Mortality in economic downturns: unobserved migration can create the false impression that recessions are good for health

by Vellore Arthi (University of Essex), Brian Beach (College of William & Mary), and Walker Hanlon (University of California, Los Angeles)

Are recessions good for health? A number of recent studies suggest that mortality actually goes down during recessions – at least in developed countries, where social safety nets help cushion the blow of unemployment and income loss.

This striking conclusion rests on one of two assumptions: either that people do not respond by migrating away from recession-stricken areas; or that if they move, these population flows can be perfectly measured. But are these assumptions realistic?

Migrant movements can be notoriously difficult to track, and famous episodes such as the Depression-era migration from the US Great Plains to California suggest that these sorts of internal population movements may indeed be a natural response to changes in local economic conditions. This raises the question: what does unaccounted migration mean for our assessment of the recession-mortality relationship?

Our research shows that unobserved migration from recession-stricken regions may actually lead us to underestimate systematically how deadly recessions really are.

To test how migration influences estimates of the relationship between recessions and mortality, we draw on a unique historical natural experiment: the temporary but severe economic downturn in the cotton textile-producing regions of Britain that resulted from the American civil war (1861-65).

The cotton textile industry was England’s largest industrial sector in the second half of the nineteenth century and, prior to the civil war, received the majority of its raw cotton inputs from the American South. The onset of the civil war sharply reduced these supplies, leading to a severe but temporary economic downturn that left several hundred thousand workers unemployed.

Digitising a wealth of historical data on births, deaths and population, and exploiting variation in both the geographical distribution of the British cotton textile industry and the timing of the civil war, we show that standard approaches yield the familiar result: the downturn, popularly termed the ‘cotton famine,’ reduced mortality.

But we also find evidence that migratory responses to this event were substantial, with much of this mobility occurring over short distances, as displaced cotton workers sought opportunities for work in nearby districts.

After making a series of methodological adjustments that account for this recession-induced migration, we show that the sign of the recession-mortality relationship flips: this downturn in fact appears to have been bad for health, raising mortality in both cotton regions and in the regions to which unemployed cotton operatives fled.

After accounting for migration bias, we find that:

• The civil war-era downturn in the cotton textile regions of Britain increased total mortality in the affected districts
• But the downturn appears to have led to improved infant and maternal mortality outcomes, probably by freeing up maternal time for breastfeeding, childcare, and other health-improving behaviours.
• Gains in infant health were offset by large and significant increases in mortality among the elderly.
• There was no net effect on mortality among working-age adults, who were also the most mobile during the downturn.
• This outcome appears to have been driven by worsening mortality due to the deteriorating nutrition and living conditions associated with income loss, which was in turn offset by improvements in maternal mortality and by fewer deaths by accidents and violence. (The latter finding is further supported by evidence that alcohol consumption and industrial accident rates fell during the recession.)

Our study provides both a methodological and factual contribution to our understanding of the relationship between recessions and health. The methodological contribution consists of showing that migration undertaken in response to a recession has the potential to introduce substantial bias into estimates of the recession-mortality relationship using the standard approach – particularly if these population flows are not well measured.

This bias is likely to be greater in settings, such as developing countries, where labour forces are more mobile, where weak social safety nets induce migration in response to recessions, and where the intercensal population data used to track these movements are poor. Studies applying the standard approach in these settings are likely to generate misleading results, which may lead to poorly targeted public health responses.

On a factual level, our study also contributes new evidence on the relationship between recessions and mortality in a historical setting, with the implication that studies focused on just one age group, such as infants, may generate results that are not representative of other segments of the population, or indeed of the overall relationship between recessions and mortality.