Is economic recovery a myth? Robust estimation of impulse responses

There is a lively debate on the persistence of the current banking crisis’ impact on output. IRFs estimated by Cerra and Saxena (2008) suggest that these effects will be long lasting. However, standard estimates of IRFs are highly sensitive to slight degrees of misspecification. Moreover, adding fixed effects complicates inference on persistence. Direct estimation of IRFs by a method similar to the local projection method of Jorda (2005) is robust to these specification errors.

Our estimates suggest that an average banking crisis leads to an output loss of up to up to 9 percent, without any recovery within seven years. There are some indications for recovery in later years, but these are insignificant. We find some evidence for heterogeneity in the effects of a banking crisis.