Abstract

Two measures of investor sentiment or risk appetite have recently attracted the attention of researchers as predictors of business cycles and recession risk. One is variance risk premium (VRP), which is the difference between the conditional variance of stock returns under the risk-neutral measure and that under the physical measure. The other is excess bond premium (EBP), which is a component of credit spreads that is not attributable to expected default risk. This paper examines their predictive power for business cycles and recession risk. Specifically, we compare their predictive power for the composite index of business conditions, the index of industrial production and the unemployment rate in Japan as well as the likelihood of an ESRI-dated recession occurring over the next 12 months. The Japanese volatility index (VXJ) published by Osaka University is used as the conditional variance of Nikkei 225 stock index under the risk-neutral measure. The conditional variance of Nikkei 225 stock index under the physical measure is calculated using the extended heterogeneous autoregressive (HAR) model with the realized volatility. EBP is calculated using the credit spreads of corporate bonds in Japan. We show that VRP performs better than EBP in the prediction of business cycles and recession risk in Japan.