Abstract
Objective. Since physical inactivity has been identified as a pandemic and a public health priority, it is crucial to understand the role of adverse economic shocks on physical activity. In this study, we examine the impact of job losses during the U.S. Great Recession from January 2008 to June 2009 on the likelihood of physical activities of young adults. Methods. We use individual fixed effects estimation on a nationally representative longitudinal data from 2005 to 2015, the Panel Study of Income Dynamics (PSID), to examine the impact of job losses of young adults, their spouses, and their parents on physical exercise of young adults aged 18 to 27 years. Results. Own job losses during the Great Recession led to a decrease in the likelihood of physical exercise among young adults. However, job losses of parents and partners had no effect on the likelihood of young adults’ physical exercise. Conclusion. Our findings indicate a negative impact of the recession on physical activity of young adults and highlights the need for policymakers to consider the impact of major economic downturns on the physical activity of young adults.
“. . . it is crucial to understand how changes in socioeconomic environments, . . ., affect individual decisions regarding physical activity.”
Insufficient physical activity, with its wide prevalence and adverse health consequences, has been identified as a pandemic and a public health priority. 1 As countries respond to the growing calls for increased implementation of national programs and policies to increase physical exercise, it is crucial to understand how changes in socioeconomic environments, such as adverse economic shocks, affect individual decisions regarding physical activity.
As the United States tries to recover from the COVID-19-induced recession, it will be useful to understand how the last major recession affected the physical activity of young adults. Therefore, in this study, we examine the impact of job losses during the Great Recession on the likelihood of physical exercise of young adults in the United States. The Great Recession, which occurred from December 2007 to June 2009, was the longest and the most severe recession in the United States in the prior 80 years, with a disproportionately high unemployment rate among young adults of 19% in 2009.2,3 For context, the last major recession before the Great Recession, which resulted from the dot-com bubble in 2001 led to an unemployment rate of 13% among young adults. This demonstrates the substantially greater impact that the Great Recession had among young adults. 3
There have been few studies that have examined the impact of economic downturns on physical activity but with conflicting results on the direction of the effect.4-7 Own job losses may negatively affect young individuals’ health through different channels. The stress from the loss of income and an uncertain future following a job loss can lead to reductions in physical exercise and negatively affect physical health. 8 On the other hand, job losses can lead to an increase in leisure-time physical activity that consequently positively affects physical health.7,9,10 Given the mixed findings in the literature on economic downturns, the likely effect of the Great Recession on physical exercise is unclear.
These studies also suffer from limitations arising from their use of aggregated area level conditions, such as unemployment rates, as independent variables instead of using individual level job loss identifiers. This makes it difficult to interpret whether own job displacement of an individual, and consequently the loss of income and the greater availability of time, affects their physical activity or whether high unemployment in an area causes employed people to work longer due to fears of future job loss. While a study by Xu 11 adopts a structural approach to overcome this limitation, their analysis is restricted to a sample of males aged 25 to 55 years with some college education or less.
Our study focuses on all young adults between the ages of 18 and 27 years. Youth is a critical phase of development that is associated with maturation and intense learning, along with psychological, social, and economic changes, 12 making it necessary to carefully examine the health impacts of the recession during this period of transition from childhood to adulthood. Moreover, a lack of physical exercise can lead to serious adverse consequences on the health, social, and economic outcomes of young adults13,14 and is also associated with a higher incidence of chronic diseases in adulthood. 15
Another motivation for our interest in this age group stems from the fact that the young adults can be head of households or be living with their parents and the effects of the recession might be very different for these 2 groups. Parents are likely to be the primary income earners, and their job loss may indirectly affect young adults through some additional channels, such as transmission of the emotional and financial stress of parents into the lives of their children, increased responsibilities of young adults, and a consequent change in their physical exercises.16-18
To have a comprehensive understanding of the effect on physical exercise, we analyze the effect of job losses of different individuals in a family on the physical activity of young adults. Specifically, we examine the effect of a young adult’s job loss on his or her likelihood of physical exercise. We additionally examine the effect of spousal job loss on the physical exercise of young adults who have started their own families. For young adults living with their parents, we separately examine the effect of parental job loss and young adult’s own job loss during the Great Recession on their physical exercise.
Methods
Data and Variables
The data used in this article come from the Panel Study of Income Dynamics (PSID). This survey was conducted annually from 1968 through 1997 and biennially thereafter. The PSID is a nationally representative survey using the 1968 US population and it has continuously tracked those individuals and their children over the years. As members of original households grow up and form their own economically independent households, they are interviewed separately, increasing the overall number of interviews conducted in each wave. While it primarily surveys the household head and his or her partner, the PSID started collecting additional data on young adults from 2005 onward through the Youth’s Transition to Adulthood (TA) supplement. This supplement collects data for young adults aged 18 to 27 years. We use employment and health-related variables from the 6 consecutive survey rounds from 2005 to 2015. We have a total sample of 624 young adults living by themselves and 934 young adults livings with their parents. Since the PSID is conducted by the Survey Research Center at the University of Michigan and is made available for researchers without any individual identifiers, this study is exempted from institutional review board approval. The study participants provided verbal informed consent to the surveyors (Survey Research Center at the University of Michigan).
The PSID is particularly suited for this study because it includes questions on whether an individual stopped working at a job during the survey year or the year prior to the survey and their reason for doing so. We define a person to have suffered from a job displacement/loss if they reported any of the following reasons for stopping work: “laid off,” “fired,” “company folded/changed hands/moved out of town,” and “employer died/went out of business.” Other reasons reported by individuals that are not part of our job displacement definition are strikes, quitting the job, retiring from employment, and the job being a temporary/contractual job. Job losses in each round were assigned a value of 1 or was 0 otherwise. We use this definition of job loss following the methodology of several prior studies19, 20 that examined the impact of job loss using the PSID data.
If young adults are household heads, we have job loss data for them and their partners. For households where young adults are living with their parents, we have job loss data of the parents and the young adult. While we do not have job loss data of partners of individuals living with their parents, only 0.5% of these young adults have a partner.
We define job losses during the recession as those occurring between January 2008 to December 2009, as reported in the 2009 survey round. We exclude December 2007 from our analysis because no job losses were reported for that month in our sample. We also examine the impact of job losses during a non-recessionary period, as reported in the survey rounds of 2005, 2007, 2011, 2013, and 2015.
Young adults were asked the frequency with which they do physical exercise. More specifically, they were asked: How often do you do vigorous/moderate/light physical activities for at least 10 minutes that cause heavy/light sweating or large/slight to moderate increases in breathing or heart rate? Based on individual reports, we create a dummy variable to indicate whether an individual exercises at least once a month.
We use PSID’s household level data on the total wealth, the total income, homeownership, amount of rent paid, and the number of household members as control variables for our estimations. We also use individual level controls from the PSID, such as age, gender, education, race, marital status, and religion of the young adult, where applicable. We supplement the PSID job loss data with the annual state unemployment data for the survey years from the Bureau of Labor Statistics. 21
Statistical Analysis
We examine the impact of job losses in the year of the survey and the year prior to the survey on physical activity at the time of the survey. More specifically, we regress the outcome variable on the job loss indicators to estimate the effect of job losses on the likelihood of physical exercise by young adults. To distinguish the effect of job losses during the Great Recession, we separate job losses into 2 sets of variables: (a) job losses during the recession and (b) job losses during nonrecessionary survey rounds. We use separate dummy variables to represent job losses for each type of individual: young adult, partner of young adult, father and mother of young adult.
This article uses a linear probability model. We use the statistical model known as the individual fixed-effects model. This model provides more robust estimates by comparing the same individual over time to control for time-invariant characteristics associated with the individual such as gender, race, religion, constant preferences, household and area characteristics, and other time-invariant factors. This estimation strategy addresses concerns on unobserved heterogeneity arising from the correlation between these time-invariant factors and both job loss and physical activity that can bias the results. Therefore, this individual fixed-effects model provides more precise and unbiased estimates when compared with standard ordinary least squares (OLS) estimation.
We also use the rich set of variables available in the PSID to account for potential time-varying household and individual level characteristics that may bias the estimations. To ensure that the controls are not biased by the job displacement variables, we use these controls from the prior survey rounds, that is, rounds 2003 to 2013. As state level economic conditions can also affect individual decisions, we also control for annual state unemployment rate for the survey years. We additionally use a set of dummy variables representing the 6 survey rounds. We use the month of interview to control for seasonal variations. Standard errors are clustered at the household level and all estimations are weighted to ensure national representativeness. All analyses were conducted using Stata 14 (StataCorp LP).
Results
Table 1 shows the average job loss rates during the recessionary and nonrecessionary periods for the 4 groups of individuals of interest in this study. Young adults who are household heads experienced more job losses during the recessionary period (16.7%) compared with nonrecessionary period (10.8%), and the difference is statistically significant. However, we observe a small difference in partner’s job loss (3.5% vs 4.6%). Young adults living with their parents also suffered significantly more job losses during the recession compared to nonrecessionary period (14.8% vs 10%), but we find small differences for father’s (3% vs 4.5%) and mother’s job loss (4.3% vs 3.8%).
Summary Statistics of Job Loss and Physical Exercise Variables. a
Recessionary period represents the 2009 survey round. Nonrecessionary period represents the other 5 survey rounds.
Indicates statistical significance at 10% level.
Indicates statistical significance at 5% level.
Indicates statistical significance at 1% level.
To provide a sense of individual and household characteristics of young adults who are living by themselves and who are living with their parents, Table 2 presents the summary statistics of the dependent variable and the controls variables for young adults who are household heads (column 1) and the ones who are living with their parents (column 2). Not surprisingly, we find that households where young adults are living with parents are wealthier and have higher income than households where young adults are household heads.
Summary Statistics of Characteristics of Young Adults.
The averages are over all 6 survey rounds. Standard deviations are in parentheses.
In Table 3, we present the results of the impact of job losses on the physical exercise of young adults living by themselves. The OLS coefficients in column 1 shows that job losses do not have a significant effect on the likelihood of doing physical exercise. However, this estimation does not account for individual or household level unobserved heterogeneity. To control for time-invariant factors that may be correlated with both job loss and health behavior, we use an individual fixed-effects model in column 2. With the fixed-effects model, we find that the own job loss of young adults during the recession reduces the likelihood of physical exercise statistically significantly by 6.3 percentage points (95% CI = −0.138 to 0.012). Job losses of spouses or partners during the recession do not have a statistically significant effect on the likelihood of physical exercise. Similarly, job losses in non-recessionary periods do not have a significant effect.
Impact of Job Loss on the Likelihood of Doing Physical Exercise Each Month (95% CI). a
Abbreviations: OLS, ordinary least squares; FE, fixed effects.
For OLS estimation in columns 1 and 3 we use some additional controls such as age, gender, race, religion, and completion high school, in the absence of individual fixed effects. In Columns 2 and 4 we use linear model with individual fixed effects. All models include controls for survey round and interview month dummies.
Indicates statistical significance at 10% level.
Indicates statistical significance at 5% level.
Indicates statistical significance at 1% level.
We next examine the impact of job losses on the physical activity of young adults who are living with their parents. We present both the OLS and fixed effects results in columns 3 and 4 of Table 3, respectively. In the fixed effects model we find that neither the young adults’ own job loss nor the parental job loss has a significant impact on the likelihood of physical exercise.
Robustness Checks
The causal interpretation of our estimations is based on the assumption that the Great Recession was unanticipated, and therefore, job losses during the recession were not foreseen. If individuals were able to anticipate their job loss, they may have changed their health behavior in response to the expected job loss, consequently biasing our results. We examine whether individuals anticipated their own job losses by reversing the main estimation, that is, by regressing the jobs loss indicator on the physical exercise indicator. As reported in columns 1 and 2 of Table 4, there is no evidence of physical exercise significantly predicting job loss for young adults living by themselves and those living with their parents. We also test whether income level in prior round predicts job loss in the current survey round, since prior research suggests that individuals may be able to anticipate job loss or layoffs because workers sometimes experience fewer hours of work, and consequently lower pay, long before the actual job loss. 19 In column 3 of Table 4, we show that prior income level does not predict job loss.
Robustness Checks.
We use individual fixed effects. All models include controls for survey round and interview month dummies. Because of data unavailability, we cannot replicate column 3 for young adults who are living with their parents.
Indicates statistical significance at 10% level.
Indicates statistical significance at 5% level.
Indicates statistical significance at 1% level.
We conducted several other tests to check the validity of our results. First, we reran our analysis using alternative definitions for the outcome variable. Instead of an indicator variable for any exercise once a month, we used dummy variables for light and vigorous exercise over various time intervals. The results continue to hold for vigorous exercise but not for light exercise. Second, we find that the results remain unchanged when we control for the health status of the young adult in our main regressions. We do this to ensure that the possibility of health problems biasing our results by affecting the job loss and the physical activity decisions. Third, we find that our results remain robust to the inclusion of interactions between a linear time trend and different individual and household level characteristics. This indicates that the results are not driven by a time trend that primarily affects individuals who are more likely to experience a job loss.
Discussion
We find that job losses of young adults living by themselves during the Great Recession led to a significant reduction in the likelihood of physical exercise. Given that these young adults are the primary income earners of their households, the disproportionately high unemployment rates during the recession likely causes greater stress among them, which then affects their physical health, as suggested in prior studies.9,22 The fact that job losses during nonrecessionary period do not have a significant effect on physical exercise also hints at this underlying reason. Another reason behind the reduction in physical exercise can be the increase in time spent by young adults looking for jobs. However, evidence from time-use data does not support this, especially for high school graduates. 23
We also find that parental job losses have no impact the likelihood of physical exercise of young adults living with their parents. This happens likely because the young adults are not directly exposed to the stress and uncertainty of losing own job when parents lose their jobs and consequently, have no reason to modify their behavior. Job losses of young adults living with their parents also do not affect their physical exercise, perhaps because their income stream is not as essential for the household’s survival as the parents are likely the primary income earners for the households.
Prior studies provide evidence of both positive and negative effects on health status and health behaviors such as smoking, drinking, drug use, and diet,24-29 in general, as well as on physical activity of adults.4-7,11 This article is the first study that examines how job losses specifically during the Great Recession affected the physical activity of young adults in the United States. In our knowledge, it is also the first to examine the effect of parental and partner’s job losses on physical health behavior of young adults. Overall, this study adds to the limited literature on the detrimental effects of the Great Recession on the youth and help shed important light on how the recession affected the health behavior of young adults.
Limitations
There are a few limitations of our study. We are unable to explore the channels through which job loss affects health behavior because of the absence of data. For the same reason, we analyze only the short-term impacts of the recession of 2008. Further research is needed to understand how recessions can impact the physical health of young adults, including the analysis of more indicators of health.
Another limitation is the lack of data on the duration of physical exercise. PSID asks about the frequency of exercising for at least 10 minutes and since it is the best available dataset for our research, we are compelled to restrict our analysis to the use of this outcome variable. Future research should consider using more standard measure of physical activity, such as exercising 75 minutes per week, for their analysis.
Conclusion
Despite these limitations, this study highlights the need for researchers and policymakers to consider the impact of major economic downturns on the physical activity of young adults, as countries race to reduce physical inactivity by 10% by 2030. 30 Our findings indicate that policies aimed at improving labor market conditions that do not account for the health-related costs of job loss are not sufficient. Moreover, focusing on the health of young adults is particularly important since ignoring this could lead to increases in future health expenditures along with decreases in future productivity. Policy assistance should not exclude the young generation, particularly those living independently, as has been common due to the belief that job loss is not a big concern for the youth. Individuals who were young adults during the Great Recession now form the millennial generation, the largest generation in the US labor force, 31 and their importance in the economy cannot be emphasized enough.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Ethical Approval
Since the PSID is conducted by the Survey Research Center at the University of Michigan and is made available for researchers without any individual identifiers, this study is exempted from institutional review board approval.
Informed Consent
The study participants provided verbal informed consent to the surveyors (Survey Research Center at the University of Michigan).
Trial Registration
Not applicable.
