M.Phil. (1998),
Economics, Yale University
M.A. (1997), Economics, Yale University
B.A. (1991), Economics, Seoul National University |
Teaching
Assistant, Introductory Microeconomics, Yale University, Fall 2000, Spring 2001, Fall
2001, Spring 2002
Mathematical Economics, Yale University, Spring 2000, Spring 1999
Theory of Resource Allocation and its Applications, Yale University, Fall 1998 |
The last several
years have witnessed the substantial growth of the Internet and the consequent expansion
of e-commerce. Some believe that the Internet has created a cyberspace, where buyers and
sellers are searched and matched with little friction. The notion of cyberspace seems to
make economic geography quite irrelevant. This dissertation studies the determinants of
Internet retailer location in the age of the Internet, which many think is diminishing the
importance of geographic location.
Does location matter for Internet retailers? The first chapter tries to answer the
question by developing a simple model of one Internet retailer and two states, where the
loophole in sales tax collection and shipping costs play a central role in the
retailer’s locational choices. It is common for Internet retailers not to collect
sales taxes on out-of-state purchases. This loophole in sales tax collection has been in
the center of debates among state authorities who are primarily concerned about the
shrinking sales tax base caused by the rapid growth of e-commerce. Despite the importance
of the sales tax loophole, there has been little research on the issue until very
recently. Goolsbee (2000) finds that, controlling for observable characteristics, people
living in high sales tax locations are significantly more likely to buy online to avoid
sales taxes. In contrast to his study on consumer side, this study investigates the role
of the sales tax loophole in Internet retailers’ decisions, focusing attention on
location choices. The basic model developed in this chapter shows that the comparison of
sales taxes and shipping costs is the key to the location choices and that the location
choices can be quite different from the ones without the loophole. In a case where demand
conditions are equal and state sales tax rates are different, the model suggests that
locating in the state with lower sales tax rate is optimal.
Ongoing debates on taxations over Internet transactions are causing substantial
uncertainty among economic agents. Based on the model, the effect of taxation over
Internet purchases on location choices and consumer welfare is studied. Adding dynamic
components into the base model, this chapter studies the role of uncertainty about the tax
policy over the Internet on the retailers’ location choices. Under a setup where a
chance of relocation exists after the announcement of tax policy, the joint role of
relocation cost and the probability of taxation is found to be crucial in determining
initial locations and relocations.
The second chapter empirically verifies the relevance of key variables found in the
location model provided in the first chapter, and describes the overall location pattern
of the Internet retail industry. The chapter begins with the introduction of a conditional
logit model of location choice. The model is applied to original data that includes the
frequency distribution of Internet retailers at state level, population, per capita
income, sales tax rate, a proxy for average transportation costs, and land area. In the
first part of the empirical analysis, a comparison between the location pattern of
Internet retailers and that of local retailers is made within the computer retail
industry. The purpose of the analysis is to check if the Internet indeed generates a
different location pattern in retailing. The second part compares the location pattern of
Internet retailers across product categories such as apparel, book, home and garden,
electronics, and computer. In the last part, using the predicted probability of new births
calculated from logit estimates, the scope of geographic concentration is discussed.
The logit estimates show that Internet retailers are attracted to states with larger
population and higher per capita income. While the coefficients on population are close to
one, which means that the new births of Internet retailers are roughly unit elastic with
respect to population, the coefficients on per capita income ranges from three to six. The
sales tax rate and average transportation costs turn out to be significant and of the
predicted sign in some product categories. Some of the other major empirical findings are
as follows. First, Internet retailers are heavily concentrated in California. Second,
bipolar clustering of Internet retailers into the East Coast and the West Coast is common
for most product categories with the exception of home and garden category where Midwest
region has a strong presence. Third, states located closer to the large population centers
attract more Internet retailers. For example, New Jersey, Connecticut, Massachusetts, and
Pennsylvania are near New York and they have a higher probability of new births of
Internet retailers.
My dissertation sheds light on factors that managers of Internet retailers should take
into account in their location choices, for example, in selecting locations for their
warehouses. In policy making, particularly related to the taxation of Internet purchases,
the dissertation suggests that the current location pattern and the possibility of
relocation after the taxation should be an important consideration in understanding the
welfare consequences.
In work currently in progress, attention is focused on the warehouse locations of Internet
retailers. This work will study the determinants of warehouse location and compare them to
those of headquarter locations of Internet retailers. |