Swfa2015.uno.edu



The Spread between the Bond Yield and the Dividend Yield and the Dividend PremiumAbstractEmpirical tests of the catering theory of dividends find that payout policy is influenced by the dividend premium, the relative market valuations of dividend paying versus non-paying firms. This paper offers the yield spread hypothesis of the dividend premium: investors tend to place lower valuations on dividend paying stocks when their yield is relatively low compared to risk-free bonds. After controlling for other variables that may affect the dividend premium, I find the spread between the one-year US Treasury note and the S&P 500 dividend yield is significantly and inversely associated with the dividend premium. The results are consistent with investors that have regular cash distribution preferences comparing yields on competing financial instruments. Key Words: Dividend premium, Payout policy, Bond yield, Dividend yieldI. IntroductionMiller and Modigliani’s (1961) theoretical paper on dividends led them to their famous proposition that dividend policy does not affect firm value. They arrived at their conclusion by assuming perfect capital markets and a fixed corporate investment plan. These restrictions are not representative of the real environment in which investors and firms operate. Relaxing them to gain insight into the dividend policy choices of managers has been the focus of much subsequent research.One recent theory of dividend policy is the catering model proposed by Baker and Wurgler (2004), from now on “BW”. BW relaxes Miller and Modigliani’s (1961) assumption that stocks are always priced efficiently. They theorize that investor demand for dividend-paying stocks varies through time and can result in the mispricing of firms that pay dividends versus those that do not. The valuation discrepancies can persist because of structural inefficiencies in the financial markets. BW call this valuation differential the “dividend premium”, which is calculated as the natural log of the ratio of average market-to-book values of payers and non-payers. BW hypothesize that managers know when their firms are mispriced along this dimension and make dividend policy decisions accordingly to maximize shareholder value. For example, BW predict that when investors place a relatively high valuation on firms that pay dividends, managers will be more likely to initiate a dividend. The empirical evidence supports the catering theory. BW show that changes in the dividend premium explain 60% of the annual variation in dividend initiations over a forty year sample period. Li and Lie (2006) apply the catering theory to a sample of dividend increases and decreases (as opposed to BW’s initiation and omission sample) and show that variation in the dividend premium is a significant determinant of these more frequent, ongoing payout policy decisions. Bulan et al. (2007) find that the dividend premium is a strong predictor of dividend initiations and Ferris et al. (2006) provide international evidence in favor of the catering theory of dividend policy. Despite these empirical findings, Lee (2011) notes that there has been little written on the factors that influence the dividend premium. This paper seeks to fill this gap in the literature and proposes that the dividend premium is associated with the spread between the yield on the one-year Treasury note and the S&P 500 dividend yield. I assume that there are classes of investors that prefer equity securities that distribute cash at regular intervals. Firms that pay dividends tend do so at regular intervals and rarely reduce them. These two features make dividends a viable substitute to interest payments from debt securities. There have been numerous stories in the financial press discussing this tradeoff between low interest rates and dividend yields during the low interest rate environment of 2008-2014. While the relative risks of bonds and stocks are generally different, the fixed-income market and dividend paying stocks form part of an investment opportunity set for investors seeking securities with cash distributions. I hypothesize that when bond securities offer relatively low yields, investors seek current income from alternative investments, such as dividend-paying stocks. These dividend-paying stocks may receive higher valuations from this increased demand which would be reflected in a higher dividend premium. Alternatively, when the yield on bond securities is high relative to the dividend yield, I predict the dividend premium will be relatively low as investors have an attractive alternative to satisfy their demand for periodic cash proceeds.The evidence is this paper broadly supports the yield spread hypothesis: after controlling for other variables that may affect the dividend premium, such as investor sentiment, age variation in the population, agency costs, tax differentials, business cycle fluctuations, and profitability, the spread variable is significantly and inversely associated with the dividend premium. I also show that the spread between a portfolio of Moody’s Baa-rated corporate bonds and the dividend yield, which is more likely to reflect investor sentiment and the risk premium, is generally not associated with the dividend premium in the main multivariate framework. These results are consistent with investors placing a premium valuation on dividend paying stocks when their yield is comparatively high compared to default risk-free bond yields and vice versa. This paper contributes to the existing literature on dividend policy in several ways. First, I offer a new factor that helps explain the variation in the dividend premium, the force behind the catering theory of dividend policy. This bond yield-dividend yield spread variable is robust to a number of specifications. Despite all of the attention on dividends over the past fifty years, I am not aware of any study that has directly postulated that the competing yields on fixed income products can have an impact on a firm’s dividend policy decisions. The omission of fixed-income yield variables in the extant dividend premium models is somewhat surprising because the tradeoff between debt and equity yields is a reasonable conjecture on what affects the demand for dividend-paying stocks.Second, the theoretical underpinnings of the yield spread hypothesis are more comprehensive than the demographic theory of the dividend premium put forth by Lee (2011). He makes a convincing argument that changes in the age structure of the population drives movements in the dividend premium. However, an age-based explanation cannot account for why investors would prefer dividend-paying equities when risk-free securities offer high yields, nor does it account for the risk-averse preferences of investors that do not fit a particular age profile. My empirical analysis shows that the significance of the age structure variable is subsumed by the spread variable. Third, my empirical tests are somewhat more powerful than previous studies on the dividend premium. For example, annual data on the age structure of the population are estimates based on census data which is officially measured only once every decade. On the other hand, spreads between bond and dividend yields can be measured more precisely and monthly, providing a sample size that is ten-times as large as prior studies on the dividend premium. II. Literature ReviewWhy dividends?Why might investors demand dividends in the first place? Miller and Modigliani (1961) argue that dividend policy is irrelevant to firm valuation because investors can create homemade dividends by selling shares to generate income. In a world with taxes, such a strategy would be efficient if the tax rate on capital gains is less than the tax rate on dividends, as is often the case. However, Black (1976) notes the preference for a cash dividend by some investors is undeniable. He briefly reviews many mainstream reasons why dividends may be preferred (or not preferred) and debunks them all leaving us with his famous “dividend puzzle”. Potential explanations for why dividends exist largely relate to market inefficiencies, tax policy, information asymmetry, and agency costs. Behavioral theorists also offer alternative explanations of the preference of dividend payments even in the absence of market frictions. For example, Shefrin and Statman (1984) use Thaler and Shefrin’s (1981) self-control framework and argue that investors with difficulties controlling their will power may prefer dividends to selling shares themselves in order to avoid the temptation of reducing the principal invested too quickly. Shefrin and Statman (1984) also argue that investors who are averse to regret will prefer to finance current consumption out of dividends lest they miss out on the potential future capital appreciation of shares they sold. Regardless of the source of investor demand for dividends, I assume that there are at least some investors that prefer stocks that pay dividends. This assumption is broadly consistent with the empirical evidence on dividends.The Dividend PremiumBW attributes the link between the dividend premium and dividend decisions to managers rationally catering to their shareholders by exploiting market mispricing. They arrived at this conclusion in several steps. First, they systematically eliminate explanations related to time-variation in firm characteristics, such as growth opportunities, and agency costs because these reasons were inconsistent with their full set of results. BW then asks “Who are managers catering to?” BW show that managers do not appear to be catering to clienteles based on tax concerns, transactions costs, and institutional investment constraints. This corresponds with survey evidence from Brav et al. (2005) that suggests that managers do not consider the demands of clienteles of any sort. On the other hand, BW provides evidence that managers cater to investor sentiment: the dividend premium is positively correlated with the closed-end fund discount. A large closed-end fund discount is associated with poor investor sentiment. When the discount is large, investors may drive up the relative valuation of dividend payers versus non-payers if they perceive the former as safer investments. While investor sentiment appears to play a pivotal role, BW admit that unresolved issues with what the closed-end fund discount truly measures detracts from its theoretical link with the dividend premium. Other studies that have attempted to identify variables that affect the dividend premium have yielded mixed results. Liu and Shan (2007) tied changes in the dividend premium to proxies for agency costs and signaling motives, but Lee (2011) points out their results were not robust when a simple time trend was added to the model. In the most promising recent research, Lee (2011) finds that the dividend premium is positively correlated with demographic variation as measured by annual changes in the proportion of persons aged above 65 to those aged under 45. His result is consistent with other research pertaining to dividend policy and demographics. For example, Becker et al. (2011) show that firms headquartered in areas in which seniors make up a large percentage of the population are more likely to initiate a dividend. Additionally, Graham and Kumar (2006) investigated retail investor stock trading behavior and found that older investors tend to buy stocks after dividend announcements and just before the ex-dividend date.III. Hypothesis DevelopmentMotivationThe demographic theory of the dividend premium put forth by Lee (2011) can be summarized as follows. Older investors tend to concentrate their portfolios in income generating investments when they turn 65 years old. The lack of labor income in this demographic could be an impetus that drives these investors to equity securities that offer income. Time variation in the older-to-younger ratio can impact the relative valuation difference between dividend payers and non-payers. Specifically, as the ratio of older-to-younger persons in the economy increases compared to the prior year, dividend payers receive a premium valuation compared to non-payers, and vice-versa. While the demographic explanation is very intuitive, I argue that it does not go far enough in several respects. First, there is no consideration of the full investment opportunity set facing investors. For example, if yields on debt securities were relatively high, there is little reason to suspect that yield-seeking, older investors would shift their portfolios to dividend-paying stocks en masse just because they are older. Additionally, the risk of receiving the interest payments and principal from bonds is lower than that of receiving dividends. It would not be surprising to find that the demographic variable is picking up the collective impact of competing yields in the bond and equity markets. Second, there is little consideration paid to the risk-return preferences of the entire population of investors in the existing demographic theory. For example, younger investors that are relatively risk-averse may have a large percentage of their portfolio concentrated in income generating investments. Lease et al. (1976) find that approximately one-third of the portfolios of highly educated young professionals are concentrated in income securities. In other words, older investors are not the only clientele that invest in securities that distribute cash. Finally, changes in the old-to-young ratio assume that portfolios are not adjusted until an age reference point is hit. It is certainly possible that investors begin to adjust their portfolios before their actual age reaches a certain number in anticipation of cash needs. The way Lee (2011) calculates the old-young ratio would not capture this dynamic.Main HypothesisThe yield spread conjecture put forth in this paper incorporates the investment opportunity set facing investors as well as the full set of market participants available to purchase securities. The level of interest rates in the economy affects every investor to a certain degree as yields on outstanding and new securities compete with other types of securities for loanable funds. This is why the yield spread hypothesis of the dividend premium is more comprehensive than the demographic theory. Dividends offer an easy way for companies to cater to investors seeking yield in times of low interest rates, especially if the firm had been on the fence about initiating a dividend. I assume that there are classes of investors that prefer dividend paying stocks, often called dividend clienteles, because they offer regular cash payments. I conjecture that these investors also consider the yields available on other types of cash distributing securities when they make their portfolio choices. A relatively low spread between fixed-income yields and the S&P 500 dividend yield may result in higher valuation on dividend paying stocks (i.e., a higher dividend premium) via a shift outward in the demand curve. On the other hand, when the bond rate is comparatively high, dividend-paying stocks lose their relative appeal since investors can obtain acceptable streams of cash from interest-bearing securities. This leads to the following hypothesis:H1: The dividend premium is inversely related to the spread between the one-year Treasury note yields and the S&P 500 dividend yield. The higher the one-year Treasury yield compared to the dividend yield, the lower the dividend premium. The lower the one-year Treasury yield relative to the dividend yield, the higher the dividend premium.I choose the one-year Treasury note yield in the dividend spread calculation to limit the confounding factors that could impact the spread. For example, using a corporate bond yield in the calculation of the spread above the dividend yield may reflect changes in investor sentiment or other factors as opposed to the cash distribution comparisons I am trying to model.IV. Method and SampleTo test the yield spread hypothesis of the dividend premium I run OLS regressions using annual times series data from 1962 to 2004. I include the same or very similar control variables found in other models of the dividend premium to test the yield spread hypothesis in a multivariate framework. The main model is (time subscripts suppressed):DP = α + β1YIELD_SPREAD + β2ΔDEMOGRAPHIC + β3CEFD + β4 GDP_GROWTH + β5 TAX_RATIO + β6 PROFIT_PREM + β7 CASH_PREM + β8 TIME + ε (1)The variables, the data sources, and the expected signs on the coefficients (when particularly relevant) are below.A. Dependent and Main Independent VariableDIV_PREM = the dependent variable in all models and is defined as the difference between the natural logs of the value-weighted market-to-book ratios of dividend payers versus nonpayers. This variable was created by BW and can be obtained from Jeffrey Wurgler’s website. RF_DIV_SPRD = the main independent variable of focus and is defined as the difference between the one-year US Treasury note yield and the dividend yield of the S&P 500 index. It is a proxy for the trade- off investors face between the risk-free yield and the dividend yield. The data on the Treasury note is obtained from the Federal Reserve Economic Data (FRED) website and the S&P 500 dividend yield is cross checked from various publicly available sources. A negative slope coefficient is anticipated: as the spread between the one-year Treasury note yield and the S&P 500 dividend yield increases, the dividend premium tends to decrease.B. Control VariablesΔDEMOGRAPHIC = the change in the older-to-younger ratio defined by Lee (2011) as the proportion of the population above 65 to those aged under 45. It is a proxy for changes in the age structure of the population. The data comes from the US Census website. The anticipated sign is positive. A marginal increase in the proportion of the population above 65 is associated with an increase in the dividend premium. CEFD = the closed-end fund discount which measures the value-weighted difference between the price of closed-end mutual funds and their net asset value. It is a proxy for investor sentiment. The data come from Jeffrey Wurgler’s website. As in empirical tests of BW and Lee (2011), I expect a positive coefficient.GDP_GROWTH = the annual real GDP growth. Lee (2011) includes this variable in his models to control for business cycle fluctuations. The data are collected from the FRED website.TAX_RATIO = the ratio of the Federal and State marginal tax rate on qualified dividends divided by the marginal capital gains tax rate based on a large sample of tax returns. It is a proxy for the actual tax considerations (as opposed to the statutory rates) facing investors when choosing between dividend-paying versus non-paying stocks. The data are generated from the TAXSIM program on the National Bureau of Economic Research (NBER) website. This control is similar to Lee (2011) but he uses statutory rates from the Citizens for Tax Justice website.PROFIT_PREM = is the profitability premium defined by Liu and Shan (2007) as the natural log of the ratio of the value-weighted future return-on-assets ratios for dividend payers versus nonpayers. It is a proxy for the signaling feature of dividends. The data are taken from Liu and Shan (2007) and the expected sign of the estimated coefficient is positive.CASH_PREM = is the cash premium defined by Liu and Shan (2007) as the natural log of the ratio of the value-weighted cash-to-asset ratio for dividend payers versus nonpayers. They use this variable as a proxy for agency costs. The data are taken from Liu and Shan (2007) and the expected coefficient is positive.TIME = is a time trend variable included by Lee (2011) to control for the nonstationarity in the dividend premium series. It is particularly important because Lee (2011) shows that the time trend subsumes the explanatory power of the cash premium variable in Liu and Shan (2007).C. Variables Included in Robustness ChecksΔALT_DEMOGRAPHIC = an alternative demographic structure variable defined by Lee (2011) as the change in the proportion of the population above 65 to those aged between 45 and 65.BW_SENT = an alternative measure of investor sentiment calculated by BW as the principal component of three standardized proxies: the closed-end fund discount, the equity share of all new issues, and the average of monthly New York Stock Exchange share turnover during the year. This variable is orthoganilized to macroeconomic variables to try and get a non-confounded measure of sentiment.Table 1 contains the summary statistics for the variables used in the OLS regressions. Both Baker and Wurgler (2004) and Lee (2011) provide motivation for why OLS regressions with a time trend variable are appropriate in modeling the dividend premium. Notice that we include summary statistics for monthly data when available which will be used in secondary regressions that provide support of the main conclusion of this paper.V. ResultsTable 2 contains the results from OLS regressions using a restricted form of the formal model laid out in equation 1. These models investigate the relationship between the dividend premium, the yield spread variables proposed in this paper, Lee’s (2011) demographic change variables, and a linear time trend variable in various combinations. The results from model 1 show that the dividend premium is inversely related to the spread between the one-year Treasury note and the S&P 500 dividend yield at the 1% significance level in a simple OLS framework. Model 2 shows that the significance of the spread variable holds when we add a time trend variable to control for impact of the nonstationarity in the dividend premium variable. Models 3 and 4 add Lee’s (2011) demographic change variables to the model. The spread variable remains significant in both models while the proxies for demographic change are both insignificant. I interpret the results as evidence that the risk-free yield spread variable is more comprehensive in modeling the dividend premium compared to the demographic structural change variable because it captures the investment opportunity set and the risk preferences of all investors.Models 5 through 8 repeat the analysis of models 1-4 but use Moody’s Baa-rated corporate bond yield as the spread variable. The results are qualitatively similar to the results in Models 1 through 4. In this reduced framework, it appears that the spread between bond yields and the dividend premium have the power to explain variation in the dividend premium. The results are consistent with the yield spread hypothesis: the relative valuation of dividend payers to non-payers is inversely associated with the difference between cash-distributing bonds and cash distributing equity. Table 3 contains the results of our main analysis. Model 1 and Model 2 show that the risk-free yield spread is inversely related to the dividend premium at the 1% significant level when controlling for investor sentiment and the linear time trend. Model 3 and 4 show that this results hold when use the corporate bond yield-dividend yield spread as the variable. Models 5 through 8 use the full multivariate framework of equation 1 with alternative measures in Lee’s (2011) proxy for demographic change and an alternative measure of investor sentiment. The results show strong support for the yield spread hypothesis of the dividend premium. The dividend premium is strongly and inversely associated with changes in the yields available to investors.Models 9 through 12 replicate the previous four models but use the corporate bond spread as the main variable of interest in the yield spread hypothesis. The results are weaker or insignificant in this case. I attribute this to the fact that the spread between the corporate bond yield and the dividend yield may be picking up some of the explanatory power the sentiment and economic variables picked up. For example, in times of poor economic sentiment, the spread on the portfolio of BAA-bonds may rise faster than the dividend yield. While a similar argument could be made for the risk-free yield spread variable, the direction of the change would appear to bias against finding my result. For example, when sentiment is low, it is likely that the yield on the safe risk free asset is relatively low and the dividend yield is relatively high as equity share prices (the denominator in the dividend yield calculation) have come down. This would bias against my finding of an inverse relationship which is not the case based on my empirical results.Table 4 contains the results from monthly regressions of the dividend premium on the variables where the data is available. Notice that sentiment spread and time seem to play a large role in the annual data and this persists on the monthly level. The significance levels on the yield spread variables are very high. While access to monthly data on the full set of control variables would diminish the significance levels, it is highly unlikely to full suppress the explanatory power given the earlier results. Thus, I take this as preliminary evidence that the results hold on a short-term basis as well.VI. Concluding RemarksI find that time variation in the relative valuation between stocks that pay dividends and stocks that do not, the “dividend premium” as defined by Baker and Wurgler (2004), is related to the investment opportunity set facing market participants. Specifically, when the spread between the one-year Treasury note and the S&P 500 dividend yield is relative small, the dividend premium tends to be high, and vice versa. In other words, investors appear to place higher relative valuations on dividend paying stocks when they cannot find attractive yields in the default risk-free fixed income market. The results are consistent with anecdotal evidence from the current low interest environment where numerous financial reporters have discussed the relative appeal of dividend-paying stocks based on their yields. The statistical significance of the risk-free yield spread variable in explaining variation in the dividend premium holds after controlling for other variables that may affect the dividend premium, such as investor sentiment, age variation in the population, agency costs, tax differentials, business cycle fluctuations, and profitability. I also show that the spread between a portfolio of Moody’s Baa-rated corporate bonds and the dividend yield, which is more likely to reflect investor sentiment and the risk premium, is weakly or not associated with the dividend premium in the main multivariate framework. These results are consistent with investors placing a premium valuation on dividend paying stocks when their yield is comparatively high compared to default risk-free bond yields and vice versa.References Al-Malkawi, Husam-Aldin, Micheal Rafferty, and Rekha Pillai. 2010. “Dividend Policy : A Review of Theories and Empirical Evidence.” International Bulletin of Business Administration 9(9):171–200Asquith, Paul and David W. Mullins Jr. 1983. "The Impact of Initiating Dividend Payments on Shareholders' Wealth." The Journal of Business 56 (1): 77-96. Baker, Malcolm and Jeffrey Wurgler. 2004. "A Catering Theory of Dividends." The Journal of Finance 59 (3): 1125-1165. Becker, Bo, Zoran Ivkovic, and Scott Weisbenner. 2011. "Local Dividend Clienteles." The Journal of Finance 66 (2): 655-683. Black, Fischer. 1976. “The Dividend Puzzle.” The Journal of Portfolio Management 2(2):5–8.Brav, Alon and J. B. Heaton. 2004. “Did ERISA’s Prudent Man Rule Change the Price of Dividend Omitting Firms?” Working paper, Duke University.Brav, Alon, John R. Graham, Campbell R. Harvey, and Roni Michaely. 2005. "Payout Policy in the 21st Century." Journal of Financial Economics 77 (3): 483-527. Bulan, Laarni, Narayanan Subramanian, and Lloyd Tanlu. 2007. "On the Timing of Dividend Initiations." Financial Management 36 (4): 31-65. DeAngelo, Harry, Linda DeAngelo, and Douglas J. Skinner. 2004. "Are Dividends Disappearing? Dividend Concentration and the Consolidation of Earnings." Journal of Financial Economics 72 (3): 425-456. Fama, Eugene F. and Kenneth R. French. 2001. "Disappearing Dividends: Changing Firm Characteristics Or Lower Propensity to Pay?" Journal of Financial Economics 60 (1): 3-43. Graham, John R. and Alok Kumar. 2006. "Do Dividend Clienteles Exist? Evidence on Dividend Preferences of Retail Investors." The Journal of Finance 61 (3): 1305-1336. Grinstein, Yaniv and Roni Michaely. 2005. "Institutional Holdings and Payout Policy." The Journal of Finance 60 (3): 1389-1426. Jensen, Michael C. 1986. "Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers." The American Economic Review 76 (2, Papers and Proceedings of the Ninety-Eighth Annual Meeting of the American Economic Association): 323-329. Lease, Ronald C., Wilbur G. Lewellen, and Gary G. Schlarbaum. 1976. "Market Segmentation: Evidence on the Individual Investor." Financial Analysts Journal 32 (5): 53-60. Lee, King Fuei. 2013. "Demographics and the Long-Horizon Returns of Dividend-Yield Strategies." The Quarterly Review of Economics and Finance 53 (2): 202-218. ———. 2011. "Demographics, Dividend Clienteles and the Dividend Premium." The Quarterly Review of Economics and Finance 51 (4): 368-375. Li, Wei and Erik Lie. 2006. "Dividend Changes and Catering Incentives." Journal of Financial Economics 80 (2): 293-308. Liu, Laura. and Liwei Shan. 2007. “What is Dividend Premium?” Financial Management Association 15 Annual Meeting 2007. Online conference proceedings.Miller, Merton H. and Franco Modigliani. 1961. "Dividend Policy, Growth, and the Valuation of Shares." The Journal of Business 34 (4): 411-433. Miller, Edward M. 1977. "Risk, Uncertainty, and Divergence of Opinion." The Journal of Finance 32 (4): 1151-1168. Shefrin, Hersh M. and Meir Statman. 1984. "Explaining Investor Preference for Cash Dividends." Journal of Financial Economics 13 (2): 253-282. Shefrin, Hersh M. and Richard H. Thaler. 1988. "The Behavioral Life-Cycle Hypothesis." Economic Inquiry 26 (4): 609. Shleifer, Andrei. 1986. "Do Demand Curves for Stocks Slope Down?" The Journal of Finance 41 (3, Papers and Proceedings of the Forty-Fourth Annual Meeting of the America Finance Association, New York, New York, December 28-30, 1985): 579-590. Stephen?P.?Ferris, Nilanjan?Sen, and Ho?Pei?Yui. 2006. "God Save the Queen and Her Dividends: Corporate Payouts in the United Kingdom." The Journal of Business 79 (3): 1149-1173. Thaler, Richard H. and H. M. Shefrin. 1981. "An Economic Theory of Self-Control." Journal of Political Economy 89 (2): 392-406. Zweig, Martin E. 1973. "An Investor Expectations Stock Price Predictive Model using Closed-End Fund Premiums." The Journal of Finance 28 (1): 67-78. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download