Mobile Edition
Print
Get the Mag
Weekly UpdatesLim et al. (2008) provide an analysis of the correlation of property returns in 15 countries, including Finland, based on the IPD property index data. Their analysis suggests that the correlations between the property returns in the countries vary significantly, from almost complete correlation (99%) between some of the studied countries to high negative correlations between some. Thus the diversification benefits obtainable vary markedly depending on the markets studied.
3. DATA AND METHODS
The study uses annual data on Finnish stocks, bonds, direct real estate and indirect real estate as well as annual real estate return data on fifteen property markets: Canada, Denmark, Finland, France, Germany, Ireland, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, U.K. and U.S. for the time period 2002-2006, i.e. the first five years where international investors have been present on the Finnish property market.
The returns of Finnish assets were obtained from KTI Finland. The bonds returns are based on EFFAS Finland Government Bond Index. Stock index is the OMX Helsinki Cap gross index and indirect property index the OMX Helsinki Real Estate gross index. All returns are total returns including income returns and capital appreciation. Finnish direct real estate returns were obtained from KTI Index, an IPD-compliant property index. International property returns are All Property-indices obtained from IPD Multinational Index Spreadsheet. International diversification benefits are analysed by using unhedged returns, as well as Euro-dominated returns.
The use of index-based property data has some drawbacks that should be taken into account when conducting the analysis. Firstly, indices reflect the investment policy of institutional investors in the country in question. Thus, the divisions between e.g. property types vary between countries (Hoesli et al., 2004). Secondly, as indices are constructed of the returns of a large amount of properties, they reflect the returns of a well-diversified portfolio. This assumption might be violated in case of international investors entering a new market. And thirdly, as most of the IPD indices are appraisal-based, they are affected by valuation smoothing. The phenomenon leads to the standard deviation of time series to be underestimated and also affects the correlations of asset returns (see e.g. Geltner, 1989, 1993; Diaz and Wolverton, 1998). There are several methods for correcting, i.e. de-smoothing, the time series to track the true volatility of property data. The shortness of the time series used in this study, however, makes the correction for the bias difficult, as the data does not cover a total cycle. Thus, no corrections for valuation smoothing were performed on the data.
This study analyses the diversification benefits provided by the Finnish real estate market using the Modern Portfolio Theory (MPT), a framework by Markowitz (1952). However, unlike most studies using MPT, this study focuses on analysing historical returns and diversification benefits. The use of ex-post data has the benefit of avoiding future predictions of the asset returns, especially in the case where the data period is too short to cover an entire business cycle. The optimisation applies a no short-selling constraint, i.e. all asset allocations in optimal portfolios must be zero or larger, and all asset weights must sum to one.
The application of mean-variance optimisation often leads to the creation of portfolios with corner solutions, i.e. few assets enjoying very large allocations and the other having zero allocations (Black and Littermann, 1992). To encounter the problem, the optimisation of international real estate portfolio is conducted in two stages. First, unconstrained optimal portfolios are created. To avoid extreme allocations in individual countries, the second stage of optimisation was conducted using a 20% maximum allocation constraint for each country.
4. DIVERSIFICATION BENEFITS IN THE FINNISH MARKET
4.1. Finnish mixed asset portfolio
The analysis of diversification benefits begins with an analysis of a mixed asset portfolio consisting of only Finnish assets, i.e. Finnish stocks, bonds, returns on investments in public real estate investment company shares (RE stocks) and real estate. If an investor is indifferent of the division of his asset into different asset classes, but prefers a country exposure, the mixed-asset portfolio offers insights in seeing if investors should invest in Finnish real estate at all. The property-related risk factors omitted by MPT, i.e. the lumpy nature of the asset class combined with the small size and thus rather low liquidity of the Finnish property market imply that should the optimal mean-variance portfolio not include real estate, the investor would prefer investing his wealth into stocks and bonds.
The descriptive statistics of the asset returns are given in Table 1. As typical, the average return on direct real estate was between that of stocks and bonds. Interestingly, the returns on RE stocks were markedly higher, more than five times the returns on direct real estate and doubled the return on stocks. The volatility of asset returns is illustrated by standard deviation and by coefficient of variation, which measures the ratio of standard deviation to average return, i.e. the average risk per unit of return. The coefficient of variation was the lowest for direct and indirect real estate. In comparison to those, the coefficient of variation was surprisingly high for bonds, whereas stocks carried, as expected, by far the most risk per unit of return.
In addition to the return levels, correlation coefficients of asset returns are of importance when choosing the structure of a portfolio. The smaller the correlation coefficient, the larger the diversification benefits. The correlation matrix for Finnish asset returns is illustrated in lower part of Table 1. As shown in the table, Finnish bonds had a high negative correlation with all the other asset classes studied, suggesting significant diversification benefits. The correlation between stocks and direct real estate, as well as stocks and RE stocks, was positive, suggesting limited diversification benefits. The correlation between direct real estate and RE stocks was high and positive; suggesting that diversification within the real estate asset class did not provide diversification benefits, but that they could have been close substitutes within the studied time period.
The structure of efficient portfolios is illustrated in Table 2. As could be expected from the risk-return structure and correlation characteristics, RE stocks dominate the efficient portfolio in all but the lowest return levels. Bonds enter the efficient portfolios only at the lowest return levels, whereas stocks are present at the medium to high return levels. Direct real estate enters the efficient portfolio at return level of 16%. The allocation decreases from this maximum of almost 70% to zero at return levels of 36%.
The results of the portfolio optimisation differ from most of the empirical research, where direct real estate is typically characterised of having a high allocation in the portfolios with low return and risk. This can be explained by the exceptional risk-return characteristics of RE stocks under the study period and the relatively poor risk-return trade-off of stock and bonds returns.
As mentioned earlier, the results of the analysis might be biased due to the appraisal-smoothing phenomenon. The bias on the optimal allocation is expected to be larger in the mixed-asset portfolio than in the international real estate only portfolio, because in the mixed-asset portfolio case direct real estate is the only asset that is exposed to the phenomenon. As the shortness of time series would limit the reliability of de-smoothing procedures, the effect of volatility increases were simply tested by doubling the standard deviation of the direct property returns, which gives a rough picture of the sensitivity of the result to asset volatility changes, but unfortunately does not take into account the change in correlation structures. In this case the increase in standard deviation leads to real estate entering the efficient portfolios later, at the return level of 22%. Between the return levels of 22 and 34% changes in the optimal allocation remained small, at less that 0,6%. It should be noted, however, that even with the doubled property return volatility the coefficient of variation of property returns was still lower than for bonds.
4.2. International real estate portfolio
Hedged returns
The second part of the analysis focuses on Finnish real estate investments in an international real estate portfolio. The analysis is started with returns in local currencies, which corresponds to the situation where the investor has hedged all the risks of currency fluctuations.
Table 3 presents the descriptive statistics of property returns in the studied property markets during 2002-2006. Concentrating only on the average returns measured in local currencies, the Finnish property market was not that attractive during the observation period. The level of Finnish property returns was below the average (10,6%), and third lowest of the countries studied. The highest returns were provided by the Irish and U.K. markets, where the return levels exceeded 15%.
In respect to the risk-return trade-off the Finnish market was more attractive, the coefficient of variation being the third lowest in the dataset. Extremely low coefficient of variation was provided by the Swiss market, which is interesting, as the Swiss index is the only IPD-index which is based on hedonic modelling, instead of appraisals, and should thus not suffer from appraisal bias. The least attractive risk-return trade-offs were provided by the German and Swedish markets.
FEED