The Determinants of Cross-Country Differences in the Severity of the Late-2000s Crisis

In this study I aim to assess variables which could be associated with output losses due to the late-2000s crisis in a broad global sample of countries. First, I estimate crisis severity as cumulative negative deviations of either output growth or output level from estimated potential trends, which is a method commonly used to estimate overall costs of financial crises to economies. Second, I explore a large number of pre-crisis conditions and structural indicators seeking for the ones that may be statistically significantly associated with cross-country differences in output losses. I apply heteroscedaticity robust Tobit estimations and calculate marginal effects on expected outcomes of output losses. Income per capita, pre-crisis output growth and credit expansion appear as consistently positive statistically significant indicators that marginally increase expected output losses. After controlling for these factors, marginal effects of a number of other indicators lose statistical significance or change signs. The ones that may be further associated with higher output losses for a large number of countries are current account deficits, inadequate international reserves relative to the extent of financial depth (monetary base) and relatively smaller equity markets.