“A Structural Analysis of Opioid Misuse: Health, Labor, Policy, and Misperception of Opioid Misuse Risk” (Job Market Paper, Link)
Abstract: This paper examines the heterogeneous responses of opioid misuse across health and labor status during 2015-2019. Three aggregate changes that characterize this period are considered: increased risk of death from opioid misuse, the spread of state-level policies on opioid prescribing, and fluctuating prices. The role of opioid misuse risk perception is highlighted as an additional channel for policy intervention. By estimating a dynamic discrete choice model of opioid misuse with stochastic perception bias, I show that labor status is just as important as health conditions in determining opioid misuse. Counterfactual analysis indicates that the decrease in opioid misuse is mainly due to the increased risk of death from opioid misuse. Policies targeting opioid prescription generally have no effect on overall misuse but alter the share of people using illegal opioids. No evidence is found for the impact of illegally traded opioid prices on opioid misuse. Lastly, correcting the perception of opioid misuse risk would be effective in decreasing opioid misuse among the unemployed and those with poor mental health, but its aggregate effect is limited due to the relative rarity of perception bias.
“Externality in Sending Children Back Home: A Structural Approach to Foster Care Incentives”
This paper examines the unintended consequences of the foster care policy on children aimed at reunifying families. Although child abuse and neglect are rare, its recurrence and foster readmission are notably high. This raises concerns over the policy’s efficacy in children’s safety and health. Leveraging the Child and Caregiver Outcomes Using Linked Data (CCOULD) published by the U.S. Department of Health and Human Services, this study aims to (1) assess the impact of child maltreatment and foster care re/admission on children’s health; (2) evaluate whether Medicaid data can be used to predict future child maltreatment; (3) build a dynamic model of foster care system that flexibly captures its institutional incentive and incentive to care for children’s welfare, and (4) examine how would additional information from Medicaid change foster care system’s decision on children and how much it would change children’s welfare, measured by Medicaid expenditure.
“Identification of Dynamic Discrete Choice Models with Quasi-Hyperbolic Discounting under Finite Dependence”
In this paper, I generalize the representation theorem in Arcidiacono and Miller (2019) in dynamic discrete choice models with quasi-hyperbolic discounting. Then I provide identification result given the two-period finite dependence and exclusion restriction in a finite horizon model. Monte Carlo simulation shows that the exclusion restriction is strong enough to separate the discount factor and present bias.
“Sufficient Conditions for Identification of Dynamic Discrete Choice Models under Finite Dependence”
This paper investigates the role of finite dependence paths in identifying dynamic discrete choice models. I prove constructively that there exists a maximum number of these paths for identification in discrete state space. The number of finite dependence paths grows exponentially by the number of states and choices. This growth explains the lack of consensus on the number of paths for reliable identification. In the one-period finite dependence setting, identification of flow utilities is achieved by the full rank condition of the linear system of equations of conditional value function differences. By examining a two-period finite dependence setting as a special case of multiple-period finite dependence, I demonstrate methods to determine identification for utility primitives.
Work in Progress
``De-biased Conditional Choice Probabilities Estimation off Short Panels” (with R. A. Miller)
``Lifecycle Decisions of Labor Supply, Homeownership, Marriage, and Fertility’’ (with R. A. Miller)