What Drives Aggregate Credit Risk?

A deep understanding of the drivers of credit risk is valuable for financial institutions as well as for regulators from multiple viewpoints. The systemic component of credit risk drives losses across portfolios and thus poses a threat to financial stability. Traditional approaches consider macroeconomic variables as drivers of aggregate credit risk (ACR). However, recent literature suggests the existence of a latent risk factor influencing ACR, which is regularly interpreted as the latent credit cycle. We explicitly model this latent factor by adding an unobserved component to our models, which already include macroeconomic variables. In this paper we make use of insolvency rates of Austrian corporate industry sectors to model realized probabilities of default. The contribution of this paper to the literature on ACR risk is threefold. First, in order to cope with the lack of theory behind ACR drivers, we implement state-of-the-art variable selection algorithms to draw from a rich set of macroeconomic variables. Second, we add an unobserved risk factor to a state space model, which we estimate via a Kalman filter in an expectation maximization algorithm. Third, we analyze whether the consideration of an unobserved component indeed improves the fit of the estimated models.