Empirical analysis of risks and returns of short-term dividend strips

Abstract

In this thesis, I examine the risk and return properties of individual dividend strips, which are claims to short-term dividends from individual companies. First, contrary to the conventional assumption that quarterly dividend payments from individual companies are sticky and certain, I document considerable variability in short-term dividends at the firm level. Uncertainty in dividends of individual stocks in the next quarter can give rise to short-term dividend risk premium at the firm level, which affects the pricing of claims on quarterly dividend payments. Then, I use exchange-traded options on individual stocks to create synthetic dividend strips and use the put-call parity relation to compute prices of dividend strips. During the sample period from 1996 to 2017, the dividend strip aggregated from all individual firms earns an average return of 4.62% per quarter, higher than the average quarterly return of the S&P 500 index during the same period. The high average return on the aggregate dividend strip is consistent with an average downward-sloping term structure of equity premium documented by prior studies using index derivatives. There are substantial cross-sectional differences in returns on dividend strips among individual firms sorted by average normalized dividend premium in the previous four quarters, which is a measure of ex-ante dividend risk premium. Average value-weighted returns on dividend strip portfolios in the highest and lowest quintiles of dividend premiums are 11.91% and -2.87% per quarter, and the spread in return is highly statistically significant. Differences in dividend strip returns are not driven by potential measurement errors in options prices, as option-implied dividends are strong predictors of future dividend payments, and are not driven by differences between dividend payers and non-payers, as the results hold for the subsample of stocks that have ever paid regular cash dividends in the past five years. Variations in returns of claims on short-term dividends do not diminish after controlling for short-sale constraints of underlying stocks and adjusting early exercise premiums in prices of American-style options. In addition, results of both the Fama and MacBeth (1973) cross-sectional regressions and the multivariate test of Gibbons, Ross and Shanken (1989) indicate that the Fama and French (2015) five-factor model can well describe average returns on dividend strips sorted by the ex-ante dividend risk premium. In contrast, the Capital Asset Pricing Model, the Fama and French (1993) three-factor model, and the Carhart (1997) four-factor model seem to be incomplete models. I also use four well-known stock return predictors, book-to-market ratio (BM), operating profitability (OP), total asset growth rate (ATG), and cumulative stock return in the previous six months (RET(-1,-6)) as alternative sorting variables. The four stock return predictors can predict subsequent dividend strip returns in the same direction of prediction on stock returns. The five-factor model performs the best in explaining variations in dividend strips of stocks with different characteristics, which indicates that the superior performance of the model is not specific to dividend strips sorted by historical dividend premium. Dividend strip returns associated with different sorts share common exposures to risk factors other than the market risk which are well captured by the profitability factor and the investment factor.