Estimating Dynamic Panel Data Models: A Practical Guide for Macroeconomists

Previous research on dynamic panel estimation has focused on panels that, unlike a typical panel of macroeconomic data, have small time dimensions and large individual dimensions. We use a Monte Carlo approach to investigate the performance of several different methods designed to reduce the bias of the estimated coefficients for the longer, narrower panels commonly found for macro data. We find that the bias of the least squares dummy variable approach can be significant, even when the time dimension of the panel is as large as 30. For panels with small time dimensions, we find a corrected least squares dummy variable estimator to be the best choice. However, as the time dimension of the panel increases, the computationally simpler Anderson-Hsiao estimator performs equally well. We apply our recommendations to a panel of countries to show that increases in income growth precede increases in savings rates and increases in savings rates precede declines in income growth. JEL Codes: C23, O11, E00 Keyword: Panel data, simulation, dynamic model, macroeconomics, growth