The 2008/09 Economic Crisis: The Impact on Psychological Well-Being in the USA

Abstract

The 2008/09 economic crisis has been the worst crisis of capitalism since the Great Depression. The causes and implications of the so-called “Great Recession” have been widely documented, but the effects of the crisis on psychological well-being have only received limited attention. Using state-level data, this paper aims to assess empirically the impact of the 2008/09 crisis on several indicators of mental health in the USA. The results indicate that unemployment and income levels have a significant and detrimental impact on mental health. This implies that social protection systems—and in particular labor market programs—play a paramount role in reducing the adverse impact of the crisis on mental health.

Keywords::

• global crisis
• mental health
• the USA

JEL classifications::

• C13
• E60
• I14

Notes

¹ Mental disorders are defined as diagnosable disorders, either mental, behavioral, or emotional (SAMHSA, 2012).

² The figure provided refers to the latest available data; in the USA, there are no systematic studies on the economic costs associated with SMI (or any form of mental illness). SMI refers to disorders that are severely debilitating, which in 2012 affected about 4% of the adult U.S. population (SAMHSA, 2013).

³ Note that the estimated costs include only measurable economic effects. They do not include the costs associated with the impact of mental illness on families and social relations. Such costs are more complex to evaluate, but are certainly relevant.

⁴ The literature on the association between unemployment and mental health has also addressed the question of possible reverse causality. The issue is about whether it is actually unemployment that causes psychological disturbances, or whether it is instead a state of mental illness (either prior or subsequent to the state of unemployment) that increases the chance of becoming or staying unemployed. Longitudinal studies do tend to support the existence of a causal link from unemployment to mental illness, rather than the reverse (Hamilton, Merrigan, & Dufresne, 1997; Murphy & Athanasou, 1999; Paul & Moser, 2009). In addition, even taking into account the link between poor mental health and unemployment, there is consensus that job losses have a causal role in generating psychological distress, depression, anxiety, low self-esteem, alcohol and substance abuse, mental illness, and suicide (Sharp, 2009).

⁵ The impact of inequality on mental health, however, remains a contested issue. A study by Weich, Lewis, and Jenkins (2001) identifies a complex relationship between income inequality and psychological disorders. They find that this association tends to be systematic only among individuals with higher incomes, and it is less frequent among individuals with the lowest incomes. More recently, Pickett, James, and Wilkinson (2006) and Pickett and Wilkinson (2010) investigate the link between mental illness and inequality in rich countries, and show that countries with greater income inequality are also the countries with a higher prevalence of mental distress and illegal drug use. The authors explain this positive association in terms of the impact of inequality on positional competition and social relationships. Greater inequality tends to increase the social pressure to acquire social status through material acquisition, and tends to weaken trust and community life. This in turn favors the development of mental illnesses and addiction to illicit drugs.

⁶ The interplay between the economy and mental health in developing countries is still an under-researched issue in the literature, but it is receiving growing attention. See Das, Do, Friedman, and McKenzie (2008) and Friedman and Thomas (2009).

⁷ According to SAMHSA, AMI among adults aged 18 or older is the result of having experienced in the past year a diagnosable mental, behavioral, or emotional disorder (excluding substance abuse behavior). SMI is defined as one that resulted in a major functional disturbance, which created serious interference or limitation in basic daily activities. SST refers to both suicide ideation and suicide attempts. SAMSHA defines alcohol and illicit drug dependence/abuse using the criteria specified in the Diagnostic and Statistical Manual of Mental Disorders of the American Psychiatric Association (1994), which include symptoms such as withdrawal, use in perilous situations, problems with the law, and interference in major responsibilities at work, school, or within the family over the past year. Illicit drugs include marijuana/hashish, cocaine (including crack), inhalants, hallucinogens, heroin, or prescription-type drugs used non-medically. These data are made available for the adult population as a whole, so in this analysis it is not possible to consider class, race, and gender as analytical categories.

⁸ The annual survey data publicly released by SAMSHA do not make state-level information available for privacy concerns. For this reason, this study relies on the estimates calculated by SAMSHA.

⁹ Given these contrasting findings in the literature, population density can be considered both as a protective and a risk factor.

¹⁰ Earnings are defined as the sum of wage or salary income and net income from self-employment, and do include interests, dividends, rental income, social security income, and public assistance income. Among the alternative measures of income available in the ACS, median earnings of the population aged 16 years and older were chosen in the American Human Development Report, 2008–2009 because it is the only income measure that is available for all the groups considered in the report (gender, race/ethnicity, gender by race/ethnicity, regions, states, and congressional districts). The income index is calculated by applying a logarithmic transformation to reflect the assumption of decreasing returns (as income increases, the marginal increase in material well-being diminishes).

¹¹ The Gini coefficient does not show significant variations across states, and therefore in this analysis it cannot be considered an informative indicator of inequality.

¹² Two-thirds weight is given to educational attainment, and one-third weight is given to gross enrollment.

¹³ In studying the impact of economic factors on mental health, there is no conclusive solution to the problem of potential endogeneity. It is theoretically possible to address possible reverse causality by using instrumental variable (IV) regression analysis, but it is very difficult to identify reliable instruments. Carrying out a sound IV regression is thus intrinsically challenging. Consistent with the evaluation by Davalos and French (2011), it is unlikely that self-reported mental health variables affect macroeconomic outcomes. Especially in the short run, where we investigate changes in macroeconomic conditions and changes in mental health indicators, the causality is most likely to run from the former to the latter.

¹⁴ In this context, spatial dependence refers to the possibility of significant spillover effects among neighboring states.

¹⁵ Through the robust regression method, the estimated coefficients are the same as the ordinary least squares, but the standard errors take into account potential issues concerning heterogeneity and lack of normality.

¹⁶ In order for the distance band (D) to include all the observations, D needs to be set between the largest minimum distance and the smallest maximum distance. In this case, the largest minimum distance is 40.62 and the smallest maximum distance is 48.12, so the distance band (D) has been set equal to 45.

¹⁷ To make the interpretation of the coefficients clear, the following is the unit of measurement of each variable. UNEMP (U-6), INEQ (the GDP share of employee compensation), HEALTHINS (the share of the population with health insurance coverage), and all the dependent variables are measured in decimals; INCOME (income index) and EDU (education index) are index numbers, so they take on a value ranging between 1 and 10.

¹⁸ For AD, a 1-percentage point increase in the share of employee compensation in GDP is associated with a 0.022 percentage point decrease in the rate of AD.

¹⁹ The interpretation of this coefficient is different from the previous cases, as the dependent and the independent variables are not measured on the same scale. The dependent variable is defined in decimals, so a one-point increase in the income index of a state leads to (100*the estimated coefficient) percentage point change in the dependent variable.