Advertising, consumer awareness, and choice: evidence from the U.S. banking industry

• Author: Ali Hortaçsu,Elisabeth Honka,Maria Ana Vitorino
• Published date: 01 August 2017
• Date: 01 August 2017
• DOI: http://doi.org/10.1111/1756-2171.12188

RAND Journal of Economics

Vol.48, No. 3, Fall 2017

pp. 611–646

Advertising, consumer awareness,

and choice: evidence from the U.S. banking

industry

Elisabeth Honka∗

Ali Hortac¸su∗∗

and

Maria Ana Vitorino∗∗∗

How does advertising influence consumer decisions and market outcomes? We utilize detailed

data on consumer shopping behavior and choices over bank accounts to investigate the effectsof

advertising on the different stagesof the shopping process: awareness, consideration, and choice.

Weformulate a structural model with costly search and endogenous considerationsets, and show

that advertising in the U.S. banking industry is primarily a shifter of awareness as opposed to

consideration or choice. Advertising makes consumers aware of more options, search more, and

find better alternatives. This increases the market share of smaller banks and makes the industry

more competitive.

∗University of California Los Angeles; elisabeth.honka@anderson.ucla.edu.

∗∗University of Chicago and NBER; hortacsu@uchicago.edu.

∗∗∗University of Minnesota; vitorino@umn.edu.

Weare grateful to Frederico Belo, Mark Bergen, Thomas Holmes, Mitch Lovett, John Lynch, Amil Petrin,Enno Siemsen,

Aaron Sojourner, and Joel Waldfogel for their numerous comments and suggestions. We also thank the participants of

the 2013 Marketing Science conference (Istanbul), the 2014 UT Dallas FORMS conference, the 2014 IIOC conference

(Chicago), the 2014 Yale Customer Insights conference (New Haven), the 2014 CIRP ´

EE Conference on Industrial

Organization (Montreal, Canada), the 2014 Summer NBER Industrial Organization Meeting (Cambridge, MA), the

2014 Summer Institute in Competitive Strategy (Berkeley), the 2014 Marketing Dynamics conference (Las Vegas),

the 2014 Annual Federal Trade Commission Microeconomics Conference (Washington DC), the 2016 Workshop on

the Economics of Advertising and Marketing (Vilnius, Lithuania), and seminar participants at the Carlson School of

Management (University of Minnesota), Brown University, Anderson School of Management (UCLA), Booth School of

Business (University of Chicago), Simon Business School (University of Rochester), Johnson School of Management

(Cornell University), Kellogg School of Management (Northwestern University), Universitat Aut`

onoma de Barcelona,

and Columbia University for their comments. Wespecifically thank the discussants Tat Chan, Judith Chevalier, Gautam

Gowrisankaran, Sanjog Misra, Sridhar Narayanan, Marc Rysman, Anna Tuchman, and Lawrence White for the detailed

comments on our work. We are grateful to RateWatch and to an anonymous market research company for providing

us with the data. Mark Egan provided excellent research assistance. This article was generously supported by the NSF

(grant no. SES-1426823) and by the Dean’s Small Grants Program at the University of Minnesota Carlson School of

Management. All errors are our own.

C2017, The RAND Corporation. 611

612 / THE RAND JOURNAL OF ECONOMICS

1. Introduction

How does advertising work? This is a central question in the economics literature (Bagwell,

2007). In his classic book, Chamberlin (1933) argued that advertising affects demand because

(i) it conveys information to consumers with regard to the existence of sellers and the prices and

qualities of products in the marketplace, and (ii) it alters consumers’ wants or tastes. Measuring

the separate contributions of these roles is, however, challenging. Incorporating the different

channels through which advertising affects demand into a model, especially one that is suitable

for estimation, is a complex task. More importantly, it is rare to have access to data that describes

the entire consumer purchase process, from awareness to consideration and finally choice, which

is necessary to identify the separate roles of advertising.

This article uses detailed survey data to empirically disentangle the roles of advertising in

the different stages of the consumer purchase process and to measure the mechanisms through

which advertising ultimately affects consumer choices. More specifically, we measure how much

advertising influences consumer behavior directly as a utility shifter versus as a way of increasing

consumer awareness for a product or brand. We do so in the context of the banking industry and

conduct our measurement through a fully specified structural model that contains the awareness-

consideration-choice stages and, in particular, endogenizes the “choice” of consideration set by

each consumer using a costly search framework.

Understanding the drivers of demand for retail banking products and services, a very large

and growing sector of the economy, is an important question. With its $14 trillion of assets,

7,000 banks, and more than 80,000 bank branches, the U.S. banking industry comprises a very

important portion of the “retail” economy with significant attention from regulators and policy

makers. Despite the importance of the banking sector, structural demand analyses to date (e.g.,

Dick, 2008; M´

olnar, Violi, and Zhou, 2013; Wang, 2016) have been based on aggregated market

share data on deposits. There has been very little research using detailed consumer-level data

to characterize consumers’ heterogeneous response to drivers of demand. Moreover, although

the banking and financial industry spends more than $8 billion per year on advertising,1and the

industry’s market value relies heavily on advertising (as shown, e.g., in Belo, Lin and Vitorino

2014 and Vitorino2014), there is little academic research that investigates the precise way in which

advertising affects consumer demand in this industry. Some recent exceptions in the literature

are Gurun, Matvos, and Seru (2016) on the marketing of mortgages and Hastings, Hortac¸su, and

Syverson (forthcoming) on retirement savings products. However, neither of these studies can

differentiate between the awareness and the utility-shifting functions of advertising.

The primary data source for our study consists of individual-level survey data from a major

market research company.The data contain infor mation about whichbanks consumers are aware

of, the set of banks consumers consider, and the identity of the banks consumers decide to open

one or more new bank accounts with. In addition, we observe a nearly complete customer profile

containing information on demographics such as age, gender, marital status, and education. We

complement these data with three additional sets of data on the retail banking industry. Data

provided by RateWatch contain information on interest rates and fees for the most common

account types for all banks over the same time period as the survey data. Advertising data

come from Kantar Media’s Ad$pender database. Kantar tracks the number of advertisements and

advertising expenditures in national media as well as in local media at the Designated Market

Area (DMA) level. Last, we collect information on the location of bank branches from the

Federal Deposit Insurance Corporation (FDIC). Together, these data give us a detailed picture of

consumers’ shopping and purchase processes and of the main variables affecting them.

The data show that consumers are, on average, aware of 6.8 banks and consider 2.5 banks

when shopping for a new bank account. There is substantial heterogeneity in consumers’ aware-

ness and consideration sets as reflected in the large variation in the sizes of these sets across

1kantarmediana.com/intelligence/press/us-advertising-expenditures-increased-second-quarter-2013.

C

The RAND Corporation 2017.

HONKA, HORTAC¸ SU AND VITORINO / 613

consumers. The relationship between the size of consumers’ awareness and consideration sets is

weak, indicating a distinct difference between the two stages. This difference is further reflected

in the large variation across consumers in what concerns which banks enter consumers’ awareness

and consideration sets. The data also show that there are large differences in the conversion rates

from awarenessto consideration and from consideration to choice/purchase across banks. Interms

of final choices, the most common account types consumers open are checking accounts (85% of

consumers), savings accounts (58%), and credit cards (26%). Finally, our data also confirm the

crucial importance of local bank presence—that is, bank branches—in the consumer’s decision

process: given that the consumer decides to consider or to purchase from a bank, we find that the

probability that a bank has a local branch within five miles of the consumer’s home is on average

81% and 84%, respectively.

To quantify the effects of advertising (and other variables) on economic outcomes, we

develop a structural model of the three stages of the consumer purchase process: awareness,

consideration, and choice. We model the decision process of a consumer who is considering to

add (or not to add) one or more bank accounts to his existing portfolio. The model incorporates

informational heterogeneity across consumers (reflected in the size and content of the consumers’

awareness and consideration sets) and allowsfor costly search (about interest rates). We showthat

incorporating both informational heterogeneity and costly search is important in our context. In

addition, we address the potential endogeneity of the banks’ advertising intensity variable using

the control-function approach.

Awareness is a function ofbank adver tising, local bank presence, and demographic factors.

A consumer searches among the banks he is aware of. Searching for information is costly for

the consumer, because it takes time and effort to contact financial institutions and is not viewed

as pleasant by most consumers. Thus, a consumer investigates only a few banks that together

represent his consideration set and makes the final decision to open one or more new accounts with

a bank from among the ones in the considered set. Our utility-maximizing modeling approach

connects and models all three outcome variables: the set of banks the consumer is aware of, the

consumer’s decision of which banks to include in his consideration set given his awareness set,

and the decision of which bank to open one or more accounts with, given his consideration set.

The estimates from the awareness stage highlight the importance of both advertising and

branch presence in driving how aware consumers are of a bank. The results from the consideration

and choice stages indicate that the average consumer search cost that rationalizes the amount of

search conducted by consumers within their awareness sets is about four basis points (0.04%) per

bank searched, which is in line with consumer search costs estimated in other financial products

settings, for example, Hortac¸su and Syverson (2004). The results also show that convenience

is a major driver in the consumer shopping process. Convenience is captured by the fact that

consumers are more likely to open bank accounts with banks located in close proximity to where

the consumers live. The positive effect of local bank presence at the choice stage suggests that,

in spite of the widespread availability and convenience of online banking, consumers still value

having the possibility of talking to a bank employee in person.

To evaluate the economic implications of our findings, we perform several analyses. First, to

answer the question of how advertising works, we compute elasticities of awareness and choice

probabilities with respect to changes in advertising. In addition, we perform a counterfactual

analysis in which we shut down advertising and investigate its effects on selected economic

variables. Overall,our results show that advertising has a large indirect effect on consumer choices

via awareness and that it affects consumers’ choices directly only marginally. Advertising makes

consumers aware of more alternatives; thus, consumers search more and find better alternatives

than they would otherwise. In turn, this increases the market share of smaller banks, making

the U.S. banking industry more competitive. This suggests that, in the retail banking industry,

the primary role of advertising is to inform consumers about the existence and availability of

retail banks and their offerings. This finding stands in contrast to other recent research that has

also investigated consumers’ demand for financial products. For example, Gurun, Matvos, and

C

The RAND Corporation 2017.