In the literature, the multivariate tests of asset-pricing models are developed focusing on the characteristics of developed markets. For example, the pioneering Gibbons’ (1982) test of the “capital asset-pricing model” (CAPM) and Harlow and Rao’s (1989) test of the “mean lower partial moment” (MLPM) model both employ a “likelihood ratio test” that assumes multivariate normality of monthly asset returns. Emerging market returns are known to be non-normal and have greater predictability than those of developed markets. Considering these stylized facts, the paper extends Harlow-Rao’s likelihood ratio test to develop multivariate tests robust to these features. In a sample of portfolio data from an emerging market, namely Pakistan, it is shown that multivariate tests of both CAPM and MLPM individually do not reject the restriction of the two financial models, respectively. However, a more powerful nested test of CAPM against MLPM with bootstrap p-values rejects the CAPM in favor of MLPM.
|Title of host publication||Financial Econometrics Modeling|
|Subtitle of host publication||Market Microstructure, Factor Models and Financial Risk Measures|
|Editors||Greg N Gregoriou, Razvan Pascalau|
|Place of Publication||Basingstoke UK|
|Number of pages||22|
|Publication status||Published - 2011|