“If… then…”: Are mental health problems a cause or effect of unemployment?

It is widely accepted that a relationship exists between mental health problems and unemployment. However, studies contest whether increased unemployment is a result of an individual living with a mental illness or a consequence of said unemployment. This is to say:

1. If you have a mental health problem, then are you more likely to be unemployed, or;
2. If you are unemployed, then it is more likely that you will develop a mental illness

The near impossibility of measuring this relationship lies within the nature of social science data itself. To measure changes in employment status and mental health conditions requires using observational panel data, as the same respondents must be tracked over a set time frame. The issue with this is that employment is not the only variable that has clear links to mental health. Unobserved heterogeneity significantly impacts the integrity of analysis focusing on a single primary cause (Schunck, 2014). Education is noted as having a positive impact on both mental health and employment, so how can the effects of this be distinguished from the effects of mental health and unemployment?

Schunck (2014) discusses various alternatives as to how to control for the issue of unobserved heterogeneity. In a perfect world, he suggests a ‘counterfactual set up’ would be used. This would mean observing the same individual when they were employed versus when they were unemployed. However, this procedure is impossible to realistically achieve. Time varying factors such as age and health would interfere with results, since no one person can experience both realities concurrently. This is the “fundamental problem of causal inference” (Holland, 1986).

It is common practice to adjust for these factors using controls including age, gender and length of lifetime unemployment (Staiger, 2018). In particular, people who suffer from mental health problems experience a “selection” effect, whereby it is more difficult to find and maintain a job.  In this regard, it becomes challenging to measure the impacts of unemployment in this perpetuating cycle (Institute for Work & Health, 2009). However, in Murphy’s (1999) study it was found that in 14 out of 16 cases job loss resulted in reduced mental health after controlling for the presence of pre-existing conditions. Thus, distinguishing two different but equally important issues:

1. The reverse causality that people with mental health issues are more likely than others to become unemployed.
2. Unemployment reduces mental health regardless of previous conditions in a manner that is statistically significant.

So, what actions can be taken to remedy these issues?

The idea of a jobs guarantee is back in vogue amongst post-Keynesian economists, which essentially involves the government hiring those individuals who are unemployed into meaningful employment. The recent resurgence of the concept has been strongly associated with the US Green New Deal as a means to move towards full employment while simultaneously combating sustainability issues (Pettifor, 2019). Pavlina Tcherneva is one of the leading academics in this area. Having written ‘The Case for a Job Guarantee’, Tcherneva contends that the non-accelerating inflation rate of unemployment (NAIRU) is an outdated measure proven to be incorrect on several occasions. The NAIRU depends on the causal link between the unemployment rate and inflation.  However, inflation has so many contributing factors that this link can become clouded.

When unemployment rates fell below the forecast NAIRU in the U.S. during the 1990s, there was no dire acceleration in inflation. Ball (2002) attributed this to other factors at play. This returns back to our discussions with regard to difficulties in measurement in social sciences, as many academics attribute this anomaly example to other factors such as demographics or technology. In turn, these additional factors must also be considered when squaring blame on unemployment rates for rising inflation – it is only fair. Perhaps it is time we re-evaluate the cost-benefit of a jobs guarantee.  This is not to say that we should abandon the use of the NAIRU completely, but perhaps take a more balanced view and remember the humanity behind unemployment.

A jobs guarantee would combat not only our second question of whether mental health issues are an effect of unemployment, but also the issues of reverse causality in regard to people with pre-existing mental health issues being more likely to become unemployed in the first place (Tcherneva, 2018). The jobs guarantee would offer liveable alternatives for employment by removing the stressors of unemployment such as affording housing and food, while concurrently maintaining dignity and pride for participants. Additionally, the program would mean people with existing mental health issues could be paired with more appropriate work, thus reducing the likelihood of short bursts of employment.

However, there are downsides to a jobs guarantee. The Keating government experimented with a form of jobs guarantee for the long term unemployed in the 1990s, which didn’t prove to be as successful as hypothesised. The main issue with the program was that participants did not acquire transferable skills that they could apply to unsubsidized jobs. This meant that only a third of participants were able to find unsubsidized employment in the three months after the initiative (Mariuz, 2020). These issues would only be magnified if the program were to be extended, as costs also increase.

Moving forward, there are components we can take from both views, and lessons to be learnt from previous implementation of a jobs guarantee. But, as we return to our initial query of the links between mental health and unemployment, it is clear that unemployment has a detrimental effect on mental health, whether pre-existing conditions are present or not. There has to be more we can do to mitigate these issues, and we must remember the faces behind unemployment as we try to put a price on it.

References:

[1] Ball, L & Mankiw, G. (2002). The NAIRU in Theory and Practice. Journal of Economics Perspectives, 16(4), 115-136. https://doi.org/10.1257/089533002320951000

[2] Holland, Paul W. (1986). Statistics and Causal Inference. Journal of the American Statistical Association, 81(396), 945–960. https://doi.org/10.1080/01621459.1986.10478354

[3] Institute for Work & Health Canada. (2009). Unemployment and mental health. https://www.iwh.on.ca/summaries/issue-briefing/unemployment-and-mental-health

[4] Mariuz, D. (2020). There’s serious talk about a “job guarantee”, but it’s not that straightforward. The Conversation. https://theconversation.com/theres-serious-talk-about-a-job-guarantee-but-its-not-that-straightforward-140632

[5] Murphy GC, Athanasou JA. (1999) The effect of unemployment on mental health. Journal of Occupational and Organizational Psychology. 72:83-99.

[6] Pettifor, A. (2019). The Case for The Green New Deal. New York: Verso.

[7] Schunck, R. (2014). Estimating causal effects with longitudinal data: Does unemployment affect mental health?. SAGE Research Methods Cases. https://www.doi.org/10.4135/978144627305014533933

[8] Staiger, T., Waldmann, T., Oexle, N. et al. (2018) Intersections of discrimination due to unemployment and mental health problems: the role of double stigma for job- and help-seeking behaviors. Social Psychiatry and Psychiatric Epidemiology 53, 1091–1098. https://doi.org/10.1007/s00127-018-1535-9

[9] Tscherneva, P. (2018). The Job Guarantee: Design, Jobs and Implementation. Levy Economics Institute Working Papers No. 902. http://www.levyinstitute.org/pubs/wp_902.pdf