Conference Sessions - Jesse H. Jones Graduate School of ...
Conference Sessions - Jesse H. Jones Graduate School of ...
Conference Sessions - Jesse H. Jones Graduate School of ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
■ TB06<br />
Legends Ballroom VII<br />
Competition II: More on Measuring the Impact in Retail<br />
Markets<br />
Cluster: Special <strong>Sessions</strong><br />
Invited Session<br />
Chair: A. Yesim Orhun, University <strong>of</strong> Chicago, Booth <strong>School</strong> <strong>of</strong> Business,<br />
5807 Woodlawn Avenue, Chicago, IL, 60637, United States <strong>of</strong> America,<br />
yesim.orhun@chicagobooth.edu<br />
1 - Entry with Social Planning<br />
Stephan Seiler, London <strong>School</strong> <strong>of</strong> Economics, Department <strong>of</strong><br />
Economics, Houghton Street, London, WC2A 2AE, United Kingdom,<br />
s.seiler@lse.ac.uk, Pasquale Schiraldi, Howard Smith<br />
In 1996 a regulatory reform was introduced in the UK that made it more difficult to<br />
open large out-<strong>of</strong>-town supermarkets. The idea behind this planning regulation was<br />
to protect town centre vitality. In this paper we analyze the consequences for<br />
consumers and firms <strong>of</strong> alternative planning policies. We start by analyzing demand<br />
for the UK supermarket industry. The industry is characterized by stores that are<br />
differentiated along various dimensions: they <strong>of</strong>fer different ranges <strong>of</strong> product and<br />
products <strong>of</strong> different quality, they vary in size etc. As most consumers regularly visit<br />
several supermarkets in the same week it is important to carefully model the<br />
interactions between different stores. To this end we propose a demand system in<br />
which we allow consumers to visit up to two stores in each week and do not impose<br />
a priori restrictions as to whether two different types <strong>of</strong> supermarkets are substitutes<br />
or complements. We use individual level data on store choice as well as expenditure<br />
to estimate the model. In the case <strong>of</strong> two-stop shopping we also use information on<br />
how weekly expenditure is split up between the two stores. In the estimation we<br />
carefully disentangle correlation in preferences from true complementarity. Finally<br />
we use the demand estimates in order to compute supermarket pr<strong>of</strong>its which are<br />
used in a model <strong>of</strong> store entry. We explicitly model the decision <strong>of</strong> the planning<br />
authority as we have data on planning applications (both accepted and rejected<br />
ones). We simulate a counterfactual in which we remove the asymmetric treatment<br />
<strong>of</strong> large and small stores in order to analyze the effect on store pr<strong>of</strong>its and consumer<br />
welfare.<br />
2 - Sleeping with the “Frenemy”: The Agglomeration-differentiation<br />
Trade<strong>of</strong>f in Spatial Location Choice<br />
Sumon Datta, Purdue University, Krannert <strong>School</strong> <strong>of</strong> Management,<br />
403 W. State Street, West Lafayette, IN, 47907,<br />
United States <strong>of</strong> America, sdatta@purdue.edu, K. Sudhir<br />
A central trade<strong>of</strong>f in location choice is the balance between agglomeration and<br />
differentiation. Should a firm co-locate (sleep) with a competitor to increase volume<br />
(competitor is a “friend” who can draw more customers to the location with<br />
agglomeration) or locate far away from a competitor in order to reduce competition<br />
(competitor is an “enemy” from who one should spatially differentiate)? Since<br />
observed co-location may be consistent with pure differentiation rationales such as<br />
(a) high demand at the location; (b) low cost at the location and (c) restrictive zoning<br />
regulations which allow entry in only small areas, it is challenging to disentangle the<br />
agglomeration- differentiation trade<strong>of</strong>f from firms’ location choices. The paper<br />
develops a comprehensive structural model <strong>of</strong> entry and location choice that helps<br />
disentangle the agglomeration-differentiation trade<strong>of</strong>f by decomposing pr<strong>of</strong>its into<br />
revenue and cost, and then further decomposing the revenue into its components <strong>of</strong><br />
consumer choice based volume and competition based price. To capture zoning<br />
effects, we introduce a new approach to obtain zoning data, an approach that should<br />
be <strong>of</strong> general interest for a large stream <strong>of</strong> spatial location applications. Our results<br />
show that the agglomeration effect explains a significant fraction <strong>of</strong> observed colocation.<br />
Surprisingly, zoning has little direct effect on co-location. But tighter zoning<br />
restrictions interact with the agglomeration effect to explain a surprisingly large<br />
fraction <strong>of</strong> observed co-location. We find that strategic firms respond in complex and<br />
nonlinear ways to a small change in zoning which could cause a discontinuous<br />
impact on the observed location pattern.<br />
3 - Does Reducing Spatial Differentiation Increase Product<br />
Differentiation? Effects <strong>of</strong> Zoning on Retail Entry and<br />
Format Variety<br />
K. Sudhir, Yale <strong>School</strong> <strong>of</strong> Management, Yale <strong>School</strong> <strong>of</strong> Management,<br />
New Haven, CT, United States <strong>of</strong> America, k.sudhir@yale.edu,<br />
Sumon Datta<br />
Zoning regulations limit the extent to which a firm can spatially differentiate. Even<br />
though the inability to spatially differentiate can lead to lower prices, a common<br />
conjecture is that zoning can be anti-competitive because fewer retailers will choose<br />
to enter tightly zoned markets. However, retailers also have the choice to pr<strong>of</strong>itably<br />
compete by choosing a higher level <strong>of</strong> format differentiation. Using estimates from a<br />
structural model <strong>of</strong> entry and location choice in the presence <strong>of</strong> zoning restrictions,<br />
we are able to perform counterfactuals to evaluate the relative impact <strong>of</strong> zoning on<br />
both the number <strong>of</strong> firms and format variety. We find that zoning impacts entry<br />
significantly more strongly when firms cannot differentiate on formats. Also, given<br />
the ability to differentiate on formats, for large ranges <strong>of</strong> zoning restrictions, retailers<br />
respond only with changes in the format mix rather than by reducing the number <strong>of</strong><br />
firms. This implies that empirically, one may find weak linkages between zoning<br />
restrictions and entry, when retailers can differentiate on format.<br />
MARKETING SCIENCE CONFERENCE – 2011 TB07<br />
13<br />
■ TB07<br />
Founders I<br />
ASA Special Session on the Marketing-Statistics<br />
Interface - II<br />
Cluster: Special <strong>Sessions</strong><br />
Invited Session<br />
Chair: Anindya Ghose, New York University, 44 W 4th Street,<br />
Suite 8-94, New York, NY, 10012, United States <strong>of</strong> America,<br />
aghose@stern.nyu.edu<br />
1 - Assessing the Validity <strong>of</strong> Market Structure Analysis Derived from Text<br />
Mining Data<br />
Oded Netzer, Philip H. Geier Jr. Associate Pr<strong>of</strong>essor, Columbia<br />
Business <strong>School</strong>, 3022 Broadway, New York, NY, 10027,<br />
United States <strong>of</strong> America, on2110@columbia.edu, Ronen Feldman,<br />
Moshe Fresko, Jacob Goldenberg<br />
Can one analyze the information posted by consumers on the Internet to allow<br />
managers to assess market structure? Web 2.0 provides gathering places for internet<br />
users in blogs, forums, and chat-rooms. These gathering places leave footprints in the<br />
form <strong>of</strong> colossal amounts <strong>of</strong> data. This type <strong>of</strong> information <strong>of</strong>fers the firm an<br />
opportunity to “listen” to consumers in the market in general and to its own<br />
customers in particular. Our objective is to utilize the large-scale consumer generated<br />
data posted on the Web, in order to allow firms to understand consumers’ brand<br />
associative network and the implied market structure insights. We first text-mine the<br />
Web exploratory data and convert them into quantifiable perceptual association and<br />
similarity between brands and products. We use network analysis techniques to<br />
convert the text-mined data into a semantic network, which can in turn inform the<br />
firm, or the researcher, about the market structure and some meaningful<br />
relationships therein. We demonstrate this approach using two cases - sedan cars and<br />
diabetes drugs - generating market structure perceptual maps, without interviewing a<br />
single consumer. The proposed approach demonstrates high degree <strong>of</strong> internal<br />
validity. We examine the external validity <strong>of</strong> the proposed approach by comparing<br />
the market structure mined from the user-generated content to those obtain from<br />
traditional market structure approaches based on sales and survey data. The<br />
comparison to traditional market structure approaches provides strong support for the<br />
external validity <strong>of</strong> utilizing online conversations and the text mining approach to<br />
derive market structure.<br />
2 - What Drives Me? A Novel Application <strong>of</strong> the Conjoint Adaptive<br />
Ranking Database System to Vehicle Consideration Set Formation<br />
using Population Statistics<br />
Ely Dahan, Princeton University, (Visiting), Princeton, NJ,<br />
United States <strong>of</strong> America, elydahan@gmail.com<br />
A new method <strong>of</strong> individual, internet-based adaptive choice-based conjoint analysis<br />
for vehicles points to a future <strong>of</strong> highly efficient questioning with a new purpose:<br />
helping customers understand themselves. Several novel approaches underpin this<br />
method: (1) the development <strong>of</strong> adaptive choice-based conjoint based on a<br />
predefined database <strong>of</strong> utilities, (2) development <strong>of</strong> a random utility generator<br />
utilizing population statistics that acts as a simulator <strong>of</strong> the actual market, including<br />
the ability to reproduce real market shares, (3) the use <strong>of</strong> actual products as conjoint<br />
stimuli, and (4) fine-tuning the trade<strong>of</strong>f between allowing for respondent error versus<br />
enforcing consistent answers using computer assistance.<br />
3 - How is the Mobile Internet Different? Search Costs and<br />
Local Activity<br />
Sangpil Han, New York University, New York, NY, United States <strong>of</strong><br />
America, sangpil78@gmail.com, Avi Goldfarb, Anindya Ghose<br />
We explore how internet browsing behavior varies between mobile devices and<br />
personal computers. Smaller screen sizes on mobile devices increase the cost to the<br />
user <strong>of</strong> reading information. In addition, a wider range <strong>of</strong> locations for mobile<br />
internet usage suggests that the <strong>of</strong>fline context can be particularly important. Using<br />
data on user behavior at a microblogging service (similar to Twitter), we exploit<br />
randomization in the ranking mechanism for user-generated microblog posts to<br />
identify a random experiment in the cost <strong>of</strong> reading information. Using a hierarchical<br />
Bayesian framework to better control for heterogeneity, our estimates show that<br />
search costs are higher on mobile devices compared to PCs. While links that appear at<br />
the top <strong>of</strong> a page are always more likely to be clicked, this effect is much stronger on<br />
mobile devices. We also find that the benefit <strong>of</strong> searching for geographically close<br />
matches is higher on mobile devices. Stores located in close proximity to a user are<br />
much more likely to be clicked on mobile devices than PCs. We discuss how these<br />
changes may affect market outcomes in local commerce and the implications for<br />
monetization <strong>of</strong> user-generated content in social media platforms.