Published/Forthcoming Papers

Estimating the Impact of School Closings on Parental Choice (with Dennis Epple and Holger Sieg
Quantitative Economics, Forthcoming

Many urban districts confront the necessity of closing schools due to declining enrollments. To address this important policy issue, we formulate a sequential game of managing a school district's student capacity. We show that a Perfect Bayesian Equilibrium exists and characterize its properties. Using data for a mid-sized district with declining enrollments, we estimate the parameters of our model. We show that consideration of student sorting is vital to the assessment of any school closing policy. We find that the district can reduce excess school capacity by closing underperforming schools. This approach inevitably leads to disruptive displacement of students in closed schools. Moreover, a significant fraction of students leaves the district. We also show that superintendents confront a difficult dilemma: pursuing an equity objective, such as limiting demographic stratification across schools, results in the exit of many more students than are lost by an objective such as closing underperforming schools. 

Does Environmental Policy Affect Scaling Laws Between Population and Pollution? Evidence from American Metropolitan Areas
 (with Nick Muller)
PLOS ONE, Forthcoming

Working Papers


Electric utilities under output price regulation purchase most of their input coal from contracts, paying contract prices higher than prevailing spot prices.  I posit that price regulation punishes extreme cost realizations, inducing regulated firms to value both a lower mean and a lower variance in total costs.  I show descriptively that spot price volatility is a determinant of plant-level coal contracting behavior.  I find that plants trade-off a $1.66 increase in expected costs for a $1 decrease in the standard deviation of costs; the benefits from lower volatility input costs would have to be substantial to justify this mean/standard-deviation trade-off.

Burning coal is known to have environmental costs; we instead quantify the environmental costs of transporting and storing coal at U.S. power plants. We first demonstrate that a 10% increase in coal stockpiles (number of deliveries) results in a 0.06% (0.12%) increase in the average concentration of fine particulates (PM2.5) within 25 miles of plants.  Using coal stockpiles and deliveries as instruments, we find that a 10% increase in PM2.5 leads to a 1.1% (6.6%) increase in average adult (infant) mortality rates. This increase in mortality rates implies environmental costs of $183 ($203) per ton of coal stockpiled (delivered).

Dynamic Regulatory Distortions: Coal Procurement at U.S. Power Plants


With risk neutral traders and zero transactions costs, the expected value of the difference between the current forward price and the spot price of a commodity at the delivery date of the forward contract should be zero. Accounting for the transaction costs associated with trading in these two markets invalidates this result. We develop statistical tests of the null hypothesis that profitable trading strategies exploiting systematic differences between spot and forward market prices exist in the presence of trading costs. We implement these tests using the day-ahead forward and real-time spot locational marginal prices from California's wholesale electricity market and use them to construct an estimate of the cost of trading in this market. During our sample period, we observe the introduction of convergence bidding, which was aimed at reducing the costs associated with exploiting differences between forward and spot prices. All of our measures of trading costs are significantly smaller after the introduction of convergence bidding. We also find that the mean of trading costs is lower for generation nodes relative to non-generation nodes before explicit virtual bidding. However, mean trading costs fell more for non-generation nodes after explicit virtual bidding, eliminating any difference in mean trading costs across the two types of nodes. We also present evidence that the introduction of convergence bidding reduced the total amount of input fossil fuel energy required to generate the thermal-based electricity produced in California and the total variable of costs of producing this electrical energy. Taken together, these results demonstrate that purely financial forward market trading can improve the operating efficiency of short-term commodity markets.

Economics and Externalities of Moving Crude Oil By Pipelines and Railroads: Evidence from the Bakken Formation (with Karen ClayNick Muller, and Randall Walsh)

In this paper, we derive and implement an empirical test for the Pareto efficiency on a nonlinear tax schedule. Building on the theoretical framework of Boadway and Sato (2011) we develop an estimable model of optimal taxation under uncertainty. We derive an inequality determining whether a given tax schedule is Pareto efficient (Werning, 2007). Empirically, we use our framework to evaluate the efficiency properties of the tax resulting from Tax Reform Act of 1986. We combine data on individual's tax returns from the SOI public use files with data on income expectations from the Survey of Consumers in order to structurally estimate this model. Surprisingly, we find that the Pareto efficiency of a tax schedule is more likely to be violated at very low incomes rather than very high incomes.

Works in Progress

A Portfolio Approach to Convergence Bidding: Evidence From California (with Frank Wolak)

Valuing Dispatchable Generation Units in Markets with Significant Intermittent Renewable Energy Shares  (with Frank Wolak)