Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Doctor of Philosophy

Program

Economics

Supervisor

Rivers, David

2nd Supervisor

Navarro, Salvador

3rd Supervisor

Conley, Timothy

Abstract

My thesis consists of three chapters studying the impact of health insurance design and public health policy on consumers’ health and health care utilization, welfare, and costs.

Chapter 2 studies an overlooked dynamic incentive that encourages health care utilization, which I term \textit{dynamic moral hazard}. Typical health plans feature high deductibles and caps on consumers' out-of-pocket spending, which generate nonlinear pricing. These nonlinearities, coupled with the uncertainty intrinsic to future health care demand, encourage consumers to increase spending since current utilization lowers future expected prices. Standard models study health care utilization decisions through the lens of annual models, which abstract from the dynamic incentives throughout the year induced by nonlinear pricing. To understand the implications of dynamic moral hazard on insurance design, I develop and estimate a dynamic, within-year model of health care demand that allows for rich, flexibly-correlated unobserved heterogeneity. I use this model to study alternative contract designs in the context of employer-sponsored health insurance. My results show that the presence of dynamic moral hazard can severely dampen the welfare gains associated with higher cost-sharing and plays a crucial role, distinct from static moral hazard, in determining optimal insurance contract design.

Motivated by the substantial unobserved heterogeneity found in Chapter 2, Chapter 3 studies the evidence for and sources of selection, both adverse and advantageous. I first propose a new method to recover family-specific distributions of multidimensional unobserved heterogeneity conditional on their health care utilization decisions and the estimated population distribution of types. Using insurance choice survey data and the model from Chapter 2, I recover family-specific measures of risk aversion. Finally, I examine the correlation between risk aversion and the other dimensions of unobserved heterogeneity to unmask potential sources of advantageous selection. I find a new source of advantageous selection: preferences for going to the doctor, which suggests advantageous selection on number of visits instead of health. These findings have implications for the plan portfolio choice problem and suggest that models with richer unobserved heterogeneity might capture the gains from offering plan choice.

The first two chapters focus on studying health care demand for insured people. However, an important fraction of the population does not have health insurance and relies on the health care safety net to get needed care. Chapter 4 estimates the causal impact of a maternity conditional cash transfer program on the choice between abortion and childbirth in a context where abortion is illegal. We leverage several sources of social security administrative micro-data matched to longitudinal hospital records to estimate the effect of participating in the Argentinean \textit{Asignación por Embarazo para Protección Social}, a conditional cash transfer program implemented in 2011 and targeted at pregnant women who are unemployed or working in the informal sector. We exploit the substantial amount of inflation in Argentina to instrument for endogenous participation in the program. We estimate that participation in the program led to a sizable reduction in the probability of abortion and in the incidence of normal birthweight. These findings are consistent with a change in composition effect, in which mothers whose abortion decision is affected have a higher risk of low birthweight children.

Summary for Lay Audience

My thesis consists of three chapters studying the impact of health insurance design and public health policy on consumers’ health and health care utilization, welfare, and costs.

Chapter 2 studies an overlooked dynamic incentive that encourages health care utilization, which I term \textit{dynamic moral hazard}. Typical health plans feature high deductibles and caps on consumers' out-of-pocket spending, which generate nonlinear pricing. These nonlinearities, coupled with the uncertainty intrinsic to future health care demand, encourage consumers to increase spending since current utilization lowers future expected prices. Standard models study health care utilization decisions through the lens of annual models, which abstract from the dynamic incentives throughout the year induced by nonlinear pricing. To understand the implications of dynamic moral hazard on insurance design, I develop and estimate a dynamic, within-year model of health care demand that allows for rich, flexibly-correlated unobserved heterogeneity. I use this model to study alternative contract designs in the context of employer-sponsored health insurance. My results show that the presence of dynamic moral hazard can severely dampen the welfare gains associated with higher cost-sharing and plays a crucial role, distinct from static moral hazard, in determining optimal insurance contract design.

Motivated by the substantial unobserved heterogeneity found in Chapter 2, Chapter 3 studies the evidence for and sources of selection, both adverse and advantageous. I first propose a new method to recover family-specific distributions of multidimensional unobserved heterogeneity conditional on their health care utilization decisions and the estimated population distribution of types. Using insurance choice survey data and the model from Chapter 2, I recover family-specific measures of risk aversion. Finally, I examine the correlation between risk aversion and the other dimensions of unobserved heterogeneity to unmask potential sources of advantageous selection. I find a new source of advantageous selection: preferences for going to the doctor, which suggests advantageous selection on number of visits instead of health. These findings have implications for the plan portfolio choice problem and suggest that models with richer unobserved heterogeneity might capture the gains from offering plan choice.

The first two chapters focus on studying health care demand for insured people. However, an important fraction of the population does not have health insurance and relies on the health care safety net to get needed care. Chapter 4 estimates the causal impact of a maternity conditional cash transfer program on the choice between abortion and childbirth in a context where abortion is illegal. We leverage several sources of social security administrative micro-data matched to longitudinal hospital records to estimate the effect of participating in the Argentinean \textit{Asignación por Embarazo para Protección Social}, a conditional cash transfer program implemented in 2011 and targeted at pregnant women who are unemployed or working in the informal sector. We exploit the substantial amount of inflation in Argentina to instrument for endogenous participation in the program. We estimate that participation in the program led to a sizable reduction in the probability of abortion and in the incidence of normal birthweight. These findings are consistent with a change in composition effect, in which mothers whose abortion decision is affected have a higher risk of low birthweight children.

Download

Share

COinS