Based on Markowitz's pioneering study [40], Sharpe [56] and Lintner [38] advanced the first positivist formulations of the capital asset pricing model (CAPM). Their models were subsequently refined by Mossin [45], Fama [15], Black [1], and others. Even though the CAPM has been studied extensively, it has not been empirically validated. According to Roll [48], the CAPM cannot be tested in an unambiguous fashion because of a number of intractable measurement and computational difficulties, and the joint nature of the hypotheses to be tested.

In a paper published in 1978, Levy [5] proposed a general capital asset pricing model (GCAPM), which he obtained by maximizing investors' utility when the number of securities held in each investor's portfolio is constrained. Although Levy's resultant asset pricing model is somewhat different in appearance than the asset pricing model proposed by Mao [8] in 1971, it can be shown that both models are not only quite comparable in content but that both result in some very promising theoretical and empirical implications. Thus, the purpose of this paper is twofold. First, these two important contributions to the literature on asset pricing in imperfect markets will be compared and contrasted. Second, it will be shown that both models can yield a “clinical” form of the traditional CAPM, which appears to be more desirable for empirical testing purposes.

This study investigates market price integration by testing per-share trade execution price or cost (TEP/C) differentials for matched intraday trades for a sample of Canadian shares cross-listed in the U.S. The TSE trade price advantage over the entire time period changed significantly after both the TSE's own minimum quotation increment reduction and that of its U.S. competitors. We show that the differential TEP/C is equivalent to the international effective spread differential and that market quality comparisons, which benchmark using the National instead of the International BBO, need to compare both national effective half-spread and midspread differences. Our cross-sectional regression results support our predictions that TEP/C differentials can be explained by differences in national midspreads and by ex ante proxies of national effective half-spreads. The TEP/C differentials vary inversely with increasing levels of our measure of signed market nonfragmentation.

This paper tests the overreaction hypothesis using monthly data for stocks listed on the Toronto Stock Exchange over the 1950–1988 period. Unlike De Bondt and Thaler (1985), (1987), it finds statistically significant continuation behavior for the next one (and two) year(s) for winners and losers, and insignificant reversal behavior for winners and losers over longer formation/test periods of up to ten years. While the systematic risks of the winners decrease significantly over all test periods, the systematic risks of the losers increase significantly for only the 12-month formation/test periods (unlike Chan (1988)). The only significant change in variance from the formation to test periods occurs for the losers for the 12-month formation/test periods. The findings are robust for January versus non-January and size-based portfolios (unlike Zarowin (1989), (1990)). The findings are robust for various performance measures (specifically, market-adjusted CAR, and the Jensen (1968) and Sharpe (1966) portfolio performance measures).

Lawrence Kryzanowski, Simon Lalancette, Minh Chau To, Performance Attribution using an APT with Prespecified Macrofactors and Time-Varying Risk Premia and Betas, The Journal of Financial and Quantitative Analysis, Vol. 32, No. 2 (Jun., 1997), pp. 205-224