Joskow Tirole 2005

THE JOURNAL OF INDUSTRIAL ECONOMICS 0022-1821
Volume LIII June 2005 No. 2
MERCHANT TRANSMISSION INVESTMENT
Paul Joskoww
JeanTirolez
We examine the performance attributes of a merchant transmission

investment framework that relies on ‘market driven’ investment to
increase transmission network capacity needed to support competitive
wholesale markets for electricity. Under a stringent set of assumptions,
the merchant investment model has a remarkable set of attributes that
appears to solve the natural monopoly problem and the associated need
for regulating electric transmission companies. We expand the merchant
model to incorporate several attributes of wholesale power markets
and transmission networks that the merchant model ignores. These
include market power in wholesale electricity markets, lumpiness
in transmission investment opportunities, stochastic attributes of
transmission networks and associated property rights definition
issues, strategic behavior by potential merchant transmission investors
and issues related to the coordination of transmission system operat-
ors and merchant transmission owners. Incorporating these more
realistic attributes of transmission networks and the behavior of
transmission owners and system operators leads to the conclusion
that several potentially significant inefficiencies may result from reliance
on the merchant transmission investment framework. Accordingly,
it is inappropriate for policymakers to assume that they can avoid
dealing with the many challenges associated with stimulating efficient
levels of investment in electric transmission networks by adopting the
merchant model.
I. INTRODUCTION
THE ECONOMIC ANALYSIS OF RESTRUCTURED COMPETITIVE WHOLESALE ELECTRI-

CITY MARKETS has mostly focused on the organization and functioning of
spot markets for electric energy. This theoretical and empirical research has
studied, among other things, the organization of day-ahead and real time
(balancing) energy markets and associated auction rules, the role of bilateral

We would like to thank Richard Green, Shmuel Oren, Masheed Rosenqvist, Steven Stoft,
Frank Wolak, the editor and three anonymous referees for helpful comments on earlier drafts.
Joskow gratefully acknowledges research support from the MIT Center for Energy and
Environmental Policy Research and the Cambridge-MIT Institute.
wAuthors’ affiliations: MIT Center for Energy and Environmental Policy Research,
Massachusetts Institute of Technology, Cambridge, Massachusetts, 02142-1347, U.S.A.
e-mail: pjoskow@mit.edu
zInstitut d’ Economie Industrielle, Université des Sciences Sociales de Toulouse, Place
Anatole France, 31042 Toulouse, France.
e-mail: tirole@cict.fr
r Blackwell Publishing Ltd. 2005, 9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden, MA 02148, USA.
233
234 PAUL JOSKOW AND JEAN TIROLE
contracts, congestion management, nodal pricing, physical and financial

transmission contracts, and associated market power issues. This work
typically takes the transmission network as given. However, in the medium
and long term as demand grows and new generating capacity is added to
replace older, less efficient, capacity or to meet growing demand efficiently,
investments in transmission capacity are likely to be necessary to minimize
the overall costs of wholesale electricity supplies, to maintain reliability, to
mitigate locational market power, and to improve the performance of
competitive wholesale and retail markets.
Despite growing problems associated with stimulating transmission
investment in many restructured electricity markets, there has been
surprisingly little research on the institutions governing transmission
investment. Early formulations of the structure for competitive wholesale
markets envisioned the creation of independent regulated regional
transmission and system operating entities (Transcos, or regulated
Transmission Companies) that would be responsible for building, owning
and operating transmission facilities and would be subject to economic
regulation (Joskow and Schmalensee [1983]). More recent research has
explored the attributes of incentive regulatory mechanisms that could be
applied to such regulated transmission monopolies (e.g., Celebi, Nasser
[1997], Léautier [2000], Vogelsang [2001]) to integrate energy price
(congestion) signals with transmission investment. The institutional
arrangements governing transmission operation and investment in England
and Wales reflect this basic institutional approach. The regulated Transco
model is necessarily subject to the classical challenges of regulated
monopoly, namely how to specify and apply regulatory mechanisms that
provide good performance incentives to the regulated firm while minimizing
the economic rents that the regulated firm can derive from its superior
information.
An alternative (or complement) to the regulated Transco model is the
merchant investment model. It relies on competition, free entry and
decentralized property-rights based institutions, and market-based pricing
of transmission service to govern transmission investment (Hogan [1992],
Bushnell and Stoft [1996,1997], Chao and Peck [1996]). In return for
investment in additional transmission capacity, merchant investors receive
property rights that allow them to collect congestion revenues equal to the
difference in nodal energy prices associated with the incremental point-to-
point transmission capacity their investments create. The value of these
rights to receive congestion revenues represents the revenues merchant
investors receive to cover the capital and operating costs of these
investments and provides the financial incentives that guide ‘market based’
transmission investment. Previous theoretical research has demonstrated
that under a fairly stringent set of assumptions, all profitable investments are
efficient (efficient investments, on the other hand, are not all profitable, due
r Blackwell Publishing Ltd. 2005.
MERCHANT TRANSMISSION INVESTMENT 235
to the lumpiness problem discussed in Section 5). This result undermines the
traditional assumption that transmission networks are ‘natural monopolies’
that must be subject to regulation. If it were correct, in practice, it would lead
to the remarkable conclusion that both the generation of electricity and the
transmission of electricity can be largely deregulated.
Research on this model has focused almost entirely on simple cases where
transmission investments are characterized by no increasing returns to scale,
there are no sunk cost or asset specificity issues, nodal energy prices fully
reflect consumers’ willingness to pay for energy and reliability, all network
externalities are internalised in nodal prices, transmission network
constraints and associated point-to-point capacity is non-stochastic, there
is no market power, markets are always cleared by prices, there is a full set of
futures markets, and the transmission system operator (SO) has no
discretion to affect the effective transmission capacity and nodal prices
over time. That is, the analysis has proceeded under assumptions equivalent
to those of a simple model of perfect competition. Unfortunately, these
assumptions do not reflect the attributes of the transmission investment
opportunities that are likely to be most conducive to merchant investment,
the stochastic properties of real transmission networks, or widely
documented imperfections in wholesale power markets.
We begin with a background discussion of the institutional environment
in which the merchant investment model has been developed, the key
assumptions upon which analyses of its properties have been based and the
primary results of this work derived (Sections 2 and 3). The paper then
examines how these results are affected when we relax several assumptions
to more accurately reflect the physical and economic attributes of electric
transmission networks.1
Section 4 shows that when there are imperfections in the competitive
wholesale electricity markets that lead nodal spot electricity prices to depart
from their efficient levels, investment incentives will be distorted. For
example, when unregulated generators have market power, nodal energy
prices will be distorted from their efficient levels. These distortions may lead
to over-investment or under-investment depending upon where on the
network electricity generators have market power. Imperfect government
interventions to control market power in competitive wholesale electricity
markets may also distort investment incentives.
Network expansion investments that are most conducive to supply by
competitive entrants are also likely to be characterized by economies of scale
or ‘lumpiness.’ Section 5 shows how economies of scale will lead to under-
1
Similar issues are likely to arise with merchant investment in other network industries such
as natural gas pipelines, railroad infrastructure and highways. However, we have not explored
these issues in the context of other network industries.
investment, to monopoly pre-emption of competitive generation or

transmission investments, and distort the timing of investments. We argue
that these problems are unlikely to be resolved by relying on bilateral or
multilateral negotiations among the market participants who benefit from
these negotiations (the ‘Coase Theorem’). Indeed, opportunities for
multilateral bargaining are likely further to distort investment incentives.
We also demonstrate how strategic interactions between merchant
transmission owners who are in a vertical relationship to one another can
lead to underinvestment in related transmission links.
Section 6 shows that the types of non-contingent transmission property
rights that have been assumed to be awarded to investors in transmiss-
ion capacity are poorly matched to the stochastic characteristics of
transmission networks and investments in them. Instead, transmission
property rights that are contingent on exogenous variations in transmission
capacity and reflect the diversification attributes of new investments would
be necessary to properly align investment incentives with the stochastic
attributes of transmission networks. Defining and allocating property rights
that provide efficient investment incentives is also likely to be inconsistent
with the development of liquid competitive markets for these rights or
derivatives on them.
Under the merchant transmission model, the ownership and maintenance
of transmission facilities are to be separated from decisions regarding
security-constrained, bid-based dispatching of generators and price-
responsive demand on the network and managing reliability criteria and
constraints. The latter decisions are to be made by a monopoly independent
system operator (ISO) that is unaffiliated with generators, energy marketers
or transmission owners. Section 7 shows that investment and maintenance
decisions made by transmission owners are interdependent with dispatch
and network reliability decisions made by the ISO. Separating these
decisions leads potentially to moral hazard problems and associated
inefficiencies. Transmission owners and system operators must have a
compatible set of incentives to avoid these inefficiencies, but designing these
incentives is challenging.2
II. MERCHANT TRANSMISSION INVESTMENT: BACKGROUND
Transmission investment institutions cannot be considered independently

of the institutions that govern the determination of energy prices, operating
reserves, contingency constraints, congestion management and the mea-
surement of transmission capacity and increments to it. In what follows, we
2
Vertical integration between transmission ownership and system operations is likely to
reduce these incentive problems and, if transmission owners are also independent of generation
and marketing of power, may be a superior organizational structure.
shall assume that a nodal pricing system for electric energy is in place with
attributes similar to those being proposed by the U.S. Federal Energy
Regulatory Commission (FERC) in its SMD proposals and what is in
operation in New York, New England and PJM in the U.S. This is the most
conducive framework for merchant investment because nodal prices provide
a measure of locational scarcity that is necessary to make this framework a
plausible option.
Under this model, an independent system operator (ISO) operates a real-
time balancing market and allocates scarce transmission capacity using bids
to increase or decrease generation or demand at each node. That is, the ISO
takes all of the bids (generation and demand) and finds the ‘least cost’ set of
uniform market-clearing price bids to balance supply and demand at each
generation and consumption node on the network using a security
constrained dispatch model. This establishes day-ahead quantity commit-
ments and nodal prices that reflect both congestion and marginal losses.3
The model recognizes that there are incumbent transmission owners (TO)
who own the existing transmission assets and requires that the system
operator (SO) and TO be separate entities and operate independently.4 The
incumbent TO receives some cost-of-service compensation for the usage of a
grid that it no longer controls to compensate it for legacy investments and
ongoing maintenance costs. New investments in transmission are antici-
pated to be made by competing merchant investors whose compensation is
based on the value of Congestion Revenue Rights (CRRs)5 created by their
investments. These financial rights represent the right to receive congestion
revenues defined as the difference between the nodal prices between the two
nodes (point-to-point) covered by the relevant CRR times the quantity of
CRRs held.
The separation between ownership (TO) and control (SO) functions in
this model is motivated by two considerations. First, a market driven
transmission system leads to multiple owners of parts of the grid; while the
owners can form a cooperative to operate the grid, their goals are in general
3
Generators may enter into bilateral contracts with marketers or load serving entities (LSEs)
and schedule supplies with the ISO separately from the organized day-ahead market. However,
they still have to pay any congestion charges associated with their schedules based on the
difference in nodal prices between the injection and receipt points. The day-ahead schedules,
nodal prices, and congestion charges are ‘commitments.’ They can be adjusted in real time by
submitting adjustment bids to the real time balancing markets (which again rely on bids, a
security constrained dispatch and nodal prices) to allow these schedules to be changed based on
real time economic conditions.
4
Exactly what functions are assigned to the TO and what functions to the SO is a subject of
continuing debate and depends in part on whether the TO is ‘independent’ of generators and
marketers that use its facilities to participate in the wholesale market.
5
Congestion Revenue Rights have also been referred to in the literature as Transmission
Congestion Contracts (TCCs) and Financial Transmission Rights (FTRs). These names are
interchangeable for our purposes.
antagonistic,6 and it is well-known that cooperatives of members with

heterogeneous interests face complex governance problems.7 Second, and
quite crucially, grid owners face a serious potential conflict of interest when
operating a transmission grid if their compensation varies directly with the
level of congestion rents.8 By conservatively ‘withdrawing’ transmission
capacity (under the cover of a safe management of the network), the system
operator can substantially raise the congestion rents. Third, in many
countries there continues to be vertical integration between generation,
power marketing and transmission. The creation of an independent SO is
thought to be a way to mitigate the potential problems that may arise from
common ownership and control of transmission and generating assets and
associated power marketing activities.
It is useful to consider two types of transmission investments that might be
governed by a merchant (or regulated) transmission investment framework,
network deepening and network expansion investments:9
Network deepening investments: These are investments that involve physical

upgrades of the facilities on the incumbent’s existing network (e.g., adding
capacitor banks, phase shifters, reconductoring existing transmission links,
new communications and relay equipment spread around the network to
increase the speed with which the SO can respond to sudden equipment
outages and relax contingency constraints). These investments are physically
intertwined with the incumbent TO’s facilities. Similar to network deepening
investments are network maintenance decisions.
Network deepening and maintenance investments can, as a practical
matter, only be implemented efficiently by the owner of the existing lines for
several reasons. Defining an efficient ‘competitive access to deepening
6
Incumbent owners of transmission lines that are being compensated based on congestion
rents will have incentives to oppose investments by others in generation or transmission that
reduce these congestion rents. Generators located in congested areas will have incentives to
oppose transmission enhancements that would reduce or eliminate the congestion.
7
See, e.g., Hansmann [1996].
8
In practice, due to the lack of market-based penalties for outages, dispatching does not
quite correspond to the least-cost optimization used in economic and engineering models;
rather, grid operators have substantial discretion over how much outage they are willing to take
while dispatching. This discretion in turn potentially provides incentives for system operators
to manipulate the congestion rents received by the owners (Glachant and Pignon, [2002]).
9
We focus on transmission investments that affect congestion on the high voltage network.
Regulators in the U.S. often break transmission investments down into additional categories.
First, there are local transmission investments ‘inside’ the demand node. These investments are
sometimes called ‘reliability’ investments. Second, interconnection investments are invest-
ments that must be made by an incumbent grid owner to connect new generators with the rest of
the network. These are often treated like radial links and are typically paid for by the generators
seeking interconnection. However, it is hard to draw bright lines between reliability
investments, interconnection investments, and investments that affect network congestion.
We recognize that, in practice, there may not always be a bright line between network
deepening and network expansion investments.
investments’ policy is likely to be extremely difficult. First, adding third-party
facilities that are fully integrated with the existing network from a
physical and maintenance perspective creates significant incentive
problems with decentralized ownership. The problems of defining a good
set of rules for investing in and maintaining facilities of this type with
decentralized ownership is further exacerbated by the heterogeneous nature
of transmission facilities. While it is theoretically possible to devise
contractual arrangements that will solve the incentive problems, including
opportunistic behavior of one or more parties, writing and enforcing such
contracts will be very difficult as a practical matter. Second and relatedly,
one would need to allocate carefully the new capacity of the line between the
initial design and maintenance choice of the original owner and the actions
of the renters who make deepening investments. This ‘moral hazard in
teams’ problem is a substantial obstacle to the design of an effective third
party access policy for this type of transmission investment. Third, the
merchant investment model effectively requires that there be free entry into
the development of new transmission capacity; entrants, though, are likely
to have less information about existing transmission lines than their owners.
These considerations suggest that network deepening investments can, as a
practical matter, only be implemented efficiently by the incumbent
transmission network owner.
Independent network expansion investments: These are investments that

involve the construction of separate new links (including parallel links) that
are not physically intertwined with the incumbent network except at the
point at either end where they are interconnected. These investments can (in
principle) be made either by incumbent transmission owners, by stake-
holders (generators, load-serving entities), or by a third-party merchant
investor. The two operating DC merchant links in Australia appear to fall
into this category as do other proposed transmission lines linking different
countries or regions and often referred to as ‘interconnectors.’ While, as in
Australia, these links may have effects on power flows on the rest of the
network, including on parallel lines, they are physically separable projects
from a construction and maintenance perspective.
While the literature underlying the merchant investment model does not
distinguish between network deepening and network expansion investments
(and some proponents of the merchant investment model appear to conceive
it as applying to all network transmission investments) for the reasons
discussed above, we think that it is unlikely that a merchant model can or will
be applied successfully to network deepening investments. Accordingly, our
analysis proceeds under the assumption that the kinds of investments to
which the merchant model will apply are investment opportunities with the
attributes of network expansion investments (e.g., interconnectors) and we
do not attempt to deal with the difficult contractual and incentive issues that
would otherwise have to be confronted if network deepening investments

were governed by a merchant investment model as well.
III. THE CASE FOR MERCHANT TRANSMISSION INVESTMENT
Let us start from the theoretical case for market driven or ‘merchant’
transmission investment. This rationale has been developed, inter alia, by
Hogan [1992] and Wu et al. [1996], Chao and Peck [1996] and examined
further by Bushnell and Stoft [1996, 1997] for simple cases. The basic
argument is conveyed in the two-node framework of figure 1. [This paper
focuses on the two-node framework for expositional simplicity. See our
discussion paper (Joskow-Tirole [2003]) for an analysis of transmission
investment when there are loop flows.]
Figure 1 depicts a simple situation in which load serving entities
(distribution companies or marketers buying on behalf of retail consumers,
or large industrial customers buying directly) in the South (say, a large city)
buy their power from cheap generation sources in the North and, possibly, in
addition, more expensive sources in the South. Alas, the capacity of the line
from North to South is limited to K and, faced with net demand/supply
curves in the North and the South, the system operator is forced to dispatch
‘out of merit’. For example, the system operator calls on expensive
generators in the South while generators in the North would be willing to
supply this amount at a lower price if more transmission capacity were
available. The rationing of the scarce North-South capacity is implemented
by setting two nodal prices, pS and pN that clear the markets in the South and
the North, respectively. The difference, Z 5 pS –pN, is the shadow price of the
transmission capacity constraint. The area ZK is the congestion rent and the
Price Congestion Rents vs Congestion Costs

Congestion Rent
pS A Congestion Cost Net Supply in
{
North
North SN
B
η
Κ
C Net Demand in
pN South DS
South
No Congestion
δΚ
Κ Κ Quantity
Figure 1
triangle ABC is the congestion or redispatch cost. The latter represents the
cost of running more costly generation in the South because less costly
imports from the North are limited by transmission congestion. The
congestion rent and the congestion cost are sometimes confused and it is
important to recognize their appropriate definitions and meaning.
Now consider a marginal (unit) increase in transmission capacity (dK).
This unit increase allows one more kWh to flow from North to South,
replacing a marginal generator in the South with cost pS by a cheaper
generator in the North producing at cost pN. That is, the social value of the
investment is given by the reduction in the area ABC in Figure 1.
Assume that the builder of this marginal capacity, whether it is a new
entrant or the incumbent TO, is rewarded through a financial transmission
right that pays a dividend equal to the shadow price of the transmission
constraint. A non-incumbent merchant company will enter to build this
extra capacity as long as Z exceeds the cost of building it. By contrast, if an
incumbent grid owner is compensated through the payment of congestion
rents, it may not want to make this marginal investment as it must compare
the extra revenue Z net of the cost of expanding the capacity with the
reduction in the congestion rent on its inframarginal transmission units
( KdZ/dK). It is only when the incumbent grid owner’s capacity has been
rated at some level K not too different from actual capacity, and that the
corresponding rights, with value ZK, have been auctioned off, that the
monopoly distortion vanishes. The incremental capacity then yields
dZ
Z þ ðK K Þ dK close to Z. As in the case of Contracts for Differences,10
forward sales restore proper incentives for a player with market power.
Hogan [1992] and Bushnell and Stoft [1996, 1997] show that under certain
conditions (besides nodal pricing, no increasing returns to scale, congestion
rights satisfying feasibility constraints, no market power in the wholesale
market, well defined property rights, a complete set of competitive liquid
forward markets to provide sufficient information on long run demand and
supply conditions and risk management, etc.), all profitable transmission
investments are efficient. This powerful result appears to transform the
transmission investment problem from one that appears to be almost
intractable to one that requires a simple implementation of a property-rights
based market system.
Accordingly, merchant investment’s appeal is that it allows unfettered
competition to govern investment in new transmission capacity, placing the
risks of investment inefficiencies and cost overruns on investors rather than
consumers, and bypassing planning and regulatory issues associated with a
structure that relies on regulated monopoly transmission companies. In
addition, in theory, it allows investment in new generating capacity in the
10
See Green (1999).
constrained area to ‘compete’ with new transmission investment that reduces

the import constraint. In this way, market driven transmission investment is
an economist’s dream, solving the problems associated with imperfect
regulation of a ‘natural monopoly’ transmission company and aligning
competitive transmission investments with the newly developed competition
in the generation segment. Unfortunately, the optimality of the market driven
approach depends on a number of strong assumptions and conditions that are
likely to be inconsistent with the actual attributes of transmission investments
and the operation of wholesale markets in practice.11
We turn now to a discussion of what we view as the most important ass-
umptions underlying the case for the merchant model and the implications
of relaxing these assumptions.
IV. IMPERFECTIONS IN ENERGY MARKETS
The reasoning above assumes that the prices that clear the markets in the
North and the South reflect the marginal costs of production (and the
marginal willingnesses to pay12) at each location, so that the congestion rent
perceived by merchant investors does reflect the social savings brought
about by the investment. That is, potential investors in new transmission
capacity see the correct locational price signals in the wholesale markets.
Factors that may distort prices include market power, regulatory interven-
tions like price caps, the absence of a complete representation of consumer
demand in the wholesale market, and discretionary behavior by system
operators when the network is constrained.
Suppose for example that there is a generator with market power in the
South, and that the latter region is import constrained. The generator
exercises market power by withdrawing capacity and driving the price in the
South up. Hence
p S > cS ;
where cS denotes the marginal cost of production in the South. The
measured congestion rent then overestimates the cost savings associated
with the replacement of one unit of power generated in the South by one unit
of power generated in the North, suggesting an over-incentive to reinforce
the link, ignoring other market imperfections.13 On the other hand, the
increased transmission capacity does not replace production in the South
one-for-one; it also leads to an increase in total energy consumption in the
11
Of course, these are attributes with which any alternative investment framework, including
any regulated investment framework, must contend.
12
We will present the argument in terms of cost savings; because what matters is net supply at
each node, the same argument would apply to the demand side.
13
For example, as we discuss below, lumpiness in transmission investment leads to under-
investment.
South, which yields a social benefit equal to (pS cS) times the expansion in
consumption. The box below shows that, under a weak assumption, the first
effect dominates, and therefore market power results in an over-incentive to
invest in merchant transmission. The assumption is that the Southern
monopolist’s reaction curve be downward sloping in a Cournot game.
Intuitively, the transmission line creates a Cournot ‘duopoly’ in the South, in
which the Southern firm faces a fixed output from its (transmission) rival. A
downward sloping reaction curve means that the Southern firm curtails its
output as the transmission capacity expands. This implies that the energy
consumption increase effect is smaller than the inflated signal effect (the two
effects would cancel if the output in the South were invariant to an increase
in imports from the North).14
The impact of locational market power on merchant investment incentives

Consider a monopoly supplier in the South producing at marginal
cost cS and facing demand function D(pS). Let P( ) denote the inverse
demand function and Sg( ) the gross surplus. Provided that the
transmission capacity K between North and South is fully utilized, the
monopolist solves:
maxfðpS cS Þ½DðpS Þ K g;
pS
equivalently, this monopolist selects a consumption qS in the South so

as to solve :
maxf½PðqS Þ cS ðqS K Þg:
qS
Letting cN denote the marginal cost in the North and neglecting

consumption in the North, social surplus is
W ¼ Sg ðqS Þ cN K cS ½qS K :
The marginal gross surplus, dSg/dqS is equal to price pS in the South,
and so when the line’s capacity is increased by dK, resulting in a
consumption change dqS, welfare changes by
dW ¼ ðpS cS ÞdqS þ ðcS cN ÞdK:
Note that, with perfect competition, pS 5 cS (and pN 5 cN) and so
dW 5 ZdK. With monopoly power in the South, though,
pS cS > 0
14
Similarly, and to the extent that reinforcing the line is akin to adding production capacity
in the South, this suggests that entrants in generation have too much of an incentive to invest in
the South as well. The box verifies that this is indeed the case.
and
cS cN < Z:
There is an over-incentive to invest if and only if
dW < ZdK;
i.e.,
ðpS cS ÞdqS þ ðcS cN ÞdK < ðpS cN ÞdK;

that is if and only if
dqS < dK;

For there to be an over-incentive to invest, the monopolist must
‘absorb’ some of the increase in imports from the North. To know
whether this is the case, differentiate the first-order condition for profit
maximization:
dqS P0
¼ 0 :
dK 2P þ ðqS K ÞP00
Thus, there is an over-incentive to invest under merchant investment if
and only if
P0 þ ðqS K ÞP00 < 0:
A sufficient condition for this is that the demand curve be concave.
More generally, this condition is the standard condition for quantities
to be strategic substitutes.
The same reasoning can be applied to generation investments in the
South. Indeed, K could alternatively denote the amount of power
produced by a competitive fringe in the South in the profit maxi-
mization exercise. And so (regardless of the fringe’s marginal cost)
dqS < dK
as long as
P0 þ ðqS K ÞP00 < 0:
Hence, there is in general an over-incentive to invest in generation in
the South as well.
Conversely, a generator with market power in the North may (while still
making full use of the link) be able to raise price pN by withdrawing
production capacity F perhaps to the level of pS if it faces no competition
in the North (Oren [1997], Stoft [1999], Joskow and Tirole [2000]). In this
case, the congestion rent underestimates the gain from expanding the line’s
capacity, resulting in an under-investment by merchant transmission
investors. At the same time, it could lead to inefficient entry of generat-
ing capacity in the North in response to the short run monopoly rents
created there.
As a second example, in the case of market power in the South (this
is the situation that will generate very high prices for consumers in the
South), the regulator may be tempted to impose a price cap.15 While the
price cap improves economic efficiency if it really is about constraining
market power, it may also distort price signals if high prices are at least
sometimes due to tight competitive supply and demand conditions rather
than market power. A tight price cap in the South then reduces the
congestion rents during those hours that are very important because they
produce the bulk of the rents to support investment, yielding under-
investment in transmission.16
Third, prices may not clear supply and demand in real time because
market clearing processes are not fast enough to respond to rapid changes in
supply and demand conditions while maintaining physical requirements for
frequency, voltage, and stability on the network. To maintain physical
network parameters, administrative rationing is then substituted for prices
to balance supply and demand as a consequence of what is effectively a
problem of incomplete markets (Wilson [2002]). Moreover, the system
operator has discretion in determining the exact nature of the responses to
operating reserve deficiencies or ‘scarcity.’ For example, Patton [2002]
shows that during tight supply conditions, the system operator takes actions
that tend to depress market-clearing prices. Whether it is administrative
rationing in response to incomplete markets or price controls motivated by
efforts to constrain market power or price distortions caused by market
power or discretionary decisions by the system operator, actual prices will
depart from the efficient prices required to give the efficient signals for new
investment. These imperfections are potentially important with regard to
transmission (and generation) investment because the prices that create
significant congestion rents tend to occur in a relatively small number of
hours and these hours also happen to be the hours when these types of price
distortions are most likely to occur.
15
Or a de facto price cap as when the system operator curtails load administratively when
prices don’t clear the market.
16
Capacity payments cum capacity obligations have been introduced or proposed in some
jurisdictions to respond to the distortions caused by price caps on spot markets. We show
elsewhere (Joskow and Tirole [2004]) that under certain conditions capacity payments may
compensate for the distortions caused by spot market price caps. If capacity payments are not
properly reflected in the prices seen by potential merchant transmission investors, their
investment decisions will be distorted.
V. LUMPY TRANSMISSION INVESTMENTS
V (i). Pre-emption
The analysis of the effects of market power on investment incentives in the
previous section, as well as most of the literature upon which the merchant
investment model relies, assumes that transmission investment is not
characterized by economies of scale or ‘lumpiness.’ However, network
expansion investments are likely to be lumpy. That is, the average cost of a
new link declines as its capacity increases, other things equal (Baldick and
Kahn [1992], Perez-Arriaga et al., [1999]). The impact of lumpiness is
illustrated in figure 2. The initial capacity is K0 and social efficiency would
command that it be brought to a level K1 in a single, discrete investment.
Assuming that the energy market participants are perfectly competitive,
so that net demand/supply curves represent the true marginal costs/
willingnesses to pay, the surplus created (or congestion cost reduced) by the
expansion of capacity from K0 to K1 is depicted by the shaded area in figure
2. The value, Z1(K1 K0) of the transmission rights (equal to the ex post
congestion rents) granted to the merchant investors building this capacity
expansion understates the social surplus S1 it creates by reducing congestion
costs. Lumpiness thus results in an underincentive to reinforce the system for
the same reason that an incumbent grid owner rewarded by congestion rents
has suboptimal incentives to remove these congestion rents.17 What is
needed to create proper incentives for investment is that the merchant
investor be granted the entire shaded area in figure 2. The corresponding
incentive mechanism requires knowledge of net supply and demand curves,
though, and does not fit directly with the merchant investment paradigm.
Another source of lumpiness for network expansion investments arises
because there may be a scarcity of rights of way, for example a unique
corridor between a cheap and an expensive area. The difficulties that new
transmission corridors face in obtaining siting authority suggests that the
available corridors for new lines through many areas will be limited in the
sense that (for example) one additional corridor may be available through
the Pyrenees between France and Spain, and it may accommodate one new
link that could be of any size between 100 MW and 1000 MW. This scarcity is
particularly problematic as demand grows. Merchant investment is then
likely to end up in a ‘pre-emption and monopoly’ situation. In a system with
growing demand, pre-emption leads to an investment at the first date at
which the discounted value of the financial rights on the additional capacity
is equal to the investment cost. The box below also contains an analysis of
the incentive to get a toehold by sinking a small investment to pre-empt
additional entry and produce monopoly rents for the merchant transmission
17
The underincentive effect associated with lumpiness is also discussed by Gans and King
[2000].
surplus S1
Net supply in
η0 η1
the North
Net demand
in the South
K0 K1
Figure 2
investor and under-investment in transmission capacity generally. A

merchant will install a small capacity on the corridor to gain a toehold
and will later expand this capacity (presumably, the merchant will
underinvest in this expansion as we have seen), to the extent that the
expansions are now deepening investments.
Lumpy investments: pre-emption and toeholds

Suppose that at some future date T, the net demand in the South
jumps up (forever) to a new level (the dotted line in figure 3); the post-
reinforcement shadow price jumps from Z1 to Z2 4 Z1. Letting r denote
the interest rate and I the investment cost, suppose that
Z1 ðK1 K0 Þ < rI < Z2 ðK1 K0 Þ:
Then, under free entry into merchant transmission investment,
investment occurs at date t o T such that

1 erðTtÞ erðTtÞ
Z1 þ Z2 ðK1 K0 Þ ¼ I
r r
(see Fudenberg and Tirole [1985]). Note that this pre-emption is
actually socially beneficial if the surplus S1 brought about by the
expansion before the increase in demand (the shaded area) exceeds the
interest on the investment cost, i.e.:
S1 > rI:
Otherwise, pre-emption is socially wasteful.

Expanded net demand in

the South (after date T )
η′2 η2
Net supply in
η0 η ′ η1 the North
Net demand
in the South
(before date T )
K0 K′ K1
Figure 3
And the point about underinvestment remains: Letting S2 denote

the surplus after demand has grown, if
Z2 ðK1 K0 Þ < rI < S2 ;
then no merchant investment ever takes place even though it is socially

desirable.
Similarly, we can show that pre-emption may encourage
inefficiently small investments. Suppose that capacity K1 can either
be reached in one stage, at cost I, as discussed above, or in two stages.
We assume that the second-stage upgrade can be performed only by the
first-stage merchant investor. The first stage costs I 0 and yields capacity
K 0 , K0 o K 0 o K1, which can then be upgraded at cost I 0 0 to K1.
Let Z0 2 ðZ1 ; Z2 Þ denote the congestion cost for capacity K 0
before demand in the South jumps up. Similarly, let Z02 denote the
shadow price after T when transmission capacity is K 0 . See figure 3. The
first-stage merchant investor has an incentive to upgrade the facility at
date T to yield total capacity K1 if and only if
ðK1 K0 Þ ðK 0 K0 Þ
Z2 I 00 > Z02 ;
r r
which we will assume. Let us look for an equilibrium in which a
merchant investor preempts at date t o T by investing a little (I 0 ) and
then upgrades the line at time T:

1 erðTtÞ 0 0 Z ðK1 K0 Þ
I0 ¼ Z ðK K0 Þ þ erðTtÞ 2 I 00 :
r r
For this, it must be the case that preemption at (t e) with the full
investment does not pay off:

1 erðTtÞ erðTtÞ
I* Z1 þ Z2 ðK1 K0 Þ;
r r
or h i 1 erðTtÞ
I I 0 þ erðTtÞ I 00 * ½Z1 ðK1 K0 Þ Z0 ðK 0 K0 Þ:
r
The right-hand side of this inequality is negative if the total value of
the rights (the total congestion rent) decreases with the capacity of the
link in the relevant region. The left-hand side represents the cost of
building all the links at once rather than in two steps, and may be
positive or negative.
Aside from the timing considerations discussed above, note that
given an entry at t, a social planner might want to jump to capacity K1
directly, as the difference in social surplus for capacities K1 and K 0
exceeds the increase (or decrease) in the flow revenue of rights
holdings. Thus, there may be too small an investment.
Remark: Merchant transmission projects that increase capacity between

an import constrained area with high nodal prices and an export constrained
area with low nodal prices are, in a sense, substitutes for generation projects
of equivalent capacity inside the import constrained area. This competition
between transmission and generation projects itself raises pre-emption
issues. Suppose for instance that a merchant investor plans to invest in a new
North-South line, whose acquisition of siting permits plus actual construc-
tion will take say ten years.18 If the transmission project is not built, assume
that a generator with equivalent capacity would be built in the South and
that such a project would take only three years to obtain siting approvals and
to be constructed. There is room for only one of these two alternative ways of
reducing the price wedge between North and South. The merchant
transmission investor is at a strategic disadvantage even if his project is
socially more valuable. If few sunk costs of building a new line are incurred
within the first two years, then the merchant investor is likely to cancel his
project if the new generation plant in the South is built. Knowing this, the
generator may well try to use his short-term investment period to pre-empt
18
Transmission lines do not take very long to build once they have obtained siting permits.
However, for major new transmission corridors, the permitting process can be very lengthy.
the transmission project even if the merchant investor has announced his
intention and has started (limited) work on this project.19
V (ii). Gaming between merchant investment projects
Section 5.1 argued that substitute merchant investments give rise to
preemption games. By contrast, complementary merchant investments,
assuming that they are best undertaken by separate entities, may give rise to
wars of attrition. As illustrated in the following example, with complemen-
tary investments each owner has an incentive to dimension his project
slightly smaller than the other owner’s project; the former project then is the
bottleneck and receives the entire congestion rent while the latter exhibits
excess capacity and generates no rent. Such a situation is conducive to wars
of attrition in which no owner wants to move first by fear of being ‘under-
dimensioned’ by the rival owner.
Gaming and coordination of merchant investment projects

An important benefit of market mechanisms is the freedom economic
agents enjoy in their investment decisions. They don’t need to coor-
dinate with other agents. In the case of electric networks, coordination
will be needed not only between transmission and generation invest-
ments, as we just discussed, but also between transmission invest-
ments. To illustrate this, consider the pair of transmission investment
projects depicted in figure 4. Complementary investments from North
to Middle and from Middle to South will allow cheap power to flow
from North to South. While this pair is really a single investment from
an economic viewpoint, the investments may be undertaken by
different entities for technological reasons (one is an AC line and the
other a DC line and the two companies have different expertise) or
other reasons (for example, through separate ownerships of rights of
way or different political jurisdictions enabling siting).
Suppose the two complementary projects are built (that is, the
standard ‘coordination failure’ does not arise) and that companies choose
their capacities KNM and KMS. The value of the point-to-point rights are
KNM ðpM pN Þ;
and
KMS ðpS pM Þ;
respectively. The incentive for gaming comes from the fact that the
lower-capacity line grabs the entire rights’ value: Suppose, for
19
Such timing issues are of course not specific to transmission investments. But the latter are
particularly vulnerable to pre-emption strategies due to their long lead time.
N pN
KNM
M pM
KMS
S pS
Figure 4
instance, that KMS o KNM. Then, pM 5 pN while ps 4 pM. This gives

rise to a game in which each would like to have a capacity slightly lower
than the other. Hence, none dares to move first as the other will then
make sure to collect the entire rent.
To be certain, this example is extreme, but it illustrates a general point:
Merchant investment is conducive to pre-emption [see section V (i)]
and war-of-attrition strategies (Tirole [1988] pp.311–314). These
detrimental behaviors can be avoided through a centralized process
involving various forms of incentives to promote incentive compat-
ibility. But, by so doing, the market moves closer to a centralized
process.
V (iii). Does the Coase Theorem Apply?

It is sometimes argued that the problems created by lumpy investments can
be resolved through negotiations between the various market participants
who will benefit from the investment; that is, that the ‘Coase Theorem’
applies. There are many reasons, exposited in the following example, to
believe that negotiations among the affected market participants is unlikely
to solve the problems.
Lumpy projects: Will winners get on board and losers be compensated?
As was discussed in section V (i), the price system does not supply the
proper signals when the transmission investment has a substantial
impact on nodal prices. Merchant investment, to be efficient, then
must involve some stakeholders’ process. Its proponents build on the
Coase Theorem to argue that winners will participate in the funding of
socially desirable projects, while losers will pay enough so to prevent
socially undesirable ones.
There are however several issues with this argument, some general
and well-known, and some more specific to the context:
Transaction costs: Coasian bargaining involves transaction costs,

especially when the number of stakeholders is large.
Asymmetric information: In order to reach an efficient agreement,
parties must make reasonable demands. When imperfectly
informed as to what others will accept, participants in a bargaining
process may end up being too greedy, resulting in an inefficient
bargaining breakdown.
Absence of future players: Efficiency requires that all parties be
present to negotiate a deal. Future newcomers in generation,
transmission and consumption by definition do not sit at the
bargaining table and so their interests will not be accounted for by
the existing parties.
Non-excludability of winners and free riding: The design of
restructured electricity markets imposes uniform prices at a given
node: this is mechanistically so in the case of pools, but it applies as
well to bilateral-transactions-based systems. Thus, a consumer in
the South benefits from the reinforcement of the North-South line
and has no incentive to join other consumers in the South and other
winners (producers in the North) to contribute to defray the
transmission investment if such free-riding does not prevent the
investment from being made. The current market design does not
call for exclusive rights for the financing parties, and thus
encourages free riding in a merchant investment context.
Hold-up of potential losers: To reach an efficient agreement when a
socially inefficient transmission investment is being considered, it
must be the case that losers be able to ‘bribe’ the winners not to
make this investment. While this bribe enables an efficient outcome
‘ex post’, when anticipated ‘ex ante’, it may provide parties with the
wrong incentives.
To illustrate the last (hold-up) point, consider a fixed demand qS in the

South (corresponding to gross surplus vqS F we assume an inelastic de-
mand for computational simplicity). The unit cost of building generation
capacity in the South is IG c0 qS , where c0 is the per-unit cost of invest-
ment. This capacity can be built by a monopoly generator (the analysis
can be altered to accommodate the case of a competitive set of generators
in the South). The variable (ex post) cost of producing qS is cqS.
Suppose now that, at cost
IT > IG ;
a link with capacity exceeding qS from another region (North) to South

can be built. There is already plenty of competitive generation in
the North at the same variable cost c. Thus, the building of the
transmission line involves a social loss of IT IG if the generation in the
South is not yet built and of IT if this generation is already in place.
Now suppose that there is investment in generation in the South (the
efficient solution). The monopolist charges v per unit, resulting in profit
(v c)qS (gross of the investment cost IT). Suppose next that the
consumers (or their LSEs) form a coalition and threaten to build the
transmission line. If they do so, Bertrand competition brings the price
down to c, and so the threat is credible if
ðv cÞqS IT > 0;
which we will assume. Coasian bargaining implies that the monopoly
generator will bribe the consumer coalition not to implement the project,
where the bribe b lies in an interval:
b 2 ½ðv cÞqS IT ; ðv cÞqS :
The lower bound of the interval corresponds to the coalition surplus
from bypass, and the upper bound to the quasi-rent enjoyed by the
monopolist in the absence of bypass. So, if b denotes the coalition’s
bargaining power:
b ¼ b½ðv cÞqS þ ð1 bÞ½ðv cÞqS IT
¼ ðv cÞqS ð1 bÞIT :
Knowing that it exposes itself to paying a ransom, the generator in

the South invests if and only if
ð1 bÞIT*IG ;
a condition that is violated unless b is small. So, if the condition does
not hold, the efficient investment in generation does not occur.
The inefficiency can be prevented by moving Coasian bargaining to an
ex ante stage F that is, in effect by having the generator and the
consumers engage in long-term contacting. (This solution of course raises
other issues, such as the absence ex ante of relevant parties at the
bargaining table; or the barrier-to-entry potential of long-term contracts
as studied by Aghion and Bolton [1987].)
VI. STATE-CONTINGENT RIGHTS AND DIVERSIFICATION
The existing analyses of merchant investment assume that the capacities

K0 and K1 are well-defined and non-stochastic, and abstract from
some important issues. In practice, even in the two-node model, the
actual capacity of the North/South link depends ex post on exogenous
environmental parameters such as temperature and other exogenous
contingencies;20 furthermore, system operators have substantial discretion

in defining and implementing security constraints, affecting the actual power
flows on the link in real time. And, of course, even a well-maintained system
will have some random outages that cause the available capacity of the link
to be reduced.
Moreover, the ex ante estimated physical capabilities of a transmission
network are defined by relatively crude administrative risk criteria (N-1, N-
2) and a set of assumed system ‘study’ conditions that are discretionary
decisions of the system operator that could change in the future.
This all raises the issue of the number of financial rights to be allocated for
the existing system and, as a consequence of new investments, how
congestion revenue deficiencies or surpluses arising from deviations between
the number of rights allocated ex ante and the actual capacity of the network
ex post are handled, and how these allocation and compensation decisions
affect investment and the ultimate performance of the system.
We consider this issue in the simple two-node case. Suppose that K varies
with the state of nature o: K 5 K(o), K 0 (o) 4 0 and o is distributed between
o and o þ . Let’s say that the line is congested for all values of o, but the
value of Z will vary with K(o). For which value of o should one compute the
number of financial rights? One could be conservative and set the number of
financial rights equal to K(o ). One would issue K(o ) financial rights and
owe the holders ZK(o ) in congestion payments. When the realized o is o ,
one satisfies the feasibility and revenue adequacy condition. But what
happens when o 4 o ? The merchandising surplus will exceed what is
owed to the rights holders. What does one do with the excess and how does
the distribution affect investment incentives? At the other extreme, one
could set the number of financial rights to reflect the maximum capacity
K(o þ ). There would be revenue adequacy when o 5 o þ but not when
o o o þ , which would be most of the time since the system operator would
owe ZK(o þ ) regardless of the actual realization of o. Where does the
shortfall come from and how does it affect investment incentives? The
20
The import capacity of Path 15 (connecting Northern and Southern California) when all of
the associated AC lines are operational varies between about 2600 MW and 3950 MW
depending upon the ambient temperature and remedial action schemes that are in place to
respond to unanticipated outages of generating plants and transmission lines. The variance
reflects both thermal limits and the application of a set of reliability criteria based on complex
system contingency studies. The rated capacity of Path 15 falls by about 600 MW as the
ambient temperature rises, other things equal. The rated capacity of Path 15 varies by about
1300 MW depending on the availability of various remedial action schemes to respond to
transmission and certain generation outages. California ISO, Operating Procedure T-122A,
November 6, 2002. It is also important to recognize that in the U.S. and Europe, there is not a
single SO controlling the network, but multiple SOs controlling independent segments of the
network. To maintain reliability and avoid free riding, less flexible contingency criteria must be
defined than might be the case if there were a single SO operating the network in real time.
answers to these questions necessarily affect the incentives merchant
investors will have to make transmission investments.
The impact of a generous [K(o þ ), say] or conservative [K(o ), say]
distribution of rights on investment incentives depends on the way the
resulting shortfall or surplus is financed or redistributed. Suppose, first, that
one appeals to the taxpayer. Even if taxation were lump-sum, there would
still be distortions in investment behavior; generous distributions over-
incentivize merchants, while conservative ones under-incentivize them. By
contrast, appropriate taxes on users of the transmission network21 make
biased distributions of rights neutral in this radial network, provided that
the dispatch is efficient: An efficient dispatch implies that in each state of
nature, the cum-tax (or subsidy) price at a given node fully utilizes the link.
And so end-users are unaffected by a generous or conservative distribution
of rights on the line. Because there is no source or sink of money outside of
the industry, rights owners receive the same overall dividend (for example,
through a smaller per-unit dividend in the case of a generous distribution).
This taxation solution is however complex, since the taxes have to be state
contingent.
Rather than dealing with the payment shortfall or surplus problems
associated with the kinds of property rights that have been proposed, it
seems more natural simply to divide the merchandizing surplus proportio-
nately among the rights’ owners based on their relative ownership shares.
The next box analyzes the optimal relative allocation rule (in the same way
percentage ownership, but not the total number of shares, matters to
determine one’s proceeds from the distribution of a firm’s dividend, the exact
number of rights does not matter as long as the merchandizing surplus is
distributed proportionately among rights’ owners). It shows that the
optimal allocation rule derives from standard asset pricing (CAPM)
principles in finance. An addition to an existing link is particularly valuable
if its actual capacity remains high when the primary link is very congested.
Its construction then creates a diversification benefit. These diversification
benefits are ignored both in the traditional definition of fixed MW point-to-
point property rights payment obligations and with a simple proportional
allocation rule.
For example, suppose that the primary North-South AC line exhibits
reduced capacity (or breakdowns) during very hot weather. Then an
addition along the same path has less social value if it is an aerial AC line
than if it is an underground DC line not subject to the same climatic shocks
21
We here have in mind proportional taxes on electricity in the South (so the price is pS þ tS,
where pS is the nodal price in the South) and a tax on exports from the North (where generators
receive pN tN ). When actual capacity (Ka) is larger than the number of rights K, this
effectively reduces the net dividend paid to rights holders so that there is sufficient revenue to
compensate them.
even if average line availability is the same for both. Allocations of rights
based only on expected link capacity therefore miss the point that, for a given
capacity, some lines may provide better insurance than others. More
generally, an allocation of rights solely in proportion to (or equal to)
expected capacities provides insufficient incentives to build lines whose
availability covaries with the shadow price less (in absolute value) than that
of the existing lines.22
Non-contingent financial rights under state-contingent capacities

Consider the North-South network (see, e.g., figure 1). The initial
expected capacity of the link is K (the actual capacity will in a moment
be assumed to be state-contingent). A merchant investor contemplates
adding a small amount dK of expected capacity to the line.
Actual dispatching depends on the realization of the state of the
world o. The state of the world encompasses the uncertainty about net
demand in the South, DS(pS, o), that about net supply in the North,
SN(pN,o), and the actual capacity of the lines:
½1 þ yðoÞK for the existing linksðsÞ
and
½1 þ mðoÞdK for the new facility;
where we can normalize the noises to have zero means:
E ½yðoÞ ¼ E ½mðoÞ ¼ 0:
Let us assume that the SO dispatches optimally given the state of
nature; and so, in states of nature in which the North-South link is
congested:
DS ðpS ðoÞ; oÞ ¼ SN ðpN ðoÞ; oÞ
¼ ½1 þ yðoÞK þ ½1 þ mðoÞdK:
Let ZðoÞ pS ðoÞ pN ðoÞ denote the (state-contingent) shadow

price of the link.
Suppose further that K total rights are distributed among all rights
owners, including the merchant investor, and that the distribution is
22
This point applies as well to the case of networks with loop flow. That is, if a transmission
investment has a diversification benefit, it must be reflected in the allocation of rights if the
rights are relied upon to stimulate efficient transmission investments.
proportional to average capacities; and so the merchant investor receives
dK
K rights:
K þ dK
The merchandizing surplus is distributed to rights owners. Needless
to say, distributions of rights that would not reflect expected capacities
could by themselves introduce a bias in merchant investors’ incentives.
For example, suppose that the incumbents in the past received a very
generous rating of the existing lines from North to South, while rating
standards are strengthened for new comers. The latter then receive a
disproportionately small share of total rights, which penalizes them
when the merchandizing surplus is distributed among rights’ owners,
and thereby gives little incentive to sink new investment. To avoid such
obvious biases, we assume an allocation of rights proportional to
average capacity. Even so, merchant incentives may be inappropriate,
as we will see shortly.
The congestion dividend, d(o) paid to the owner of a right is therefore:
K dðoÞ ¼ ½½1 þ yðoÞK þ ½1 þ mðoÞdK ZðoÞ:
The merchant investor’s expected revenue, R, is therefore

dK
R ¼ Eo K dðoÞ
K þ dK

½1 þ yðoÞK þ ½1 þ mðoÞdK
¼ ðdK ÞEo ZðoÞ
K þ dK
’ ðdK ÞEo ½½1 þ yðoÞZðoÞ

or dK small.
By contrast, the increase in social welfare is
dW ’ ðdK ÞEo ½½1 þ mðoÞZðoÞ;
Since in state o, the merchant investor’s expansion delivers
½1 þ mðoÞdK units of transmission capacity which each have value
Z(o).
Hence,
R)dW () covðm; ZÞ* covðy; ZÞ:
Let us draw the implications of this simple characterization in

specific environments:
Example 1 (diversification effect). Suppose that all uncertainty results
from line availability. Existing line(s) may exhibit reduced capacity due
to harsh weather (extreme heat) conditions. The merchant investor’s

line, by contrast is not (or is at least less) subject to these harsh weather
conditions (or is better protected against them). For example, the new
line could be underground, or cross a climatically distinct area. Then
y(o) and Z(o) are (in a first approximation) perfectly negatively
correlated, while Z is not perfectly correlated with y:
m ¼ ky þ e
with
k < 1 and EðejyÞ ¼ 0:
Hence:
covðm; ZÞ*covðy; ZÞ;
implying
R < dW:
Non-contingent rights create an under-incentive to invest. Intuitively,
the new line supplies a disproportionately high share of the
transmission capacity in those states of nature in which transmission
capacity is scarce and therefore very valuable. This contribution
however is not reflected in the distribution of dividends which is based
on fixed (non state-contingent) shares. It is only when the availabilities
of the lines (old and new) are perfectly correlated that the private and
social incentives coincide.
The analysis can be generalized to encompass uncertainty about
energy market participants’ demand and supply curves. Suppose
that
SN ¼ aN þ bN pN þ eN
:
DS ¼ aS bS pS þ eS :
So o 5 (y,m,eN,eS). Under efficient dispatching

aS aN eS eN 1 1
ZðoÞ ’ þ þ þ þ Kð1 þ yÞ:
bS bN bS bN bS bN
This implies that the analysis above generalizes when line availabilities
are independent of demand and supply shocks (cov (ei, y) 5 cov
(ei, m) 5 0 for i 5 N, S).
On the other hand, line availability may be related to demand and
supply shocks. For example, it may be that a line (old or new) is subject
to the same climatic shock as the demand node. Hot weather may
simultaneously increase demand and limit the capacity of the line
bringing electricity from a cheaper node (precisely when the line is most
needed). Such a line obviously has a lower social worth than one whose
availability is less negatively correlated with increases in demand at the
expensive node.
Example 2 (uncertainty about energy market players only). Suppose
that there is no uncertainty about the actual capacities of the lines:
yðoÞ ¼ ZðoÞ ¼ 0 for all o:
Hence all uncertainty comes from generation and consumption. In
this case, the private and social incentives coincide:
R ¼ dW:
VII. NOMINAL AND ACTUAL TRANSMISSION CAPACITY
The difficulty in putting a number in front of a line’s ‘capacity’ (in the two-
node case) raises other issues. In a nutshell, the actual capacity of the line
depends on discretionary choices, and under a merchant transmission model
control is likely to be separated from ownership due to the conflict of interest
associated with the measurement of congestion rents. This section considers
one such discretionary action: dispatching.
As we have already noted, rewarding merchant investment through
congestion rents requires separating ownership and dispatch in order
to obtain an unbiased measure of this rent. But this separation of owner-
ship and dispatch raises a moral-hazard-in-teams problem. The electric
system’s state-contingent output (to simplify, the intensity of power in the
absence of outage and the probability and duration of an outage) depends on
both the care and the forecasts of the owner (the quality of the line, its
maintenance, and its adequacy to consumers’ needs) and the quality of the
management of the grid by the system operator, as the latter must use her
acumen to get lots of power through without creating a high risk of outage.
This raises two points: First, one cannot consider incentives given to
merchant investors without also specifying those of the system operator.
Second, moral hazard in teams reduces accountability. An outage can be
claimed to result from poor line maintenance or from imprudent
dispatching. Conversely, high power prices may be due to an undue
conservatism of the system operator or to a proper dispatching motivated by
low line quality.
There is also a potential moral-hazard-in-teams problem among line
owners. Recall that merchant investment incentives are better aligned with
the public interest when merchants own inframarginal transmission facilities
whose congestion rent is to be preserved. The total North-South capacity
may then belong to different owners. The same value of a given actual
capacity K selected by the independent system operator may correspond to
different quality configurations of the various components of the network

with multiple owners. The question is then one of allocation of total capacity
and congestion rents among the different owners.23
Moral hazard in teams : transmission owners and system operator

Consider the North-South network. Let K denote the nominal
capacity of the line. In a first step, we assume that this capacity is
known to the system operator (for example, the line’s maintenance is
perfectly observed by the SO). The system operator chooses to allow an
amount Kb to flow through the link. The greater is K, b the higher is the
probability that the link breaks down. We assume that with
probability xðKb KÞ the link breaks down and no power flows
through it. With probability 1 xðKb KÞ; Kb flows through. The
function x is increasing in K. b Let LðKÞ b denote the out-of-merit
b [For example, in
dispatch cost when the realized capacity of the line is K.
figure 1, L was equal to the area of the triangle ABC (for capacity K)]. We
assume that there is no market power at either node and so L represents
the social loss attached to the inability to import power without
constraint from the North. Note that
L0 ¼ Z:
The socially optimal dispatch solves, for a given K,
n h i o
min x Kb K Lð0Þ þ 1 x Kb K L Kb :
b
K

And so Kb ¼ Kb is given by
h i h i
x0 Kb K Lð0Þ L Kb þ 1 x Kb K L0 Kb ¼ 0:
The marginal social gain from capacity expansion is then (using

the envelope theorem):
dW
¼ x0 ½Lð0Þ LðKÞ ¼ ð1 xÞZ:
dK
And so, if the marginal investment is rewarded by the congestion cost
in the absence of outages, merchant investors face the proper signal for
investment.
23
At an abstract level, one can view transmission owners and the SO as a team (in the sense of
Holmström [1982]) jointly delivering an output F state-contingent powerF to the final

consumers. A general principle is that proper incentives require that each member of the team
be made the residual claimant for the team performance ðxLð0Þ þ ½1 xL Kb in the notation
of the next example). Making each member residual claimant may however be very costly
because of adverse selection and collusion.
Dispatcher with conservative incentives.
Turn now to the system operator’s incentives. Suppose that the SO
is penalized more for outages than she is rewarded for increases in the
amount of power flowing through the network ; that is, she solves:
n h i o
min x Kb K yLð0Þ þ 1 x Kb K L Kb ;
b
K
where y 4 1. This yields first-order condition:

h i
x0 yLð0Þ L Kb þ ½1 xL0 Kb ¼ 0:
The marginal social gain from capacity expansion is
dW h i d Kb
¼ x0 Lð0Þ L Kb x0 ð1 yÞLð0Þ :
dK dK
And so
" #
dW d b
K
¼ ð1 xÞZ þ x0 ð1 yÞLð0Þ 1 ;
dK dK
using the SO’s first-order condition.

Next, rewrite the SO’s optimization program as the choice of a risk
factor D Kb K:
minfxðDÞyLð0Þ þ ½1 xðDÞLðK þ DÞg:
b
K
The cross-partial derivative of the minimand with respect to K and D
is positive as L 0 0 4 0, and so, by a revealed preference argument, D is
decreasing in K, or:
d Kb
< 1:
dK
In words, the system operator takes less risk as K increases, because
the marginal gain from increased throughflow decreases. We therefore
conclude that
dW
< ð1 xÞZ;
dK
and so congestion rent payments over-incentivize merchant investors.
In a sense, the SO’s conservative behavior implies that insufficient use
will be made of the added capacity and so the shadow price of the link
overstates the value of additional capacity. This result shows that one
cannot properly analyze merchant investment (or, for that matter, the
incentives of a Tranco company not responsible for dispatching)

without considering the independent system operator’s incentives.
VIII. CONCLUSION
We conclude that the apparently attractive properties of the merchant

transmission investment model are seriously undermined when more
realistic assumptions about the attributes of transmission networks,
transmission investment and unregulated power markets are introduced.
Clearly, policymakers cannot proceed under the assumption that they can
avoid dealing with the difficult issues associated with stimulating efficient
investment in electric transmission networks simply by adopting the
merchant investment model. The merchant model ignores too many
important attributes of transmission networks and the behavior of
transmission owners and system operators. In principle, a regulated
Transco model can deal directly with issues associated with lumpy
investment, market power in wholesale power markets, gaming behavior
of merchant investors and stochastic attributes of transmission capacity,
and avoids the need to separate transmission ownership and system
operations. However, a regulated Transco model will necessarily confront
inefficiencies resulting from asymmetric information and political inter-
ference in planning and investment processes and may be less effective than a
merchant model in providing the high powered incentives that lead to the
identification of innovative transmission investment options, construction
costs minimization and efficient tradeoffs between generation and
transmission investments. The role of reliability rules driven by the unusual
physical attributes of electricity and the need to maintain network
frequency, voltage and stability parameters on a continuous basis leads to
additional complications that have hardly been addressed in either the
merchant investment or regulated Transco models.
We believe that it is unlikely that policymakers will be able to rely
primarily on a merchant model to govern transmission investment and the
limited amount of merchant investment that has been forthcoming to date in
electricity markets where it is permitted and encouraged, is consistent with
this view. An important research challenge is to develop good regulatory
mechanisms to apply to regulated Transcos that also provide opportunities
for merchant investors to develop projects when they are the most efficient
options.
REFERENCES
Aghion, P. and Bolton, P., (1987), ‘Contracts as a Barrier to Entry,’ American Economic
Review, 77(3), 388–401.
Baldick, R. and Kahn, E., (1992), ‘Transmission Planning in the Era of Integrated
Resource Planning: A Survey of Recent Cases,’ Lawrence Berkeley Laboratory, LBL-
32231.
Bohn, R. E.; Caramanis, M. C. and Schweppe, F. C., (1984), ‘Optimal Pricing in Electrical
Networks over Space and Time,’ Rand Journal of Economics, 15(3), 360–376.
Bushnell, J., (1999), ‘Transmission Rights and Market Power,’ Electricity Journal, 12(8),
77–85.
Bushnell, J. and Stoft, S., (1996), ‘Electric Grid Investment Under a Contract Network
Regime,’ Journal of Regulatory Economics, 10, 61–79.
Bushnell, J. and Stoft, S., (1997), ‘Improving Private Incentives for Electric Grid
Investment,’ Resource and Energy Economics, 19, 85–108.
Celebi, M., (undated), ‘An Analysis of Incentives to Provide Line Capacity and
Reliability in Deregulated Power Networks,’ mimeo, Brattle Group.
Chao, H. P. and Peck, S., (1996), ‘A Market Mechanism for Electric Power
Transmission,’ Journal of Regulatory Economics, 10(1), 25–59.
Chisari, O., et al. (2001), ‘High-Tension Electricity Network Expansion in Argentina:
Decision Mechanisms and Willingness-to-pay Revelation,’ Energy Economics, 23,
697–715.
Fudenberg, D. and Tirole, J., (1985), ‘Preemption and Rent Equalization in the Adoption
of a New Technology,’ Review of Economic Studies, 52(170), 383–401. Reprinted in
Readings in Industrial Economics, L. Cabral ed. (2000), Blackwell Publishers.
Gans, J. and King, S., (2000), ‘Options for Electricity Transmission Regulation in
Australia,’ Australian Economic Review, 33, 145–160.
Glachant, J-M. and Pignon, V., (2002), ‘Nordic Electricity Congestion’s Arrangement as
a Model for Europe: Physical Constraints and Operators’ Opportunism’, mimeo.
Green, R., (1999), ‘The Electricity Contract Market in England and Wales,’ Journal of
Industrial Economics, 47(1), 107–124.
Green, R., (1997), ‘Transmission Pricing in England and Wales,’ Utilities Policy, 6(3),
185–93.
Hansmann, H., (1996), The Ownership of Enterprise, Belknap Harvard.
Hogan, W., (1992), ‘Contract Networks for Electric Power Transmission,’ Journal of
Regulatory Economics, 4, 211–242.
Hogan, W., (2002), ‘Financial Transmission Rights Formulations,’ mimeo, Harvard
University.
Holmström, B., (1982), ‘Moral Hazard in Teams,’ Bell Journal of Economics, 13, 324–340.
Holmström, B. and Tirole, J., (2000), ‘Liquidity and Risk Management,’ Journal of
Money, Credit and Banking, 32(3), 295–319.
Joskow, P., (2002), ‘Electricity Sector Restructuring and Competition: A Transaction
Cost Perspective,’ in Eric Brousseau and Jean-Michel Glachant (eds.) The Economics of
Contracts: Theories and Applications. Cambridge: Cambridge University Press.
Joskow, P. and Schmalensee, R., (1983), Markets for Power, MIT Press.
Joskow, P. and Tirole, J., (2000), ‘Transmission Rights and Market Power on Electric
Power Networks,’ Rand Journal of Economics, 31(3), 450–487.
Joskow, P. and Tirole, J., (2003), ‘Merchant Transmission Investment,’ mimeo, MIT and
IDEI.
Joskow, P. and Tirole, J., (2004), ‘Reliability and Competitive Electricity Markets,’
mimeo, MIT and IDEI.
Laffont, J-J. and Tirole, J., (1993), A Theory of Incentives in Procurement and Regulation,
MIT Press.
Léautier, T., (2000), ‘Regulation of An Electric Power Transmission Company,’ The
Energy Journal, 24(1), 61–92.

Nasser, T. O., (1997), ‘Imperfect Markets for Power: Competition and Residual
Regulation in the Electricity Industry,’ PhD Dissertation, MIT Department of
Economics.
Oren, S., (1997), ‘Economic Inefficiency and Passive Transmission Rights in Congested
Electric Systems with Competitive Generation,’ The Energy Journal, 18, 63–83.
Oren, S.; Spiller, P.; Varaiya, P. and Wu, F., (1995), ‘Nodal Prices and Transmission
Rights: A Critical Appraisal,’ The Electricity Journal, 8(3), 24–35.
Patton, D., (2002), ‘Summer 2002 Review of New York Electricity Markets’, October 15,
2002, pp.25, 42–64, New York Independent System Operator.
Perez-Arriaga, I. J., et al. (1995), ‘Marginal Pricing of Transmission Service: An Analysis
of Cost Recovery,’ IEEE Transactions on Power Systems, 10(1), 546–553.
Stoft, S., (1999), ‘Financial Transmission Rights Meet Cournot: How TCCs Curb Market
Power,’ The Energy Journal, 20, 1–23.
Tirole, J., (1988), The Theory of Industrial Organization, MIT Press.
Vogelsang, I., (2001), ‘Price Regulation for Independent Transmission Companies,’
Journal of Regulatory Economics, 20(2), 141–165.
Wilson, R., (2002), ‘Architecture of Power Markets,’ Econometrica, 70(4), 1,299–1,340.
Wu, F. P.; Varaiya, P.; Spiller, P. and Oren, S., (1996), ‘Folk Theorems on Transmission
Access: Proofs and Counterexamples,’ Journal of Regulatory Economics, 10(1), 5–24.

Joskow Tirole 2005

Uploaded by

Copyright:

Available Formats

Joskow Tirole 2005

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Joskow Tirole 2005

Uploaded by

Copyright:

Available Formats

THE JOURNAL OF INDUSTRIAL ECONOMICS 0022-1821

Volume LIII June 2005 No. 2

MERCHANT TRANSMISSION INVESTMENT

We examine the performance attributes of a merchant transmission

THE ECONOMIC ANALYSIS OF RESTRUCTURED COMPETITIVE WHOLESALE ELECTRI-

contracts, congestion management, nodal pricing, physical and ﬁnancial

investment, to monopoly pre-emption of competitive generation or

II. MERCHANT TRANSMISSION INVESTMENT: BACKGROUND

Transmission investment institutions cannot be considered independently

antagonistic,6 and it is well-known that cooperatives of members with

Network deepening investments: These are investments that involve physical

Independent network expansion investments: These are investments that

would otherwise have to be confronted if network deepening investments

III. THE CASE FOR MERCHANT TRANSMISSION INVESTMENT

Price Congestion Rents vs Congestion Costs

pS A Congestion Cost Net Supply in

constrained area to ‘compete’ with new transmission investment that reduces

IV. IMPERFECTIONS IN ENERGY MARKETS

The impact of locational market power on merchant investment incentives

equivalently, this monopolist selects a consumption qS in the South so

Letting cN denote the marginal cost in the North and neglecting

ðpS cS ÞdqS þ ðcS cN ÞdK < ðpS cN ÞdK;

dqS < dK;

V. LUMPY TRANSMISSION INVESTMENTS

investor and under-investment in transmission capacity generally. A

Lumpy investments: pre-emption and toeholds

r Blackwell Publishing Ltd. 2005.

Expanded net demand in

And the point about underinvestment remains: Letting S2 denote

Z2 ðK1 K0 Þ < rI < S2 ;

then no merchant investment ever takes place even though it is socially

Remark: Merchant transmission projects that increase capacity between

Gaming and coordination of merchant investment projects

instance, that KMS o KNM. Then, pM 5 pN while ps 4 pM. This gives

V (iii). Does the Coase Theorem Apply?

Transaction costs: Coasian bargaining involves transaction costs,

To illustrate the last (hold-up) point, consider a ﬁxed demand qS in the

a link with capacity exceeding qS from another region (North) to South

Knowing that it exposes itself to paying a ransom, the generator in

VI. STATE-CONTINGENT RIGHTS AND DIVERSIFICATION

The existing analyses of merchant investment assume that the capacities

contingencies;20 furthermore, system operators have substantial discretion

Non-contingent ﬁnancial rights under state-contingent capacities

Let ZðoÞ  pS ðoÞ pN ðoÞ denote the (state-contingent) shadow

K dðoÞ ¼ ½½1 þ yðoÞK þ ½1 þ mðoÞdK ZðoÞ:

The merchant investor’s expected revenue, R, is therefore

’ ðdK ÞEo ½½1 þ yðoÞZðoÞ

Let us draw the implications of this simple characterization in

to harsh weather (extreme heat) conditions. The merchant investor’s

So o 5 (y,m,eN,eS). Under efﬁcient dispatching

VII. NOMINAL AND ACTUAL TRANSMISSION CAPACITY

different quality conﬁgurations of the various components of the network

Moral hazard in teams : transmission owners and system operator

The marginal social gain from capacity expansion is then (using

where y 4 1. This yields ﬁrst-order condition:

The marginal social gain from capacity expansion is

using the SO’s ﬁrst-order condition.

incentives of a Tranco company not responsible for dispatching)

We conclude that the apparently attractive properties of the merchant

Let ZðoÞ pS ðoÞ pN ðoÞ denote the (state-contingent) shadow

K dðoÞ ¼ ½½1 þ yðoÞK þ ½1 þ mðoÞdK ZðoÞ:

’ ðdK ÞEo ½½1 þ yðoÞZðoÞ