Dokument: Three Essays on Empirical Industrial Organization in Grocery Retailing

Titel:Three Essays on Empirical Industrial Organization in Grocery Retailing
URL für Lesezeichen:https://docserv.uni-duesseldorf.de/servlets/DocumentServlet?id=43467
URN (NBN):urn:nbn:de:hbz:061-20170918-154917-9
Kollektion:Dissertationen
Sprache:Englisch
Dokumententyp:Wissenschaftliche Abschlussarbeiten » Dissertation
Medientyp:Text
Autor: Lu, Anna [Autor]
Dateien:
[Dateien anzeigen]Adobe PDF
[Details]1,18 MB in einer Datei
[ZIP-Datei erzeugen]
Dateien vom 14.09.2017 / geändert 14.09.2017
Beitragende:Prof. Dr. Duso, Tomaso [Gutachter]
Prof. Dr. Haucap, Justus [Gutachter]
Dewey Dezimal-Klassifikation:300 Sozialwissenschaften, Soziologie » 330 Wirtschaft
Beschreibungen:In the first essay, titled “Retail Mergers and Assortment Repositioning”, I study how supermarkets compete in product variety and what this means for merger control. Authorities previously focused their attention on price effects of mergers. However,
the recent years saw a shift in attention: For example, the new U.S. Horizontal Merger Guidelines (2010), issued by the Department of Justice and the Federal Trade Commission, now emphasize the importance of merger effects on product variety. So far, the literature has found ambiguous effects of mergers on store level assortment,
i.e. the average number of products carried by a store (Pires and Trindade 2015; Argentesi et al. 2016).
I extend the analysis to look at strategic repositioning between stores. I study a series of local mergers in the U.S. and find that, while mergers increased store-level assortment by 7.4%, they also increased the assortment overlap between stores by
3%, suggesting that gains from business-stealing exceeded losses from cannibalization. Effectively, a merger raised the number of available products in the average market by 6.9%, i.e. mergers, on average, improved consumer access to variety. The analysis is based on a data-driven market definition: Unlike the existing literature, I
use machine-learning methods to cluster stores into local markets according to their geographic location, thus yielding highly flexible markets. All in all, the findings in this chapter support the notion that product variety is an important factor in merger control. Merger effects on store-level assortments tell only one part of the story – in order to get the full picture, regulators should also look at the merger effect on substitutability between stores. This is particularly relevant in industries in which consumers switch frequently between stores and in which store assortment is one of the major drivers of shopper patronage (Hoch, Bradlow and Wansink 1999; Fox, Montgomery and Lodish 2004; Briesch, Chintagunta and Fox 2009).

The second essay, titled “Consumer Stockpiling and Sales Promotions”, looks at consumer stockpiling and retailer pricing in storable goods markets. It builds on the well-established finding that consumers tend to stockpile for future use when they face price promotions, anticipating that prices will go up in later periods. While
the literature has developed sophisticated models to describe and estimate such forward-looking consumer behavior (e.g. Erdem, Imai and Keane 2003; Hendel and Nevo 2006; Seiler 2013), there is very little empirical work on how retailers should respond to it, and in particular, how they should best design promotions. I am the first to compare the effects of promotion length with the effects of
promotion depth on long-run consumer purchases and seller revenues. I study the U.S. laundry detergent market which is characterized by frequent price promotions and substantial consumer stockpiling. I estimate a dynamic discrete-choice model of strategic stockpiling in which consumers are forward-looking with rational expectations over future prices. Consumers can keep storage but incur storage costs. I use the estimates from this model to simulate pricing policies of varying promotion length and depth. I find that, in the detergent market, the revenue elasticity with respect to promotion depth is about four times larger than with respect to promotion
length. My results suggest that retailers should increase promotion depth rather than length in markets with steady consumption rates, few demand shocks, and large heterogeneity in storage costs and price sensitivity.

The third essay, titled “Inference of Consumer Consideration Sets”, investigates how consumers make decisions when they face a large number of alternatives and how the underlying, typically unobserved decision process can be inferred from observed choice data. It is motivated by the fact that, in the modern market place, consumers
often have to choose from an overwhelmingly large variety of products. In order to simplify this decision problem, consumers tend to reduce the number of objectively available alternatives to a subset of relevant alternatives. The literature refers to this subset as a “consideration set”. In structural models of demand, we typically have
to make assumptions about consideration sets because they are usually unobserved. However, these assumptions are not innocuous: If the consideration model is misspecified, the demand estimates can be severely biased (Sovinsky 2008; Draganska and Klapper 2011; Conlon and Mortimer 2013). I propose a novel framework to formally test any two competing models of consideration against one another in order to determine which model fits the data best. My test follows the intuition of a menu approach (e.g. Nevo 2001; Villas-Boas 2007)
and uses supplemental data on marginal cost shifters to construct overidentifying restrictions. To illustrate my approach, I show an application to German grocery retailing in the categories of coffee and milk. Specifically, I test two popular models against each other. The first model is the workhorse model in empirical industrial organization; it assumes that consumers consider all products in the market. The
second model is more popular in quantitative marketing research and models a two-stage consideration process in which consumers first choose a store and then choose a product within this store. I find that consideration sets differ fundamentally across product categories: The two-stage model outperforms the model of global consideration
in the milk category but not in the coffee category. I relate this finding to differences in demand and supply conditions of the two product markets.

Ph.D. thesis
Lizenz:In Copyright
Urheberrechtsschutz
Fachbereich / Einrichtung:Wirtschaftswissenschaftliche Fakultät » Volkswirtschaftslehre
Dokument erstellt am:18.09.2017
Dateien geändert am:18.09.2017
Promotionsantrag am:22.05.2017
Datum der Promotion:16.08.2017
english
Benutzer
Status: Gast
Aktionen