Title: An Empirical and Theoretical Analysis of Nonlinear Public Policies
- Camille Landais, London School of Economics
- Andrew Atkeson, University of California, Los Angeles
- Christian Hellwig, UT1 Capitole – TSE
- Nicolas Werquin, UT1 Capitole – TSE
This thesis proposes empirical and theoretical methods to optimize public policies and applies them to subsidies for renewable energies and disease-control measures. In particular, my thesis focuses on optimal nonlinear policies, which, in many applications, are more efficient than linear policies. The main challenge in the design of optimal nonlinear policies is to know how citizens react to them. Theoretically, I address this issue by characterizing optimal policies as a function of observable statistics. Empirically, I propose novel quasi-experimental estimators for citizens’ responses. The thesis is organized into three chapters.
The first chapter’s title is "Kinks Know More: A Policy Evaluation Beyond Bunching with an Application to Solar Energy." The chapter estimates agents’ intensive and participation margin responses to nonlinear incentive schemes. The proposed nonparametric estimator allows evaluating nonlinear pricing schedules when existing kink and discontinuity methods are inapplicable because both margins are present. The chapter’s first contribution is to show that the reactions of agents to kinks or discontinuities in the incentive scheme identify responses at both margins simultaneously. The only observable needed for estimation is the distribution of agents’ choices. The second contribution of the chapter is to evaluate the German subsidy for solar panels - a cornerstone in the country’s energy transition efforts. I find sizable elasticities along both margins and an optimal subsidy close to linear.
The second chapter’s title is "Optimal Case Detection and Social Distancing Policies to Suppress COVID-19." This chapter finds that the combination of case detection and social distancing is crucial for efficiently eradicating a new infectious disease. Theoretically, I characterize the optimal suppression policy as a simple function of observable statistics, which eases its implementation.
Together with the number of infected, optimal social distancing decreases over time. The fundamental trade-off is between its intensity and its duration. Quantitatively, I calibrate the model to the COVID-19 pandemic in Italy at the end of the first lockdown on May 10th, 2020. Given the observed prevalence and detection efficiency, eliminating the virus costs 11 % of annual GDP. Efficient digital contact tracing reduces the cost to 0.4 %. This cost is by one order of magnitude lower than the cost of optimal mitigation strategies.
The third chapter is joint work with Jean-Pierre Florens. Its title is "Nonparametric Identification of Supply or Demand Using Nonlinear Budget Sets." The chapter finds that kinks and discontinuities in an incentive scheme non-parametrically identify the utility-function underlying citizens’ responses. We relax the restrictive iso-elasticity assumption so far standard in the literature. The utility function is the solution to a "Schröder Equation," a functional equation not yet used in Econometrics. The result allows non-parametrically estimating the intensive margin response to taxes, subsidies, and other incentive schemes, which, in turn, allows evaluating and
optimizing these policies.