Since the decision to adopt and become an online shopper is non random, it is likely that a sample of Internet customers is a selected one, making non adopters an inappropriate control group for them. Even within adopters, the endogeneity of the selection of the channel prevents us from estimating the effect of the Internet channel by simply comparing behavior between online and brick-and-mortar purchases. The chapters of this dissertation tackle different sides of the impact of e-commerce on economic outcomes; the unifying trait is a common theoretical framework aiming to address both of the methodological concerns mentioned above.
The first chapter provides background information on the supermarket chain that provided the data, on their online division, and on the grocery industry. I introduce the main scanner data whose features are instrumental to my identification strategy. It contains information on purchases for a panel of customers of a supermarket chain who shop with the chain both online and in traditional stores. I am in a privileged position to reduce concerns over sample selection as I am comparing the same people shopping on different channels for identical good at the same grocery chain; an almost ideal control. Moreover, observing the same households through time allows me to model the channel selection decision, addressing potential biases arising from its endogeneity. The last part of Chapter 1 sketches the theoretical framework used in the rest of the dissertation. The model allows customers behavior to be affected by selection—since agents' choice on the distribution channel they want to use for their shopping is endogenous—that has to be disentangled from the treatment effect of the online channel itself.
Chapter 2 investigates the effect of shopping online on customers disposition towards exploring new brands. I show that households are 9% less likely to purchase a cereal brand they have not tried before when they are shopping online. I estimate a fully parameterized version of the model introduced in Chapter 1, modeling consumers selecting the shopping channel where they want to buy and their demand, conditional on that decision. The results suggest that the driving force behind the reduced form result is the search for reduction of the shopping cost. The cost is particularly salient for a repeated activity involving low price items, such as grocery. This causes preference for usage of the past shopping history list feature on the website that allows reduction in cost of browsing but only allow to buy previously purchased items. The counterfactual simulations gauge the role played by website design in affecting competitive outcomes.
While Chapter 2 tries to disentangle the treatment effect of the Internet channel on demand, Chapter 3 focuses on the selection part. I study how decisions on the bundle of grocery goods to be bought affect the decision on the shopping channel for the purchase. I focus on heavy and bulky items conjecturing that they should raise the utility of shopping online, since the service offers home delivery. I show that a large, significant correlation exists between the number of heavy items purchased in a trip and the decision of shopping online. Said correlation is robust to different specifications, and holds even when product categories are considered separately. Instrumental variables results suggests that more than 30% of this correlation is due to selection: that is to households selecting the online channel because they want to buy heavy items. I conclude by investigating the implications of this finding on price elasticity and showing that the complementarity between online purchases and heavy product categories can lower price elasticity for such items online.
Finally, Chapter 4 relies on a reduced form representation of the model proposed in Chapter 1 to ask how much of the sales generated by the online channel are just cannibalization of sales which would have occurred anyway in the traditional stores. The detailed information on usage of online shopping allows to pursue the research question without incurring in sources of measurement error that have plagued previous contributions. I show that the amount of crowding out is non negligible and can be quantified in about 30 cents for each dollar spent online. This still leaves a margin big enough to justify the viability of the hybrid retailer business model. I show that the amount of crowding out is heterogeneous with respect to the different groups of households and product categories. This exercise allows to answer questions that are crucial for the retailer, such as quantifying the revenue value of the online channel, as well as providing insights for regulators by documenting the degree of substitutability between the Internet and the traditional sector. (Abstract shortened by UMI.)
|School Location:||United States -- California|
|Source:||DAI-A 70/07, Dissertation Abstracts International|
|Keywords:||Brand trial, Cannibalization, E-commerce, Grocery industry, Panel data, Price elasticity, Structural model|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be