Measuring the Shadow Economy: Endogenous Switching Regression with Unobserved Separation

We develop an estimator of unreported income that relies on much more flexible identifying assumptions than those underlying previous estimators of the shadow economy using household-level data. Assuming only that evading households have a higher consumption-income gap than non-evaders in surveys, an endogenous switching model with unknown sample separation enables the estimation of both the probability of hiding income and the expected amount of unreported income for each household. Using data from Czech and Slovak household budget surveys, we find the size of the shadow economy to be substantially larger than estimated using other techniques. These results are robust under a number of alternative specifications.