Do real estate prices and stock prices move together? An international analysis. by Daniel C. Quan and Sheridan Titman

Transcription

1 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 1 by Daniel C. Quan and Sheridan Titman A significant relationship exists between stock returns and real estate prices. This type of relationship, which is in contrast to findings in most of the existing studies, can be attributed to the added power resulting from the large cross section of the study. Also, it can be attributed to the use of lags in time-series regressions and to the longer holding periods utilized in the cross-sectional regressions. Changes in the relationship between stock returns and real estate prices are most likely to occur if corporate profits and rents are simultaneously affected by business cycle variables. COPYRIGHT 1999 American Real Estate and Urban Economics Association A number of authors have argued that commercial real estate offers diversification benefits to institutional investors because of its low correlation with commonly used stock price indexes. For example, using annual U.S. data from 1947 to 1982, Ibbotson and Siegel (1984) found real estate s correlation with SP stocks to be -.06, whereas Hartzell (1986), using quarterly data from 1977 to 1986, estimated the correlation to be Worzala and Vandell (1993), using the Frank Russell Index and more recent quarterly data from 1980 to 1991, estimated the correlation to be Worzala and Vandell (1993) estimated the U.K. real estate correlation with stock returns to be.039. More recently, Fu and Ng (1997) cite a low contemporaneous correlation between a transactions-based real estate index and stocks for Hong Kong over the period 1980 to Eichholtz and Hartzell (1996) document correlations of -.10, -.08 and -.09 for Canada, the U.K. and the U.S. between property and stock indexes(1). The low correlation between real estate and stock price indexes is somewhat surprising, given that both are affected by the level of economic activity and interest rates. However, other factors can reduce the correlation between the two time series. For example, stock prices may increase because of increased investment opportunities in an economy s corporate sector. This increase in investment opportunities could in turn lead to increases in real interest rates, which could reduce the value of commercial buildings even if their rental prices increase. Changes in the cost of labor could also induce a negative relation between stock prices and commercial real estate values. For example, foreign competition may lead to decreases in domestic wage rates, which in turn lead to increased corporate profits and higher stock prices. However, if wage rates in the building sector also decrease, then construction costs decline, and the value of commercial real estate will fall. Hence, changing labor costs can lead to negative correlations between commercial real estate values and stock prices. Real estate will indeed provide substantial diversification benefits for pension funds and other institutions if the low correlations between real estate and stock price indexes arise because the values of these capital assets react very differently to economic factors. However, the low observed correlation could alternatively be an artifact of the data. Because real estate trades infrequently, researchers must rely on indexes based on appraisals or inferred prices. Due to the nature of this inferring process, such indexes are often smoothed and thus underestimate the true volatility of the commercial real estate time series as well as the covariance between real estate price changes and stock returns. Hence, commercial real estate may provide less diversification benefit than would be indicated by a naive interpretation of the data. This paper provides new evidence on the relation between stock returns and real estate price changes by analyzing real estate price indexes, stock prices and macroeconomic data from 17 countries. These countries include most of the world s largest industrialized economies as well as some of the smaller economies in Asia s emerging markets. In contrast to earlier studies, we find a significant positive relation between stock returns and changes in commercial real estate values. After establishing this relation, we estimate additional regressions that provide some insights into why these two return series are correlated. One hypothesis is that real estate and stock prices are both driven up and down by changing expectations (either rational or irrational) of future economic growth that is independent of current fundamentals, like current rents and GDP. For example, in Japan, people often refer to the "bubble economy" in the mid-1980s which was characterized by real estate and stock prices being bid up to levels that could not be sustained. This argument suggests that real estate price changes should be more closely related to stock prices than to changes in the

2 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 2 macroeconomy. Alternatively, as illustrated by the recent situation in Asia, stock prices and commercial real estate values can rise and fall together because of changing political and economic fundamentals. This argument suggests that the relation between stock prices and real estate should weaken when one controls for changes in the macroeconomy(2). Our analysis suggests that the observed correlation between real estate values and stock prices arises because of current economic fundamentals and not because of expectations about future growth. Indeed, after we control for changes in rental rates and various macro variables, the relation between real estate values and stock prices weakens considerably. Moreover, we find that rental rates, which are the primary determinant of real estate values, are strongly correlated with GDP growth rates as well as stock returns. The analysis in this paper complements other recent studies of international real estate that have focused on the performance of the stocks of foreign real estate firms as opposed to the actual real estate assets.(3) Although these real-estate-related securities are interesting in their own right, and may offer an attractive alternative to owning foreign real estate, it is still somewhat unclear to what extent these assets offer investors a pure real estate play given their very low correlations with any of the appraisal-based indexes. The recent research focus on international real estate stocks rather than on actual real estate is attributable in part to the reliability and availability of commercial real estate price indexes outside of the U.S. Although we share these concerns, our analysis suggests that, by examining a large set of countries over a 14-year period, we are able to overcome some of the obstacles that have hindered earlier research that examined only a small number of countries. Data Description Our database is composed of capital values and rental indexes of prime office buildings from selected cities in 17 countries. The countries (and cities) are the Netherlands (Amsterdam), Spain (Madrid and Barcelona), Belgium (Brussels), Germany (Berlin, Dusseldorf, Frankfurt, Hamburg and Munich), France (Paris and Lyon), Italy (Milan), U.K. (London), Australia (Sydney and Melbourne), New Zealand (Auckland), Malaysia (Kuala Lumpur), Japan (Tokyo), Singapore (Singapore), Hong Kong (Hong Kong), Taiwan (Taipei), Thailand (Bangkok), Indonesia (Jakarta) and the U.S. (Boston, Chicago, Houston, Los Angeles, New York, Philadelphia, San Francisco and Washington). Except for Malaysia and New Zealand, we used Morgan Stanley s Capital International s Composite stock return indexes, which are market capitalization-weighted indexes of selected stocks traded in each country s exchanges. For Malaysia we used the Kuala Lumpur Composite Index, which is a capitalization-weighted index of 100 stocks listed in the Kuala Lumpur Stock Exchange, and for New Zealand we used the New Zealand Stock Exchange 40 index provided by Bloomberg, which is a capitalization-weighted index of 40 of the largest and most liquid companies listed in the New Zealand stock exchange. With a few exceptions, all GDPs, exchange rates, discount rates and inflation rates were obtained from the IMF International Financial Statistics Yearbook. The interbank money rate was used as the interest rate for France, while data for Thailand and Malaysia s GDP and Thailand s CPI were from Bloomberg. Hong Kong s exchange rate and CPI were from the UN s Statistical Yearbook. All real estate data were obtained from Jones Lang Wootton (JLW). JLW estimates of capital and rental values arise from consensus opinions, based on market transactions, about the price and rents paid for prime commercial properties within their respective markets. The capital series tracks the top values achievable for notional prime office property in the best location and fully let to a first-class tenant at the current market rent. No allowances were made for size discounts, acquisition costs, taxes or depreciation. The rental indexes similarly track the top open market net rent achievable for notional prime quality office space located in the primary office market in each city. The rents represent net "face" rents and do not factor in free rent and tenant improvement concessions. For each of the 17 countries, summary statistics for the 1984 to 1996 income and capital value indexes are provided in Table 1. The table reports the mean annual changes in commercial real estate capital values, commercial rents or income and the stock index, as well as the first-order serial correlation of the capital appreciation series. These statistics were calculated both in U.S. dollars and in each country s home currencies. The table reveals that some Asian countries, such as Hong Kong and Taiwan, experienced dramatic increases in real

3 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 3 estate values during this period, whereas most other countries experienced only moderate growth. Table 2 presents the same information over two subperiods - from 1983 to 1989, when most of the real estate and stock markets did exceptionally well, and from 1990 to 1996, when they did less well. The difference in the return patterns in the two periods is much more pronounced for real estate, and in some countries the actual prices fell in the second time period. To compare our sample with data used in earlier work, we calculate simple correlations between real estate price changes and rental rate changes and stock returns from 1983 to 1996 for each country. These correlations were calculated two ways: first, with all values measured in U.S. dollars, and second, with all values calculated in the home currencies. If the currency rate changes reflect mainly differences in inflation rates and if real changes in both stock and real estate values are independent of the inflation rate, then the correlations between stock returns and real estate price changes should be higher when the values are measured with a country s home currency. In this case, inflation will push stock prices and real estate prices up and induce a positive correlation. Although inflation-induced correlation will not arise if returns are all measured in U.S. dollars, an additional source of correlation will exist in this case if exchange-rate changes reflect changes in real purchasing power. In this case, dollar-denominated real estate and stock prices will move up and down together as the country s currency strengthens and weakens relative to the U.S. dollar. For example, if Japanese real estate and stock prices were unchanged, when measured in yen, during the weakening of the yen in 1997 and 1998, then their values calculated in U.S. dollars would drop. As it turns out, the correlations between stock returns and real estate price changes are similar when estimated with U.S. dollars and home currencies. These correlations, presented in Table 3, are relatively high for a number of [TABULAR DATA FOR TABLE 1 OMITTED] [TABULAR DATA FOR TABLE 2 OMITTED] countries; however, with the exception of Japan, where the correlation between dollar-denominated real estate value and stock price changes is.84, the correlations are not statistically significant. Table 3 * Correlations of capital and income values with stocks, Capital Income Capital Income Corr. Corr. Corr. Corr. Country (U.S. $) (U.S. $) (Domestic) (Domestic) Australia Belgium France Germany Hong Kong.13(a).25.12(a).25(a) Italy -.02(a) (a).33(a) Japan (a) Malaysia Netherland New Zealand Singapore Spain U.K U.S Mean a Sample is from 1984 to Empirical Tests The insignificance of the contemporaneous correlations described in the last section, as well as in earlier studies, is due in part to limitations of the available real estate data. In particular, as we saw earlier in the last column of Table 1, there is significant serial correlation in changes in the real estate value and rental rate time series. This serial correlation probably arises because the JLW real estate indexes are based, in part, on appraisals that reflect past transactions. As Quan and Quigley (1991) discuss, the appraisal process is likely to lead to indexes that are too smooth. In contrast, a number of empirical studies have suggested that stock returns are excessively volatile over short intervals.(4) Empirical studies of

4 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 4 the correlation between real estate price changes and stock returns must, therefore, face the unfortunate task of attempting to detect a relationship between excessively volatile stock price indexes with artificially smoothed real estate indexes. A possible way to overcome this problem is to compare stock and real estate price changes over longer holding periods. A longer holding period reduces the effect of smoothing on the real estate index as well as the excess volatility of the stock market index. In essence, although we may have little confidence in an appraiser s estimate of quarter-to-quarter or even year-to-year price changes, it is likely that measured price changes over longer intervals (like 5 to 10 years) reflect accurately the true price changes over these intervals; the actual price change should increase as the interval increases while the pricing error remains constant. A second approach is to estimate the effect of stock prices on real estate values using lagged as well as contemporaneous stock prices. By doing this we directly allow for the fact that the reported real estate price changes in any given year partially reflect price changes that actually occurred in the previous year. Cross-Sectional Tests With the exception of Indonesia, Taiwan and Thailand, which only had data from 1988 till 1996, the largest common period over which price changes can be calculated in our database is from 1983 until Due to this lack of data for the three countries, they were excluded from our cross-sectional regressions. For the remaining 14 countries we calculated the change in the value of the stock and real estate indexes as well as the changes in the other macro variables over this entire 14-year period. The following cross-sectional regressions examine the determinants of the long-term appreciation rates as well as changes in the rental rates of commercial real estate. We estimate simple regressions of real estate value and rental rate changes on stock returns as well as multiple regressions that include, as independent variables, changes in GDP, price levels and interest rates and - for the capital-value regression - changes in rental rates: RE = [[Beta].sub.0] + [[Beta].sub.1] x Stock + [[Beta].sub.2] x Inflation + [[Beta].sub.3] x Int. Rate + [[Beta].sub.4] x GDP + u (1) All values in these regressions are expressed in U.S. dollars. The results reported in Table 4 indicate that in the simple regression of changes in stock values on changes in real estate values the coefficient is [TABULAR DATA FOR TABLE 4 OMITTED] positive but only marginally significant. An inspection of the residuals indicates that Singapore appears to be an outlier; and when it is removed, the significance level increases substantially. Because we are dealing with a relatively small sample with potential outliers, we performed non-parametric tests to estimate the significance of the relation between stock returns and real estate value changes. We calculated the Spearman s rho statistic for the correlation between changes in capital value and stock returns for the whole sample as well as the two subsamples. The values obtained were.375,.714 and.648 for the full sample, the sample and the sample, respectively. The two subsample statistics were significant at the 99% confidence level, while the full-sample estimate was significant at the 90% level. Thus we reject the hypothesis that the two series are independent. When the macro variables are included in the regression, the coefficient on the stock variable becomes considerably more significant. This is because the change in GDP is very strongly related to changes in real estate values, and the addition of this variable substantially increases the precision of the other coefficient estimates. However, the inflation and interest-rate variables are both economically and statistically insignificant. Given that commercial real estate price changes as well as the stock returns in this regression are expressed in U.S. dollars, the insignificance of the inflation-rate coefficient estimate is consistent with commercial real estate being a good long-term hedge against realized inflation. In addition, the insignificance of the interest-rate variable suggests that changes in anticipated inflation do not have long-term effects on commercial real estate values. The final cross-sectional regression included changes in rental rates in the equation explaining changes in real estate values. As one would expect, changes in rental rates are highly correlated with changes in real estate values, so we expect the significance of the stock returns variable to decline when rental rates are added to the regression. However,

5 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 5 since stock prices and real estate values both incorporate expectations of future economic growth that is not captured by either the rental rates or the macro variables, stock returns should still have some marginal explanatory power.(5) This marginal explanatory power of the stock-returns effect is likely to depend on the length of the measurement interval. Over very short time periods, changes in expectations about the future are likely to be large relative to changes in actual rents and cash flows. Hence over very short intervals, stock prices are likely to offer a better prediction of changes in real estate values than changes in rental rates. However, over long time periods, changes in rental rates become much more important, so the marginal explanatory power of stock returns should become less important. As this argument predicts, including rent in our cross-sectional regression over the entire sample period makes the stock return variable statistically insignificant. Indeed, all of the variables become statistically insignificant. This lack of significance can, in part, be attributed to the possible presence of multicollinearity; rent is strongly related to stock returns as well as changes in GDP, thus making it impossible to separately determine their independent effects on changes in real estate values. Table 5 presents estimates of these same regressions for two different seven-year subperiods: and In the simple regressions of changes in real estate values on stock returns, the coefficient estimates using data from each of the two sample periods are about the same and are twice as large as the coefficient estimates using data from the entire period. However, when macro variables are added to the subperiod regressions, stock returns appear to be much more important in the later time period. As we saw for the entire-period regressions, the growth in GDP and changes in rental rates appear to be the most important determinants of changes in [TABULAR DATA FOR TABLE 5 OMITTED] real estate values in both time periods. Indeed, stock prices have no marginal explanatory power in the first subperiod when the other variables are included in the regression. However, in the second time period, stock returns have a significantly positive effect on changes in real estate values, even after controlling for changes in macro variables and rental rates. Time-Series Tests The cross-sectional regressions estimated in the previous subsection establish that over long time periods there is a positive relation between stock returns and real estate price changes. However, after controlling for rental rate changes and macro variables, stock returns have explanatory power only in the last time period. As we discussed in the last subsection, we have reason to believe that after controlling for macro variables and rental rates, stock prices are more likely to explain changes in real estate values as the intervals over which value changes are measured become shorter. This argument suggests that if we had accurate data, we should find a stronger relation between stock returns and real estate price changes in yearly data. However, this argument is somewhat offset by the smoothed nature of our real estate data, which, for returns measured at shorter intervals, should lower the correlation between real estate price changes and stock returns. This section presents time-series regressions that allow us to estimate the co-movements between stock prices and real estate prices on a year-to-year basis. These regressions are estimated for our entire sample as well as in various subsamples that include the two subperiods examined in the cross-sectional regressions, as well as separate regressions for each time period for Asian and European countries. By breaking the sample period in half, we are able to examine the stability of our estimates. Examining Asian and European countries separately is also of interest because the Asian stock market indexes include more real estate stocks and might therefore be more correlated with changes in real estate values. The estimated equation, given below, includes lagged stock returns as well as contemporaneous stock returns to take account of the fact that the real estate price indexes contain information from past transactions:(6) RE = [[Beta].sub.0] + [[Beta].sub.1] x Stock + [[Beta].sub.2] x Stock(-1) + [Beta] + u (2) Tables 6 and 7 present fixed-effects estimates of the above equation.(7) These estimates, which assume that the

6 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 6 coefficients are equal across countries, indicate that both real estate values and rents are correlated with stock returns, contemporaneously as well as with a lag. These results hold for the entire sample as well as for each of the subsamples. For the entire sample [TABULAR DATA FOR TABLE 6 OMITTED] period, the relation between stock returns and real estate price changes is much stronger in Asia than in Europe. However, in the individual subperiods, the two continents exhibit no significant difference in this relation. Tables 8 and 9 present fixed-effects regressions which include contemporaneous and lagged macro variables (GDP growth, the inflation rate and changes in interest rates) used previously in the cross-sectional regressions: RE = [[Beta].sub.0] + [[Beta].sub.1] x Stock + [[Beta].sub.2] x Stock(-1) + [[Beta].sub.3] x GDP + [[Beta].sub.4] x GDP(-1) + [[Beta].sub.5] x Inflation + [[Beta].sub.6] x Inflation(-1) + [[Beta].sub.7] x Int. Rate + [[Beta].sub.8] x Int. Rate(-1) + u (3) The regression estimates for the entire sample indicate that the positive relation between stock prices and real estate values and rents is still significant after controlling for macro variables. For the full sample period [TABULAR DATA FOR TABLE 7 OMITTED] [TABULAR DATA FOR TABLE 8 OMITTED] [TABULAR DATA FOR TABLE 9 OMITTED] in the regression described in the last three columns of Table 8, we see that the joint hypothesis that both contemporaneous and lagged stocks are insignificant is rejected at the 99% confidence level. The magnitudes of the coefficients, however, are about half as large as the estimates were without the macro variables. In contrast to the earlier simple regressions, we find that after controlling for the macro variables there is very little difference between the coefficient estimates for the European and Asian countries. These regressions suggest that the higher concentration of real estate stocks in the Asian indexes does not induce a higher correlation between real estate price changes and aggregate stock returns in these countries. The relation between the macro variables and real estate price changes is also of interest. Consistent with the cross-sectional regressions, real estate values and rents are significantly related to GDP growth rates. Moreover, the interest-rate variable is again insignificant. However, in contrast to the cross-sectional regressions, inflation does seem to be important. The evidence indicates that real (recall that rental rates and capital values are expressed in U.S. dollars) year-to-year rental rate and value changes are negatively affected by inflation. In other words, although real estate appears to be a good long-term hedge against inflation, it does not appear to be a good hedge on a year-to-year basis. Finally, in unreported fixed-effects regressions, we included changes in rental rates as an additional explanatory variable. As one would expect, changes in rental rates are very significantly related to changes in real estate values. However, the relation between rents and values is so strong that when rents are included in the regressions, all other variables become statistically significant. This is true for our Asian as well as our European subsamples, and for the whole sample period as well as in the two subperiods. These results suggest that changes in future expectations do not play a significant role in determining the correlation between real estate price changes and stock returns. Conclusion This paper examines the relation between stock returns and real estate prices in 17 countries. In contrast to the evidence in most of the existing studies, we find significant positive relations between both real estate values and rental rates and stock returns. Our ability to detect a significant relation between real estate prices and stock prices, not found in previous studies, is partly due to the added power arising from our large cross section. It is also due to the longer holding periods used in our cross-sectional regressions and the use of lags in our time-series regressions. The correlation between stock returns and real estate price changes will arise if business cycle variables simultaneously affect corporate profits and rents. In addition, stock and real estate prices will also move together if expectations (either rational or irrational) about future profits and rents move together. Indeed, in an environment where speculators push real estate and stock prices up and down because of changing expectations, there will be a significant relation between stock returns and real estate price changes that is independent of both business-cycle fundamentals and changes in rents.

7 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 7 Our results suggest that a large fraction of the observed positive correlation is due to economic fundamentals that affect both real estate and stock prices. Specifically, changes in real estate values and rents are both strongly correlated with changes in GDP. Moreover, the correlation between the year-to-year real estate value changes and yearly stock returns becomes insignificant when we control for both business-cycle variables and changes in rent, suggesting that changes in expectations about future growth is not an important determinate of the correlation between stock returns and real estate. Our results also have implications about real estate as an inflation hedge. The cross-sectional regressions allow us to investigate real estate s long-term sensitivity to realized inflation, an issue previous studies have not addressed. The time-series regressions examine the year-to-year sensitivity of real estate values to inflation - an issue examined in a number of single-country studies. Our findings show that although commercial real estate seems to be a good long-term inflation hedge, it is not a particularly good short-term inflation hedge.(8) The time-series regressions indicate that this latter result is due to the fact that dollar-denominated rents decline with inflation. It should also be noted that interest-rate changes are insignificant in all of the regressions, suggesting that changes in anticipated inflation have no discernible affect on real estate prices. Although our results are suggestive, our conclusions are ultimately no better than the quality of our data. We feel reasonably confident about the cross-sectional regressions using 7-year and 14-year price changes, since we expect the signal-to-noise ratio in observed price changes over long periods to be quite good. However, the yearly price changes are potentially quite noisy, so these regressions must be interpreted with more caution. Since the real estate price changes enter our regressions as the dependent variable, the estimates are suspect only to the extent that the estimation error in this variable is correlated with the estimation error in the independent variables. We have no reason to believe that this estimation error is correlated with either the stock-market indexes or the macro variables, and we thus feel reasonably confident about the yearly regressions that do not include the rental rate as an independent variable. However, there is likely to be some correlation between the measurement errors in the real estate price series and the corresponding measurement errors in the rental-rate series which could potentially bias the final regressions that examine whether stock returns have a significant effect on real estate values after controlling for rents. We think this question is quite important, since it provides some insights into how expectations in the real estate market and the stock market correspond. This is an important issue, and additional research, hopefully with better data, would certainly be of interest. We are extremely grateful to Katherine McNeil of Jones Lang Wootten New York, Nigel Roberts of JLW London, Morgan Stanley International and the Equity Index Group of Salomon Brothers for providing us with our data. The views presented are solely those of the authors and do not necessarily represent those of the Federal Reserve Board or its staff 1 In one of the broadest international diversification studies to date, Goetzmann and Wachter (1995), using a database composed of commercial asking rents and inferred yields for 21 countries over 1985 to 1993, argues that there is substantial diversification benefit from holding international real estate. 2 Goetzmann and Wachter (1995) have made similar arguments that global real estate activity is heavily influenced by each country s business cycle. 3 See, for example, Eichholtz and Hartzell (1996), Liu, Hartzell and Hoesli (1997) and Liu and Mei (1998). 4 See Shiller (1981) and Fama and French (1988). 5 In addition, there will be marginal explanatory power if there is a significant correlation between the discount rates applied to stocks and real estate in our sample of countries. 6 Gyourko and Keim (1992) also discuss appraisal lags and include lagged REIT returns as an independent variable in their regressions explaining the FRC index of U.S. commercial real estate values. 7 These specifications allow for country as well as year effects by adding country-year dummy variables to a regression that constrains the coefficients of the stock return variables to be fixed across both countries and time.

8 Real Estate Economics Summer 1999 v27 i2 p183(2) Page 8 8 The finding that commercial real estate is a poor year-to-year inflation hedge is inconsistent with prior studies that argued that real estate is a good inflation hedge (Hartzell, Hekman and Miles 1986; Gyourko and Linneman 1988). However, the results are consistent with more recent multi-country studies that concluded that real estate securities are a poor hedge against inflation (Liu, Hartzell and Hoesli 1997). References Eichholtz, P. and D. Hartzell Property Shares, Appraisals and the Stock Market: An International Perspective. Journal of Real Estate Finance and Economics 12: Fama, E. and K. French Permanent and Temporary Components of Stock Prices. Journal of Political Economy 96: Fu, Y. and L. Ng Fundamental News and Price Adjustment in the Real Estate and Stock Markets: Evidence from the Hong Kong Market. Working paper. Department of Economics and Finance, City University of Hong Kong. Goetzmann, W. and S. Wachter The Global Real Estate Crash: Evidence From an International Database. Working paper. Yale School of Management. Gyourko, J. and D. Keim What Does the Stock Market Tell Us About Real Estate Returns. AREUEA Journal 20: Gyourko, J. and P. Linneman Owner-Occupied Homes, Income-Producing Properties, and REITs as Inflation Hedges: Empirical Findings. Journal of Real Estate Finance and Economics, 1: Hartzell, D Real Estate in the Portfolio. E J. Fabozzi, editor. The Institutional Investor: Focus on Investment Management. Ballinger: Cambridge, MA. Hartzell, D., J. Hekman and M. Miles Diversification Categories in Investment Real Estate. AREUEA Journal 14: Ibbotson, R. and L. Siegel Real Estate Returns: A Comparison with Other Investments. AREUEA Journal 12: Liu, C., D. Hartzell and M. Hoesli International Evidence on Real Estate Securities as an Inflation Hedge. Real Estate Economics 25: Liu, C. and J. Mei The Predictability of International Real Estate Markets, Exchange Risks and Diversification Consequences. Real Estate Economics 26: Quan, D. and J. Quigley Price Formation and the Appraisal Function in Real Estate Markets. Journal of Real Estate Finance and Economics 4: Shiller, R Do Stock Prices Move Too Much to be Justified by Subsequent Changes in Dividends? American Economic Review 71: Worzala, E. and K. Vandell Internation Direct Real Estate Investments as Alternative Portfolio Assets for Institutional Investors: An Evaluation. Paper presented at the 1993 AREUEA meetings, Anaheim, CA.