Mike Palazzolo. Ph.D. in Business Administration Marketing Expected 2016 Stephen M. Ross School of Business, University of Michigan

Transcription

1 Mike Palazzolo University of Michigan, Stephen M. Ross School of Business 701 Tappan Street R5323, Ann Arbor, MI Mobile: Education Ph.D. in Business Administration Marketing Expected 2016 Stephen M. Ross School of Business, University of Michigan Master of Business Administration 2009 Graduate School of Management, University of California, Davis Bachelor of Science Management Science 2004 University of California, San Diego Research Interests Consumer search Consideration set formation Intertemporal substitution Bayesian estimation of discrete choice models Manuscripts Under Review & Working Papers Palazzolo, Mike and Fred M. Feinberg Modeling Consideration Set Substitution Job Market Paper. Revising for invited resubmission to Marketing Science. Orhun, A. Yeşim and Mike Palazzolo. The Constraining Effect of Poverty Working Paper. To be submitted Fall Additional Work in Progress The First is Always the Toughest: The Managerial Implications of Search Cost Savings. Co-authored with Fred M. Feinberg. Data analysis in progress. Leveraging Consumer Uncertainty to Improve Technology Adoption Forecasting. Co-authored with Koert Van Ittersum and Fred M. Feinberg. Data analysis in progress.

2 Dissertation Search Costs and their Implications for Consideration Set Substitution Committee: Fred Feinberg (Chair), Yeşim Orhun, Rich Gonzalez, Peter Lenk, Ryan Kellogg Proposal defended on April 7, 2014 In categories with broad competitive landscapes, consumers may not expend the effort necessary to learn about all alternatives. They may instead construct a consideration set of alternatives to search for information about in an effort to balance the expected benefits of search (e.g., finding a high quality product or low price) with the costs (e.g., time needed to collect information or the mental cost of evaluating alternatives). A wealth of literature has demonstrated that modeling search and consideration set formation is critical to accurately measuring important quantities such as price elasticities. Yet doing so can be challenging, particularly for markets with a large number of alternatives the very markets for which consumers likely need to limit their search. Many models used in the literature consequently abstract away from an important aspect of the consumer search process that the net (expected) benefit of search may not be constant in the number (or type) of alternatives considered. This dissertation investigates the importance of accounting for two ways in which the net expected benefit of (simultaneous) information search for an alternative may change as a consequence of how many and which other alternatives are also being considered. Essay One: Modeling Consideration Set Substitution If the marginal expected benefit from search decreases in the number of considered alternatives, marketing actions that encourage consideration of one alternative may discourage consideration of another. This paper develops a model of consideration set formation that can account for this consideration set substitution. Using stated consideration set and observed purchase data from the automotive industry, we measure the impact of marketing actions and market events on consideration and purchase. We benchmark our model against one that is commonly used in the literature but that does not account for consideration set substitution. We show the benchmark model misestimates the impact of studied marketing actions and events on market share by as much as 13%. Additionally, although the benchmark model appears to fit the data well on many dimensions, it does not capture the distribution of consideration set sizes well. This causes it to misestimate quantities that are a function of consideration set size, such as price elasticities (which are underestimated by nearly 10%). Essay Two: Managerial Implications of Search Cost Savings Within certain subsets of alternatives (e.g., those from one manufacturer s product line), some costs associated with search need not be repeated if a consumer searches for information about multiple alternatives from that subset. For example, a consumer who learns about the iphone may not need to expend as much effort to learn about the ipad (which shares features in common with the iphone) as s/he would to learn about a tablet from another manufacturer. Similarly, the incremental opportunity cost of time spent to test drive a Honda Civic would be less substantial for a consumer already planning to visit a Honda dealership to test drive the Accord than for a consumer who was not. This essay investigates the latter of these (search) cost saving examples. It extends the model from Essay One to account for the fact that alternatives within the same dealership (or automall) may be considered together more often because the cost of considering N vehicles from one location are lower than the cost of considering N vehicles from different locations. We will examine how (and the degree to which) accounting for such cost savings can improve measurement of substitution patterns stemming from marketing actions and market events. These improvements are derived from the model s ability to recognize that an increase (decrease) in consideration frequency for an alternative may contribute to an increase (decrease) in consideration frequency for others at the same location.

3 Honors and Awards 2015 Haring Symposium Presenter, Indiana University Honorable mention for best presentation 2014 AMA-Sheth Foundation Doctoral Consortium Fellow, Northwestern University 2014 ISMS Marketing Science Doctoral Consortium Fellow, Emory University 2013 ISMS Marketing Science Doctoral Consortium Fellow, Özyeğin University 2012 Milton G. Kendrick and Josephine H. Kendrick Award, Ross School of Business Marketing Dept. Award for excellent second-year research paper Conference Presentations 2015 Mike Palazzolo* & Fred Feinberg, Modeling Consideration Set Substitution. ISMS 37th Marketing Science Conference in Baltimore, MD Mike Palazzolo* & Fred Feinberg, Modeling Consideration Set Substitution. Haring Symposium in Bloomington, IN Mike Palazzolo* & Fred Feinberg, A Search Cost Model of Consideration Set Formation. ISMS 36th Marketing Science Conference in Atlanta, GA Mike Palazzolo & Fred Feinberg*, The First is Always the Toughest: The Managerial Implications of Search Cost Savings. ISMS 36th Marketing Science Conference in Atlanta, GA Mike Palazzolo* & Fred Feinberg, Forecasting Substitution Patterns when Consumers Diversify Their Consideration Sets. ISMS 35th Marketing Science Conference in Istanbul, Turkey. * = Presenting author Teaching Experience Teaching Assistant, Ross School of Business, University of Michigan 2015 Marketing 630: Marketing Engineering 2013 Graduate Student Instructor, Ross School of Business, University of Michigan Marketing 300: Marketing Management Teaching Rating: 4.9/5.0 (average across all instructors: 4.5/5.0) 2010 Guest Lecture, Graduate School of Management, University of California Davis MGT 293: Product Management (invited by Professor David Bunch) Excelling in the Markstrat Simulation 2008 Teaching Assistant, Graduate School of Management, University of California Davis Management 203A: Data Analysis for Managers (Introduction to Statistics)

4 Service 2014 Current Reviewer for Marketing Science PhD Forum Officer (student government for Ross School of Business doctoral program) Industry Experience 2008 Palantir Technologies: Business Development Intern Stockton Ports Baseball Club: Director of Special Projects Fresno Grizzlies Baseball Club: Intern, Project Manager, Marketing Manager Doctoral Coursework Marketing Choice Theory and Modeling Fred Feinberg Empirical Modeling using Bayesian Methods Anocha Aribarg Structural Models Srinivasaraghavan Sriram Quantitative Methods Puneet Manchanda Behavioral Research in Advertising Rajeev Batra Social Influences in Consumption David Wooten Learning and Memory Carolyn Yoon Consumer Sensation and Perception Aradhna Krishna Economics Microeconomic Theory Yusufcan Masatlioglu Game Theory Daisuke Nakajima Econometric Analysis Lutz Kilian Applied Microeconometrics Jeff Smith Industrial Organization & Public Policy I Ying Fan & Natalia Lazzati Industrial Organization & Public Policy II Jeremy Fox & Dan Ackerberg Statistics Applied Bayesian Inference Tim Johnson Advanced Topics in Biostats: Bayesian Statistics Tim Johnson Generalized Linear Models Douglas Schaubel Computation Methods in Statistics Yves Atchade

5 References Fred Feinberg Handleman Professor of Marketing Professor of Statistics University of Michigan Ross School of Business 701 Tappan Street, R5324 Ann Arbor, MI Ph: (734) Yeşim Orhun Assistant Professor of Marketing University of Michigan Ross School of Business 701 Tappan Street, R5460 Ann Arbor, MI Ph: (734) Rich Gonzalez Professor of Pyschology Professor of Marketing Professor of Statistics University of Michigan Department of Psychology 530 Church St. Ann Arbor, MI Ph: (734)

6 Abstracts for Working Papers and Works in Progress The Constraining Effect of Poverty Households can save on everyday storable items by choosing cheaper brands, buying larger packages that offer quantity discounts, and/or taking advantage of temporary price cuts. We expect lower income households to be more incentivized to take advantage of these strategies. However, studying the purchases of toilet paper across a panel of nationally representative households, we find that compared to higher income households with the same total consumption levels, lower income households buy smaller packages in general, and stockpile less when purchasing on sale. Therefore, they hold a smaller inventory at a higher per-unit-cost given their brand choices. We also find that lower income households are less able to strategically wait or accelerate their purchase timing in order to take advantage of a sale. We argue that liquidity constraints provide a unifying explanation of these results. We show that lower income households buy larger packages, stockpile more, and accelerate their purchases in order to take advantage of a sale during the first week of the month, when their liquidity constraints are exogenously and temporarily relaxed by the arrival of food-stamps and pay-checks and that higher income households do not. These results cannot be explained by differences across income groups in myopia, storage costs, or access to products, none of which differ by time of the month. Leveraging Consumer Uncertainty to Improve Technology Adoption Forecasting Uncertainty has been shown by previous research to be a critical element of consumers development of purchase expectations and intentions. Our goal is to develop a model that incorporates how consumers uncertainty about their purchase expectations and intentions changes over time. Consider an example: A consumer is highly uncertain about whether they will adopt a new technology in the next few months, but is fairly certain they will do so before the end of the year. Econometric models built upon the expectations of consumers like this one may benefit from a more detailed treatment of those consumers uncertainty. Though some work has utilized consumers reported uncertainty about their expectations of future purchase, as of yet no research known to the authors has examined how consumers uncertainty about their behavioral expectations may shift over time, or how such timevarying information about uncertainty might help improve product adoption forecasts constructed from consumer self-reports. We ask consumers to provide a holistic appraisal of their intention to adopt a new product (the ipad) or a new service (a bicycle service near a large university in the southern United States), and their corresponding uncertainty, over a two-year period of time. This appraisal comes in the form of an "optimistic" (upper bound) and "pessimistic" (lower bound) forecast of their probability of adopting the product or service. We find that the "optimistic" forecasts strongly match forecasts from a separate group of consumers who were either asked to provide an independent, time-invariant measure of their uncertainty, as well as the forecasts from a group that were not asked for any uncertainty measure at all. In other words, consumers that are not asked for both an upper- and lower-bound forecast tend to report an upper-bound forecast, not an average of the two. We aim to test if an econometric model built upon our more detailed reporting scheme outperforms more common approaches that ask for certainty measures that are not represented as bounds and that are invariant over time.