Inequality Convergence : How Sensitive are Results to the Choice of Data?

This paper examines the extent to which estimates of inequality convergence are sensitive to the choice of welfare concept, inequality indicator, database, country coverage, and time period. Moreover, we explore the sensitivity of the estimated rate of convergence by testing five hypotheses using a series of pair-wise F-tests. The main takeaways are as follows. First, estimates appear to be more sensitive to the choice of welfare concept than to the choice of inequality measure. Second, different international inequality databases frequently produce different results, even when the countries, the welfare concept, the inequality measure, and the time period are held constant. Third, while there is a rather large amount of evidence that estimated rates of convergence differ by region and by time, even this result is sensitive to the database that is used to perform the analysis