On the dynamics of poverty and income inequality in US states

Purpose – This study seeks to provide answers to the following questions: Is there a relationship between poverty and income inequality in the short run/long run? Is the relationship unidirectional from income inequality to poverty as the previous studies assume, or is it bidirectional? Design/methodology/approach – The paper investigates the causality between income inequality and poverty within a multivariate framework using a panel data set of 50 US states over the period 1980 to 2004. Findings – The results reveal that a bidirectional relationship exists between poverty and income inequality both in the short run and in the long run. With respect to the short‐run dynamics associated with poverty, both income inequality and the unemployment rate have a positive and statistically significant impact on poverty, a negative and statistically significant impact for real per capita personal income and level of education, while corruption is insignificant. In terms of the short‐run dynamics associated with income inequality, poverty, the unemployment rate, real per capita personal income, and the level of education have a positive and statistically significant impact, while corruption has a statistically insignificant impact on income inequality. With regard to the long‐run dynamics, the statistically significant error correction terms indicate the presence of a feedback relationship between poverty and income inequality. Originality/value – This study contributes to the existing empirical literature on several fronts. First, while previous studies rely on cross‐country data, both the cross‐sectional and time series variation of a panel data set of all 50 US states spanning the period 1980 to 2004 is exploited. Second, while the previous studies implicitly assume unidirectional causality from income inequality and poverty in their model specifications, the short‐run and long‐run causal dynamics between poverty and income inequality within a panel vector error correction model are explicitly examined.