Unemployment raises the risk of death by about 63%, a meta-analysis found

Being out of work is associated with dying sooner: a systematic review and meta-analysis of 42 studies covering more than 20 million people found that unemployed working-age adults had a mean hazard ratio for death of 1.63 — roughly a 63% higher risk during follow-up than the employed [s1]. The effect is not evenly spread, it is only partly causal, and it becomes visible at population scale during downturns, when the mortality and suicide statistics move measurably with the economy [s2][s3].

The core number, and its shape

The 1.63 figure comes from a 2011 meta-analysis in Social Science & Medicine that pooled 235 mortality-risk estimates and modelled what moderates the association [s1]. Several features of the finding matter as much as the headline. The mean effect was larger for men than for women [s1]. It was concentrated in early and mid career and smaller for those near the end of their working lives, consistent with the loss of an identity and income stream that later-life workers are closer to shedding anyway [s1]. And the excess risk was highest in the first 10 years after job loss, then decayed [s1].

Crucially, the authors could partly explain the association away: among the subset of studies that adjusted for health-related behaviours such as smoking and drinking, the mean hazard ratio was 24% lower [s1]. That points to one of the main pathways — unemployment changes behaviour and mental state, and those changes carry the cardiovascular and other risks — rather than joblessness killing directly.

Association runs both ways

The meta-analysis cannot settle causation, and the reason is a familiar confound. People already in poor health are more likely to lose or leave work, so some of the mortality gap reflects sickness causing unemployment rather than the reverse [s1]. This "reverse causation" is why the strongest single number in the literature — a 63% higher death rate — should be read as an upper bound on the causal effect, not a measurement of it. The behavioural adjustment above, and the decay of risk over time, both suggest the true causal share is real but smaller than the raw association.

The population signal: deaths of despair

Where individual-level studies hit the causation wall, population trends offer a different kind of evidence. In 2015, Anne Case and Angus Deaton documented a reversal that had no parallel in other rich countries: all-cause mortality among middle-aged white non-Hispanic Americans rose between 1999 and 2013, undoing decades of decline [s2]. The increase was concentrated in that group, was largest among the least educated, and was driven by drug and alcohol poisonings, suicide, and chronic liver disease and cirrhosis — the cluster later labelled "deaths of despair" [s2]. Self-reported declines in physical and mental health, rising chronic pain, and deteriorating liver function accompanied the death rates, so the trend was not an artefact of how deaths were coded [s2].

Case and Deaton framed the deterioration as having economic roots — the erosion of stable, decently paid work for less-educated Americans — while stopping short of claiming a single cause [s2]. The pattern is the macro-scale echo of the micro-scale finding: economic dislocation and the behaviours it feeds show up in the mortality data.

Recessions and suicide

The clearest natural experiment is a sharp downturn. A time-trend study across 54 countries estimated 4,884 excess suicides in 2009 — the year after the 2008 financial crisis — above what pre-crisis trends (2000–07) would have predicted, with a 95% confidence interval of 3,907 to 5,860 [s3]. The rise fell mainly on men: suicide rates ran about 4.2% above expectation in European countries and 6.4% above in the Americas, while women showed little change [s3]. And the size of the increase tracked the size of the jump in unemployment, particularly in countries that had enjoyed low unemployment before the crash [s3].

That dose-response link — more job loss, more excess suicide — is the strongest available evidence that the economic shock, not some coincident factor, is doing the work [s3].

What it means, and what it does not

None of this is medical advice, and none of it says an individual job loss will harm a given person's health; these are population associations with real limits [s1][s3]. What the three studies together establish is that unemployment and recessions leave a mortality signature, that it is heaviest among men and the less educated, and that suicide and substance-related deaths are the sharp end of it. The same downturns that produce job loss also produce the medical debt and social disconnection — loneliness carries its own measured mortality risk — that compound the effect.

What to watch

The policy question the evidence raises is whether the mortality cost of downturns is fixed or modifiable. Chang and colleagues note that the countries and periods with stronger labour-market protection showed smaller suicide rises [s3], which is a testable claim rather than a settled one. The metric to follow is whether future recessions reproduce the 2009 suicide spike, or whether active labour-market and mental-health measures blunt it.

Sources

• [s1] Losing life and livelihood: a systematic review and meta-analysis of unemployment and all-cause mortality — Social Science & Medicine, published March 2011. https://doi.org/10.1016/j.socscimed.2011.01.005
• [s2] Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century — Proceedings of the National Academy of Sciences, published 2 November 2015. https://doi.org/10.1073/pnas.1518393112
• [s3] Impact of 2008 global economic crisis on suicide: time trend study in 54 countries — BMJ, published 17 September 2013. https://doi.org/10.1136/bmj.f5239