We assess whether the subprime crisis was in some way unique or unprecedented by testing for an impact of a set of determinants on the probability of banking crises in OECD countries since 1980. The hypothesis of an exogenous probability of a crisis is not considered in the literature, and we also address it, along with testing for the effect of variables that explain crises in emerging markets. We find that a parsimonious logit model can be estimated for OECD<br />
crises, including the subprime period, featuring bank capital adequacy, liquidity, the current<br />
account and changes in house prices as predictors, thus showing the patterns immediately<br />
preceding the subprime crisis were not unprecedented. Not only was the subprime crisis not unique, it was also predictable using data available five years earlier; our model, even if estimated over 1980-2003, could have helped authorities to forecast the subprime crisis and take appropriate regulatory measures.