This paper tests conditional asset pricing models in international markets on value, momentum, and the COMBO anomaly of Asness, Moskowitz and Pedersen (2013) (AMP). We find that incorporating instruments to capture the time variation in risk exposure can significantly reduce the bias in unconditional alpha documented in recent international studies. Particularly, employing the instrumental variables regression approach of Boguth Carlson, Fisher and Simutin (2011) to estimate the conditional Fama-French model can successfully explain returns on COMBO portfolios in North America, Europe, Japan, and the global market. Furthermore, instrumenting the global Fama-French model with lagged component betas can reduce the unconditional AMP's 50–50 COMBO alpha by 11–72%, pointing to the efficacy of this instrumental variable in international markets. Our findings have important implications for international asset pricing theory.
- Conditional asset pricing models
- Lagged component betas
- Multi-factor risk models