Market Risk Premiums and the Macroeconomy: Canadian Evidence of Stock Market Predictability

Article excerpt


There has been much interest in the academic community regarding stock market predictability. Two broad streams of research have evolved. First, there is the stream that shows stock returns are predictable from previous returns. Fama and French (1988b), Lo and MacKinlay (1988), Poterba and Summers (1988), and Jegadeesh (1990) show that short-term stock returns exhibit significant positive serial correlation and long-term stock returns exhibit significant negative serial correlation. Brock, Lakonishok, and LeBaron (1992) and Jegadeesh and Titman (1993) examine technical trading rules and relative strength models, respectively, and demonstrate they may be implemented profitably. Second, many researchers have examined the predictability of stock returns using economic variables both endogenous and exogenous to the equity markets. For example, Campbell (1987), Fama and French (1988a, 1989), Fama (1990), Chen (1991), Harvey (1991), Campbell and Hamao (1992), and Ferson, Foerster, and Keim (1993) examine the predictability of stock returns from various economic variables.

This study follows the second stream by examining the predictability of stock market returns, or more specifically stock market risk premiums, using observable macroeconomic state variables. Risk premiums have been shown by Ferson and Harvey (1991) to account for the vast majority of the predictability of stock returns at the portfolio level and therefore are used instead of real stock market returns in this paper. The paper proceeds under the premise that if it is possible to find a set of financial state variables that predict real Gross Domestic Product (GDP) growth, then it may be possible to use these same financial state variables to predict stock market premiums. The study is unique in that few investigations have focused on the predictability of the Canadian stock market. Furthermore, few Canadian or U.S. studies have examined the profitability of such predictability.

This research is motivated by Chen's (1991) examination of the relationships between U.S. real output growth, equity market risk premiums, and five macroeconomic state variables. Chen shows that a set of information variables commonly used to determine conditional expected stock returns is related to U.S. output growth in a manner that is consistent with general intertemporal asset pricing models. As Fama (1991, p. 1610) notes, establishing a link between expected returns and business conditions remains one of finance's more important areas of future research. This paper attempts to add another small piece to the puzzle. The analysis presented here is similar to Chen's in that it examines the same relationships and state variables, but it uses Canadian data over a longer sample period and thus endeavors to provide valuable out-of-sample support for his work.

The study extends Chen's (1991) analysis by examining the ex ante profitability of implementing the market premium prediction model with an operational trading strategy. This helps determine if stock market predictability may be used to generate statistically significant returns above a buy-and-hold market portfolio. If it can be shown that economic agents can earn higher risk-adjusted returns than the market portfolio by using publicly available information variables, then the market efficiency debate becomes more than an academic exercise. If higher returns can be generated from these information variables, then the market is either inefficient or economic agents rationally give up this potential return. Investors may give up this return because the loss of expected utility from a less smooth consumption path is higher than the utility gained from capturing the return. Thus, if the world is assumed rational and efficient, this potential excess return is a measure of the return that is foregone by rational economic agents to meet their objective of smoother consumption.


Although most papers examining the predictability of stock returns are empirical studies with little theoretical foundation, several asset pricing models relate expected stock returns to economic variables. …