Project Details
Projekt
Measuring Consumer Behavior Using Big Data
Applicant
Professor Dr. Florian Heiß
Subject Area
Economic Policy, Applied Economics
Statistics and Econometrics
Statistics and Econometrics
Term
since 2023
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 462020252
In classical discrete choice models, consumers are assumed to make a comprehensive and careful choice between the available alternatives in terms of their utility consequences. It is well known that this is not an accurate description of the real-world consumer. Brand loyalty and inertia are often found to have a substantial on consumer choice. If inertia and other deviations from strict rationality are strong determinants of daily shopping decisions, this has important consequences for (i) understanding consumers themselves, (ii) the validity of standard choice analyses, and (iii) the strategic behavior of producers and retailers. Our goal in this project is to shed light on all of these aspects. While previous literature mostly focusses on single product categories, the fact that during a shopping trip consumers make dozens of choices among different categories suggests great opportunities to identify a more wholistic picture of behavior from a wealth of repeated and interrelated decisions and large amounts of data. By building on techniques from the machine learning literature, demand can be modelled simultaneously for many product categories. The approach is based on splitting behavioral parameters across individuals and categories. This allows the individual parameters to be identified from choice behavior in all categories at the same time: If we observe many yoghurt purchases of a household, this helps to identify the individual behavior for toothpaste. We will study various aspects of joint decision making using a large consumer panel data set by Nielsen. How often do people automatically pick the same kind of cereal or tooth paste without a deliberate choice? How loyal to a brand are they? How much do these kinds of effects differ across individuals? Is this an individual trait which is correlated over product categories? How much are future choices affected by current shocks such as advertising or product availability? How does a consumer react to bundling and similar store strategies? Finally, we will use the developed methods and behavioral insights to study the strategic behavior of producers and retailers in highly concentrated markets and discuss whether there is any role for legislation or regulation.
DFG Programme
Research Units
