Industrial Organization
Applied Econometrics
Econometrics

Industrial Organization
Applied Econometrics
Econometrics
Microeconomics
Financial Economics

1998 (Oral, with Distinction) Econometrics, Industrial Organization
1997 (Written, with Distinction) Microeconomic and Macroeconomic Theory

Empirical Essays on Bargaining and Price Discrimination

Professor Ariel Pakes
Professor John Rust
Professor Steven Berry

Summer 2002

M.Phil., Economics, Yale University, 1998
M.A., Economics, Yale University, 1998
B.S.Sc. (with First Class Honors), Economics, Chinese University of Hong Kong, 1996

Yale University Dissertation Fellowship, 2001
Cowles Foundation Graduate Fellowship, 1998
Yale University Graduate Fellowship, 1996 – 1999
Nanyang Commercial Bank Ltd. Prize for Top Graduate in Economics, 1996
Head of College Creativity Prize, 1996

Teaching Assistant, Econometrics II, Yale University, 1999
Teaching Assistant, Econometrics I, Yale University, 1998
Tutor, Summer Economics Workshop, Chinese University of Hong Kong, 1997–1998

Senior Associate, Charles River Associates, 2000–2001
Associate, Charles River Associates, 1999–2000
Performed economic and econometric analyses related to mergers and acquisitions, illegal bundling, predatory pricing, exclusionary deals, and patent infringement.

Research Assistant to A. Pakes and M. Pessendorfer, Yale University, 1998 – 1999
Assessed the uncertainty in the acquisition plans of the Department of Defense and the effects of mergers activities on such plans.

Research Assistant to J. Rust, Yale University, 1997–1998
Conducted research related to Social Security Disability Insurance that leads to three joint papers.

"The Effect of Bargaining on Price and Profit: A Case Study on a U.S. Steel Service Center," manuscript, Yale University, 2000. [Job Market Paper]

"Analysis of Variance in Steel Sale Price," manuscript, Yale University, 2000.

"An Empirical Analysis of the Social Security Disability Application, Appeal, and Award Process," with H. Benitez-Silva, M. Buchinsky, J. Rust, and S. Sheidvasser, Labour Economics, 1999.

"How Large is the Bias in Self-Reported Disability Status?" with H. Benitez-Silva, M. Buchinsky, J. Rust, and S. Sheidvasser, NBER Working Paper 7526, revised and submitted to Review of Economic Studies.

"How Large are the Classification Errors in the Social Security Disability Award Process?" with H. Benitez-Silva, M. Buchinsky, J. Rust, and S. Sheidvasser, submitted to American Economic Review.

Steel service centers (SSCs) act as intermediaries between steel producers and steel buyers. In North America, SSCs form the single largest steel customer group, buying approximately a quarter of all steel produced. Like many other markets in the wholesale industry, bargaining is prevalent in the steel service center (SSC) market. SSCs bargain with both steel producers and steel buyers. The empirical literature on bargaining has remained very limited in scope and to my knowledge no work has been done on the steel intermediary industry. Taking advantage of detailed transaction level data provided by a SSC in the U.S., this dissertation analyzes the market equilibrium as a result of the bargaining process between the SSC and steel buyers.

The first essay offers background information and explores stylized facts about the SSC industry and the SSC of this study. Even though steel is a relatively homogenous good, considerable price dispersion is observed in steel sales. While cost differentials certainly explain part of the variance, other factors also contribute significantly to the price dispersion. Cost related factors by and large are time-specific. Nevertheless, a decomposition of variance in sale price shows that variance due to customer-specific sources is of the same order of magnitude as variance due to time-specific sources, signaling the presence of price discrimination.

The SSC in fact has a well-established mechanism for price discrimination among steel buyers and over time. The SSC, like its competitors, does not post or advertise price. Buyers have to call a sales agent for a price quote and they can then bargain with the sales agent. The SSC provides incentive for sales agents to bargain for a high price by giving them commissions based on the profit they make over accounting cost. There are no long-term contracts so buyers have to bargain on a transaction-by-transaction basis. Sales agents can therefore charge different prices to different customers on the same business day or to the same customer at different points of time.

In the second essay, I model the bargaining game between a sales agent and a steel buyer as an alternating-offer game with complete information. Estimation of the parameters in the bargaining model allows me to evaluate the division of trade surplus between the two parties and identify the sources for price discrimination. I find that the trade surplus is split between buyers and sales agents in a five-to-three ratio. Large quantity buyers, buyers more distant from the SSC, buyers in the Wholesale Industry, and buyers with a worse credit rating have a lower reservation value on steel and therefore get a more favorable deal. Buyers also obtain a better deal when the accounting cost is low.

The estimation results of the bargaining model also enable me to assess the effect of bargaining on the SSC’s profit. After deriving point and variance estimates of the steel buyer’s reservation value in each transaction, I estimate what the sales quantity and revenue would be if the SSC were to impose a fixed markup over accounting cost for all transactions, instead of letting sales agents determine the price for each transaction separately through bargaining. Without information on the true economic cost of steel, I exploit the differences in sales quantity and revenue over the entire sample period to calculate the critical per unit cost that will make the profit from bargaining the same as the profit from a fixed markup. Comparing the critical cost with the purchase cost of steel over the sample period, I conclude that the SSC makes a higher profit from bargaining than from a fixed markup not greater than 15%. However, for any markup rate higher than 15%, my analysis cannot determine which of the two pricing strategies is more profitable.

The SSC uses accounting cost instead of economic cost in the incentive scheme for sales agents. Although the owner and top executives of the SSC realize that economic cost instead of accounting cost should be used, they find it hard to understand and compute economic cost. To get the economic cost of steel, the SSC has to solve a dynamic programming problem that involves variables such as the spot purchase cost, the expected future purchase cost, the expected demand, and the length of delivery time. Let alone the difficulties in obtaining some of the relevant information, it is not easy to figure out and solve the dynamic programming problem.

The accounting principle that the SSC employs to calculate cost, first-in-first-out (FIFO), is in fact adopted by many businesses as a proxy of economic cost when economic cost is uncertain. While FIFO historical cost appears to be a good candidate to be used by the SSC in the incentive scheme for sales agents because bargaining over FIFO historical cost can at least guarantee no loss over a long enough period of time, there are other easy-to-use alternatives that have their own appealing aspects. The third essay (in progress) conducts simulation exercises to compare the SSC’s profitability from the current incentive scheme based on FIFO historical cost with the profitability from incentive schemes based on alternative notions of cost, such as most recent purchase cost and weighted average historical cost, under various scenarios of market structure. This analysis will shed light on whether businesses may improve their profits by adopting cost principles other than FIFO.