The business cycle and mortality: Urban versus rural counties
Introduction
The association between the business cycle and mortality has been extensively studied. Such studies include Ruhm, 2000, Ruhm, 2015 for the United States, Neumayer (2004) for Germany, Tapia Granados (2005) for Spain, Gonzalez and Quast (2010) for Mexico, Ariizumi and Schirle (2012) for Canada, Lin (2009) for Pacific Asian countries, and Gerdtham and Ruhm (2006) for OECD countries. Using unemployment as a proxy for the business cycle, these studies report a pro-cyclical pattern of mortality at the state or national level. Mortality falls when unemployment is high, a claim first reported over 90 years ago (Ogburn and Thomas, 1922). Such a finding, however, is not universal. Brenner, 1973, Brenner, 1975, Brenner, 1979 finds a countercyclical association. Moreover, many studies that use either family-level data (Strully, 2009) or individual-level data (Halliday, 2014, Sullivan and Von Wachter, 2009, Gerdtham and Johannesson, 2005) find a countercyclical pattern with mortality rates rising during recessions.
How could mortality be pro-cyclical? For one, the opportunity cost of going to the doctor, of exercising, and taking time to eat healthy is, presumably, higher during expansions than during recessions. Alternatively, people might push themselves harder during expansions such as work overtime or work multiple jobs. Such activities could cause more stress or allow them to become more susceptible to disease. During expansions people become wealthier and that might encourage them to take on risky activities such as excessive drinking or driving more recklessly thereby increasing fatality rates (Ruhm, 1995). In all of these cases, people's behavior changes across the business cycle and such changes hold ramifications for health and mortality. Rising pollution or changes in the quality of medical care could also play roles as factors external to an individual's behavior.
When examining the U.S., the typical approach is to consider state-level variations in unemployment and mortality which is the approach first taken by Ruhm (2000) although Ruhm (2015) and Lindo (2015) use both state and county-level data. We, instead, employ county-level data as do Fontenla et al. (2011). County-level data holds both advantages and disadvantages over state-level data. The degree of within-county variation is likely to be smaller than within-state variation allowing for less heterogeneity within the unit of analysis. Moreover, a greater number of observations can increase the power of statistical tests. On the other hand, larger units of analysis are likely to better filter out random errors since one is averaging over larger units. Pierce and Denison (2006) identify reporting errors from Texas using county-level data. People are also more likely to migrate and commute across county lines as opposed to state lines. See Lindo (2015) for further discussion.
A second reason to conduct a county-level analysis is that it can allow us to better understand what could be driving previous results by uncovering differences across heterogeneous settings, in our case urban versus rural ones. Examining such differences across settings could help to offer explanations as to why mortality is countercyclical. For example, one reason is that the opportunity cost of going to a doctor or seeking medical treatment is relatively high during economic booms as people might find it costly to take time off from work. These opportunity costs could differ between urban and rural settings, especially if one from a rural area needs to travel long distances to receive medical care or see a specialist. If true, then the pro-cyclical association between mortality and unemployment should be stronger in rural areas. On the other hand, to the extent that stress contributes to mortality, that stress levels are higher in urban areas, and that stress is higher during expansions then the association between mortality and the business cycle should be stronger in urban areas. To the extent that pollution rises during economic booms thereby contributing to mortality, then associations should be stronger in urban areas where pollution levels are higher. In fact, Davis et al. (2010) find that emissions of particulate matter from trucking in New Jersey were higher during economic booms. Heutel and Ruhm (2013) find evidence at the state level that lower air pollution during recessions provides a partial explanation for why mortality is pro-cyclical (although Sameem and Sylwester (2016) find little evidence that pollution is what drives the pro-cyclicality of mortality).
Thinking of reasons why the overall mortality rate as well as mortality for specific types of death could differ between urban and rural areas is not difficult. As just suggested, more air pollution in cities could contribute to respiratory and related problems (Calderon-Garciduenas et al., 2015, Zhou et al., 2015, Heutel and Ruhm, 2013), especially in infants (Currie and Schmieder, 2009, Foster et al., 2009, Currie and Neidell, 2005, Chay and Greenstone, 2003). Similarly, the higher number of vehicles in metropolitan areas adds to traffic accidents and motor vehicle fatalities (French and Gumus, 2014). In this paper, we consider whether associations between mortality and the business cycle also differ between urban and rural areas. We find substantial differences in mortality rates between urban and rural settings, especially for women and the elderly. We also find significant differences regarding deaths due to heart disease as these deaths are more pro-cyclical in urban areas. External causes of death such as accidents are found to be more pro-cyclical in rural counties.
This analysis could be especially enlightening when comparing findings from individual-level studies that often find that being unemployed raises mortality for individuals. See Winkleman and Winkleman, 1998, Burgard et al., 2007, Sullivan and Von Wachter, 2009, Strully, 2009 and Tapia Granados et al. (2014). Job loss can be associated with depression, greater risks of disease, and deviant behaviors that diminish health and income thereby increasing mortality. An explanation to reconcile these contrasting views is that relatively few people become unemployed during a recession as an increase in the unemployment rate from 5% to 9%, for example, still only directly impacts a minority of the labor force. So even if the newfound unemployed suffer greater mortality, overall mortality could still decrease if the slowing economy lowers pollution levels (which affects all residents) or lowers stress at work (for the majority who remain employed) as people find themselves less busy. Therefore, examining differences between rural and urban areas can help narrow explanations for the macroeconomic associations reported above.
The remainder of the paper is organized as follows: Section 2 describes the data and section 3 presents the methodology. Section 4 provides results and Section 5 concludes.
Section snippets
Data
Our sample spans the 24 years from 1990 to 2013 and includes three recessions:1990–91, 2001, and 2007–09. Data comes from the Bureau of Labor Statistics and the Compact Mortality Files (CMF) of the National Center of Health Statistics. Data on unemployment before 1990 is not compatible with subsequent data and the BLS cautions against using them together. The unemployment rate we use corresponds to U-3 (the official unemployment rate) and is calculated as the number of unemployed people as a
Methodology
To analyze the impact of cyclical fluctuations upon mortalities across urban and rural counties, we relate the natural log of mortality rate for the th type of mortality in county at time () to the natural log of the annual county unemployment rate () and several county-year demographic control variables () along with time-invariant county fixed effects (), county-invariant time fixed effects () and an error term (). Use of natural logs allows one to interpret coefficient
Baseline results
Table 2 presents the estimates of equation (1) for all counties (column 1) and then for rural and urban counties separately (columns 2 and 3, respectively). We concur with previous findings that report a negative association between unemployment and mortality. Nevertheless, we also find distinctions as to the strength of these associations. The coefficient on the unemployment rate is 75% higher for urban counties, −0.014, than it is for rural ones, −0.008. Also of note is that the coefficients
Conclusion
Using U.S. county-level data from 1990 to 2013, this study provides a nuanced story to the recent findings of the pro-cyclical behavior of mortality. We find that negative associations between unemployment and mortality are stronger in urban counties. In some cases including women and the elderly, the coefficients differ by more than a factor of 3. Therefore, our main conclusion is that the association between unemployment and mortality found by others appears to more strongly hold in urban
References (42)
- et al.
Are recessions really good for your health? evidence from Canada
Soc. Sci. Med.
(2012) - et al.
Megacities air pollution problems: Mexico city metropolitan area critical issues on the central nervous system pediatric impact
Env. Res.
(2015) - et al.
Macroeconomic fluctuations and motorcycle fatalities in the US
Soc. Sci. Med.
(2014) - et al.
Deaths rise in good economic times: evidence from the OECD
Econ. Hum. Bio
(2006) - et al.
Business cycles and mortality: results from Swedish microdata
Soc. Sci. Med.
(2005) - et al.
Mortality and business cycles by level of development: evidence from Mexico
Soc. Sci. Med.
(2010) Unemployment and mortality: evidence from PSID
Soc. Sci. Med.
(2014)Aggregation and the estimated effects of economic conditions on health
J. Health Econ
(2015)Recessions lower (some) mortality rates
Soc. Sci. Med.
(2004)Economic conditions and alcohol problems
J. Health Econ
(1995)
Recessions, healthy no more?
J. Health Econ
The associations between ambient air pollution and adult respiratory mortality in 32 major Chinese cities, 2006-2010
Env. Res.
Comparability of cause of death between ICD-9 and ICD-10: preliminary estimates
Natl. Vital Stat. Rep.
Variation in Physician Office Visit Rates by Patient Characteristics and State, 2012
The urban-rural disparity in nursing home quality indicators: the case of facility-acquired contractures
Health Serv. Res.
Mental Illness and the Economy
Trend in alcohol consumption and associated illnesses: some effects of economic changes
Amer. J. Pub. Health
Mortality and the national economy
Lancet
Utilization of Ambulatory Medical Care by Women: United States, 1997-98
Toward a better estimation of the effect of job loss on health
J. Health Soc. Beh
Occupational Heart Disease
Cited by (25)
Native American “deaths of despair” and economic conditions
2024, Research in Social Stratification and MobilityThe spatially and temporally varying association between mental illness and substance use mortality and unemployment: A Bayesian analysis in the contiguous United States, 2001–2014
2022, Applied GeographyCitation Excerpt :Populations in different areas might respond differently to unemployment due to variations in local contexts (Shoff et al., 2012; Trgovac et al., 2015). A spatially explicit approach is thus needed to explore the potential spatial disparity within the relationship (Heutel & Ruhm, 2016; Sameem & Sylwester, 2017). With the research opportunities posed by the literature in mind, we examine two research questions.
Death on the job: The Great Recession and work-related traffic fatalities
2021, Social Science and MedicineCitation Excerpt :Even the estimate for the overall fatality rate in rural crashes in column (1) is small in magnitude and marginally significant. This finding is consistent with that of Sameem and Sylwester (2017) who report that motor vehicle crashes are more procyclical in urban counties compared to rural ones. The explanation may be related to the fact that working individuals in rural areas are more likely to be engaged in agriculture, which is less sensitive to fluctuations in the aggregate economy (e.g., Da-Rocha and Restuccia, 2006).
Are U.S. lung cancer mortality rates converging?
2020, Quarterly Review of Economics and FinanceCitation Excerpt :Therefore, the present study also looks at the convergence of lung cancer mortality across the states located on east and west coasts and those located elsewhere. Last but not the least, cancer deaths are one of the common exceptions in the literature on the pro-cyclical pattern of mortality, that is, many major types of mortality rates have been found to be declining during recessionary periods, but cancer deaths appear to be unrelated to transitory cyclical fluctuations in the economy (Ariizumi & Schirle, 2012; Gonzalez & Quast, 2010; Gerdtham & Ruhm, 2006; Granados, 2005; Heutel & Ruhm, 2016; Neumayer, 2004; Ruhm, 2015, 2000; Sameem & Sylwester, 2017). This study looks at the long term trends of lung cancer mortality which is a major share of all cancer deaths, hence, it could possibly fill in the gap in this literature.
Neighborhood social and economic change and diabetes incidence: The HeartHealthyHoods study
2019, Health and PlaceCitation Excerpt :In particular, mortality increases with short term economic growth, an effect that is independent of individual-level change in economic conditions (Tapia Granados et al., 2014). The association between economic growth and cardiovascular mortality is three times stronger in urban environments compared to rural areas (Sameem and Sylwester, 2017). A potential mechanism for the association between economic growth and mortality association is that rapid change in urban areas is associated with increasing cardiovascular risk factors, including diabetes.