The Social Determinants of Choice Quality: Evidence from Health Insurance in the Netherlands
Market provision of impure public goods such as insurance retirement savings and education is common and growing as policy makers seek to offer more choice and gain efficiencies. This approach induces an important trade-off between improved surplus from matching individuals to products and misallocation due to well documented choice errors in these markets. We study this trade-off in the health insurance market in the Netherlands, with a specific focus on misallocation and inequality. We characterize choice quality as a function of predicted health risk and leverage rich administrative data to study how it depends on individual human capital, socioeconomic status and social and information networks. We find that choice quality is low on average, with many people foregoing options that deliver substantive value. We also find a strong choice quality gradient with respect to key socioeconomic variables. Individuals with higher education levels and more analytic degrees or professions make markedly better decisions. Social influence on choices further increases inequality in decision making. Using panel variation in exposure to peers we find strong within firm, location and family impacts on choice quality. Finally, we use our estimates to model the consumer surplus effects of different counterfactual scenarios. While smart default policies could improve welfare substantially, including the choice of a high-deductible option delivers little welfare gain, especially for low-income individuals who make lower quality choices and are in worse health.
We thank the Central Bureau of Statistics of the Netherlands and especially Annemieke Redeman for help with the data. Savannah Berquist, Chloé de Meulenaer, Miguel Fajardo Steinhauser and William Parker provided excellent research assistance. We thank seminar participants at Amazon, Edinburgh University, IO2 Stanford seminar series, Erasmus University Rotterdam, KULeuven, LSE, Statistics Netherlands, the KVS, the NBER Health Care and the Paris-London Public Economics meetings for excellent comments. We gratefully acknowledge funding by ERC (grant #716485), ESRC and STICERD. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.