نوسانات بازار کار محلی و سلامت: آیا یک اتصال و برای چه کسی وجود دارد؟

We examine the relationship between local labor market conditions and several measures of health and health behaviors for a sample of working-aged men living in the 58 largest metropolitan areas in the United States. We find evidence of procyclical relationships for weight-related health and mental health for men with low ex ante employment probabilities. Separate estimates suggest worsening labor market conditions lead to weight gains and reduced mental health among African-American men and lower mental health among less-educated males. Among our findings, those related to mental health are most pronounced.

Economists have devoted much attention to the impact of macroeconomic fluctuations on a variety of outcomes, including earnings and their distribution, employment, criminal activity and human capital investment. While interest is rising, they have paid less attention to a possible connection to health. Using repeated cross-sectional data from the National Health Interview Surveys (NHISs), we estimate relationships between local labor market conditions and several measures of health and health behaviors for a sample of individuals living in the 58 largest metropolitan statistical areas (MSAs) in the United States. The paper's primary contributions are threefold: (1) we consider local rather than more aggregate labor market fluctuations, (2) we explore more detailed measures of mental health, and (3) we examine whether economic conditions have larger impacts on individuals with relatively poorer labor market prospects. The paper proceeds as follows: in the remainder of this section, we discuss why health may vary with local labor market fluctuations, whose health might be most affected and briefly review the relevant literature. In Section 2, we describe our data, focusing on key variables and the construction of our analysis sample, which consists of working-aged men. Section 3 presents our empirical strategy which relates local labor market conditions, via MSA-level unemployment rates, to measures of health and health behaviors that may vary over short periods of time. Since the effect of labor market conditions on health may depend on the extent to which one's present or prospective employment is impacted by them, we divide our sample into groups whose employment prospects are potentially more and less likely to be affected by such fluctuations. For example, we allow the effect of local labor market conditions to vary by race and education group since previous research suggests the labor market outcomes of non-white and less educated individuals are relatively more impacted by economic fluctuations. In addition, we allow this effect to vary by one's potential “exposure” to labor market fluctuations, as proxied by their predicted employment status. Section 4 presents our principal findings and briefly discusses them. For those men least likely to be employed, we find evidence of a procyclical relationship for weight-related health and mental health. Consistent with these findings, we present evidence that worsening labor market conditions lead to weight gains and reduced mental health among African-American men and lower mental health among less educated males. Among these findings, those related to mental health are most pronounced. This is significant given the connection of mental health to other phenomena such as homelessness, drug abuse and criminal activity. It is also important since it may provide information to policymakers on how best to allocate scarce health resources. Section 5 discusses our most prominent findings and Section 6 concludes the paper. 1.1. Why might local labor market conditions affect health? Conceptually, local labor market conditions may affect health for a variety of, potentially conflicting, reasons. Two general explanations have gained prominence in related work. The first can be classified as a “behavioral” explanation since it implies that health impacts propagate through changes in individual behavior, while the second can be considered a “structural” explanation as it implies labor market conditions can affect health absent any explicit behavioral changes by individuals. To elaborate, we briefly consider each in the context of a labor market contraction. First, local labor market fluctuations might impact health through changes in the opportunity cost of time. When the unemployment rate rises, employment is reduced on intensive and extensive margins. Such reductions lower the opportunity cost of other, non-market activities including household production. One form of household production that is very time-intensive is the production of health.1 Facing lower time costs, affected individuals may spend more time in activities intended to improve their health (e.g., exercising, producing and consuming homemade rather than mass-produced or restaurant meals, or using preventive medical services). If investment in such activities improves health and does so in a reasonably short period of time, a countercyclical relationship between labor market conditions and health will obtain.2 Another channel through which fluctuating labor market conditions might affect health is sometimes referred to as the “economic stress” hypothesis (c.f., Catalano and Dooley, 1983 and Catalano, 1991). In general, the idea is that a weaker economy leads to increased stress due to greater uncertainty of present and future income receipt. In turn, this greater stress level leads to reductions in health.3 In addition, such uncertainty over income may have “feedback” effects in that it may increase the likelihood of life events such as bankruptcy or marital dissolution which may add to the stress associated with a downturn in the labor market. If the stress hypothesis is operative and if greater stress reduces health in the short-run, a procyclical relationship between labor market conditions and health will obtain.4 1.2. Who might be most affected by labor market fluctuations? While these two general explanations are not mutually exclusive and certainly do not exhaust the mechanisms through which labor market conditions may affect health, they do indicate that their directional impact on health is an empirical question. A separate issue is whose health is most likely to be impacted by such fluctuations. Since the question of interest is whether labor market conditions impact health, individuals whose current employment or employment prospects are most affected by labor market fluctuations may be most likely to experience corresponding changes in health, if such effects exist. But who are these individuals? Previous work suggests the labor market outcomes of “lower-skilled” individuals are disproportionately affected by economic fluctuations. Of these studies, the ones that use MSA-level variation in labor market conditions to examine labor market outcomes such as earnings and employment are most relevant to this study ( Bartik, 1991, Bartik, 1993a, Bartik, 1993b, Bartik, 1994, Bartik, 1996, Bound and Holzer, 1993 and Bound and Holzer, 1995). Generally speaking, these authors find greater sensitivity to economic fluctuations among non-whites, younger individuals and individuals with lower education levels. That is, these groups are relatively more likely to have positive labor market outcomes during economic expansions and negative ones when labor market conditions deteriorate. 5 While not as directly relevant, studies which use national-level variation or focus on younger individuals tend to find similar patterns (c.f., Blank, 1989, Acs and Wissoker, 1991 and Freeman, 1991). More recently, a comprehensive study by Hoynes (2000) finds that the labor market outcomes of non-whites and those with lower levels of education are relatively more impacted by changes in local labor market conditions. 6 In particular, she finds that these groups are more likely to experience reductions in employment and earnings in a contraction, and more likely to experience gains in these areas in subsequent recoveries, relative to their white and more educated counterparts. Based on this evidence, we allow the impact of local labor market conditions on various measures of health and health behaviors to vary across race and education groups, as discussed in Section 3. Note, however, that individuals whose employment status or prospects are not directly impacted by labor market fluctuations might also experience health effects. Consider the following examples in the context of a labor market expansion. First, for individuals who remain without jobs in an expansion, government programs that provide cash or other in-kind benefits may be less likely to expire or otherwise be curtailed (e.g., unemployment insurance, job training, etc.). On the other end of the spectrum, those with relatively secure employment may be impacted since such individuals may experience improved job mobility or job characteristics (e.g., higher real wages or more generous fringe benefits) in an expansion. Such individuals may also be impacted because of more intense work schedules (e.g., more hours worked per week) associated with an economic expansion. So, while it seems reasonable to expect that labor market conditions might affect the health of individuals whose labor market fortunes are most directly impacted by them, it is also possible that they affect health across a wider range of individuals and that they do so in a heterogeneous fashion. Finally, note that such potential indirect effects of labor market conditions on health (i.e., impacts on health that do not work through explicit changes in employment status) affect our choice of empirical strategy. As described in Section 3, we estimate a reduced form model of the impact of local unemployment rates on various measures of health and health behaviors, rather than a more structural model (e.g., instrumenting endogenous employment status with local unemployment rates) which would miss the kind of indirect impacts on health discussed above. In taking a reduced form strategy, we are consistent with the relevant literature which we describe below. 1.3. Related work While the present work is related conceptually to the literature that investigates the impact of employment status on health, we limit description to those studies that examine the connection between labor market conditions and health. In particular, we describe three recent studies most closely related to ours. In the first rigorous study of its kind, Ruhm (2000) examines the impact of state-level unemployment rates on state-specific measures of total mortality and 10 specific causes of death which account for roughly three-fourths of all deaths in the United States.7 He finds evidence of a countercyclical relationship for total mortality and 8 of the 10 specific causes examined.8 While automobile-related fatalities account for a substantial portion of the impact of changes in state unemployment rates on total mortality, the author finds that preventable causes of death account for an even greater portion of total deaths. Moreover, he also examines age-specific death rates and finds that fatalities among those aged 20–44 are most sensitive to changes in state labor market conditions, consistent with the idea that his estimates are capturing a labor market phenomenon. Finally, he finds that the suicide rate varies directly with the state unemployment rate, suggesting that mental health is procyclical in nature. Second, Ruhm (2001), using data on individuals residing in 31 “large” MSAs from the 1972 to 1981 NHISs finds evidence of a countercyclical relationship between state unemployment rates and several indicators of physical health including medical care utilization (e.g., hospital episodes and doctor visits), unhealthy days (e.g., restricted-activity days and bed days) and whether an individual experienced an acute, but not chronic, medical condition. The author allows the impact of state unemployment rates to vary across certain groups and finds these relationships are most pronounced for males, employed persons and working-aged individuals.9 Finally, consistent with his earlier finding regarding suicide, he finds that non-psychotic mental disorders rise with increases in state unemployment rates and concludes that this represents “some evidence that mental health is procyclical.” Of the studies described, Ruhm (2005) is most relevant to the present work because of the generally greater overlap in outcomes examined.10 Using data on individuals aged 18 and older from the 1987 to 2000 waves of the Behavioral Risk Factor Surveillance Survey (BRFSS), the author finds countercyclical relationships between state unemployment rates and several health behaviors. In particular, he finds systematic relationships for smoking, physical inactivity and weight-related health. Consistent with Ruhm (2001), he finds that these relationships are, generally speaking, most pronounced for males and employed persons. Finally, the author presents evidence that suggests the impacts are considerably larger in the first half of the period in question. In particular, estimates from models that include only observations for the years 1987–1994 are considerably larger in magnitude than estimates that include all years. This is especially true in models that examine smoking behavior and obesity. In general, the preceding work suggests opposite effects of economic conditions on physical and mental health. In particular, while there is relatively less evidence related to mental health, what is available suggests that mental health is procyclical while physical health, and the health behaviors that may influence it, are countercyclical.

In this paper, we present systematic evidence of procyclical relationships for weight-related health and, especially, mental health for a sample of men living in the 58 largest MSAs in the U.S. We find these relationships are most pronounced for groups previously found to be most affected by changing labor market conditions. In particular, our evidence is most consistent for those least likely to be employed and African-Americans. As discussed, our findings have similarities to and differences with existing work, which suggests that physical health is countercyclical in nature while mental health is procyclical. Given the consistency of the findings regarding mental health, a deeper understanding of the long-run implications of changes in mental health is appropriate. For example, how long does the detrimental effect of worsening labor market conditions reduce mental health? Moreover, do such changes in mental health have implications for longer-run physical health or other non-health issues such as homelessness, illegal drug use or criminal activity, all of which have been linked to mental illness? To the extent possible, such extensions should be examined in future related work.