Economic Mobility and Expenditure Growth: The Effects of Measurement Error in Peruvian Data

Abstract

Although several authors have shown evidence that expenditure and income data obtained from household surveys is typically measured with a large amount of error1, there are not many studies using data from Latin America that address this issue. As a result, studies that calculate different economic indicators, such as poverty or inequality indexes or economic mobility for Latin America countries, are usually contaminated by measurement error in the data (Duval Hernández et.al., 2006).

This paper uses nationally representative panel data from Peru to estimate the effect of measurement error on estimated economic mobility and the growth rate of per capita expenditures for years 2004 to 2006. More specifically, two alternative methods proposed by Luttmer (2002) and Glewwe and Dang (2005) are implemented to estimate the impact of the measurement error on the observed variance of the logarithm of per capita expenditures. These results are then used to assess the relative importance of measurement error in the estimation of economic mobility in Peru between 2004 and 2006 by simulating the join distribution of the logarithm of per capita expenditures corrected for the effect of measurement error. Finally, the magnitude of the bias in the growth of rate of per capita expenditures by quintiles is also evaluated.