Unemployment and Suicide in the United States: The Import of Addressing Cross-Sectional Dependence

Abstract Recent reviews of the sociological and economic-based ecological studies of suicide find cyclical unemployment to be a key suicide risk factor, though the evidence presented is mixed at best. The ambiguity of the ecological associations appear to stem from faulty statistical methodologies. Panel treatments offer advantages over conventional time-series methods by exploiting cross-section variation. However, if the added cross-section units are cointegrating (dependent) and independence is presumed, incorrect statistical inference and inconsistent coefficient estimation can result. Herein, we fully address the import of cross-sectional dependence on the ecological relationship between U.S. unemployment rates and suicide rates using an 81-year panel of the 48 contiguous states and the District of Columbia. When proper allowances are made for cross-section dependence at each step of the examination, we find no significant statistical association, short-run or long-run, running from unemployment to suicide rates in the U.S. Results of this sequential analysis highlight potential sources of the ambiguity found in the literature. JEL classification number: B55, C31, C49. Keywords: Suicide rates, Cointegration, Cross-section dependence.