We investigate the extent to which it is possible to detect asset-price booms and banking crises according to alternative identification strategies and we assess their robustness. We find some evidence that house price-booms are more likely to turn into costly recession or to trigger a banking crisis than stock-price booms. Resorting both to a non-parametric approach and a discrete-choice (logit) model, we analyze the ability of a wide set of indicators to robustly explain costly asset-price booms. According to our results, real long-term interest rates and real stock prices tend to increase the probability of a costly housing-price boom, whereas real GDP tends to increase the probability of a costly stock-price boom. Interestingly, the credit-to-GDP gap indicator, sometimes put forward in the literature as a key reference for setting countercyclical capital buffers, does not seem to be a robust leading indicator of costly booms or banking crises.

A bank that lends on the unsecured market requires compensations for facing the default risk of the borrowing bank (credit risk) and the risk associated to its own future funding needs (liquidity risk). In this paper, we propose a quadratic term-structure model of the spreads between unsecured and risk-free interbank rates. Our no-arbitrage econometric framework allows us to decompose the term structure of spreads into credit and liquidity components and to identify risk premia associated with each of these two risks. Our results suggest that, over the period 2012–2013, most of the reduction in interbank spreads comes from a decrease in liquidity-related risk components.