Testing Convergence in Income Distribution*

Abstract

The generalized method of moments (GMM) estimator is often used to test for convergence in income distribution in a dynamic panel set-up. We argue that though consistent, the GMM estimator utilizes the sample observations inefficiently. We propose a simple ordinary least squares (OLS) estimator with more efficient use of sample information. Our Monte Carlo study shows that the GMM estimator can be very imprecise and severely biased in finite samples. In contrast, the OLS estimator overcomes these shortcomings. Copyright (c) Blackwell Publishing Ltd and the Department of Economics, University of Oxford, 2008.

Inequality convergence in a panel of states

“…Thus, the preliminary results lend very weak support, if any, to the convergence in income inequality by using the 'top1' indicator. This may not appear too surprising as Bao and Dhongde [1] have shown that the GMM estimates can be severely biased in finite samples and the associated variances (or standard errors) can be very imprecise as well. This is particularly true in our application where there are only 48 states (N = 48).…”

“…As such, Bao and Dhongde [1] propose to estimate the parameters by using an OLS procedure based on the following first-differenced model…”

“…In a recent paper, Bao and Dhongde [1] demonstrate that the GMM estimator, though consistent, can utilize the sample observation in an inefficient way. In response to the above concern, this paper applies the novel OLS estimator, proposed by Bao and Dhongde [1], to reinvestigate the validity of the inequality convergence hypothesis in a dynamic panel set-up.…”

“…In response to the above concern, this paper applies the novel OLS estimator, proposed by Bao and Dhongde [1], to reinvestigate the validity of the inequality convergence hypothesis in a dynamic panel set-up. The advantage of the OLS estimator over the GMM counterpart is that it makes use of a greater number of observations over time.…”

“…Despite its potential importance, nevertheless, this issue has been largely ignored in the current literature that assesses the validity of the convergence hypothesis implied by the neoclassical growth model, yet with a few exceptions. 1 Of the relevant literature, we notice that Panizza [13] follows the approach of Caselli et al ([7], hereafter CEL) to test inequality convergence for the U.S. in the panel data framework. In particular, CEL's approach shows that incorrect treatment of sectionspecific effects can cause the problem of omitted variable bias.…”