Measuring consumer switching costs in the television industry

• Published date: 01 May 2016
• Author: Oleksandr Shcherbakov
• DOI: http://doi.org/10.1111/1756-2171.12131
• Date: 01 May 2016

RAND Journal of Economics

Vol.47, No. 2, Summer 2016

pp. 366–393

Measuring consumer switching costs

in the television industry

Oleksandr Shcherbakov∗

In this article, I develop and estimate a model of dynamic consumer behavior with switching

costs in the market for paid-television services. I estimate the parametersof the structural model

using data on cable and satellite systems across local US television markets over the period

1992–2006. The results suggest switching costs range from $159 to $242 for cableand from $212

to $276 for satellite providers in 1997 dollars.Using a simple dynamic model of cable providers,

I demonstrate that switching costs of these magnitudes can significantly affect the firms’optimal

strategies.

1. Introduction

Switching costs are likely to exist in a variety of important industries, including computer

software and hardware development,banking, and telecommunications. However, how consumer

switching costs affect competition is not clear from a theoretical standpoint. Do markets become

more competitive due to producers’ incentives to invest in their customer base? When does the

“harvesting” motive (from already locked-in consumers) begin to dominate? Should such indus-

tries be regulated and,if so, then when and how? A careful analysis of market structure and industry

conduct and design of an efficient policy environment requires not only a qualitative assessment

but also a precise quantitative measure of switching costs and their effects on market outcomes.

In this article, I quantify switching costs in the paid-television industry and explore potential

implications for competition. My primary contribution is the development of an empirical

framework that can be used to identify and estimate consumer switching costs using market-level

data. I then use estimates of the structural parameters and a supply-side model to illustrate

differences in the optimal policy of cable televisionproviders under alternative market structures.

In many markets, consumers make repeated purchases from the same provider because

switching to an alternative supplier involves additional monetary and utility costs. When

these costs are substantial and product characteristics change rapidly over time, rational

consumers should recognize the effects of their current decisions on their future utility flows.

∗University of Mannheim; ashcherb@mail.uni-mannheim.de.

I am deeply indebted to my supervisors, Gregory Crawford and Gautam Gowrisankaran, for their help at all stages of

my work. I greatly acknowledgevaluable advice and suggestions made by Daniel Ackerberg, Keisuke Hirano, and Aviv

Nevo. Useful comments and suggestions made by Kathleen Nosal, Yuya Takahashi, Tim Lee, Nicolas Schutz, Philipp

Schmidt-Dengler, and anonymous referees are greatly appreciated. All errors are my own.

366 C2016, The RAND Corporation.

SHCHERBAKOV / 367

Therefore, allowing for forward-looking behavior of consumers is important for an internally

consistent empirical model. For example, in a model where consumer expectations about future

product attributes are ignored, consumer decisions can be rationalized only by current period

variables. This may result in biased estimates of the structural parameters and incorrect policy

implications.

In this article, I develop a method to identify switching costs from aggregate data. When

individual decisions are not observed, a researcher has to separately identify the distribution

of persistent consumer preferences and the switching costs from aggregate statistics. As in

static decision models, I use the variation in observable product characteristics across markets

to identify parameters of the distribution of consumer heterogeneity. One way to help identify

these parameters is if some consumer decisions are not directly affected by the switching costs.

For example, if a consumer can switch between alternative products from one supplier at no

cost, product-specific market shares provide important information about the distribution of

preferences. By contrast, I identify the switching cost parameters using exogenous shifters of

the previous period’s consumer decisions because in a model without switching costs (or other

sources of state dependence), previous period decisions must be irrelevant for current period

choices.

Developments in the US paid-television industry between 1992 and 2006 make it a good

setting to study consumer switching costs incurred when changing service providers. Although

cable television in the United States has a fairly long history, entry of new satellite technology

in the 1990s challenged the previous monopoly position of the cable providers. The new market

structure provides an opportunity to explore the interplay between consumer switching costs and

competition in an important industry.1

To empirically evaluate the effect of switching costs on the optimal policy by the cable

providers, it is crucial to consistently estimate consumer switching costs and other parameters

of the consumer utility function, which is the main focus of the empirical application in this

article. The empirical model of consumer decisions I use to quantify consumer switching costs

is tailored to accurately represent the institutional details of the paid-television industry in the

US, including differences in technologies used to deliver TV signals and vertical differentiation

of products offered by the same firm.2

To motivate estimation of the structural model, I provide reduced-form evidence of

state dependence in consumer decisions. In particular, I find that the current period values

of exogenous variables affect not only contemporaneous consumer decisions but also future

decisions, which is consistent with substantial consumer switching costs in the industry.

The results suggest that consumer switching costs amount to approximately $190 for cable

and $240 for satellite (in 1997 dollars). These correspond to slightly more than one half of

the annual service cost for each of the providers. Comparing these estimates to the average

cable service installation fees in 1992–2002 implies that the fees account for about 20% of the

estimated monetary value of switching costs, with the remaining 80% of the costs explained by

the unobserved hassle costs of switching. Importantly, static and myopic models of consumer

behavior overestimate short-run price elasticity because they ignore the effect of price increases

in the current period on the future flow utility values.

To illustrate the importance of accounting for consumer switching costs, I develop a

dynamic model of a cable service provider and recover the cost structure of cable firms. Then,

I use the model together with estimates of the demand-side parameters to evaluate the effect of

consumer switching costs on the optimal policy of cable providers under duopoly and monopoly

scenarios. Although this is not meant to be suggestive of actual policies, it does suggest that the

magnitudes I estimate would be material to firms’ optimal strategies. For example,holding prices

1In 2007, cable operators had revenues of about $79.1 billion nationwide.

2The methodology can be applied to other industries as well. For example, Nosal (2012) extends the method to

estimate consumer switching costs in the health insurance industry.

C

The RAND Corporation 2016.