CandidatesBack to Candidates
Job Market Paper
Valuing School Choice: Using a Randomized Experiment to Validate Welfare Evaluation of Private School Vouchers
(Joint with Peter Arcidiacono, Karthik Muralidharan, and John D. Singleton)
While parents and students value schools for many reasons, evaluation of school choice programs often focuses on impacts on students’ test scores alone. In this paper, we pursue a unique research design to credibly identify the welfare benefits of private school vouchers using data from Andhra Pradesh School Choice Program, a randomized controlled trial of private school vouchers in rural India. We first develop and estimate a model of school choice with and without credit constraints using rich control group data. We then externally validate the estimates by simulating a voucher program in the control group to compare with the experimental outcomes. We find that, although both models with and without credit constraints fit the control group data well, accounting for credit constraints is important for accurately predicting voucher takeup and implies substantial welfare gains to recipients. Our preliminary results show that the voucher program generated 1 to 5 dollars welfare gain per dollar spending when credit constraints were considered. Ignoring credit constraints led to predictions of lower voucher takeup rate and net negative welfare effect of the program. Preliminary welfare decomposition results show that the gain from the change in short-term test scores only explains 4 percent of the total welfare gain for the beneficiary families.
The Impact of Conditional Cash Transfer Programs under Risk-Sharing Arrangements: Schooling and Consumption Smoothing in Rural Mexico
An important reality of the lives of poor families in developing countries is that, in the absence of formal lending opportunities, they manage income risk through transfers among relatives and neighbors, often referred to as informal risk sharing. There is evidence that public transfers targeted to specific groups, such as old age pension programs or conditional cash transfers (CCT), are shared by others through informal risk sharing. Most of the existing studies about the effect of CCT programs on children’s schooling have modeled the behavior of households without consideration of the presence of informal risk sharing.
This paper develops and estimates a model of informal risk sharing with limited commitment that incorporates children's school attendance choices. The model is estimated using Mexican rural villages data from the PROGRESA experiment and is used to analyze how the presence of informal risk sharing influences schooling and child labor choices, as well as the effectiveness of CCT programs. In particular, I compare the outcomes (schooling, child labor, and consumption) generated under the informal risk-sharing model with those that would be obtained, forcing households to make choices under autarky.
I find that the number of years of schooling completed at age 18 is 0.3 years lower under autarky than with risk sharing. Also, given a set of parameter values the effect of CCT on schooling outcomes and welfare of households is larger under autarky than under risk sharing, and CCT increases consumption volatility under risk sharing. When the autarky model parameters are re-calibrated to fit the observed schooling outcomes in the data, the predicted difference in the effect of CCT on schooling outcomes under autarky and under risk sharing disappears. However, the difference in the welfare effects persists.
Measuring Heterogeneity in Demand for In-Kind and Cash Transfers: Evidence from Two Choice-Based Experiments in India
(Joint with Karthik Muralidharan, Paul Niehaus, and Sandip Sukhtankar)
Although it is well understood that estimating heterogeneity in treatment effect is important in policy evaluation, not much effort has been made to incorporate heterogeneous treatment effect in designing optimal policy. In this paper, we measure heterogeneity in beneficiary preference using revealed preference approach enabled by a unique choice-based experiment in the context of food security policy in India and argue that beneficiary welfare can be maximized by allowing beneficiaries to make a choice between in-kind (status quo) and cash transfer instead of imposing one or the other. The distribution of cash-equivalent value of rations were nonparametrically estimated based on the experimental and stated outcomes. The estimated distribution displayed wide dispersion of beneficiaries’ valuation from its fiscal value within each state. We found that if the current system is replaced by cash transfer system, the total monthly welfare gain will be 0.3 billion Rupees in Bihar and negative 0.3 billion Rupees in Rajasthan. On the other hand, if beneficiaries are allowed to choose between cash and in-kind transfers, monthly welfare gain will reach 2.3 billion Rupees in Bihar and 1.3 billion Rupees in Rajasthan.
Theoretical Appendix for “Quality and Accountability in Healthcare Delivery: Audit-Study Evidence from Primary Care in India”
(The authors of the main paper are Jishnu Das, Alaka Holla, Aakash Mohpal, and Karthik Muralidharan. Forthcoming in AER)
In this theoretical appendix, I develop a theoretical framework for a doctor-patient transaction in rural India. To provide a potential mechanism to interpret the findings of the paper, I apply Bayes learning process where a doctor exerts effort to draw a noisy signal to update belief about the true underlying illness of a patient (consultation stage). The accuracy of the signal depends on the level of effort. Given the posterior belief about the true illness, the doctor determines the treatment to maximize own payoff, partially internalizing the patient’s health outcome (treatment stage). Because public doctors have less incentive to exert effort, the signal they draw is nosier. They decide how to treat the patient with larger uncertainty about the true illness, and thus, prescribe many medicines to increase the chance of treating the illness. On the other hand, private doctors exert more effort, have better idea what the illness is, but because of financial incentive, they also prescribe many medicines. Given the number of medicines prescribed, private doctors choose the correct medicine with higher probability. The prediction of this mechanism also supports the authors’ argument that in an environment where public-sector effort level is low as in rural India, the benefits of unregulated market incentive for higher diagnostic effort may outweigh the costs of over-treatment.
Teaching Assistant/Head Teaching Assistant/Recitation Instructor for:
Business in the Global Political Environment
Business Environment in Developing Countries
Advanced Econometrics Techniques and Applications
Topics in Development
Introduction to Economics for Business
Software: FORTRAN, MATLAB, STATA, R
Languages: Fluent in English and native in Korean
Eligible to work in U.S. without sponsorship
Kenneth I. Wolpin
Professor of Economics, Rice University
Professor of Economics, University of Pennsylvania
Professor of Economics, University of Pennsylvania
Associate Professor of Economics, UC San Diego
I recently found a position and will be no longer available for interviews at the AEA meetings.