Abstract
This paper shows that an income effect can drive expenditure switching between domestic and imported goods. We use a unique Latvian scanner-level dataset, covering the 2008-09 crisis, to document several empirical findings. First, expenditure switching accounted for one-third of the fall in imports, and took place within narrowly-defined product groups. Second, there was no corresponding within-group change in relative prices. Third, consumers substituted from expensive imports to cheaper domestic alternatives. These findings motivate us to estimate a model of non-homothetic consumer demand, which explains two-thirds of the observed expenditure switching. Estimated switching is driven by income, not changes in relative prices.