Predicting Recessions in Europe: Predicting Recessions Using the Yield Spread and the VIX Index

Abstract

Recessions lead to extensive consequences for countries and individuals. Its ramifications can be decreased and possibly avoided through monetary and fiscal policies, given that the recession is forecasted enough in advance. It is established that the financial indicator that best predicts a recession is the ten-year minus three-month yield spread. The hedging needs of investors and the increased economic uncertainty make the yield spread decrease, ultimately leading to a flat or inverted yield curve before a recession. It has recently been suggested that the VIX index, a forward-looking stock market volatility measure, can improve prediction accuracy for recessions in the United States. This paper shows that the same pattern can be observed in Europe.
The yield spread and the VIX index has shown to have coordinated movements in counterclockwise circles that align with the business cycle. This relationship enables predictions for the European economy to be made based on the countries´ location on the VIX- yield cycles. Recessions appear when the yield spread is low, and the volatility is high. Several probit models using different combinations of the yield spread and VIX index as explanatory variables have been tested, aiming to find the optimal model with the highest predictive power on European recessions. The approach employing ten models, covering different European countries, and spanning six time-horizons, generated a total of 186 test results. The prediction accuracy of the models is illustrated through the bootstrapped ROC curves. This results in out-of-sample AUROC values, which will be supported by the evaluation metrics AIC and BIC to ensure objective conclusions.
The four biggest economies in Europe have been chosen for this study, with them being Germany, the United Kingdom, France, and Italy. There is a correlation between the size of the country and the prediction accuracy of the yield spread model. Nevertheless, the high AUROC values for the biggest economies confirm its predictive power in Europe. However, a model including the yield spread, the VIX index, and the interaction term between the two was found to significantly outperform the industry-standard yield spread model. The interaction term enhances the pre-recession state of an inverted yield curve and high stock market volatility. The model shows great abilities in predicting recessions from the 1990s.