We use a simple financial friction in an economy with high degree of liability dollarization - and currency mismatch - to show that the negative balance-sheet effect of an exchange rate depreciation may be observable only if the magnitude of the depreciation is large enough. This result justifies the difficulty to find strong empirical evidence for balance-sheet effects and suggests the convenience of including a "large depreciation" term in empirical analyses. We review some of the related empirical literature and provide some new evidence of this large depreciation effect. © 2011 Elsevier Ltd.