Do tightenings of bank lending standards permanently reduce bank lending? We construct a measure of a bank’s level of lending standards using micro-data from the sample of banks participating in the Eurosystem Bank Lending Survey in The Netherlands and show that this level measure affects business lending. The level effect is statistically robust and economically relevant; a one point tightening reduces a bank’s quarterly growth rate of business lending by about half a percentage point until bank lending standards are eased. This level effect of bank lending standards helps to explain low bank lending growth after a period of prolonged tightening as well as high bank lending growth in a period of prolonged easing. As such, the analysis provides another potential indicator for macroprudential policy.

This paper contributes to the literature on the relation between bank profitability and economic activity. When allowing for stronger co-movement of bank profit with economic activity during deep recessions, we find a much larger impact of output growth on bank profitability than commonly found in the literature. Among the different components of bank profit, loan losses are the main driver of this result. We also find long-term interest rates in previous years to be important determinants of bank profit in times of high economic growth. Our findings are robust to the use of aggregate or individual bank data.

This paper examines which variables have predictive power for financial stress in 25 OECD countries, using a recently constructed financial stress index (FSI). First, we employ Bayesian model averaging to identify leading indicators of stress. Next, we use those indicators as explanatory variables in a panel model for all countries and in models at the individual country level. It turns out that panel models can hardly explain FSI dynamics. Although better results are achieved in country models, our findings suggest that (increases in) financial stress is (are) hard to predict out-of-sample despite the reasonably good in-sample performance of the models.