Abstract
Introduction:
This study explored the associations between midlife obesity and an array of common financial stressors related to wealth loss, debt, and bankruptcy.
Methods:
The analysis was conducted in 2022 with data from the National Longitudinal Survey of Youth 1979 (NLSY-79). The final sample included 36,122 observations on 8,059 respondents as they aged from 31–39 in 1996 to 51–59 in 2016. Associations between obesity and financial stressors were estimated with logistic regression models adjusting for a comprehensive set of relevant control variables.
Results:
The analysis found significant relationships between obesity and multiple types of financial stressors, including property debt, unsecured debt, and bankruptcy. Property debt and unsecured debt increased the odds of obesity by 29% and 20%, respectively, and bankruptcy increased the odds of obesity by 43%. Average Marginal Effects (AMEs) in combination with model fit statistics confirmed that these stressors—unsecured debt, property debt, and bankruptcy—were important financial correlates of midlife obesity among the NLSY79 cohort.
Conclusions:
The financial correlates of obesity included multiple financial stressors, but the magnitude of associations varied substantially across types of financial stressors. Results suggest that future interventions aimed at reducing obesity disparities should target populations with high levels of debt and bankruptcy.
INTRODUCTION
In response to concerning trends in the economic well-being of aging adults, researchers across social and health sciences have sought to better understand the impact of financial stressors on population health.1–5 Financial stressors include a range of negative financial events such as large wealth losses (or “wealth shocks”), indebtedness (negative net worth or simply debt), and bankruptcy.5 Current research on financial stressors generally provides evidence of a negative association between financial stressors and health,6–9 but less is known about the associations between financial stressors and midlife obesity.
A prominent explanation for the association between financial stressors and adult health emphasizes the impact of stress on the body, which has implications for midlife obesity.2,6,10 Stressful experiences or social conditions (stressors) “get under the skin” by triggering a hormonal response (stress) capable of increasing one’s chances of health problems and obesity.11–13 Stress is a known catalyst of unhealthy behaviors with established links to obesity, e.g., problematic alcohol use and “emotional” eating.14–16 Given the established links between financial well-being, stress, and weight gain, a growing body of research has begun examining the financial correlates of obesity.5,17–19 This study extends this line of research by identifying which specific financial stressors have the largest and most robust associations with midlife obesity. To this end, the analysis explored variation in the associations between common types of financial stressors and midlife obesity in the the National Longitudinal Surveys of Youth 1979 (NLSY9). The NLSY79 is a cohort of Baby Boomers, born 1957 to 1964,20 who transitioned into middle adulthood during the Great Recession of 2007–2009. This cohort thus experienced multiple types of wealth loss, debts, and bankruptcies with relationships to stress and poor health outcomes in later adulthood.
Wealth shocks denote large drops in net worth, the value of one’s total assets minus their total debts. Wealth shocks have significant associations with mental health and mortality risk.3–5 Wealth loss, however, can occur within any one of the components of net worth. For example, housing-related financial problems such as a large drop in the value of one’s primary residence can be a stressful experience with multiple implications for individuals’ later financial, emotional, and physical well-being.1,21,22 Although homeownership is the most common source of wealth, one’s net worth can also drop due to a loss in one’s savings or in the value of other properties such as vehicles, real estate (beyond one’s primary residence), and businesses or farms. Recent studies suggest that these components of net worth, as well as large drops in their value, can have independent associations with health and longevity,17,23,24 but research has not considered whether this finding extends to obesity.
Debt is not necessarily bad for health,25 but among midlife and older adults, the stress of debt—especially indebtedness—is consistently associated with mental and physical health problems.8,9,26,27 Debt can be disaggregated by its relationship to asset ownership.28 Secured debts are linked to material assets such as real estate or vehicles, whereas unsecured (or noncollateralized) debts are not tied to any physical asset. In general, secured debts are related to financial investments such as homeownership and have lower interest rates than unsecured debts, but even secured debts can become financial stressors under certain conditions. Small business owners, for example, can easily find themselves with large amounts of debt while trying to keep a struggling business afloat. Unsecured debt is typically created by, e.g., credit card use, medical expenses, or high-interest loans used to supplement low wages. In other words, unsecured debt typically denotes a financial liability and a future drain on resources among those with few resources to spare.6,29
Bankruptcy offers debt relief but at an apparent price for health.19,30 Filing for bankruptcy prohibits creditors from collecting outstanding debts, but filers generally find that their financial troubles continue. Bankruptcy carries stigma in both social and financial networks, highlighting a person’s previous financial problems and casting doubt on their “credit worthiness.”31 As a result, bankruptcy can cause problems in the housing and labor markets, often leaving those who file for bankruptcy with many of the same financial strains that the bankruptcy process was supposed to alleviate.7 Multiple types of bankruptcy exist, with debtors generally filing Chapter 7 or Chapter 13 bankruptcy.30 Chapter 7 bankruptcies lead to an immediate repossession of assets, and Chapter 13 bankruptcies reorganize debt into a repayment plan wage-earning debtors can afford to pay off over time. Paying one’s debt down and holding onto assets might reduce stress and lower anxiety or potentially become a recurring strain on limited resources.7
In summary, this study extends research by identifying which specific financial stressors are associated with obesity among a cohort of Baby Boomers entering midlife. Given the established relationships of multiple financial stressors with poor health and obesity, the analysis considered an arrary of stressors related to wealth loss, debt, and bankruptcy. Althouh this analysis is exploratory in nature, mapping the relative associations between multiple financial stressors and obesity provides useful information for future research and public policies, e.g., future attempts to reduce obesities disparities should target populations with high levels of debt and bankruptcy.
METHODS
Study Sample
To assess the association between financial stressors and health, data from the National Longitudinal Survey of Youth 1979 (NLSY79) was used. The NLSY79 is a nationally representative, longitudinal study of U.S. adults that began in 1979 with 12,686 respondents between the ages of 14 and 22. The NLS is a program of the U.S. Bureau of Labor Statistics with data collection provided by the Center for Human Resource Research, a multidisciplinary research organization affiliated with The Ohio State University. The NLS data are available to the public and can be downloaded for free at www.nlsinfo.org.
The analysis only used the 1996, 2000, 2004, 2008, 2012, and 2016 survey years for 2 reasons. First, 1996 was the baseline because it represents the first year of the NLS in which all respondents were in their 30s (31–39 years old). Second, the analysis relied on 4-year intervals because the NLSY79 began collecting wealth data every 4 years in 2000. The analysis excluded the 1998 survey to ensure that results were not an artifact of change in the interval length between measurements of wealth.
The final analysis sample included 36,122 observations of 8,059 NLSY79 respondents. In 1990, the NLS dropped 3 supplemental samples from data collection (N=2,700), reducing the total sample from 12,686 to 9,986 respondents. To arrive at the final analysis sample, another 1,927 respondents were dropped due to missingness (N=1,065) or attrition (N=862). To ensure that listwise deletion did not substantively impact results, a supplemental analysis was conducted using respondents’ BMI as the outcome with Full Information Maximum Likelihood linear regression.32,33 This supplemental analysis with 54,744 observations of 9,124 respondents (862 respondents who did not participate in any survey from 1996 to 2016 were dropped) yielded the same overall conclusions as the analysis of obesity presented here.
Measures
Obesity was constructed by taking the most recently reported height and respondents’ self-reported weight in each survey to create each respondents’ BMI (kg/m2), with BMI ≥ 30 kg/m2 coded as a 1 and all others 0. Although measured height and weight are preferable, research indicates that self-reports are a reasonable proxy for measured height and weight.34,35 The BMI of women was set to missing if they reported a pregnancy.
Wealth shocks included large drops in net worth and the value of assets including savings, housing, and other properties. The NLSY79 collected information on the market value of respondents’ homes; savings, stocks, investments, and bonds; businesses and farms; and vehicles; in addition to the value of debts related to homes, cars, real estate, and businesses. Net worth denotes the total value of assets minus total debts. Savings refers to the total amounts in savings and checking accounts. Property wealth denotes the total market value of vehicles, real estate (other than primary residence), businesses, and farms. Shocks denote a 75% or larger drop in assets since respondent’s last survey (yes=1, no=0).3,4 For example, if a respondent reported a 75% drop in the value of their savings from their 1992 to 1996 interview, they would be given a 1, indicating a savings shock.
Debts included indebted, housing debt, property debt, and unsecured debt. Indebtedness was operationalized using information collected from respondents about their financial situation if they were to pay off all outstanding debts. Respondents could reply that they would have money left over, break-even, or be in debt. With this information, a binary indicator of indebtedness was created by combining those who reported having money left over or breaking even into a single reference group (coded 0).8 Indicators of housing and property debt were created by taking the difference in the property’s market worth and the debt owed on it and creating binary indicators of negative net housing and property wealth. Unsecured debt was measured as a binary indicator of respondents who reported a debt of over $500 due to credit cards, educational loans, or medical debts.
Bankruptcies were operationalized with 3 binary variables that identified whether respondents had filed for any type of bankruptcy, a Chapter 7 bankruptcy, or a Chapter 13 bankruptcy. In 2004 and then from 2008 to 2016, the NLSY79 collected information on respondents’ experiences with bankruptcy. This information includes whether the respondents have ever declared bankruptcy, the type of bankruptcy, and the date it occurred. To operationalize bankruptcy, a binary variable for each year was created in the analysis period that respondents experienced a personal bankruptcy of any type. Likewise, binary indicators were created denoting the type of bankruptcy (Chapter 7 or 13).
The analysis adjusted for several known socioeconomic and sociodemographic correlates of health and financial stressors (Table 1).5,36 Socioeconomic controls included average family income, average wealth, education, employment, job loss, work hours, and health insurance. To normalize the distributions of income and wealth, the analysis used the natural log of a rolling average. Education was measured as a categorical variable denoting respondents with less than 12, 12 years, 13 to 15 years, and 16 or more years of schooling. The analysis also controlled for current employment status and work hours in the previous year (none, part-time, or full-time). Demographic controls included sex (male/female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, or other race/ethnicity), age (in years), and marital status (married, previously married, or never married).
Table 1.
Descriptive Information for Analysis Variables at 1996 Baseline
| Variable names | Description | Min, Max | Mean or % |
|---|---|---|---|
| Outcomes | |||
| Obesity | BMI ≥ 30 kg/m2 is 1, otherwise 0 | 0, 1 | 24% |
| Wealth shocks | |||
| Net worth shock | 75% or larger drop in net worth | 0, 1 | 12% |
| Savings shock | 75% or larger drop in savings | 0, 1 | 21% |
| Housing shock | 75% or larger drop in home wealth | 0, 1 | 5% |
| Property shock | 75% or larger drop in property wealth | 0, 1 | 11% |
| Debts | |||
| Indebted | Negative net worth | 0, 1 | 14% |
| Housing debt | Negative net home wealth | 0, 1 | 2% |
| Property debt | Negative net property wealth | 0, 1 | 6% |
| Unsecured debt | Unsecured debt over $500 | 0, 1 | 37% |
| Bankruptcies | |||
| Bankrupt | Filed for bankruptcy | 0, 1 | 3% |
| Chapter 7 | Filed for a Chapter 7 bankruptcy | 0, 1 | 2% |
| Chapter 13 | Filed for a Chapter 13 bankruptcy | 0, 1 | 1% |
| Socioeconomic controls | |||
| Wealth | Natural log of average wealth | 0, 13.87 | 8.38 |
| Income | Natural log of average of income | 0, 13.79 | 10.24 |
| Education | Completed years of schooling | 0, 3 | |
| Less than 12 years | 13% | ||
| 12 years | 43% | ||
| 13 to 15 years | 24% | ||
| 16 or more years | 20% | ||
| Employed | Currently employed | 0, 1 | 81% |
| Work hours | Hours worked in the previous year | 0, 2 | |
| None | Zero hours worked | 14% | |
| Part-time | Less than 2,080 hours | 29% | |
| Full-time | 2,080 or more hours | 57% | |
| Demographic controls | |||
| Age | Age in years | 31, 39 | 34.55 |
| Female | Women=1, men=0 | 0, 1 | 49% |
| Race/ethnicity | 0, 3 | ||
| White | Non-Hispanic, White | 48% | |
| Black | Non-Hispanic, Black | 29% | |
| Hispanic | 19% | ||
| Other | All other races/ethnicities | 4% | |
| Marital status | 0, 2 | ||
| Married | 57% | ||
| Previously married | Divorced, separated, or widowed | 20% | |
| Never married | 23% |
Statistical Analysis
To correctly identify the most influential financial stressors in terms of adult obesity among the NLSY79 cohort, the analysis examined statistical significance, coefficient magnitude, and model fit. The analysis thus proceeded in 3 steps. First, to test statistical significance, the analysis regressed obesity on each type of wealth loss, debt, and bankruptcy using logistic regression with socioeconomic and demographic controls. To address the non-independence among observations of individuals over time, the analysis estimated the SEs with the clustered sandwich estimator, allowing for intragroup correlation by relaxing the usual requirement that the observations be independent. In other words, the analysis assumes that observations are independent across respondents but not necessarily within respondents over time.37 To further improve the accuracy of parameter estimates, the analysis also included lagged obesity as a control variable in all statistical models.38,39
The second step of analysis used the logistic coefficients from the first step to estimate the Average Marginal Effects (AMEs) of each financial stressor on midlife obesity. AMEs denote the differences in the probability of midlife obesity between those who experienced a particular financial stressor and those who did not experience the stressor, holding all other variables at their observed values in the data. AMEs reveal the relative magnitude of associations between stressors and obesity. Although ORs offer a useful way of presenting patterns in significance, examining predicted probabilities such as AMEs is a critically important step when working with logistic regression.40,41
The third and final step of the analysis compared the differences in model fit between (1) a model of health with controls but no financial stressors to (2) models of health with controls and a single stressor. Model fit is measured with the Akaike Information Criteria (AIC). Lower AIC values indicate better fitting models.42 Model fit was measured with the AIC because the Bayesian Information Criteria (BIC) penalizes model complexity, and the goal of analysis was identifying which financial stressor provided the best improvement in fit regardless of model complexity. All data preparation and statistical estimation was done in Stata 17.43
RESULTS
The analytic sample included 8,059 participants, who contributed a total of 36,122 observations between 1996 and 2016. The characteristics of the sample at baseline (1996) are presented in Table 1. Table 2 shows the percentage of obesity and financial stressors in the sample, by survey year. Obesity ranged from 24% in 1996 to 41% in 2016. Financial stressors related to savings losses and unsecured debt were most common. Nearly a quarter of the sample experienced a savings loss in 1996 and 2012, and approximately half of the sample reported unsecured debts of $500 or more from 2004 to 2016. Even financial stressors that occurred less frequently, such as housing debt and Chapter 13 bankruptcy, were still reported by about 1% of the sample each year.
Table 2.
Percentage of the Sample That Experienced Obesity or Financial Stressors, by Year (N=8,059; Observations=36,122)
| Years | ||||||
|---|---|---|---|---|---|---|
| Variable names | 1996 | 2000 | 2004 | 2008 | 2012 | 2016 |
| Outcome | ||||||
| Obesity | 24% | 30% | 32% | 36% | 39% | 41% |
| Wealth shocks | ||||||
| Net worth shock | 12% | 11% | 15% | 14% | 19% | 13% |
| Savings shock | 21% | 16% | 20% | 21% | 22% | 17% |
| Housing shock | 5% | 6% | 8% | 8% | 9% | 6% |
| Property shock | 11% | 14% | 15% | 12% | 14% | 13% |
| Debts | ||||||
| Indebted | 14% | 11% | 11% | 12% | 15% | 13% |
| Housing debt | 2% | 1% | 1% | 1% | 2% | 1% |
| Property debt | 6% | 6% | 5% | 5% | 4% | 4% |
| Unsecured debt | 37% | 38% | 50% | 53% | 50% | 48% |
| Bankruptcies | ||||||
| Bankrupt | 3% | 4% | 4% | 3% | 3% | 1% |
| Chapter 7 | 2% | 3% | 2% | 2% | 1% | 1% |
| Chapter 13 | 1% | 2% | 2% | 1% | 1% | 1% |
Table 3 presents the ORs of interest from 13 models of obesity, including 11 models that adjust for a single financial stressor and 2 models, Models 12 and 13, that estimate the association between obesity and multiple financial stressors simultaneously. Although not shown, all statistical models included controls for lagged obesity,19,39 age, age squared,44,45 and the socioeconomic and demographic controls listed in Table 1.
Table 3.
ORs From Logistic Regression Models Predicting Obesity, 1996 to 2016 (N=8,059; Observations=36,122)
| Variable | Models | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | |
| Net worth shock | 0.97 (0.04) | 0.96 (0.04) | |||||||||||
| Savings shock | 1.00 (0.03) | 1.00 (0.04) | |||||||||||
| Housing shock | 0.94 (0.05) | 0.94 (0.05) | |||||||||||
| Property shock | 0.95 (0.04) | 0.95 (0.04) | |||||||||||
| Indebted | 1.05 (0.05) | 1.05 (0.05) | |||||||||||
| Housing debt | 1.00 (0.13) | 0.97 (0.13) | |||||||||||
| Property debt | 1.33*** (0.09) | 1.29*** (0.08) | |||||||||||
| Unsecured debt | 1.21*** (0.04) | 1.20*** (0.04) | |||||||||||
| Bankruptcy | 1.43*** (0.11) | 1.43*** (0.11) | |||||||||||
| Chapter 7 | 1.44*** (0.15) | 1.44*** (0.15) | |||||||||||
| Chapter 13 | 1.40** (0.17) | 1.41** (0.17) | |||||||||||
| Model fit | |||||||||||||
| AIC | 33,188 | 33,188 | 33,187 | 33,186 | 33,187 | 33,188 | 33,168 | 33,146 | 33,167 | 33,175 | 33,180 | 33,169 | 33,119 |
| AIC0 – AICM | −2 | −2 | −1 | 0 | −1 | −2 | 19 | 40 | 20 | 11 | 6 | 18 | 67 |
Notes: Boldface indicates statistical significance (*p<0.05; **p<0.01; ***p<0.001 [2-tailed tests]). Robust SEs are in parentheses. Every model included lagged obesity, age, age squared, and the socioeconomic and demographic controls listed in Table 1. AIC0 – AICM denotes the differences between a model that excludes all financial stressors (AIC0) and models with a single financial stressor included (AICM).
After controlling for socioeconomic and demographic variables, a loss of 75% or more in net worth was not significantly associated with an increase in odds of obesity by a factor of 1.06 (p>0.05) in Model 1. This result in terms of significance held for every type of wealth loss, including savings, housing, and property shocks. In terms of debt, only property and unsecured debts were significantly associated with obesity (p<0.001), whereas bankruptcy and its 2 component types, Chapters 7 and 13 bankruptcy, were all significantly related to obesity (p<0.001 for bankruptcy and Chapter 7 and p<0.01 for Chapter 13).
Models 12 and 13 in Table 3 considered multiple stressors in a single model to assess whether any significant associations between financial stressors and obesity were due to a correlation with another financial stressor. In Models 12, the analysis considered overall measures of wealth shock, debt, and bankruptcy in a single model. In Model 13, the analysis the analysis examined the multiple types of wealth shocks, debts, and bankruptcies in a single model. Even after adjusting for other, related financial stressors, the results were the same in terms of statistical significance. In Model 12, bankruptcy remained the only financial stressors related to obesity, and in Model 13, property debt, unsecured debt, and Chapters 7 and 13 were all significantly related to obesity.
Table 3 also includes the AIC statistic for model fit. The analysis used a model to assess the improvement in fit provided by adding each of the varieties of financial stressors to a new model with AICM. The change in fit is denoted by AIC0 – AICM, where AIC0 represents a model without a financial stressor. A difference of 10 or more indicates strong support for AICM, a difference of 4 to 9 provides moderate support for AICM, and a difference of 3 or less indicates little to no improvement in fit from AIC0.42 Unsecured debt provided the largest improvement in fit, a difference of 40. Property debt and bankruptcy provided the next best improvements in fit, a difference of 19 and 20, respectively. Although relatively smaller improvements in fit, the AIC statistics still suggested strong to moderate support for models that included Chapters 7 and 13 bankruptcy.
To compare the magnitude of the associations between financial stressors and obesity, Figure 1 presents differences in the predicted probability of obesity. Average Marginal Effects (AMEs) are the expected increase in the probability of obesity for those who experienced a financial stressor. If the CIs include 0, then the AME is not significant at the 0.05 significance level. The largest AMEs were related to unsecured debts, property debts, and bankruptcy, which had AMEs ranging in size from approximately 0.03 (unsecured debt) to 0.05 (bankruptcy). The magnitude of AMEs related to wealth shocks were all relatively small and not statistically significant. When considered together, the ORs, AMEs, and model fit indices all confirmed that unsecured debts, property debts, and bankruptcy were the principal financial correlates of obesity during the transition to middle adulthood for the NLSY79 cohort.
Figure 1.
Average marginal effects (AMEs) of financial stressors on midlife obesity with 95% CIs (N=8,059; Observations=36,122).
Notes: AMEs (x-axis) were calculated from logistic regression estimates (Table 3). If an AME is positive and its CIs do not overlap 0, then the probability of midlife obesity was significantly larger among those who experienced the financial stressor.
DISCUSSION
This study extends past obesity research by using the breadth of the NLSY79 data to create a detailed accounting of the financial stressors experienced by a cohort of Baby Boomers as they aged from 31 to 39 in 1996 to 51 to 59 in 2016. Results revealed that the magnitude of associations between financial stressors and midlife obesity varied significantly. Property debt, unsecured debt, and bankruptcy had significant and relatively large associations with midlife obesity, whereas wealth shocks, as well as various types of wealth shocks, were not significantly related to obesity. These results and their implications and limitations are discussed in more detail below.
First, results confirmed that debt was among the more robust correlates of midlife obesity.7,46,47 Debt is a critically important topic in academic research, as well as contemporary policy discussions, and results from this study provide further evidence that certain forms of debt—property debt and unsecured debt specifically—are associated with midlife obesity. Unsecured debt was the most common financial stressor considered and had a particularly robust association with midlife obesity compared to other financial stressors. Although the housing crisis and subsequent “Great Recession” of 2007 exacerbated problems with indebtedness in the U.S. population, consumer debt has been growing for decades as the traditional trappings of middle-class life, e.g., a college degree and homeownership, became considerable financial undertakings that require large loans for the average person.28 Obesity appears to be an additional consequence of this unprecedented growth in consumer debt and financial distress among lower middle-class families during the early 21st century.
Second, the analysis found evidence that bankruptcy and the different types of bankruptcy, Chapters 7 and 13, had nearly identical associations with midlife obesity—an approximately 5-point increase in the predicted probability of obesity. Although larger than other financial stressors, the AMEs for bankruptcy had notably wider CIs due to the small sample of respondents who experienced bankruptcy. This was especially the case for Chapter 13, which had AMEs ranging from under 0.02 to well over 0.08. Furthermore, the AIC statistic suggested that models using bankruptcy as the single operationalization of financial stressor had a substantially poorer fit than models of obesity that included unsecured debt. Thus, bankruptcy appears to be an important—though rare—correlate of midlife obesity, while unsecured debt appears to be a more common but less influential correlate of midlife obesity.
Third, wealth shocks had no association with obesity after adjusting for control variables. To ensure this absence of a relationship was not due to recoding BMI into a binary indicator of obesity, the results were reproduced in a supplemental analysis using linear regression and BMI. The analysis using BMI came to the same conclusions, i.e., wealth shocks had no significant relationship to BMI after controlling for past obesity, family income and wealth, education, employment, and other demographic characteristics. This finding suggests that the association between wealth shocks and obesity did not exist independently of other indicators of economic stability, e.g., average income and wealth. Wealth shocks may represent more acute financial experiences than either debt or bankruptcy, which often act as long-term, chronic stressors.5
Finally, results from this study offer support to examining the distinct components of broad constructs such as wealth and debt.5,17,36,48 Researchers often use overall indicators of wealth and debt when studying health-related disparities.8 Using a single operationalization, e.g., net worth, of wealth simplifies theoretical and statistical models but may conceal important relationships. By examining multiple types of debts, the analysis revealed a robust relationship between unsecured debt and obesity that did not exist for the overall measure of indebtedness.
Limitations
Although the analysis found significant relationships between financial stressors for midlife obesity, the study has several limitations that should be noted and addressed by future work. First, the analysis controlled for race/ethnicity and gender but did not examine race or gender differences in the associations between financial stressors and obesity. Given differences in obesity rates and experiences of financial stressors, a closer examination of these social groups will be important. Second, even though the analysis considered a wide variety of financial stressors and controls, the results are still based on self-reports of weight, height, and wealth, which can be unreliable and thus introduce bias. Future research should consider extending this line of research with data that verifies self-reports with medical records and tax information. Third, to compare results across an expansive set of financial stressors, the financial stressors were all recoded into binaries (i.e., either 0 or 1 values). This is the only option for bankruptcy and its types, but this meant collapsing the distributions of debt variables. Although supplemental analysis found that higher levels of unsecured debt had a similar association to obesity as the indicator of unsecured debt presented here, but future research should still test whether the association changes significantly across the distribution of debt. Finally, reverse causality is always a possibility with observational data. Health conditions related to obesity could constrain economic attainment and thus increase one’s chances of experiences debt.49 Establishing causal relationships was not the aim of this study, but future research should extend the results of this study by testing the associations between debt, bankruptcy, and obesity are causal in nature.
CONCLUSIONS
Despite the limitations above, results found a clearer picture of the relationship between financial stressors and obesity emerged when net worth, debt, and bankruptcy were disaggregated into their distinct components. More specifically, although shocks to net worth and indebtedness were not significantly associated with midlife obesity, unsecured debt and property debt were associated with midlife obesity. Results also revealed the need for obesity prevention interventions targeting populations experiencing financial difficulties related to debt and bankruptcy. As obesity inequalities persist and grow in the coming years, it is imperative that researchers should also explore the legal and political systems that help or hinder individuals resolving financial problems.
Supplementary Material
ACKNOWLEDGMENTS
The author is especially grateful for the support of Obesity and Health Disparities Research Center (OHDRC) and the members of OHDRC who provided excellent feedback on prior drafts of this study. Research reported in this manuscript was supported by the National Institute on Minority Health and Health Disparities of the NIH under Award Number U54MD000502. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. No sponsor had a role in study design; collection, analysis, and interpretation of data; writing the report; or the decision to submit the report for publication.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Credit Author Statement
Joseph D. Wolfe: Conceptualization, Methodology, Data curation, Visualization, Writing
No financial disclosures were reported by the authors of this paper.
REFERENCES
- 1.Yilmazer T, Babiarz P, Liu F. The impact of diminished housing wealth on health in the United States: evidence from the Great Recession. Soc Sci Med. 2015;130:234–241. 10.1016/j.socscimed.2015.02.028. [DOI] [PubMed] [Google Scholar]
- 2.Boen C, Yang YC. The physiological impacts of wealth shocks in late life: evidence from the Great Recession. Soc Sci Med. 2016;150:221–230. 10.1016/j.socscimed.2015.12.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Pool LR, Needham BL, Burgard SA, Elliott MR, de Leon CFM. Negative wealth shock and short-term changes in depressive symptoms and medication adherence among late middle-aged adults. J Epidemiol Community Health. 2017;71(8):758–763. 10.1136/jech-2016-208347. [DOI] [PubMed] [Google Scholar]
- 4.Pool LR, Burgard SA, Needham BL, Elliott MR, Langa KM, Mendes de Leon CF. Association of a negative wealth shock with all-cause mortality in middle-aged and older adults in the United States. JAMA. 2018;319(13):1341–1350. 10.1001/jama.2018.2055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wolfe JD, Baker EH, Uddin J, Kirkland S. Varieties of financial stressors and midlife health problems, 1996–2016. J Gerontol B Psychol Sci Soc Sci. 2022;77(1):149–159. 10.1093/geronb/gbab108. [DOI] [PubMed] [Google Scholar]
- 6.Sun AR, Houle JN. Trajectories of unsecured debt across the life course and mental health at midlife. Soc Ment Health. 2018;10(1):61–79. 10.1177/2156869318816742. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Addo FR. Seeking relief: bankruptcy and health outcomes of adult women. SSM Popul Health. 2017;3:326–334. 10.1016/j.ssmph.2017.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sweet E, Nandi A, Adam EK, McDade TW. The high price of debt: household financial debt and its impact on mental and physical health. Soc Sci Med. 2013;91:94–100. 10.1016/j.socscimed.2013.05.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.O’Rand AM, Hamil-Luker J. Landfall after the perfect storm: cohort differences in the relationship between debt and risk of heart attack. Demography. 2020;57(6):2199–2220. 10.1007/s13524-020-00930-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kahn JR, Pearlin LI. Financial strain over the life course and health among older adults. J Health Soc Behav. 2006;47(1):17–31. 10.1177/002214650604700102. [DOI] [PubMed] [Google Scholar]
- 11.Lupien SJ, McEwen BS, Gunnar MR, Heim C. Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nat Rev Neurosci. 2009;10(6):434–445. 10.1038/nrn2639. [DOI] [PubMed] [Google Scholar]
- 12.Ferraro KF, Shippee TP. Aging and cumulative inequality: how does inequality get under the skin? Gerontologist. 2009;49(3):333–343. 10.1093/geront/gnp034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Tomiyama AJ. Stress and obesity. Annu Rev Psychol. 2019;70(1):703–718. 10.1146/annurev-psych-010418-102936. [DOI] [PubMed] [Google Scholar]
- 14.Hermes G, Fogelman N, Seo D, Sinha R. Differential effects of recent versus past traumas on mood, social support, binge drinking, emotional eating and BMI, and on neural responses to acute stress. Stress. 2021;24(6):686–695. 10.1080/10253890.2021.1877271. [DOI] [PubMed] [Google Scholar]
- 15.Spinosa J, Christiansen P, Dickson JM, Lorenzetti V, Hardman CA. From socioeconomic disadvantage to obesity: the mediating role of psychological distress and emotional eating. Obesity (Silver Spring). 2019;27(4):559–564. 10.1002/oby.22402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Stephens MAC, Wand G. Stress and the HPA axis: role of glucocorticoids in alcohol dependence. Alcohol Res. 2012;34(4):468–483. [PMC free article] [PubMed] [Google Scholar]
- 17.Wolfe JD, Baker EH, Scarinci IC. Wealth and obesity among US adults entering midlife. Obesity (Silver Spring). 2019;27(12):2067–2075. 10.1002/oby.22625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Munster E, Ruger H, Ochsmann E, Letzel S, Toschke AM. Over-indebtedness as a marker of socioeconomic status and its association with obesity: a cross-sectional study. BMC Public Health. 2009;9:286. 10.1186/1471-2458-9-286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kuroki M. Obesity and bankruptcy: evidence from US counties. Econ Hum Biol. 2020;38:100873. 10.1016/j.ehb.2020.100873. [DOI] [PubMed] [Google Scholar]
- 20.Dimock M Defining generations: Where Millennials end and Generation Z begins. Pew Research Center. https://www.pewresearch.org/fact-tank/2019/01/17/where-millennials-end-and-generation-z-begins. Published January 17, 2019. Accessed November 19, 2019. [Google Scholar]
- 21.Desmond M, Kimbro RT. Eviction’s fallout: housing, hardship, and health. Soc Forces. 2015;94(1):295–324. 10.1093/sf/sov044. [DOI] [Google Scholar]
- 22.Downing J The health effects of the foreclosure crisis and unaffordable housing: a systematic review and explanation of evidence. Soc Sci Med. 2016;162:88–96. 10.1016/j.socscimed.2016.06.014. [DOI] [PubMed] [Google Scholar]
- 23.Boen C, Keister L, Aronson B. Beyond net worth: racial differences in wealth portfolios and Black-White health inequality across the life course. J Health Soc Behav. 2020;61(2):153–169. 10.1177/0022146520924811. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wolfe JD, Thomeer MB. Divorce, economic resources, and survival among older Black and White women. J Marriage Fam. 2021;83(1):173–190. 10.1111/jomf.12702. [DOI] [Google Scholar]
- 25.Dwyer RE, McCloud L, Hodson R. Youth debt, mastery, and self-esteem: class-stratified effects of indebtedness on self-concept. Soc Sci Res. 2011;40(3):727–741. 10.1016/j.ssresearch.2011.02.001. [DOI] [Google Scholar]
- 26.Zurlo KA, Yoon W, Kim H. Unsecured consumer debt and mental health outcomes in middle-aged and older Americans. J Gerontol B Psychol Sci Soc Sci. 2014;69(3):461–469. 10.1093/geronb/gbu020. [DOI] [PubMed] [Google Scholar]
- 27.Richardson T, Elliott P, Roberts R. The relationship between personal unsecured debt and mental and physical health: a systematic review and meta-analysis. Clin Psychol Rev. 2013;33(8):1148–1162. 10.1016/j.cpr.2013.08.009. [DOI] [PubMed] [Google Scholar]
- 28.Dwyer RE. Credit, debt, and inequality. Annu Rev Sociol. 2018;44:237–261. 10.1146/annurev-soc-060116-053420. [DOI] [Google Scholar]
- 29.Berger L, Collins JM, Cuesta L. Household debt and adult depressive symptoms in the United States. J Fam Econ Issues. 2016;37(1):42–57. 10.1007/s10834-015-9443-6. [DOI] [Google Scholar]
- 30.Dobbie W, Goldsmith-Pinkham P, Yang CS. Consumer bankruptcy and financial health. Rev Econ Stat. 2017;99(5):853–869. 10.1162/REST_a_00669. [DOI] [Google Scholar]
- 31.Maroto M The scarring effects of bankruptcy: cumulative disadvantage across credit and labor markets. Soc Forces. 2012;91(1):99–130. 10.1093/sf/sos095. [DOI] [Google Scholar]
- 32.Allison P Listwise deletion: it’s NOT evil. Statistical Horizons. https://statisticalhorizons.com/listwise-deletion-its-not-evil#comments. Published June 13, 2014. Accessed March 10, 2022. [Google Scholar]
- 33.Allison PD. Missing Data. Sage; 2001. 10.4135/9781412985079. [DOI] [Google Scholar]
- 34.Stommel M, Schoenborn CA. Accuracy and usefulness of BMI measures based on self-reported weight and height: findings from the NHANES & NHIS 2001–2006. BMC Public Health. 2009;9:421. 10.1186/1471-2458-9-421. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Craig BM, Adams AK. Accuracy of body mass index categories based on self-reported height and weight among women in the United States. Matern Child Health J. 2009;13(4):489–496. 10.1007/s10995-008-0384-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Wolfe JD. The effects of socioeconomic status on child and adolescent physical health: an organization and systematic comparison of measures. Soc Indic Res. 2015;123(1):39–58. 10.1007/s11205-014-0733-4. [DOI] [Google Scholar]
- 37.Stata 17. Base Reference Manual. Stata Press; 2021. [Google Scholar]
- 38.Wilkins AS. To lag or not to lag?: Re-evaluating the use of lagged dependent variables in regression analysis. Political Sci Res Methods. 2017;6(2):393–411. 10.1017/psrm.2017.4. [DOI] [Google Scholar]
- 39.Gius MP. The prevalence of obesity and overweight among young adults: an analysis using the NLSY. International Journal of Applied Economics. 2011;8(1):36–45. [Google Scholar]
- 40.Long JS, Mustillo SA. Using predictions and marginal effects to compare groups in regression models for binary outcomes. Sociol Methods Res. 2018;50(3):1284–1320. 10.1177/0049124118799374. [DOI] [Google Scholar]
- 41.Long JS, Freese J. Regression Models for Categorical Dependent Variables using Stata. Third ed. Stata; 2014. [Google Scholar]
- 42.Burnham KP. Multimodel inference: understanding AIC and BIC in model selection. Sociol Methods Res. 2004;33(2):261–304. 10.1177/0049124104268644. [DOI] [Google Scholar]
- 43.StataCorp. Stata Statistical Software: Release 17. StataCorp LP; 2021. [Google Scholar]
- 44.Frongillo EA, Fawcett SB, Ritchie LD, et al. Community policies and programs to prevent obesity and child adiposity. Am J Prev Med. 2017;53(5):576–583. 10.1016/j.amepre.2017.05.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Keyes KM, Utz RL, Robinson W, a G L. What is a cohort effect? Comparison of three statistical methods for modeling cohort effects in obesity prevalence in the United States, 1971–2006. Soc Sci Med. 2010;70:1100–1108. 10.1016/j.socscimed.2009.12.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Drentea P, Reynolds JR. Where does debt fit in the stress process model? Soc Ment Health. 2015;5(1):16–32. 10.1177/2156869314554486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Drentea P, Reynolds JR. Neither a borrower nor a lender be: the relative importance of debt and SES for mental health among older adults. J Aging Health. 2012;24(4):673–695. 10.1177/0898264311431304. [DOI] [PubMed] [Google Scholar]
- 48.Wolfe JD, Thomeer MB. Divorce, economic resources, and survival among older Black and White women. J Marriage Fam. 2021;83(1):173–190. 10.1111/jomf.12702. [DOI] [Google Scholar]
- 49.Wilkinson LR, Ferraro KF, Mustillo SA. Wealth in middle and later life: examining the life course timing of women’s health limitations. Gerontologist. 2019;59(5):902–911. 10.1093/geront/gny048. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.