Linear Asset Pricing Models
FROM THE VERY beginning of the application of statistical methods to financial market data, expected returns have occupied a central place in empirical finance. This is natural given the fundamental role of agents’ expectations about future asset prices in portfolio theory. We begin our econometric analysis of DAPMs by examining the economic underpinnings of, and historical evidence for, two widely studied restrictions on expected returns: (1) expected holding-period returns on investments are constants (implying, among other things, that asset returns are unpredictable), and (2) the expected returns on two different investment strategies are equal. The former restriction is on the time-series properties of the return on a single asset. The latter is a restriction across securities and, by itself, it does not restrict the time-series properties of either return.
The interest in the first hypothesis of unpredictable returns has arisen primarily in the literature on returns on equity and currency positions, whereas the link between expected returns on different investment strategies has been the central issue in the literature on the term structure of interest rates. After briefly discussing the economic underpinnings of these two hypotheses, we turn to the empirical evidence from equity and bond markets.
Examining Asset Return Predictability
Early analyses of the conditional means of security returns (e.g., Fama, 1965) focused on the null hypothesis that expected holding-period returns on investments are constants. The economic motivation for this null hypothesis was that rational investors should use available information “efficiently” in predicting future stock prices and, as a consequence, stock returns (the changes in logarithms of prices) should be unpredictable.