The Dynamic Relationship Among Health Knowledge, Health Sector, Health and Education

Presenter: Steven Lehrer, Queen's University and NBER

Abstract

The idea that healthy children are best able to learn and develop is widely accepted and underlies numerous public policies such as Medicaid and school lunch programs. Understanding not only whether health status impacts education outcomes but the exact pathway remains an open debate. The health sector is often hypothesized as a potential pathway and the sector itself has trumpeted their role, successfully placing the sector on the social and political agenda. Our interest is to investigate whether greater frequency of contact with the health sector leads first to improvements in health and second to academic outcomes in both early childhood and adolescence. However, since the use of health sector services reflects parents’ behavioral choice we require a variable that can directly influence these decisions. To identify the causal pathways between these factors a source of exogenous variation that only affects health knowledge, health sector visits or measures of health or education is required.
In this study, we first exploit exogenous variation from the availability of a randomized intervention that provided parents of these children with information on the benefits and availability of medical, dental, and social services in their community. To accomplish these goals we use a unique matched panel data set from French Canada that provides detailed data on academics, health status and interactions with the health sector. Most importantly, health sector usage is obtained from numerous sources including hospital administrative records and government claims records. In Canada, all health sector services are provided by the government thereby circumventing selection issues related to health insurance. These records provide accurate information on the date of service, type of medical provider, provider setting, diagnosis code, treatment or service received code and provider reimbursement for each encounter our subjects had for a 13 year period. Finally, the parental training program was assigned randomly to low socioeconomic schools and was offered in 28 of 53 schools in which our subjects attended Kindergarten. With detailed data on program attendance our estimation approach can recover multiple causal parameters including the average treatment effect for the treated.
Our main empirical findings are as follows. First, we find that the impacts of the parent training intervention on health sector usage are sensitive to the choice of time-period and measurement of the health sector. In general, we find that the intervention has large impacts early on and may have given a continued advantage in only preventive care visits. Second, we did not find any systematic evidence that on average gains in parental health knowledge led to improved measures of child or adolescent health on average. Third, we find significant heterogeneity in these relationships as children with poorer birth outcomes receive larger benefits from the intervention on later health. Finally, we contrast the impacts of physical and mental shocks to academic performance highlighting under which situations encounters with the health sector boost academic outcomes.