We estimate the impulse response function (IRF) of GDP toa banking crisis, applying an extension of the local projectionsmethod developed in Jorda (2005). This method is shown to bemore robust to misspecification than calculating IRFs analytically. However, it suffers from a hitherto unnoticed systematicbias which increases with the forecast horizon. We propose asimple correction to this bias, which our Monte Carlo simulations show works well. Applying our corrected local projectionsestimator to a panel of 99 countries observed between 1974-2001,we find that an average banking crisis yields a long-term GDP lossof around 10 percent with little sign of recovery within 10 years.GDP losses to banking crises are even more severe in Africancountries. Like the original Jorda's (2005) method, our extensionof it is quite widely applicable.