Testing asset pricing models with individual stocks: An instrumental variables approach

Abstract

This study empirically tests time-varying asset pricing models in an emerging market with individual stocks. We employ a recently proposed instrumental variables (IV) technique that uses individual stocks as test assets while consistently estimating ex-post risk premiums. This method differs from constructing test portfolios, a common practice employed to mitigate errors-in-variables bias, and, instead, uses factor sensitivity estimates from alternating even and odd months as IVs. Applying this approach, we observe statistically insignificant factor risk premiums under various multifactor models in asset pricing tests at Borsa Istanbul, after accounting for asset characteristics. Our method facilitates the inclusion of essential risk or return-related characteristics of individual stocks in tests, raising insights usually obscured by conventional test portfolios. The results contribute to empirical asset pricing by highlighting the failure of classical models to explain risk premiums at Borsa Istanbul, a significant emerging stock market, when tested with individual stocks using an IV approach.