solutions are optimal and the selection process can be constrained by practical considerations, such upper and lower bounds, which can be written as linear constraints. The weakness of the MV approach, however, is that the underlying assumptions of multivariate normality or that investors have a quadratic utility function are not sustainable. This has led researchers to develop portfolio asset allocation models using other measures of risk that have many theoretical and practical advantages over MV. Even so, the MV approach remains the most popular approach to the asset allocation problem. This may be because deciding which measure of risk is “best” is still unresolved (Stone, 1973). Cheng and Wolverton (2001) for example, highlight the difficulty of comparing portfolios based on different risk criteria. They find that each approach
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 501-511 q Emerald Group Publishing Limited 1463-578X DOI 10.1108/14635780410569489
JPIF 22,6
502
produces results that minimise risk, but only in its own space. When the portfolio compositions of one risk measure are used to calculate the risk and return trade-off in another risk space the results are always “inferior” to the solutions produced inside that risk space. They argue that any attempt to find the portfolio model that offers the best risk return trade-off is likely to be futile until a common risk measure can be identified. In this paper, another approach, which overcomes this, is considered (Phillips, 1993). The portfolio asset holdings and weights produced by different risk measures are compared, rather than the conventional risk/return trade-off. In this way it is possible to see whether different risk measures produce asset allocations that are essentially the same or radically different. So, for instance, if the portfolio compositions produced by different risk models, i.e. the assets chosen and the weights assigned to the them, are essentially the same then there is little to be gained from using one risk measure or another. In contrast, if the portfolio compositions produced by the different risk models are substantially different, the choice of risk measure becomes crucial to the investor. The paper is structured as follows. The next section discusses the various risk measures used in the study. Section 3 presents the data. The following section provides a brief discussion of the various optimisation models used and shows the average results of the various optimisations. A number of similarity indices are then calculated and discussed in section 5. Section 6 concludes the paper. 2. Risk measures The variance remains the most commonly used risk measure in portfolio optimisation models. Markowitz (1952) showed that if risk is measured by the variance of returns and expected return by the mean of returns, then uncertain investments can be ordered by their ranking in MV space. The variance is defined as:
Var ¼
T 2 1X Rt R : T t¼1
ð1Þ
Although the MV model is the most popular approach, it relies on the assumptions that returns are either normally distributed or that the investor’s utility function is quadratic. If either of these conditions hold, it can be shown that choosing among risky investments is compatible with the maximisation of an investor’s expected utility (Tobin, 1958). Many authors have pointed out, however, that both of the assumptions underlying the MV model generally do not hold – either theoretically or in practice. Apart from the criticisms of the assumptions underpinning the MV model, it has been argued also that the use of the variance as a measure of risk implies that investors are indifferent between returns above and below the mean. Clearly, however, most risk-averse investors are generally more concerned with risk below some target level of return, be it the mean or some other benchmark. In order to overcome the difficulties associated with the MV model, Markowitz (1952) and others proposed the semi-variance (SV). The SV concentrates on the returns below the mean (expected return) so that the SV is defined as:
T 2 1X SV ¼ min 0; Rt E ð RÞ subject to Rt , E ð RÞ: T t¼1
ð2Þ
Bawa (1975) and Bawa and Lindenberg (1977) generalised this idea by suggesting models based on lower partial moments (LPM) over n orders (see for example, Sing and Ong, 2000). LPMs of order 2 are measures of portfolio risk that focus on returns below some target level, so that, for example, the SV is just a special case LPM when Rt equals the E(R). The LPM (n ¼ 2) is thus: LPM ¼
T 2 1X min 0; Rt Rt again; subject to Rt , Rt : T t¼1
ð3Þ
In the investment literature, the target Rt is usually the minimum return that the investor would be willing to receive. The target is often set to the risk-free rate or it could be zero: that is negative returns are to be avoided[1]. Alternatively, the target could be set to the return of the benchmark index (B) that the manager is expected to outperform. In that case a new variable (R 2 B) can be defined and the target set to zero. A target of zero can be considered as the general case and is used as the target for the lower partial moment: (LPMZ). The MV and SV models require the use of complex non-linear numerical algorithms to solve the portfolio problem. The practical application of such models was severely limited until computers were powerful enough to handle even the smallest problems. Sharpe (1971) commented that if the portfolio problem could be formulated as a linear programming problem, the prospect for practical application would be greatly enhanced. Young (1998) has proposed such a solution based on the “minimax” (MM) rule. The MM rule has a long tradition in models of uncertainty because, in situations with conflicting alternatives, the most rational choice is that which seeks to minimise the maximum loss (negative gain). Given an historic time series of returns, the optimum portfolio under the MM rule is defined as that which would minimise the maximum loss over all past periods, subject to a restriction that some minimum average return is achieved across the observed time periods. Hence the LP can be defined as follows: max M P subject to: N X
wi r it M P $ 0; t ¼ 1; . . .; T;
i¼1 N X i¼1
and as usual:
wi ri $ B;
ð4Þ
Different risk measures
503
N X
JPIF 22,6
wi ¼ 1;
i¼1
and: wi $ 0; i ¼ 1; . . .:; N :
504
The MM portfolio model has a number of advantages over the MV model. Young shows that the MM rule corresponds approximately to an expected utility function that is more extreme than that implied by the MV rule, having a strong absolute aversion to downside risk. The MM approach is akin to the LPM rule, especially if the level of maximum loss that the investor requires is set to zero or the risk free rate. Zero is again used as the target rate here. MM linear programming solutions can also accommodate complex decision variables such as integer values, e.g. fixed transaction costs, which are difficult to incorporate into the quadratic programming model used by the MV rule. In addition, the MM rule has logical advantages over the MV rule if returns are non-normally distributed. Based on the results of simulation studies, Young concludes that the MM rule provides a convenient and useful approach to portfolio selection. All these measures of risk are sensitive to outliers in the data because the mean differences are squared. Squaring gives such outliers a disproportionate influence in the calculation of the measures of risk. In a similar way, LPM measures are sensitive to observations that are distant from their target. The MM rule also will be affected strongly by such outliers, because it specifically minimises the maximum negative return. A measure of variability that is less sensitive to outliers is the mean absolute deviation (MAD). This prompted the development of portfolio optimisations that use the MAD as a measure risk (Konno, 1989). Such models have a number of advantages. First, Konno and Yamazaki (1991) show that the MAD approach is equivalent to the MV model if the returns are multivariate normally distributed. Second, the MAD model produces optimal portfolios without the need to calculate the covariance matrix and so can be used in situations when N, the number of assets, is greater than T, the number of time periods over which the analysis is performed. Finally, Konno and Shirakawa (1994) show that the MAD model can handle large problems in real time. However a limitation of this approach is that the computational savings from the use of MAD objective functions may in some cases be outweighed by the loss of information from the (unused) covariance matrix (Simaan, 1997). The MAD is defined by: MAD ¼
T 1X j Rt E ð RÞj: T t¼1
ð5Þ
Other portfolio models using different measures of risk have been and continue to be developed. MV, SV, LPMZ, MM and MAD models are considered in this study to keep the comparisons to a reasonable number. 3. Data The data used in this study are the total monthly returns for the ten market segment indices used by the Investment Property Databank (IPD) in their standard performance
analysis reports to investors. The IPD monthly indices are based on the individual property data from 55 institutional investors and cover more than 2,500 properties valued at £2.7 billion at the end of 2002. Details of the construction methods are available in IPD (2002). Tests performed by IPD have suggested that this ten-segment categorisation maximises the explanatory variance in returns across individual properties and is the most effective split for asset allocation optimisation (Frodsham and Key, 1996). The ten market segments are: (1) standard retail Southeast (SRSE); (2) standard retail rest of UK (SRRUK); (3) shopping centres (SHC); (4) retail warehouses (RW); (5) offices in the city of London (OCITY); (6) offices in the West End (OWE); (7) offices rest of Southeast (ORSE); (8) offices rest of UK (ORUK); (9) industrials southern and eastern (ISE); and (10) industrials rest of UK (IRUK).
Different risk measures
505
The data cover the period January 1987 to December 2002, a total of 192 monthly returns. Summary statistics for the data are presented in Table I. Table I shows that the segment with the highest average return overall was IRUK, while that the segment with the lowest risk was SRRUK. However, the segment with the highest overall risk, OCITY, also had the lowest average returns. In general the market segments display positive skewness, with IRUK showing the highest value, but two market segments SHC and OCITY do display negative skewness, OCITY showing significant negative skewness. All the assets display significant positive kurtosis, (i.e. they are leptokurtic), that is more peaked than the normal distribution. This is a well-known feature of real estate data. Lizieri and Ward (2000) argue that this comes from the presence of a high proportion of zero returns and too few larger negative and positive returns, which can be attributed to the thinly traded nature of direct property, where new information is infrequent and is only slowly impounded into valuations. Consequently, Jarque-Bera tests show that the data were not normally distributed at the usual levels of significance.
Mean SD Skewness Kurtosis JB test
SRSE
SRRUK
SHC
RW
OCITY
OWE
ORSE
ORUK
ISE
IRUK
0.72 0.8 0.93 4.88 56
0.71 0.72 1.26 8.11 259.2
0.81 0.86 20.68 6.86 134.1
1.12 1 0.08 6.57 102.4
0.49 1.55 -0.97 9.37 355.3
0.84 1.43 0.38 5.44 52.3
0.77 0.99 0.53 4.33 23.2
0.92 1.03 1.28 5.52 103
1.01 0.95 0.7 4.31 29.36
1.16 0.99 1.72 6.87 214.5
Table I. Statistics of the IPD ten-segment categorisation: monthly data 1987:1-2002:12
JPIF 22,6
506
Table II. Average portfolio holdings and risk/ return trade-off: 53 optimisations MR portfolio
Figure 1. Asset composition of minimum risk portfolios
4. Asset allocation comparisons In order to investigate the potential differences between the portfolio allocations produced by the alternative risk measures, a simple asset allocation problem is examined using the data discussed in the previous section. Specifically, this is for an investor whose portfolio is updated on a quarterly basis using the previous 36 month’s data (hence there are 53 points at which rebalancing occurs). Rather than considering the whole efficient frontier however, as Cheng (2001) does, the minimum risk (MR) portfolio is examined here since this optimisation involves only the measure of risk. The results for the 53 optimisations are shown in Table II. Table II contains the average portfolio holdings in each “asset” and the comparative risk of the optimisations in MV space. Figure 1 and Table II show that the mean allocations produced by the alternative risk models are significantly different to the MV approach, except for those in the MAD model. This is to be expected because the data display considerable non-normality. In particular, the LPMZ shows a dramatic shift in allocation away from the market
MV MAD MM SV LPMZ
SRSE
SRRUK
12.1 10.8 – 21.9 9.3
20.8 19.8 0.4 13.8 3
Average percentage holding in market segment SHC RW OCITY OWE ORSE ORUK 14.4 15.5 2.1 4.8 3.6
0.5 0.2 19.3 0.2 28.3
9.2 9.7 3.1 45.5 –
7.1 7.6 12.8 1.4 3.9
3.1 3.3 0.3 2.6 3.8
22.2 24.6 12.4 8 6.8
ISE
IRUK
MV Risk
3.3 2.7 27.4 1.1 4
7.4 5.8 22.2 0.7 37.3
0.48 0.50 0.65 0.58 0.74
segment with significant negative skewness (OCITY) to the segment with significant positive skewness (IRUK). The MM model also shows a similar aversion to negatively skewed segments, with 63 per cent of the average portfolio allocation in ORUK, ISE and IRUK. This highlights the link between downside risk and negative skewness. The portfolio weights produced by the alternative risk models were then used to calculate the average risks and returns in MV space. The results in Table II show a number of features of interest. First, all the alternative risk measures produce risks that are “inferior” to those produced by MV analysis. This supports the conclusions of Cheng and Wolverton (2001) that the risk measures, although optimal in their own risk space, always produce results that are inferior in another risk space and the question of which risk measure is the “best” cannot be addressed adequately by such comparisons. Second, the results from the MAD and SV models are the closest to those of the MV solution. This agrees with Byrne and Lee (1999) who show that that the MAD approach produces portfolio weights and has risk/return tradeoffs that are similar to the MV model. 5. Asset allocation similarity To see whether similar portfolio allocations occur when different risk measures are used, two issues need to be addressed (Phillips, 1993). First, to what extent do the same assets appear in each optimisation model? Second, for those assets that are contained in each solution, to what extent do they appear in similar proportions? In order to assess this, the number of assets and their weights in each optimal portfolio were identified. From these it was possible to find those assets that are common, (overlap), in each model and also the extent to which the weights are similar between the different risk measures. The average results are presented in Table III. Panel A of Table III shows that on average the 53 MV solutions contained 4.1 assets, with a maximum of 7 and a minimum of 1. By comparison, the MAD solution was made up, on average, of 3.3 assets, with a maximum of 6 and a minimum of 1. Panel A shows that all the models produce average solutions which, based on a one-tailed t-test, are significantly less than the MV solutions. Panel B of Table III shows that the number of assets in the minimum risk portfolios that overlap, or are common to both the MV and MAD optimisations, is 3.1. This means that on average, 3.1 of the 4.1 assets that make up the MV portfolios are also part of the 3.3 assets that form the MAD portfolios. Similar calculations are made for the other risk measures. Panel B shows that the risk measure with the least assets in common with the others, is the LPMZ. Using these raw numbers portfolio overlap, weight and similarity indices were then calculated (see Phillips (1993) for details). As an example, the average portfolio overlap index for the MAD against the MV is defined as the ratio of the number of assets that overlap between the two risk measures to the average number of assets, in the union between the two risk measures. The union between the two risk measures is equal to the average number of assets in the MV solution (4.1) plus the number of assets in the MAD solution (3.3), minus the number of assets in common (3.1), which in this case is equal to ð4:1 þ 3:3 2 3:1Þ ¼ 4:3. Thus while on average the MV and MAD solutions contained (4.1) and (3.3) assets, only 3.1 were in common to both results, leaving 4.3 assets that appear in only one solution. The ratio of the overlap between the two risk measures to the union between the two risk measures is therefore 3.3/4.3¼ 72.4 per cent.
Different risk measures
507
JPIF 22,6
508
Table III. Average portfolio overlap, weight and similarity indices: compared with the minimum risk MV solution
MV
MAD
MM
SV
LPMZ
Panel A: average holdings Average Max Min
4.1 7 1
3.3 6 1
3.1 5 1
3.0 6 1
3.2 8 1
Panel B: holdings in common MV MAD MM SV LPMZ
4.1 3.1 1.4 2.5 1.6
3.3 1.1 2.2 1.1
3.1 1.0 1.4
3.0 0.9
3.3
Panel C: overlap indices MV MAD MM SV LPMZ
100.0 74.3 25.6 56.4 20.8
100.0 22.2 56.2 14.5
100.0 19.3 31.2
100.0 12.3
100.0
Panel D: weight indices MV MAD MM SV LPMZ
100.0 80.2 16.6 48.9 15.3
100.0 15.1 46.6 11.2
100.0 8.2 45.1
100.0 6.9
100.0
Panel E: similarity indices MV MAD MM SV LPMZ
100.0 60.5 7.4 31.7 7.0
100.0 5.2 29.9 4.1
100.0 3.0 15.3
100.0 1.8
100.0
Thus, on average, 72.4 per cent of holdings contained in the MV solution are also contained in the MAD solutions. Other portfolio overlap indices were calculated in the same way. The results are given in Panel C of Table III. The calculation of portfolio overlap indices only addresses one facet of the similarity or dissimilarity of portfolio compositions. When two portfolios contain exactly the same assets the portfolio overlap index will be 100 per cent. However, the weights within such portfolios could vary markedly. This has important investment implications. To test the similarity between the weights attached to assets held in common by two portfolios a portfolio weight index can be constructed. The index is measured by summing the minimum weight attached to each asset that overlaps two portfolio solutions. For example, column 2 of Table III shows a portfolio overlap index of 74.3 per cent between the MV and MAD models. On the other hand, the sum of the minimum weights found in the MR portfolios between the two risk measures for the holdings that are common in both solutions is 80.2 per cent. In other words, 80.2 per cent of the holdings in the MR portfolio of the MV solution are common to the MAD
solution. The portfolio weight indices for the alternative risk measures are calculated in the same way. The results are presented in Panel D of Table III. Multiplying the portfolio overlap indices by the portfolio weight indices gives the proportion of assets in common to both risk measures with similar weights, i.e. a portfolio similarity index. These are shown in Panel E of Table III. In the case of the MR portfolios for the MV and MAD solutions, 60.5 per cent of the holdings in the MV optimisation are also in the MAD optimisation with similar portfolio weights. The average portfolio similarity indices of the risk measures, compared with the alternatives, are calculated in the same way. Table III shows a number of features of interest. First, given the results in Table II, by comparison with the MV solutions, the MAD model shows the largest similarity with a value of 60.5 per cent, while LPMZ shows the least similarity; 7 per cent. Second, in comparison with the other models, LPMZ with MM show the largest similarity (15.3 per cent), but LPMZ with SV the least (1.8 per cent). In addition, Figure 2 shows that the average values (Table III: Panel E) hide a great deal of variability over the 53 optimisations. Nonetheless, the MAD values are closest to the MV solutions; closely followed by the SV results, with the MM and LPMZ models showing similarity values that are substantially lower than the MV optimisations. This may be the result of setting the target rate of return for these two risk measures to zero, whereas the target rate in the other three measures is the mean return over the period.
Different risk measures
509
6. Conclusions The measure of risk used traditionally in portfolio optimisation models is the variance. This is in spite of the fact that other measures of risk have many theoretical and practical advantages. Given the advantages that these other measures of risk present, it seems puzzling that they are not used more frequently. However, if it can be shown that these alternative approaches produce asset allocations that are essentially the
Figure 2. Similarity indices relative to MV portfolio compositions
JPIF 22,6
510
same as those produced by the MV model then there would be little to be gained from its continuing use. In order to investigate this issue this study tested the proposition that different measures of risk produce minimum risk portfolios that are essentially the same in terms of asset allocations, using monthly data over the period January 1987 to December 2002. The results indicate that the portfolio compositions produced by different risk measures vary markedly from risk measure to risk measure. Of the various alternatives examined the one that comes closest to the MV model is the MAD. The MAD approach produced asset allocations that are most similar to those produced by MV optimisation and a risk level that is very similar to that of the MV model in MV space. In contrast, the LPMZ and MM models produce asset allocations that are least like those of the MV model, and risk values in MV space that are substantially higher. These findings have a practical consequence for the investor because they suggest that the choice of model depends very much on individual attitudes to risk rather than any theoretical and practical advantages of one model over another. Note 1. The risk-free rate is not used as the target return in the optimisations below because this would mean that the target rate would be a non-constant value when a constant value is assumed. References Bawa, V.S. (1975), “Optimal rules for ordering uncertain prospects”, Journal of Financial Economics, Vol. 2 No. 1, pp. 95-121. Bawa, V.S. and Lindenberg, E.B. (1977), “Capital market equilibrium in a mean-lower partial moment framework”, Journal of Financial Economics, Vol. 5 No. 2, pp. 189-200. Byrne, P.J. and Lee, S.L. (1994a), “Computing Markowitz efficient frontiers using a spreadsheet optimiser”, Journal of Property Finance, Vol. 5 No. 1, pp. 58-66. Byrne, P.J. and Lee, S.L. (1994b), “Real estate portfolio analysis using a spreadsheet optimiser”, Journal of Property Finance, Vol. 5 No. 4, pp. 19-31. Byrne, P. and Lee, S.L. (1999), “The place of property in an Australian multi-asset portfolio: a comparison of the MPT and MAD optimisation methods”, Australian Land Economics Review, Vol. 5 No. 1, pp. 21-8. Cheng, P. (2001), “Comparing downside-risk and mean-variance analysis using bootstrap simulation”, Journal of Real Estate Portfolio Management, Vol. 7 No. 3, pp. 225-38. Cheng, P. and Wolverton, M.L. (2001), “MPT and the downside risk framework: a comment on two recent studies”, Journal of Real Estate Portfolio Management, Vol. 7 No. 2, pp. 125-31. Frodsham, M. and Key, T. (1996), “Segmentation of the UK property market”, paper presented at the European Real Estate Society Conference, Belfast. Investment Property Databank (IPD) (2002), UK Annual Index, IPD, London. Konno, H. (1989), “Piecewise linear risk functions and portfolio optimization”, Journal of the Operations Research Society, Japan, Vol. 33, pp. 139-56. Konno, H. and Shirakawa, H. (1994), “Equilibrium relations in a capital asset market: a mean-absolute deviation approach”, Financial Engineering and the Japanese Markets, Vol. 1, pp. 21-35. Konno, H. and Yamazaki, H. (1991), “Mean-absolute deviation portfolio optimization model and its applications to the Tokyo stock market”, Management Science, Vol. 37 No. 5, pp. 519-31.
Lizieri, C. and Ward, C.W.R. (2000), “The distribution of real estate returns”, in Knight, J. and Satchell, S. (Eds), Return Distributions in Finance, Butterworth-Heinemann, London, pp. 47-74. Markowitz, H. (1952), “Portfolio selection”, Journal of Finance, Vol. 7 No. 1, pp. 77-91. Phillips, H.E. (1993), “Portfolio optimization algorithms, simplified criteria, and security selection: a contrast and evaluation”, Review of Quantitative Finance and Accounting, Vol. 3, pp. 91-7. Sharpe, W. (1971), “Linear programming formulation of the general portfolio selection problem”, Journal of Financial and Quantitative Analysis, Vol. 8, pp. 621-36. Simaan, Y. (1997), “Estimation risk in portfolio selection: the mean variance model versus the mean absolute deviation model”, Management Science, Vol. 43, pp. 437-46. Sing, T.F. and Ong, S.E. (2000), “Asset allocation in a downside risk framework”, Journal of Real Estate Portfolio Management, Vol. 6 No. 3, pp. 213-23. Stone, B. (1973), “A general class of three-parameter risk measures”, The Journal of Finance, Vol. 28, pp. 675-85. Tobin, J. (1958), “Liquidity preference as behaviour towards risk”, Review of Economic Studies, Vol. 25, pp. 65-86. Young, M.R. (1998), “A minimax portfolio selection rule with linear programming solution”, Management Science, Vol. 44 No. 5, pp. 673-83.
Different risk measures
511
The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-578X.htm
JPIF 22,6
The level of direct property in Hong Kong property company performance
512
Graeme Newell
Received January 2003 Accepted March 2003
School of Construction, Property and Planning, University of Western Sydney, Penrith South, Australia, and
Chau Kwong Wing and Wong Siu Kei Department of Real Estate and Construction, University of Hong Kong, Hong Kong Keywords Management styles, Hong Kong, Property management, Performance appraisal Abstract Hong Kong is one of the most dynamic property markets in the world, and now provides the economic gateway to China. Using style analysis, the level of direct property in Hong Kong property company performance is shown to be approximately 15 per cent over 1984-2000, with the level of direct property increasing to approximately 25 per cent in recent years. The level of direct property in Hong Kong property company performance is significantly below that seen for the USA, Europe and Australia. This highlights a number of key strategic property investment issues over 1984-2000, relating to the level of direct property in Hong Kong property company performance. Also assesses the level of direct property at the individual property company level, as well as the property company sector level, further emphasising the strategic role of Hong Kong property companies in an investment portfolio. This research complements the previous research by Brown and Chau on excess returns in the Hong Kong property market, as well as highlighting the issues and role of both direct and indirect property for inclusion in diversified investment portfolios; these being key areas of Gerald Brown’s extensive property research agenda.
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 512-532 q Emerald Group Publishing Limited 1463-578X DOI 10.1108/14635780410569498
Introduction Hong Kong is a major economic force in the Asia-Pacific region, having been transformed from a manufacturing-based economy to a services-based economy over the last 25 years, with this being further enhanced by the economic integration of Hong Kong and China in recent years (Newell and Chau, 1996). This has seen over 600 multinational organisations establish their regional headquarters in Hong Kong (Newell and Chau, 1996), with Hong Kong less affected by the Asian economic crisis of the late 1990s than other Asian economies such as Indonesia, Thailand, Malaysia and South Korea (JLW, 1998). Hong Kong gross domestic product (GDP) per capita is currently $25,581. In recent years, this economic prosperity has seen Hong Kong consistently ranked among the most competitive economies in the world. For example, the World Economic Forum (WEF, 2002) survey sees Hong Kong ranked 18th internationally for overall competitiveness, 4th for macroeconomic environment and 13th for growth competitiveness, with an overall competitiveness rank (across 20 factors) of 9th in the IIMD (2002) survey. Among the Asian economies, only Singapore enjoys a higher competitiveness ranking (IIMD, 2002; WEF, 2002). The economic profile of Hong Kong
at June 2002 is shown below, reflecting the recent downturn in the Hong Kong economy (JLL, 2002a, b): (1) Economy contracted 0.5 per cent over 2001-02 (versus 0.5 per cent growth over 2000-2001). (2) Currently HK economy is only ranked 12th of 13 major Asian economies (1st ¼ China; 13th ¼ Japan). (3) Exports declined 1.6 per cent over 2001-2002. (4) Unemployment: 7.7 per cent (versus 4.7 per cent at June 2001). (5) Inflation: 2 3.3 per cent (versus 2 1.4 per cent at June 2001). (6) GDP growth: 0.7 per cent (versus 0.2 per cent at June 2001). (7) HK Central office market profile: . vacancy rate: 9.6 per cent (versus 3.0 per cent at 3Q: 1994 peak); . net rent: HK$4,697 per m2 p.a. (versus HK$10,600 at 3Q: 1994 peak): 56 per cent decline; . office rent ¼ 2nd highest in Asia (1st ¼ Tokyo); . capital value: HK$ 63,271 per m2 p.a. (versus HK$ 155,000 at 2Q: 1994 peak): 59 per cent decline; and . investment yield: 3.9 per cent-5.2 per cent. (8) HK Causeway Bay retail market profile: . net rent: HK$43,979 per m2 p.a. (versus HK$90,000 at 3Q: 1994 peak): 51 per cent decline; . retail rent ¼ highest in Asia (2nd ¼ Singapore); . capital value: HK$610,838 per m2 p.a. (versus HK$1,650,000 at 2Q: 1994 peak): 63 per cent decline; and . investment yield: 6.5 per cent-8.5 per cent. (9) HK luxury residential market profile: . net rent: HK$3,794 per m2 p.a. (versus HK$4,482 at June 2001 peak): 15 per cent decline; . residential rent ¼ highest in Asia (2nd ¼ Mumbai); and . capital value: HK$91,945 per m2 p.a. (versus HK$100,268 at June 2001 peak): 8 per cent decline. Similarly, Hong Kong is a major financial market, ranked 4th after London, New York and Tokyo, with the seventh largest stockmarket by market capitalisation (Chau et al., 2001). At March 2002, this market capitalisation stood at HK$3,855 billion (US$501 billion), with over 740 companies listed on the Hong Kong stockmarket, including over 50 China-incorporated enterprises. The commercial property market in Hong Kong is recognised as one of the world’s most dynamic and volatile property markets, with unique property market characteristics (Chau et al., 2001; Newell and Chau, 1996). It is also one of the deepest and most liquid property markets, with factors influencing this liquidity including short leases, low agents fees, no CGT and low tax rates (Chau et al., 2001).
Level of direct property in Hong Kong 513
JPIF 22,6
514
The compact, homogeneous nature of this market, with readily available sales evidence further enhances this property market liquidity (Chau et al., 2001; Newell and Chau, 1996). This sees the property sector in Hong Kong accounting for over 20 per cent of Hong Kong GDP (Walker et al., 1995), as well as enjoying a high level of property market transparency, only exceeded by Singapore in the Asian region (JLL, 2002a). The Hong Kong property profile at June 2002 is shown in Table I. While the Hong Kong market generally remains competitive in the Asian region, the significant downturn in the Hong Kong property market in recent years is clearly evident; particularly since the market peak in 1994. This has seen Hong Kong office and retail rents and capital values drop by up to 63 per cent since the market peak (JLL, 2002a, b). The dynamics of the Hong Kong property market has seen this important international property market as the focus for considerable property research in recent years. This has included: . property market dynamics (Brown and Chau, 1997; Ganesan and Chiang, 1998; He and Webb, 2000; Newell et al., 1996; Tse and Webb, 2000, 2001; Webb et al., 1997); . portfolio diversification issues (Brown and Chau, 1997; Brown et al., 2000; Chiang and Ganesan, 1996; Newell and Chau, 1996); and . significance of securitised property (Chau et al., 2001; Newell and Chau, 1996; Tse and Webb, 2000, 2001). This research demonstrated that excess returns existed in Hong Kong commercial property markets (Brown and Chau, 1997) and that Hong Kong property company returns reflected important transaction-based information about property market fundamentals that was subsequently impounded into Hong Kong commercial property performance (Newell and Chau, 1996). However, recent research (Chau et al., 2001) has highlighted Hong Kong property companies conveying a lesser level of information about Hong Kong property returns, after the effect of international capital market variables is included. Importantly, a substantial portion of Hong Kong commercial property is securitised in the Hong Kong stockmarket (Chau et al., 2001). Property companies currently account for 15 per cent of Hong Kong’s stockmarket capitalisation, compared to levels of up to 29 per cent seen in 1984-1996 (Chau et al., 2001). Including consolidated enterprises which are involved in property development and property investment activities, this contribution has been as high as 45 per cent of total stockmarket capitalisation in the mid-1990s (Walker et al., 1995). Given the significant role of global securitised property in an investment portfolio (Steinert and Crowe, 2001), and the significant benefits from diversifying a property securities portfolio internationally within Asia, both on a short-term and long-term basis (Garvey et al., 2001), Hong Kong property companies make a significant contribution to the Hong Kong economy and property industry (Walker et al., 1995). Given the significance of commercial property and property companies in the Hong Kong economy, it is essential that the recent investment dynamics of Hong Kong property companies are critically evaluated. As such, the objectives of this research are to: . examine the investment dynamics of Hong Kong property company performance over 1984-2000; and
4 6 13 17 25 38 44 51 69
163,869 150,057
59,242 44,033 22,255 14,011 10,523 8,092 4,935
Ranking
b
67,074 75,803 31,007 46,427 31,965 22,968 28,481
70,661 194,295
Property portfolio (HK$ million)
24 40 73 74 32 23 56
13 23
Level of gearing (%)
Property activities
26 42 20 17 18 56 49
12 25 27 30 29 29 26 21 16
15 21 1 3 2 4 6 0 0
1 4 35 19 39 31 41 21 2
60 42 11 6 11 19 9 2 33
12 9
38 85 63 74 59 97 74
26 43
62 15 37 26 41 3 26
74 57
Office Retail Industrial Residential Otherc Investment Development (%) (%) (%) (%) (%) (%) (%)
Property portfolio
Note: a HK$1 ¼ £0.09; b Ranking by property market capitalisation on Hong Kong Stock Market; c Other includes hotels, China infrastructure, car parks, China property development container terminals Source: Author’s compilation from property company annual reports and www.hsi.com.uk
Cheung Kong Sun Hung Kai Henderson Land Wharf Hang Lung New World Sino Land Hysan Great Eagle
Property company
Market capitalisationa (HK$ million)
Level of direct property in Hong Kong 515
Table I. Profile of Hong Kong property companies: March 2002
JPIF 22,6
516
.
determine how much of Hong Kong property company performance is attributable to direct property performance.
Significance of property companies in Hong Kong Property companies in Hong Kong[1] are a significant component of the Hong Kong stockmarket. At March 2002, there were 95 property companies listed on the Hong Kong stockmarket, accounting for HK$576,585 million (US$74,956 million) and representing 15 per cent of the total Hong Kong market capitalisation. While this current contribution is down on the levels of up to 30 per cent seen over 1984-1996 (Chau et al., 2001), this is largely attributable to recent poor property market performance, no new listings of property companies, and the increased number of large state enterprises and China-incorporated enterprises now listed on the Hong Kong stockmarket. Table I presents a profile of the nine leading property companies in Hong Kong (Cheung Kong, Sun Hung Kai, Henderson Land, Wharf, Hang Lung, New World, Sino Land, Hysan and Great Eagle); with these leading property companies forming the basis for the subsequent style analysis in this paper. These nine property companies had a total property portfolio of over HK$568,680 million (US$73,928 million), with their market capitalisation representing 12 per cent of the total market capitalisation and 82 per cent of the property company sector market capitalisation. The stature of these property companies is reflected in Cheung Kong and Sun Hung Kai being the 5th and 6th largest companies respectively on the Hong Kong stockmarket, with Henderson Land and Wharf also in the top 20 largest companies on the Hong Kong stockmarket. Of these nine property companies, eight are included in the Hang Seng index (HSI), which is the overall Hong Kong stockmarket indicator. Office, retail and residential property make up the substantive components in these property portfolios, with industrial property, hotels, China infrastructure, carparks, China property development and container terminals also represented to a lesser extent. Residential property figures prominently, with the seven largest property developers in Hong Kong supplying over 60 per cent of private sector housing in Hong Kong (Chau et al., 2001). Table I also lists the level of property investment and property development by these nine Hong Kong property companies. Typically, Hong Kong property developers have been characterised by property acquisition and trading, while Hong Kong property investors have focused on property acquisition and management for long-term growth (Newell and Chau, 1996). Recent years have also seen a trend from property development activities to more property investment activities among these property companies. The property development orientation was most evident in Cheung Kong, Henderson Land and Sun Hung Kai, while the property investment orientation was most evident in Hysan, Wharf and New World. Overall investment strategies ranged from conservative (e.g. Wharf) to aggressive (e.g. Sino Land). Institutional investors are the major players on the Hong Kong stockmarket. Typically, they have favoured the liquidity of indirect property investment via Hong Kong property companies, rather than direct property investment (Chau et al., 2001). This has further enhanced the investment stature of property companies in Hong Kong.
The performance of the Hong Kong property companies is assessed and benchmarked with the Hang Seng property sector sub-index, which accounts for 15 per cent of the overall HSI. The other available Hong Kong stockmarket sub-indices are for the commerce and industry sector (44 per cent of HSI), finance sector (34 per cent of HSI) and utilities sector (7 per cent of HSI), with the HSI representing 75 per cent of the total market capitalisation and 70 per cent of the transaction volume of the Hong Kong stockmarket. Methodology Data sources Quarterly property company performance in Hong Kong was assessed using the Hang Seng property company index, as well as a range of leading Hong Kong property companies examined, including Hang Lung, New World, Sino Land, Cheung Kong, Sun Hung Kai, Henderson Land, Great Eagle, Wharf and Hysan over June 1984-December 2000. Details of the property portfolios for these nine Hong Kong property companies were given previously in Table I. Equivalent Hong Kong quarterly performance indicators for: stocks: HSI, and the commerce and industrial, finance and utilities sector sub-indices and cash: 90-day bill rates were also utilised. Hong Kong bonds were not included in this analysis, as the Hong Kong bond market is very immature and thin. The Hong Kong direct property return series used in this study are the Jones Lang LaSalle (JLL) capital value indices reported quarterly over 1984:Q2-2000:Q4. The property types analysed were office, retail and industrial property. The JLL property indices are widely acknowledged as the property performance benchmarks for Hong Kong commercial property (Newell and Chau, 1996) and have been used extensively in Hong Kong property research (e.g. Brown and Chau, 1997; Chau et al., 2001; Newell and Chau, 1996; Newell et al., 1996). These JLL property indices incorporate both valuation-based and transaction-based property information, and do not have the same level of appraisal-smoothing that is typically evident in USA, UK and Australian commercial property indices (Brown and Matysiak, 1995, 1998, 2000; Newell and Chau, 1996; Newell et al., 1996). This reflects a number of key structural property market features in Hong Kong which minimise appraisal-smoothing (Chau et al., 2001; Newell and Chau, 1996; Newell et al., 1996). Statistical analysis To assess the changing investment dynamics of the inter-relationship between Hong Kong property companies, direct property and shares, rolling correlations with five-year performance periods were used. To assess the changing investment dynamics of property company performance in Hong Kong over time, style analysis via multi-factor asset allocation mix models over 1984-2000 were used. The general asset allocation mix model (Sharpe, 1992) is given by: R ¼ b1 F 1 þ b2 F 2 þ ::: þ bk F k þ e where: R ¼ property company return.
ð1Þ
Level of direct property in Hong Kong 517
JPIF 22,6
Fi ¼ return on ith financial or stock market factor. bi ¼ model coefficient that represents financial/stock market factor weighting in constrained asset allocation mix. e ¼ residual component.
518
Constrained asset allocation mix models were utilised using the “Solver” routine in Excel. The constrained asset allocation models ensure model coefficients or weightings are positive and sum to 100 per cent to reflect the asset allocation mix in practice. The performance technique of style analysis for evaluating property portfolios has also been effectively used in previous USA, UK, European and Australian studies (Chiang and Lee, 2002; Gallo et al., 2000; Lee, 1999; Liang and McIntosh, 1998; Myer and Webb, 1996; Newell, 2001a, b; Stevenson, 2001). Based on this style analysis and asset allocation mix model, the allocation to property in each period can be determined as shown below. Given the high level of property companies in the overall HSI, a purer representation of the non-property stockmarket effect is shown by using the three stockmarket sub-sector indices (commerce and industrial, finance and utilities). As such, the model involving shares (three factors) and cash is given as: RPC ¼ bCI RCI þ bF RF þ bu Ru þ bc Rc þ e
ð2Þ
where the subscripts used are: PC:
property company sub-sector;
F:
finance sub-sector;
U:
utilities sub-sector; and
C:
cash.
While property companies and property have different pricing mechanisms, it is highly likely that the unexplained variation in equation (2) is property company-specific and can be logically assumed to be largely attributable to direct property performance, although this cannot be tested conclusively. Fitting this constrained model in equation (2), and using 1 2 R 2 as the percentage allocation to direct property, then: RPC ¼ b0CI RCI þ b0F RF þ b0u Ru þ b0c Rc þ b0P RP
ð3Þ
0 where bCI , bF0 , bu0 , bc0 , bP0 are the adjusted portfolio weights such that 0 0 bCI þ bF þ b0u þ b0c þ b0P ¼ 1:0. Specifically, bP0 is the resulting percentage asset allocation to direct property in this multi-factor asset allocation model, with this statistical approach to determine the level of direct property previously used by Liang and McIntosh (1998) for USA real estate investment trusts (REITs) and Newell (2001a, b) for Australian LPTs and European property companies.
Results and discussion Traditional performance analysis Table II presents the investment performance analysis over 1984-2000 for Hong Kong property companies, shares and direct property. While the property company sector
53.16 43.84 56.45 56.14 49.51 44.03 46.89 64.96 57.66 44.02 31.24 21.12 17.83 13.45
38.52 38.41 41.31 24.71 21.88 22.97 22.92 40.93 32.08 29.83 23.10 10.96 9.72 5.73
Annual risk (%)
Note: Ranking based on risk-adjusted performance analysis
Property companies Sun Hung Kai Cheung Kong Henderson Land New World Hysan Hang Lung Wharf Sino Land Great Eagle Stockmarket sectors Property companies Hang Seng (overall HK stockmarket) Direct property Office Retail Industrial
Sector
Average annual return (%)
1.93 1.83 2.35
1.35
1.48
1.38 1.14 1.37 2.27 2.26 1.92 2.05 1.59 1.80
10 8 14
2
5
4 1 3 13 12 9 11 6 7
Risk-return ratio Rankinga
0.29 0.28 0.07
0.59
0.57
0.64 0.77 0.65 0.36 0.35 0.41 0.39 0.56 0.47
Sharpe index (1)
12 13 14
4
5
3 1 2 10 11 8 9 6 7
Rankinga
0.52 0.61 0.45
2021
2019
2020 2012 2023 2025 2015 2008 2006 2006 2019
1Q
2 026 2 029 2 030 2 029 2 033 2 026 2 041 2 024 2 027
4Q
0.22 2 033
0.19 2 032
0.11 0.16 0.23 0.17 0.12 0.25 0.12 0.14 0.09
3Q
0.34 0.20 2 005 0.20 2 002 2 018 0.32 0.39 0.28
2005
2009
2010 2013 2003 2004 2009 2008 2003 0.01 0.11
2Q
Serial correlation
Level of direct property in Hong Kong 519
Table II. Hong Kong property company performance analysis: 2Q: 1984-4Q: 2000
JPIF 22,6
520
provided higher average annual returns than the overall stockmarket, the volatility of the property companies sector (44.02 per cent) was 41 per cent higher than the overall stockmarket volatility (31.24 per cent). On a risk-adjusted basis (using Sharpe index), Cheung Kong, Henderson Land and Sun Hung Kai were the best performed property companies, as well as the overall property company sector being ranked as the 5th best performed, marginally behind the overall Hong Kong stockmarket (4th). The Hong Kong office, retail and industrial property sectors were the worst performed asset classes. As evidenced by the serial correlation structure in Table II, the extent of appraisal smoothing is not large in these JLL Hong Kong property series and much lower than that typically seen in most international property series such as the IPD and NCREIF series (Brown and Matysiak, 1995, 1998, 2000). The lack of appraisal smoothing is reflected in the ready availability of market evidence from sales in the Hong Kong market, instead of a strong dependence on valuations as a market proxy. The extensive use of strata title for commercial property and the resulting frequent transactions in a trading-oriented property market are major catalysts to the availability and use of sales evidence in reporting Hong Kong property performance (Chau et al., 2001; Newell and Chau, 1996; Newell et al., 1996). In addition, a number of key structural property market features are clearly evident in Hong Kong to minimise the effect of appraisal smoothing (Chau et al., 2001; Newell and Chau, 1996). These features include: . normal office lease terms are for a maximum of three years, with an efficient rental market; . short-term holding periods, with market previously dominated by property traders (not long-term investors); . degree of market liquidity, resulting from high level of strata titling in many major CBD buildings; . small, homogenous property market with large number of transactions; . no restrictions on leasehold purchases by foreign investors; . stamp duty of up to 2.75 per cent of sale price; . agent’s fees of approximately 1 per cent; . 2 per cent tax deduction for depreciation in commercial property; . no capital gains tax; . maximum tax rate of 15 per cent (personal) and 16.5 per cent (corporations); and . ready availability of property information from both government and private sources. Table III presents the Hong Kong inter-asset correlation matrix over 1984-2000. Property company performance was more highly correlated with the stockmarket (r ¼ 0:93) than the direct property market (r ¼ 0:02 2 0:22). This confirms the general view that property companies in Hong Kong are more closely related to stockmarket performance than direct property performance (Chau et al., 2001; Newell and Chau, 1996). However, the above static analysis takes no account of how the dynamics of property companies in Hong Kong may vary over time.
1.00 0.84 0.77 0.75 0.77 0.65 0.79 0.77 0.93 0.91 0.10 0.19 0.01
1.00 0.87 0.90 0.90 0.86 0.79 0.72 0.87 0.76 0.97 0.86 0.04 0.18 2 0.003 1.00 0.85 0.78 0.81 0.72 0.82 0.81 0.93 0.88 0.06 0.14 0.07
HL
1.00 0.84 0.79 0.81 0.79 0.72 0.91 0.84 0.09 0.17 2 0.002
NW
1.00 0.71 0.73 0.85 0.65 0.87 0.77 0.17 0.28 0.04
H
1.00 0.69 0.77 0.71 0.85 0.80 0.11 0.22 0.14
HLD
1.00 0.64 0.67 0.77 0.77 0.18 0.22 0.04
W
1.00 0.80 0.88 0.78 0.15 0.26 0.12
SL
1.00 0.82 0.82 0.20 0.18 0.19
GE
1.00 0.93 0.11 0.22 0.02
PC
1.00 0.12 0.21 0.08
HS
1.00 0.65 0.54
Office
Industrial
1.00
Retail
1.00 0.53
Note: SHK ¼ Sun Hung Kai; CK ¼ Cheung Kong; HL ¼ Henderson Land; NW ¼ New World; H ¼ Hysan; HLD ¼ Hang Lung Development; W ¼ Wharf; SL ¼ Sino Land; GE ¼ Great Eagle; PC ¼ Property companies sector; HS ¼ Hang Seng (overall HK Stock Market)
SHK CK HL NW H HLD W SL GE PC HS Office Retail Industrial
CK
SHK
Level of direct property in Hong Kong 521
Table III. Hong Kong inter-asset correlation matrix: 2Q: 1984-4Q: 2000
JPIF 22,6
522
Correlation dynamics Using rolling five-year performance windows over 1984-2000, the correlation between Hong Kong property companies and the overall stockmarket was consistently in the range of r ¼ 0:94 2 0:97 over 1984-1998, reflecting the significant contribution of Hong Kong property companies to the overall Hong Kong stockmarket. Only over 1999-2000 has the correlation steadily declined from 0.96 to 0.91, reflecting some lesser alignment with the overall stockmarket, with this being largely attributable to the increased significance of China-incorporated enterprises on the Hong Kong stockmarket since 1998. While reducing in recent years, this correlation result is in marked contrast to the correlation dynamics for US REITs (Liang and McIntosh, 1998) and Australian property trusts (Newell, 2001a) over similar periods. In both cases, the correlations reduced significantly over this period from 0.75 to 0.25 and 0.50 respectively, with these correlations having also reduced further in subsequent time periods. Hong Kong property company performance “style” Using the general asset allocation mix model of equation (1) (Sharpe, 1992), the three stockmarket sub-indices (commerce and industrial, finance, utilities) and cash were used to explain Hong Kong property company performance over 1984-2000. These stockmarket sub-indices were chosen, because of the high correlation between the property companies index and the overall Hang Seng index (r ¼ 0:93), to obtain purer explanatory factors. Using rolling five-year performance windows and attributing the unexplained variation to direct property, Figure 1 shows the performance style for Hong Kong property companies over 1984-2000. The key feature to emerge from this performance style analysis is that Hong Kong property companies performed similarly to a portfolio of 15 per cent property and 85 per cent shares (reflecting the aggregation of the three non-property sub-sectors on the Hong Kong stockmarket) over this period of 1984-1999. Since 1999, this level of direct property in Hong Kong property company performance has steadily increased from this level of 15 per cent to the current level of 25 per cent in 2000. Figures 2-10 show the level of direct property reflected in the performance of the nine individual Hong Kong property companies considered in this study over 1984-2000. The level of property reflected in individual property company performance ranged from 20 per cent-50 per cent. These direct property levels tended to be reasonably stable over the period of 1984-1999. Since 1999, these levels of direct property have increased in most instances, with the levels of direct property at the end of 2000 being 25-47 per cent. In particular, Great Eagle (47 per cent), Hang Lung (45 per cent), Wharf (45 per cent) and Sino Land (45 per cent) had the highest levels of direct property in their performance at the end of 2000. Typically, property investors (45 per cent) saw higher levels of direct property in their performance than property developers (33 per cent). Overall, these style analysis results indicate that Hong Kong property companies have attributed more of their performance to direct property in 2000 than in previous years, but the magnitude of this contribution by direct property remains small relative to the overall stockmarket contribution. The current levels of direct property are 25 per cent, compared to average levels of only 15 per cent over 1984-1999. These style
Level of direct property in Hong Kong 523
Figure 1. Style analysis: Hong Kong property company sector: 1984-2000
Figure 2. Style analysis: Sun Hung Kai: 1984-2000
JPIF 22,6
524
Figure 3. Style analysis: Cheung Kong: 1984-2000
Figure 4. Style analysis: Henderson Land: 1984-2000
Level of direct property in Hong Kong 525
Figure 5. Style analysis: New World: 1984-2000
Figure 6. Style analysis: Hysan: 1984-2000
JPIF 22,6
526
Figure 7. Style analysis: Hang Lung: 1984-2000
Figure 8. Style analysis: Wharf Holdings: 1984-2000
Level of direct property in Hong Kong 527
Figure 9. Style analysis: Sino Land: 1984-2000
Figure 10. Style analysis: Great Eagle: 1984-2000
JPIF 22,6
528
Figure 11. USA style analysis: 1984-1999
analysis results reflect the closer alignment of Hong Kong property companies with the Hong Kong stockmarket and also reinforce the Chau et al. (2001) results, which saw lesser evidence of Hong Kong property companies conveying information about Hong Kong direct property performance after the effect of international capital market variables was included. Also the style analysis results for Hong Kong are significantly different to those seen for US REITs (Liang and McIntosh, 1998), Australian listed property trusts (Newell, 2001a) and European property companies (Newell, 2001b), both in terms of the level of direct property and the trend over 1984-2000. In particular, for US REITs (see Figure 11), the level of direct property in REIT performance has averaged 50 per cent over 1984-1998, with the current level at 65 per cent (Liang and McIntosh, 1998). For Australian LPTs (see Figure 12), the level of direct property has averaged 65 per cent over 1984-1999, with the current level of 75 per cent (Newell, 2001a). Similarly, Figures 13 and 14 present the style analysis results for The Netherlands and France (Newell, 2001b). Both The Netherlands and France exhibit comparable trends, with direct property levels in property company performance increasing since 1999 to 55 per cent and 85 per cent respectively in 2000, after experiencing significant declines in the level of direct property contribution over 1992-1998. While recognising that property companies are different property investment vehicles to REITs (USA) and listed property trusts (Australia), the results reflect the closer alignment of Hong Kong property companies with the Hong Kong stockmarket than the Hong Kong property market.
Level of direct property in Hong Kong 529
Figure 12. Australian style analysis: 1986-2000
Figure 13. The Netherlands style analysis: 1983-2000
JPIF 22,6
530
Figure 14. France style analysis: 1984-2000
Property investment implications This study has used style analysis to identify the level of direct property in Hong Kong property company performance over 1984-2000. These levels averaged 15 per cent over 1984-1999, having increased to 25 per cent in 2000. The levels for direct property are significantly below the equivalent style analysis results for USA, Australia and Europe. These differences in results are likely to be attributable to the specific dynamics of the Hong Kong markets, as well as the functional differences as property entities between REITs/LPTs and property companies. Given the well-established view of direct property providing diversification and risk-reduction benefits in a mixed-asset portfolio, and the liquidity benefits of investing in property companies (Brown and Matysiak, 2000), the results of this Hong Kong study over 1984-2000 have significant implications for both investors and fund managers in Hong Kong and internationally. In particular, with Hong Kong property companies only delivering lesser levels of direct property (compared to USA REITs and Australian LPTs), it is essential for investors to have exposure to both Hong Kong property companies and direct property to achieve effective exposure to property performance and receive the resulting portfolio diversification and risk-reduction benefits. This is in marked contrast to the USA REIT and Australian LPT scenarios, in which investors are achieving significant levels of direct property exposure by investing in these indirect property investments, as well as achieving the traditional stockmarket exposure for these REITs and LPTs.
Note 1. Details of the contribution of Hong Kong property companies to the Hong Kong stockmarket were obtained from the Hang Seng Index Web site (www.hsi.com.hk) References Brown, G. and Chau, K.W. (1997), “Excess returns in the Hong Kong commercial real estate market”, Journal of Real Estate Research, Vol. 14 No. 2, pp. 91-105. Brown, G. and Matysiak, G. (1995), “Using commercial property indices for measuring portfolio performance”, Journal of Property Finance, Vol. 6 No. 3, pp. 27-38. Brown, G. and Matysiak, G. (1998), “Valuation smoothing without temporal aggregation”, Journal of Property Research, Vol. 15 No. 2, pp. 1-15. Brown, G. and Matysiak, G. (2000), Real Estate Investment: A Capital Markets Approach, Prentice-Hall, Harlow. Brown, R., Li, L.H. and Lusht, K. (2000), “A note on intercity geographic diversification of real estate portfolios: evidence from Hong Kong”, Journal of Real Estate Portfolio Management, Vol. 6, pp. 130-40. Chau, K.W., MacGregor, B. and Schwann, G. (2001), “Price discovery in the Hong Kong real estate market”, Journal of Property Research, Vol. 18 No. 3, pp. 187-216. Chiang, K. and Lee, M.L. (2002), “REITs in the decentralised investment industry”, Journal of Property Investment & Finance, Vol. 20 No. 6, pp. 496-512. Chiang, Y.H. and Ganesan, S. (1996), “Property investment in Hong Kong”, Journal of Real Estate Portfolio Management, Vol. 2 No. 2, pp. 141-58. Gallo, J., Lockwood, L. and Rutherford, R. (2000), “Asset allocation and the performance of real estate mutual funds”, Real Estate Economics, Vol. 28 No. 1, pp. 165-84. Ganesan, S. and Chiang, Y.H. (1998), “The inflation-hedging characteristics of real and financial assets in Hong Kong”, Journal of Real Estate Portfolio Management, Vol. 4 No. 1, pp. 55-67. Garvey, R., Santy, G. and Stevenson, S. (2001), “The linkages between real estate securities in the Asia-Pacific”, Pacific Rim Property Research Journal, Vol. 7 No. 4, pp. 240-58. He, L.T. and Webb, J. (2000), “Causality in real estate markets: the case of Hong Kong”, Journal of Real Estate Portfolio Management, Vol. 6 No. 3, pp. 259-71. International Institute of Management Development (IIMD) (2002), World Competitiveness Yearbook, IMD, Lausanne, available at: www.imd.ch/wcy Jones Lang LaSalle (JLL) (2002a), Asia Pacific Property Investment Guide, JLL, Hong Kong. Jones Lang LaSalle (JLL) (2002b), Asia Pacific Property Digest: 2nd Quarter 2002, JLL, Sydney. Jones Lang Wootton (JLW) (1998), Property Investment in Asia: Future Prospects, JLW, Sydney. Lee, S. (1999), “Style analysis and property fund performance”, Journal of Property Investment & Finance, Vol. 17 No. 2, pp. 145-56. Liang, Y. and McIntosh, W. (1998), “REIT style and performance”, Journal of Real Estate Portfolio Management, Vol. 4 No. 1, pp. 69-78. Myer, N. and Webb, J. (1996), “Management style and asset allocation in real estate portfolios”, Journal of Real Estate Portfolio Management, Vol. 2 No. 2, pp. 119-25. Newell, G. (2001a), “LPTs: a matter of style”, Property Australia, Vol. 15 No. 10, p. 20. Newell, G. (2001b), “How much property performance are European property companies delivering?”, paper presented at the IPD/GPR European Property Strategies Conference, Wiesbaden.
Level of direct property in Hong Kong 531
JPIF 22,6
532
Newell, G. and Chau, K.W. (1996), “Linkages between direct and indirect property performance in Hong Kong”, Journal of Property Finance, Vol. 7 No. 4, pp. 9-29. Newell, G., Chau, K.W. and Pretorius, F. (1996), “Adjusting the volatility of the Hong Kong property market”, Journal of Real Estate and Construction, Vol. 6 No. 1, pp. 1-16. Sharpe, W. (1992), “Asset allocation, management style and performance measurement”, Journal of Portfolio Management, Winter, pp. 7-19. Steinert, M. and Crowe, S. (2001), “Global real estate investment: characteristics, optimal portfolio allocation and future trends”, Pacific Rim Property Research Journal, Vol. 7 No. 4, pp. 223-39. Stevenson, S. (2001), “Evaluating the investment attitudes and performance of property companies”, Journal of Property Investment & Finance, Vol. 19 No. 3, pp. 251-66. Tse, R. and Webb, J. (2000), “Public versus private real estate in Hong Kong”, Journal of Real Estate Portfolio Management, Vol. 6 No. 1, pp. 53-60. Tse, R. and Webb, J. (2001), “Public versus private real estate in Hong Kong using adaptive expectations”, Journal of Real Estate Portfolio Management, Vol. 7 No. 1, pp. 143-9. Walker, A., Chau, K.W. and Lai, L. (1995), Hong Kong in China: Real Estate in the Economy, Brooke Hillier Parker Research, Hong Kong. Webb, J., Chau, K.W. and Li, L.H. (1997), “Past and future sources of real estate returns in Hong Kong”, Journal of Real Estate Research, Vol. 13 No. 3, pp. 251-71. World Economic Forum (WEF) (2002), “Hong Kong SAR: competitiveness profile”, available at: www.weforum.org
The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-578X.htm
PRACTICE BRIEFING
The valuation of specialised property A review of valuation methods
The valuation of specialised property 533
Nick French The Department of Real Estate and Planning, The University of Reading Business School, Reading, UK Keywords Asset valuation, Property marketing Abstract Provides a brief overview of the methods that used in real estate valuation with a particular emphasis on the valuation of specialised property. Proposes that the underlying requirement is to estimate market value and that the role of the valuer is to choose the method that is the best model to achieve this objective. Concludes that a valuer must work with the recognised techniques and, in the case of specialised property, these are methods that go back to analysing value from first principles by identifying the value of the property to the business.
Introduction The valuation of real estate is central tenet for all businesses. Land and property are factors of production and, as with any other asset, the value of the land flows from the use to which it is put, and that in turn, is dependent on the demand (and supply) for the product that is produced. Valuation, in its simplest form is the determination of amount for which the property will transact on a particular date. However, there is a wide range purposes for which valuations are required. These range from valuations for purchase and sale, transfer, tax assessment, expropriation, inheritance or estate settlement, investment and financing. The objective of this paper is to provide a brief overview of the methods that used in real estate valuation with a particular emphasis on the valuation of specialised property. Specialised properties are properties that are more heterogeneous than homogonous. That is, the nature of the property is such that the type of property concerned do not transact sufficiently to be able to determine value by comparison of previous sales. In such circumstances the valuer needs to resort to a valuation model that addresses the underlying fundamentals of that property so that its value can be determined by reference to the wealth-producing qualities of the asset. For most non-specialised property, the value of the property is based on its income producing potential as an investment. For most specialised property, the value is based on the owner-occupier’s views of the worth of the property, i.e. the contribution it will make to business profit, as well as subjective issues such as status and feelings of security. Valuers, with hardly any transaction evidence, can only attempt to replicate these calculations of worth in arriving at an estimate of exchange price by reliance on an accepted valuation model.
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 533-541 q Emerald Group Publishing Limited 1463-578X DOI 10.1108/14635780410569506
JPIF 22,6
534
Price, worth and market value For any valuation model to have validity it must produce an accurate estimate of the market price. The model should therefore reflect the market culture and conditions at the time of the valuation. It should be remembered that the model should be a representation of the underlying fundamentals of the market and that the resulting figure of the valuation is “value”. However, much confusion arises from the fact that, in common parlance, the word “value” is often used to describe three distinct, albeit related, concepts. To clarify matters, the following convention is adopted: . price is the actual observable exchange price in the open market (it is historic comparable information); . value or market value is an estimation of the price that would be achieved if were the property to be sold in the market; and . worth is a specific individual’s perception of the capital sum that he/she would be prepared to pay (or accept) for the stream of benefits that she/he expects to be produced by the property. In the language of economics worth can be considered as value in use, whereas price or market value can be considered as value in exchange. In a perfect market, where all investors have the same information and the same requirements, “price” and “worth” should be the same figure. Value is a price. It is a price that would prevail under specified market conditions as a result of the interaction of the forces of supply and demand. Worth is the present value of future benefits anticipated or forecast to be receivable from the ownership of an asset. This is the basis for value in use. In the perfect market of economic theory, informed and rational buyers would pay no more, and informed and rational sellers would accept no less, than the present worth of the anticipated future benefits from ownership of an asset (discounted at market-determined rates). Thus, all transactions would take place at prices that reflected value in use, and represented value in exchange. Value in use would equal value in exchange, and price would be synonymous with value. However, the property market is not perfect and there is a natural divergence between the two figures in certain markets. Indeed, depending on the type of property, the valuation model may have its origin in comparing previous sale prices and thus deriving an investment value (value in exchange) by reference to observed payments in the market. Whereas other properties, which do not transact sufficiently often to produce reliable comparable information, need to use valuation models which reflect the thought process of the principal players; this relates to worth (value in use). There is one internationally accepted definition of market value: Market value is the estimated amount for which an asset should exchange on the date of appraisal between a willing buyer and a willing seller in an arm’s length transaction after proper marketing wherein the parties had each acted knowledgeably, prudently and without compulsion.
This is the definition of the International Valuation Standards Committee (IVSC) and has been fully adopted by most national property organisations. The most common methodologies of estimating market value are direct capital comparison, the
investment method, the contractor’s method, the residual method and the profits method. Valuation is simply a model to try to determine price. Value is the end result. It is the quantification of an understanding of the market; the legal impact; the physical constraints; the planning regime; the availability of finance; the demand for product and the general economy all influence the value of property. Valuation is the process of determining market value; an estimation of the price of exchange in the market place. Thus, the intent of any valuation is constant. It is the best estimate of the trading price of the property. The distinction between the valuation of non-specialised and specialised property stems from the nature of the model used. With non-specialised property there is sufficient trading activity to observe the level of prices without the need to interpret the underlying fundamentals. Price is determined by comparison. However, given that price should reflect the thought process of a potential purchaser, it is not unreasonable that where there is no established trading market then cost of replacement or an analysis of the property as an asset to the business will become the principal forms of pricing. This is the basis of the valuation models used for the valuation of specialised property. Purpose of valuations Valuations are required for many different purposes ranging from open market transaction to compulsory purchase. Although the underlying preferred method of valuation should not be dependant on the purpose of the valuation, it is important that the purpose is determined before undertaking any calculation. Market value – sale report The most common purpose for requesting a valuation is for sale. Although, this is often referred to as a valuation, it is actually more akin to marketing advice as normally the estimate of price is given for a future date after the property has been fully marketed. Conversely, a valuation for purchase is, by its nature, an estimate of the individual’s best bid and thus is a calculation of worth. Market value – accounting purposes A more correct use of the term valuation is the value of property as reported in company (or public) accounts. The majority of property owners have to prepare valuations of their properties for the purposes of their accounts. This is a statement of the company’s wealth on a particular date. Thus the value of the property element within the business is an estimate of the market value on the date of the accounts. Market value – loan security Banks and other lenders commission the valuation of property acting as collateral for a loan. They want a market value on which they can judge the amount of the loan based on a “loan to value” ratio. They are attempting to manage the risk of the loan by ensuring that the property has sufficient value to act as security for the amount lent. Market value – minimum price or auction reserve Often when a company or public body is selling its assets by tender or auction, they are obliged to only accept offers in excess of their valuation of the asset. Thus a market
The valuation of specialised property 535
JPIF 22,6
value has to be determined as a guide. Similarly, in cases where an owner has a property that is unusual or where there are special circumstances pertaining to it, the valuer may be instructed to place a reserve value on the property for auction.
536
Market value – insurance All property must be insured in the case of replacement - but this is really unconnected to the sale price – which of course includes the land. For insurance purposes the normal basis of valuation adopted will be the cost of replacing the building in the event of destruction or partial destruction. Market value – taxation Valuers frequently have to value property for tax purposes. The principal taxes fall into groups: capital and revenue. Often these valuations are formula based and diverge form normal market value calculations. Market value – compulsory purchase Often public schemes involve the purchase of land and this acquisition is subject to a compulsory order. The principal basis of compensation for the land and buildings taken is based on open market value. This is often a case where the valuer has to deal with specialised property such as churches, schools and the like. Valuation models Each country will have a different culture and experience, which will determine the methods adopted for any particular valuation. Valuation methods can be grouped as follows: (1) Comparable method. Used for most types of property where there is good evidence of previous sales (non-specialised property). (2) Investment/income method. Used for most commercial (and residential) property that is producing, or has the potential to produce, future cash flows through the letting of the property (non-specialised property). (3) Accounts/profits method. Used for trading properties (other than normal shops) where evidence of rents is slight as they tend not to be held as investments. The accounts method determines an appropriate rent, which is then used in the investment method (specialised property). (4) Development/residual method. Used for properties ripe for development or redevelopment or for bare land only. Determines the value of the asset undeveloped relative to the potential sale price of the completed development (non-specialised and specialised property). (5) Contractor’s/cost method. Used for only those properties not bought and sold on the market and for technical (accounts and statutory) purposes only (specialised property). For a full discussion on each of the five methods, see Pagourtzi et al. (2003). Globally there are a number of different models used. In the UK, there are five recognised methods (above) whereas in the USA and in Germany, they divide this down into only three methods. The principal three methods are capital comparison,
investment method and depreciated replacement cost methods. The addition two UK methods, the residual method and the profits method, are actually sued in the other markets but are considered to be sub-sets of the investment method. Given that the latter two are often used to value specialised property, this paper will use the UK convention of splitting them into five approaches. We have already determined that a fundamental valuation model should therefore reflect this thought process of determining the worth of the asset to the potential owner. However, in a market where there are frequent transactions it is possible to observe the level of prices without the need to interpret the underlying fundamentals. Price is determined by comparison. This is the principal “unit of currency” for the capital comparison and the investment methods both of which are used for the valuation of non-specialised property. However, the investment method has moved away from modelling the thought processes of the players in the market, and instead assesses the market value of a subject property by reference to observed recent transactions of similar properties in the same area. If there are insufficient sales to determine a comparable value and if there is no rent produced because the property is in owner-occupation, then the valuer must determine the value by returning to a detailed market analysis and it is the valuer’s role to assess the economic rent for the property from first principles. This is calculated by assessing the potential revenue to be expected each year from the asset, and deducting all other costs of a prudent entrepreneur in realising that cash flow. The residue will be an estimation of the economic rent for the property. The capital value can then be derived by multiplying the annual rent by an appropriate multiplier. This process reverts to a fundamental analysis of the worth of the property to the business. The economic rent is a derivative of the supply and demand for the final product. The same principle will apply to any type of property where the market value of the property is intrinsically linked to the business carried out within that property. This is the basis of the profits method. A further way in which it is possible to estimate the market value of land and property is the contractor’s method or the replacement cost method. If the property being valued is so specialised that properties of that nature are rarely sold on the open market, it will be effectively impossible to assess its value by reference to comparable sales of similar properties. Similarly, if there is no rental produced, the investment method will also be inappropriate. The profits method could be applied if the property is intrinsically linked to the business carried out in the property. However, where that business is one of production (rather than service) it is difficult to determine the contribution of the property to the overall usage. Thus, once again, the valuer must revert to understanding the thought process of the user of the building. Here the nature of the business is so specialised that there are no comparisons, thus, the owner of the building will simply assess the market value of the building by reference to its replacement cost. How much would it cost to replace the property if the business was deprived of its use. In simple terms, market value will equate to reconstruction costs. The valuer will assess the market value of the raw land (by reference to comparable land values in an appropriate alternative use), add to this value the cost of rebuilding a new building which could perform the function of the existing structure and from this then make subjective adjustments to allow for the obsolescence and depreciation of the
The valuation of specialised property 537
JPIF 22,6
538
existing building relative to the new hypothetical unit. It is reasonable to assume that this mirrors the thought process of the owner-occupier and thus should be viewed as a valid and rational method of valuation. This is the contractor’s or cost method. Choice of method Thus the valuer has the choice of a number of methods and the method used will be a reflection of the available information in the market place. Generally the less information, in the form of comparable sales, the more the valuer will inclined to use a model that reflects the role of property as the asset to the business. These types of property tend to be referred to as specialised property. Conversely, where there is a lot of comparable transaction data (either in the form of capital value and/or rents/yields) then the valuer will value without reference to the original thought process of the occupier. They take “second-hand” information from comparables and interpret this information within the context of the current market to estimate the price of the subject property. This is used for non-specialised properties. Non-specialised property The type of property that would be referred to as non-specialised are the dominant property types of residential, offices, shops, industrial units and warehousing. These will be valued by either the Investment or comparative methods or, in some markets, by the replacement cost approach. This paper is not concerned with these property types. Similarly, while some valuations are carried out against the backdrop of a statutory or legal framework. For example, properties being acquired by compulsory purchase will be valued in accordance with acquisition laws in the acquiring country, yet the valuation methods underpinning them will not differ simply because the acquisition is compulsory; likewise with valuations for taxation. The general principal is that a property is either specialised or non-specialised regardless of the reason for the valuation. It is the property type that determines method, not the purpose of the valuation. Specialised property The types of property that would be referred to as specialised are those properties where there are insufficient market data to value them by some form of comparison. The assumption with all valuations of specialised buildings is that they are to be valued on the assumption that the existing use of the building will continue. On this basis, there are a number of assets that could be described as specialised. These may be: . Agricultural land. Although, in its purest form, agricultural land can be valued by comparison often the market is distorted by governmental policy and the value of the land may relate to the payments that are made in the form of grants and set-asides or quota allocations. As such, the value is determined on an accounts or profits method. . Telecommunications. This may incorporate a whole host of facilities. Aerial masts are now so common that comparative valuations are now the norm. While, cabling, overhead wiring and relay or booster sites may, in the absence of
.
.
.
.
.
.
.
comparison, be valued by reference to their contribution to the business. As such, their value is determined on an accounts or profits method. Mineral extraction. This is a classic case of land as a factor of production. The land is the core element of the business and as such the value of the land is based on the likely profits arising from the extraction of the mineral(s) in the ground relative to the costs of extraction. As such, the land value is determined on an accounts or profits method. Alternatively, it is possible that a residual approach could be adopted, which is a simple variation on the accounts basis. Land fill. As this is the reverse of the above, mineral extraction, then it follows that the same thought process will apply. Except in this case, profits are generated by what you can put into the ground not take out. Once again, the land is core to the business and the appropriate method will be the accounts or profits method. Bars and restaurants. In many countries, the sale of bars and restaurants has become commonplace and as such there can be sufficient comparable information available for it to be valued by either the comparable or investment methods. However, in areas where there is a paucity of comparables, then the valuer needs to resort to an analysis the likely profits arising from the use of the building as a leisure business from the sale of food and drink. As such, the property value is determined on an accounts or profits method. Casinos and clubs. Although both these concerns rely heavily on the sales of food and drink as above, they also have other ways inn which to generate income. In the case of clubs, there will be an entry fee in addition to the services provided in the club. In the case of casino, the income is generated through gambling receipts. This is just a variation on a theme and the correct valuation method will be the accounts or profits method. Cinemas and theatres. A variation on the above. The facility charges an entry fee but in addition derives a substantial amount of its revenue from the concessions stands through the sale of snack, sweets and drinks. Again, the correct valuation method will be the accounts or profits method. Hotels. This is another leisure property where the building is integral to the business. The room charge is only one component to the revenue producing potential of a hotel. Generally, the larger the hotel, the more variation in the ways in which they generate income. They can offer, food, drink and entertainment. But many of them also offer conference facilities, health clubs and swimming pools. All of which generate additional, and often substantial, income. Once again the appropriate valuation method will be the accounts or profits method. Leisure properties (private). An all encompassing heading to cover health clubs, tennis courts, swimming pools, football pitches, golf clubs, athletic clubs and the like. Some of these are now sold sufficiently often to generate comparables allowing them to be valued by either the comparable or investment methods. However, as with hotels, it is more normal to view the property associated with the business as an asset to generate income by changing a market price (which may be high to reflect exclusivity) for its use and as such the most common valuation method will be the accounts or profits method.
The valuation of specialised property 539
JPIF 22,6
540
.
.
.
.
.
.
.
Leisure properties (public). Most local or municipal authorities have a brief to provide leisure facilities to the general public and generally at a subsidised price. As such, they are non-profit making organisations and thus the use of an accounts or profits method would be inappropriate. In these cases, the only way in which value can be assessed is by reference to the replacement cost of the building, and thus the contractor’s or cost method should be used. Care/nursing homes. As with leisure properties there is a distinct split in the market between private and public nursing homes. The former are valued as income generating properties by the accounts or profits method. The latter, the public nursing/care homes, are non-profit organisations and will be valued by the contractor’s or cost method. Hospitals. Again we have a split between private and public hospitals. The former are valued as income generating properties by the accounts or profits method. The latter, the public hospitals, are non-profit organisations and will be valued by the contractor’s or cost method. Development property. Development property is on the cusp between specialised and non-specialised property. Obviously, the end use of the completed development may be either a specialised or non-specialised use and as such the calculation of the completed development value might, in the case of specialised property, rely on a profits or contractor’s method. However, the overall method adopted to determine the land/property value in its existing state will be the residual method. Petrol stations. Petrol stations are income-generating business and as such they are valued by the accounts or profits method. Woodlands. As with agricultural property, it is possible for woodland to be valued by comparison. However, most woodland is an income-generating business which benefits form grants and tax incentives and as such they tend to be valued by the accounts or profits method. Churches. The majority of churches are non-profit organisations and in most countries are recognised as charitable institutions. As such, they will be valued by the contractor’s or cost method.
Conclusion In this paper we have reviewed the methods that have been used for estimating real estate property’s value for specialised property. The underlying requirement is to estimate market value. The role of the valuer is to choose the method that is the best model to achieve this objective. A valuer must work with the recognised techniques. In the case of specialised property, these are methods that go back to analysing value from first principles by identifying the value of the property to the business. Reference Pagourtzi, E., Assimakopoulos, V., Hatzichristos, T. and French, N. (2003), “Real estate appraisal: a review of valuation methods”, Journal of Property Investment & Finance, Vol. 21 No. 4, pp. 383-401.
Further reading Adair, A., Downie, M.L., McGeal, S. and Vos, G. (Eds) (1996), European Valuation Practice: Theory and Technique, E. & F.N. Spon, London. Askham, P. (Ed.) (2003), Special Properties and Purposes, Estates Gazette, London. Baum, A., Mackmin, D. and Nunnington, N. (1997), The Income Approach to Property Valuation, 4th ed., Thompson, London. French, N. (1996), “Investment valuations: developments from the Mallinson Report”, Journal of Property Valuation and Investment, Vol. 14 No. 5, pp. 48-58. Peto, R., French, N. and Bowman, G. (1996), “Price and worth: developments in valuation methodology”, Journal of Property Valuation and Investment, Vol. 14 No. 4, pp. 79-100. Rees, W.H. and Hayward, R. (Eds) (2000), Principles into Practice, Estates Gazette, London. Royal Institution of Chartered Surveyors (RICS) (2003), RICS Appraisal and Valuation Standards, RICS, London. Trott, A. (Ed.) (1986), Property Valuation Methods: Research Report, Polytechnic of the South Bank/Royal Institution of Chartered Surveyors, London.
The valuation of specialised property 541
JPIF 22,6
Book review Real Estate Investment: A Capital Market Approach Gerald Brown and George Matysiak
542
Journal of Property Investment & Finance Vol. 22 No. 6, 2004 pp. 542-543 q Emerald Group Publishing Limited 1463-578X
Knowing Gerald as a colleague at the National University of Singapore when he was writing the final stages of the book provided me with valuable first hand impressions of the book Real Estate Investment: A Capital Market Approach. The best way to describe the book is that it is a labor of love for Gerald. First, the book is developed from Gerald Brown’s earlier book Property Investment and Capital Markets published in 1991, but it is clear that the this book represents a significant improvement. Second, Gerald took great pains in ensuring that the book is readable – technical readings are conveniently found in the Appendices following each chapter; favoring the use of the second person, etc. Third, Gerald chose the cover after much careful thought. The original cover shows a spiraling staircase that “spirals down to nowhere!” Such is the care that went into this book. Last, but not least, the book was written with sweet jazz music in the background – a labor of love. The book is divided into three basic sections. The first is an introduction to financial concepts, the second is on risk and return while the third section focuses on portfolio management. The coverage on cash flow analysis, valuation models, mechanics of mortgages and interest rates, investment and capital budgeting analysis are easy to read, especially for undergraduates pursuing an interest in real estate. It should be noted that the first chapter provides a good link between valuation and financial theory. Although it may seem on the surface that any basic real estate investment text should focus on risk and return, chapters 7 and 12 in Real Estate Investment: A Capital Market Approach mark a departure. I find the material in chapter 7, which focuses on the relation between valuation (appraisal) and price, a refreshing distinction from the traditional treatment of valuation. Chapter 12 examines valuation smoothing, a topic that has drawn much attention in the real estate literature over the 1990s. Valuation smoothing is an area of research that reflects Gerald contribution to the literature, and I find the material refreshingly well written and succinct. The book provides arguably one of the best references on valuation smoothing. Personally, I like the way the optimal updating strategy for valuers is used to motivate valuation smoothing, drawing on works from game theory and economics. In addition, the text provides an excellent explanation on how the smoothing parameter can be estimated, how to detect smoothing and how property index aggregation is influenced by smoothing, and vice versa. The portfolio management section of the book covers the efficient markets hypothesis, inflation hedging, portfolio strategy and portfolio measurement, not unlike most finance textbooks. Chapter 14 examines an important feature of real estate – its inflation hedging property. The appendices to Chapter 14 contain many useful empirical studies covering the USA, UK, Switzerland and Hong Kong. Chapter 15 provides a good coverage of portfolio strategies. I find the chapter on performance measurement particularly interesting. Although it may seem somewhat trivial, performance measurement for real estate is not as
straightforward as it seems. Take the example of property company performance. There is evidence (in the book) to suggest that property companies should be measured with time-varying parameters. The book also provides thoughtful sections for readers who are not very mathematically inclined. The sum of some important series, probability distributions and how to graph functions are some examples. I find these to be very well written and would serve as good introductions for undergraduate students. Did I mention readability? This book is highly readable, period. Although I have agreed to review this book some time ago, I have put off this task time and again, for a simple reason – every time I read the book, I cannot help but imagine Gerald speaking to me through the text. The clarity of language, the kindly patience and the scholarly care that epitomize Gerald and his work came through ever so poignantly. Perhaps that is the main reason I keep coming back to Real Estate Investment: A Capital Market Approach. Seow Eng Ong
Book review
543