Macroprudential policy is now based around a countercyclical buffer, relating capital requirements for banks to the degree of excess credit in the economy. We consider the construction of the credit to GDP gap looking at different ways of extracting the cyclical indicator for excess credit. We compare different smoothing mechanisms for the credit gap, and demonstrate that some countries require an AR(2) smoother whilst other do not. We embed these different estimates of the credit gap in Logit models of financial crises, and show that the AR(2) cycle is a much better contributor to their explanation than is the HP filter suggested by the BIS and currently in use in policy making. We show that our results are robust to changes in assumptions, and we make criticisms of current policy settings.
