Testing capital asset pricing models using functional-coefficient panel data models with cross-sectional dependence

This paper proposes a functional-coefficient panel data model with cross-sectional dependence motivated by re-examining the empirical performance of conditional capital asset pricing model. In order to characterize time-varying property assets’ betas and alpha, our proposed allows be unknown functions some macroeconomic financial instruments. Moreover, common factor structure is introduced which an attractive feature under regression setting as different assets or portfolios may affected same unobserved shocks. Compared existing studies, such classic Fama–MacBeth two-step procedure, can achieve substantial efficiency gains for inference adopting one-step procedure using entire sample rather than single at each time point. We propose local linear correlated effects estimator estimating pooling data. The consistency asymptotic normality estimators are established. Another methodological challenge in how test constancy significance errors, we echo this constructing L2-norm statistic models allowing dependence. show that new has limiting standard normal distribution null hypothesis. Finally, method applied Fama French (1993) Fama–French 25 100 portfolios, sorted size book-to-market ratio, respectively, dated from July 1963 2018.