The individual and contextual determinants of unemployment in the U.S.: a multilevel analysis

Abstract

The main objective of this thesis is to empirically examine the effects of individual demographic characteristics and state characteristics on people's unemployment status in the United States using data derived from the 1995 June Current Population Survey (CPS) and the State and Metropolitan Area Data Book 1997-98 (SMADB). The individual-level independent variables include age, sex, race, the highest educational attainment, and foreign-born versus native-born status. The state-level independent variables are: the state average educational level, the state immigration rate, the gross state product per labor force participation, the state-supporting financial aid and investment per capita, and the percentage of manufacturing employment in the state employed civilian labor force. Guided by human capital theory and human ecology theory, the approach of the thesis is to integrate the variables from both of these two levels by using Hierarchical Generalized Linear Model (HGLM) to conduct multi-level analyses. The results from the multi-level analysis indicate that both personal demographic characteristics and state characteristics affect people's unemployment status, although the most deciding factors lie in the individual-level predictors. Specifically, for the level-1 variables, age and educational attainment both have negative effects on the risk of unemployment. Whites, males and more educated people are less likely to be unemployed than their non-white, female, and less educated counterparts. However, an unexpected finding is that native-born people are more likely to be unemployed than immigrants. As to the level-2 variables, both the state immigration rate and the gross state product per labor force participation have positive effects on the likelihood of unemployment, but the percentage of manufacturing employment in the state employed civilian labor force has a negative correlation with unemployment. Moreover, there are a lot of cross-level interactions, showing contextual factors having indirect effects on people's unemployment status, which have significant policy implications. Finally, the limitations of the study are discussed and future research is indicated.