Consumer Search Ability, Price Dispersion and the Digital Divide
| Date | 01 April 2017 |
| Author | Colin Wren,Stephen McDonald |
| DOI | http://doi.org/10.1111/obes.12151 |
| Published date | 01 April 2017 |
234
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 79, 2 (2017) 0305–9049
doi: 10.1111/obes.12151
Consumer SearchAbility, Price Dispersion and the
Digital Divide*
Stephen McDonald,† and Colin Wren†
†Economics, Business School, Newcastle University, 5 Barrack Road, Newcastle Upon Tyne,
NE1 4SE, UK (e-mails: stephen.mcdonald@ncl.ac.uk; colin.wren@ncl.ac.uk)
Abstract
The ‘digital divide’ in online activities is believed to arise from differences in Internet ac-
cess, but this paper advances an alternative explanation that is related to consumer search
ability. It argues that this leads to greater price dispersion, causing some consumers to be
discriminated against. It analyses price data for the UK Internet motor insurance market,
collecting data on 32,255 prices for 110 sub-markets, where differences in price dispersion
across these by age, occupation and sex of the driver are argued to reflect differences in
search ability. Allowing for price dispersion to also depend on the insurance risk, it finds
greater price dispersion for consumers with weaker search abilities, i.e. older, unemployed,
retired or female consumers. As this is not explained by alternative hypotheses, the pa-
per concludes that improved Internet access alone will not close the ‘digital divide’. The
implication is that policymakers should address the online search abilities of individuals
as well as Internet access.
I. Introduction
The existence of a ‘digital divide’ is well documented, with older and poorer households
less able to take economic advantageof the Inter net due to their lowerlevels of online access
(Hoffman and Novak,2000; Chinn and Fairlie, 2006). The ‘digital divide’will lead to price
dispersion as the benefits of lower online search costs will not be distributed evenly. The
evidence is that the uptake of the Internet reduced price dispersion (Brown and Goolsbee,
2002), and indeed Goldfarb and Prince (2008) find that the ‘digital divide’ no longer exists
once Internet access is controlled for.This has meant that UK policy interest has centred on
improving online access (OFT, 2007; ONS, 2012). Despite this, price dispersion remains
pervasive in Internet markets for homogeneous goods and services (Baye, Morgan and
Scholten, 2004; Haynes and Thompson, 2008), and this suggests that some consumers still
do not take advantage of the loweronline search costs and are being discriminated against.
JEL Classification numbers: L11, D83, D40
*Earlier versions of this paper were presented at the RoyalEconomic Society conference and to Economics staff at
Newcastle University. The authors thank participants at these events, as well as Robert Hudson, Ginger Jin, Francis
Kiraly, Claudio Piga, SteveThompson and Mike Waterson for comments. The authors are grateful to referees of this
journal, including the editor, Dr Debopam Bhattacharya, but remain responsible for the paper’scontent.
Search ability and price dispersion 235
This paper explores an alternative explanation for price dispersion in online markets due
to consumer search ability. It implies that improving Internet access alone is insufficient
to close the ‘digital divide’.
To directly relate price dispersion to individual characteristics, the paper takes a novel
approach as it gathers price data for an online market by adopting assumed personal iden-
tities for different individual types. This is for the UK online motor insurance market in
which the quotes are related to personal and car characteristics. As these individual types
vary in their frequency of Internet use, it is argued that they vary in their search ability.
Further, it is argued that the motor insurance firms learn this through their experience of
making quotes and that greater price dispersion will be observed for the consumers with
weaker search abilities. The price data are collected from the leading UK motor insurance
comparison website and from the leading sellers that choose not to list on these online
aggregators. The quotes are potentially contractible and they are collected for 22 individ-
ual types by age, occupation and sex.This is for each of 5 car types, giving 110 sub-markets.
Further, the data are collected at 12 monthly intervals, giving 1,320 price sets and a total
of 32,255 price observations.
The use of insurance price data means that the insurance risk affects price dispersion, as
well as the effectiveness of consumer search, although this is rarely considered in studies
using insurance data (e.g. Dahlby andWest, 1986).This is because in a sub-market, in which
the individual characteristics are held constant, each insurance contract carries the same
risk, so that in equilibrium, a firm must sell more lower-priced contracts to get the same
expected return as from selling higher-priced contracts. This affects price dispersion and
as higher risks mean that lower-priced contracts are less likely to be offered, it also affects
consumer search. In this paper, we regress the price data across sub-markets to examine
the variation in price dispersion, controlling for this risk. Overall, we find that online price
dispersion is greater for consumers that are older, unemployed, retired or female, indicating
the existence of a ‘digital divide’.These results are consistent with differences in consumer
search ability, as measured by the frequency of Internet use, and they are not otherwise
explained such as by unobserved heterogeneity. The implication is that improved Internet
access alone will not be sufficient to close the ‘digital divide’, so that the policy should
also focus on online search ability.
In the next section, the basis for price dispersion at the sub-market level due to search
behaviour is considered and the hypothesis set out. Section III outlines the data and section
IV presents the results. These are discussed in sectionV, after which conclusions are drawn.
II. Consumer search and price dispersion
The relationship between consumer search behaviourand price dispersion is much explored
in the literature and overall it finds that greater search will lead to less price dispersion (e.g.
Stahl, 1989; Janssen and Moraga-Gonz`alez, 2004).1This literature is characterised by many
models of search behaviour, although not all of these are relevant to our particular setting.
Thus, the main paper that relates to search in the motor insurance market by Dahlby and
1Price dispersion may also arise from other factors, such as differences in marginal costs (Reinganum, 1979;
Carlson and McAfee, 1983), but in insurance markets a substantial part of these costs will be the expected claim on
a contract, which will again be constant at a sub-market level as the individuals carry identical risks.
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial