The Predictability of Dividend Yield over Stock Return - Literature review Example

The predictability of dividend yield over stock return Table of Contents Literature Review 3 Historical studies 3 Long term and short term 5 Factor affecting the predictability of dividend yield through stock returns 7 Small sample bias 7 Time horizon/holding period 8 Less dividend involving higher amounts of share repurchase 9 Time series 9 Cross-sectional studies 10 Reference List 12 Literature Review Huge amount of literature is available in regards to the empirical studies that have been conducted to assess the predictive power of dividend yields so as to suggest stock returns. The behaviour of time series data on dividend yield was observed by several researchers in due course of the last 25 years. These extensive researches have supported the idea that dividend yields can be used as a prime ratio for measuring the expectation of stock returns. Rozeff (1984) had provided evidences for the fact that equity risk premium is predicted by use of the dividend yield measure. Gordon Shapiro’s constant growth model uses the spreads in bond classes and he employed this model for providing suggestions to the idea that the dividend yields on a specific stock can be used for the approximation of equity risk premium, which should be quite helpful. The model also suggested that the expected return rate in the stoic market is simply equal to the dividend yield of the variable on the markets with a predicted rate of growth for the dividends. The particular case of using dividend yield as a construct for estimating stock returns might be one of the most common predictor of returns, which has been examined in literature and also more because the econometric problems associated with dividend yields can be generalised for other kinds of commonly used predictor variables. The predictability regressions inferences are to be based on the dividend yields, where the econometric problems caused due to data generation issue are highly persistent. Historical studies Prominent claims about predictability of the stock market returns have been made since 1965, which was initiated in the studies of Eugene Fama (1965). The efficient Market Hypothesis postulated by Fama stated that stock markets were efficient and it was possible to completely reflect the prices of stock market in future with the publicly available information. This statement was well-accepted within the financial community and the model was subsequently used to approach various studies in the market. There have been numerous studies in the past, which have been conducted to test predictability of the dividend yield for returns over the stock market in future. The traditional mode of the dividend price ratio has been used as a measure for estimating the expected return on common stock. In a research conducted by Rozeff (1984), it was found that when dividend yields was expressed as a ratio of short-term interest rates, the results had significant predictability for the annual returns of common stock. According to the studies conducted by Fama and French (1988), the regression framework developed within the paper proved that dividend yields have a significant probability to predict a sizable portion of stock returns for multiple years when measured on the NYSE index. It was further observed within the study that the dividend yield rises with the time horizon of returns when the four year data was analysed. R squared range varied from a low of 19% to a high of 64%. The results of the Fama and French model were also strengthened by the studies of Campbell and Shiller (1988) and Flood, Hodrick and Kaplan (1987). In one of their studies, Campbell and Shiller found that use of valuation ratios for determination of stock market returns in the long run was highly effective in the US markets. They considered valuation ratios to be a better measure because they could then value the extreme fluctuation in stock market prices within the ratios as well as facilitate comparisons across times. Campbell and Shiller (1988) assumed that the distribution of such valuation ratios was stable within their study and hence, were in conformity with the mean reversal theory. This notion implies that when prices were high, they were expected to eventually fall because the ratios would have to come back to their means at historical levels. Thus, the authors also established that the US markets were overvalued during the time period of their study and it was quite certain that they had a gloomy outlook for the future. In another study by Claessens, Dasgupta and Glen (1998), data pertaining to many different risk factors associated with the stock market returns was collected for about 19 countries. It was found that the trading volume and size of the stock had a notable impact on predicting stock market returns. The results had also provided confirmation to the idea that the dividend yield variable plays a critical role in determining explanation to stock returns in about 7 out of the total 19 nations that were selected for research. Kiem (1985) also examined the relation of stock returns over the dividend yields in his empirical tests and found a non-linear relation existing between the two variables. Some models might predict a non-zero for dividend yields in most studies because of the consideration for impact of differential taxation with reference to capital gains and dividends. However, the findings of Kiem was restricted to January results, which is unique for its after tax impacts on asset pricing and it was derived that that after tax differentials were more important in January that in any other month. Long term and short term Hodrick (1992) assessed the long run predictability of stock market returns based on dividend yields. The Monte Carlo analysis has shown that Hansen and Hodrick (1980) have used a procedure that is biased towards the long run predictability of stock returns. The statistical properties of three basic methodologies were estimated using the Monte Carlo experiments by Hodrick in his study. Such methods were applied in order to conclude the long run patterns of forecasting stock prices through dividend yields. He made use of the OLS or the ordinary least squares measure, which was also used by Fama and French (1998), while the prime method used was the Vector Autocorrelation or VAR as done by Cambell and Shiller (1988), Cambell (1991) and Kandel and Stambaugh (1988). In his study, Cambell (1991) suggested that unexpected returns on stocks might be an occurrence due to expected future dividend changes. He conducted a vector autoregressive examination, where the unexpected stock returns were attributed towards the variance of changes in the expectations of dividends, along with a third of variance of changes in expected stock returns and that of the covariance between the two variables. It was observed that altering expectations on returns have a huge affect on the prices of stocks mainly because such changes are predictable. Here, the changes in expected stock returns were found to be negatively correlated with any change in expected dividends. This finding would contribute towards increasing the reaction of stock market towards occurrences in the markets. Through establishment of a link between the short run and long run predictability of returns, he showed that the consistency existed largely with the long run returns in the forecasting experiments, while results reflected small degree of consistency when the short run data was analysed. In the study, for example, the data between 1952 and 1987 predicted that there was a strong support given by the VAR measure so as to predict the month long returns of stock prices. The significant amounts of stock return changes have been forecasted by the changes in dividend yields and such results of the VAR experiment was also supported by estimates and conclusions derived by the Monte Carlo experiment. In another study conducted by Chan, Hamao and Lakonishok (1991), a cross-sectional study was done to estimate returns predictability of stocks based on the behaviour of four variables, namely dividends yields, size, cash flow and book to market ratio. It was found that the entire variable shared a significant relationship with expected returns and the market of Japan in general. The relationship was stronger in case of book to market ratio and cash flows with the expected returns than in case of other variables. Factor affecting the predictability of dividend yield through stock returns Small sample bias Nelson and Kim (1993) have identified a small sample bias existing in the use of VAR systems for predicting returns based on dividend yields, that is used by the hypothesis of zero returns predictability. The authors explained that there could be a possibility where the biases occurring from small samples shall have an important role to play in judgments, which are directed towards inferring predictability of stock returns with the use of financial fundamentals such as, dividend yields, in this case. The t values used within the researches directed towards establishment of a predictive relationship between dividend yields and stock market returns was analyzed in the study of Nelson and Kim. Consequently, it was observed that t value could be affected by two kinds of sample biases, typical for small samples. When the predictor becomes endogenous, the regression coefficient is biased. When the periods are overlapping, the standard error gets biased. In both the cases, the biases have worked in a similar direction, thereby indicating that the returns are more predictable in statistical estimations, as opposed to that in reality. The annually sampled return was considered for the US market and accounted for upward displacements of a simulated distribution set for the t statistics measure. Despite this fact, there were traces of predictability with standardized levels of significance. In another study by Banz (1981), the empirical relationship between the market value and the returns of common stocks from the NYSE were estimated. It was found that the small firms have a higher capacity to adapt to risk adjusted returns compared to that of the larger ones when checked on an average. Hence, the affect of size is also said to exist when a sample of 40 years is taken with the evidence that the CAPM model has been mis-specified. It was also observed that such size effects are in a non-linear relationship with its primary affect observed in the smallest firms. The affect variations were difficult to observe in average sized or large firms. However, it was challenging to determine whether such a size affect had any impact on the dividend payouts and the expected stock returns. Time horizon/holding period Fama and French had found that dividend yields have the ability to predict stock returns. Dividend yield is also used to project returns in an equal and value weighted variants of portfolios that consist of NYSE stocks and those that have a holding period. Here, holding period refers to the time horizon taken for yielding returns as dividends and varies from 4 months to 4 years. Considering the use of overlapping returns spread over multiple years, there was very strong evidence indicating that dividend yield did have an impact on predicting stock returns. Evidences also suggested that the predictive power increased with rise in the horizons for returns. Contrarily, the short run regressions of a month or a quarter were not too strong in predicting the power for stock returns. They were capable of predicting not more than 5% of the returns. Studies conducted over a period of 20 years have reflected that there exists some amount of econometric difficulty in tests, which involve the holding period return or the horizon return. This also resulted in a rejection towards the biasness towards the basic null hypothesis. Stambaugh (1986) had indicated another deficiency in the independent variable or the explanatory variable, which is the dividend yields contained and reflects the prices which have already appeared and been regressed in historical data; hence, cannot be regarded as a proper externally originating variable. Such variable error problems have been pointed out by Fama and French (1988), who have also stated that some information about predictability of the future returns and of the dividend growth, that shall be seen as a predictor variable, has already been seen in the yield. Consequently, these results might reflect some bias towards the coefficient of regression in the regression model of dividend yield (Schwert, 1989). Less dividend involving higher amounts of share repurchase The impact of dividends on stock returns is a critical issue pertaining to finance. It has been argued that lower amount of yields have to be associated with higher returns primarily because the investors have a perception about the values of dividends, which shall be higher than retained earnings. Modigliani and Miller (1962) had demonstrated that in the absence of any transaction costs, equality in tax treatment for dividends and capital gains shall have little or no impact on welfare accrued to holders of the security. In this case, it is assumed that investors act in rationality. Early experiments done in order to assess the impact of dividends, as that of Friend and Puckett (1964), have relied upon cross-sectional regressions for the prices of common stock and its impact on the dividends paid per unit of stock as well as several other measures that estimate the retained earnings per unit of share. Time series Fama French model as developed within their study conducted in 1988 was revisited by Cochran (1992) for the purpose of making an extension to the previous results and was able to derive two vital results that in most critical of the stock markets across the world, dividend yields have the power to predict stock market returns and the lengths of such return horizons could vary between 1 month to 48 months. Such predictability power also rises with presence of a positive correlation between the length of return horizon and the predictability factor (Cochran, 2008). Kothari and Shanken (1997) evaluated the B/M ratio or the Book to Market Ratio for tracking the variation in time series of the expected returns of the market index and such was compared to the ability of forecasting the B/M ratio with that of the dividend yields ratio. With a view to examine the statistical significance of evidences predicted by the regression analysis showing the predictive capacities of stock returns, the study also employed the VAR framework. The null hypothesis stated no predictability and was aimed at testing the bootstrap procedure for simulation. The statistical significance levels as well as the degree of the relationship between B/M and the expected returns in association with the dividend yield were also evaluated using the bootstrap simulation procedure. The conventional bootstrap procedure was further extended in order to develop the Bayesian simulation of bootstrap and the same was utilized to determine slope values for estimating the historical least squares method or the OLS. Such predictive capacities of the dividend yields are also forecasted by the use of univariate version of the least squares method. The results obtained from the OLS as well as the developed method of bootstrap provided with the confirmation that dividend yields have the power to predict the values as well as returns, which are equally weighted (Campbell and Yogo, 2002). Black and Scholes (1974) have tested the general affect of dividends on the returns of common stock with the use of time series methodology. They had found that the method could possibly avoid a number of difficulties in estimations while measuring stock returns. The results were similar to those arrived at while estimating through cross-sectional studies (Hess, 1981). Cross-sectional studies Many researchers have also provided evidences supporting the idea that dividend yields have certain levels of prediction powers, where the returns ratio is associated with the component or underlying asset. Such a predictive power of the dividend yield was re-examined by Ang and Bekaert (2001), where they attempted to forecast cash flows, excess returns and rates on interest. The findings indicated that end result of the research conducted worked on to separate data sets, where the short data set undertook a sample of four countries, namely the UK, the USA, France and Germany; whereas the long data included Germany, the UK and the USA. Past earnings data and dividends data were considered for the construction of earnings yield and dividend yield. In this case, the seasonal component of quarterly and monthly figures was not taken into account. The growth estimated for earnings as well as dividends was determined from the ratios constructed, while the annual dividend rate or the growth of the company earnings over a specified time period, like, a quarter or a month, was developed. The results obtained from such analysis indicated that the dividend yield factor had the power to predict the excess returns on stocks over a short period of time, but the predictability was also possible over a short rate. The four countries that were examined in the research did not help in reflection of any long-term predictability of results in the scenarios of excess returns experienced in these nations. The power of predictability of the dividend yield was, thus, not too strong in regards to forecasting future growth of dividends among the countries that were studied and also within the time period analysed. As a result, the model of present value as analysed by Ann and Bekaert was successful in showing that there was not a huge role played by the rate of discount and the short rate for providing the explanation in support of the variations that occur in dividend yields. In the study conducted by Lewellen (2002), the focus was primarily laid upon short horizon test of the cross-sectional study for estimating regression results. The same was chosen for the purpose of avoiding any complication that might arise in the event of overlap of returns, when the monthly figures of returns were regressed over the lagged figures of dividend yields. The findings showed significant and strong predictive results related to the power of predictability for a data set ranging over 1946-2000. Reference List Ang, A. and Bekaert, G., 2001. Stock Return Predictability: Is It There? Manuscript. Columbias University. Banz, R. W., 1981. The relationship between returns and market value of common stocks. Journal of Financial Economics, 9(1), pp. 3-18. Black, F. and Scholes, M., 1974. The Effect of Dividend Yield and Dividend Policy on Common Stock Price and Returns. Journal of Financial Economics. 1, pp. 1-22. Cambell, J. Y., 1991. A Variance Decomposition for Stock Returns. NBER Working Paper. 3246. January. Cambell, J. Y., and Shiller, R. J. 1988. Stock Prices, Earnings, and Expected Dividends. The Journal of Finance, 43(3), 661-676. Campbell, J. Y., and Yogo, M., 2002. Efficient Tests of Stock Return Predictability. Manuscript, Harvard University. Chan, L. K. C., Hamao, Y. and Lakonishok, J., 1991. Fundamentals and Stock Returns in Japan. The Journal of Finance, 46(5), 1739-1764. Claessens, S., Dasgupta, S. and Glen, J., 1995. Return behavior in emerging markets. World Bank Economic Review, 9, pp. 131–151 Cochran, J. H., 1992. Explain the Varianceof Price Dividends Ratios. Review of Financial Studies, 5. Pp. 243-280. Cochran, J. H. 2008. The Dog That Did Not Bark: A Defense of Return Predictability. Review of Financial Studies, 21(4), 1533-1575. Fama, E. F., & French, K. R. 1988. Dividend yields and expected stock returns. Journal of Financial Economics, 22(1), 3-25. Fama, E., 1965. Portfolio analysis in a stable Paretian market. Management Science, 11, pp. 404–419. Flood, R. P., Hodrick, R. J. and Kaplan, P. 1987. An evaluation of recent evidence on stock market bubbles. National Bureau of Economic Research, Cambridge, Unpublished manuscript. Friend, I. and Puckett, M., 1964. Dividends and Stock Prices. American Economic Review, 54, pp. 656-682. Hansen, L. P. and Hodrick, R. J., 1980. Forward exchange rates as optimal predictors of future spot rates: An econometric analysis. Journal of Political Economy, 88, pp. 829–853. Hess, P. J., 1981. Dividend Yields and Stock Returns. NBER Working Paper, 649. December. Hodrick, R. J., 1992. Dividend Yields and Expected Stock Returns: Alternative Procedures for Inference and Measurement. The Review of Financial Studies, 5(3), pp. 357-386. Kandel, S. and Stambaugh, R., 1988. Modelling Expected Stock Returns for Long and Short Horizons, Working Paper, University of Chicago and University of Pennsylvania. Kiem, D. B., 1985. Dividend yields and stock returns: Implications of abnormal January returns. Journal of Financial Economics, 14(3), pp. 473-489. Kothari, S. P. and Shanken, J., 1997. Book-to-market, dividend yield, and expected market returns: A time-series analysis. Journal of Financial Economics, 44(2), pp. 169-203. Lewellen, J., 2002, “Predicting Returns with Financial Ratios,” Forth-coming in the Journal of Financial Economics, 15(2), pp. 533-564. Nelson, C. R. and Kim, M. J., 1993. Predictable Stock Returns: The Role of Small Sample Bias. The Journal of Finance, 48(2), 641-661. Rozeff, M. S., 1984. Dividend yields are equity risk premiums. Journal of Portfolio Management, 10, pp. 68-75. Schwert, G. W., 1989. Why Does Stock Market Volatility Change Over Time? Journal of Finance, 44, pp. 1115-1153. Stambaugh, R. F., 1986. Bias Regressions with Lagged Stochastic Regressors. CRSP Working Paper 156, University of Chicago. Read More