Household surveys underreport incomes from the upper tail of the distribution, affecting our assessment about inequality. This paper offers a tractable simulation method to deal with this situation in the absence of extra information (e.g., tax records). The core of the method is to draw pseudodata from a mixture between the income empirical distribution and a parametric model for the upper tail, that aggregate to a preestablished top income share. We illustrate the procedure using Peruvian surveys that, as in the rest of Latin America, have displayed a sustained decrease in the Gini index since the 2000s. In a number of experiments, we impose a larger top income share than the one observed in the data, closer to corrected estimates for less egalitarian neighbors (e.g., Colombia and Chile). We find that even though the point estimates of the Gini index are biased, the corrected indices still decrease in time.
Nota bibliográficaFunding Information:
We are indebted to Gustavo Yamada for insightful discussion. We would also like to thank Marco Terrones, Javier Torres, the editor Daniel Waldenström and two anonymous referees for very useful comments. The financial support of the Research Center of Universidad del Pacífico (Project PPA 2018) is gratefully acknowledged. We alone are responsible for the views expressed in this paper and for any remaining errors.
© 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.