Econometric Methods for Health Economics
The availability of large-scale insurance claims data provides new opportunities for causal inference in health economics, while posing additional challenges on estimation, model building and variable selection. This session consists of three papers that examine modern econometric methods to deal with these challenges in the context of health care and health insurance. The first paper proposes a new estimation algorithm for models that entail multiple high-dimensional fixed effects, unbalanced panel structure, instrumental variables, and clustered standard error corrections. Applying the algorithm to a sample of over 1.4 million patients using more than 150,000 distinct primary care doctors over a 47-month period, the authors find that the breadth of provider networks dominates cost sharing in influencing consumers’ monthly utilization of care. The second paper is focused on the measurement of breadth of provider networks itself. While narrow network plans have been well known to gain market shares in the US, less is understood about their impacts partly due to a lack of reliable measures of narrow networks. This paper builds a statistical model to infer plan level provider choice using insurance claims data from the US employer-based health insurance market. The third paper models individual-level health care costs in China using a unique insurance claims data set covering more than 280,000 individuals from 2006 to 2014. A variety of econometric models are explored, with a particular focus on the evaluation and selection of risk adjustment models. Their findings have significant implications for reforming health care and payment system in China.