Price-Earnings Ratios, Dividend Yields and Real Estate Stock Prices Executive Summary. Both dividend yields and past returns have predictive power for P/E ratios; hence they can be used as tools in forming a market timing and asset allocation strategy in stock markets. This study examines the extent to which changes in real estate returns, reflected in changes of property value and dividend yields, can have great effects on P/E ratios. The study is confined to four major real estate stocks in Hong Kong. It shows that a low dividend yield appears to be associated with a relatively high price-to-earning ratio. Variance of dividend yields tends to increase relative to the variance of earnings yield, with a rapid dividend adjustment at higher dividend payout ratios. *Hong Kong Institute of Real Estate and International City University of America, Hong Kong or Chairman@Hkir.Com. by Raymond Y. C. Tse* Introduction Compared to the capitalization rate of real estate, little is known about the forecasts of price-toearnings ratios (P/E) of real estate stocks. This is because capitalization rates of real estate are fundamentally different from real estate stock s P/ E ratios. Real estate rents are determined in space market whereas earnings reflect the performance of a company. There are several reasons why real estate returns have a significant impact on the performance of property companies. Most developers have significant amounts of real estate loans; real estate returns should logically have a significant impact on their profitability. A weak real estate market will eventually lead to a reduced cashflow for the property company. Tse and Webb (2000) have shown that real estate stocks, with a larger portion of their assets invested in residential property, are most sensitive to changes in residential property prices. While a good prospect of real estate markets justifies a higher P/ E ratio, concern has arisen that the real estate stock prices may be headed for a downturn because firms share prices have become very high relative to their earnings when the real estate market was booming. Alternatively, it has been argued that high priceearnings ratios have usually been followed by slow growth in stock prices. Campbell and Shiller (1998) found that price-to-earnings ratio was negatively correlated with subsequent stock price growth but uncorrelated with subsequent earnings growth. Furthermore, a high P/ E ratio means the earnings yield is low, thus indicating that stocks are expensive relative to alternative investments. Stock market investors are interested in the forecasting power of variables like dividend yields and Journal of Real Estate Portfolio Management 107
Raymond Y. C. Tse past returns as tools in market timing in highly volatile stock markets. Fama and French (1998) note that the growth style is associated with stocks with relatively high P/ E ratios, high profitability and consistent growth. However, real estate stocks are differentiated from common stocks with respect to the heterogeneous nature of real estate and inherent volatility of real-estate markets. A major problem in microeconometric analysis of real estate demand involves the capital market constraints that restrict their abilities to borrow against their entire stream of future income. Moreover, the high transaction costs of real estate tend to impede adjustments to changed prices and incomes. In the 1980s, the real estate literature began to pay attention to the choice between real estate and real estate stocks. Unlike the common stocks, various characteristics of the real estate market complicate the transaction process. Among these characteristics are: (1) locational differences of real estate; (2) heterogeneity of real estate; (3) infrequent transactions among sellers and buyers; and (4) complexity arising from the financial and legal dimensions of the transactions. Unlike the real estate market, stocks trade in continuous auction markets with low transaction costs, large volume, and large pools of informed buyers and sellers. In short run equilibrium, there is some distribution of market prices for identical properties (Quan and Quigley, 1991). In the real estate market, the interaction of agents is more complex than in the stock market (Mantrala and Zabel, 1995). For instance, vacant real estate is more illiquid than occupied real estate (Zuehlke, 1987). Previous research has tended to focus on either how stock returns can be predicted by P/E ratios (Fama and French, 1988b; Fama and French, 1998; Ferson and Harvey, 1997; and Rouwenhorst, 1999) or how stock and real estate markets are related to each other (He, Myer and Webb, 1996; Eichholtz and Hartzell, 1996; Quan and Titman, 1999; and Tse and Webb, 2000). Earlier literature (such as Owen and Grundfest, 1977) shows that there is an analogy between the real estate market and the stock exchange. While important differences exist in these markets, both are subject to similar competitive pressures. In particular, Ross and Zisler (1991) show that real estate returns are roughly comparable to stocks. This study develops an extension of the dividend adjustment model based on Lintner (1956). The objective is to investigate the ability of dividend yields and past returns to predict future P/E ratios of real estate stocks at the firm level after controlling for error corrections. The results indicate that both dividend yields and past returns have predictive power for P/E ratios; hence they can be used as tools in forming a market timing and asset allocation strategy in stock markets. A particular contribution of this study is that for the first time it is found that changes in real estate returns, reflected in changes of property value and dividend yields, can greatly affect the P/E ratios. The impact of the price growth on the P/E ratios between different real estate stocks is also compared. Data This study is confined to four major real estate stocks in Hong Kong, namely Cheung Kong, Sun Hung Kai, Henderson and New World Development. The four major property companies mainly invest in Hong Kong. Thus, these companies are expected to have characteristics similar to that of real estate stocks. However, differences in the market structures between real estate and stocks may affect the behavior of public real estate as a proxy for private real estate. Note that there is a very high concentration of real estate in the stock market in Hong Kong. For instance, the four real estate companies constituted about 13% of the total market capitalization of the blue chips in March 2000. End-of-month average market P/ E, stock prices and dividend yields were obtained from the Hong Kong Securities Report between 1991 and 2000. The stock price growth rate represents a onemonth return. The data set spans January 1991 through January 2000. In the data set, no negative P/ E ratios are observed. Exhibit 1 presents summary statistics for the P/ E ratios, dividend yields and stock price growth rates 108 Vol. 8, No. 2, 2002
Price-Earnings Ratios Exhibit 1 Data Statistics P/E Yield Price Growth Panel A: Cheung Kong Mean 11.67 2.62 8.74 Median 9.94 2.69 9.15 Maximum 37.12 4.25 71.82 Minimum 4.51 1.18 41.67 Std. Dev. 6.37 0.74 22.14 Panel B: Henderson Mean 12.23 4.97 7.87 Median 12.23 4.56 7.34 Maximum 22.70 12.05 140.51 Minimum 4.51 2.18 45.11 Std. Dev. 3.40 2.25 30.63 Panel C: New World Mean 14.06 3.958 4.92 Median 12.85 3.450 3.28 Maximum 28.23 11.54 97.12 Minimum 3.63 1.71 45.16 Std. Dev. 5.22 1.91 30.65 Panel D: Sun Hung Kai Mean 15.18 3.69 7.55 Median 15.11 3.34 7.19 Maximum 23.29 9.58 105.80 Minimum 4.63 1.55 40.66 Std. Dev. 4.34 1.65 27.37 for the four real estate stocks over the period January 1991 through January 2000. Variation across stocks for all variables is remarkable. For example, simple average P/ E is 11.67 for Cheung Kong, 12.23 for Henderson, 14.06 for New World and 15.18 for Sun Hung Kai with a standard deviation of 6.37, 3.40, 5.22 and 4.34, respectively. The dividend yield is about 3% to 5%. It is interesting to note that the mean price growth rate is about 5% to 9% with a standard deviation of as high as 30%. For all four stocks, the maximum price decline is about 40%, indicating that they were subject to similar downside risk. Tests for Stationary Component Early tests of market efficiency only examined autocorrelations of stock returns. This study examined the autocorrelations of P/E ratio with a time-varying framework. The P/ E ratio pe(t) was modeled as the sum of a random walk, (t), and a first-order autoregression (AR1) with a slowly decaying parameter,. pe(t) pe(t 1) (t), (1) (t) (t 1) (t), (2) where is expected drift and (t) is white noise. The long temporary pe(t) swings assumed in models of an inefficient market imply a slowly decaying. It is hypothesized that P/E ratio would not take long temporary swings away from fundamental values, which is translated into the statistical hypothesis that P/ E has slowly decaying stationary components. The tests are based on the proposition that the mean-reverting property of the P/ E ratio implies that is less than 1.0. The equation system can be estimated with a time-varying framework. In the results for the sample period (Exhibit 2), 0 1 is consistent with the hypothesis that mean-reverting P/ E ratios are important in the Journal of Real Estate Portfolio Management 109
Raymond Y. C. Tse Exhibit 2 Estimation of Time-Varying Parameter Model Cheung Kong Henderson New World Sun Hung Kai 1.084*** ( 2.731) 0.972*** ( 3.013) 1.299*** ( 5.798) 1.432*** ( 7.334) 0.454*** (3.628) 0.400*** (2.896) 0.513** (5.231) 0.542*** (6.144) Log Likelihood 18.55 4.15 14.08 6.98 Adj. R 2 0.170 0.142 0.242 0.276 SE 0.419 0.278 0.369 0.301 DW 1.520 2.058 1.998 1.826 Notes: t-statistics appear in parenthesis. * Significant at the 0.1 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level. variation of P/ E ratios. The estimates suggest that predictable variation of the P/E ratio due to mean reversion is about 40% 55% of its own previous value. Thus, the results add to mounting evidence that P/ E ratios are predictable. Furthermore, it is expected that P/ E ratio is more predictable than sheer prices because prices cannot take long temporary swings away from their earning power. The Model Following Lintner (1956), a firm s target dividend D*(t) for year t is assumed to be a constant fraction of earnings E(t), D*(t) ke(t). (3) In addition, the change in the actual dividends from t 1tot is assumed to follow a partial adjustment model: D(t) D(t 1) (D*(t) D(t 1)). (4) Combing Equations (3) and (4) yields: D(t) (1 )D(t 1) ke(t). (5) Equation (5) yields: j D(t) k (1 ) E(t j ), (6) j 0 which implies that dividend at time t is in fact a distributed-lag function of earnings. For 0 1, the effects of earnings diminish over time. Dividing both sides by P(t) produces: Thus, where: D(t)/P(t) (1 )D(t 1)/P(t) ke(t)/p(t). (7) y(t) (1 )[D(t 1)/P(t 1)]/(1 g) k/pe(t), (8) y(t) D(t)/P(t), pe(t) P(t)/E(t) and g(t) (P(t) P(t 1))/P(t 1). Thus, y(t) (1 )y(t 1)/(1 g) k/pe(t). (9) Alternatively, pe(t) k(1 g)/((1 g)y(t) (1 ) y(t 1)). (10) Hence Equation (10) indicates that P/ E ratios are a function of y(t), y(t 1) and g(t), such that: 110 Vol. 8, No. 2, 2002
Price-Earnings Ratios Exhibit 3 Regression Results for Price-to-Earning Ratio Cheung Kong Henderson New World Sun Hung Kai Constant 2.767*** (0.264) 3.016*** (0.092) 3.767*** (0.074) 3.675*** (0.043) g(t) 0.874*** (0.329) 0.525*** (0.091) 0.175** (0.074) 0.197*** (0.052) log(y(t)) 0.513* (0.272) 0.368*** (0.058) 0.938*** (0.056) 0.831*** (0.033) Error-correction (t 1) 0.352*** (0.303) 0.753*** (0.101) 0.632*** (0.124) 0.858*** (0.100) Adj. R 2 0.277 0.809 0.916 0.962 SE 0.382 0.128 0.121 0.068 DW 1.432 1.847 2.114 2.113 Notes: t-statistics appear in parenthesis. The number of observations 37. * Significant at the 0.1 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level. pe(t) f(y(t), y(t 1), g(t)). (11) Fama and French (1988a) suggest that dividend yields are highly autocorrelated but slowly meanreverting stationary processes. There is much evidence that variation in the expected stock return generates autocorrelation in the dividend yield. However, if the price growth rate and dividend yield are stationary (mean-reverting), the P/ E ratio is also stationary. Thus, it is expected that y(t) and y(t 1) tend to be highly correlated, then the following regression is used: log( pe(t)) log( y(t)) g(t) 0 1 2 Error Correction. (12) t 1 This requires a two-step estimation procedure: log( pe(t)) 0 1 log( y(t)) 2 g(t) u(t) is regressed yielding: log( pe) ˆ ˆ 0 1 log( y) ˆ 2 g. The following equation is then regressed: log( pe(t)) log( y(t)) g(t) 0 1 2 (log( pe ) ˆ 3 t 1 0 ˆ log( y ) ˆ g ). (13) 1 t 1 2 t 1 Equation (13) provides a simple intuition for Equation (11), and produces approximately unbiased estimates of the P/ E ratio under more general assumptions. Equation (13) suggests two possible sources of determinants of P/ E ratio: dividend yields and stock price growth rates. An increase in the price-to-earnings ratio could occur in two ways through faster growth in stock prices or slower growth in dividend yields. In Equation (13), the current price growth rate g(t) is relevant. This suggests that today s high P/ E ratio signals faster growth in stock prices not only in the long term but also in the short term. The implication is that even if investors have a long-term horizon, they still have to make short-term investment decisions. Exhibit 3 presents the results of Equation (13). All the coefficients are significant, but the regressions are difficult to interpret. The coefficients of the intercept are all positive and relatively stable. The slopes on the two variables have opposite signs, positive in the regression for stock price growth rates and negative for dividend yields. The coefficient on price growth rate g(t) is highest at 0.874 for Cheung Kong, compared to 0.525 for Henderson, 0.175 for New World and 0.197 for Sun Hung Kai. The effect of dividend yields on P/E is strongest in New World and SHK at 0.938 and 0.831, respectively. The negative coefficient on (log) dividend yield suggests some predictability of the P/ E ratio due to the mean reversion of dividend yields. The negative P/ E slopes say that P/ E ratio is Journal of Real Estate Portfolio Management 111
Raymond Y. C. Tse likely to be high when current dividends are low relative to current stock prices. High P/ E ratios tend to reduce the earnings yield on stocks relative to returns on other investments. A lower dividend yield may indicate that firms retain cash for internal use rather than for dividend payments. Dividends provide a cheaper source of capital and, if wisely invested, will possibly increase earnings and dividends in the future. Thus, lower dividends are usually paid to avoid higher costs of external financing (Myers, 1984). All of these changes suggest possibly higher P/E ratios if earnings are expected to grow persistently faster than before. There is no apparent trend in P/E ratios during the period under concern. Based on the historical behavior of stock prices, P/ E and dividend yields, there is little evidence that P/E ratio is high when dividend yield is high. However, there is evidence that P/E ratio is high when price growth rate is high. The P/E ratio was high relative to the dividend yield before 1997. The sudden decreases in the difference between the P/ E and dividend yield in 1998 was largely due to the Asian Financial Crisis, resulting in a substantial drop in stock prices. Some of the decline in the P/E ratio was probably due to decreases in stock prices and increases in dividend yields, but most of it is unexpected. Note that dividend yield is driven by expected future returns and expectations about future dividends. However, dividend yields are essentially unpredictable (Campbell, 1991; and Cochrane, 1991). There is, however, a problem in using Equation (10) to estimate the P/E ratio, since both and k remains unknown. In general, stock prices, dividends and earnings have different volatilities. Short-term changes in P/ E ratios and dividend yields are largely due to changes in stock prices, rather than earnings and dividends respectively. In other words, changes in stock prices lead to simultaneous changes in P/ E ratio and dividend yields. How is the volatility of these three variables connected? Put a bit differently, the variances of price growth rate, P/E ratio and dividend yields are closely connected. Taking the variance of both sides of Equation (10) produces: Var( pe(t)) Var{ k/[y(t) (1 )y(t 1)/(1 g(t))]}. (14) However, Var( y(t)) Var( y(t 1)) Var( y). Thus, 2 2 Var( pe) ( k) /(Var( y) (1 ) Var( y)/var( g)), 2 2 2 Var( pe)var( y) ( k) /(1 (1 ) / g ), (15) where g2 Var( g). Since 0 k 1 (for, k 1), and (1 ) 2 / g2 0, it can be determined that: Var( pe)var( y) 1. Since the reciprocal of the P/E ratio 1/pe earnings/price, Var(1/pe) 1/Var(pe) represents the variance of the earnings yield (earning-to-price ratio). However, shareholders prefer a steady stream of dividends. This result is also consistent with Fama and French (1988b) that dividend tends to be more stable than earnings, implying that variance of dividend yield should be smaller than the variance of earning yield. Equation (16) implies variance of dividend yields tends to increase relative to the variance of earning yield, when the dividend adjustment speed ( ) and dividend payout ratio (k) increase. Conclusion This article presents a methodology for investigating the behaviors of P/E ratio on real estate stocks. The results have important implications with respect to the predictability of P/ E ratios. However, this study focuses on real estate stocks at the firm level. In Hong Kong a good deal of stock market volatility can be attributed to the real estate market (Tse, 2001). For example, due to the financial turmoil in October 1997, the Properties Stock Index performed poorly during this period. The property stock P/E ratio decreased from 15.7 in July 1997 to less than 5.0 in July 1998, down by nearly 70%, which is the lowest among all the Hong Kong stock sub-indices. Real estate stocks are, however, differentiated from common stocks with respect to the heterogeneous nature of real estate and the inherent volatility of real-estate markets. This study shows that an increase in a P/E ratio could occur in two ways through faster growth in 112 Vol. 8, No. 2, 2002
Price-Earnings Ratios stock prices or slower growth in dividend yields. The implication is that there is some predictability of the P/E ratio due to the mean reversion of dividend yields. A low dividend yield appears to be associated with a relatively high P/E, since a lower dividend yield may indicate that dividends are retained for internal use. Thus, lower dividends are paid to avoid higher financing costs and increase future profits. Consequently, a higher profit will lead to higher stock prices and P/E ratio. These findings are consistent with Myers s (1984) pecking order and stakeholder hypotheses of dividend behavior. There is also evidence that P/ E ratio is high when price growth rate is high. This study indicates that a high P/ E ratio signals faster growth in short-term stock prices. The results also indicate that variance of dividend yields tends to increase relative to the variance of earning yield, with a rapid dividend adjustment and higher dividend payout ratio. The extent to which this model can be applied to different types of stocks must nevertheless await further research. Additional research in this area could extend the results to measure the impact of the differences on pricing or trading customs. References Campbell, J. Y., A Variance of Decomposition for Stock Returns, Economic Journal, 1991, 101, 157 79. Campbell, J. Y. and R. J. Shiller, Valuation Ratios and the Long-Run Stock Market Outlook, Journal of Portfolio Management, 1998, 24:2, 11 26. Cochrane, J., Explaining the Variance of Price-Dividend Ratio, Review of Financial Studies, 1991, 15, 243 80. Eichholtz, P. and D. Hartzell, Property Shares, Appraisals and the Stock Market: An International Perspective, Journal of Real Estate Finance and Economics, 1996, 12, 163 78. Fama, E. F., Term-structure Forecasts of Interest Rates, Inflation, and Real Returns, Journal of Monetary Economics, 1990, 25, 59 76. Fama, E. F. and K. R. French, Permanent and Temporary Components of Stock Prices, Journal of Political Economy, 1988a, 96:2, 246 73.., Dividend Yields and Expected Stock Returns, Journal of Financial Economics, 1988b, 22, 3 25.., French (1998), Value Versus Growth: The International Evidence, Journal of Finance, 53, 1975 99. Ferson, W. E. and C. R. Harvey, Fundamental Determinants of National Equity Market Returns: A Perspective on Conditional Asset Pricing, Journal of Banking and Finance, 1997, 21, 1625 65. He, L. T., F. C. N. Myer and J. R. Webb, The Sensitivity of Bank Stock Returns to Real Estate, Journal of Real Estate Finance and Economics, 1996, 12, 203 20. Lintner, J. M., Distribution of Incomes of Corporations among Dividends, Retained Earnings and Taxes, American Economic Review, 1956, 46, 97 113. Mantrala, S. and E. Zabel, The Housing Market and Real Estate Brokers, Real Estate Economics, 1995, 23:2, 161 85. Myers, S. C., The Capital Structure Puzzle, Journal of Finance, 1984, July, 575 92. Owen, B. M. and J. Grundfest, Kickbacks, Specialization, Price Fixing, and Efficiency in Residential Real Estate Markets, Standard Law Review, 1977, 29, 931 67. Quan, D. C. and J. M. Quigley, Price Formation and the Appraisal Function in Real Estate Markets, Journal of Real Estate Finance and Economics, 1991, 4, 127 46. Quan, D. C. and S. Titman, Do Real Estate Prices and Stock Prices Move Together? An International Analysis, Real Estate Economics, 1999, 27:2, 183 207. Ross, S. A. and R. C. Zisler, Risk and Return in Real Estate, Journal of Real Estate Finance and Economics, 1991, 4, 175 90. Rouwenhorst, K. G., Local Return Factors and Turnover in Emerging Stock Markets, Journal of Finance, 1999, 54, 1439 64. Tse, R. Y. C., Impact of Property Prices on Stock Prices in Hong Kong, Review of Pacific Basic Financial Markets and Policies, 2001, 4:1, 1 15. Tse, R. Y. C. and J. R. Webb, Public Versus Private Real Estate in Hong Kong, Journal of Real Estate Portfolio Management, 2000, 6, 53 60. Zuehlke, T. W., Duration Dependence in the Housing Market, Review of Economics and Statistics, 1987, 69:4, 701 4. Journal of Real Estate Portfolio Management 113