Memorandum submitted by the University
of East Anglia
1. SUMMARY OF
MAIN POINTS
We believe that there are important doubts about
the intensity of competition in the retail market. On structural
grounds, there are concerns about high concentration (fewness
of effective competitors), and barriers to the entry of potential
new competitors. There is at least a case to be answered that
the vertical integration of the six remaining firms provides them
with the ability to foreclose any future entrants. Following liberalisation
in the late 1990s, it was hoped that cross-region and cross-fuel
entry by the traditional electricity incumbents and British Gas
would replace monopoly with competition amongst a large number
of sizeable firms. In the event, however, cross-entry by electricity
incumbents has been relatively insignificant, and all regional
markets remain effectively duopolies—British Gas and the
previous electricity incumbent. For this and other reasons, the
characteristics of the market are such that "tacit collusion"
is a possible outcome. It is important to stress that this term
does NOT imply a cartel, but rather a "soft" form of
competition—loosely speaking, a mutual recognition that
it is in no-one's interest to compete very aggressively. Tacit
collusion, so defined, may not be illegal in competition law,
but it is most certainly not in consumers' interests. The evidence
on various features of "performance" also merits deeper
scrutiny. In particular, the extent of switching by consumers
is controversial. There are those, including OFGEM, who argue
that effective switching has been substantial, and that this is
direct evidence of healthy competition. However, this is unconvincing
for three reasons: (i) much of the switching that has occurred
reflects the introduction of dual-fuel bundles, which would be
cheaper than single-fuel supply even in the absence of competition;
(ii) survey evidence suggests that much switching by consumers
has been ill-advised and ill-informed, (iii) with the exit of
so many firms since 1999, it is increasingly doubtful that consumers
have any genuinely low price suppliers to whom they can switch.
A second topic on which sound evidence would be invaluable is
retail price cost margins and profitability, but here the data
are difficult to unravel—not least because of the integrated
structure of all the majors, raising uncertainty about what is
the appropriate transfer price from wholesale to retail. However,
we do know—and this seems to be uncontested—that the
traditional incumbents are not only still dominant in their traditional
markets of incumbency (in terms of market share), but, even to
this day, they continue to offer the consumer price tariffs which
are more expensive than those offered by non-incumbents. Other
things equal, high market shares coupled with high price are inconsistent
with vibrant competition—especially when, as here, all firms
sell essentially an identical product (what economists call an
"homogenous" good).
2. BRIEF INTRODUCTION
OF AUTHORS
Stephen Davies is Professor of Economics in
the School of Economics at UEA and a Principal Investigator of
the ESRC Centre for Competition Policy. His research interests
include: the economics of competition policy; European industrial
structure; mergers and merger simulation; and multinational firms.
He is a member of the Academic Panel which advises the Office
of Fair Trading. He was formerly the General Editor of The Journal
of Industrial Economics. He is currently working on: mergers,
competition with non-linear pricing in the electricity market;
and the resale price maintenance.
Catherine Waddams (formerly Price) is founding
Director of the ESRC Centre for Competition Policy at the University
of East Anglia and Professor of Regulation in the School of Management.
She joined UEA from the University of Warwick in 2000. She has
studied the development of utility privatisation and regulation,
and is particularly interested in the effect on different income
groups of introducing competition to such industries, both in
developed and developing economies; and in the role of consumers
in competition policy. Much of Catherine's research centres on
the energy sector. She is a part time member of the reporting
panel of the UK Competition Commission.
The ESRC Centre for Competition Policy (CCP)
at the University of East Anglia (UEA) undertakes interdisciplinary
research into competition policy that has real-world policy relevance
without compromising academic rigour. Our members are drawn from
a range of disciplines, including economics, law, business and
political science.
The Centre was established in September 2004,
building on the pre-existing Centre for Competition and Regulation
(CCR), with a grant from the ESRC (Economic and Social Research
Council). It currently includes a total of 17 faculty members
(including the Director and a Political Science Mentor), seven
full- and part-time researchers and 14 PhD students.
The Centre produces a working paper series,
runs weekly seminars, holds a number of events throughout the
year (including a yearly conference), and publishes a regular
newsletter and e-bulletin. Its members welcome links with academics,
practitioners, policy makers and the voluntary sector.
3. FACTUAL INFORMATION
FOR THE
COMMITTEE
The committee asks:
— Whether the current market structure
encourages effective competition in the retail markets for gas
and electricity;
1. How vigorous is competition?
3.1 There is concern that the retail energy
suppliers may be competing less aggressively than they might.
In general, the ability of consumers to choose a better (usually
cheaper) supplier is an important curb on any unilateral market
power which incumbents of previously monopolised markets might
otherwise wield. However such choice provides no countervailing
power if suppliers co-ordinate their prices. There is little evidence
of any explicit collusion, which is illegal, but there is a suggestion
of tacit co-ordination: this is sometimes referred to as "tacit
collusion", in which firms do not compete vigorously with
each other, resulting in "co-ordinated effects". The
UK Competition Commission (following the decision of the European
Court of First Instance judgement in the Air tours case, 2002)
describes three necessary conditions for co-ordinated effects.
First, firms must be aware of the behaviour of others. In energy
markets the prices are highly transparent: publicly available
information on offerings clearly provides firms with information
about each other's actions. Secondly it must be costly for firms
to deviate from the prevailing (non competitive) behaviour. Since
six large firms interact repeatedly in 15 markets (14 regional
electricity markets and a national gas market), there is plenty
of opportunity for any deviation from the "non competitive"
position to be detected, so that the market will revert to the
less profitable but more competitive situation as firms note any
individual departure from the "collusive" situation.
And, thirdly, competitive constraints are weak. Small independent
companies have struggled to enter and survive in the household
energy market; if they do manage to enter they generally exit
(through failure or takeover) within a couple of years. The national
recognition of the brand names of the main leaders can act as
a significant barrier for new entrants.
3.2 Add to these market characteristics,
the homogeneity of energy, and the conditions certainly seem ripe
for co-ordinated effects. The energy regulator noted such concerns
in a 2004 review (Ofgem, 2004) but has not publicly revisited
this issue in detail since then. Companies interact regularly
to discuss "best practice" in areas such as social concerns.
There is little evidence to support allegations (as reported in
the Sunday Times in January 2008) that such meetings are used
to fix prices, but they do provide an opportunity for the industry
to share common approaches. Such legitimate interaction is likely
to increase as the environmental and fuel poverty agendas become
more pressing, raising such concerns further. Moreover, the problem
is not just in the UK. Four of the six large energy companies
are major European players (EdF, EOn, RWE and Iberdrola each own
incumbent suppliers in three of the electricity regions) and interaction
is at international as well as UK level. This is likely to be
a continuing issue for the European Commission as well as the
UK authorities.
3.3 The analysis of such problems depends
partly on whether the energy markets are regional or national.
The evidence points strongly in the direction of markets being
regional. We estimate that, in the typical electricity region,
10 times more consumers buy from the original electricity incumbent
than from other, entrant, suppliers who were previously incumbents
outside that region. This means that, within each region, market
structure still amounts to little more than a duopoly involving
British Gas and the local regional electricity incumbent.
2. Consumer switching
3.4 Householders continued to change energy
supplier after the removal of price caps in 2002, unsurprisingly
at a slower rate than when the market was initially opened, and
not always away from the incumbent. Interestingly, there has been
only a slight increase in switching rates in the last two years.
Given the rapid recent acceleration in retail electricity prices,
one might have expected a sharp upturn in switching as more and
more consumers were galvanised into actively searching for cheaper
suppliers. While the figure does suggest an upturn, it is only
very moderate.
3.5 By September 2007 the market share of
the incumbent gas supplier remained as high as 47% on a national
level; and at the same date the average market share of the incumbent
electricity suppliers in their "home" area ranged from
39% for direct debit to 53% for standard credit customers. Across
the markets, incumbents retained exactly half the prepayment customers.
3.6 The role of switching in a newly opened
market is clearly crucial. If consumers are loyal to their incumbent
suppliers, then no amount of entry by those offering alternatives
can exert competitive pressure on the firm. In energy, the first
decade of consumer choice has shown a little less than half of
consumers remain loyal to the incumbent at the kind of savings
which entrants are able to offer.
3.7 The energy regulator has encouraged
consumers to switch supplier and at various times has hailed high
switching rates as a sign that competition is working well in
the household energy market. However, if consumer errors are large,
high rates of switching are not necessarily a sign of a well functioning
market—nor are they necessary for the market to function
effectively. The threat of switching could act as a powerful discipline
on the incumbent's behaviour, and if it responded by lowering
its price to a competitive level to retain consumers, then the
market might indeed be functioning well. In that case there would
be little mark-up by incumbents. However, as noted below, consumers
often appear to make erroneous switches, incumbent markups remain,
and it is therefore improbable that the threat of switching is
in fact disciplining the market.
3. Relation of prices to costs
3.8 One indication of the competitiveness
of the retail market is provided by analysing how closely the
charges to consumers reflect the costs of the suppliers. The closer
prices are to costs, the more competitive the markets are. Such
analysis also provides an indication of whether (and how much)
the incumbent is able to charge more than other suppliers, once
other factors are taken into account.
3.9 Such an analysis was undertaken for
three consumption levels and three payment methods (ie nine times
in all) in the electricity market, seeing how the charges for
high/medium/low consumption level consumers, and for different
payment methods, depended on the charges paid by retailers for
use of the distribution system in December 2006 (results tables
shown in Annex 1).
3.10 Results across the markets are very
similar, with distribution costs, the main cost category that
we can identify, being reflected almost one to one in retail charges,
except for medium consumption prepayment consumers. Other costs
are transmission, which accounts for only 3% of retail tariffs,
and generation and supply costs, which we cannot observe. Generation
costs at least are likely to be similar for each company in each
region, so if they are just reflecting costs in their prices we
would expect the "company effect" to be similar across
consumption levels and price methods.
3.11 All credit customers pay more if they
are supplied by the incumbent. The differential, compared to British
Gas, is between £9 and £33 per year, and, in turn, British
Gas prices are higher than those of (non incumbent) electricity
competitors by an average of around £63 a year. This reveals
a fair degree of incumbency power. The only place where there
was no incumbent electricity mark-up was for low consumption prepayment
consumers, suggesting that entrants are not targeting consumers
in this group. This was also the only market in which British
Gas's price was not higher than all other non incumbent suppliers.
This analysis applies to charges for electricity only and not
to dual fuel deals, to which many consumers switch.
4. Consumer errors
3.12 Switching (or its threat) disciplines
the market because consumers choose suppliers who meet their needs
better, usually by being cheaper for a homogenous product like
electricity. However evidence from CCP questions the assumption
that consumers necessarily make good choices for themselves. Even
amongst those who switched electricity supplier purely to save
money, consumers made surprisingly poor decisions: at most half
selected the tariff that was best for them; those who switched
captured less than half the available benefits; and more than
a fifth switched to a more expensive supplier. This raises concerns
not only for the consumers themselves, and for potential overestimates
of the direct benefits from switching (if it is calculated by
multiplying the number of switchers by potential benefits). But
also because the discipline of switching on suppliers is likely
to be less beneficial if switching is not based on the maximum
benefit available to the consumer. The assessment of the benefits
and the effects of competition depend partly on the nature and
extent of consumer response to the incentives in the market, and
in this market we can see that such response is often faulty,
and the consequent discipline correspondingly lax.
5. Does ownership of the electricity network
by the incumbent hamper competition?
3.13 Accounting unbundling and separate
licensing of the incumbent and the distribution network in each
electricity region has been enforced since 2000, and since then
there has been separation of ownership in half of the regions
(seven of the 12 regions in England and Wales). In the other five
in England and Wales the incumbent supplier and the distribution
wires are owned by the same company, while full vertical integration
remains in the two Scottish regions. This provides a natural experiment
to identify whether such co-ownership hampers the development
of competition. In any one year since entry was permitted, the
amount of market which entrants had managed to capture was 4%
less where the incumbent was co-owned with the distribution company
than where ownership was separate. Similar effects are found if
Scotland (where vertical integration is more extensive) is excluded.
These simple statistics suggest that co-ownership of the distribution
system, even with the rigorous license and accounting separation
now required in the UK, may still endow some (albeit relatively
small) advantages on the incumbent in repelling entrants to its
home territory. If further exploration confirms these findings,
it might be appropriate to explore consider whether ownership
separation should be imposed. However in this case (as compared
with voluntary separation) some of the costs of separating ownership
might have to be borne by the consumer, rather than shareholders,
which would affect the valuation of its benefits.
— The implications of growing consolidation
in the energy market;
3.14 Since the market was opened in 1999,
the number of distinct retail suppliers to households has fallen
dramatically, so that there are now five consolidated survivors
of the previous electricity incumbents (which may reduce further,
depending on the outcome of discussions about the future of Iberdrola),
and one gas incumbent. Our discussion above shows that this has
created conditions where tacit collusion is likely to flourish,
and that other indications on prices and market share confirm
this possibility.
— The interaction between the UK and
European energy markets;
3.15 As a general principle, any move to
a genuinely integrated single European energy market would be
competition enhancing, ceteris paribus. However, we noted above
the trend towards consolidation in ownership of the leading firms
across member states. Clearly, if this trend were to continue
unabated, the end result may be that the bigger integrated market
is dominated by just a few large multinational suppliers. The
potential problem of tacit collusion may then resurface at a supranational
level.
— The effectiveness of regulatory oversight
of the energy market;
3.16 Many questions were raised about the
benefits of opening household retail energy markets to competition
at the time, and more recently David Newbery has noted that "retail
margins have widened considerably since the domestic franchise
ended in 1999" (2005), indicating that the consolidated energy
market is exerting less competitive pressure on the liberalised
market than did the previous regulatory regime. The dilemma for
the regulator is that entrants will only be attracted by high
margins, and so some increase may be necessary to "kick start"
a change from regulated monopoly to competition. Consumers have
certainly been offered a much wider choice of tariffs since 1999.
But if the market is functioning well, margins should in turn
be reduced through competition. Since the regulator has itself
become a champion of competition (its duties include the protection
of consumers by promoting competition wherever appropriate), there
is a danger that it has an incentive to defend the success of
competition as a measure of its own effectiveness. We therefore
welcome the announcement of OFGEM's own probe into the energy
supply markets, and the microscope which they promise to bring
to the exercise.
— Progress in reducing fuel poverty
and the appropriate policy instruments for doing so.
3.17 Fuel poverty depends, both in a statistical
and a practical sense, on income and on energy prices. In the
early years of the century, while prices fell and incomes rose,
fuel poverty also fell. Since energy prices have been rising from
2004, fuel poverty has correspondingly risen. As the rise in incomes
slows, as seems likely, in 2008, this counterbalance to rising
energy poverty will be less effective than in the recent past.
3.18 It is very difficult to see how the
government can meet its target to eliminate fuel poverty amongst
vulnerable households by 2010 with the present and likely future
changes in energy prices. The underlying world price of oil and
the increasing importance of the environmental agenda are both
relentlessly increasing the upstream cost of energy. The signals
from such prices will help to curb demand and further green house
gas emissions; but for the fuel poor they may depress demand below
a level which is desirable from a social perspective and/or cause
real hardship for households with low levels of disposable income.
3.19 The most effective way to help the
fuel poor is to raise their income. Compared with offering them
lower prices (for the same effect), this also has the advantage
of not distorting their demand, so that they purchase "too
much" energy relative to its real cost to society. Since
lower income households are more responsive than average to changes
in price, subsidising their prices has a larger than average effect
in raising demand.
3.20 However if fuel poverty is to be addressed
through lower prices, voluntary schemes are unlikely to be effective.
The recent move to encourage companies to offer social tariffs
to low income consumers has had little effect, and it seems unlikely
that the Chancellor's exhortations for them to increase their
provision will do so. Any company which makes a significant commitment
to supply one group at lower profit margins than others, puts
itself at a competitive disadvantage for other parts of the market,
where prices would have to be higher than otherwise in order to
subsidise such offers. It is interesting that the only company
which does offer significant cross-subsidies (Ebico) is a not
for profit subsidiary of one of the major players, and does not
pay to be included as a full participant in the "search and
switching engines" which are approved by energywatch. If
the government believes that low income energy consumers should
be subsidised by other energy consumers, then it should mandate
such schemes, so that the impact on the market (effectively a
tax on other energy users) is equal across players, and does not
cause further distortions. One immediate result of such a policy
will be to make the subsidised consumers less attractive for companies,
who, as we have seen from the history of the prepayment market,
can take a number of steps to try to avoid recruiting such consumers.
APPENDICES
A: RESULTS TABLES FOR THE ESTIMATED RELATIONSHIP
BETWEEN RETAIL CHARGES AND DISTRIBUTION CHARGES
(I) DETERMINANTS
OF ANNUAL
BILL FOR
STANDARD CREDIT,
DECEMBER 2006
| | |
|
| Annual Consumption | 1650kWh
| 3300kWh | 4950kWh
|
| | |
|
| Constant | 23194.31***
(1182.11)
| 35778.68***
(1817.53) | 54088.27***
(2520.73)
|
| Distribution charge | 1.098***
(0.253)
| 1.436***
(0.222) | 1.042***
(0.207)
|
| Distribution area (km2) | -0.036**
(0.014)
| -0.069***
(0.017) | -0.074***
(0.204)
|
| Distribution customers (000) | -0.927***
(0.229)
| -0.339
(0.356) | -0.612
(0.549)
|
| Incumbent | 948.71**
(378.45)
| 2121.79***
(463.04) | 3266.66***
(681.92)
|
| Suppliers | |
| |
| London | -4631.86***
(484.33)
| -7847.52***
(592.57) | -11085.71***
(872.69)
|
| Npower | -2731.86***
(484.33)
| -3811.81***
(592.57) | -4914.28***
(872.69)
|
| Powergen | -2453.29***
(484.33)
| -3947.52***
(592.57) | -5421.42***
(872.69)
|
| Scottish Power | -2964.10***
(480.54)
| -4353.11***
(587.94) | -5816.66***
(865.87)
|
| SSE | -5967.58***
(484.33)
| -8826.09***
(592.57) | -11778.57***
(872.69)
|
| Adjusted R2 | 0.7271 | 0.8188
| 0.7918 |
Notes: Standard errors in parentheses.
* Significant at 10%; ** Significant at 5%; *** Significant at
1%
(II) DETERMINANTS
OF ANNUAL
BILL FOR
DIRECT DEBIT,
DECEMBER 2006
| | |
|
| Annual Consumption | 1650kWh
| 3300kWh | 4950kWh
|
| | |
|
| Constant | 22939.57***
(1182.07)
| 34333.56***
(3914.55) | 52540.69***
(2970.20)
|
| Distribution charge | 0.878***
(0.253)
| 1.707***
(0.479) | 1.062***
(0.244)
|
| Distribution area (km2) | -0.029**
(0.014)
| -0.108***
(0.038) | -0.744**
(0.029)
|
| Distribution customers (.000) | -1.038***
(0.229)
| -1.072
(0.768) | -0.878
(0.647)
|
| Incumbent | 900**
(378.44)
| 4001.92***
(997.28) | 3215.38***
(803.51)
|
| Suppliers | |
| |
| London | -4228.57***
(484.31)
| -7264.69***
(1276.27) | -10181.87***
(1028.30)
|
| Npower | -3278.57***
(484.31)
| -3978.98***
(1276.27) | -4574.72***
(1028.30)
|
| Powergen | -1950***
(484.31)
| -3564.69***
(1276.27) | -5081.86***
(1028.30)
|
| Scottish Power | -4592.85***
(480.52)
| -7178.84***
(1266.29) | -9973.62***
(1020.26)
|
| SSE | -5614.28***
(484.31)
| -7144.69***
(1276.27) | -12517.58***
(1028.30)
|
| Adjusted R2 | 0.7236 | 0.5212
| 0.7581 |
Notes: Standard errors in parentheses.
* Significant at 10%; ** Significant at 5%; *** Significant at
1%
(III) DETERMINANTS
OF ANNUAL
BILL FOR
PREPAYMENT, DECEMBER
2006
| | |
|
| Annual Consumption | 1650kWh
| 3300kWh | 4950kWh
|
| | |
|
| Constant | 21357.97***
(1398.01)
| 36822.46***
(1988.79) | 57343.09***
(2647.43)
|
| Distribution | charge 1.451***
(0.301)
| 1.669***
(0.243) | 1.279***
(0.217)
|
| Distribution area (km2) | -0.055***
(0.018)
| -0.080***
(0.020) | -0.079***
(0.027)
|
| Distribution customers (.000) | -1.033***
(0.284)
| -0.351
(0.395) | -0.411
(0.580)
|
| Incumbent | 715.38
(481.00)
| 1910.89***
(546.24) | 3112.17***
(750.78)
|
| Suppliers | |
| |
| London | -3453.29***
(615.56)
| -9988.04***
(699.06) | -16552.61***
(960.81)
|
| Npower | 1360.98**
(615.56)
| -2388.04***
(699.06) | -6181.18***
(960.81)
|
| Powergen | -1460.44**
(615.56)
| -6195.19***
(699.06) | -10931.18***
(960.81)
|
| Scottish Power | -1987.91***
(610.75)
| -6601.55***
(693.59) | -11280.31***
(953.30)
|
| SSE | -2760.44***
(615.56)
| -9645.19***
(699.06) | -16538.32***
(960.81)
|
| Adjusted R2 | 0.5994 | 0.8295
| 0.8592 |
Notes: Standard errors in parentheses.
* Significant at 10%; ** Significant at 5%; *** Significant at
1%
APPENDIX B
DOES OWNERSHIP UNBUNDLING MATTER? EVIDENCE FROM UK ENERGY
MARKETS[372]
Stephen Davies and Catherine Waddams Price ESRC Centre
for Competition Policy, University of East Anglia November 2007
Ownership unbundling of vertical stages in the energy sector
has become a contentious topic of debate at the end of 2007. To
illustrate the issues, this paper focuses on ownership separation
between the distribution and retail parts of the energy supply
chain, where a mixed experience has emerged in the UK. Ten years
ago both the national gas incumbent and all the electricity incumbents
(monopoly suppliers before the markets were opened to competition)
in each region shared ownership with the local pipes/wires (though
accounting separation had been imposed some time earlier). In
1997 the incumbent gas supplier voluntarily disinvested the pipeline
business, and seven of the fourteen regional electricity companies
have followed suit since then, once separate licenses for the
distribution and retail functions were introduced. If co-ownership
confers advantages on the incumbent, higher incumbent market shares
would be expected in regions where there had been no separation.
This paper explores the evidence for such exploitation of integration,
but first considers the general issues involved and the structure
of the UK energy industry.
ARGUMENTS FOR
AND AGAINST
INTEGRATION
The debate around unbundling in energy concerns the separation
between parts of the industry which have an element of natural
monopoly (national transmission and regional distribution) and
those where there are no obvious economic reasons why the market
should not be competitive (generation and retail). There are four
vertical stages to the energy industry: generating the fuel (from
exploiting gas deposits or imports for gas, from a variety of
sources for electricity); transmission (generally at high pressure
or voltage over fairly long distances); distribution (more local
transportation of energy at lower pressure/voltage, generally
to customers' houses or premises); and the retail function of
selling and billing to the final customer, which generally includes
obtaining the fuel and necessary transportation en route. Most
energy industries have a history of vertical integration over
at least some of these functions, and of established monopolies,
so introducing effective competition may involve some separation
of different vertical (and perhaps horizontal) elements. The essential
arguments in principle can be identified by focusing on this boundary
between distribution and retail, but they should be broadly applicable
to other parts of the supply chain.
In a general model of an upstream natural monopoly and a
potentially competitive downstream market, there are three possible
patterns, each of which has different implications for integration.
If the upstream monopoly is not regulated and the downstream market
is competitive, the upstream distribution company will extract
all the monopoly rent, the downstream retailer is constrained
by competitive pressures, and the outcome will be the same whether
or not the company is integrated. However if the downstream retailer
has some monopoly power (for example from incumbency advantages)
there is a danger that if they are separated both the unregulated
distribution company and the retailer will try to raise price,
resulting in so called "double marginalisation", and
a higher price for the end consumer than if the company were integrated.
In this case of market power in both parts of the supply chain,
the perhaps counterintuitive conclusion is that it would be better
both for consumers and for overall economic welfare to integrate
the two parts of the chain. The third situation is the most common
in practice and relevant to the current discussion. This involves
a regulated monopoly distribution company, and an incumbent who
retains some market power in the retail market. In this case there
is concern about whether a vertically integrated company can influence
the effectiveness of the regulation and so "lever" its
monopoly advantage to deliver (or protect) market power in the
downstream market.
Whereas regulation can in principle ensure that the regulated
distributor does not confer any advantage on a co-owned retailer,
the integrated company has an incentive to increase the price
of the monopoly product and lower the downstream price, thus raising
its rivals' costs in the downstream market, and making its own
retailer more relatively attractive (Bradley and Price, 1991,
Noll and Owen, 1994). Much regulatory theory and practice has
been concerned with addressing such issues. The efficient component
pricing rule (originally developed for the telecoms market, Baumol
and Sidak, 1994) identifies ways of ensuring that an upstream
distributor with monopoly power levies a price which allows efficient
downstream entry but deters inefficient entry. In general, regulators
responsible for such integrated entities require accounting separation
between the two functions, to minimise the chances of exploitation
by reducing the inherent information asymmetry in such situations.
Nevertheless while common ownership persists, so does both the
incentive and the ability to distort prices. The latter can be
achieved by the allocation of costs disproportionately to the
regulated function to raise the charges in that sector. If such
costs are in some sense "common", it is difficult for
the regulator to detect or correct such "biases". The
main concern about allowing common ownership in such cases is
thus that the firm has both the incentive and ability to distort
emerging competition in the downstream market.
However there are counterarguments which may indicate that
integration is better. The natural monopoly of the distribution
pipes means that the efficient price to charge for this element
is below the average cost, and some cases of vertical integration
might enable this. Such pricing would be the reverse of the incentives
to raise the distribution costs discussed above. Nevertheless
there are cases where it would be more efficient to keep the firm
integrated, if the access charge for using the network is (positively)
related to the degree of entry downstream (de Fraja and Waddams
Price, 1999). Proponents of integration also often argue that
common ownership can deliver important sources of efficiency gain.
One example is the transactions costs which arise in cases where
it is very difficult to specify complete contracts between the
different parts of the industry, and so it makes sense to bring
these "in house". Some commentators (BBC, 2006) suggest
that such difficulties account for some of the problems experienced
by the segregated privatised British rail system, where responsibility
has sometimes been difficult to attribute. There may also be information
efficiencies from integration; here the general rule is that decisions
should be made where the information lies. If information is needed
about retail customers, for example for safety purposes, by gas
and electricity distributors, can such information really be effectively
hidden from the retail activities of the same company? "Chinese
walls", designed to separate such activities, are notoriously
difficult to seal in practice, particularly when the employees
on each side of the wall are former colleagues.
Policy makers also need to take into account any "one
off" costs of changing from the current situation. If these
are imposed on unwilling firms, who will bear the costs? Here
the experience of the UK is of some interest. Since divestiture
between the distribution and retail function has been voluntary,
the costs have been borne by the shareholders. However if separation
is imposed by regulators or governments, shareholders might argue
that they should not bear the costs, but that these should be
passed on to consumers.
In the UK, the story of separation is associated with that
of privatisation, but not in a clearly deterministic sense.
THE UK ENERGY
SECTOR AND
INTEGRATION
One of the major criticisms of the 1986 privatisation of
the UK gas industry, which had been nationalised since 1949 and
a national monopoly since 1972, was that the opportunity for both
horizontal and vertical separation was missed: the privatised
incumbent proudly announced that it was responsible for gas and
its delivery "from beach head to meter", ie for the
last three stages in the supply chain. By the time the electricity
industry was privatised four years later some vertical separation
was imposed in England and Wales (between generation and transmission)
but the distribution and retail function remained integrated under
a single license for another 10 years. In Scotland two fully vertically
integrated companies (one serving the north and one the south
of the country) were created, each providing generation, transmission,
distribution and the retail function. Throughout Great Britain
(ie excluding Northern Ireland), the electricity industry retained
its nationalised structure as fourteen separate companies (distributors
and incumbent retailers) in distinct regionally defined markets.
Despite these initial integrated positions, over the last 10 years
the gas incumbent and seven of the fourteen regional electricity
incumbents (table 1) have voluntarily separated themselves from
the associated distribution function. In the case of the gas incumbent
this was under some regulatory pressure, but the mixed result
in the electricity case shows that both common and separated ownership
are chosen outcomes. It is this range of ownership patterns than
enables a test of whether integration adversely affects the development
of downstream competition.
Government ministers had rejected a recommendation by the
Monopolies Commission in 1993 that the gas industry should be
vertically separated before retail competition was introduced,
and instead enacted primary legislation to introduce competition
from 1996 while the incumbent supplier was still vertically integrated
with the transmission and distribution provider. The regulator
sent clear messages that the retail function of the company would
fare better if it was separately owned, and in 1997, in the midst
of market opening, the company itself divested the distribution
and retail functions. Commentators at the time believed that the
retail arm would not prosper, and that the separation was partly
to protect the assets invested in distribution and transmission
from the much riskier retail function. In practice the retail
arm has retained almost half the gas market, and is now the largest
single electricity retailer, supplying about a quarter of the
market (Ofgem, 2007).
Table 1
THE UK ELECTRICITY SUPPLY REGIONS AND OWNERSHIP OF INCUMBENT
AND DISTRIBUTION WIRES IN 2007
| Area | Distribution Wires Owners
| Incumbent Supply Owners | Same Ownership?
|
| East Midlands | Central Networks of E.ON
| PowerGen of E.ON | Y |
| East England | EDF Energy |
PowerGen of E.ON | N |
| London | EDF Energy | EDF Energy
| Y |
| Merseyside, Cheshire & North Wales |
Scottish Power | Scottish Power
| Y |
| Midlands (west) | Central Networks of E.ON
| Npower of RWE | N |
| North East England | CE Electric
| Npower of RWE | N |
| North West | United Utilities
| PowerGen of E.ON | N |
| North Scotland | Scottish and Southern Energy
| Scottish and Southern Energy | Y
|
| South Scotland | Scottish Power
| Scottish Power | Y |
| South East England | EDF Energy
| EDF Energy | Y |
| Southern England | Scottish and Southern Energy
| Scottish and Southern Energy | Y
|
| South Wales | Western Power Distribution (WPD)
| Scottish and Southern Energy | N
|
| South West England | WPD |
EDF Energy | N |
| Yorkshire | CE Electric |
Npower | N |
| |
| |
The retail market in gas was opened on a regional basis between
1996 and 1998, and in electricity across all regions in 1998 to
1999. From May 1999, therefore, all energy consumers have been
able to choose between a range of suppliers. All the incumbents
entered each others' (gas and regional electricity) markets, and
since then there has been considerable consolidation in both retail
and distribution, so that there are now six main retailers (five
consolidated regional electricity incumbents and the national
gas incumbent) and seven distribution company owners. Of these,
four are also major retailers. All companies were required to
impose accounting separation between their distribution and retail
functions. In its review of electricity distribution companies
in 1999, just as competition was starting in the retail market,
the regulator, Ofgem, intervened in the company attributions,
and reallocated over a fifth of companies' costs from the distribution
to the retail function. One company was told to transfer over
one third of its costs. This action by the regulator suggests
that the companies both had incentives to load costs more heavily
onto the distribution function in anticipation of competition,
and that they acted on these incentives.
During the many post-privatisation transactions in which
electricity companies changed hands, a mixture of ownership patterns
for the incumbent suppliers and distribution companies emerged.
The original 14 regional incumbents had reduced to five through
takeover by 2003, and the main suppliers, as they stand in mid
2007, in addition to British Gas, are shown in table 1, along
with their ownership. One main retailer owns no distribution assets;
one owns distribution assets only in (both) the areas where it
is incumbent; one owns them for two of its three incumbency regions,
but not elsewhere; and the remaining two own distribution assets
in some areas where they are incumbent and some where they are
not. In this paper the main focus is in the seven areas where
there is common ownership between the incumbent and the distributor.
In particular, is there any evidence that the incumbent retains
higher market share in those regions where it shares ownership
with the distributor?
Does integration protect incumbent market share?
Figure 1 shows the evolution of incumbent market share in
the 14 regions, labelled according to the status of their joint
ownership (solid lines) or not (dotted lines) in 2007. This graph
provides a useful preliminary overall picture, but it is simplified
because, where ownership did become separated, it happened at
different times since market opening commenced in 1998. Nevertheless,
it does reveal that the region in which the incumbent has retained
the largest market share in 2007 (North of Scotland) is integrated,
while as the regions with the four lowest incumbent market shares
(Midlands, the North West, Northern and Yorkshire) are not; however,
the evidence between these two extremes is mixed.
Figure 1
REGIONAL* INCUMBENT MARKET SHARES
*Solid lines indicate regions where the incumbent and the local
distributor are owned by the same company in 2007; dotted lines
indicate regions where the incumbent is not owned by the same
company as the local distributor.
Source: Ofgem, 2007 and predecessor Ofgem reports
Therefore, to examine this further, a least squares panel
regression has been used to explore whether the market share retained
by the incumbent in each year was related to whether or not it
was integrated with the distributor up to and including that year.
The results are shown in Table 2, in which the dependent variable
is the incumbent's market share, in a given region at a given
point in time, integrated is a binary dummy variable, indicating
whether or not the retailer was integrated with the distributor
in that year. The equation also includes a time trend, to allow
for the natural erosion of market share over time, which will
typically occur in any, previously monopolised, market into which
new entry is introduced. However, this is modelled using a quadratic
time trend (including time squared), to allow for the possibility
that, as consumers become increasingly familiar with the market,
the rate of switching will perhaps slow down after the initial
few years. Since this is a panel model, the equation also controls
for any other differences between the regions, which may remain
even after taking account of integration and the time trend (for
instance, consumers in certain regions of the country may exhibit
more or less loyalty to the incumbent, perhaps because it has
a strong regional identity.
Table 2
MARKET SHARE, TIME TREND AND VERTICAL INTEGRATION 1999-2007
| | |
| |
| Market share | Coef.
| Std. Err. | t
| P>t |
| | |
| |
| Time | -9.125303 | 0.6835588
| -13.35 | 0.000 |
| Time squared | 0.5128004 |
0.0644144 | 7.96 | 0.000
|
| integrated | 4.104698 | 1.664571
| 2.47 | 0.015 |
| constant | 93.93349 | 2.178107
| 43.13 | 0.000 |
| sigma_u | 6.1550725 |
| | |
| sigma_e | 4.0301183 |
| | |
| Rho | 0.69992939 |
| | |
| | |
| |
| | |
| |
The estimated equation includes very striking, and statistically
significant, results on both the time trend and the role of incumbency.
First, as expected, the incumbent's market share does indeed
tend to decline over time: typically, then, incumbents lost market
share year-on-year in all regions. However, the particular values
and signs of the coefficients on time and time squared reveal
that the rate of decline gradually slowed down over the period,
so that, in the last year (2007, year 9), the annual rate of loss
had almost levelled out. On average over the whole time since
market opening, the annual loss of market share by the incumbent
was around 4%, but at much higher rates in the opening years,
and much lower rates in the later years.
Second, and most important for the current discussion this
general reduction in market share, though experienced in all regions,
is found to be significantly slower for companies which are integrated
(as indicated by the positive coefficient on the "integration"
variable.) Thus, on average, in any one year, the market share
of an integrated firm has been more than 4 percentage points higher
than that of a counterpart where different companies own the incumbent
retailer and the associated regional distribution company.
These are the "headline" results, but the estimated
equation also reveals considerable background variation between
regions (not shown in the table). Five regions show similar patterns
of market share reduction: Manweb, Northern, North Western, South
Eastern and East Midlands. Incumbents in the other nine regions
retain significantly higher market shares, even after accounting
for whether or not the incumbent is integrated. In particular,
the north of Scotland, whose incumbent is Scottish Hydro, shows
particularly high incumbent market share, over 20% above that
of the comparator regions, in addition to the higher market share
attributable to its integrated status. Scottish Power, the incumbent
in the southern part of Scotland, also retains a higher market
share than the comparator regions. Both these companies are vertically
integrated not only with distribution, but also with transmission,
which is not allowed in England and Wales.
CONCLUSION
The analysis above appears to provide clear evidence that
those UK incumbent electricity suppliers who remained vertically
integrated with their local distributor have retained a higher
market share than those where these functions have been undertaken
by separately owned companies. This result is evident even after
region specific characteristics, such as different levels of consumer
loyalty, have been included. Competitors have been slower to gain
market share where there is common ownership despite considerable
intervention by the regulator. Its actions have included reallocating
costs (originally attributed to the distribution function by companies)
to the potentially competitive retail function, a regulatory regime
for distribution which is generally regarded as robust, and constant
vigilance by the regulator in the retail market.
We should stress that the above statistical model is relatively
simplistic, and it should be viewed as a piece of documentary
evidence—to be put alongside any other information which
becomes available results. It certainly does not prove that the
companies concerned have been indulging in illegal or improper
behaviour. Nevertheless, the results do suggest that, even with
vigilant regulation and clear accounting separation, incumbents
who are vertically integrated appear to exhibit an advantage in
retaining their market share against the inroads of entrant firms.
As the debate about ownership separation continues in Europe,
this summary of UK experience provides one piece of evidence which
suggests that joint ownership of the distribution function may
indeed confer competitive advantage on the incumbent.
REFERENCES
BBC, News 5 November 2006, http://news.bbc.co.uk/1/hi/uk/6117728.stm,
last accessed 13 November 2007
Baumol, W J and G Sidak, The Pricing of Inputs Sold to Competitors,
YALE J. REG. 171 (1994)
Bradley, I and C Price (1991) Partial and Mixed Regulation of
Newly Privatised UK Monopolies in W Weigel (ed): Economic Analysis
of Law—A Collection of Applications, pp 212-221, Schriftenreihe
der Bundeswirtshaftskammer, Wien, 1991
De Fraja, G and C Waddams Price (1999) Regulation and access pricing:
comparison of regulated regime, Scottish Journal of Political
Economy, 46,(1), pp 1-16
Noll, R G & Bruce M Owen (1994) The Anticompetitive Uses
of Regulation: United States v. AT&T, in THE ANTITRUST
REVOLUTION: THE ROLE OF ECONOMICS 328 (John E Kwoka & Lawrence
J White eds)
Ofgem, 1999, Reviews of Public Electricity Suppliers 1998-2000,
Distribution Price Control Review Draft Proposals, August
Ofgem, 2007, Domestic Retail Market Report—June
APPENDIX C
DO CONSUMERS SWITCH TO THE BEST SUPPLIER?[373]
Chris M Wilson* and Catherine Waddams Price** * Department
of Economics, University of Oxford ** ESRC Centre for Competition
Policy, University of East Anglia
Abstract: This paper suggests that the ability of consumers
to choose accurately between alternative suppliers is substantially
limited even in a relatively simple and transparent market. Across
two independent datasets from the UK electricity market we find,
on aggregate, that those consumers switching exclusively for price
reasons appropriated between a quarter and a half of the maximum
gains available. While such outcomes can be explained by high
search costs, the observation that at least a fifth of the consumers
actually reduced their surplus as a result of switching cannot.
We consider and reject several alternative explanations to pure
decision error.
1. INTRODUCTION
Competition policy and other policy initiatives in markets
as diverse as health and education are increasingly based on the
presumption that consumers can play a positive role in generating
market competition by choosing to trade with the supplier that
best suits their needs. However, consumers may be unable to perform
this role and competitive forces may be consequently weakened
for several reasons. Consumers may be unwilling to change suppliers
because of switching costs, unaware of alternative suppliers because
of search costs or may face difficulties in evaluating and comparing
different suppliers' offers because of cognitive decision-making
costs[374]. While previous
empirical research has largely focussed on identifying the effects
of switching costs, this paper investigates the importance of
the last two possibilities by analysing empirically the accuracy
with which switching consumers choose their best available alternative
supplier.
We exploit two independent datasets from the UK electricity
market where consumers have been free to switch away from their
regional incumbent to one of several entrants since the market's
liberalisation in 1999. In such a market, we would expect consumers'
switching decisions to be relatively accurate for several reasons.
First, almost all households consume electricity and for many,
it forms a significant part of their household budget. Second,
the market is relatively simple as firms supply a near-homogenous
good and at the time of our surveys each supplier effectively
offered only a single tariff option. Third, the market is transparent
with the industry regulator and several online price comparison
services providing many forms of advice and tariff information.
Yet, despite such market conditions, this paper suggests that
the inaccuracy of consumers' switching decisions remains substantial.
Even when focussing only on the consumers who, when asked, indicated
that they had switched suppliers exclusively for price reasons,
we find that across the two datasets and under a range of assumptions,
only 8-19% of consumers switched to the firm offering the highest
surplus and, in aggregate, switching consumers appropriated only
between 28% and 51% of the maximum gains available to them. While
such behaviour is wholly consistent with the behaviour of rational
consumers facing high search costs, the additional finding that
20-32% of switching consumers appear to have lost surplus through
their choice of supplier is not. These consumers lost an average
£14-35 per year in increased bills, apart from any other
switching costs they may have incurred.
Very little previous research has examined empirically the
switching accuracy of consumers. As part of a much wider investigation
into the effects of entry in the New York State telephone market,
Economides et al (2005) suggest that 42% of consumers switched
to a more expensive supplier, resulting in an average loss of
$4.32 per month. Giulietti et al (2005) suggest there may be consumer
inaccuracy in the UK gas market by showing that consumers' (binary)
switching decisions appear unrelated to the monetary gains available
from doing so, especially for consumers who expect price differences
to be transitory. A larger literature however, has analysed the
widespread potential for consumers to select a non-cost minimising
option from a menu of tariffs offered by the same firm. Agarwal
et al (2006), for example, suggest that over 40% of consumers
selected the more expensive tariff when offered the option of
two credit card contracts in a market experiment by a US bank,
while Lambrecht and Skiera (2006) use data from a German internet
provider to estimate that around a third of consumers chose a
more expensive fixed rate tariff, and over half of these paid
more than double the cheapest alternative tariff. The proposed
explanations for such choices fall into three broad categories.
First, consumers may show a preference for certain tariff structures,
such as flat-rate fees (Lambrecht and Skiera 2006). We find no
support for such an explanation as the gains from switching are
largely unrelated to any associated change in tariff structure.
Second, in comparing tariffs, consumers may weight inappropriately
the various components of a tariff or price, such as the introductory
rate, shipping charge or state-tax rate (eg Ausubel 1999, Hossain
and Morgan 2006, Ellison and Ellison 2006, respectively). This
explanation is not supported by our data which show that the gains
made by consumers who switched to suppliers offering a potentially
focal "dual-supply" discount are not significantly different
from the gains made by other consumers. Third, consumers may evaluate
alternative suppliers' tariffs using an incorrect prediction of
their own future consumption (Miravete 2003, Della Vigna and Malmendier
2004, 2006). This explanation also appears unconvincing as all
results are derived from consumers' own (expenditure) beliefs
and remain robust across consumption variations of plus and minus
10 percent.
Highlighted by the recent widespread allegations about such
practices within the industry, one plausible explanation of the
results concerns the pressurising or misleading influence of suppliers'
sales activities. However we find that the accuracy of consumers'
choices are not significantly related to the self-reported influence
of a sales agent; nor does an increased number of regional competitors,
which might result in increased sales activity, consistently reduce
the accuracy of decisions. Instead, the paper concludes that consumers'
switching inaccuracy is consistent with pure decision error. This
finding underlines the importance of the growing research into
the incentives firms may face to exploit or induce consumer confusion—see
Ellison and Ellison (2004) or Armstrong and Spiegler (2007) for
a further discussion.
Section 2 provides a brief theoretical foundation for the
measures of the gains from switching that are later calculated.
Section 3 introduces the market, the data and the calculation
procedures. The descriptive results are presented in section 4.
Section 5 proposes some potential explanations for the results
and presents some further analysis to test them; section 6 concludes.
2. THEORY
To analyse the accuracy of consumers' switching decisions
it is necessary to calculate both the actual gains in surplus
that each consumer made through their choice of new supplier and
the maximum possible gains that each consumer could have achieved
by switching to their best supplier (given their demand characteristics).
We now present some simple measures to form the basis of such
calculations.
Consider consumer i's decision to switch away from
his old supplier, o, to a new supplier, n, chosen
from his set of alternative suppliers, Si. Assuming
that consumer i cares only about the tariff offered by
each supplier, equation (1) describes the approximate annual gain
in consumer surplus (excluding switching costs) from deciding
to switch from supplier o to supplier, n n,
where the consumer surplus received at any firm j consists
of the utility from consuming Cji homogenous
units of electricity annually, ui(Cji),
minus the associated bill expenditure, E(Cji;Tj),
which depends on firm j's tariff, Tj. With the use
of a revealed preference argument to ensure that ui(Coi)—E(Coi
; To) ≥ ui(Cni) —
E(Cni;To) an upper bound for
the actual gains made from such a switching decision, xswi,
is constructed by comparing the expenditures that would result
from consuming the level of post-switching consumption, Cni,
at each supplier, (2). Such an upper bound is very close to the
approximate change in surplus described by (1) when demand is
highly price inelastic, as in the electricity market (Baker et
al 1989).
Similarly an upper bound for the maximum possible
gains that consumer could have made by switching away from supplier,
o, , can be constructed by comparing
the expenditure at i's old supplier with the lowest possible expenditure
available from the set of alternative suppliers, Si
, (3). One final upper bound measures the gains consumer i would
have expected to make by randomly selecting an alternatively supplier,
 . (4) compares the expenditure
at i's old supplier with the average expenditure across supplier
i's set of alternative suppliers.
Fully rational and informed consumers who care only about
the tariffs offered by each firm would select the alternative
supplier that offers the maximum reduction in expenditure, xswi
= . However if they are
rational but not fully informed, perhaps due to the existence
of search costs, switching consumers may be willing to select
a supplier that does not offer the maximum reduction in tariff
expenditure, xswi ∈ [0,]
. As consumers always retain the option of not switching, one
should never observe switching consumers making negative gains,
xswi < 0
3. CALCULATIONS
This section uses the measures constructed in section 2 to
analyse the switching accuracy of two sets of consumers in the
UK electricity market. After an introduction to the market in
section 3.1, section 3.2 presents the data and illustrates how
the UK electricity market is particularly well suited for such
an analysis. Section 3.3 explains how the final calculations are
made.
3.1 The Market
Since liberalisation of the UK residential electricity market
was completed in mid 1999, electricity suppliers have been permitted
to enter each of the fourteen regional markets to compete with
the original regional incumbent. While few new suppliers chose
to enter the industry, many regional incumbents took the opportunity
to enter most, if not all, of the regions in which they had not
previously been incumbent, as did the national gas supplier, British
Gas. Consumers were free to switch away from their regional incumbent
(or any subsequent supplier) with 28 days notice and no financial
penalty. In the subsequent eight years about half of all energy
consumers moved away from their regional incumbent.
An example of the range of tariffs on offer to consumers
is displayed in Table 1. As tariffs vary by region and by time,
Table 1 presents a typical snapshot of the tariffs offered within
an example region, the Midlands, in June 2000. Suppliers are obliged
to offer tariffs for three possible consumer payment methods—standard
credit, direct debit and prepayment, but in practice, only offered
a single tariff per payment method[375].
Suppliers typically offer two-part tariffs, with some offering
three-part tariffs that contain an additional marginal rate for
higher levels of consumption beyond some threshold. The majority
of electricity suppliers who are also active in the gas market
increasingly participate in mixed bundling by offering a dual-supply
discount to those consumers who choose to buy both forms of energy.
While it is common for suppliers to approach consumers directly
in the hope of persuading them to switch, it is rare for suppliers
to use upfront discounts or incentives.
Since liberalisation, many internet-based price comparison
sites have offered consumers advice in choosing between suppliers.
Despite the industry regulator and consumer body endorsing the
use of several comparison sites, their popularity remained limited
in the period of our studies, with only 10% of surveyed consumers
having used them in 2003 (OFGEM 2004).
EXAMPLE SET OF TARIFFS (MIDLANDS REGION, JUNE 2000, IN
PENCE)
| Payment Method:
| | | |
| | | |
| |
| Credit |
| | Direct Debit
| | Prepayment
| | | |
| Electricity Supplier: | Fixed
| Rate1 | Rate2 |
Fixed | Rate1 | Rate2
| Fixed | Rate1 |
Rate2 | Threshold |
Dual-Supply
Discount |
| MEB (Regional Incumbent) | 2159
| 6.72 | - | 2094
| 6.52 | - | 3734
| 6.72 | - | - |
- |
| British Gas | 0 | 10.57
| 5.65 | 0 | 9.01
| 5.65 | 0 | 10.28
| 6.17 | 900 | 1460
|
| Eastern TXU Energi | 2848 |
6.38 | 6.28 | 1856
| 6.38 | 6.28 | 3713
| 6.72 | - | 2392
| - |
| East Midland | 3541 | 5.99
| - | 2491 | 5.99
| - | 5116 | 5.99
| - | - | 250 |
| Independent | 4982 | 5.46
| - | 4026 | 5.46
| - | 4497 | 7.77
| - | - | - |
| London Electricity (1) | 3048
| 5.86 | - | 3048
| 5.86 | - | 9202
| 7.80 | - | - |
- |
| Northern Electric and Gas | 0
| 9.14 | 5.68 | 0
| 8.19 | 5.68 | 3990
| 6.52 | - | 1092
| - |
| Norweb Energi | 4922 | 5.30
| - | 4637 | 5.21
| - | 3734 | 6.72
| - | - | - |
| Seeboard (2) | 0 | 11.97
| 5.34 | 0 | 10.82
| 5.34 | 4112 | 6.72
| - | 728 | - |
| Scottish Hydro | 1873 | 6.08
| - | 1873 | 6.08
| - | 3990 | 6.52
| - | - | - |
| Scottish Power | 5408 | 5.26
| - | 4883 | 5.01
| - | 3734 | 6.72
| - | - | 1050 |
| Southern | 3116 | 6.29
| - | 3053 | 6.16
| - | 3990 | 6.52
| - | - | - |
| SWALEC | 1966 | 5.67
| - | 1886 | 5.44
| - | 3734 | 6.71
| - | - | - |
| SWEB | 3045 | 5.86
| - | 2954 | 5.68
| - | 4523 | 7.39
| - | - | - |
| Utility Link | 3595 | 7.25
| - | 2595 | 7.25
| - | 7388 | 7.68
| - | - | - |
| Yorkshire | 4721 | 5.72
| - | 4091 | 5.76
| - | 8669 | 5.76
| - | - | - |
Each supplier offers a tariff across three payment methods. Each
tariff consists of an (possibly zero) annual fixed fee, Fixed,
with an additional marginal rate, Rate1 in pence/kWh, and,
in some cases, a second marginal rate, Rate2, for consumption
over and above some annual breakpoint, Threshold (in kWh).
Dual supply discounts are offered only to credit or direct debit
consumers (except by East Midland/Powergen who offer them to all
consumers). Additional discounts are labelled with numbers in
brackets—(1) 3% off Direct Debit if bill exceeds £10.50
(2) £8.40 off credit and direct debit.
3.2 Data
Two datasets were constructed from two independent, cross-sectional,
face-to-face surveys of consumers in England, Scotland and Wales.
The EA survey (Cooke et al 2001) was conducted between March and
August 2000 and was intentionally biased towards low-income consumers[376].
Of the 3417 consumers surveyed, 523 had switched electricity suppliers
and, of these, 373 had a full set of responses to questions relevant
for the analysis. In contrast, the CCP survey, was designed to
be representative of the general population and was conducted
for the ESRC Centre for Competition Policy in June 2005[377].
Of the 2027 consumers surveyed, 370 had switched suppliers in
the previous three years, and 245 furnished useable responses.
While the presence of a low-income bias and missing information
limit our ability to draw general inferences about how switching
behaviour varies with consumer characteristics, we view the measurement
of switching accuracy within each of these samples as informative.
A major constraint on the ability to measure consumers' switching
accuracy arises from the possibility that consumers switched for
reasons other than price. Whilst non-price gains are likely to
be small in a near-homogeneous market like electricity, they may
arise from two sources. First, although the reliability of supply
is independent of the supplier (since it depends upon the vertically
separated distribution function), consumers may perceive that
firms vary in attributes such as customer service or environmental
awareness. Second, in addition to the possible monetary benefits
of being supplied electricity and gas by the same supplier, for
which we account for, consumers may perceive some non-price, practical
benefits from having to deal with only one supplier. To eliminate
these possibilities, we restrict our analysis to a subset of consumers
who stated that their switching decision was motivated purely
by price. Specifically, two sub samples are created that contain
318 and 154 consumers respectively who, when asked, cited only
differences in price as a reason for switching and did not mention
factors such as the quality of service, the provision of "environmental"
tariffs or the practical benefits of being dual-supplied. A full
summary of the consumers' (multiple) reasons for switching suppliers
is presented in Tables 2a and 2b.[378]
Tables 2a and 2b
REASONS FOR SWITCHING SUPPLIERS ACROSS THE TWO DATASETS
| Reason for Switching (EA) | Mean
| Reason for Switching (CCP) | Mean
|
| Cheaper | 0.77 | Better Prices/Rates
| 0.86 |
| Dual Supply Discounts | 0.10
| Better Service/Quality | 0.19
|
| Influence of Sales Agent | 0.10
| Not Satisfied with Old Supplier | 0.11
|
| "Conned"/Unaware of switching |
0.03 | Dual Supply | 0.06
|
| Poor Service from Old Supplier | 0.03
| Environmental Tariffs | 0.03
|
| Better Service | 0.02 | Other
| 0.10 |
| No Standing Charge | 0.01 |
n | 245 |
| Other | 0.05 |
| |
| n | 373 | |
|
| | |
|
3.3 Calculating the Gains from Switching
This section provides further details of how the bound measures
constructed in section 2 are used with the selected data samples
to calculate consumers' switching accuracy.
To focus only on the accuracy of consumers' choice of supplier
and not on the choice of payment method or gas supplier, all calculations
are made by comparing suppliers' relevant tariffs whilst treating
each consumer's known choice of payment method(s) and gas supplier
as given. Specifically, the calculations are made using equations
(5)-(7), where the tariff of each supplier, Ttr (m,g),
varies according to the consumer's date of switching, t, electricity
supply region, r, choice of gas supplier, g, and choice of payment
method, m, (both before and after switching).
Using a time series of the unique tariff offered by each
supplier per payment method[379],
an estimate of consumption, Cni,
was calculated from each consumer's own estimate of their average
electricity expenditure[380].
Such an approach offers two advantages. First, it is probably
more accurate as consumers are more likely to recall their expenditure
than their consumption. Second, and more importantly, all gains
are calculated in a way that is consistent with consumers' own
consumption beliefs, so that any inaccurate consumer choices cannot
be attributed to consumers' incorrect consumption estimates. A
potential drawback, however, comes from the possibility that each
consumer's expenditure beliefs may have changed in the intervening
period between the time of the switching decision and the time
of the survey. We take two approaches to allow for this possibility
and to add further robustness to the findings. First, we identify
a subgroup of the EA consumers whose survey responses indicated
that their consumption was highly price inelastic, and stable
over time, and demonstrate that these do not differ significantly
from the rest of the sample[381].
The insignificant difference supports the claims that i) the constructed
upper bounds form close approximations to the true gains from
switching and ii) consumption is likely to be stable between the
time of switching and the time of the survey. Second, we repeat
the three measurements for all consumers using consumption levels
which are plus and minus ten percent of our original estimate.
Whilst the CCP dataset is sufficiently rich to provide all
the required information, the EA dataset does not provide all
the necessary variables directly from the survey because of uncertainty
about the exact date of switching and of any change in payment
method. To proceed we derive the EA calculations under the four
most likely scenarios and compare the results for robustness.
This leads to the specifications, Oct99nochange, Oct99change,
Jun00nochange and Jun00change, which are detailed fully in the
appendix.
4. DESCRIPTIVE RESULTS
Figure 1 plots the estimated actual gains from switching
against the maximum gains available for all consumers (averaging
across the EA specifications outlined above). Two immediate observations
can be made. First, many of the consumers have not appropriated
the maximum gains available, as indicated by the points located
below the 45° line. This is consistent with the behaviour
of rational consumers facing search costs and with experimental
evidence that suggests consumers often search too little (Sonnemans
1998 and Tenorio and Cason 2002). Second, however, a significant
fraction of switchers appear to have actually lost surplus by
switching to a more expensive supplier, as indicated by the points
below the x-axis, a finding which is inconsistent with the behaviour
of rational consumers motivated to switch only by price. To explore
the findings in more detail, Table 3 displays the main results
derived from the original estimates of consumption and Table 4
includes the results with the alternative consumption levels.
Figure 1
THE ACTUAL GAINS MADE FROM SWITCHING RELATIVE TO THE MAXIMUM
GAINS AVAILABLE, CCP AND EA (POOLED SPECIFICATION) DATASETS
Table 3
DESCRIPTIVE STATISTICS OF THE GAIN MEASURES ACROSS A RANGE
OF DATASETS AND SPECIFICATIONS
| Data Specification | CCP
| | EA |
| EA | | EA
| | EA |
| EA | |
| | | Pooled
| | Oct 99 no change
| Oct 99 change | Jun 00 no change
| Jun 00 change |
| Average | (StDev)
| Average | (StDev)
| Average | (StDev)
| Average | (StDev)
| Average | (StDev)
| Average | (StDev)
|
Number of Switchers | 154
| | 318 | |
318 | | 318 |
| 318 | | 318 |
|
| Average Maximum Gains Available (annual, £)
| 49.04 | (39.20) | 44.22
| (42.65) | 43.02 | (42.84)
| 41.42 | (39.91) | 47.08
| (42.85) | 45.35 | (45.00)
|
| Average Mean Gains Available (annual, £)
| 11.43 | (31.16) | 8.80
| (27.14) | 8.62 | (28.02)
| 7.01 | (27.62) | 10.64
| (29.25) | 8.92 | (33.06)
|
| Average Actual Gains Made (annual, £) |
17.92 | (43.18) | 19.41
| (38.56) | 21.36 | (41.57)
| 19.75 | (38.99) | 19.13
| (35.61) | 17.40 | (38.09)
|
| Average Mean Gains/Average Maximum Gains |
0.23 | | 0.20 |
| 0.20 | | 0.17
| | 0.23 | |
0.20 | |
| Average Actual Gains/Average Maximum Gains |
0.37 | | 0.44 |
| 0.50 | | 0.48
| | 0.41 | |
0.38 | |
| Proportion of Switchers with Perfect Gains |
0.18 | | 0.14 |
| 0.18 | | 0.18
| | 0.10 | |
0.10 | |
| Expected Proportion if Random Alternative Selected
| 0.14 | | 0.07
| | 0.07 | |
0.07 | | 0.07 |
| 0.07 | |
| Proportion of Swtichers with Negative Gain |
0.31 | (0.46) | 0.25
| (0.43) | 0.24 | (0.43)
| 0.26 | (0.44) | 0.22
| (0.41) | 0.29 | (0.45)
|
| Average Gain given Negative Gain | -26.96
| (32.99) | -17.56 | (19.16)
| -16.78 | (20.77) | -19.23
| (19.80) | -15.76 | (16.93)
| -18.47 | (19.14) |
| Proportion of Switchers with Non-Negative Gain
| 0.69 | (0.46) | 0.75
| (0.43) | 0.76 | (0.43)
| 0.74 | (0.44) | 0.78
| (0.41) | 0.71 | (0.45)
|
| Average Gain given Non-Negative Gain | 37.64
| (30.55) | 31.85 | (35.29)
| 33.13 | (39.27) | 33.52
| (34.53) | 28.98 | (33.24)
| 31.78 | (34.10) |
| Proportion of Switchers with Dominated Choice
| 0.01 | | 0.06
| | 0.07 | |
0.08 | | 0.03 |
| 0.04 | |
Maximum Gains Available refers to the change in surplus
that would have been realised by a switcher had they switched
to their cheapest alternative supplier. Mean Gains Available
refers to the change in surplus that a switcher would expect to
gain by selecting a supplier randomly. The Proportion of Switchers
with Perfect Gains refers to the proportion of consumers who
appropriated all of the maximum gains available. This is compared
to the expected probability of doing so had the consumer randomly
selected an alternative supplier. The Proportion of Switchers
with Dominated Choice refers to the proportion of consumers
that switched to a tariff that could not be cheaper than their
previous tariff for any level of consumption.
Table 4
COMPARING THE CALCULATED GAIN MEASURES WITH THE PERTURBED
CONSUMPTION LEVELS
| Data Specification | CCP
| | EA |
| EA | | EA
| | EA |
| EA | |
| | | Pooled
| | Oct 99 no change
| Oct 99 change | Jun 00 no change
| Jun 00 change |
| Using Estimated Consumption | Average
| (StDev) | Average | (StDev)
| Average | (StDev) | Average
| (StDev) | Average | (StDev)
| Average | (StDev) |
Average Maximum Gains Available (annual, £)
| 49.04 | (39.20) | 44.22
| (42.65) | 43.02 | (42.84)
| 41.42 | (39.91) | 47.08
| (42.85) | 45.35 | (45.00)
|
| Average Actual Gains Made (annual, £) |
17.92 | (43.18) | 19.41
| (38.56) | 21.36 | (41.57)
| 19.75 | (38.99) | 19.13
| (35.61) | 17.40 | (38.09)
|
| Average Actual Gains/Average Maximum Gains |
0.37 | | 0.44 |
| 0.50 | | 0.48
| | 0.41 | |
0.38 | |
| Proportion of Switchers with Perfect Gains |
0.18 | | 0.14 |
| 0.18 | | 0.18
| | 0.10 | |
0.10 | |
| Using Estimated Consumption -10% |
| | |
| | | |
| | | |
|
| Average Maximum Gains Available (annual, £)
| 47.47 | (37.56) | 42.04
| (38.00) | 41.17 | (41.66)
| 40.97 | (36.27) | 42.44
| (38.21) | 43.57 | (35.85)
|
| Average Actual Gains Made (annual, £) |
20.76 | (41.19) | 18.51
| (34.89) | 20.72 | (40.53)
| 19.27 | (37.05) | 17.42
| (31.99) | 16.64 | (29.99)
|
| Average Actual Gains/Average Maximum Gains |
0.44 | | 0.44 |
| 0.50 | | 0.47
| | 0.41 | |
0.38 | |
| Proportion of Switchers with Perfect Gains |
0.16 | | 0.13 |
| 0.19 | | 0.14
| | 0.10 | |
0.08 | |
| Proportion of Switchers with Negative Gain |
0.25 | (0.43) | 0.23
| (0.42) | 0.24 | (0.43)
| 0.24 | (0.43) | 0.22
| (0.41) | 0.23 | (0.42)
|
| Using Estimated Consumption +10% |
| | |
| | | |
| | | |
|
| Average Maximum Gains Available (annual, £)
| 53.30 | (49.22) | 53.23
| (59.92) | 44.12 | (44.46)
| 43.88 | (38.75) | 51.81
| (47.86) | 73.09 | (108.62)
|
| Average Actual Gains Made (annual, £) |
17.98 | (52.50) | 21.36
| (39.39) | 22.42 | (42.48)
| 20.82 | (39.27) | 21.64
| (39.19) | 20.56 | (36.63)
|
| Average Actual Gains/Average Maximum Gains |
0.34 | | 0.42 |
| 0.51 | | 0.47
| | 0.42 | |
0.28 | |
| Proportion of Switchers with Perfect Gains |
0.14 | | 0.13 |
| 0.19 | | 0.15
| | 0.10 | |
0.08 | |
| Proportion of Switchers with Negative Gain |
0.32 | (0.47) | 0.22
| (0.42) | 0.24 | (0.43)
| 0.24 | (0.43) | 0.20
| (0.40) | 0.21 | (0.41)
|
Maximum Gains Available refers to the change in surplus that would
have been realised by a switcher had they switched to their cheapest
alternative supplier. The Proportion of Switchers with Perfect
Gains refers to the proportion of consumers who appropriated all
of the maximum gains available.
The results shown in tables 3 and 4 are remarkably robust
across datasets, across specifications and across consumption
levels, providing support for the chosen measurement methodology.
Despite including only decisions based exclusively on price, many
consumers failed to switch to the cheapest supplier. Across datasets,
specifications and consumption levels, the reported percentage
of consumers selecting their cheapest supplier ranges between
only 8 and 19%. Although consumers as a whole made positive average
gains of between £16 and £22 per annum, in aggregate,
consumers appropriated only between 28 and 51% of the maximum
benefits available to them.
They have only achieved a little more than would have been
expected by switching to a randomly selected supplier; this would
have offered consumers a 7-14% chance of picking the cheapest
supplier[382] and appropriated
17-23% of the maximum gains available.
More startlingly, even without taking into account the (financial
or non financial) costs of making the switch, between 20 and 32%
of consumers switched to a more expensive supplier, losing, on
average, approximately £14-35 per year. Further, between
3 and 31% of these loss-making consumers actually switched to
a "dominated" tariff that could not have offered them
a reduction in expenditure at any level of consumption. Finally,
although it is difficult to make robust comparisons given the
biases within each of the samples, our data provide no evidence
that switching accuracy improved over the five years which elapsed
between the two surveys.
5. POTENTIAL EXPLANATIONS
The existence of search costs can explain why consumers did
not select the best possible supplier, but the choice of a more
expensive supplier remains puzzling. In this section we explore
the validity of four possible explanations:
i) consumers exhibited some bias or preference for particular
tariff structures;
ii) consumers were overly-attracted to suppliers offering
dual-supply discounts;
iii) consumers were influenced by misleading sales activity;
and
iv) consumers made genuine decision errors.
First, we consider the possibility that consumers' choices
could be explained by a bias or preference for different tariff
structures, as proposed in the literature documenting consumers'
inaccurate tariff choices (eg Lambrecht and Skiera 2006). While
the potential for such biases is limited in our market due to
the narrow range of available tariff structures, we investigate
the potential for consumers to have displayed a preference for
tariff structures in two respects—the number of parts in
the tariff (two or three) and whether or not there is a positive
fixed fee. The evidence for such biases seems limited. Table A1
in the appendix indicates that the estimated switching gains are
largely unrelated to the choice of a two- or three-part tariff;
the only weak evidence of such a bias occurs in the EA June specification
where the 40 consumers who switched from a three- to a two-part
tariff made significantly less accurate decisions than other switchers.
Table 2a shows that only 1% of consumers cited the existence of
a zero fixed fee as a reason for switching and Table A2 in the
Appendix shows that the estimated switching gains are, for the
most part, unrelated to the magnitude of the chosen fixed fee.
The only possibility of a bias occurs within the EA dataset where
the 18 consumers who switched to a positive fixed fee made significantly
worse decisions.
Second, we examine the possibility that consumers could have
overestimated or have been overly sensitive to the dual-supply
discount, as emphasised as an explanation in other contexts by
Ausubel (1999) and Hossain and Morgan (2006). Despite excluding
any consumer who cited the existence of a dual-supply discount
as a reason for switching, this explanation may seem persuasive
since 74% of the consumers in the sample who changed supplier
switched to their gas provider. However, Table A2 in the appendix
indicates that the dual-supplied switchers made, if anything,
higher gains than the non-dual supplied consumers, contradicting
such an explanation. This evidence also eliminates the potential
explanation that consumers may have switched to their gas supplier
to receive some unmeasured non-price benefit.
Third, could consumers have been influenced by suppliers'
mis-selling activity? Such an explanation is particularly plausible
in the UK electricity market where there have been many allegations
of mis-selling. While some complaints have been targeted at internet
price comparison sites for misleading consumers by favouring certain
suppliers[383], most
allegations have been aimed directly at the use of more direct
mis-selling tactics by suppliers themselves. Indeed, the problem
of aggressive or misleading "cold-calling" or doorstep
selling was considered so serious that several bodies conducted
investigations (energywatch 2002, OFGEM 2002 and OFT 2004) and
OFGEM subsequently fined London Electricity two million pounds[384],[385]
5.1 Potential Mis-selling
In this section we estimate whether the consumers'
switching accuracy is related to two sets of test variables associated
with potential mis-selling. We analyse each in turn. First, we
explore whether the accuracy of consumers' switching decisions
is adversely affected by the self-reported influence of suppliers'
sales activity, as captured by two dummy variables from the EA
survey. These correspond to consumers either reporting that they
had been "conned" into switching without their consent,
connedi, or that a sales agent had been active
in their switching decision,[386].
Consumers could cite both influences. To analyse how these variable
relate to switching accuracy, two procedures are used to estimate
variations of equation (6), where the gains from switching,  *, are modelled as a function of the
two test variables agenti and connedi
while controlling for a vector of consumer demographics, Di,
and each consumer's maximum available gains,  .
Table 5
SUMMARY STATISTICS OF THE DEMOGRAPHIC AND TEST VARIABLES
| Variable Name | Variable Definition
| Mean | (StDev)
|
| highsoc | Household social grade: A, B or C1
| 0.28 | (0.45) |
| midsoc | Household social grade: C2 or D
| 0.49 | (0.50) |
| lowsoc | Household social grade: E
| 0.22 | (0.42) |
| highinc | Household income: £25,000+
| 0.13 | (0.33) |
| midinc | Household income: £12,500-£25,000
| 0.25 | (0.43) |
| lowinc | Household income: Less than £12,500
| 0.43 | (0.50) |
| incref | Income status refused
| 0.20 | (0.40) |
| age | Age of respondent |
44.86 | (15.96) |
| single | The household respondent is single
| 0.15 | (0.36) |
| married | The household respondent is married
| 0.62 | (0.49) |
| exmar | The household respondent is widowed or divorced
| 0.23 | (0.42) |
| arrears | The household has electricity arrears
| 0.04 | (0.21) |
| gassw | The household has previously switched gas supplier
| 0.51 | (0.50) |
| rent | The household lives in rented accommodation
| 0.43 | (0.50) |
| disable | The household has some form of disability benefit
| 0.19 | (0.47) |
| agent | The household cited the influence of a sales agent
| 0.11 | (0.31) |
| conned | The household switched without consent
| 0.03 | (0.18) |
| n | The number of regional competitors
| 14.75 | (0.85) |
| Number of Obersevations |
318 | |
| | |
|
A further variable, stablei, is included
to investigate whether the measured switching accuracy of the
sub group of consumers who reported highly price inelastic and
stable consumption differs from the rest of the sample. This variable
is later reported to be insignificantly different from zero, as
discussed previously in Section 3.3. All relevant variables are
described and summarised in Table 5.
We use equation (6) to explore how consumers' switching
gains depend on a set of independent variables in two ways. In
the first case, * is treated
as a latent variable and we estimate the probability of a consumer
making a positive gains using a probit model, and in the second
case, we model the gains from switching as a continuous variable
using OLS with heteroscedasticity-consistent standard errors.
For robustness, the two estimations are conducted across each
of the four EA data specifications and the results are reported
in Tables 6 and 7.
The self-reported incidences of sales and "conning activity"
have no significant effect on switching accuracy across all specifications.
The estimations also indicate, in line with the findings of Economides
et al (2005) and Miravete (2003), that very few demographic variables
are useful predictors of the ability of consumers to make accurate
decisions. Consumers living in rented property make less accurate
decisions, probably because they expect to enjoy any benefits
for a shorter time. Some of the specifications suggest that consumers
with higher incomes (and those who declined to reveal their incomes)
appropriate less of the available gains. Consumers are less likely
to make a loss from switching suppliers if the maximum gains available
are higher, a finding consistent with consumers having a higher
incentive to make an accurate decision when the rewards from doing
so are greater.
Table 6
ESTIMATIONS OF THE PROBABILITY OF MAKING A POSITIVE GAIN[387]
| June
| June | October
| October |
| No Method Change
| Method Change | No Method Change
| Method Change |
| M.Effct | z
| M.Effect | z |
M.Effect | z | M.Effect
| z |
agent | 0.03 | 0.53
| -0.16 | -1.62 | 0.08
| 1.39 | 0.04 | 0.61
|
| conned | -0.18 | -1.16
| -0.23 | -1.24 | 0.07
| 0.79 | -0.07 | -0.45
|
| gainmax | 0.00 | 4.23**
| 0.01 | 7.16** | 0.01
| 5.52** | 0.01 | 7.18**
|
| stable | -0.03 | -0.55
| -0.02 | -0.46 | -0.05
| -1.04 | -0.06 | -1.31
|
| highsoc | -0.01 | -0.11
| -0.07 | -0.74 | 0.02
| 0.21 | -0.08 | -0.89
|
| midsoc | -0.02 | -0.39
| -0.07 | -1.00 | -0.05
| -0.79 | -0.14 | -2.12*
|
| highinc | -0.24 | -2.03*
| -0.22 | -1.78 | -0.13
| -1.21 | -0.16 | -1.37
|
| lowinc | -0.05 | -0.69
| -0.04 | -0.55 | -0.03
| -0.43 | -0.09 | -1.40
|
| incref | -0.09 | -1.13
| -0.11 | -1.21 | -0.08
| -1.05 | -0.10 | -1.17
|
| age | 0.00 | 0.63
| -0.01 | -0.83 | 0.00
| 0.30 | 0.00 | 0.21
|
| age2 | 0.00 | -0.71
| 0.00 | 0.78 | 0.00
| -0.01 | 0.00 | 0.12
|
| disable | -0.05 | -0.96
| -0.07 | -1.25 | 0.00
| -0.01 | -0.04 | -0.70
|
| single | -0.10 | -1.17
| -0.08 | -0.86 | -0.12
| -1.33 | -0.21 | -2.07
|
| exmar | 0.01 | 0.09
| 0.03 | 0.50 | 0.02
| 0.29 | 0.02 | 0.29
|
| rent | -0.15 | -2.87**
| -0.16 | -2.58** | -0.10
| -1.93 | -0.14 | -2.55**
|
| arrears | 0.03 | 0.27
| -0.01 | -0.05 | 0.09
| 1.29 | 0.08 | 1.02
|
| gassw | -0.12 | -2.77**
| -0.12 | -2.44* | -0.05
| -1.20 | -0.04 | -0.84
|
| n | 318 | |
318 | | 318 |
| 318 | |
| Log-Lik | -141.7 |
| -145.6 | | -144.3
| | -137.0 | |
| LR(17) | 51.90** |
| 89.65** | | 58.78**
| | 91.07** |
|
| McF R2 | 0.15 |
| 0.24 | | 0.17
| | 0.25 | |
| |
| | | |
| | |
There is no evidence that previous experience improves decision
accuracy. While Giulietti et al (2005) suggest that consumers
are more likely to switch in a given market if they have previously
switched in others, we find that a past experience of switching
gas suppliers does nothing to improve (and sometimes reduces)
switching accuracy.
Table 7
ESTIMATIONS OF THE GAINS MADE FROM SWITCHING[388]
| June
| June | October
| October |
| No Method Change
| Method Change | No Method Change
| Method Change |
| Coeff | t
| Coeff | t |
Coeff | t | Coeff
| t |
| agent | 0.70 | 0.14
| -2.57 | -0.49 | -2.10
| -0.41 | -3.34 | -0.58
|
| conned | 0.22 | 0.05
| -0.03 | -0.01 | -3.64
| -0.45 | -3.75 | -0.49
|
| gainmax | 0.01 | 9.43**
| 0.01 | 11.05** | 0.01
| 14.24** | 0.01 | 10.55**
|
| stable | 0.93 | 0.31
| 0.79 | 0.27 | -1.22
| -0.42 | -1.54 | -0.54
|
| highsoc | -4.21 | -0.90
| -2.98 | -0.61 | -2.26
| -0.56 | -2.28 | -0.54
|
| midsoc | -3.88 | -1.00
| -3.91 | -0.95 | -3.08
| -0.85 | -4.45 | -1.16
|
| highinc | -13.90 | -2.21*
| -13.23 | -2.08* | -1.08
| -0.19 | -0.36 | -0.06
|
| lowinc | -5.12 | -1.39
| -5.80 | -1.50 | 1.89
| 0.52 | 1.55 | 0.41
|
| incref | -13.57 | -3.22**
| -13.73 | -3.22** | -6.87
| -1.57 | -5.63 | -1.41
|
| age | -0.02 | -0.04
| -0.27 | -0.51 | 0.39
| 0.81 | 0.27 | 0.55
|
| age2 | 0.00 | 0.18
| 0.00 | 0.66 | 0.00
| -0.25 | 0.00 | 0.04
|
| disable | -4.87 | -1.30
| -4.52 | -1.16 | -6.53
| -1.77 | -6.30 | -1.71
|
| single | -5.66 | -1.25
| -4.94 | -1.06 | -0.33
| -0.08 | -3.25 | -0.75
|
| exmar | -0.49 | -0.16
| -0.33 | -0.10 | 0.16
| 0.05 | -0.44 | -0.13
|
| rent | -6.08 | -2.17*
| -4.54 | -1.58 | -8.40
| -2.77** | -7.71 | -2.46*
|
| arrears | -8.98 | -1.21
| -8.22 | -1.08 | -4.17
| -0.66 | -4.48 | -0.72
|
| gassw | -3.92 | -1.33
| -3.44 | -1.15 | -4.27
| -1.53 | -3.32 | -1.20
|
| constant | 5.28 | 0.38
| 7.29 | 0.52 | -15.52
| -1.23 | -12.03 | -0.92
|
| n | 318 | |
318 | | 318 |
| 318 | |
| F(17,300) | 10.34** |
| 14.06** | | 18.37**
| | 14.06** |
|
| R2 | 0.58 | |
0.61 | | 0.68 |
| 0.64 | |
| |
| | | |
| | |
To provide a further (less direct) test of the effects of
mis-selling, the estimations are repeated with the inclusion of
a different test variable—the number of competitors in each
consumer's regional market. While conventional theories of consumer
search do not predict any negative relationship between consumers'
ability to appropriate the gains available and the number of competitors[389],
it is reasonable to conjecture that mis-selling strategies may
be more attractive to firms as the profits from more standard
forms of competition are reduced from increases in the number
of suppliers. In a related sense, recent work by Spiegler (2005)
illustrates how firms face an increased incentive to obfuscate
by increasing the variance of their utility offers when faced
with more competitors, while Miravete (2007) offers evidence to
suggest that firms are more likely to employ dominated tariff
options when competition increases. To test for such an effect,
we exploit the fact that the number of regional competitors varied
between twelve and sixteen at the time of the EA survey[390].
If mis-selling were an explanation, consumers would make less
accurate decisions in regional markets with a higher number of
competing suppliers[391].
Formally, the two estimation procedures are repeated with
the replacement of the previous test variables, agenti
and connedi, with the new test variable, ni,
measuring the number of regional suppliers faced by each consumer[392].
As the estimated coefficients differ very little from those previously
reported, only the effects of the test variable are displayed
in Tables 8 and 9.
Table 8
ESTIMATED MARGINAL EFFECTS OF THE NUMBER OF REGIONAL COMPETITORS
ON THE PROBABILITY OF SWITCHING TO MAKE A POSITIVE GAIN[393]
| June
| June | October
| October |
| No Method Change
| Method Change | No Method Change
| Method Change |
| M.Effct | z
| M.Effect | z |
M.Effect | z | M.Effect
| z |
n | -0.01 | -0.54
| 0.03 | -0.96 | -0.04
| -1.43 | -0.05 | -1.77
|
| |
| | | |
| | |
Table 9
ESTIMATED MARGINAL EFFECTS OF THE NUMBER OF REGIONAL COMPETITORS
ON THE ACTUAL GAINS MADE FROM SWITCHING
| June
| June | October
| October |
| No Method Change
| Method Change | No Method Change
| Method Change |
| Coeff | t
| Coeff | t |
Coeff | t | Coeff
| t |
n | -3.76 | -2.47*
| -3.84 | -2.47* | -1.87
| -0.99 | -2.66 | -1.36
|
| |
| | | |
| | |
While there is no evident relationship between the number
of regional competitors and the probability of making a positive
gain by switching, Table 9 suggests that in two out of four specifications,
consumers appropriated relatively less of the maximum available
gains in regions with a higher number of suppliers. However as
much of the variation in the number of regional competitors arises,
however, from the relative lack of market entry in the two Scottish
electricity regions, such a finding is also consistent with the
presence of some unobserved characteristic of firms or consumers
within the Scottish markets. The results are therefore unclear
and do not provide direct evidence that mis-selling explains the
inaccuracy of consumers' switching decisions.
The evidence presented in this section does not indicate
that consumers' poor switching choices are explained by tariff
biases or suppliers' mis-selling activity. We deduce that much
of the switching inaccuracy results from genuine consumer confusion
and decision error.
6. CONCLUSION
Using two independent datasets from the UK electricity market
our results show that the capacity of consumers, to choose efficiently
between suppliers may be limited, even when switching purely for
price reasons. While the results are not necessarily representative
of the general population, our estimations show that, at best,
a fifth of the consumers in our samples actually lost surplus
as a result of switching; and that, in aggregate, switching consumers
appropriated only half of the maximum gains available to them.
Such a failure of consumers to compare accurately between alternative
suppliers can damage their welfare, both directly in lost savings,
and indirectly by delivering firms with a source of market power.
Indeed, together with the well established effects of switching
costs in reducing the willingness of consumers to switch suppliers,
such behaviour may seriously impede the competitive process, even
after a market has been liberalised or made subject to standard
competition policy (as recently argued by Waterson 2003).
We have examined and rejected several explanations of consumer
errors, including preferences for particular tariff structures
or dual fuel supply, and misleading sales activities by firms.
Instead, despite the apparent simplicity and transparency of the
market, consumers' poor choices seem more consistent with an explanation
of pure decision error. This finding casts doubt on the ability
of consumers to generate competitive forces through accurate switching
decisions and raises many important policy concerns. Future research
would be valuable in understanding how competition and consumer
authorities should respond to consumer errors, if at all, and
in investigating the implications for current policies aiming
to increase competition in less familiar markets, such as health
and education.
References
Agarwal S, Chomsisengphet S, Liu C and Souleles N S (2006) "Do
Consumers Choose the Right Credit Contracts?" Federal
Reserve Bank of Chicago Working Paper 2006-11
Armstrong M and Spiegler R (2007) "Boundedly Rational
Consumers" June 2007
Ausubel L M (1999) "Adverse Selection in the Credit Market"
Working Paper University of Maryland
Baye M R, Morgan J and Scholten P (forthcoming) "Information,
Search and Price Dispersion" in Handbook on Economics
and Information Systems, Elsevier
Baker P, Blundell R and Micklewright J (1989) "Modelling
Household Energy Expenditure Using Micro-Data" Economic
Journal vol.99 p.720-738
Chang Y and Waddams Price C (forthcoming) "Search and
Switching Across Relationship Markets" ESRC Centre for
Competition Policy Working Paper
Cooke D, Ferrari A, Giuletti M, Sharratt D and Waddams Price C
(2001) "Affording Gas and Electricity: Self Disconnection
and Rationing by Prepayment and Low Income Credit Consumers and
Company Attitudes to Social Action" www.ccp.uea.ac.uk/publications.asp#2001
Della Vigna S and Malmendier U (2004) "Contract Design
and Self-Control: Theory and Evidence" Quarterly Journal
of Economics vol. 119(2) p.353-402
Della Vigna S and Malmendier U (2006) "Paying Not to Go
to the Gym" American Economic Review vol. 96 (3) p. 694-719
Economides N, Seim K and V B Viard (2005) "Quantifying
the Benefits of Entry into Local Phone Service" NET Institute
Working Paper, October 2005
Ellison G and Ellison S F (2004) "Search, Obfuscation
and Price Elasticities on the Internet" NBER Working
Paper 10570
Ellison G and Ellison S F (2006) "Internet Retail Demand:
Taxes, Geography and Online-Offline Competition" Working
Paper April 2006
energywatch (2002) "Doorstep and Telephone Selling: A
Guide for Gas and Electricity Consumers" Information
Sheet
Farrell J and Klemperer P (2006) "Coordination and Lock-In:
Competition with Switching Costs and Network Effects"
Forthcoming in Handbook of Industrial Organization Vol 3, M Armstrong
and R Porter (eds), North-Holland
Gabaix X, Laibson D and Li H (2005) "Extreme Value Theory
and the Effects of Competition on Profits" Working Paper
March 2005
Giulietti M, Waddams Price C and Waterson M (2005) "Consumer
Choice and Competition Policy: A Study of UK Energy Markets"
Economic Journal vol. 115 p.949-968
Hossain T and Morgan J (2006a) "Plus Shipping and Handling:
Revenue (Non) Equivalence in Field Experiments on eBay"
Advances in Economic Analysis and Policy vol. 6(2)
Iyengar S S and Lepper M (2000) "When Choice is Demotivating:
Can One Desire Too Much of a Good Thing?" Journal of
Personality and Social Psychology vol.79 p.995-1006
Iyengar S S and Kamenica E (2006) "Choice Overload and
Simplicity Seeking" Working paper, October 2006
Kohn M G and Shavell S (1974) "Optimal Adaptive Search"
Journal of Economic Theory vol. 9 p.93-123
Lambrecht A and Skiera B (2006) "Paying Too Much and Being
Happy About It: Existence, Causes and Consequences of Tariff Choice
Biases" Journal of Marketing Research, forthcoming, May
2006
Miravete E J (2003) "Choosing the Wrong Calling Plan?
Ignorance and Learning" American Economic Review vol.
93(1) p.297-310
Miravete E J (2007) "The Doubtful Profitability of Foggy
Pricing" Working Paper October 2007
OFGEM (2002) "OFGEM Keeps up Pressure on Suppliers to
Reduce Misselling" The Office of Gas and Electricity
News, Press Release 30 January 2002
OFGEM (2004) "Domestic Competitive Market Review"
April 2004
OFT (2004) "Doorstep Selling: A Report on the Market Study"
Office of Fair Trading, May 2004
Sonnemans J (1998) "Strategies of Search" Journal
of Economic Behavior and Organization vol.35, p.309-332
Spiegler R (2005) "Competition Over Agents with Boundedly
Rational Expectations" Theoretical Economics vol. 1(2)
p.207-231
Tenerio R and Cason T N (2002) "To Spin or Not to Spin?
Natural and Laboratory Experiments from The Price is Right"
Economic Journal vol.112 p.170-195
Waddams-Price C (2003) "Spoilt for Choice? The Costs and
Benefits of Opening UK Residential Energy Markets" Working
Paper December 2003
Waterson M (2003) "The Role of Consumers in Competition
and Competition Policy" International Journal of Industrial
Organization vol.21 p. 129-150
APPENDIX C1
IDENTIFYING TARIFFS FOR THE EA DATASET
Two aspects of the EA dataset make it difficult to identify
directly the exact set of tariffs relevant for each consumer's
switching decision. The first is the exact date of the switching
decision. (Economides et al (2005) faced the same problem and
were forced to assume that consumers had switched at the date
of information collection.) The second problem arises from the
timing of the change in payment method for the 32% of consumers
who reported such a change. To calculate the gains on switching
we need to know whether they, changed their payment method before,
after, or at the same time as they switched suppliers. To resolve
these uncertainties and to enhance the robustness of our findings
we report the results over four different specifications. As the
EA survey was conducted in March-August 2000, very soon after
liberalisation, consumers could have switched using one of only
four possible tariff sets, namely those commencing in June 1999,
October 1999, April 2000 and June 2000. Consumers are most likely
to have switched under either the October 1999 tariffs, as these
were stable for the longest period (October 1999-April 2000),
or the June 2000 tariffs, as the proportion of consumers switching
suppliers was rising over the period. Using both of these time
periods, the calculations are then made under two further assumptions
to provide a total of four specifications. These two assumptions
concern whether the 32% of consumers who had changed their payment
method, changed either before they switched suppliers (the consumers
traded with both their original and current supplier under their
current payment method) or, perhaps more realistically, at the
time of switching (the consumers traded with their original supplier
using their previous payment method but traded with their current
supplier under their current payment method)[394].
The four specifications are respectively labelled as Oct99nochange,
Oct99change, Jun00nochange and Jun00change (see appendix for further
details).
APPENDIX C2[395]
Table A1
SWITCHING ACCURACY BY CHANGES IN CHOSEN TARIFF STRUCTURE
| No Change in Tariff Structure
| Three-part to Two-part |
Two-part to Three-part |
| CCP Data | Average
| (StDev) | Average
| (StDev) | Average
| (StDev) |
Number of Switchers | 74
| | 50 | |
30 | |
| Proportion of Switchers | 0.48
| | 0.32 | |
0.19 | |
| Average Maximum Gains Available (annual, £)
| 44.25 | (32.46) | 53.16
| (44.10) | 53.99 | (45.35)
|
| Average Actual Gains Made (annual, £) |
14.82 | (41.96) | 23.18
| (50.25) | 16.80 | (32.63)
|
| Average Actual Gains/Average Maximum Gains |
0.33 | | 0.44 |
| 0.31 | |
| Proportion of Switchers with Negative Gain |
0.30 | (0.46) | 0.28
| (0.45) | 0.37 | (0.49)
|
| EA Data (Pooled June Specification) |
Average | (StDev) |
Average | (StDev) |
Average | (StDev) |
| Number of Switchers | 169 |
| 40 | | 109
| |
| Proportion of Switchers | 0.53
| | 0.13 | |
0.34 | |
| Average Actual Gains Made (annual, £) |
12.50 | (29.24) | -3.38**
| (27.69) | 35.14** | (41.33)
|
| Average Maximum Gains Available (annual, £)
| 42.17 | (36.40) | 34.69
| (29.29) | 56.72* | (54.29)
|
| Average Actual Gains/Average Maximum Gains |
0.30 | | -0.10 |
| 0.62 | |
| Proportion of Switchers with Negative Gain |
0.31 | (0.45) | 0.55**
| (0.46) | 0.06** | (0.20)
|
| EA Data (Pooled October Specification)
| Average | (StDev)
| Average | (StDev)
| Average | (StDev)
|
| Number of Switchers | 226 |
| 78 | | 14
| |
| Proportion of Switchers | 0.71
| | 0.25 | |
0.04 | |
| Average Actual Gains Made (annual, £) |
19.85 | (43.01) | 23.28
| (24.11) | 16.84 | (32.72)
|
| Average Maximum Gains Available (annual, £)
| 41.42 | (42.66) | 41.85
| (28.43) | 57.19 | (47.58)
|
| Average Actual Gains/Average Maximum Gains |
0.48 | | 0.56 |
| 0.29 | |
| Proportion of Switchers with Negative Gain |
0.27 | (0.43) | 0.18
| (0.37) | 0.21 | (0.43)
|
| |
| | | |
|
Table A2
SWITCHING ACCURACY BY CHANGES IN CHOSEN FIXED FEE TARIFF
STRUCTURE[396]
| No Change in Tariff Structure
| Pos. Fixed Fee to Zero Fixed Fee
| Zero Fixed Fee to Pos. Fixed Fee
|
| CCP Data | Average | (StDev)
| Average | (StDev) | Average
| (StDev) |
| Number of Switchers | 69 |
| 29 | | 56
| |
| Proportion of Switchers | 0.45
| | 0.19 | |
0.36 | |
| Average Actual Gains Made (annual, £) |
19.46 | (43.29) | 19.43
| (32.43) | 15.25 | (48.19)
|
| Average Maximum Gains Available (annual, £)
| 46.97 | (33.38) | 55.08
| (45.67) | 48.46 | (42.55)
|
| Average Actual Gains/Average Maximum Gains |
0.41 | | 0.35 |
| 0.31 | |
| Proportion of Switchers with Negative Gain |
0.28 | (0.45) | 0.28
| (0.45) | 0.36 | (0.48)
|
| EA Data (Pooled June Specification) |
Average | (StDev) |
Average | (StDev) |
Average | (StDev) |
| Number of Switchers | 156 |
| 144 | | 18
| |
| Proportion of Switchers | 0.49
| | 0.45 | |
0.06 | |
| Average Actual Gains Made (annual, £) |
8.79 | (29.44) | 32.78**
| (37.93) | -15.72** | (22.08)
|
| Average Maximum Gains Available (annual, £)
| 39.93 | (35.04) | 53.73**
| (50.27) | 40.64 | (40.53)
|
| Average Actual Gains/Average Maximum Gains |
0.22 | | 0.61 |
| -0.39 | |
| Proportion of Switchers with Negative Gain |
0.36 | (0.46) | 0.08**
| (0.23) | 0.78** | (0.39)
|
| |
| | | |
|
Table A3
SWITCHING ACCURACY OF DUAL AND NON-DUAL SUPPLIED CONSUMERS
| Not Dual Supplied
| Dual Supplied |
| CCP Data | Average
| (StDev) | Average
| (StDev) |
| Number of Switchers | 29 |
| 125 | |
| Proportion of Switchers | 0.19
| | 0.81 |
|
| Average Actual Gains Made (annual, £) |
15.36 | (62.37) | 18.52
| (37.68) |
| Average Maximum Gains Available (annual, £)
| 48.07 | (49.43) | 49.27
| (36.66) |
| Average Actual Gains/Average Maximum Gains |
0.32 | | 0.38 |
|
| Proportion of Switchers with Negative Gain |
0.45 | (0.51) | 0.27
| (0.45) |
| EA Data (Pooled Specifications) |
Average | (StDev)
| Average | (StDev)
|
| Number of Switchers | 96 |
| 222 | |
| Proportion of Switchers | 0.30
| | 0.70 |
|
| Average Actual Gains Made (annual, £) |
10.45** | (43.17) | 23.29
| (30.95) |
| Average Maximum Gains Available (annual, £)
| 46.87 | (50.42) | 43.07
| (34.82) |
| Average Actual Gains/Average Maximum Gains |
0.22 | | 0.54 |
|
| Proportion of Switchers with Negative Gain |
0.39** | (0.40) | 0.19
| (0.34) |
March 2008
| | | |
|
372
Published in December 2007 edition of Intereconomics. Back
373
Published as CCP working paper 07-6 Back
374
See Farrell and Klemperer (2006) for a review of the market power
effects of switching costs, Baye et al (forthcoming) for search
costs, and Gabaix et al (2005) for cognitive costs. Back
375
More recently suppliers have offered a wider choice of tariffs,
including "capped" tariffs, but these were not available
at the time of the consumer decisions analysed here. Back
376
The EA survey and its initial analysis were funded by the Electricity
Association-an early description of consumers' choices and errors
is contained in Waddams Price (2003). Back
377
The CCP survey was designed to analyse search and switching behaviour
across eight different product markets as analysed by Chang and
Waddams Price (forthcoming). Here, only the data from the electricity
market is used. Back
378
The EA respondents were asked to provide an unstructured explanation
for why they had switched, which was later coded into an exclusive
list of reasons, whereas the CCP respondents were asked to indicate
up to three reasons from a list of possible options. No distinction
was made between price and non-price benefits of dual-supply and
so all consumers who cited dual-supply as a reason for changing
suppliers are eliminated from the sample. Back
379
The tariff dataset builds on that used by Giulietti et al (2005)
and was obtained by either contacting suppliers directly or downloading
bimonthly tariffs from a consumer advice website, www.which.co.uk
or the energy consumer body, www.energywatch.org.uk. Back
380
Consumers were asked to provide an estimate of their expenditure
on a weekly, fortnightly, monthly or quarterly basis as they preferred. Back
381
The subgroup of consumers indicated high price inelasticity by
replying "the same" to the following questions: Q. If
the cost of electricity went down would you use more electricity
or use the same electricity and use the savings for something
else?, and Q. If the cost of electricity went up would you use
less electricity or use the same electricity?, and further indicated
a stable consumption pattern by replying "No" to the
following questions, Q. Has there been any change in your household's
circumstance in the last two to three years that affected your
fuel consumption?, and Q. Has your household's electricity ever
been disconnected because of unpaid electricity bills? Back
382
This figure was calculated by finding the reciprocal of the number
of alternative suppliers, averaged across consumers, given their
respective regions. The probability doubles to 0.14 for the later
CCP dataset due to the heavy market consolidation in recent years. Back
383
See http://business.guardian.co.uk/story/0,,1975484,00.html. 19
December 2006. Back
384
See http://news.bbc.co.uk/1/hi/business/2315115.stm. 10 October
2002. Back
385
We find no evidence that those consumers who switched to London
Electricity made significantly different gains to those who switched
to other suppliers. Back
386
The CCP data do not include these variables Back
387
All significant tests are indicated by * for the 5% level and
by ** for the 1% level. Where applicable, all marginal effects
are calculated for the average switcher relative to the base case
of a consumer who is married, of low social class and with middle
income. Back
388
All significant tests are indicated by * for the 5% level and
by ** for the 1% level. Where applicable, all coefficients are
estimated relative to the base case of a consumer who is married,
of low social class and with middle income. Back
389
Indeed, for any given price distribution and cost of search, a
consumer should accept any discovered price below the optimal
reservation price which is defined independently from the number
of firms (Kohn and Shavell 1974). Back
390
These numbers refer to the number of large firms that were patronised
by consumers in the EA sample and do not include some smaller
firms that also operated across all regions. Including such firms
in the estimations increases the number by a constant and does
not affect our qualitative results. No such variation in firm
numbers existed at the time of the CCP survey due to later market
consolidation. Back
391
It is feasible, but unlikely given the limited variation in the
number of firms, that consumer inaccuracy may also be prompted
by a `choice overload' effect from the increased complexity of
the decision (eg Iyengar and Lepper 2000 and Iyengar and Kamenica
2006). Back
392
Both the number of competitors and the maximum gains can be included
as explanatory variables, since they have a negligible correlation
of approximately 0.02 across specifications. Back
393
Significance is denoted at 5% by * and at 1% by **. Back
394
The most commonly reported method changes are moving from credit
to direct debit (41%) and credit to prepayment (38%). We do not
allow for the unlikely possibility that the change was made after
the process of changing suppliers. Back
395
Notes for Tables A1-A3. ** and * are used to indicate a significant
difference in means under both a standard t-test and a non-parametric
Mann-Whitney U test at the 5% and 1% respectively. Back
396
Tariffs with positive fixed fees were so common within the EA
Pooled October Specification that all consumers switched to such
tariffs, preventing us from testing such a hypothesis. Back
|
