The impact of business cycles on multi-factor models

Abstract

Many empirical studies to date paint a picture of the economy as having a consistent form at every single time-series. Contrary to that view, we have seen financial markets undergo lots of movements and some of these unpredictable. These fluctuations can range from local disturbances to yearlong tendencies.

This thesis demonstrates empirically the effects of considering different business cycles on the accuracy of traditional (multi-)factor models, especially in the European Monetary Union. Indeed, when a market shifts to another state, factor sensitivities and factor premiums do not remain static. Therefore, statistical proof is put forward to support the fact that for some specific cycles conditional versions have better explanatory power in the cross-section of stock returns. Next to this, some market anomalies showed to still be present in certain states.

Before considering the integration of new factors, the developed conditional model aims to improve the predictability of future stock returns. Regarding this, some leading indicators have been attributed key roles in the final model.