High Dividend Yield Does Predict Lower Dividend Growth: A Natural Experiment

Abstract

Today, many financial economists believe that the time variability in expected returns is the dominant component of stock price variability. This modern-day view of stock prices is, nevertheless, dissatisfactory. According to Cochrane (2001, p. 398), is nonetheless an uncomfortable fact that almost all variation in price/dividend ratios is due to variation in expected excess returns. How nice it would be if high prices reflected expectations of higher future Very recently, dividend predictability has been critically re-examined by Ang and Bekaert (2007), Ang and Liu (2007), Bansal and Yaron (2004), Lettau and Ludvigson (2005), and Menzly, Santos, and Veronesi (2004). These efforts have so far shown that dividend growth is predictable, but have stopped short of presenting evidence of a desirably negative relation between dividend yields and expected dividend growth. In fact, Ang and Bekaert (2007) found the dividend yield coefficient in dividend growth predictive regression over the 1952-2001 period to be surprisingly positive. Why is it that the apparently intuitive relation between high prices relative to dividends today and high dividend growth in the future cannot be successfully uncovered in stock market data? This study argues that a potential reason for this is that dividends are usually poor measures of true value-relevant cashflows. It is well known that dividends are often smoothed, manipulated, censored at zero, related to perceived mispricing, and subject to structural shifts in corporate financial policy. Based on this observation, this study contributes to the current surge of interest in re-examining the predictability of dividend growth from dividend yields via a different empirical strategy. Specifically, we focus our study on a unique subset of U.S. stocks-equity real estate investment trust (REIT) stocks, whose dividends are arguably good measures of true value-relevant cashflows. REITs typically adopt a stable dividend policy and pay out almost all of their cashflows from operations as dividends. During the 1980-2000 period, REITs paid out an average 110% of taxable income. The study finds that, when dividends reasonably capture true value-relevant cashflows, expected dividend growth is forecastable from dividend yields. Furthermore, the empirical relation between dividend yields and expected dividend growth is desirably negative, which is consistent with the usual prediction of the present value model. At the three-year predictive horizon, the bootstrapped adjusted dividend yield coefficient is -1.09 with a bootstrapped t-statistic of -14.14. The OLS R2 value is 35.48%.