Expected credit losses under imperfect foresight: insights from a partial equilibrium model

ABSTRACT We develop a partial equilibrium model to assess how backward-looking provisioning, loan loss provisioning and expected credit losses determine the temporal pattern of recognition of credit losses and the optimal credit spreads for banks. We apply our model to a dataset covering the Spanish economy between 1984 and 2018. Backward-looking provisioning results in highly correlated recognition of credit losses with the real economy and in low credit spreads in good times. Our loan loss provisioning would decouple the recognition of credit losses from the evolution of the real economy and would lead to substantially higher credit spreads in normal times. Expected credit losses with imperfect foresight do not significantly anticipate the recognition of credit losses before recessions. In general, optimal credit spreads under expected credit losses would not materially depart from backward-looking provisioning.