This project will examine the effects of the deteriorating employment and earnings in local labor markets on maladaptive social behavior and political extremism.
One of the long-standing goals of the Foundation’s program on the Future of Work has been to assess the broader social consequences of the long-term structural changes in the U.S. labor market that have left many American workers, particularly those with no more than a high school education, without the prospect of a stable, high-quality job. There is no shortage of correlational studies showing that individuals who lose their jobs or experience marginal employment show high rates of depression, alcohol and drug use, criminal involvement, and other sorts of self-damaging behavior, but it is hard to disentangle cause and effect in these studies. It may well be true that job loss and marginalization cause people to become depressed and behave maladaptively, but it may equally well be that people with mental and behavioral problems are most vulnerable to being fired or marginally employed.
Economists David Autor, David Dorn, and Gordon Hanson propose to address exactly this causal question by building on their impressive recent research showing that the rapid increase in imports from China since 1990 has depressed employment and earnings and increased the use of social safety-net programs most significantly in those local U.S. labor markets that are most vulnerable to import competition from China. Autor and colleagues mapped the national labor market into some 720 "commuting zones" (or CZs) which are clusters of counties that are characterized by strong within-zone commuting ties and weak between-zone commuting ties. By comparing the distribution of employment across industries in each CZ prior to the acceleration of trade with China against the distribution of industrial products subsequently imported from China, Autor et al. were able to derive a measure of import exposure per worker for each CZ. As Chinese imports rose (more than eleven-fold between 1990 and 2007), those CZs with the greatest import exposure showed the largest declines in manufacturing employment, the largest wage losses outside of manufacturing, the steepest drops in household income, and the sharpest increases in transfer payments from multiple federal and state programs, such as federal disability insurance, Social Security and Medicaid. These effects were not small. To take just one example, Autor et al. estimate that the employment-to-population ratio declines about 1.1% for every additional $1,000 of import exposure in a CZ. In the 2000s, as imports from China accelerated, this effect reduced employment in a highly trade-exposed CZ like Providence, Rhode Island, by about 5%.
Autor, Dorn, and Hanson now propose to use the same methodology to investigate the possible social and political consequences of the labor market shock arising from increased trade with China between 1990 and 2007. Their goal is to exploit the exogenous nature of this shock to employment and earnings to get a better handle on difficult causal questions about the social and political ramifications of diminished labor market prospects. Holding constant relevant demographic and labor market features of the CZs under study, they plan to estimate the degree to which those CZs that suffered the most severe shocks to employment and job opportunities also suffered significant declines in physical and mental health, experienced increases in the incidence of maladaptive social behavior, and embraced more extreme political positions – on the left or right.
To study the effects on mental and physical health and maladaptive behavior, Autor, Dorn, and Hanson plan to use the Center for Disease Control’s Behavioral Risk Factor Surveillance System (BRFSS), a national telephone survey, which tracks person-level data on general physical health status, mental health, anxiety and depression, physical activity, diet and weight, health care access, tobacco and alcohol use, diabetes, blood pressure, and heart condition as well as basic demographic information on family structure, education, and employment status. The BRFSS sample has expanded substantially, from 82,000 respondents in 1990 to over 430,000 in 2007. County identifiers are available from 1994 on, which will allow Autor et al. to link individual respondents to the commuting zones in which they reside. To measure the incidence of out-of-wedlock childbearing the PIs will additionally incorporate Vital Statistics data which provide a census of births and demographics of the birth parents for most U.S. counties.
To examine whether labor market shocks trigger changes in political orientation, Autor and his colleagues plan to employ both data on voting in presidential elections and legislative votes by members of Congress. They will assemble presidential voting totals by county and link these to commuting zones. This will allow them to chart changes in average political orientation by CZ. For more fine-grained analysis they plan to employ data on presidential voting at the census block level, available since 2000, which will allow them to investigate whether trade shocks have distinct effects on voting behavior in high- and low-income neighborhoods, or in neighborhoods that are more, or less, directly affected by trade shocks. To analyze changes in the political orientation of Congressional representatives for each local labor market, Autor et al. will rely on Poole and Rosenthal’s DW-Nominate scores, which track the ideological position of members of congress based on their patterns of legislative votes. Using GIS techniques to compute the overlap between Congressional districts and commuting zones, Autor and his colleagues will test the hypothesis that CZs with the highest levels of trade exposure tended to elect increasingly extreme members of Congress, or to re-elect incumbent representatives who adopt increasingly extreme political positions, as the labor market shocks from Chinese imports grew from 1990 to 2007.