Mo Alloush is a Ph.D. candidate at the University of California, Davis.
Nearly one in five adults is affected by psychological disorders every year. However, the role that psychological well-being plays when it comes to earnings and economic decision-making is not well understood partially because the likely bi-directionality makes estimating causal links using representative data difficult: psychological well-being can affect income and economic decision-making and, at the same time, economic well-being may play a role in determining one’s state of mental health. In addition to this, there is a dearth of representative data at the individual-level that has information on mental health and socio-economic variables, especially panel data.
While this makes causal identification difficult, it makes the implications of this relationship all the more important: causal impacts in both directions could create a feedback loop between income (and by extension poverty) and psychological well-being that, if strong enough, has the potential to affect income dynamics and put individuals on vicious or virtuous cycles. Understanding this relationship may help explain prolonged poverty spells and low resilience to shocks, and it can pave the way to designing more effective poverty-alleviation programs.
Depression and Income
In my job market paper, I explore the relationship between income and psychological well-being using a four-wave panel dataset from South Africa that – in addition to rich information on many socio-economic variables – has a psychological well-being module (the Center for Epidemiologic Studies Depression (CES-D) scale) for adults in all four waves. This is unprecedented for a nationally representative panel dataset in a developing country.
The CES-D scale is a widely-used measure of depressive symptoms that screens for depression in the general population. Depression is the most common psychological disorder where the cross-sectional rate is nearly 5% among adults worldwide: that’s over 200 million people, and even more will suffer from it at some point in their life! In addition to a diminished quality of life and increased rates of mortality, depression is associated with significant functional impairment in occupational and social roles. How does this affect earnings/income?
Having four reasonably-spaced waves of data (or more!) allowed me to develop an econometric approach that extends established dynamic panel data methods to estimate the relationship between income and psychological well-being as a system of dynamic simultaneous equations to answer two main questions:
- Do psychological well-being and economic well-being causally affect each other in a significant way?
The answer I find is yes on average, but with significant nonlinearity:
- Psychological Well-being → Income: The results indicate that a 1 standard deviation (SD) increase in depressive symptoms decreases income by nearly 17% for an average individual. These effects are larger near or past the threshold that psychologists use to screen for depression. In the paper, I show that one avenue through which this occurs is a decreased labor supply at the extensive and intensive margins.
- Income → Psychological Well-being: The results show that a 20% decrease in household income per capita increases depressive symptoms by 0.1 SD on average which back-of-the-envelope calculations suggest is similar (but a bit lower) to the effect of cash transfers experimentally estimated by Haushofer and Shapiro (2016). This corresponds to a 6% increase in the likelihood of depression. The results are robust to the use of other measures of economic well-being such as food expenditure and wealth. I also find marked heterogeneity: this effect is larger among the poor, especially the extremely poor!
Together, these results show that income and psychological well-being are intertwined and that a particularly vulnerable group – the poor with low levels of psychological well-being where the estimated effects are large – may be disproportionately affected by shocks. This leads to the second question:
- How does this relationship affect income and poverty dynamics?
I answer this question using three different methods, all of which suggest that this bi-directionality can have important effects on the dynamics of income and poverty by amplifying shocks to either variable. The results suggest nonlinearity in the relationship between income and psychological well-being; long-term impacts of large positive and negative shocks may be markedly different. For the sake of blog brevity, I will highlight just one of these methods:
- I run simulations using the estimated system of dynamic simultaneous equations and independently and randomly drawn income and CES-D scales (calibrated by the data). The simulations show that the relationship between psychological well-being can explain some heterogeneity in long-term poverty: starting with low levels of psychological well-being and a per capita income that is poor or near-poor implies a 15-40 percentage point increase in the likelihood of poverty after 10 years compared to the prediction of simulations that do not account for effect of psychological well-being (the difference is shown in Figure 1).
The increasing availability of large-scale representative panel datasets that track mental health allows for the development and use of econometric approaches that answer important policy-relevant questions. This work sheds light on the relationship between psychological well-being and poverty in a general population in a way that cannot (or should not) be done experimentally. However, by extending dynamic panel methods to estimate the relationship, I require two assumptions to hold: a lack serial correlation in the error terms and no omitted variable bias after controlling for individual fixed effects. Where possible, I relax these assumptions to allow for first-order serial correlation and/or use alternative instruments to check the robustness of my estimates and find that the results hold up.
A Positive Note
While the results do mainly stress the potential negative consequences of this relationship, there is a positive story to tell. Poverty-alleviation programs have an added benefit of positive impacts on psychological well-being, which is an important goal in itself, and may also enhance the beneficiary’s capability to further increase their economic well-being. Thus, the results in my job market paper reaffirm the importance of considering psychological variables in poverty-alleviation programs. Early evidence on the economic impact of interventions such as cognitive behavioral therapy among specific subpopulations is encouraging (Baranov et al. 2017; Blattman 2017)! My hope is that the results from this paper guide and encourage future research on this topic and that organizations begin to consistently and seriously track mental health.