There’s a new study on the Earned Income Tax Credit and its impact on maternal health that’s worth some attention (ungated working paper version here). The EITC is a refundable tax credit available to poor families and individuals; the credit is most generous for families with children and an EITC expansion in 1993 disproportionately benefitted families with two or more children relative to those with one child. William Evans and Craig Garthwaite attempt to exploit this expansion to identify the effect of EITC on both self-reported and objective measures of maternal health.
The authors’ effort to observe changes in objective health metrics is an especially interesting aspect of the study. They’re able to use results from NHANES, a survey that collects information on biomarkers. Biomarkers are physical indicators—like blood pressure or the presence of acute-phase proteins indicating inflammation—measured by a medical professional, thus eliminating any uncertainty or potential bias inherent to self-reported measures. And biomarkers matter:
[I]ndividuals can experience significant reductions in health even without the presence of identifiable chronic conditions. Often, these decreases in health can be identified through the use of biomarkers even when specific diseases are not detectable.
These metrics are valuable, but come at a cost: labor-intensive as they are to collect, the survey suffers from a relatively small sample size. NHANES has a sample of about 10,000; by contrast, BRFSS (the survey used for the study’s self-reported measures) boasts a sample of 450,000.
What did the authors find?
Utilizing self- reported data from the large sample of respondents to the BRFSS, we find that the expansion of the EITC decreased the number of reported bad mental health days for mothers with a high school degree or lower and two or more children compared to a similar woman with only one child. Suggestive evidence was also found that the increase in payments increased the probability of reporting excellent or very good health status. We also find strong evidence that the expansion of the EITC lowered the counts of the total number of risky biomarkers for women with two or more children and a high school degree or less compared to similar women with only one child. These effects were strongest for measures of inflammation and suggestive evidence was found for a decrease in women with risky levels of diastolic blood pressure.
The authors’ findings throughout the paper were a mix of significant and “marginally significant” (a semi-technical way of saying “close, but not quite there”). Since analysis was restricted to mothers assumed to be receiving EITC, the sample size creates certain unavoidable limitations, similar to the Oregon Medicaid study. But taken as a whole—and taken in the context of previous work—it makes sense for the EITC to improve health. The authors offer this explanation:
Receiving a large EITC benefit may loosen a liquidity constraint regarding important expenditures that would not occur with a much smaller transfer. For example, previous analyses of the expenditure patterns of EITC recipients found increased consumption in categories such as housing, transportation, and durable goods. In a survey of EITC recipients, Smeeding, Phillips, and O’Connor (2000) found that over half of the respondents spent their refunds on goods to improve social mobility such as moving and transportation.
In short, cash transfers can get at social determinants of health in a way that medical care can’t. It reflects the core thesis of Bradley and Taylor The American Health Care Paradox: the United States is “exceptional in how little it spends on social services relative to health.” But research on social determinants and the effect of social policies on health is difficult—difficult and important. This study is ambitious, and we need more like it.