Home Address:
  940 State Street, Apt. 4
  New Haven, CT 06511

Telephone: (203) 464-0431 (cell)

Applied Microeconomics
Applied Econometrics
Industrial Organization
Experimental Economics

Microeconomics
Industrial Organization
Econometrics and Statistics

May 2002 (Oral): Industrial Organization, Econometrics
May 2001 (Written): Microeconomic Theory, Macroeconomic Theory

"Empirical Studies of the Effects of Information Asymmetry"

Professor Fiona Scott Morton
Professor Steven Berry
Professor Dean Karlan
Professor Donald Green

May 2006

M.A. (2003), Department of Economics, Yale University, New Haven, CT
B.S. (1996), Physics (high honors), Wesleyan University, Middletown, CT

Yale Dissertation Fellowship, Fall 2004
John F. Enders Award, Summer 2004
Yale University Fellowship, 2000–2006

Teaching Assistant, Microeconomics with Environmental Applications, Yale University, 2005
Teaching Assistant, Econometrics and Data Analysis, Yale University, 2004 (Spring and Fall)
Teaching Assistant, Introduction to Microeconomics, Yale University, 2003
Teaching Assistant, General Physics, Wesleyan University, 1995

Statistics Consultant, Statlab, Yale University, 2003–2005
Consultant to Edmunds.com, 2003
Research Assistant to Patrick Bayer and Christopher Timmins, Yale University, 2002–2003
Assistant Economist, Domestic Research, Federal Reserve Bank of New York, 1997–1999
Research Associate, National Economic Research Associates (NERA), New York, NY, 1996–1997

Professor Fiona Scott Morton
School of Management
Yale University
PO Box 208200
New Haven, CT 06520-8200
Phone: +44 (0)131-650-8361
Fax: +44 (0)131-650-4514
(visiting University of Edinburgh)
fiona.scottmorton@yale.edu

Professor Dean Karlan
Department of Economics
Yale University
PO Box 208209
New Haven, CT 06520-8209
Phone: (203) 432-4479
Fax: (203) 432-3296
dean.karlan@yale.edu

Information asymmetries threaten the efficient functioning of markets. Yet for many types of markets, such as expert services markets and used goods markets, there is an ongoing debate in the empirical literature about the extent of inefficiencies resulting from asymmetries. My dissertation informs this debate by measuring the effects of information asymmetries in markets for auto repair, used cars, and vehicle leasing.

In my first chapter, I investigate agency problems and possible mitigating factors in the auto repair market by examining data I collected in a field experiment. In my second chapter, I study whether information asymmetry hinders the efficient trading of vehicles in the used car market, and in doing so, address identification challenges that exist in the empirical literature. In my third chapter, I test whether moral hazard explains why taxicab drivers who lease their vehicles have worse driving outcomes than those who own their vehicles. This study is collaboration with the New York City Taxi and Limousine Commission.

I find large differences in the magnitude of inefficiencies across these markets. These disparities result in part from differences in the duration of information asymmetries. For example, in the auto repair market, buyers rarely ever know the quality of the service they receive, while in the used car market, buyers quickly learn the quality of their purchase. Correspondingly, I found large inefficiencies in auto repair but, in fact, none at all in the used cars market. These outcomes suggest that buyers must be able to learn about the quality of their purchase, even if after the transaction, in order for reputational factors — in the form of sellers’ pursuit of an honest reputation — to be effective at limiting agency problems.

A Field Experiment to Measure Agency Problems in Auto Repair

In many service markets, the seller of the service is also the expert who diagnoses how much service is needed. This dual relationship provides incentives for experts to supply a level of service that is not optimal for the customer. While a large body of empirical literature on agency problems in diagnosis-cure markets exists, data limitations have prevented clear conclusions about even the existence of these problems from emerging (Gaynor and Voigt (2000)).

I attempt to fill this void by conducting a field experiment in auto repair. With guidance from the Canadian consumer advocacy organization, Automobile Protection Association (APA), I prepared a test vehicle with a simple-to-diagnose defect and submitted it to garages for repair recommendations. Garages were randomly assigned to receive low and high-reputation treatments in which the undercover researcher appeared as one-time or repeat business. I implemented a flexible sample design that allowed data collection to stop after 40 garage visits. I then examined thoroughness of inspection, diagnoses, and costs for the presence of agency problems. I also tested for differences in outcomes between garages that were assigned to high and low-reputation settings. These findings were supplemented with an analysis of data on 124 APA undercover garage visits.

I find that unnecessary repairs occurred in 32% of visits and represented 52% of all charges. I also find that substantial underdiagnosis occurred in 71% of visits. When the undercover researcher represented the possibility of repeat business, upfront inspection costs were almost a full standard deviation lower. However, the possibility of repeat business had little effect on repair recommendations or overall costs. While these results confirm Hubbard’s (1998, 2002) conclusion that mechanics signal friendliness to customers, they reveal that reputation does not eliminate inefficiencies: Even in settings in which garage owners benefit from a reputation for good service — setting in which repeat visits and referrals are possible — adequate service is rare and large inefficiencies occur.

Estimating the Effects of Adverse Selection in the Used Car Market

Efficient sorting in the used car market causes cars of different quality to end up with owners who value them most highly. The efficiency of this process, however, is threatened by adverse selection, an information asymmetry that hinders a buyer’s ability to evaluate car quality. Separately identifying the effects of efficient sorting and adverse selection has proven difficult because the two effects usually occur simultaneously and both affect the same market outcomes.

I estimate the joint effect of adverse selection and efficient sorting, and find these results to be highly informative about adverse selection. Since both mechanisms increase the rate of price depreciation (Hendel and Lizzeri (1999)), their combined effect represents an upper bound on the effect of adverse selection. Finding a small joint estimate would reveal that adverse selection has an immaterial effect on this market, while a large estimate raises the possibility that adverse selection is present.

The transfer of vehicle repair costs from manufacturer to owner upon warranty expiration provides an opportunity to estimate this joint effect. By covering the cost of repairs, warrantees, in effect, equalize the condition of cars, thereby limiting the gains from trade than lead to efficient sorting. Warrantees also eliminate adverse selection by removing incentives for owners to trade low unobserved quality cars to unsuspecting buyers. Both mechanisms come into play when the warranty expires, and if either is important, the rate of price depreciation should increase.

I examine proprietary data on more than one million franchised dealer used car transactions, and find that the rate of price depreciation remains unchanged upon warranty expiration. This indicates that adverse selection is absent in the dealer segment of the market. A similar conclusion holds for vehicles traded in to dealers. An analysis of Consumer Expenditure Survey data on vehicle repairs supports the conclusion that the quality of used cars obtained from dealers is similar to equivalent continuously-held cars. Privately traded vehicles, however, had 64% more repairs after transaction than equivalent continuously-held vehicles, raising the possibility that adverse selection is important to this segment of the market.

Moral Hazard in New York City Taxicab Leasing

Since the New York City Taxi and Limousine Commission (TLC) lifted the ban on the leasing of taxicabs to drivers in 1979, the leaser-driver arrangement has displaced the owner-driver arrangement as the dominant contracting structure. There are systematic differences in driving outcomes between contract types: Leaser-drivers have almost twice as many violations and accidents per mile driven, and higher failure rates on TLC-mandated vehicle inspections, than owner-drivers.

These differences may be explained by a relationship between driving ability and contract choice, or a moral hazard in the leasing contract that encourages inferior driving behavior. This project investigates the importance of each factor in order to identify policies that would improve taxi safety and service. If driving ability explains outcomes, then policies that improve driver quality would help. If poor incentives associated with leasing are relevant, then modifying the incentive structure should be the priority.

I examine data on 250,000 individuals that have driven or owned taxicabs, including all driver records, vehicle inspection outcomes, and financial transactions recorded by the TLC. I estimate a model that shows the effect of taxicab leasing choice on two driving outcomes: driving violation rates and vehicle inspection failure rates. I control for vehicle usage and observed measures of driving ability such as age and taxi driving experience. Since unobserved driving ability may be correlated with leasing choice, I use instrumental variables to predict a driver’s leasing choice; these include taxi leasing rates among neighbors, and measures of assets and credit availability. I supplement this analysis with an examination of before and after outcomes of drivers who switch from leasing to owning. I find that the leasing contract itself has a larger effect on driving outcomes than driving ability. Incentives inherent to the leasing contract cause leaser-drivers to have 21% more driving violations and 27% higher vehicle inspection failure rates than owner-drivers.