Do the production-based factors capture the time-varying patterns in stock returns?

As a summarization of previously suggested production-based approaches, Chen et al. (2010) propose two production-based factors. We examine whether the proposed factors explain the time-varying patterns in stock returns, captured by the common conditioning variables. With a variety of test portfolios, we find that the fitted conditional expected return (fit) is always statistically significant in the presence of the production-based factors. Moreover, when the fit is included in the analysis, the magnitude of the production-based factors becomes consistently smaller and the fit drives out the significance of the production-based factors. Our empirical results cast some doubt on the validity of the production-based model as a conditional benchmark for risk adjustment.