Testing factor models when asset bubbles occur A time-varying perspective-张培刚发展研究院

Financial market conditions vary over time, which makes it important to consider a factor model’s performance in different situations. The existing literature typically tests factor pricing models over a long time horizon. In this study, we test a three-factor model on Chinese data in two market periods: the bubble and the normal. We find that stock returns are more volatile during the bubble period, which results in severe power loss in the Gibbons–Ross–Shanken (GRS) test. To this end, we propose a wild bootstrap GRS test to address the effect of time-varying volatility on stock returns. The model has excellent explanatory power in the normal period but performs poorly in the bubble period. Our approach is applicable to testing factor models in time-varying real stock returns.

ISSN号：0264-9993

链接：Testing factor models when asset bubbles occur: A time-varying perspective - ScienceDirect