Estimating switching costs: the case of banking
Introduction
Switching costs are costs induced to economic agents when they change their suppliers. As such, ex-ante homogeneous products become ex-post heterogeneous. These costs originate from a host of reasons, economic as well as psychological, such as various addictions and cognitive dissonance problems. Intertemporal product and service compatibility, network externalities, informational investment in business relationships are a few examples of the economic sources of switching costs. From the theoretical perspective, customer switching costs confer market power on firms. Thus, firms may face a trade-off between charging low prices to attract customers and lock them in, and high prices to extract supranormal rents from its already locked-in customers. The vast theoretical literature on switching costs is summarized in Klemperer (1995).1 Studies dealing with this phenomenon attempt to gain insight into the main issues of industrial organization such as entry deterrence, and the command over supranormal rents. Shapiro and Varian (1998) provide numerous examples of the impact of switching costs on market behavior.
The available empirical studies investigate the effect of switching costs on prices and market power. Ausubel (1991) provides some information that switching costs may explain the high interest rates on credit card balances, and Stango (1998), using variables related to switching, finds that switching costs have a significant impact on pricing in that market. Knittel (1997), using some proxies for switching costs, shows that they have provided long distance telephone carriers with market power. Sharpe (1997) finds that (banking) retail deposit-rates are positively affected by a proxy of switching costs. Dahlby and West (1986) support the effect of costly search on price dispersion in liability insurance, and Schlesinger and von der Schulenburg (1993) document a similar result in autoinsurance. Greenstein (1993) estimates the probability of “lock-in” in commercial mainframe computer systems acquired by federal agencies. His results may indirectly confirm the existence of switching costs for that sector but no quantification of the magnitude of switching costs is attempted. Another interesting empirical example is that of Borenstein (1991) for the gasoline market, where price discrimination is possible due to differences in the willingness of customers to switch stations. In a recent article, Shum (1999) measures the effect of advertising on habit persistence in the purchasing behavior of various brands of cereals. Shum finds that advertising encourages switching behavior at the household level. His main empirical question, however, concerns the way advertising affects brand substitutability thereby enhancing competitive conduct and lowering margins, and not the measurement of switching costs. Finally, a very recent paper by Israel (2001) develops and estimates a behavioral model of consumer–firm relationships in autoinsurance.2
Although the aforementioned empirical studies do point to the importance of switching costs in the determination of conduct and to the effect of various firms' policies on switching costs, they are generally silent regarding the magnitude and significance of switching costs. Whether switching costs are empirically important probably depends on the specific environment, industry, product type, and time period. One possible reason for the lack of empirical documentation of the magnitude and significance of switching costs is that the necessary micro data on individual-level transitions are rarely, if ever, available to researchers.3 In the context of estimating switching costs, the unobservables are individual customers' purchase decision histories. More specifically, we lack information on the identity of customers' previous suppliers.
The task of the present research is to complement existing theoretical models with an empirical investigation capable of highlighting the process of customer's switching behavior when customer-specific data are absent, and then embed it in a general behavioral model of the firm. As a matter of illustration, we estimate the magnitude and significance of switching costs in the market for bank loans and empirically explore various counterfactuals related to bank and customer behavior in this market.4
It should be noted that from an empirical perspective, switching costs may be more pronounced when they contribute to, and may result from, long-term relationships and repeated contacts between firms and their customers. In such cases, customers' switching among suppliers may entail not only psychological costs but also costs related to the loss of capitalized value of established relationships. As such, the market for bank loans may provide a natural habitat for the investigation of the magnitude and significance of switching costs.
Banking is a major sector in the economy in which switching costs seem to be prevalent due to information asymmetry. A high quality borrower switching to a competing (uninformed) bank may be pooled with low quality borrowers and confront unfavorable conditions Sharpe, 1990, von Thadden, 1998. This phenomenon (also known as the ‘lemons’ problem) may be exacerbated during periods of systemic wide banking problems or rescission periods. Thus, a switch between suppliers in the market for loans may entail direct (transaction-related) costs of closing an account with one bank and opening it elsewhere, as well as the unobserved, and perhaps the most significant costs associated with the foregone capitalized value of (previously established) long-term customer–bank relationship.5 Indeed, the extensive discussion in recent literature about the importance of “relationship banking” and its significant impact on borrowers' values James, 1987, Vale, 1993, Petersen and Rajan, 1994, Boot, 2000 may point to the existence of severe switching costs in this sector.6
The paper is organized as follows: in Section 2 we describe the model, and in Section 3 the empirical methodology. Section 4 describes the data used, and Section 5 provides the results and some counterfactuals. Section 6 concludes the paper.
Section snippets
A model of competition with switching costs
The empirical setup we present in this section builds on theoretical investigations related to the effect of customer switching costs on market conduct (see (Klemperer, 1987)). The theoretical relations between firm and customer behavior in the presence of switching costs require, however, some adjustment for empirical application. First, such relations are usually posited in terms of rarely observable individual switching activity. Secondly, while in most theoretical models switching costs
Empirical methodology
The model presented in Section 2 yields the following equations:
- (i)
the first-order condition (2.36):
- (ii)
the market share Eq. (2.14):
The banking industry
Our database consists of a panel of annual observations for the Norwegian banking industry, spanning nine years from 1988 to 1996. The panel covers all banks operating in Norway in that period.30 Table 1 describes the banking industry characteristics.
The number of banks declined from 177 in 1988 to 139 in 1996. The reduction in the number of banks is almost only due to mergers.31
Estimation and results
There are two major issues to be attended in the estimation process. The first relates to the proper length of period in which switching may take place. The second important issue relates to the definition of the geographic and product market, i.e., to account for two aspects:
- (i)
customer location preferences and the corresponding banks' branch-network size, and
- (ii)
the ability of banks to provide the entire continuum of loan sizes demanded.
Regarding the first issue, the loan maturity may help us in
Concluding remarks
We have proposed an empirical model of firms' strategic behavior in the presence of switching costs. Customers' transition probabilities embedded in firms strategic interaction were used in a multiperiod model to derive estimable equations of a first-order condition and market share (demand) equations. The novelty of the proposed model is in its ability to extract information on both the magnitude and significance of switching costs, as well as on customers' transition probabilities, from
Acknowledgements
We are grateful to Emanuel Barnea, Pedro Barros, Benjamin Bental, Luı́s Cabral, Clive Granger, Giora Hanoch, Patricia Jackson, Eugene Kandel, Benny Levikson, Loretta Mester, Espen Moen, Dan Peled, Christian Riis, Asbjørn Rødseth, Agnar Sandmo, Steven Sharpe, Lars Sørgard, Erling Steigum, Siri Pettersen Strandenes, Dag Tjøstheim, Ana Lozano-Vivas and Yitzhak Weit for their comments. We thank participants of the Annual Research Meeting of the Norwegian Economic Association and workshop
References (40)
Relationship banking: What do we know?
Journal of Financial Intermediation
(2000)- et al.
Endogenous switching costs in a duopoly model
International Journal of Industrial Organization
(1990) - et al.
Dynamic duopoly with slowly changing customer loyalties
International Journal of Industrial Organization
(1996) The importance of relationships to the availability of credit
Journal of Banking and Finance
(1998)- et al.
Relationship lending within a bank based system: evidence from European small business data
Journal of Financial Intermediation
(2000) Some evidence on the uniqueness of bank loans
Journal of Financial Economics
(1987)Banking technology and the existence of a consistent output aggregate
Journal of Monetary Economics
(1986)- et al.
The duration of bank relationships
Journal of Financial Economics
(2001) Mixed pricing in oligopoly with consumer switching costs
International Journal of Industrial Organization
(1992)Revisiting dynamic duopoly with consumer switching costs
Journal of Economic Theory
(1995)
The failure of competition in the credit card market
The American Economic Review
Multi-period competition with switching costs
Econometrica
Banks as multioutput oligopolies: An empirical evaluation of the retail and corporate banking markets
Journal of Money, Credit, and Banking
Selling costs and switching costs: explaining retail gasoline margins
RAND Journal of Economics
Switching costs and bidding parity in government procurement of computer system
Journal of Law, Economics and Organization
Price dispersion in an automobile insurance market
Journal of Political Economy
Did installed base give an incumbent any (measurable advantage in federal computer procurement
The RAND Journal of Economics
Cited by (214)
Counteroffers and Price Discrimination in Mortgage Lending
2023, Journal of Empirical FinanceEnergy price shocks and stabilization policies in the MATRIX model
2023, Energy PolicyBanking market power and its determinants: New insights from MENA countries
2023, Emerging Markets ReviewEnter the MATRIX model:a Multi-Agent model for Transition Risks with application to energy shocks.
2023, Journal of Economic Dynamics and ControlIt takes more than two to tango: Multiple bank lending, asset commonality and risk
2022, Journal of Financial StabilityBuffering B2B service failure: The role of customer engagement
2022, Industrial Marketing Management