Estimation of Poverty Transition Matrices with Noisy Data

Nayoung Lee, University of Southern California
Geert Ridder, University of Southern California

Poverty transition matrices have been regarded as powerful tools to estimate economic mobility and poverty dynamics. Very few researchers, however, have investigated potential measurement error bias in these transition matrices constructed by survey data, although the presence of measurement error in survey data has been widely acknowledged. This study uses data from the Korean Labor and Income Panel Study (KLIPS) to examine whether the measurement error in reported income and consumption has the potential to generate biases for poverty transition matrices, an empirical measurement of poverty dynamics. We simulate income and consumption transition matrices first assuming that no time-varying measurement error exists and then allowing for it. This study suggests that poverty transition matrices based on income or consumption survey data are biased due to the existence of measurement error.

  See extended abstract

Presented in Poster Session 2