Errors in self-reported earnings: the role of previous earnings volatility

I report the measurement error in self-reported earnings for a developing country. Administrative data from the Federated States of Micronesia's (FSM) Social Security office are matched to the FSM Census data for the wage sector employed. I find that the error in annual self-reported earnings is centered on zero but less efficient than results from the US. Additionally the error is not classical in nature - I find evidence for mean reversion in the data. Using previous annual earnings history contained in the FSM Social Security data, I construct accurate measures of past deviations of administratively recorded earnings to identify the impact of transitory income on current reporting of earnings. Prior earnings volatility is an important determinant of the error in earnings for the current period. However, the effect of prior shocks diminish significantly over time - suggesting that information on transitory income shocks will be helpful in evaluating the usefulness of self-reported earnings measures in applied work. Finally, I use information on an exogenous and transitory shock to FSM household incomes (typhoons) to correct for errors in self-reported earnings. I find that the coefficients from these corrected regressions approach those that use administrative data on earnings in a consumption regression. -- Measurement error ; earnings ; instrumental variables ; transitory income