A nonparametric analysis of the personal income distribution across the provinces and states in the U.S. and Canada

Abstract

Baumol (1986) and Quah (1993) have all been strong proponents of the idea of grouping similar countries or economies into different convergence clubs, such as the rich country club, the poor country club etc.While overall convergence may not exist, countries within a convergence club may show signs of convergence (club convergence). regression results.From the above one may realize that the test of multi-modality can be very useful in the research of income distributions.As indicated by Zhu ( 2002) important hypotheses such as income convergence, polarization, poverty traps and "vanishing middle class" can' be formulated as a question of multi-modality.In this paper, several nonparametric approaches, specifically, an adaptive kernel density estimator, one mode 4 testing procedures, stochastic kernel and Markov transition probability matrix were used to investigate convergence issues and analyze distribution dynamics of the personal income distribution across the 59 provinces and states in the US and Canada.The non-parametric approach used in this paper is different from a conventional one.No particular functional form is imposed on the underlying density as in parametric approach.The empirical results of this work indicate that over the study period 1950-2000, multimodality is a significant feature of the regional personal income distribution across the U.S. and Canada.Convergence (unimodality) is found over the 1950s and most of the 1960s.Evidence in favor of bimodality is found during the 1970's and the 1980's.In the 1990s, there is a strong indication that the number of the poor regions increased dramatically.A mode is defined as a point at which the gradient of the density changes from positive to negative.