Quantitative models of sovereign debt predict that countries should default during deep recessions. However, empirical research on sovereign debt has found a surprisingly large share of "good times" defaults (i.e., defaults that happen when GDP is above trend). Existing evidence also indicates that, on average, defaults happen when output is close to potential. This paper reassesses the empirical evidence and shows that the detrending technique proposed by Hamilton (2018) yields results that are closer to the predictions of standard quantitative models of sovereign debt.