You are on page 1of 13

Making Markets Work:

The Special Case of Electricity


In many ways electricity markets resemble markets
generally, but they also have some distinctive features that
John Kwoka is the Neal F.
Finnegan Distinguished Professor of make pure reliance on market processes problematic. Thus,
Economics at Northeastern prescriptions for deregulation and restructuring that
University. His areas of
specialization are industrial might succeed in other markets cannot be relied upon to
organization, regulation, and have the same effect in electricity. At worst, the result of
antitrust economics. His research
covers a range of topics, but most
unthinking reform can be a perfect storm of forces
recently has focused on the effects of leading to serious problems in market operation.
various restructuring policies in the
electricity industry. He is also a
Research Fellow of the American John Kwoka and Kamen Madjarov
Antitrust Institute and a member of
the Board of Directors of the
Industrial Organization Society, and
has previously served at the Federal I. Introduction and overall performance in
Trade Commission, the Federal particular industries. Where
Communications Commission, and Economics looks to competitive market problems arise, policy
the Antitrust Division of the Justice markets for its performance prescriptions are usually based on
Department. He can be contacted at benchmark. Economic theory, a return to the preconditions of
j.kwoka@neu.edu empirical evidence, and competition.
Kamen Madjarov is a Ph.D. experience all demonstrate that The benefits of competitive
student at Northeastern University. competitive markets generally markets have been realized
He received his undergraduate degree result in least-cost production, through deregulation of many
in Economics from Adelphi appropriate quantity and quality industries or, where full
University and is currently writing of product, optimal pricing to deregulation is not feasible, by
his dissertation on topics that include consumers, and strong incentives introducing competition more
restructuring in the electricity
for innovation. As a result, selectively.1 Electricity is the last
industry.
regulatory and antitrust policies of the major industries to be
are generally guided by the restructured and deregulated for
competitive model in assessing the simple reason that it has some
market structure, firm conduct, problematic features that make

24 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal
conventional competition firms. This condition is less easily of the full social costs of their
unlikely and competition in any measured than concentration, but activity. Otherwise, even what
form quite possibly fragile. equally important.4 appears to be market equilibrium
These problematic features (3) Homogeneous product. will reflect only private costs and
have drawn renewed attention as Homogenous products are benefits, not their full economic
it has become clear that in more direct competition valuation.
restructuring to date has with each other, while If all of these conditions hold, a
failed to deliver the promised differentiated products face fundamental theorem of
benefits.2 Here we take a closer weaker competitive constraints. economics demonstrates that total
look at the question of the Product heterogeneity is often surplus in the market is
problematic, and perhaps unique, measured indirectly, as with maximized. That is, the maximum
features of electricity markets. advertising intensity for total of consumer and producer
consumer goods.5
I n what follows, we first review
the preconditions for
competition together with some
benefit is created, given the cost of
inputs, consumer preferences, etc.
Output is produced in an efficient
variations in assumptions and manner and allocated to those
implications of the competitive
This is in its own consumers who value that output
model. Then we discuss the way a truly remarkable most highly. This theorem forms
factors often noted as proposition, and can the basis for the preference for
distinctive and problematic competitive markets. It states that
provide a benchmark
about electric power markets certain preconditions, together
and propose a framework against which to with surplus maximizing
for understanding which factors measure the operation behavior by consumers and
are most responsible for the of actual markets. producers, will by themselves
difficulty of creating result in the most efficient market
well-functioning electricity outcome. No other output or price
markets. or production method can
(4) Perfect information about improve on this outcome in terms
current and future market condi- of an unambiguous surplus
II. Competitive Markets tions. Rational choice requires full increase (though many may
information about alternatives redistribute surplus between
A conventional list of available to consumers as well as groups).8
preconditions for competitive
markets includes the following
assumptions:
about future market conditions.
The absence of perfect informa-
tion can have a variety of adverse
T his is in its own way a truly
remarkable proposition.
When interpreted and used
(1) Large numbers of buyers effects.6 properly, it provides a benchmark
and sellers, none dominant. (5) The absence of transaction against which to measure the
Large numbers of relatively small costs. Perfect competition operation of actual markets.
firms are usually necessary to assumes that there are no costs to Where market operation is
ensure against pricing power. The the act of exchanging of goods and deficient, it also provides
degree of market concentration is services. The frictions associated guidance for examining the
typically measured by the with market operation can sub- reasons for market failures. In
Herfindahl index.3 stantially affect market equilibria.7 these respects, this proposition is
(2) Free entry and exit. Cheap (6) The absence of externalities. enormously useful. But it is also
and easy entry constrains the Good market operation requires important to note several
pricing discretion of existing that all agents behave on the basis important qualifications about

November 2007, Vol. 20, Issue 9 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 25
this proposition and the case in which the ultimate particular effects, rather than
assumptions that underlie it. implications of an assumption assuming that nearly satisfying an
First, while this theorem might in principle be satisfied in assumption means nearly
explains how maximum some alternative manner. achieving perfectly working
efficiency can be achieved, Third, in contrast to the above, markets.
competitive markets do not
necessarily achieve other possible
societal objectives, such as equity
the implications of some
assumptions may be extremely
sensitive to minor variations in
F inally, some observers have
focused on assumptions
which can be slightly violated
and technological progress.9 It whether those assumptions hold. without reversing the competitive
should be noted that competitive For example, Stiglitz shows that implications of the model, and
markets are not necessarily even slight information offered lists of somewhat weaker,
inconsistent with these objectives, imperfections can cause sudden but supposedly more realistic,
but there is nothing in the market and complete reversal of the usual conditions. The resulting list of
mechanism that ensures these conditions defines what was
other objectives are satisfied. The originally termed workable
proposition described above competition by Clark.12 While
focuses exclusively on economic
Competitive markets different authors suggest slightly
efficiency, as measured by total do not necessarily different lists, one authoritative
surplus. achieve other possible source has suggested the
Second, some of these following:13
societal objectives,
conditions are necessary as stated, Structural criteria:
whereas the essential purpose of such as equity and  Number of traders is as large
others can be realized in technological as scale economies permit.
alternative ways. The most progress.  No artificial impediments to
important example of the latter mobility or entry.
possibility involves the first two  Moderate, price-sensitive
assumptions above, those quality differentials among the
concerning large numbers of conclusions with respect to products.
participants and ease of entry. competitive markets, instead of a Conduct criteria:
These assumptions jointly ensure smooth and continuous decline  Some uncertainty by rivals as
that no single seller has pricing from maximum benefits as to whether price initiatives will be
discretion, but that result can also information is increasingly followed.
be achieved even when seller imperfect.11 He notes, for example,  Firms strive to attain goals
numbers are small so long as that in the presence of imperfect independently.
entry is instantaneous, costless, information, markets can be  No unfair, exclusionary, pre-
and fully and costlessly characterized by demand datory, or coercive tactics.
reversible. In that case termed a exceeding supply, or by supply  Inefficient suppliers should
contestable market any possible exceeding demand. Similarly, the not be shielded permanently.
effort by an incumbent firm to implication that price will be  Sales promotion should be
raise price is immediately constant per unit of output or that informative, not misleading.
undercut by entry before the profit goes to zero with entry can  No persistent harmful price
incumbent can benefit.10 This be shown not to hold under discrimination.
proposition is of some theoretical conditions of imperfect Markets satisfying these criteria
interest, but its requirements are information. Such facts make it will likely produce satisfactory,
not met in any real-world market. important to analyze each though less than perfect,
We note this here to illustrate the deviation in assumptions for its performance results.

26 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal
III. Markets and that factor are not necessarily space over which the market must
Electricity decisively adverse except for clear.15 In the case of electricity,
being magnified by other the market interval is essentially
With the prior discussion as characteristics in the case of determined by technology,
background, we now turn to electricity. The result of this specifically, by the need for near-
electricity to evaluate the reasons combination of forces is poor instantaneous electrical balance at
why well-functioning markets market operation, but to repeat, all points on the electrical grid.
may be difficult to achieve for this the causal mechanism is not That is, physical supply from
sector. A number of observers captured by a simple listing of the generators and physical demand
have offered lengthy lists of sort found in standard sources. from consumers must equilibrate
characteristics of electricity
markets that supposedly create
difficulties.14 These lists include
W e shall set out this
economic argument,
beginning with the key factor,
at all points on the electrical grid
and at all points in time, with
essentially zero tolerance.
such factors as the non-storability Electricity delivered to the grid
of electricity, low demand must be taken off the grid, and
elasticity, and high capital electricity demanded must be
intensity. The problem with all
There are a few other satisfied by electricity from the
these lists is that, while the factors markets with some grid, with each of these strictly
on them may well hold for the similar properties say, time- and location-dependent.
case of electricity, the same factors There are a few other markets
telecom services, with
are equally true for other markets with some similar properties. The
that function reasonably well. But an obvious and market for telecom services has
if these factors are neither unique important time similarities, including an obvious
to electricity nor necessarily dimension. and important time dimension.
associated with market failures Yet not all possible transactions
elsewhere, they cannot logically need to be consummated. Excess
represent the true source of any call volume is simply truncated
problems with electricity markets. then followed by the various by blocking calls, essentially

T he analysis to follow focuses


on this question: the precise
reason or reasons for the difficulty
contributing forces. This
represents an economically sound
framework for understanding the
rationing available capacity, and
so the market does not need to
clear in the sense of satisfying all
in making electricity markets challenges of operating a well- demand (or supply) that appears.
work reasonably well. By functioning market for electricity. The stock market is another
reasonably well, we mean possible example, since
something like workably A. The key characteristics transactions there are generally
competitive, with performance consummated in each moment.
that may fall short of perfection We begin by setting out and But both demand and supply can
but nonetheless is recognized as discussing the key feature that and often do shift between
satisfactory, especially by distinguishes the market for moments, allowing for smoother
comparison with alternatives. We electricity from most other market clearing.16 The airline
argue that there is not one single markets. That feature is the length industry illustrates the case where
truly unique factor responsible for and scope of the market interval, supply expires at the moment
problems in electricity markets. that is, the time and space over when the plane departs. Yet as
Much does depend upon one which physical supply and with the stock market, air travel
unusual, but not entirely unique, physical demand manifest demand shifts between moments
factor. Moreover, the effects of themselves, and the time and of time, and as with telecom

November 2007, Vol. 20, Issue 9 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 27
excess demand can be truncated. externalities that are much less impracticality for consumers to
Thus, despite some similarities, amenable to market resolution. write contracts that ensure against
these other markets differ from Blackouts, for example, can be such outcomes.
electricity in significant ways.
If the consequences of the
failure to maintain balance in time
viewed as externalities over both
space and time. At any point in
time they may cover substantial
T he implication of these
considerations is that there
really is not a meaningful market
and space in electricity markets geographic regions having for electricity. Rather than a
were small or limited in scope, nothing to do with the triggering single market that clears in
they might be tolerated and event. This externality is a isolation, there is a complex web
internalized. In contrast to the consequence of the electrical grid, of interacting momentary
telephone, stock, and airline which creates the opportunity for markets. These markets must
markets, however, in electricity negative (as well as positive) clear in a physical sense
disruption of electrical frequency effects over a wide territory. In instantaneously and at all nodes.
is likely to have several significant Failure to do so can affect
adverse effects. These include the operations more widely in space
following:17 and across time, creating external
 Equipment may be damaged.
The implication costs to third parties. This
 Product quality delivered to of these characteristic forms the basis for
consumers can be degraded considerations is the further analysis to follow.
(flickering, etc.).
that there really
 Transmission lines may B. Complicating factors
become overloaded. is not a meaningful
 The power system could market for The above discussion has noted
collapse. electricity. that most other markets allow for
Some of these effects are demand truncation or
modest in magnitude and largely intertemporal shifting of supply
internal to consumerse.g., home or demand to moderate the effects
computer damage. Such effects addition, system failures and of otherwise short market
do not pose threats to market blackouts at any one point in time intervals with externalities. Here
operation, since the consequences affect operations in the next we argue that in the case of
can be resolved by contractual period, since blackouts are not electricity, no such helpful factors
penalties and damage claims by necessarily easily reversed. Hours exist. Instead of helpful factors,
adversely affected parties. Other or even days may be required to electricity exhibits two important
effects may be much more bring the system back up, factors that further complicate
substantial in physical magnitude illustrating the interdependency market operation. The first factor
and economic cost, as with of markets over time. is inelastic short-run supply at
shutdown of commercial office System overload and blackouts capacity, the second is inelastic
buildings or of major production are sufficiently serious and costly demand.
processes dependent upon that they are viewed as requiring That short-run supply at
power. Even these potential avoidance at a very high level of capacity is inelastic is not in
outcomes, however, do not probability and at dispute. Figure 1 illustrates this
necessarily represent correspondingly high cost. The phenomenon from actual
insurmountable problems to desired degree of reliability is a operating data for all thermal
satisfactory market operation. public good due to the generating plants in Califormia in
Certain other effects, however, externalities over time and space three months in 1998 and 2000.
may involve important created by blackouts, and the Marginal costs follow fuel costs

28 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal
periods of low demand in order to
make it available in periods of
high supply. Hydroelectric
pumped storage is an exception to
this statement, but its impact is
generally minor.
Collectively, these three
underlying forces cause the costs
of incremental supply to rise very
sharply at capacity, as shown in
Figure 1. But it should not be
Figure 1: Rising Cost of Incremental Supply as Capacity is Approached. concluded that low supply
elasticity by itself creates any
fairly closely until capacity is producers except in the very long adverse effects on, say, price.
approached. Near capacity, and run. As summarized in Table 1, Consider, for example, Figure 2,
even more so at capacity, supply an efficient coal-fired plant takes with inelastic supply as
costs rise sharply and elasticity about four years and $775 million previously shown and demand
falls correspondingly. Similar to construct. Combined-cycle gas that shifts between high and low
supply curves and elasticities are plants are considerably less periods.19 It is evident that
commonplace for electricity expensive but still require as instantaneous market clearing is
markets. much as three years to bring now still possible with only

F rom the underlying


economics of electricity, we
can identify three sources of very
online. Both involve additional
time for licensing and approval.
Based on recent experience in
moderate price variation, as P1
(which is at or near marginal cost)
is similar to P2. Thus under
low elasticity of supply at capacity: other countries, nuclear plants certain conditions, supply
 The production technology, can be expected to take six years inelasticity at capacity need not
which makes generation capacity to construct and cost nearly $3 jeopardize relatively smooth
a nearly fixed parameter. Genera- billion in construction costs. market operation.
tion plants have rated capacities
that cannot be increased by the
application of other inputs. This
Other estimates suggest another
five to six years are required for
licensing approval.18 None of
T he cause of a more serious
problem is a combination of
very inelastic demand and very
creates a hard capacity con- these alternatives represent rela- inelastic supply. This case is
straint, that is, output past some tively quick and cheap sources of illustrated in Figure 3 where D3
amount simply cannot be obtained new supply. and D4 denote inelastic demands
from existing plant.  The virtual inability to store in the low- and high-demand
 Substantial sunk costs in pro- output, which precludes shifting periods, respectively. Whereas
duction, which prevent getting output between periods. This inelasticity makes only a modest
additional output from new prevents inventorying output in difference during the low-
demand period (P3 here is not too
Table 1: Time and Cost Factors for Minimum Efficient Scale of Generation Plants different from P1 in Figure 2), in
combination with inelastic peak
Technology Lead time (years) Scale (new) Capital Costs ($M) supply, a shift of demands result
Scrubbed coal 4 600 775 in proportionally far larger
Advanced gas/oil combined cycle 3 400 238 changes in equilibrium market
Advanced nuclear 6 1350 2800 price when demand is already
Source: EIA Annual Energy Outlook, Electricity Market Module, Table 39, April 2007. strong. This is illustrated by the

November 2007, Vol. 20, Issue 9 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 29
that a 10 percentage point
increase in price results in a 9
percent decrease in quantity
demandedconsiderably more
than in the short run. Long-run
supply elasticity is limited only by
the timeframe one allows, since in
Figure 2: Effect of Time Shifting on Market Clearing Prices. a sufficiently long time period
much new plant can be
constructed at constant long-run
marginal cost. The relevant time
frame for analysis, however, is not
the long run, but rather the very
short interval in which the market
must clear. During that time, both
demand and supply (at capacity)
are quite inelastic. It is the
Figure 3: Effect of Very Inelastic Demand and Very Inelastic Supply. combination of these forces that
jeopardizes a well-functioning
market.
contrast between P4 in Figure 3 most consumer goods.20 Inelastic The considerations are
and P2 in Figure 2. demand for electricity is rooted in illustrated in Figure 4. That

A s an empirical matter,
demand in electricity
markets is quite inelastic,
business and residential
commitment (i.e., sunk costs) to
its use, in long-lived capital
schematic illustrates the fact that
the most fundamental and
relevant characteristic of
reflecting the largely non- equipment with fixed energy electricity markets is their very
deferrable nature of consumption efficiency, and in consumer short market interval. It also
in the short term. A considerable behavior toward electric power indicates the role of elasticities,
amount of econometric evidence, usage that resists quick change. storability, barriers to entry, etc.,
summarized in Table 2, suggests
that short-run demand elasticity
is about 0.35. This value implies
B oth supply elasticity and
demand elasticity are
greater in the long run, as
in making good operation
difficult. Causation runs from
factors such as non-storability to
that every 10 percentage point consumers and producers low elasticities, and then from low
increase in price reduces respond more completely to price elasticities to market operation
consumption by only about 3.5 changes. Consumers can switch to that is problematic.
percent, considerably less than more energy-efficient appliances
the demand responsiveness of or more fully adjust their own C. Making matters worse
behavior, while producers can
Table 2: Demand Elasticity alter existing capital equipment or It should be stressed that, even
bring more capacity online over under the above circumstances,
Short Run Long Run
time. Table 2 also reports the markets may work in the literal
Low 0.20 0.60 results of a survey of long-run sense that supply and demand do
Medium 0.35 0.90 electricity demand elasticity intersect, there is a price that
High 0.60 1.20 studies. That review concludes results, and the market clears. The
Source: Chris S. King and Sanjoy Chatterjee, Predict-
ing California Demand Response, Public Utilities Fort-
that long-run demand elasticity is outcome of that process, however,
nightly, July 1, 2003. on the order of 0.90. This implies may have undesirable properties.

30 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal
periods. As shown in Figure 3,
demand given by D3 results in
price P3 while D4 causes a much
different P4. The degree of price
change depends on the magni-
tude of demand variation as well
as on the supply and demand
elasticities. Demand for electricity
varies substantially over hourly,
daily, and seasonal cycles, as
Figure 4: Effect of Short Market Interval on Electricity Market. illustrated in Figures 5 and 6.
Figure 5 shows hourly loads, by
day, for four different weeks in
2006 in PJM. Retail load over most
days varies by about 60 percent in
the seasonal off-peak of April, and
by about 70 percent in the July
peak period. Moreover, these
weeks differ in level, as shown in
the daily load chart for PJM in all
of 2006, shown in Figure 6. Few
other markets exhibit such enor-
mous, and inflexible, time-
dependent variations.

T his demand variation in


combination with fixed
capacity causes substantial
Figure 5: Hourly Loads for Four Weeks in 2006. price volatility. This is illustrated
for PJM wholesale market prices
This section examines various (1) First, low elasticities on an hourly basis in Figure 7
features of the outcome, become more troublesome to the and on a daily basis in Figure 8.
explaining how some may arise degree that demand (or supply) Substantial price volatility is an
that are judged to be undesirable. shifts significantly between undesirable outcome for
residential, commercial, and
industrial consumers. There are
two reasons why volatility is
undesirable. One is adjustment
costs to consumers who are
dependent upon electricity.
Large and sudden changes
require adaptation of
consumption and use
patterns in ways that are
themselves costly.21 While
adjustment costs are rarely
Figure 6: Daily Load Chart for PJM. addressed in economics, their

November 2007, Vol. 20, Issue 9 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 31
some circumstances, this transfer
serves a straightforward
economic purpose. In a market
that is competitive, such a transfer
sends a signal that greater
capacity is required in this
market. In at least two other cases,
however, it might not be
economically justified or socially
acceptable. For one, in markets
such as electricity, the capacity
expansion just described may not
be forthcoming because the
price/revenue signal is viewed as
Figure 7: PJM Prices on an Hourly Basis. too transient and too arbitrary to
prompt new entrants (or their
financiers) to bet on the expensive
and lengthy process of capital
investment. If so, the revenues
constitute a simple transfer from
consumers to producers
windfall profits and the
outcome might be judged socially
unacceptable. As a second point,
the market may not be
competitive, in which case the
price increase itself might be
economically unjustified. In that
Figure 8: PJM Prices on a Daily Basis. case the resulting windfall profits
cannot and will not prompt new
entry and therefore have no
practical implications may be producers. Figure 9 illustrates the justification as part of the
significant. price change that might result ordinary process of market

I n addition, such volatility


creates substantial and
seemingly arbitrary transfers of
from a shift from D1 to D2. One
result of this price change would
be an enormous revenue transfer
adjustment.
(2) Second, an important com-
plicating factor to price volatility
revenues between consumers and from consumers to sellers. Under is the high fixed costs of the
electricity production process. A
given level of total costs with high
fixed costs implies lower mar-
ginal costs and hence lower price
up to capacity. In high-demand
periods, however, the demand
curve intersects the marginal cost
line in the sharply increasing
Figure 9: Price Changes Resulting from D1 to D2 Shift. segment near capacity, resulting

32 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal
(4) A rather different concern
is that the electricity markets
just described are unusually vul-
nerable to the exercise of unilat-
eral market power. Unilateral
market power arises when a
single company controls suffi-
Figure 10: Scenario with Modest Fixed Costs. cient capacity in a market so that
shutting down some part of that
capacity causes a profitable price
in a much higher price than in the (3) Yet another complicating rise. This phenomenon is illu-
low-demand period. This case of factor in the case of electricity strated in Figure 11. Suppose the
volatile prices is illustrated by is the unpredictability of market is supplied by 10 units of
Figure 9. By contrast, if fixed costs demand fluctuations. Apart from capacity, two of which are owned
were modest, then marginal costs their magnitude, the unpredict- by the same company. With
would be higher and more uni- able nature of those fluctuations inelastic demand as shown, mar-
form (indeed, if all costs were violates the competitive ket price is a modest P1, but the
variable, then the marginal cost assumption of perfect informa- company can take one unit offline,
curve would be flat). Now even tion and limits consumers abil- causing price to rise to P2. While
with low demand elasticity and ity to avoid or contend with the there is lost profit on the deacti-
the same demand variation, the resulting price volatility. Of vated unit, the added profit on the
high-demand price would not course, certain determinants of remaining unit may more than
diverge as much from that in the demand are well enough compensate. Neither high
low-demand period. This latter understood to lend some pre- market concentration nor a
case is illustrated in Figure 10 dictability, but many others are large market share is required
where the two prices resulting not. Residential demand for for the exercise of unilateral
from demand D1 and D2 do not electricity, for example, varies market power. Under conditions
differ anywhere near as much as with climate conditions that are of sufficiently low demand and
in Figure 8 where fixed costs are known with only modest, if any, supply elasticity, such as electric
higher. lead times. Seasonal demand is power, unilateral market power

E lectricity production is, of


course, characterized by
very high fixed costs. Table 3
more predictable than hourly
load for any particular day.
The inability to predict these
may occur in a wide range of
market structures, not requiring
conventional market power.22
reproduces a tabulation of capital changes limits consumers (5) Yet another complicating
intensity for various prominent ability to protect against their feature is that the geographic
industries, measured in terms of adverse effects. extent of electricity markets can
assets per dollar of revenue.
This list is headed by investor-
owned electric utilities. Only
mining and railroads have capital
intensities anywhere close to
that for IOUs, with all other
industries trailing well behind.
This evidence confirms common
belief about electricity
production. Figure 11: Price Effect of Unilateral Withholding.

November 2007, Vol. 20, Issue 9 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 33
Table 3: Capital Intensity peak, and off-peak periods, which
Industry Assets per Revenue Dollar themselves differ by season.
Investor-owned electric utilities $2.26 Other analyses would further
Mining $2.07 subdivide geographic markets
Railroads $1.71 based on likely transmission
Investor-owned gas utilities and pipeline industry $1.20 constraints that effectively
Communications $0.99 isolate territories. The essential
Nondurable manufacturing industries, average $0.98 point is that electricity markets
Durable manufacturing industries, average $0.87 are constantly and unpredictably
Retail trade $0.50 being redefined, making it
Wholesale trade $0.43 difficult to sustain a normal
Source: Electric Perspectives, Edison Electric Institute, 1996. market process. It might also be
noted that transmission
congestion has proved vulnerable
to market power and other
manipulation by opportunistic
traders, so that this feature of
electricity markets can be
exploited.23

T hese considerations are


summarized in the
schematic in Figure 12. Together
with the preceding discussion
and description, it should be
apparent that the key
Figure 12: Effect of TKTK on Electricity Market. characteristic of electricity
markets the short length and
narrow scope of the market
vary continuously and unpredic- its termination may be unpre- interval becomes a more
tably as the result of transmission dictable, since congestion can substantial impediment to good
congestion. Overloaded trans- arise with little prior market operation as a result of an
mission lines result in the abrupt warning and lead to immediate unusual combination of other
termination of electricity imports loss of supply into the congestion forces.
into a load area or exports from a area.
generation site. While many
other markets of course face
transport difficulties with
T hese factors result in a
multiplicity of geographic
markets at any one point in time,
IV. Counter-Measures
and Further
respect to out-of-region supply, as well as a changing set of such Complications
typically these cause rising markets over time. These are
marginal costs of supply. In illustrated by filings in the Duke- All of the above factors make
electricity momentary Cinergy proposed merger, for electricity markets unusually
transmission constraints example, which defined three difficult to implement and fragile
result instead in truncation of all broad geographic markets based to operate. It should be
outside supply. Moreover, the on established transmission emphasized, however, that
amount of electricity thus con- constraints. Each of these is then measures can be taken to resolve
strained may be substantial, and subdivided into peak, shoulder in whole or part some of the

34 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal
difficulties described above. (3) The complete inelasticity of assigning property rights in this
At least the following three supply at capacity can be mod- context complicates the task of
should be noted: erated by holding capacity in orderly markets for long-distance
(1) The need for physical reserve for unexpectedly high transmission.
clearing of the momentary demand. This strategy is an alter- (2) Vertical economies are
market does not necessarily imply native to reliance upon outside sacrificed as a result of the
that economic clearing must take supply to meet peak demand. separation of the generation,
place in the same time frame. Instead of some outside capacity transmission, and distribution
Agents generally cannot take being quickly converted to pro- functions under restructuring.24
action or respond in as short a duction not easily done in elec- These economies derive from the
period of time as the market tricity here dedicated capacity is need for tight coordination
interval for electricity, and so they among stages, a task that the
enter into contractual relation- market or institutions such as
ships that cover a series of such power pools and independent
momentary markets. Contracts service operators find difficult to
help consumers contend duplicate.
with imperfect information
and price volatility by
providing some certainty V. Conclusions
and moderating any adverse
effects, though there are In many ways electricity
explicit and implicit costs of markets resemble markets
contracting. Clearly, contracts generally, but they also have some
(and regulation itself) have distinctive features that make
served to ensure electricity pure reliance on market processes
consumers more moderate simply held in readiness and then problematic. Foremost among
price fluctuations than underly- operated only part of the time. these features is the limited time
ing conditions would have Typically, about 15 percent of and space over which the physical
produced. available capacity is held in supply and physical demand
(2) The inelasticity of reserve, but this is maintained at manifest themselves, and over
demand can be moderated by some significant cost to the system. which the market must clear.
such measures as interruptible On the other hand, there are These characteristics make
supply or priority use contracts at least two other factors that electricity markets uniquely
and by time of use metering. further complicate the vulnerable to poor operation and
Interruptible contracts are introduction of markets into even failure, especially when
typically negotiated with electricity. These are loop flow other factors come into play. Low
large industrial users, who externalities and vertical supply and demand elasticities,
agree to cessation of some economies. volatile and unpredictable
supply under conditions of (1) Loop flow externalities demand, high fixed costs and
extreme network imbalance. arise as a consequence of the fact sharp capacity limits in
Time-of-use metering that electron flow over the trans- production, and constraints on
confronts consumers with the mission grid is governed by the transmission capacity all
real-time cost of power, thereby laws of electrical engineering, compound the core problem.
evoking some response. Both
therefore impart some elasticity to
demand.
often resulting in flows that across
the lines of non-parties to a
transaction. The difficulty of
T his array of forces cautions
that prescriptions for
deregulation and restructuring

November 2007, Vol. 20, Issue 9 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 35
that might succeed in other 8. J. CHURCH AND R. WARE, INDUSTRIAL INDUSTRIAL ORGANIZATION, 3rd Ed.,
ORGANIZATION: A STRATEGIC APPROACH, Ch. 17.
markets cannot be relied upon to 2000, Ch. 2.
have the same effect in electricity. 16. Where supply and demand do not
9. Scherer and Ross, supra note 5, Ch. shift sufficiently smoothly over time,
At a minimum, the particulars of 2, 18. as in panics, the stock market more
proposed reforms in electricity closely resembles electricity. Yet even
10. Strictly, the incumbent suffers
need to be carefully examined losses as a result of the further in that case, the stock market has
before the market mechanism can assumption that its response time is instituted blocks on transactions that
slower than that of entrants. For threaten disruption. These so-called
be trusted to produce benefits. At circuit breakers work since there is
further discussion, see J. Church and
worst, the result of unthinking R. Ware, op. cit, pp. 50713. no technological reason why the
reform can be a perfect storm of transaction must be consummated in a
11. J. Stiglitz, supra note 6, at 1067. single moment.
forces leading to serious problems Related to this point, Stiglitz criticizes
in market operation.& 17. This discussion borrows heavily
from B. Kirby, J. Dyer, C. Martinez,
R. Shoureshi, R. Guttromson and J.
Dagle, Frequency Control Concerns in the
Endnotes: North American Electric Power System,
Oak Ridge National Laboratory, Dec.
1. A useful account of the 2002.
deregulation movement in general
and with respect to various industries 18. M.S. Blanton, Challenges to the
can be found in C. Winston, Economic Licensing, Construction, and
Deregulation: Days of Reckoning for Operation of a New Nuclear Fleet, M.
Microeconomists, J. ECON. LITERATURE, Sanford Blanton, Balch & Bingham,
Sept. 1993. downloaded presentation dated April
18, 2007.
2. J. Kwoka, Restructuring the U.S.
19. These might designate night vs.
Electric Power Sector: A Review of Recent
day, weekend vs. weekday, or
Studies, American Public Power Assn.,
summer vs. winter (or vice versa, in
Nov. 2006.
some climates.)
3. For discussion of these and other the typical simple listing of
20. C. King and S. Chatterjee,
concentration measures, see J. Kwoka, assumptions found in many sources,
Predicting California Demand Response,
The Herfindahl Index in Theory and noting that too often they contain
PUB. UTIL. FORTNIGHTLY, 2003. See also
Practice, ANTITRUST BULLETIN, 1985. important hidden assumptions, for
D. BOHI, ANALYZING DEMAND BEHAVIOR
example, embedded in definitions
4. For discussion, see JOE BAIN, (Baltimore: Johns Hopkins Univ.
rather than made explicit.
BARRIERS TO NEW COMPETITION Press, 1982), Ch. 2.
(Cambridge, MA: Harvard Univ. 12. J. M. Clark, Toward a Theory of 21. Residential consumer demand is
Press, 1956); W.G. SHEPHERD, Workable Competition, AMER. ECON. REV., usually insulated from the price
THE ECONOMICS OF INDUSTRIAL 1940. volatility that characterizes wholesale.
ORGANIZATION (Englewood Cliffs, NJ:
Prentice-Hall, 1997), Ch. 9; and 13. This list is taken from Scherer and 22. Some conditions for the exercise
Douglas Needham, Potential Entry into Ross, supra note 5, at 525. of unilateral market power are
Oligopoly, in READINGS IN developed in J. Kwoka, Unilateral
MICROECONOMICS, W. Breit and H. 14. See for example S. Borenstein, The
Withholding: Market Power and
Hochman, Eds., 2d edition (NY: Holt, Trouble with Electricity Markets:
Californias Electricity Crisis, George
Rinehart: 1971). Understanding Californias Restructuring
Washington Univ. Economics Dept.
Disaster, J. ECON. PERSPECTIVES, Winter
5. F.M. SCHERER AND D. ROSS, INDUSTRIAL Working Paper, May 2001.
2002. Also P. Joskow, The Difficult
MARKET STRUCTURE AND ECONOMIC Transition to Competitive Electricity 23. J. Weaver, Can Energy Markets Be
PERFORMANCE 3rd ed. (Boston: Markets, in ELECTRICITY DEREGULATION: Trusted? The Effect of the Rise and Fall of
Houghton Mifflin, 1990). CHOICES AND CHALLENGES, Eds. J. Griffin Enron on Energy Markets, HOUSTON BUS.
6. J. STIGLITZ, WHITHER SOCIALISM and S. Puller (Chicago: Univ. of & TAX LAW J., 2004.
(Cambridge, MA: MIT Press, 1994), Chicago Press, 2005).
24. J. Kwoka, Vertical Economies in
Ch. 3.
15. For discussion of market clearing Electric Power: Evidence on Integration
7. P. Joskow, Transmission Policy in the issues, including timing, see D. and Its Alternatives, INTL. J. INDL. ORG.,
United States, UTIL. POLICY, 2005. CARLTON AND PERLOFF, MODERN 2002.

36 1040-6190/$see front matter # 2007 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2007.10.008 The Electricity Journal

You might also like