Head Teaching
Assistant Yale University
Econ 161b: Econometrics and Data Analysis, Spring 2000
Coordinated and led a team of graduate teaching
assistants.
Managed coverage, presented course material in
weekly sections, evaluated coursework and exams.
Teaching Assistant for Yale Graduate Course
Econ 558a: Graduate Class in Statistics and Econometrics, Fall 1999
Taught new material in weekly sections,
clarified communication between students and professor,
assessed coursework.
Mentored and assisted students individually. |
Yale
University
Research Assistant, Professor Martin Pesendorfer, Yale University Department of Economics,
1999-2000, on the project, "Bidding Behavior in a Repeated Procurement Auction"
Analyzed highway procurement auctions in California to asses efficiency
losses due to bidder asymmetry
and intertemporal effects.
Research enables government agencies to enhance design of their
auctions and bring in better prices
Stanford University
Research Assistant, Professor Thomas Hellmann and Professor Manju Puri, Stanford Graduate
School of Business, 1996-1997, member of the Venture Capital Research Team
Assessed information needs. Designed Macros in Excel to enable
efficient data extraction.
Educated new research assistants.
Research tracked the organizational development of 173 startup
companies in Silicon Valley and analyzed
the effects of firm financial structure to the timing and
success of its IPO. |
Since 1984,
AT&T’s market share of long distance telephony has decreased from 90.1% to 37.9%.
Such a change could only have taken place with the help of legislation and the competitive
opposition of smaller, new entrants. This dissertation examines the competitive tactics of
the long distance telephony fringe. Using an unusually detailed original dataset, it
uncovers new evidence of price discrimination, it models heterogeneous switching costs for
subscribers, and it determines the underlying causes of tenure dependence on consumer
switching behavior. The findings provide implications for optimal firm response in a
changing environment and for the efficacy of telecommunications market policy.
The first paper reviews certain salient attributes of long distance telephony before, and
immediately after, the 1996 Telecommunications Act. This legislation allows new entrants
to purchase facilities, to lease lines or facilities, and to purchase resalable voice
minutes for a reasonable wholesale cost. The resulting reduction in cost of capital lowers
the barriers to entry. The subsequent ease of market entry following 1996 paved the way
for many new entrants, including a certain interesting subset of aggressive new firms.
This subset makes up what I call the competitive resale fringe. These resellers lease time
from other carriers’ networks for resale on the retail market. Practically
nonexistent seven years ago, resale firms have become a major influence in increasing
competition within the market. The new environment has been unusually favorable for new
entrants—at least for the short term. Even though both consumer price and wholesale
cost of long distance service is temporarily declining, retail prices have been sticky and
have decreased at a slower rate. This unique combination has provided a convenient
disguise for certain competitive tactics.
One tactic that has enabled resale firms to compete so effectively in the current market
is price discrimination. This first paper finds new evidence of price
discrimination. The paper explains findings that may only be uncovered by utilizing a data
set that contains individual prices, costs, and a large number of observations. An
original detailed data set from a proprietary company (containing retail price, cost, and
consumer characteristics for each and every phone call from over 180,000 customers during
1998 to 2000) is used. Zip Code census data are added to enhance demographic
characterization. Price dispersion is analyzed in a heteroskedastic-consistent fixed
effects model against subscriber demographics, while accounting for cost effects, time
effects, and individual credit records. Results indicate that price variation is partially
due to customer demographics, such as income and location, and therefore imply 3rd
degree price discrimination. Interviews with firm management corroborate the empirical
results by explaining that the long distance resale segment has a well-established
mechanism for price discrimination among prospective subscribers. Here, sales agents
telemarket. They offer different prices based upon location demographics, as well as upon
answers given to telemarketing questions (including willingness to pay and ability to
search for alternative carriers).
Once the customer is signed up, switching costs discourage subscribers from leaving, even
as market-wide prices decrease. In an environment of decreasing costs, it is unnecessary
to raise prices for tenured subscribers in order to create price discrimination. Carriers
need only to offer prospective subscribers a lower price (while sustaining stable price
levels for old subscribers). In addition, because small resale firms have a lower profile
than do major carriers, resale subscribers are generally less informed about their
carrier’s pricing than are subscribers of major carriers. This circumstance gives
resale firms an advantage for price discrimination. As a result, price discrimination is
an important tactic contributing to a small resale firm’s ability to extract the
profits necessary to stay in business and to compete effectively against the much larger
incumbents.
The second paper develops a model of subscriber switching behavior. Each individual
subscriber’s departure is used to predict the price differential necessary to lure
subscribers away, or alternatively, to prevent subscribers from leaving (given subscriber
specific cost factors, demographics, and credit history). Subscribers optimize each period
as both market-wide costs and market-wide prices change. I develop an original method
(called switch indexing) to account for the effects of separate price differentials
in various sectors of the telephony market (knowing that some subscribers would look to
premium prices, like AT&T’s, as a possible alternative, while other subscribers
would look to other small resale firm prices as possible alternatives). Since I observe
subscriber movement to and from our observed firm, but do not observe specific
originations or destinations of subscribers, I adapt the likelihood framework to optimize
over the binary actions I do observe. The model has the unique advantage of being able to
focus more on time-specific cost differentials, rather than being constrained to assume a
parametric function to characterize expected change in the environment. This time
flexibility feature is important for the long distance telephony market, where cost and
relative price velocity is quite volatile. In addition, this framework allows for more
facile forecasting in a market with environmental volatility. More accurate forecasting
can enable both better public policy analysis and more effective individual firm response.
The second paper goes on to capture switching cost heterogeneity among subscribers.
Here the paper introduces a mixture model. I begin by modeling consumers as one of
two types: (1) a subscriber with low switching costs or (2) a subscriber with high
switching costs. Two separate sets of parameters are recovered, as well as the
distribution probabilities. Alternative specifications are explored. The extensions
include increasing the number of subscriber types and allowing distribution probabilities
to vary as a function of consumer demographics and behavior.
The third paper (in progress), accounts for tenure dependence and isolates its
origin. Tenure dependence can result from two different causes: (1) relationship effects
and/or (2) selection. First, tenure dependence can occur when a subscriber’s
preference for a carrier increases over time as a relationship is developed.
Alternatively, the distribution of subscribers may shift over time to include more
subscribers with either higher switching costs or a greater matching preference for the
firm. Since these two separate causes have markedly different implications for consumer
behavior, it is important to distinguish them empirically. Due to an unusual amount of
variation in the data, this paper will be able to decipher between the two motivations by
examining consumer reaction to time-varying relative prices (each producing differing
patterns of movement over each consumer portfolio).
The estimated models allow examination of pro-competitive policies. The analysis of the
survival tactics of small resale firms improves our understanding of the effectiveness of
regulation such as the 1996 Telecommunications Act and the Robinson-Patman Act (which
prohibits price discrimination for the purpose of competition enhancement). We also
discover how fringe firm competitive tactics may be fettered under a change in environment
(like the inevitable stabilization of costs). As a result, this study of resale fringe
activity provides insights that are useful for both individual firm response and for
telecommunications market policy. |