Politics, Transaction Costs, and the Design of Regulatory Institutions.
Antonio Estache, David Martimort,
Politics, Transaction Costs, and the Design
of Regulatory Institutions,
(World Bank and ECARE, Brussels)
(Harvard University and INRA, Toulouse).
JEL Classification Nos.: D82, L51
Keywords: Government Organization, Regulation, Multiprincipal, Collusion.
We are indebted to Omar Chisari, Claude Crampes, Gines de Rus, Andrea Goldstein, Richard Green, and Martin
Rodriguez-Pardina for useful discussions and comments.
With the wave of infrastructure privatization that has spread throughout the world, many
countries are working on the creation of new regulatory institutions to monitor the behavior and
performance of their newly privatized infrastructure monopolies. Little conceptual work has been done on
the optimal design of these new institutions. To try to improve our understanding of what this design
should take into account, it seems reasonable to try to review the work on organizational design that grew
from developments in principle-agent theory in the 1980’s and draw some policy lessons. In this
literature, regulation is generally viewed as a game between various players with different degrees of
knowledge and information required to make the choices that lead to the efficient and fair allocation of
resources. The understanding of the incentive problems underlying these games provides some useful
lessons on the optimal design of the regulatory regime. But this is not enough. We also need to improve
our understanding of the internal organization of the government to assess the importance of the
institutions responsible for the implementation of these regimes.
To progress in this direction and provide better analytical support for the recommendations made
by practitioners, we have to move away from the standard normative approach, which has been the
cornerstone of the economics of the public sector so far. In the standard normative context, the
government has generally been viewed as a single entity with a larger possible set of policy tools, a
perfect ability to commit, and endowed with a clear objective function (the maximization of social
welfare).1 This is not enough to assess the efficiency of the internal organization of a government.
This understanding requires, as suggested by Coase (1937) and developed by Williamson (1975,
1985), the recognition that the organization of hierarchies is the result of a minimization of transaction
costs. 2 These transaction costs have to be isolated in order to draw any policy-relevant lessons from
these theories. The first transaction costs we will look at are dependent on the extent of the informational
problem faced by the government, and on the limitations in the scope governments have in relying on the
market to implement some of its regulatory responsibilities. For instance, the right of regulating an
industry cannot be sold to outsiders. Yet, while the state exerts regulatory rights, some aspects of this
responsibility can be subcontracted (auditing, detailed studies leading to rate or price cap revisions, etc.
In addition, a more policy-relevant theory of regulatory institutions has to deal with transaction
costs reflecting difficulties in writing fully contingent contracts. Contracts often cover only a finite
amount of time, and are therefore not necessarily binding to future generations and future governments.
This often leads to a renegotiation of some administrative and incentive costs. Moreover, a unified,
centralized government body ruling the economy in its entirety does not exist. Politics is, at its core,
competition between the several principals involved in public decision-making. As noted by Moe (1986)
and Wilson (1989), bureaucrats are subject to a multitude of influences and regulatory rights, and work
1 Baron and Myerson (1982) and Laffont and Tirole (1993),
2 Importantly, transaction costs only matter because government intervention takes place under asymmetric information.
Under symmetric information, the internal organization of the government has no bites on the outcome of its intervention.
Coasian bargaining within the government makes the internal distribution of regulatory rights within the government be
irrelevant. Transaction costs are therefore dependent on the extent of the informational problem faced by the government at the
time of intervening. When different institutions are associated with different values of these transaction costs, there is scope
for their comparison and for a somewhat rough optimization over regulatory structures.
on an issue is often shared among several agencies.3 Last, there is no government intervention, not even
regulation, without an attempt by coalitions to manipulate its outcome. Regulatory contracts and
institutions have to account for the possibility of a capture of the bureaucracy by interest groups.
Our thesis is this: When these transaction costs are taken into account, structures and processes
will affect regulatory outcomes and hence should be recognized explicitly in the design of incentives
based regulation. First, the structures affecting the regulatory outcome include the distribution of
regulatory rights among different levels of the government, the objectives given to agencies, and the voting
procedures used to elect political principals, all of which influence regulatory decisions.4 This is not
recognized by the normative approach to Public Economics. Second, processes matter to regulatory
outcomes. The timing of government intervention, the length and the span of control of different
regulatory bodies, and the design of the communication channels within the regulatory hierarchy affect the
regulatory outcome. A failure to address these influences is bound to over- or under-estimate the effective
impact of incentive based-regulation on regulatory outcomes.
To explain more clearly how structure and processes matter to regulatory outcomes, the paper is
organized as follows. Section 2 sets up the stages of our analysis. It describes the four-layer hierarchy
(voters-political principals-regulators-regulated agents), which has been the fundamental source of
analysis of New Regulatory Economics so far. At each of these layers, transaction costs can influence the
optimal design of the organization of the government. The rest of the paper demonstrates how the main
types of transaction costs can influence this design. Section 3 describes the government failures that arise
from contract incompleteness. Section 4 proposes possible solutions to each of these failures, beginning
with the observation that structures and processes matter. The point of this section is to show how
different forms of contract incompleteness may countervail each other. Section 5 concludes by developing
2. A Stylized Multi-Layer View of the Regulatory Process in Restructured Infrastructure
Restructuring processes in infrastructure generally leave monopolistic providers in the
distribution of electricity, gas and water. In the absence of any regulation of price, quantity and quality,
monopolists would restrict the firm’s quantity and would charge an excessively high price to consumers.
Regulation is needed to reach a higher level of social welfare but it cannot be done directly by consumers
because of the well-known free-rider problem in collective actions.5 It must therefore be designed and
enforced by institutional regulators who act in the name of voters and who can only handle a limited
number of tasks because of standard bounded rationality problems. As an immediate consequence, the
control of the quality of service, of the price structure and the decision on the rate of return on investment
are left to a utility regulator.
3 Landis (1960) in his report to the President newly elected offers a remarkable survey of these overlapping responsibilities.
See also Kahn (1988) seminal book on regulation.
4 In the UK for instance, the electricity regulator was given specific ranking of its responsibilities. He/She is responsible to
secure all reasonable demand, proper financing and promote competition and the defense of the interest of consumers is
subject to these three primary responsibilities. (see Electricity Act 1989)
5 Even if sometimes governments subsidize consumer associations to create a bilateral monopoly.
However, this control is imperfect. Regulators face informational asymmetries in their
relationship with the firm they regulate. They do not know the exact technologies of the firm and the
elasticity of its demand. They have little knowledge about the internal incentive structure of the firm and
the contracts that it may have with non-integrated input suppliers. Regulators have limited tasks to
perform, but they often have limited instruments to carry out their functions. For instance, the
Environment Protection Agency (EPA) in the US is only allowed to ban the use of some chemical
products, but in principle it could use more fine-tuned incentives to limit pollution damages.
Control imperfections also stem from the fact that regulators are not accountable for their acts
directly to consumers, but instead to “political” principals. Often, these political principals belong either
to the legislative or to the executive branch of the government. They can be endowed with a Federal or a
State mandate and they can have tenures with quite different lengths. They have difficulties in controlling
regulators who are generally at an informational advantage because of their expertise or simply because
of their access to historical information.6 In most countries, the control exercised by the Congress on
these regulators is indirect. There is a system of Committees and Subcommittees representing various
constituencies, which is subject to various influences. In some cases (as in the US), part of the control is
not under the Congress itself, but instead under a government budget office which exerts budgetary
control of these agencies.
Political principals themselves can hardly be put on explicit incentive schemes for maximizing
social welfare because of the free-riding problem that voter-consumers would be facing in controlling
these political entrepreneurs.7 Instead, consumers use the rather rough and imperfect procedures inherent
in voting to impose some accountability on the political principals. Only career considerations and the
threat of political takeovers can adjust the political decision-makers’ incentives. It should be noted that
the political arena is not a perfectly competitive market. Enormous fixed costs must be incurred before
entering this market. This reduces the scope for an efficient control of the incumbents.8 Moreover, voters
are often only imperfectly informed regarding the effort and efficiency of the political decision-maker.
They must therefore rely on rather imperfect measures of their own level of satisfaction before deciding
on their voting strategy.9
Each of these layers of the government can be viewed as a principal-agent relationship with its
own informational problem. The simplified overview of society that we get from the above picture is that
society and its government are based on a sequence of vertical contracts that solve, more or less easily, a
number of agency problems. Loss of control results as incentives trickle down the hierarchy through a
system of delegation. However, government is not only vertical, but, as discussed in the introduction,
much of its peculiarities come from its Multi-principal nature.10 Regulators share the control of the firm.
6 See Kiewert and McCubbins (1991) for a thorough discussion of the issues raised by the delegation
of some powers from the Congress to the bureaucracy.
7 There is a clear parallel between this discussion and of the firm as a monitoring technology due to Alchian and
Demsetz (1972) and Holmstrom (1982).
8 See Myerson (1997) for a discussion of entry in the political market.
9 See the discussion of sociotropic voting rules of voting in Ferejoh (1986).
10 See Moe (1986), Baron (1985) and Martimort (1996b).
Political principals share the control of the regulatory policies at a given date, but they also share this
control with their successors. Legislature are themselves divided into numerous Committees and
Subcommittees with their own objectives, each of them trying to influence the bureaucracy.11 Moreover,
the bureaucracy is not a unified body. Often agencies and bureaus compete with one another for
resources, while at the same time bureaucrats compete for influence within a given industry.
With this complex description of the regulatory organization in mind, we can move to an analysis
of the different contractual incompleteness’ which impede the efficiency of the government’s intervention.
3. Transaction Costs in the Government Organizational Structure
The overall organization of the government can be seen as a nexus of more or less explicit
contracts linking stakeholders together. The absence of an overall grand and complete regulatory contract
creates opportunities for hidden gaming which bring their own set of transaction costs. The contractual
externalities that arise from the incompleteness of regulatory contracts is a reality that departs from the
more idealized view of society in which unified control is feasible. This section reviews the relevance of
the institutional design of transaction costs stemming from various types of government failure: lack of
government commitment, the Multi-principal nature of government, and the excessive discretion of the
various decision-makers in the regulatory process.
3.1 Lack of Commitment and The Need for Renegotiation
Because future contingencies may not all be foreseen at the time the government proposes a
regulatory contract to the firm, and because new information becomes available as the regulator learns
about earlier realizations of the contractual targets, renegotiation of the contract matters and can improve
ex post efficiency. This may, however, lead to some counter-intuitive results for practitioners. To
understand why, it is useful to draw a comparison between situations with and without full commitment.
Under full commitment, the optimal regulation of a public utility repeats the way the optimal
static contract deals with the fundamental trade-offs between efficiency and rent extraction. In order to
reduce the costly informational rent that can be obtained by efficient firms in mimicking less efficient ones
through choosing the same menu of cost targets and lump sum transfers, regulators distort the costs and
production targets towards less efficient values to make them less attractive for an efficient firm.
Without full commitment, after the first period of the relationship the regulator knows more about
the private information of the firm and is willing to renegotiate the regulatory contract to improve second
period efficiency. This search for increases in ex post efficiency may nevertheless introduce some
perverse incentives from an ex ante point of view. Anticipating an increase in the power of incentives
later on, efficient agents behave suboptimally in the earlier periods of the relationship. They tend, in
particular, not to reveal their information and its revelation is slowed down in the hierarchy.12 This is
11 Moran and Weingast (1983) show empirically that the Congress has much of the bargaining power when dealing with the
12The intertemporal inconsistency and its impact on decision rules have been forcefully made in the macroliterature which
has analyzed the issue of rules versus discretion. See the seminal contribution of Kydland and Presscott (1977) and the survey
where the counterintuitive results show. For instance, rates of return may best be increased following a
good performance. Regulated firms may best be allowed to plan new investments or to change their
prices as new information on demand is learned. Otherwise, a public utility anticipating that a regulator
will use the information revealed in the first period of their relationship to readjust the contractual terms
has fewer incentives to reveal any information that can help improve the rent efficiency trade-off ex post.
The main point is that incentive compatibility constraints are hardened by the mere possibility of
renegotiation. As a results, first period inefficient choices are made; underinvestment in specific assets,
periods of inefficient technological choices, prices far above the marginal cost in order to manipulate the
beliefs of the regulator on its marginal cost before the renegotiation stage, etc…. More generally, the
inability to commit increases the transaction costs of contracting, something which is often underestimated
by practitioners. Moreover, these higher transaction costs may be justified by the need to retain some
flexibility in order to ensure that past regulators or political principals are not allowed to bind the future
of society to a given regulation. This is important because consumers tastes may change in the future, or
the political principals themselves will may change after an election and these new decision makers may
favor other groups. Hence, society as a whole agrees on imposing limited terms for the mandates of its
political principals to limit the possible intertemporal abuse of powers.13
3.2 The Multiprincipal Nature of Government
An important step toward assessing the real nature of the rent-efficiency trade-off discussed
above, is to consider government intervention as coming from a whole set of different principals, each
with its own objective. 14 As a whole, these principals may have for a collective objective the
maximization of the same social welfare function as that of a single benevolent regulator. However, each
single principal has only a limited mandate to fulfill. For instance, the Environmental Protection Agency
is concerned with protecting the environment, which is only one aspect of the consumers’ welfare. At the
same time, the Utility Regulator is concerned with controlling the rates of return, or the price cap and
prices structure of the utility. Each of these agencies has only a partial view of the regulatory stake.15
The importance of this complex structure of government can be illustrated by three main themes
in the literature dealing with multi-principals agency relationships: allocative inefficiency, rent distribution
and regulatory rights, and the optimal levels of regulatory decentralization and the timing and voting right
of the various agencies. These factors are all influenced by the design of regulatory institutions.
of the literature presented in Persson and Tabellini (1994, Introductory Chapter). In an agency context, it has been more fully
understood with the works of Dewatripont (1989) and Laffont and Tirole (1990).
13 Constitutional rules exist which impose such limits. In France, there is no limit in the number of mandates that a political
decision-maker may have in a row. The only limitations are on his span of control, i.e., on the number of mandates he may
have simultaneously. In the US, a President can only have two mandates in a row.
14 This difference is may be not so clear cut for the only reasons that within firms the central management is also part of a
bunch of bilateral explicit contractual relationships with different stakeholders (creditors, customers, regulators, owners,
etc…) having their own specific interest in the firm.
15 Baron (1985) and Martimort (1996b) analyze contractual settings describing these situations. Their basic assumption is
that the splitting of responsibilities among the various regulatory agencies comes from a splitting of their monitoring
technologies. As a result of this distribution of regulatory rights each of these agencies is only able to contract on its own
sphere of responsibilities. Their non-cooperative behavior in regulating the industry follows. The regulation implemented is
then a Nash equilibrium among various regulations offered in a decentralized way.
Allocative Inefficiency: Under this complex structure, the regulatory process introduces new
distortions with respect to what would have been implemented under unity of command. The rentefficiency
trade-off, which results from the regulators’ interactions, must capture the fact that a regulator
typically does not take into account what the other regulators’ schemes are when he offers his own
regulatory incentive scheme. Allocative efficiency is affected by the separation of powers between those
Rent Distribution and Regulatory Rights: The non-cooperative behavior of rival regulators also
results in the firm being offered excessively low or excessively high powered incentives, depending on the
set of activities that they respectively control. When several regulators control complementary activities
of the firm, they extract too much informational rent from it and the power of incentives tends to become
excessively low. 16 Each single regulator exerts a negative externality on the other. Cost-plus
arrangements result from this overregulation.17 When regulators instead control substitute activities, the
reverse phenomenon happens as a result of regulatory competition. Each regulator competes with the
other for attracting the agent toward the activity under his own control. The common agent can play one
principal against the other to escape their global control. In equilibrium, because now a principal exerts a
positive externality on the others, higher powered incentives than what would have been collectively
optimal end up being offered. Decisions rules are then closer to their full information values and
allocative efficiency improves.18 As a result, the amount of informational rent that is kept by the firm is
larger than what would be socially optimal.19
Levels of Decentralization: The number of agencies in control of the firm and the timing of their
interventions change the outcome of this structural separation. Administrative processes and structures
affect the outcome of the regulatory intervention. The larger the number of agencies controlling a given
firm – allowing permits, subsidizing or monitoring it – the greater the inefficiency of the regulatory
outcome. Indeed, an extreme version of the free-rider problem may arise when control is highly
decentralized. It may not be worth allowing a given project to be done if one expects other agencies to
restrict their own contribution to its financing. There is at least casual evidence in the US that such
outcomes arise.20 The experience of the EEC provides other examples in which local decisions are
16 Stole (1991) and Martimort (1992) and (1996a) and (1996b) present a general theory which analyzes the contractual
externalities that appear under adverse selection when regulatory powers are shared between non-cooperating agencies.
17 See also Bernheim and Whinston (1986) for an analysis of a common agency game under moral hazard case which also
exhibits this under-provision of effort. Dixit (1996) discusses also a moral hazard model with linear schemes featuring also
lots of the insights of the present model. He applies it to the organization of the bureaucracy. We discuss this model in
18 This is true when a pure equilibrium of the contracting game between the principals exists, Martimort and Stole (1997)
show that it may be possible to find information structures such that only mixed-equilibria exist with substitutes. In this case,
only distributions of regulatory rules are offered in equilibrium by each principal. This suggests that the policy outcome may
be quite fuzzy.
19 This discussion in terms of positive versus negative externalities between different government centers may offer an
unifying framework to read the emerging literature on federalism and budget constraint (see Qian and Roland (1994)). This
literature argues that the allocation of fiscal and monetary policies to the regional and the national levels is due to the nature
of the externalities (respectively positive and negative) that decentralizing these policies imply on the budget constraint of
20 For instance, in the US, transports has been said to be under an extremely inefficient regulation (Kahn (1988)).
Noticeably, it is under several overlapping jurisdictions concerning the ICC, the CAB, the Federal Marine Commissions, the
financed by the supranational institutions in an environment in which goals and incentives are often not
Timing and Voting Right: The sequentiality of the intervention of different agencies calls also for
excessive rent extraction and allocative distortions that are now even larger than in the simultaneous
game.21 For instance, the Stackelberg leader position of the EPA in designing environmental taxes vis-àvis
local States agencies worsens the case for the separation of the regulatory responsibilities on
environmental issues between the State and the Federal levels of the government. Regulation becomes
extremely stringent. An interesting interpretation of these sequential timings is that the Stackelberg leader
benefits somewhat from a veto right and only accepts regulations that guarantee his constituency a
minimal utility. As long as the regulators have quasi-linear utility functions independent of the regulated
firms’ private information, the identity of the vetoing principal has no effect on the regulatory distortions.
But when this assumption is rejected, the decentralization of regulatory responsibilities can lead to an
optimal assignment of regulatory instruments that is also hierarchical.22
In an open rule system, Congress is free to amend any proposals made by a committee. This
means that the regulators face multiple principals. In this type of setting, regulatory agencies are able to
exploit one of these principals against the other and increase their own powers regarding the decision
being implemented. To shift the policy closer to its most preferred outcome, which turns out to be that of
the median-voter, the veto of the presidency is needed as an institutional option. 23
3.3 Discretion of the Decision-Makers
In the regulatory hierarchy, discretion matters at two different levels: First, at the level of the
political principal; and second at the level of the regulatory agency itself. We sometimes distinguish
between formal and real authority. As discussed in Aghion and Tirole (1997), real authority lies in the
hands of agents implementing the final regulatory decision, i.e., in our case the regulators and
bureaucrats. Instead, political decision makers are endowed with the formal authority to make public
decisions. Both are important and need to be recognized to understand how the overall design of
regulatory institutions influences regulatory outcomes.
3.3.1 Discretion of the Political Principal
The Received Theory: Frictionless Influence. A line of research commonly associated with the
Chicago School of economics is often quoted by practitioners to argue the importance of political risks
and discretion in the design of regulatory institutions. This research illustrates very clearly the idea that
political principals who have some discretion regarding the set of policies that can be implemented, are
FPC, the Bureau of Public Roads, the Department of Commerce, etc See also, Landis (1961) for a discussion of the
externalities associated to these separation of powers among agencies and its consequences on regulatory efficiency: Delays in
decision-making, overloads due to inter-regulatory agency conflicts to be settled, etc.
21 See Martimort (1996b) for a formal proof of this assertion.
22 See Estache and Zheng (1995) for a model discussing the optimal instrument assignment in the context of multiple lawyers
of regulators playing Stackeberg game.
23 This veto can be easily modeled in an agency framework as a constraint on the well-being of the president.
subject to a number of conflicting influences exerted by different pressure groups, i.e. consumers, firms,
taxpayers etc… 24 This line of research has put forward the idea that the political principal puts some
values on the contributions or the amount of bribes that he receives from the interest groups, but also
considers social welfare. Contributions are valuable for the political decision-maker, first because they
help him finance his electoral campaigns and second because these bribes may be only pure perks. Social
welfare matters because political decision-makers have reelection concerns and it gives an aggregate
measure of the satisfaction of society.
The first prediction of these papers are that the interest groups are powerful, i.e., they affect the
economic outcome when they are able to circumvent the celebrated free-riding problem in collective
actions. These papers are, however, unable to assess the true reasons why free-riding is less an issue for
some group forming than for others. The true parameters underlying the influence of an interest group
are not characterized. As a result, the outcome of the game is almost immediate: Lobbies influence the
policy outcome by moving the choice of policy toward their own preferred outcomes. The political
decision-maker maximizes a welfare function which favors interest groups.25 Importantly, even if he does
not provide a formal treatment of this issue, Posner (1976) nevertheless also stresses the view that the
force of an interest group may depend on the institutional setting. Structures matter in defining the power
of a group. The main problem is that it is also difficult to see how the contract between an interest group
and the political decision-maker is more enforceable than any other contract aimed at controlling this
political decision maker in the first place. Because it ignores the agency problem that makes the control of
these political principals impossible, this line of research is unable to give a clear account of the role of
the regulatory institutions in constraining their behavior, as well as that of the different pressure groups.
Hence, it has little to offer in terms of specific policy advice.
Political Principals and the Distribution of Rents. An interesting alternative approach is
proposed by Baron (1989) and Laffont (1995). They do not restrict a priori the set of instruments
available to the political decision-makers. However, they make explicit the assumption that the choice of
the political principal is made through a voting procedure. Society is composed by voters with different
preferences on the level of regulation that should be implemented. In fact, these preferences may directly
come from ideology or from the different shares of the regulated firm that each group of voters
individually gets. As a result of the voting procedure, the regulatory choice of the median voter is
implemented. The case to reduce discretion is due to the fact that this political principal does not
maximize social welfare defined as an aggregate of the voters’ utilities, but instead the welfare of the
median-voter, his constituency, and not society as a whole.
An important corollary of this line of research is that suboptimal contracts like average cost
pricing or simple quotas can now find a rationale. Even if these tools are imperfect ways of transferring
wealth between groups, they can also be used to tie the hands of the political principal by reducing his
discretion. But a full endorsement of these instruments would require a better understanding of the way in
which different constituencies can try to align the preferences of the political decision-maker with their
own. In other words, what is lacking is a complete full-fledged treatment of how campaign contributions
and other forms of influence by interest groups affect the outcome of the political game, and therefore the
incentives of a political candidate to align his policy with that of an interest group. Indeed, these political
24 Grossman and Helpman (1994).
25 See Helpman (1997) for a survey of this literature.
principals are putting some value on reelection which may depend on the number of voters who choose
them and the preferences of the latter are likely to be affected by the stringency of the regulation.26
Another unsolved issue is the exact role played by institutions on the final economic outcome. The voting
procedure used to select the political principal is somewhat exogenous. Yet it would seem important to
know how different districts which are affected differently by regulation aggregate their preferences
through the voting procedure.
3.3.2. Discretion of the Regulators
An interest group can be active not only before the enactment of a regulatory policy, as discussed
above, but also at its implementation stage. In other words, interference with the regulatory process can
also arise within the regulatory institution and this is something practitioners tend to be very much
concerned with. This concern is the main anchor to the debate on accountability, independence and
autonomy. The problems can arise from three main sources: internal vs. external influences on the
regulators; conflicts between the legislative powers and the bureaucracy; and the complex dynamics of
Internal and external influences: Bureaucracies are the nexus of conflicting influences.
Regulatory agencies are no different and are subject to several influences, in particular, the internal
influence of a political principal representing the voters’ interests, and the external influence of interest
groups. As a result of this multiplicity of efforts in curbing his behavior, the “regulators-bureaucrats”
end up being vested in the status quo and it becomes extremely hard to provide any incentives.27 There
are two ways to influence the bureaucrats’ incentives: first, wages; second implicit incentives, i.e., career
concerns. In addition, exogenous restrictions on the set of instruments available to the different political
principals controlling the bureaucracy may help to reduce the scope of the bureaucrats in regard to rentextraction.
28 When different principals, for instance different Committees, are affected by the activities
of the bureaucrat-regulator, which are substitutes, the latter can play one principal against the other to get
more freedom. The bureaucracy goes its own way. The solution is to increase accountability by exposing
the bureaucrat and making available private information on the effectiveness of the bureaucrat’s behavior.
Simple institutional rules like the public release of regulatory information may allow this kind of
information sharing between multiple principals. In a nutshell, regulatory processes can matter in curbing
the bureaucracy and practitioners are right to emphasize this point.
The Conflict between the Legislative Power and the Bureaucracy: An agency problem between
the political principal, the Congress and the regulatory agencies can also create the conditions necessary
for the capture of these regulators by an interest group.29 The agency is used by Congress as a
monitoring system observing signals regarding the private information of the public utility. This
26 Laffont and Boyer (1995) build such a model. They assume that a political decision-maker is reelected with a
probability which is proportional to the amount of campaign contributions he receives. This amount is in turn linked to the
informational rent that the interest group gets from regulation. Grossman and Helpman (1996) try to endogenize the weight
that a political decision makers put on an interest group as coming from a campaign contributions game.
27 Spiller (1991) and also Wilson (1989)
28 Dixit (1996)
29 Laffont and Tirole (1993, Chapter 13). Kiewert and McCubbins (1991) make the important point that the delegation
process from Congress to the bureaucracy leads surely to the abdication of any power.
discretion in the hand of the regulator gives him some power or, put differently, some real authority. 30
Indeed, by bribing its regulator, the firm can insure a misuse of his power. The regulator will prefer
taking perks or a position in the industry later on rather than obey the regulatory mandate he has received.
Again, this is why it makes so much sense to restrict future interactions between a regulatory and the
firms it is regulating.
This means that on top of the usual incentive constraints, a regulatory contract must also satisfy a
set of collusion-proof constraints to insure that the regulator does not misuse his power. This does not
come cheap to society.31 The existence of communication costs in the formation of the capturing coalition
or the presence of some transaction costs due to the non-enforceability of the side-contracts will generally
prevent the regulator from getting the full informational rent of the firm. Capture can only be effective
when a stake exists in concealing from the political principal information which is favorable to the
regulated agent. This happens quite often in the context of utility privatization for instance.
The problem is that the constraints needed to avoid collusion/corruption affect the terms of the
trade-off between efficiency and rent-extraction. To limit the regulator’s incentives to collude with an
efficient firm which earns some informational rent from regulation, the political principal has to reduce
further the regulatory stake. This in turn increases the distortions in the production decisions when
compared to the problem free environment. Moreover, these distortions have to be greater when collusion
is relatively more efficient, i.e., when the transaction costs of side-contracting are low. Low-powered
regulation is then called for to limit the scope for collusion. They indeed imply lower levels of rent for
efficient types and therefore less possibilities for corrupting regulators. These low-powered incentives
also correspond to the choice of more bureaucratic rules for the agency. The latter is given less power
and less discretion. The point of this all is that processes also matter and can be quite costly to an
The Dynamics of Bureaucratic Behavior: Part of the reality of regulation is that there are sidecontracts
between a regulator and the interest group he regulates. Since they are not legal, they cannot be
enforced by a court of justice.32 The regulator is lenient with the firm as long as his prospects for getting
“bribes” or future employment opportunities in the industry are high. The firm is ready to “bribe” the
regulator as long as he remains lenient in his stances. Modeling capture or influence of the interest group
as an enforceable side-contract hides some features of the dynamics of capture. Martimort (1997b)
shows that side-contract agreements become easier to enforce when the regulator is better informed about
the firm or when the preferences for the future of the administrative branch of the government are more
pronounced than the preferences for the future of the political principal himself. The discrepancy
between the long-term objectives of the bureaucracy and that of the political principals matter and affect
the regulatory outcome. In sum, transaction costs of side-contracting are endogenous and depend on mere
regulatory institutions. Again, the point is that structures matter to regulatory outcomes.
30 As Niskanen (1971) has forcefully argued, information is the key which allows the bureaucracy to pursue other objectives
than social welfare maximization.
31 The regulatory agency is assumed to have all the bargaining power when offering the side-contract to the interest group. In
other words, there is little competition within the bureaucracy for the exercise of power.
32 A noticeable exception is given by political campaign contributions.
4. The Whole Picture: Transaction Costs at Play
In this section, we take a broader perspective on regulatory institutions to try to see how a more
coherent design of regulatory institutions can be used to reduce the influence of the government failures
discussed in section 3 on regulatory outcomes. Following an argument already applied with some success
by Williamson (1985) regarding the theory of the firm, we argue that the optimal regulatory institutions
are those which minimize the overall transaction costs of contracting. To test the implications of this
paradigm for the internal organization of the government, we use the results highlighted in the last section
to better understand how through playing contractual externalities against one another, the other may help
improve social welfare. This second-best approach is obviously subject to the same weaknesses as any
other second-best analysis. We nevertheless think that this second-best approach turns out to be
extremely useful for getting a richer picture of the internal organization of the government.
4.1. How to Improve Commitment?
The first step in the search for a regulatory institution that minimizes transaction costs, and hence
interference with desirable regulatory outcomes, is to look for an approach that guarantees a commitment
by all parties involved to deliver on their responsibility. Since contracts between the firms, the regulators,
the executive and the legislative powers are likely to be quite incomplete, the specification of renegotiation
rules is quite critical and so are the systems of checks and balances as discussed below.
4.1.1 Renegotiation Rules and Separation of Powers
A well-known result among political scientists, and part of the recent folklore of the industrial
organization economists, is that the design of rules and processes at the renegotiation stage can improve
commitment.33 One particular choice is the separation of powers between different regulatory agencies.
This is not something that practitioners like to recommend. Indeed, the idea of unifying regulatory
responsibilities under a single umbrella institution is quite a common recommendation. Theoretical
research suggests that this may not be the best strategy when commitment capacity by the government to
the regulatory contract is limited or when renegotiation is a likely outcome of the reform process.
No Commitment at All: Under an extreme form of incompleteness, in which regulatory contracts
are limited to cover only the current period because of some type of constitutional constraint for instance,
Olsen and Torsvick (1993) show that it may be better to split up the control of the firm between two
regulatory agencies rather than leaving all the control rights in the hand of a single omnipotent regulator.34
Under separation, both regulators control the firm’s output. As a result of this complementarity between
their regulatory tools, their free-riding in providing incentives to the firm implies that there is an excessive
extraction of the firm’s rent in the second period of the relationship. The firm receives excessively lowpowered
incentives under separation. The main benefit is that this makes it less valuable for an efficient
firm to hide its type in the first period of the relationship. As a result, there is more information revelation
in earlier periods and total intertemporal welfare may increase. It should be clear however, that second
period benefit should be traded off against the cost of the first period competition between the different
33 This point has for instance been forcefully made by McCubbins, Noll and Weingast (1986) and (1989). See also Spiller
(1991) and Spiller and Levy (1995)
34 More exactly, their model deals with the issue of privatization and the consequences of the structural separation between
the owners of the firm and its regulator.
agencies. In sum, the overall benefits of separation are finally unclear when one assesses its impact from
an intertemporal point of view. The more general result is that there may exist an optimal degree of
decentralization in the regulatory charter, i.e., an optimal number of agencies in control of a given firm.
This number optimally trades off the cost of a larger static inefficiency coming from the coarser sharing
of regulatory rights against its dynamic benefits. 35
Renegotiation: Renegotiation improves ex post efficiency, however, the free-rider problem among
regulatory agencies at the time of renegotiating the contract makes this ex post improvement harder to
achieve.36 Separation improves the commitment ability of the government, but there is a twist.
Intertemporal welfare increases with separation only when the latter takes place at the renegotiation stage.
The optimal regulatory charter requires, therefore, a cooperative offer of the regulatory contracts when
regulation of a firm starts. It also requires the splitting up of the regulatory rights among various
agencies at the renegotiation stage. Generally, this type of optimal regulatory charter is unfeasible.37 The
more important result that practitioners may want to keep in mind is that the excessive splitting of the
responsibilities of regulatory agencies may be good for regulatory outcomes.
The solution may be to go for sequential moves of the regulatory decisions. For instance, the
federal government may act as a leader in renegotiating regulatory contracts. Then local governments
follow by offering complementary regulations. This type of institutional framework corresponds to what
is used in the Common Agricultural Policy in the European Community. In sum, the case for separation
as a way to improve commitment through the precise fine tuning of the regulatory process is an
ambiguous one. Nevertheless, we believe that it has strong relevance.
4.1.2. Checks and Balances
The recommendation in favor of separation also comes up in the discussion on checks and
balances. Sometimes the lack of commitment comes from the inability to contract ex ante on a variable
which provides a valuable signal on the firm’s performance, one that becomes only available at the interim
stage. For instance, the future profitability of a firm or the quality of its products cannot be assessed at
the time of the writing of the contract.
Tirole (1994) shows that creating different agencies with each having a specific objective function
improves commitment. Indeed, these agencies intervene differently depending on the realizations of this
signal. For instance, an agency concerned with consumer’s surplus intervenes following evidence that the
firm will charge a high price in the future. A Ministry of Finance, concerned with the profitability of the
firm will terminate projects that turns out to have intermediate cost overruns. Allocating the right to
intervene between “tough” and “soft” agencies according to the ex post realization of the signals
readjusts the ex ante incentives of the firm and improves intertemporal welfare. In a similar vein, Lewis
and Sappington (1991) analyze the intertemporal separation of powers between two short lived principals.
They take the allocation of tasks between both regulators as given but show that this separation, i.e., the
impossibility of regulators transferring money from one period to the other because they are under the
35 Olsen and Torsvick (1995)
36 Martimort (1997) shows why and how the separation of powers acts as a commitment at the renegotiation stage to have a
tough renegotiation among the different non-cooperating agencies controlling the regulatory agenda.
37 See Kahn (1988) for a discussion of the expansion of regulation over time.
control of different regulators, may be helpful in protecting the non-verifiable investment of the regulated
In both of these papers, a system of Checks and Balances is constructed in which different
agencies react differently to the information they may receive. The design of the objective function of the
regulators then becomes an important tool to control both the regulated firm and readjust the incentives of
the regulator to intervene. The discussion of the specific objectives is, however, something practitioners
seldom deal with explicitly.
4.1.3. Elected Political Principals and Independent Agencies
The above discussion focused only on the case of a benevolent principal unable to commit
himself to an intertemporal behavior. The splitting of the social welfare function among several agencies
was helpful but things become different in the case of a political principal. Elected political principals
cannot commit the future of society to a particular regulation. However, because they have a majority,
they often have the possibility of creating independent agencies and relinquishing some of their powers
and responsibilities to them. Of course, the threat of losing future elections makes them create
independent agencies that will still represent their constituencies even in case of political defeat.
Examples abound in regulation, but perhaps the most striking one belongs to the macroeconomics arena.
The choice of relinquishing the control of monetary policy to an independent central bank is a way for a
government in favor of a monetarist policy to commit future generations to it, even if future governments
might be more willing to allow higher levels of inflation.
We are not aware of any formal theoretical analysis along these lines in the framework of the
New Regulatory Economics. We nevertheless conjecture that relinquishing regulatory rights to
independent agencies may have social benefits, at least from an ex post point of view. More groups are
represented in the regulatory decisions that are taken ex post and future minorities are somewhat
protected. The ex post regulatory outcome is closer than that which would have been chosen by a
benevolent regulator. However, at the time of enacting a regulatory agency with similar objective
functions as its own, a non-benevolent government creates his own competition. As we have shown
earlier, this competition entails inefficiency. Competition between government bodies extracts too much
rent from the constituency it is supposed to protect. Relinquishing regulatory powers has also ex ante
4.1.4 Credibility of Commitment to a Regulatory Charter
An important problem of the models we have just surveyed is that they assume that the commitment to
creating an agency and incurring the administrative costs that it requires, is interpreted by the regulatory
players as a credible signal that the newly defined objectives and responsibilities of this agency are not
going to be overruled in the future. Institutional designs have to be credible to have any bite.
The credibility of the separation between various agencies can be enforced by standard
reputation-like arguments. As argued by the reputation literature, because they are long-term players,
agencies should be able to impose their most preferred regulatory outcome, i.e., should be able to render
their structural separation credible. This argument is a little loose and suffers from two underpinnings:
First, if reputations could be build by agencies to improve their collective commitment abilities, a single
agency could certainly also build a reputation for not renegotiating any regulatory contracts in the first
place; second, agencies themselves may be long-term players but bureaucrats often have limited tenures.
One should understand how collective reputations could establish among cohorts of regulators, especially
when they sometimes have different intrinsic preferences on the stringency of regulation.
4.2. How to Limit the Scope for Non-Benevolence?
Since non-benevolence is to be expected, it seems reasonable to look for ways to limit its scope.
One way of doing so is by focusing on the simplicity of the instruments of regulation. We already
discussed this possible remedy to the non-benevolence of political principals. Regulatory contracts could
be simplified to leave less leeway to the political decision-makers. This suggests that simple rules like the
constraints on the use of discretionary transfers and the short-term mandates of political principals can be
written as constitutional rules aimed at controlling these non-benevolent principals. More generally, the
use of instruments which are not sensitive to information, like for instance quotas, is a way to curb the
non-benevolence of political decision-makers, even if it creates other problems when contract
enforcements are a problem, as is typical in many reforming countries and in environments in which
information is limited.
Another way of dealing with non-benevolence, which is important in countries with an inefficient
public sector in need of reform, is to figure out what is the optimal speed of reform. Dewatripont and
Roland (1991) show that a benevolent principal willing to impose some layoffs on an industry is likely to
prefer to gradually impose these layoffs rather than impose them right away when its policy has to be
accepted by a majority of voters/workers within the sector. Under unanimity, it is a well known result
from the incentive literature that the optimal policy should be implemented with full commitment. 38
Under majority rule, the political principal prefers to impose its reform gradually, i.e., workers leave an
industry at different dates depending on their efficiency. By doing so, the principal learns some
information on the workers’ efficiency and can use it to credibly threaten to shift majorities in the future.
He can impose some concessions from some groups in the majority today by threatening them with being
in the minority tomorrow. Allowing the political principal to use such credible threats increases his
bargaining power. Reform processes are important when the political game imposes only majority voting.
4.3 Separation of regulatory powers as a way to prevent capture
Under integration of the regulatory powers, collusion is quite efficient and society must incur a
large cost of capture. For instance, bad projects are too often allowed and the budgetary burden on
society increases. Instead, asymmetric information between partially informed regulators and the firm
they regulate weakens their ability to extract advantages from the firm. Asymmetric information
implicitly increases the transaction costs of capture. The overall cost of capture borne by society is
diminished by the introduction of competition between asymmetrically informed bureaucrats.
The structural separation between different regulatory agencies can act as a commitment to
prevent regulatory capture by some interest groups. For instance, splitting the control rights on the firm’s
output between a Public Utility Commission and an Environmental Protection Agency helps prevent the
capture of the regulatory process by the producer. Several regulatory agencies with specific missions are
38 See Baron and Besanko (1992).
now each controlling only one dimension of the overall performance of the firm. Their incomplete
knowledge about the dimensions that they do not directly observe puts them at a disadvantage vis-à-vis a
the firm in the collusion game. Their individual ability to extract rents from the firm is then weakened.
Instead, if the regulators have been merged, they would be able to fully observe the performance of the
firm. Their demands for bribes could perfectly match the supply, i.e., the informational rent which is left
to the firm if the regulator behaves in a lenient way.
The argument for separation has to be considered much more carefully than what this quick
assessment suggests. The various interactions between separation and regulatory outcomes are discussed
next in some detail.
4.3.1 The efficiency and the distributional consequence of separation
The efficiency and distributional consequences of the structural separation are somewhat
ambiguous. On the one hand, the separation of powers can be a way for the interest group to escape the
control of its incompetent or corrupted regulators. This is the standard argument given by practitioners
on why a public utility commission, for instance, is better than sector specific regulators.39 Efficient firms
can get rents from every dimension of information observed by the regulators. Henceforth, they can better
fool them when regulators do not cooperate. On the other hand, the discretion of partially informed
regulators who may be corrupted may justify a reduction in their individual power which can be obtained
through separation. Because of the asymmetric information they face, regulators exert their individual
power in trying to capture benefits from the industry only when the latter is inefficient. Hence, to
diminish the discretion of these regulators, only the informational rent of these less efficient firms needs to
be reduced. Therefore, less efficient firms unambiguously lose from the structural separation of powers
and this is how regulatory outcomes improve under the separation of regulatory powers. Inefficient firms
suffer from the more bureaucratic rules followed by each single agency to reduce the cost of collusionproof.
Separation advocates for the use of more bureaucratic rules. The choice of communication
channels and the choice of more bureaucratic rules are two complementary tools used to improve the
design of the internal organization of the government. Again, structures and processes move hand in
4.3.2. Separation and the firm’s budget constraints
Separation has not only ambiguous distributional consequences, but it also has an ambiguous
allocative impact. Indeed, separation may both harden and soften the firm’s budget constraint depending
on the extent of the adverse selection problem that the government faces. Separation may indeed call for
less or more projects being accepted by the regulatory charter when several agencies are used to help the
decision making process. For small (resp. large) uncertainties, less (resp. more) projects are still allowed
under separation. In this case, rules are not too bureaucratic under integration because even if the single
regulator gets captured, the stake of collusion and the cost for society remain small. The single regulator
under integration keeps enough discretion to allow the pursuit of some bad projects. As we discussed
above, rules are more bureaucratic under separation and bad projects are more likely to be stopped by a
charter with several regulatory agencies.
39 Laffont and Martimort (1995) show how efficient firms may be able to play advantageously one regulator against the other.
This result is in sharp contrast with the one obtained by Dewatripont and Maskin (1995). These
authors show that some form of decentralization of the rights of refinancing projects hardens the firm’s
budget constraint.40 Similarly, Schleifer and Vishny (1993) discuss how corrupted regulators holding
complementary permits are able to extract an excessive amount of bribes from the firm leading to
insufficient entry on the market, i.e., hardening its budget constraint.41 In sum, the theoretical jury is still
out on the impact of separation on specific regulatory outcomes. Practitioners have generally already
decided against it in developing countries with limited institutional capacity.
4.3.3. Separation and communication problems
Until now, we have assumed that communications between each regulator and the principal were
taking place simultaneously. The benefits of separation also exist when we allow one regulator to observe
other regulators’ information before deciding on the degree of collusion with the firms regulated. This
hierarchical organization captures the often observed asymmetry in the role of different regulators
intervening at different stages of the regulatory process. For instance, the procedures for merger control
differ across countries but in general require two stages to be completed sequentially; notification and
investigation. According to Nutall and Seabright (1993), who compare the different organizations of
regulations, in some countries, such as the United Kingdom, two distinct bodies are in charge with each of
these tasks, respectively the Director General for Fair Trading and The Monopolies and Merger
Commissions. In the European Economic Commission, a unique body, the Merger Task Force performs
both activities. Even if one could agree with these authors and recognize that a concentration of the tasks
speeds up the regulatory process, it may also involve greater costs in terms of capture. Laffont and
Martimort (1995) show that a sequential separation between regulators still helps in fighting capture.
However, the gains of reducing the cost of collusion-proof constraints are only obtained vis-à-vis the first
intervening agency, which is only partially informed and therefore at a disadvantage in the bribery game
with the regulated firm. The sequential separation between the regulators comes with an asymmetry
between their real authority and the level of bureaucratic procedures they follow. Regulators moving first
follow more bureaucratic procedures and are entitled to less formal authority than regulators moving last.
Again, structure and processes matter on the regulatory outcome.
4.3.4. The Life Cycle of Regulatory Agencies and Transaction Costs
40 The main focus of Dewatripont and Maskin (1995) is the organization of the banking sector. They show that a banking
sector in which lenders are constrained in the size of the loans they offer can be used as an effective incentive device. When
lenders have a limited ability to finance projects, a venture needs to look for refinancing by other lenders who are at an
informational disadvantage vis-a-vis initial lender. As a result, the cost of capital increases and this hardens the ex ante
incentives of the firm to complete projects earlier. Obviously, the logic of their argument also applies to regulatory agencies
with limited budgets.
41 Their approach is cast in terms of exogenous stakes of collusion and exogenous bribery strategies. We already address in
Section 3.2 this excessive rent-extraction which arises in a multiprincipal models when the latter control only
complementarity activities of the agent. The weakness of this approach is that it overestimates the problem of control of the
bureaucracy by modeling no control at all. Because, it has no normative underpinnings, there is no place for a welfare
comparison between regulatory institutions.
Multiplying the number of regulators to induce inefficiencies in the bribery game is also an argument that the Public Choice
literature has put forward (Miller, Shughart and Tollison (1984)). This approach suffers nevertheless from the same
weaknesses as Schleifer and Vishny (1993).
Institutions affect the efficiency of the collusion game between the interest groups and the
regulatory agencies. The key to getting a richer vision of the capture transaction is to see it as a selfenforceable
side-contract between the regulator and the firm it regulates. This is the natural theoretical
framework to analyze this collusion, but it is also the most realistic way, the formalism of regulatory
processes and rules are often dominated by less formal processes and rules. 42 This is because there is no
court of justice to enforce these illegal agreements and the party to the contract must rely on “words of
honor” to fulfill their agreements. The common view argues that the repetition of the regulatory game is
good because it allows for less opportunistic behaviors by the regulator. 43 This is in fact not the case in a
repeated game environment in which the regulator is always the same. The expectations of the regulator
on future advantages deriving from friendly behavior toward the industry, and the expectation of the
industry on lenient behavior from the regulator are the glue that allows these illegal agreements to be selfenforcing.
These informational rents, which will be under the control of the regulator in the future, create
the stakes for starting collusive relationships immediately. The higher the stakes in the future, the higher
the cost of capture today. Since more and more opportunities for collusive behavior may emerge over
time, the optimal regulatory charter calls for smaller and smaller regulatory stakes as time goes on.
Therefore, one should also impose more and more bureaucratic rules as agencies grow older. In other
words, the bureaucratization of the government intervention is an unavoidable outcome.45 Agencies
follow a so-called “Life Cycle”. They begin by behaving in the public interest and then become more and
more under the influence of interest groups. They then become vested in the status quo and overly
This Life Cycle view of regulatory agencies modeling as capture also gives firmer theoretical
foundations to the Laffont and Tirole (1993, Chapter 13) model of capture, which instead assumes
enforceable side-contracts, although the long-run outcome of the regulation is the same as that obtained
with a static model of capture.46 The key novelty of the analysis is to derive the efficiency of the sidecontracts
from the institutional structure. These side-contracts, because they are not enforceable and rely
on a dynamic process to establish, suffer from some transaction costs. These transaction costs of capture
change with simple institutional constraints: they decrease with the amount of information that the
regulator gets on the firm; they decrease with the discount factor of the principal, i.e., with the
discrepancy between the time horizon of the political principal and that of the firm; and, they decrease
also with the discount factor of the bureaucracy. Changing these parameters increases the value of the
continuation of capture for the regulator and the firm. The capture transaction is now easier to be
The consequences of this endogenization of the transaction cost of capture for institution design
are immediate. First, arm’s length regulation, in which the political principal exerts the regulatory rights
42 See Tirole (1992) and Martimort (1997) for models along these lines.
43 Salant (1992), Salant and Woroch (1993) and Gilbert and Newbery (1993)
44 See Fudenberg and Maskin (1986) for a general proof of the Folk Theorem in repeated games.
45 As suggested by numerous political scientists: Lierson (1949), Bernstein (1955), Huntington (1966) and Downs (1967)
among others) but also by sociologists and finally by practitioners (Kahn (1988)
46 Martimort (1997)
itself and where there is little room for an active bureaucracy, could be preferred to a more day-to-day
relationship with the firm, which occurs under regulation. The recent move away from regulation and
towards antitrust policies observed in European countries can be provided with a rationale along these
lines. Indeed, antitrust policies make little use of ex ante monitoring devices and correspond in fact to the
use of ex post monitoring controls. Courts, and often poorly informed judges, are given large degrees of
discretion in an absolutely non-predictable way in order to break any scope for capture by the industry.
The reinforcement of antitrust agencies throughout Latin America for instance could be seen as a
significant potential improvement of the potential efficiency of regulatory outcomes for the same reason.
The Life-Cycle view of regulatory agencies also shows that the composition of the agency itself
matters. As discussed by Wilson (1989), we can isolate three different types of regulators; the so-called
“careerists,” who are more likely to move to the industry they regulate once they have finished their
tenure in the civil service; the “professionals,” who are former industry managers willing to keep some
influence on the regulatory process; and lastly the “politicians,” who see their civil service as a first step
before taking a political position. All these types respond to different implicit incentives. Clearly, the two
first groups are more likely to give a high continuation value on the fact of colluding with the firm. This
suggests that shifting the balance of powers towards the “politicians” within the regulatory agency itself
is optimal, which is not quite consistent with the standard recommendations given by practitioners, as
suggested by the overview prepared by Smith (1996). Alternatively, playing on the tenures of the
bureaucrats helps to fight capture. Short-term involvement in the civil service should nevertheless be
traded off with the gain of longer lasting relationships like a better learning of the industry conditions.
4.3.5. Limited Commitment and the Degree of Honesty of the Policymakers
Simple “constitutions” which limit the intertemporal mandates of a government can be used as
constraints to prevent capture. 47 At the time of designing the constitution, the “founding fathers” have to
decide whether a government should be able to bind the future of society to a given policy or whether they
should be given the regulatory rights only for a shorter periods. The cost of commitment is that a badly
composed government could extract more rent from its ability to commit. Instead, short-term mandates
are such that a badly composed government can be replaced with some probability by an honest one that
is more willing to prevent capture with the industry. Political competition acts as a potential threat of
regulatory capture. The cost of short-term contracts is, instead, that they are prone to the standard hold
up problem.48 Firms underinvest if they are ex post expropriated from their investment at the time of
change in the regulatory policy.
When the political decision maker is, in all likelihood, dishonest, the optimal complete
intertemporal contract or constitution calls for collusion-proof regulatory mechanisms. Instead, when
dishonesty is less likely, letting collusion happen is optimal since it leads to an avoidance of paying the
costs of an increase in the firm’s budget to fight collusion. This outcome can also be implemented as a
47 Laffont and Tirole (1993), Chapter 16
48 See Williamson (1975) et Grossman and Hart (1986).
simple constitution which chooses ex ante between long-term and short-term contracting. In particular, a
more efficient technology of collusion, i.e., a lower transaction cost of capture, calls for the choice of
short-term contracts. Indeed, the dead-weight loss due to capture is less important under short-term
contracting. This result is reinforced by Olsen and Torsvick (1995), who show that the benefits of
offering a sequence of two short-term contracts with respect to writing a long-term contract increases with
the transaction costs of side-contracting. A more efficient ex post capture induces a tougher reaction of
the political principal and the regulatory stake is strongly reduced. In other words, there is an implicit
competition between the political principal for the second period and the regulator when they make deals
with the interest group. An increase in the efficiency of side-contracting hardens this competition. By
playing the bureaucracy against a political principal who is unable to commit helps to improve welfare.
The increase in the efficiency of the bureaucracy becomes a commitment device.49 Collusion-proofness
calls for little discretion and this is good when only short-term contracts can be written.
4.3.4. Delegation and Decentralization
One largely unanswered question in the theory of government remains its optimal degree of
decentralization. The shrinking share of the government sector in all western economies, the failures of
the socialist economies, and the discussion of the costs and benefits of federalism are all evidence that this
debate is very current. 50
The Folklore of the Public Economics literature argues that decentralization is good because it
allows local powers entitled with regulatory rights to use their local information to improve the provision
of regulation, redistribution, or the production of public good at the local level. The standard argument is
that these benefits of decentralization must be traded off against the costs coming from the lack of
coordination in the regulatory policies of the competing states. Externalities arise from this decentralized
exercise of regulatory rights.51
This argument is at best incomplete because it implicitly assumes that local principals are unable
to communicate the information they learned at the local level to the upper level of the regulatory
hierarchy. If such communications were available, a grand and centralized mechanism would be enough
to coordinate all jurisdictions. Moreover, under seemingly innocuous assumptions like risk-neutrality,
unrelated local shocks and acceptance of the federal contracts at the interim stage, i.e., after that local
governments have learned their own information, there are decentralized ways of implementing the
optimal coordinated regulatory policy. With a convenient set of budget allocations and grants at the local
level, the Federal level can achieve the efficient policy by letting local jurisdictions have the formal
authority for implementing local regulation.52
49 Sappington (1987) builds a related model.
50 See Stiglitz (1994, Chapter 10) for a discussion of the centralization versus decentralization problems in this context.
51 We do not aim here at discussing the huge literature which, following Tiebout (1955), has analyzed these externalities.
Few of these papers have really taken the agency perspective which is necessary to an institutional analysis. Klibanoff and
Morduch (1994) is a valuable exception. They provide a model of decentralization based on informational constraints and
discuss the meaning of residual regulatory rights in this context. Additional ideas are discussed in Estache, A., ed., “
Decentralizing Infrastructure, Advantages and Limitations”, World Bank Discussion Paper, 290.
52 Baron and Besanko (1992) and Moockherjee and Reichelstein (1996) show this result in the context of the theory of the
There are three possible ways to prove the optimality of decentralized structures. The first argues
that the main benefit of decentralization comes from the ability of the local governments to collude with
specific interest groups at the local levels.53 This collusion is indeed socially optimal because it allows
the overall contractual arrangement to use the shared information of the local behavior to improve on the
centralized arrangement. Capture is not a curse for society, but on the contrary, it allows regulatory
contracts to be completed.54 A second solution is to recognize that implicitly, communication between
the local and the centralized governments is assumed to be limited because, for instance, the so-called
revelation mechanisms are not available. 55 Contracts are then incomplete. The mere existence of a
communication constraint creates the scope for collusion at the local level between the regulated firm and
its regulator. The structural asymmetry that decentralization introduces between these two layers of the
hierarchy helps the central level use the local regulators to complete their regulatory contract. A third,
less normative, perspective about the optimal organization argues that delegation may also help in the
case of non-benevolent political principals elected through voting procedures. The basic idea is that the
optimal organization trades off the incentive costs of decentralization (modeled as a moral hazard problem
between the local regulators and the centralized one) and the benefits of decentralization, which is a better
representation of the preferences of the local median voter by the local elected principals.
5. Lessons for a pragmatic approach to the design of regulatory institutions
Table 1 summarizes the main discussions covered in this paper. It suggests that there is a
reasonable degree of consensus between practitioners and theorists on many of the issues covered here. In
a nutshell, everyone seems to agree that utility regulation has to promote (static and dynamic) efficiency
while it protects consumers from potential monopolist abuses, and investors and operators from political
influence. Everybody also agrees that there is a trade-off between the credibility of the regulatory
commitments and the flexibility required to rebalance, as needed, the interests of the various actors. Some
degree of flexibility is desirable but the track record of governments in their use of flexibility is generally
perceived as being so problematic that the rules built in the various privatization instruments are designed
to limit this flexibility.
Since some degree of discretion is always available since regulatory contracts cannot foresee all
future occurrences, safeguards are needed. One of the key components of these safeguard mechanisms
analyzed, in some detail, by practitioners is the specific design of regulatory institutions. One of the key
points made in this paper is that these safeguards sometimes reflect and sometimes imply transaction costs
which influence or should influence the optimal rent-efficiency trade-off in ways which are sometimes
ignored by practitioners.
The first set of generic issues practitioners usually cover on institutional design is the importance
of the independence, autonomy, and accountability of the regulatory institutions for the sustainability of
the reforms of regulated sectors. Most of the literature on transaction costs due to government failures
53 Caillaud, Jullien and Picard (1996)
54 See Itoh (1993) for this point in the framework of the theory of the firm.
55 See Maskin (1976). Also, Laffont and Martimort (1995) takes explicitly this route and model collusion and limits of
covered here would agree with the need to meet these three criteria. It would, however, emphasize that the
way in which the criteria are met is determined by the way in which the transaction costs are minimized,
and this in turn drives the desirable design of the regulatory framework. The time dimension, for instance,
is crucial. In practice, it means that if there are commitment problems, short term institutional contracts
between the various players are more likely to ensure independence and autonomy. This has an impact on
the duration of the nomination of the regulators. Short term contract may be better, yet typically,
contracts for regulators are between 4 and 8 years and often with possibilities of renewal.
But the empirical debate on the design of the regulators’ job goes further and this is a potential
source of tension. According to one of the schools of thought discussed above, it would seem that what
practitioners typically recommend – i.e. that ensuring that the regulators are appointed on the basis of
professional rather political criteria – may not be the optimal strategy to minimize capture since the
professional experts are likely to come from the sector they are supposed to regulate and are likely to
return to it sooner or later. This is the case in most developing countries for instance. Most practitioners
would reply that the idea of electing regulators may be just as dangerous in developing countries. Unless
the democratic process is well oiled and has the required transparency and accountability, elected
regulators are unlikely to be much more independent than professional regulators; they will simply
represent different interests. What is interesting to note however is that while practitioners and theorists
emphasize different sources of capture, both agree that one way of dealing with this risk of capture is to
ensure that the selection process involves both the executive and the legislative branches.
Table 1: Summary of the main approaches to minimize transaction costs in the design of regulatory frameworks
Impact on regulatory outcome if not taken into account How to minimize transaction costs
Lack of commitment of government
and need for renegotiation
·with more info in 2nd period regulatory want to improve incentive
·this reduces incentive for efficiency in 1st period
·“overincentivate” in period 1
·maintain flexibility to renegotiate to avoid overbinding commitments
to private operators
Multiprincipal Nature of
·allocative inefficiency: because of horizontal coordination problems, more or less
·sub-optimal degree of decentralization and vertical coordination problems: cuts
efficiency rent to firms and move towards cost plus regime
·too low incentives build in contract when multiple regulators cover
complementary activities because of overregulation
·too much incentives build in contract when multiple regulators cover substitutes
activities because of regulatory competition
·sequentiality of regulatory decisions matter:
·improve information and monitoring (formal yardstick competition
will be a useful tool)
·reduce number of players
·let regulators share information to soften their competition
·let regulators share information to soften their competition
·give a veto right to head of state or institutions to which regulators are
Discretion of Political Principal · structure driven direction of incentives because they drive the distribution of
power of interest groups
· politician favor median voter rather than majority of voters
·how to fix structures? separation?
·fix process by specifying decision making processes (including voting
mechanism to ensure proper representation in decision making
suboptimal contracts (average cost pricing or quota) can be useful ways
of tying the hands of politicians
Discretion of Regulator ·internal influences: wages and promotions can be used to alter rent distribution;
increase internal incentive to perform to reduce external influence and risk of
·conflict between legislative and regulators; need to reduce rent to cut risks of
·regulatory dynamics: side payments
·increase accountability (disseminate information through any which
way it is possible; including the media but this requires also educating
·cut rent and move towards cost- plus
·fix structures to reduce risks
Autonomy covers various issues. Practitioners will generally argue that agencies need to
have access to their own funding sources. Relying on budgetary transfers decided by politicians
is often viewed as a threat to the independence of the regulators since an easy way to reduce the
effectiveness of a regulator would be to cut its budgetary allocation. Levies on the regulated
firms or the consumers of the regulated services are the most common alternatives, and can be
viewed as user fees to be paid for the protection services provided by the regulators. To some
extent, the regulatory fee approach also increases the accountability of the regulators since there
is a more direct link between the source of their financing and what is expected from them. 56
But autonomy has to be more than just financial. It should also mean that the regulator
can recruit its own staff. The achievement of staffing autonomy will often require an exemption
from civil service salary and recruitment rules. It may also imply that these agencies have to be
able to recruit external consultants when the required skills are not available locally to address
specific needs. Autonomy in monitoring of compliance and enforcement deserves to be
highlighted because it requires that specific instruments be assigned to the regulators. This match
between instruments and institutions is also a recurring theme addressed by the theoretical
literature, which does not provide closure, so there is little policy advise that be drawn beyond
what practitioners have to say. To be effective in his role, the regulator must be able to impose
penalties according to clearly defined rules. This is consistent with the emphasis on simple
transparent rules that emerged from the literature reviewed in the paper. The other theme
addressed here is that there is an ideal sequence in the decision-making process that depends on
the distribution of information among the actors
Both practitioners and theoreticians agree that accountability requires transparency in the
decision making process, which is unfortunately too often counterintuitive to many bureaucrats. It
also requires an operating environment subject to simple and clear procedural rules, including
stipulated deadlines for reaching decisions, detailed justifications of decisions, non-political
reviews of decisions, opportunities for all concerned parties to be heard through public hearings
(and hence greater interactions with consumer defense groups), and venues for appeal and
provisions permitting removal of regulators in case of proven misbehavior. The practical
challenge, however, is no longer the staffing. The challenge is to ensure that the information
generated by this process is relevant to allowing the required accountability. This is,
unfortunately, not the case in many countries since process oriented regulators tend to deal with
details that have little to do with the distribution of the rent generated by the private or public
monopolies providing utilities services. They deal with the formal regulatory issues rather than
the real regulatory issues. In this case, transparency requirements become an alibi rather than an
instrument that can be used by regulators.
Finally, practitioners tend to discuss their views on the optimal number of agencies. For
all or most infrastructure sectors, they tend to recommend the creation of multisectoral agencies,
i.e. a single regulatory agency as in the case of state-level regulators in the US, Canada and
Australia, and national regulators in Costa Rica and Jamaica, rather than a sector or industry
specific agency. This is one of the recommendations that appears to be at odds with what theory
suggests. The case for some degree of separation of regulatory roles is made strongly by many of
56 Also, there are potential problems here too since regulation is a public good and this will probably lead to some
degree of free riding.
the papers surveyed here (but not too much since this can lead to unclear and complex
overlapping of divisions of responsibilities). This apparent inconsistency can be reconciled in two
ways. First, the practitioners’ concern with the need to share regulatory resources (regulatory
economists and lawyers, etc.) to deal with the limited regulatory capacity is generally not
addressed by the theorists surveyed in this paper since their focus in on the US or the UK,
probably the largest producers of “regulatory skills”. 57 Second, the fact that some degree of
overlapping of responsibilities is unavoidable (e.g. the regulation of environmental issues
typically involves multiple government institutions) guarantees that the gains achieved through
some degree of separation are achieved.
Considered jointly, the impact of transaction costs resulting from the difficulties involved
in organizing governments often imply that less rent and less incentive needs to be built into the
regulatory frameworks if transaction costs in general and collusion problems in particular are to
minimized. This is an important conclusion since it rejoins the recommendation from theory on
the optimal strategy to deal with excessively high risk levels in privatization projects. There
seems to be some truth that there can be too much of a good thing when considering the
appropriate level of incentives to build into regulatory regimes.
57 Exceptions include Spiller and Levy and Spiller (1993), Guasch and Spiller (1997), and Kahn and Kessides
Alchian, A. and Demsetz, 1972, “Production, Information Costs and Economic Organization”, American
Economic Review, 62: 777-795
Aghion, P. and J. Tirole, 1997, “Formal and Real Authority in Organizations”, Journal of Political
Economy, 105: 1-29
Atkinson, C., and J. Stiglitz, 1980, Lectures in Public Economics, Mc Graw and Hill, London.
Baron D., and R. Myerson, 1982, Regulating a Monopolist with Unknown Costs, Econometrica, 50: 911-
Baron, D., 1985, “Non-Cooperative Regulation of a Non-Localized Externality,” Rand Journal of
Economics, 16: 269-282.
Baron, D. 1989, “Regulation and Legislative Choice”, Rand Journal of Economics, 20: 467-477.
Baron D., and D. Besanko, 1992, “Information, Control, and Organizational Structure”, Journal of
Economics and Management Strategy, 1: 237-276.
Baron, D., 1995, “The Economics and Politics of Regulation: Perspectives, Agencies and Approaches”, in
Banks and Hanushek, Modern Political Economy, 1-60.
Becker, G., 1983, “A Theory of Competition among Pressure Groups for Political Influence,” Quarterly
Journal of Economics, 98: 371-400.
Bernheim, D. and M. Whinston, 1986, “Common Agency,” Econometrica, 54: 923-943.
Bernheim, D. and M. Whinston, 1987, “Menu Auctions, Resource Allocations, andEconomic Influence,”
Quarterly Journal of Economics, 101: 1-31.
Bernstein, M.,1955, Regulatory Business by Independent Commissions. Princeton, University Press,
Buchanan, J.,] 1980, “Rent-Seeking under External Diseconomies,” in Towards a Theory of the Rent-
Seeking Society, ed. J. Buchanan, R. Tollison and G. Tullock. College Station: Texas A end M Press.
Coase, R., 1937, “The Nature of the Firm,” Economica, 4: 386-405.
Congleton R., 1984, Committees and Rent-Seeking Effort, Journal of Public Economics, 25: 197-209.
Cremer, J., A. Estache and P. Seabright 1996,“Decentralizing Public Services: what can we learn from the
Theory of the Firm?”, Revue d’Economie Politique, 106 (1).
Dasgupta, P., P. Hammond, and E. Maskin, 1979, “The Implementation of Social Choice Rules,” Review
of Economic Studies, 46: 185-216.
Dewatripont, M.,1989,“Renegotiation and Information Revelation over Time,” Quarterly Journal of
Economy, 103: 589-620.
Dewatripont M., and E. Maskin, 1995, Credit and Efficiency in Centralized and Decentralized
Economies, Review of Economic Studies, 62: 541-555.
Dewatripont, M. and J. Tirole, 1994, “A Theory of Debt and Equity: Diversity of Securities and Manager-
Shareholder Congruence,” Quarterly Journal of Economics, 109: 1027-1054.
Dixit, A., 1996, The Making of Economic Policy: A Transaction Cost Politics Perspective,MIT Press,
Downs, A., 1967, Inside Bureaucracy, Little Brown and Company, Boston.
Estache, A. (1997), “Designing Regulatory Institutions for Infrastructure -Lessons from Argentina”,
Viewpoint No114, Private Sector Development Department, The World Bank Group.
Estache, A., ed., 1995, ”Decentralizing Infrastructure, Advantages and Limitations”, World Bank
Discussion Paper, 290.
Ferejohn, J., 1986, Incumbent Performance and Electoral Control”, Public Choice, 86, 5-26
Fudenberg, D. and D. Levine, 1989, “Reputation and Equilibrium Selection in Games with a Patient
Player,” Econometrica, 57: 759-778.
Fudenberg, D., D. Levine and E. Maskin, 1994, “The Folk Theorem with Imperfect Public Information,”
Econometrica, 62: 997-1039.
Greenwald, and J. Stiglitz, 1986, “Externalities in Economies with Imperfect Information and Incomplete
Markets,” Quarterly Journal of Economics, 101: 229-264.
Grossman, G. and E. Helpman, 1994, “Protection for Sale,” American Economic Review 84: 833-850.
Grossman, G. and E. Helpman, 1996, “ Electoral Competition and Special Interest Politics,” Review of
Economic Studies 63: 265-286.
Grossman, S. and O. Hart, 1986, “The Costs and Benefits of Ownership: A Theory of Lateral and
Vertical Integration,” Journal of Political Economy, 94: 691-719.
Guasch, J.L and P. Spiller, 1997, Managing the Regulatory Process: Design, Concepts, Issues and the
Latin American and Caribbean Story”, mimeo
Hart, O. and J. Tirole, 1988, “Contract Renegotiation and Coasian Dynamics “, Review of Economic
Studies, 45: 509-540.
Helpman (1997), “Politics and Trade Policy”, in Kreps, eds., Advances in Economics and Economietrics:
Theory and Applications”, 19-45
Holmstrom, B, 1982, “Moral Hazard in Teams”, Bell Journal of Economics, 10: 74-91
Holmstrom, B. and J. Tirole,] 1988, “,” in Handbook of Industrial Organization, vol 2, Chapter : -.
Homlstrom, B. and P. Milgrom, 1991, “Multitask Principal Agent Analyses:Incentive Contracts, Asset
Ownership, and Job Design,” Journal of Law, Economic and Organization, 7: 24-52.
Huntington, S., 1966, “The Marasmus of the ICC: The Commission, the Railroad and the Public
Interests” in P. Woll, ed., Public Administration Policy: Selected Essays. Harper and Row, New-York.
Itoh, H., 1993, “Coalitions, Incentives, and Risk Sharing,” Journal of Economic Theory, 60: 410-427.
Kahn, A., 1988, The Economics of Regulation: Principles and Institutions, 2nd edition, MIT Press,
Kiewert, D. and M. McCubbins, 1991, The Logic of Delegation: Congressional Parties and the
Appropriations Process, Chicago University Press, Chicago.
Kofman F., and J. Lawarree 1993, Collusion in Hierarchical Agency, Econometrica, 61: 629-656.
Kydland, F. and E. Prescott, 1977, “Rules rather than Discretion: The Inconsistency of Optimal Plans”,
Journal of Political Economy, 85, 3.
Laffont J.J., 1994, The New Regulatory Economics: Ten Years After, Econometrica, 62: 507-538.
Laffont, J.J. and D. Martimort, 1995, “Separation of Regulators against Collusive Behavior,” mimeo
Laffont, J.J., 1995, “Industrial Policy and Politics”, International Journal of Industrial Organization, 14: 1-
Laffont, J.J. and D. Martimort, 1998, “Collusion and Delegation,” forthcoming Rand Journal of
Laffont, J.J. and J. Tirole, 1990, “Adverse Selection and Renegotiation in Procurement,” Review of
Economic Studies, 75: 597-626.
Laffont J.J., and J. Tirole, 1991, The Politics of Government Decision Making: A Theory of Regulatory
Capture, Quarterly Journal of Economics, 107: 1089-1127.
Laffont J.J. and J. Tirole, 1992,“Should Government Commit?,” European Economic Review, 36: 345-
Laffont J.J. and J. Tirole, 1993, A Theory of Incentives in Procurement and Regulation, MIT Press,
Landes W., and R. Posner, 1975, The Independent Judiciary in an Interest-Group Perspectives, Journal of
Law and Economics, 18: 875-901.
Landis J., 1960, Report on Regulatory Agencies to the President Elected, U.S. Senate, Committee on the
Judiciary, Committee Print.
Lewis, T. and D. Sappington, 1991,“Oversight of Long–Term Investment by Short–Lived Regulators,”
International Economic Review, 32: 579-600.
Lierson, A., 1946, “Interest Groups in Administration,” in F. Morstein Marx, ed., Elements of Public
Administration. Prentice Hall, New-York.
Martimort, D.,1992, “Multi-Principaux avec Anti-Selection,” Annales d’Economie et de Statistiques, 28:
Martimort, D., 1996a, “Exclusive Dealing, Common Agency and Multiprincipal Incentive Theory”, Rand
Journal of Economics, 27: 1-31.
Martimort, D., 1996b, “The Multiprincipal Nature of the Government,” European Economic Review, 40:
Martimort, D.1997a, “A Theory of Collusive Behavior”, Scandinavian Journal of Economics, 99, 555-
Martimort, D. 1997b, “The Life Cycle of Regulatory Agencies: Dynamic Capture and Transaction Costs”,
Martimort, D. and L. Stole, 1997, “A Note on the Revelation Principle in Common Agency Games,”
Martimort, D., 1998, “Renegotitation Design with Multiple Regulators”, mimeo IDEI, Toulouse.
McAfee, P. and J. McMillan, 1995, “Organizational Diseconomies of Scope,” Journal of Economics and
Management Strategy 4: 399-426.
McCubbins, R., R. Noll and B. Weingast, 1987, “Administrative Procedures as Instruments of Political
Control,” Journal of Law, Economics, and Organization, 3: 243-277.
McCubbins, R., R. Noll and B. Weingast, 1989, “Structure and Process, Politics and Policy:
Administration Arrangements and the Political Control of Agencies,” Virginia Law Review, 75: 431-482.
Melumad, N., D. Mookherjee, and S. Reichelstein, 1995, “Hierarchical Decentralization of Incentive
Contracts,” Rand Journal of Economics, 26: 654-692.
Meyrson, R., 1997, “Economic Analysis of Political Institutions”, in Kreps, ed., Advances in Economics
and Economietrics: Theory and Application, Seventh World Congress, 46-65.
Miller J., W. Shughart and R. Tollison, 1984, “A Note on Centralized Regulatory Review,” Public
Choice, 43: 83-88.
Mirrlees J., 1971, “An Exploration in the Theory of Optimum Income Taxation,”
Review of Economic Studies, 38: 175-208.
Moe, T., 1984, “The New Economics of Organization,” American Journal of Political Science, 28: 739-
Moe T., 1986, “Interests, Institutions, and Positive Theory: The Policies of the NLRB,” Studies of
American Political Department.
Myerson, R., 1979, “Incentive Compatibility and the Bargaining Problem,” Econometrica 47: 61-73.
Neven D., R. Nutall and P. Seabright, 1993,Regulatory Capture and The Design of European Merger
Policy, in Merger in Daylight, eds.; Neven D., R. Nutall and P. Seabright, CEPR Press, London.
Niskanen, W., 1971, Bureaucracy and Representative Government, Aldine-Atherton, New-York.
Olsen, T. and G. Torsvick, 1993a, “The Ratchet Effect in Common Agency: Implications for Privatization
and Regulation,” Journal of Law, Economic and Organization, 9: 136-158.
Olsen, T. and G. Torsvick, 1993b, “Collusion and Ratcheting in Hierarchies,” mimeo, Norvegian
Research Center in Organization and Management.
Olsen, T. and G. Torsvick, 1995, “Intertemporal Common Agency and Organizational Design: How
much Centralization,” European Economic Review, 39: 1405-1428.
Olson, M., 1965, The Logic of Collective Action: A Theory of Interest Groups in Public Goods, Harvard
University Press, Cambridge.
Olson, M., 1982, The Rise and Decline of Nations: Economic Growth, Stagflation and Social Rigidities,
Yale University Press, New Haven.
Peltzman, S., 1976, “Towards a More General Theory of Regulation,” Journal of Law and Economics, 19:
Posner, R., 1976, “Theories of Economic Regulation,” Bell Journal of Economics, 5: 335-358.
Qian, Y. and G. Roland, 1994, “Regional Decentralization and the Soft Budget Constraint: The Case of
China,” mimeo ULB Bruxelles.
Salant, D., 1995, “Behind the Revolving Door: A New View of Public Utility Regulation,” Rand Journal
of Economics, 26: 362-377.
Salant, D. and G. Woroch, 1992, “Trigger Price Regulation,” Rand Journal of Economics, 23: 29-51.
Shleifer, A. and R. Vishny, 1993, “Corruption,” Quarterly Journal of Economics, 109: 599-617.
Smith, W. (1996), “Utility Regulators: Creating Agencies in Reforming and Developing Countries”, paper
presented to the International Forum for Utility Regulators, Oxford, England, June.
Spiller, P., 1991, “Politician, Interest Groups, and Regulators: A Multiple-Principals Agency Theory of
Regulation, or “Let Them Be Bribed”, Journal of Law and Economics 33: 65-101.
Levy, B. and P. Spiller (1993), “Regulation, Institutions and Commitment in Telecommunications: A
Comparative Analysis of Five Country Studies”, Annual Bank Conference on Development Economics, May 3-4,
The World Bank.
Stigler, G., 1971, “The Theory of Economic Regulation,” Bell Journal of Economics, 2: 3-21.
Stole, L., 1990, “Mechanism Design under Common Agency,” mimeo, University of Chicago.
Tirole, J., 1986, “Hierarchies and Bureaucracies: On the Role of Collusion in Organizations,” Journal of
Law, Economic and Organization, 2: 181-214.
Tirole, J., 1992, “Collusion and the Theory of Organizations,” in Advances in Economic Theory, vol. 2 ,
ed. J.J. Laffont, 151-206. Cambridge University Press.
Tirole, J., 1994, “The Internal Organization of Government,” Oxford Economic Papers 46: 1-29.
Williamson, O., 1975, Markets and Hierarchies, Free Press.
Williamson, O., 1985, The Economic Institution of Capitalism.
Wilson, J., 1989, Bureaucracy: What Government Agencies Do and Why They Do It., Basic Books, New