Economic impact of poverty and social isolation: an exploration with US State data

Abstract

A. Objectives: The policy debates surrounding the apparent trade-off between economic growth and income inequality, and consequent increase in poverty rate, have gained momentum since the financial debacle in 2008. In this paper, we explore the causal link between poverty, economic performance and social isolation. We contribute to the literature in two ways; first, our analysis overcomes the limitations of cross-country studies by using state-level data for three years-2000, 2005 and 2010; second, we include a variable for social isolation, to explain the differences in economic performances across the US states. We posit that social isolation, defined as poor English language proficiency, restricts economic integration and hence has a negative economic impact. B. Data and Methods: Employing data from the US Bureau of Census, we explore if differences in the poverty rate and linguistic isolation can explain the differences in Gross State Product (GSP) per capita across 50 US states. We estimate equation (1): [see presentation page 8 for equation] In equation (1), α2 represents state fixed effects (for state specific unobserved factors) ,D₁ and D₂ are time dummy variables for year 2005 and 2010, respectively, and s(s,t)is a vector of control variables (educational attainment, employment in manufacturing industries and density of population).The dependent variable is GSP per capita (in 2005 prices) and the main causal variable x s,t is defined as the headcount index. The interactive term measures the possibility that the marginal impact of poverty on GSP depends on social isolation z(s,t). We argue that the negative impact of poverty will be higher for states with a higher proportion of socially isolated people. C. Results & Concluding Remarks: We conduct a Breusch-Pagan (BP) Test and a Hausman test to check for Fixed Effect (FE) specification. Both tests reject the null hypotheses for pooled OLS and Random Effect model. FE estimation of equation (1) reports that the coefficients for x(s,t) and its interactive term with z(s,t)are negative and are statistically significant. Our estimates indicate that a 1% increase in poverty rate, GSP per capita reduces by 1.4%, and it could further reduce by a 0.08 % in states with a higher proportion of linguistically isolated people. We checked for model robustness, against omitted variables, and estimation with 48 continental states. Our results provide an economic rationale for reducing social isolation, as part of the poverty reduction strategies in the US, and countries like the UK, with large immigrant population.