Does investor sentiment as conditioning information help to explain stock returns behaviour? A test of alternative asset pricing models

Purpose - The purpose of this paper is to use investor sentiment (IS) as a conditioning information variable for the cross-sectional return predictability tests of alternative asset pricing models (APMs). Design/methodology/approach - Cross-sectional tests of alternative APMs in the linear beta representation and stochastic discount factor specifications, Fama and Macbeth and generalized method of moments techniques have been used. Findings - Results reveal that IS as a conditioning information variable contains significant information for making the discount factors time varying. Model comparison test statistics suggests that among the alternative APMs, the conditional five-factor model (FFM) performs better. Research limitations/implications - Empirical analysis does not extend to the inclusion of the business-cycle conditioning information variables for the test of APMs. Practical implications - The potential benefit of the conditional FFM can be leveraged upon for cost of capital determination, and mutual fund manager's portfolio performance evaluation when the portfolio is heavily weighted with sentiment-sensitive hard to value and difficult to arbitrage stocks. During volatile and boom periods in stock markets the IS scaled conditional APMs may be useful for the fundamental value determination of sentiment-sensitive stocks. Originality/value - This study extends available literature in the context of both developed and emerging equity markets by exploring the cross-sectional tests of conditional APMs using IS as the conditioning information variable. To the author's knowledge, this is perhaps the first study that examines IS as conditioning information for the cross-sectional tests of alternative APMs.