Executive Summary: “Inequality and Health: Long-Run Evidence from a Panel of Countries” by Christopher Jencks and Andre Leigh

Is mortality higher in countries that are more unequal? Leigh and Jencks draw on tax data for changes in top income shares, which have been shown to proxy changes in other inequality measures, to investigate the relationship between inequality, life expectancy, and infant mortality, They examine data on the richest 10 percent of the population in Australia, Canada, France, Ireland, the Netherlands, New Zealand, Switzerland, the UK, and the US between 1905-2002.

While previous studies have focused on the relationship between inequality and health at a single point in time, Leigh and Jencks’ work offers a view of this relationship over time. They are thus able to control for both country- and year-specific effects which often include such common unobservables as wars, technological innovations that diffuse rapidly, such as measles and polio vaccines, and major epidemics such as influenza and HIV/AIDS. When country- or year-fixed effects are overlooked, for both life expectancy and infant mortality, Leigh and Jencks find that a rise in inequality is associated with a statistically significant rise in mortality. However, when these effects are included in the analysis, the relationship between inequality and health becomes insignificant for both life expectancy and infant mortality. Leigh and Jencks therefore conclude that the relationship between inequality and mortality may be driven by unobserved factors affecting both inequality and health, rather than being a causal relationship. In other words, while there may be some consensus that there is a positive correlation between income and health, there is less agreement over the relationship between income inequality and health.

If inequality affects health by affecting public spending on healthcare or by weakening the social fabric, for instance, these effects would likely take some years to show up in mortality statistics. Leigh and Jencks’ 97-year survey, however, does not corroborate this suspicion. Though higher GDP is associated with better health outcomes, and this effect declines as GDP rises, holding country or year-fixed effects constant renders this relationship statistically insignificant. It appears that neither the ‘absolute income hypothesis’ nor the ‘relative income hypothesis’ captures the channels through which inequality affects health. In the former, health depends solely on individual income, where marginal health gains from an extra dollar of income diminish as income rises. Transfers from a richer individual to a poorer individual would therefore raise the health of the poor more than it would lower the health of the rich. Yet Leigh and Jencks do not find that more equal societies as such have better health, holding average income constant. In the latter, however, inequality has an indirect impact on health, on top of income. These include increased crime, reduced social capital, and public spending on healthcare, and what scholars have termed ‘relative depravation,’ wherein individuals’ comparisons with more affluent members of their cohort engender a loss of self-esteem and increased stress which in turn negatively affects health.

Leigh and Jencks point out that the problem in assessing these channels is that data on inequality and health often only exist in aggregate. It is extremely difficult to distinguish between the impact of so-called absolute income and relative income. Moreover, since sicker individuals are less likely to work, countries with lower health standards may have more unequal family incomes. The causal relationship between inequality and health can therefore run either from health to inequality as well as from inequality to health. Leigh and Jencks conclude that perhaps the underlying relationship between inequality and health is either non-existent or too fragile as to show up in a specification such as theirs, which would be consistent with a number of other careful cross-country analyses by Judge, Mulligan, and Benzeval and Deaton and Paxson.