On the Effect of Unemployment on Mortality

Abstract

It is well known that unemployed persons generally have more health problems than people in employment. Prior literature has indicated that unemployment has an adverse effect on individuals' mental and physical health, leading to higher morbidity among the group of unemployed. However, persons with ill health are more likely to become unemployed and they have lower exit rates to employment. Failing to account for this selection into unemployment may lead to an overestimated effect of unemployment on mortality. In this paper I account for the selection by including information on the pre-unemployment health status.

The paper utilises administrative register data on a 10 per cent random sample of the Danish labour force between 20 and 59 years of age in 1990-91 (N = 244,535). In addition to a rich set of personal characteristics, the data include the number of hospital admissions, GP visits, days on sickness benefit, and prior unemployment records from 1987 and on. I monitor survival status until the end of 2004.

Given the rich dataset, I argue selection-on-observables and estimate the average treatment effect for the treated (ATT) by propensity score matching, where a comparison group is constructed by matching each treated to one or more persons with a similar probability of being unemployed. A person is defined as unemployed if he or she was unemployed more than 20 weeks during 1990-91.

My two main findings are: First, unemployed men face a significantly higher risk of mortality during the follow-up period than otherwise. A much lower - and statistically insignificant - relative risk is found for unemployed women. Second, the relative risk of unemployment peaks when men are between 30 and 50 years old, whereas for women already during their twenties. Using the most conservative findings, unemployment increases the risk of mortality during the follow-up period by 38 per cent for both men and women in their twenties.

When men enter their thirties the relative risk increases from 38 to 46 per cent and remains at this level until their fifties where the relative risk drops to 26 per cent. For woman the relative risk drops from 38 to 21 per cent when they enter their thirties and then gradually declines to an 8 per cent relative risk at the age of fifty.

Keywords: mortality, unemployment, propensity score matching

JEL Classification: E24, I12