Property portfolio briefing

Property Management

ISSN: 0263-7472

Article publication date: 1 May 2000

Citation

Wyatt, P. (2000), "Property portfolio briefing", Property Management, Vol. 18 No. 2. https://doi.org/10.1108/pm.2000.11318bac.002

Publisher

:

Emerald Group Publishing Limited

Copyright © 2000, MCB UP Limited


Property portfolio briefing

A Henry Stewart Property Portfolio Briefing on "Latest thinking on measuring risk" was held on 9 December in London. The briefing consisted of seven presentations by leading practitioners and academics in the field of property portfolio risk analysis.

The first speaker was Pam Craddock who, until earlier this year, was Director of Property at National Provident. Craddock reported the results of a questionnaire and interview survey undertaken earlier in 1999 to examine investors' attitudes to risk. The survey revealed that investors have a range of views on what risk is, many of which centred on the failure to meet a prospective target or benchmark. There was recognition that risk was a measure of volatility. However, investors are more concerned with total return performance and regard risk as the chance that performance objectives are not met. This is perhaps unsurprising given that most investors are judged by total return performance rather than risk volatility. Awareness of risk among investors focuses on the achievement of a benchmark whilst minimising exposure to downside risk. The measurement of risk at both the portfolio and property level is not attempted by approximately 90 per cent of respondents. So, despite the widespread use of DCF appraisal techniques, incorporation of risk measurement is not undertaken in general. Where risk measurement is attempted it usually takes the form of standard deviation, downside risk analysis and hurdle rates at asset and portfolio levels. An encouraging corollary to the lack of risk measurement was that 80 per cent of respondents were actively diversifying their portfolios. This is achieved by comparison with market indices. There was evidence of more sophisticated analysis at this level; rather than comparison to the total IPD Annual Index, many respondents were comparing their portfolios with customised sub-sets of the IPD universe. Investors, by comparing their funds against tailored benchmarks, may think that they are matching risk liabilities too. In this way benchmarking may be a proxy to other risk measurement techniques.

Stephen Palmer, a director at Guardian Properties, was next to speak. Palmer argued that property, as an investment asset, is different and risk measurement techniques used in other asset classes are not yet appropriate for property. Having said this, Palmer argued that there are competition, globalisation and securitisition pressures on property, an essentially deal-driven business, to align with other investment classes. There is, therefore, demand for greater market transparency and more research at the market level and this will lead to explicit pricing of risk. At the portfolio level risk management takes the form of diversification of assets and therefore MPT and CAPM are key tools. Palmer suggested that these tools, devised for the equity investment market, are not appropriate for property because of a lack of transaction data, reliance on valuations as a proxy for transaction data and the market leading index having far fewer data points than the equivalent stock market index. In practice Palmer argued that the main problem with diversification of property investment is illiquidity - the cost of adjusting betas in different markets is now (given recent increases in stamp duty) quite high.

Property risk is managed by reference to portfolio performance objectives, portfolio structure relative to market and risk characteristics of individual assets. However, it should be remembered that fund managers are rewarded for performance, not adherence to risk parameters. The question posed is how do you separate them? Palmer demonstrated, by analysis of the Guardian Properties portfolio, how a series of analyses help identify risk in the portfolio. Palmer suggested that perhaps too much attention has been paid to sector/region weightings rather than examining income, covenant, lease term, etc. of individual assets. This was suggested because more specific risk might impact on the property portfolio than would be the case with other asset classes.

Alan Patterson, Director and Property Analyst for WestLB Panmure Stockbrokers, provided a pragmatic view of risk from the equity investment market perspective. Comparison of the property risk premium to the equity risk premium was the theme of the presentation. The risk premium attempts to quantify risk in the risk return trade-off decision. Diversification of equity stock is no different to diversification in the direct property investment market. Patterson considered three methods of estimating the equity risk premium:

  1. 1.

    The CAPM measures risk in terms of standard deviation of returns. As you diversify away non-market risk to be at the point on the efficient frontier that touches the capital market line, higher betas mean more market risk. The essence of the model is that there is a linear relationship between risk and return. Patterson identified a number of problems with the theoretical assumptions of the CAPM and suggested that betas are very difficult to estimate as they vary over time and actually what is required is a "future" beta. In practice, not only might the risk-free rate change as the yield on gilts changes but also there is a significant difference between the heuristic equity risk premium (2-4 per cent) and the historic average of 8 per cent. This is known as the "equity risk premium puzzle" and is perhaps due to market expectation of low inflation dampening the risk premium. In practice the CAPM is not used by equity investors.

  2. 2.

    An equity risk premium based on accounting fundamentals is used when trying to determine the cost of equity or capital of a company.

  3. 3.

    Arbitrage pricing theory is considered useful because when property market sectors are regressed against the market there is very low correlation. Patterson considered the range of variables that affect property returns in the three main sectors and the primary variables that affect property yield shifts.

Patterson concluded by suggesting one further method of examining a property's risk premium; think of it as a bond. The return on a bond consists of a risk-free element, a risk premium and a default premium in the same way that a property return can be considered to consist of a risk-free element, risk premium, tenant default premium and a re-letting risk premium.

Dr Ian Cullen, a founding director of IPD and the next speaker, asked whether the assessment and control of risk should be central to property portfolio management. Cullen pointed out that only half of IPD investor clients request customised portfolios for performance benchmarks. The IPD Annual Report includes risk measurement statistics but these are limited to one page and include standard deviations of returns, tracking error, Sharpe ratios, risk-adjusted returns and return covariance. Cullen suggested that there are further risk analyses that should be of interest. These include portfolio lot size concentration and asset profile idiosyncrasy, stock specific out-performance, yield dependent capital growth, income concentration, growth potential and sources of income insecurity. Stock selection is a major factor affecting return, not asset allocation. Cullen concluded by suggesting that current risk analyses miss a great deal of the point of property risk management and most of the diagnostics used for explaining performance differences raise questions of risk identification, measurement and management. Consequently a better approach to property market risk is needed.

Paul Richards, European Director of Property and Portfolio Research (PPR), began by stating that PPR use CAPM and MPT to analyse investment portfolios. Richards suggested that risk at the portfolio level is usually managed at the property level through sector and location diversification but went on to point out that risk at this level stems from the occupier's market as well as the property investment market. Once an investor's risk/return requirements have been defined Richards suggested techniques for managing investments to balance the forecast of risk and return. These techniques include cluster analysis of the IPD portfolio to provide sectors and locations that are substitutable in terms of investment behaviour, standard deviations to measure volatility/dispersion of returns around the mean and skewness to measure the "shape" of returns. By using IPD data, Richards presented a series of measures for the above to exemplify risk/return profiles of locations and sectors of the property investment market.

Richards went on to ask if there was a risk/return trade-off (as posited by CAPM) in the property market. Charting returns against standard deviation of returns of IPD sectors and regions reveals no obvious bottom left-top right trend. A more discernible trend is revealed in the US data but it must be noted that return and risk change over time so the time period of measurement is very important. Richards concluded by suggesting that MPT principles can be used to combine uncorrelated assets to manage risk but pragmatism in the application of the theory is needed.

Colin Lizieri, Professor of Real Estate Finance at the University of Reading, described how three statistical analysis techniques, namely cluster analysis, discriminant analysis and factor analysis, may be used to identify homogenous groups of properties as an aid to asset allocation within a property portfolio. Cluster analysis assigns properties to a cluster if they exhibit similar behaviour, e.g. investment returns over time. Discriminant analysis tests whether properties are correctly assigned to clusters. Factor analysis identifies common patterns of variation in order to reduce the number of variables to a smaller number of factors or components. Having introduced these techniques, Lizieri gave examples of their practical application.

CB Hillier Parker rent and yield data were used in a cluster analysis to see if a superior clustering of sectors and regions than that used by IPD (as measured by investment performance) can be generated. The initial cluster analysis selected nine clusters, which were a combination of sectors and regions, e.g. Scottish Retail or City offices. A clustering solution, revised in light of intuitive adjustment, was compared to the IPD classification using MPT. The results demonstrated slightly superior performance with the new groupings being perhaps more homogenous. It is important to note that economic change may warrant revised clustering.

Street-level data on rents and yields for City, Mid-Town and West End of London were analysed to see if the City market behaved in a coherent manner, i.e. were the boundaries of that area correct? Cluster analysis grouped the data and discriminant analysis was used to test the results. The City appeared to behave consistently with well over 80 per cent of streets correctly classified. Behaviourally, it is perhaps inevitable that the boundaries will be correctly classified. Occupiers wishing to locate in the City will select a City address - if the City boundary were moved out then occupiers wishing to locate in the City would locate in the extended area and behave in a similar fashion to existing occupiers. Nevertheless the implications for property portfolio management are clear.

An example of the use of factor analysis can be found in Goetzmann and Wachter's study of office rents and yields for a sample of cities around the world. The results showed that a single "real estate" factor explained a high proportion of the variation in returns. The authors suggest the result (which Lizieri says should be treated with some caution) implies that there is a high level of systematic risk and less diversification potential than might be expected. Another application of factor analysis by Lizieri and Lee examined possible links between sub-sectors of the stock market and the property market, e.g. City offices and financial firms. In other words the same economic forces might drive performance in these sub-sectors. This has obvious implications for risk management and diversification. Applying factor analysis to IPD property sector and region returns and Datastream stock market returns by sector results in two factors explaining 61 per cent of the total variation. Removal of these two factors (stock market and property market) reveals long-run relationships between sub-sectors of the property and stock markets (e.g. between telecommunications and utilities shares and industrial and offices shares in the South East, between general retail firms and southern shop property and between city offices and financial firms). The implications of the study are that to avoid excess specific risk that may arise if these long-run linkages are ignored, there needs to be better communication between property and equity teams in a top-down asset allocation strategy.

Summarising, Lizieri argued that clustering may be used to structure portfolios and help manage risk, discriminant analysis provides a formal test of the clustering process and factor analysis is useful in explaining patterns of performance and market behaviour.

The final speaker was Stephen Ellis, of King Sturge, who gave the surveying practitioner's view of some of the risk management techniques discussed so far. Ellis suggested that diversification strategy (attempting to hold assets with negative covariance with one another to reduce market risk) was difficult to implement in practice for all assets but it is done intuitively. To reduce specific risk at least 30-40 properties are needed to avoid over-dependence on one or two assets in a portfolio. Portfolio benchmarking is undertaken and the primary objective is to outperform the benchmark; often risk is increased to achieve this objective. MPT is not undertaken because the efficient frontier is difficult to plot for property assets, it is difficult to calculate variance and risk objectives are generally not quantified by the client. Trading assets in a bid to manage volatility in rising and declining markets is not easy in practice because of illiquidity.

To summarise, investors are primarily concerned with return performance, typically measured against a benchmark. This is because investors are remunerated on the basis of their performance, not risk. They are less concerned with assessment of volatility of their returns. When risk measurement is undertaken, risk is pragmatically regarded, rather simplistically, as the chance of not achieving benchmark return. The main measure of risk is standard deviation and the focus is always on downside potential. More sophisticated measures of risk in terms of volatility are not in general use in the property investment market yet.

Peter Wyatt