The effect of job loss on risky financial decision-making

Abstract

Job loss is a common and disruptive life event. It is known to have numerous long-term negative effects on financial, health, and social outcomes. While the negative effects of becoming unemployed on health and well-being are well understood, the influence of job loss on financial decisions has received little attention. Across a large-scale survey ([Formula: see text]), spending data from a bank ([Formula: see text]), and two online experiments (total [Formula: see text]), we find that job loss increases financial risk-taking. First, in survey data, job loss is associated with elevated levels of self-reported financial risk-taking and lottery ticket purchases. Next, using administrative data from a large bank, we find consistent causal evidence of the influence of job loss on gambling spending. Although total spending decreases after job loss, gambling spending is less affected than our control categories. Finally, we turn to two incentive-compatible manipulations of job loss operationalized in a lab setting. We find that this experimental manipulation increases the take-up of financial risks. The current finding that job loss increases financial risk-taking could accentuate long-term negative financial effects of job loss.

Keywords: employment; financial decision-making; job loss; risk.

Fig. 1.

The figure plots the results from our key measures in the cross-sectional survey. Gray bars show the no job loss group $(N=30,674)$ and red bars show the job loss group $(N=7,180)$. Panel A plots the number of gambles taken in our gamble task. Panel B plots the self-reported number of lottery tickets purchased. Error bars represent 95% CI.

Fig. 2.

Panel A shows the change in the likelihood of spending before and after the month in which the job loss group lost their job in both gambling and control categories. Panel B shows the change in the average amount of spending before and after the month in which the job loss group lost their job in both gambling and control categories. Gray bars represent the change for control categories, red bars represent the change for gambling spending. For the full model without controls, see SI Appendix, Table S11.

Fig. 3.

The figure plots the proportion of participants choosing to gamble their earnings from Experiment 1 in panel A$(N=201)$ and Experiment 2 $(N=1,202)$ in panel B. Gray bars show the no job loss group and red bars show the job loss group for both experiments. For experiment two, the gold bar shows the reduced rate condition, and the blue bar shows the compensation condition. Error bars represent 95% CI.