regulatory capture: a review - berkeley haas

203Oxford Review of Economic Policy vol. 22 no. 2 2006© The Author (2006). Published by Oxford University Press. All rights reserved.

REGULATORY CAPTURE: A REVIEW

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 22, NO. 2DOI: 10.1093/oxrep/grj013

ERNESTO DAL BÓHaas School of Business and Travers Department of Political Science, University of California,Berkeley1

This article reviews both the theoretical and empirical literatures on regulatory capture. The scope is broad,but utility regulation is emphasized. I begin by describing the Stigler–Peltzman approach to the economics ofregulation. I then open the black box of influence and regulatory discretion using a three-tier hierarchicalagency model under asymmetric information (in the spirit of Laffont and Tirole, 1993). I discuss alternativemodelling approaches with a view to a richer set of positive predictions, including models of common agency,revolving doors, informational lobbying, coercive pressure, and influence over committees. I discuss empiricalwork involving capture and regulatory outcomes. I also review evidence on the revolving-door phenomenonand on the impact that different methods for selecting regulators appear to have on regulatory outcomes. Thelast section contains open questions for future research.

1 E-mail address: [email protected] am grateful to Dieter Helm, two anonymous referees, Clare Leaver, Martín Rossi, and Justin Tumlinson for useful comments

and suggestions.

I. INTRODUCTION

The economics profession has devoted substantialeffort to answering two broad questions. First, whydo we observe state intervention in economic life?And second, what are the politico-economic proc-esses that shape state intervention? In a broadsense, regulation encompasses all forms of stateintervention in the economy, although a more nar-row definition concerns the control of natural mo-nopolies. Thus, the term ‘regulatory capture’ alsoreceives both a broad and a narrow interpretation.According to the broad interpretation, regulatory

capture is the process through which special inter-ests affect state intervention in any of its forms,which can include areas as diverse as the setting oftaxes, the choice of foreign or monetary policy, orthe legislation affecting R&D. According to thenarrow interpretation, regulatory capture is specifi-cally the process through which regulated monopo-lies end up manipulating the state agencies that aresupposed to control them.

Most of the literature that is explicitly concernedwith regulatory capture has been developed in thecontext of utility regulation, although a literature on

204

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 22, NO. 2

political influence has grown alongside it. The domi-nant normative argument for the regulation of utili-ties posits that natural monopoly situations call for asole firm. Then regulation is needed to prevent thatfirm from exploiting its market power (Demsetz(1968) challenged this view; see Berg and Tschirhart(1988) for a clear exposition of the economics ofnatural monopolies, and Joskow (2005) for a morecomplete explanation of the normative argument).Highly complementary to this argument is the viewthat regulators will be motivated by the duty toprotect consumers from monopolistic abuse. Thisperspective on regulatory motivation came to beknown as the ‘public interest’ view (see, for in-stance, McCraw (1975) for an exposition). Eventu-ally, observers of regulation came to challenge thisperspective. Even when a regulatory body has beenset up to prevent monopolistic abuse, regulationends up being ‘captured’ by the firms it is supposedto discipline. This view was articulated in an influen-tial paper by Stigler (1971), and since then it hasbeen refined and expanded in a variety of ways.

The purpose of this review is twofold: first, toprovide a technically accessible account of the maintheoretical views on regulatory capture held byacademic economists, and, second, to provide anoverview of available evidence on causes and con-sequences of capture. The scope of this review isbroad although contributions focusing on utility regu-lation are emphasized. Some of the work on captureoverlaps with the literature on corruption. We touchon the latter when connected to the problem of thepurchase of official favours, but abstract from thatpart of the corruption literature dealing with harass-ment or extortion by public officials.

This review is structured as follows. The nextsection discusses briefly the argument on regulatorycapture put forward by Stigler (1971) and theformalization of these ideas by Peltzman (1976).This initial theory of regulatory capture is, in fact,common to other areas of public policy and, likemany of the theory’s developments, it may beapplied to the problem of political influence as it isbroadly understood. Peltzman (1976) made progressby leaving a number of elements in a black box. Inother words, important links in the theory’s logicalchain are assumed to work in a particular way,without our knowing whether those assumptions arereliable. Two elements that deserve to be probed

further are the precise nature of influence and theorigin of regulatory discretion. Section III presentsa three-tier principal–agent model where influenceand regulatory discretion are linked to the exchangeof favours and asymmetric information. This theoryhas a strong normative component. We brieflydiscuss the alternative ‘common agency’ approach.Both approaches emphasize influence through in-centives that can be construed as bribes, such asdirect payments or lucrative post-agency employ-ment, and model regulators as a single actor. Insection IV I discuss alternatives to influence throughbribes, such as the provision of information andcoercion, and briefly review work studying influ-ence over collective bodies. Section V discussesmodels on the ‘revolving door’ phenomenon, whichinvolves a tendency of regulators to favour industrywhen they have an industry background or whenthey expect rewards in the form of future industryemployment.

Section VI switches attention to empirics. First, Idiscuss the scarce evidence on the effects of asym-metric information (or the lack of transparency).Then I review evidence of a connection betweencapture and regulatory outcomes; I survey empiri-cal studies on the revolving-door phenomenon andon the potential role of regulators’ personal charac-teristics. Lastly, I review evidence that regulatoryoutcomes may respond to the methods used toselect officials who have an impact on regulation,such as regulators and judges. Section VII providesa summary and highlights open questions for futureinvestigation.

II. A SIMPLE MODEL OFREGULATORY CAPTURE

(i) Stigler’s Approach

Stigler (1971) advocated an economic theory ofhow the regulation of business comes to be. Thisincluded, but was not limited to, the regulation ofmonopolies. Stigler’s starting point was the obser-vation that, as a rule, regulation is acquired by theindustry and is designed and operated primarily forits benefit. An additional piece of data was thatregulation is not only observed in association withnatural monopolies, for which a normative rationaleexists. Many other economic activities attract state

205

E. Dal Bó

intervention, in the form of price-fixing and entrycontrol. In order to account for these observations,an economic theory of regulation had to specify thedeterminants of the supply and demand for regula-tion. The demand for regulation would be connectedprimarily to two features of the group of beneficiar-ies. First, whether the beneficiary group is large and,second, whether the group has large stakes inregulation. Excessive group size could hamper suc-cessful organization of the beneficiary group. Suchorganization requires the resolution of the well-known collective-action problem. This problem arisesfrom the temptation that each potential beneficiaryhas to shirk his obligations in the common lobbyingeffort. (This echoes Olson’s (1965) view that largegroup sizes can hinder collective action.) Large stakescould mobilize group members and give them anincentive to demand regulation. On the supply side, onewould have to pay attention to the machinery thatproduces regulation: the public sector, which re-sponds to political pressures. Stigler viewed politi-cians as potential suppliers of regulation who pursueselfish objectives, among which one should countthe desire to augment their power. If power dependsboth on money and votes, the costs of regulationcould include its potential to bring unpopularity to thepolitician that promotes it. However, Stigler countedone factor that could mitigate this cost: the fact thatthe very many voters that are only marginallyaffected by regulation would have poor incentives tobe well informed about the regulation in question.

Stigler illustrates his way of thinking through asimple study of the regulation of trucks in the UnitedStates. By the late 1920s, trucks emerged as acompetitor to railroads on inter-city freight. Railroadsresponded by influencing public authorities to im-pose limits on the weight trucks could carry in inter-city hauls. Stigler listed various factors that shouldaffect the demand for regulation.

(i) In states with numerous farm trucks, railroadswould have a harder time overcoming the agricul-tural interests that relied on truck services, makingweight limits less likely, or less stringent.

(ii) Trucks posed less of a threat on long-haulfreights than on short-haul. So, states where the

average haul is shorter would be associatedwith stronger demand for restrictive limits byrailroads.

(iii) States with better-quality roads would haveless of a reason to worry about truck weight, sothat the friends of railroad interests would havea harder time making the case that weight limitswere urgently needed to keep roads usable.

Stigler measured the degree of regulatory favours torailroads by looking at the weight limits on four- andsix-wheel trucks by 1932–3 in each state. Theexplanatory variables corresponding to the argu-ments in points (i)–(iii) were, respectively: the numberof farm trucks per 1,000 labour-force members (by1930); the average haul of railroad freight (by 1930);and the percentage of state roads of high-type (ahigher-quality paved surface) by 1930. All threeexplanatory variables appeared significant in thecross-state regressions (except for high-type surfaceon the six-wheel trucks weight-limit regression).

It is worth noting that Stigler’s ideas were neithertypical at the time they were published, nor whollywithout precedent. The idea that economic policymay not stem from a benevolent planner (as in thepublic-interest view) was not as well established inthe mainstream economics literature as it is today.2

The Stiglerian approach has been related to theChicago view of public policy, even when it could beseen as complementary to the emerging literatureon public choice, which was, in turn, associatedwith the names of Buchanan, Tullock, and theVirginia school. This literature focused on the role ofpolitical institutions in affecting collective decisionmaking in the face of heterogeneous societal inter-ests. The view of public policy emerging from theStiglerian and the public-choice school emphasizedthe idea that regulators could be swayed by specialinterests. This idea had an impact on a largerongoing debate on the optimal size and scope ofgovernment. On the microeconomics side, insightsdue to Pigou and Samuelson justified state interven-tion: in the presence of externalities or public goods,the decentralized allocations determined by marketswith private property will be inefficient.3 It was atthat stage that Stigler’s call for an economic theory

2 This was so even when libertarian and Marxist mistrust for state-sponsored goals was far from new.3 A counterpoint was offered on the macroeconomic side. The monetarist school was beginning to upset the prevailing views

of the macroeconomy as a system in need of activist intervention.

206


of regulation was made. If regulators are the nexusof a set of special-interest pressures, should we callfor even larger state controls in the hope of neutral-izing the potential special-interest bias? Or shouldwe roll back existing regulation and live with marketfailures? A lot of the work in economics in thefollowing decades can be organized in terms of theposition taken regarding these questions, and wecan view Stigler’s proposed programme as an im-portant step towards answering them.

As indicated by Posner (1974), a problem withStigler’s approach was the lack of clear implicationsas to the profile of groups benefited by regulation.Stigler emphasized that industries with concen-trated ownership would have an easier time atovercoming the hurdles facing collective action.However, large groups could attract favourableregulation by vote-seeking politicians. What consid-eration should prevail to define the profile of regu-lated industries?

(ii) Peltzman’s Model

Peltzman (1976) refined and expanded Stigler’sideas. He offered three distinct formal set-ups, thesecond of which focuses on price-entry regulation.This model rationalizes intervention in industries thathave both small and large numbers of beneficiaries.It is this second set-up that is of interest here.

The model comprises three classes of players: apolitician who holds the coercive power of the state,an undefined quantity of producers, and an unde-fined quantity of consumers. The politician wants tomaximize his ‘majority’ or ‘power’ as given by M =M(p,π), where p is the price paid by consumers andπ is the profits for producers. Assume that thepolitician’s majority decreases with higher prices toconsumers (high prices are vote losers), but that itincreases with the profits of producers (presumablybecause those producers vote, but also becausethey can mobilize voters or other elements of thecommunity through their financial power). Both thepower-losing effects of higher prices and the power-increasing role of profits are assumed to be ofmarginally decreasing intensity. Formally, theseassumptions on the majority M(p,π) take the form:Mp < 0,Mπ > 0,Mpp < 0,Mππ < 0. Assume also that themarginal effects of prices are unchanged by thelevel of profits and vice versa: Mpπ = 0. All supply

and demand information is captured in the functionπ = f ( p,c), where c = c (q) are the production costsof firms. Assume fp ≥ 0, fpp < 0,fc < 0.

The politician then chooses the price he will allowproducers in order to maximize his majority M(p, π)subject to the constraint π = f (p, c). Or,

The first-order condition for this problem is

For a firm facing no competitors, at prices lowerthan monopolistic pricing, an increase in price in-creases profits (i.e. df /dp > 0). Recalling theassumptions Mp < 0 and Mπ > 0, the first-ordercondition then tells us that, starting from a position ofmonopolistic pricing (i.e. a relatively high price), theofficial will lower the price to gain consumer votesuntil the marginal vote gain equals the marginalpower loss from dissatisfying producers. Inversely,starting from a competitive price, the official willraise the price to please industry, up to the pointwhere the marginal value of the favours fromindustry equals the marginal loss of consumer votes.This effect will also benefit a multi-firm industrywhen prices are fixed and enforced above competi-tive prices. (In such a case, the first-order conditionabove can be seen to reflect the interests of theentire industry, rather than those of a single firm.)

The model predicts that regulation will typicallyentail less than 100 per cent producer protection butalso less than perfect protection of consumersagainst market power. The ‘political’ price willusually lie between competitive and pure monopolylevels. Moreover, incentives for ‘regulator entry’appear highest when industry is fully monopolistic orperfectly competitive to begin with, because thepower gains from moving price towards a middlerange are highest (this is a result of the concavityassumptions on the function M). In other words,monopolies such as utilities will attract regulation,but not for the normative reason commonly ac-cepted, but because a politician can act as a brokerthat trades moderate political losses with producersfor a large political gain among consumers. In thiscase, and as long as the political price allows for thereposition of capital, the regulator’s intervention is

)].,(,[ cpfpMMaxp

.0dpdfMM

dpdfMM pp ππ −=⇒=+

207

E. Dal Bó

beneficial for social welfare (understood as profitsplus consumer surplus). The reverse of the pictureof entrepreneurial regulatory intervention is thatcompetitive industries will attract regulation as well,because the initial marginal losses of power owingto angry consumers can be small relative to the gainsowing to grateful producers. Examples of the lattertype of regulatory presence are when the stateengages in price fixing in competitive industries andwhen it controls entry (through licensures) in occu-pations. The model can then account for the appar-ent puzzle of regulation being present in technologic-ally diverse industries.

The main feature in Peltzman’s model is that theofficial appears to be trading off the benefits fromfavouring two different masters: consumers andproducers. However, little is said about how thosebenefits materialize. Neither of the two stakeholdersin the model is explicitly considered an active,strategic player. Moreover, no detail is given as tohow a politician can equally depart from monopolis-tic or competitive pricing. In the case of a naturalmonopoly, for instance, a regulator may be limitedby the fact that he is supposed to leave the firm areasonable rate of return, but allowing for higherprices may be denounced as a give-away to indus-try. The following section covers models tacklingthese limitations.

III. PRIVATE INFORMATION ANDCOLLUSION IN HIERARCHICALAGENCY

(i) Overview

One way to understand capture is by thinking of athree-tier hierarchy comprising a political principal(the government), a regulator, and an agent (thefirm). Why use a principal–agent model? This is theset-up used when we want to analyse incentiveproblems in detail. The simplest possible modelwould have only two parties: a firm and a (potentiallycapturable) government. Can we justify complicat-ing the model by considering more players? Theintroduction of a third element allows differentiationbetween the government and the regulator (the

reader may think that a similar structure can be usedto think of a hierarchy featuring citizens, govern-ment, and the firm). The extra player (the regulator)allows one to analyse how the political principalmight want to respond to the risk that its delegatemay be captured.

Tirole (1986) was probably the first to analyseregulatory capture in a three-tier set-up. As I willpresent it, the model has two building blocks. Thefirst is the model of regulation of a monopolist withunknown marginal costs (Baron and Myerson, 1982).In this set-up, the firm has private information on itscosts. Thus, the government is unsure about howmuch of a cost reimbursement (or how high a price)the firm should be allowed. The optimal best re-sponse for the regulator is to offer a second-bestcontract to the firm, one that attempts to limit thebenefit the firm derives from the asymmetric infor-mation. However, because the firm is valuable toconsumers even when costs are high, and becausecosts are high with some probability, the contractoffered to a firm when the government does notknow the state of nature necessarily leaves the firmsome rents in situations when costs are actually low.The second building block, as proposed by Tirole(1986), then further developed by Laffont and Tirole(1993, ch. 11), assumes that the regulator, whospecializes in learning about the industry, may findout the true costs of the firm. Then the firm has anincentive to bribe the regulator into not telling thegovernment when costs are low. If the regulatorsays he has learned nothing, the best possible con-tract that the government can offer the firm is onethat leaves rents to the firm. The government thenhas an incentive to offer the regulator a contractinducing him not to lie and simultaneously to offer acontract to the firm that reduces incentives forcollusion with the regulator.

(ii) The Model

In this set-up, the firm has private information aboutthe marginal cost of providing a service to consum-ers, and the government wants to maximize anobjective including consumer surplus.4 In keepingwith the notation of last section, the firm’s operatingprofits are π( p) = q ( p) p – C (q ( p)), where q ( p)

4 Spiller (1990) analyses a related problem where Congress lacks information on the regulator’s amount of effort directed atimproving regulatory outcomes.

208


is the demand function, p is the allowed price, and C= F + qc is the cost of provision, where F is a fixedcost. The marginal cost c takes one of two values inthe set {cl , ch}, where 0 ≤ cl < ch. Marginal costs aredetermined by unpredictable developments that arebeyond the players’ control, such as oil discoveries,weather changes, etc. So, we will say they aredetermined by ‘nature’. Nature picks low costs withprobability γ and high costs with probability 1 – γ. Inaddition to approving a price, the government hasthe ability to make transfers T to the firm, which arepaid for by consumers. The total pay-off for the firmis Π = π(p) + T. The pay-off for consumers is theirconsumer surplus minus their tax payments to fi-nance transfers to the firm (we assume the govern-ment keeps a balanced budget; the taxes paid byconsumers equal the transfers to the firm). Thegovernment seeks to maximize social welfare asgiven by

S = s(p) – T.

Hence, the government would like to use transfersto reimburse the firm just for its fixed cost and setprices equal to marginal cost in order to maximizenet surplus. When the government knows the reali-zation of c, such an arrangement is feasible. But itis not when the government does not know therealization of c. Suppose for a second that thegovernment has offered the full information con-tracts {[T = F, pl = cl ] , [T = F, ph = ch]}. Then evenwhen costs are low the firm has an incentive to claimthat costs are high and obtain better compensation.The informational rent the low-cost firm wouldobtain in the asymmetric information regime is

Πl = q(ph)(ph – cl) = q(ph)(ph – ch) + q(ph)(ch – cl)= q(ph)(ch – cl) ≡ q(ph)∆c,

where ∆c = ch –cl. Clearly, q (ph)∆c > 0, and thelow-cost firm obtains a positive rent owing to itsinformational superiority.

The government can do better than simply offer thefull information contracts and have firms obtain aninformational rent whenever costs are low. Oneway of doing this is to raise the price that is allowed

in high-cost situations, so that lower demandedquantities will reduce the informational rent. Theproblem with this is that it reduces consumer surpluswhen costs are high. Hence, the trade-off facing thegovernment can be characterized as being betweenlowering informational rents in low-cost situationsand not reducing consumer surplus in high-costsituations. There are many potential arrangementsthe government could choose from. However, thegovernment loses nothing by restricting itself toimposing arrangements that are tailored to eachtype, while respecting the condition that each typeshould have incentives not to misrepresent its pri-vate information.5 Therefore, whatever arrange-ment the government offers must entail,

Πl(pl,Tl) ≥ Πl(ph,Th),

which is called the incentive compatibility constraintof the low-cost firm. Analogously (although notbinding in equilibrium), the offered arrangementmust give the high-cost firm incentives for sincerity,i.e. Πh(ph,Th) ≥ Πh(pl,Tl). Simultaneously, the par-ticipation constraint on the two types of firm is thatthe government must offer terms that will not induceany type to shut down: Πl(pl,Tl) ≥ 0, Πh(ph, Th). Infact, the government will offer the following ar-rangement,

A sketch of the proof is in the Appendix. Twoobservations are worth making. First, in order toreduce the informational rent that the firm gets inlow states, the government raises the price in high-cost states by (γ/(1 – γ))∆c (which reduces de-manded quantities and thus the informational rentq(ph*)∆c). Note that the government raises ph fur-ther when γ is high, so the expected benefit of areduced informational rent is high, and the expectedcost of a reduction in consumer surplus is small.Second, marginal cost pricing is preserved in the low-cost state and no rents are left to the high-cost firm.6

.

1

)(;0

*

*

*

ll

hh

hlh

cp

ccp

cpq

=

∆γ−

γ+=

∆=Π=Π

5 The revelation principle (Myerson, 1979) implies that the optimization over all possible arrangements can be reduced to onewhere the player with private information simply reveals its type truthfully. See Laffont and Tirole (1993, ch.11) for a completeproof in a very similar context to that presented here.

6 The solution is completed with the characterization of transfers: Tl = F + q (ph*)∆c ; and Th = F – q (ph*) (γ/(1 – γ))∆c.

209

E. Dal Bó

The government should be interested in finding aregulator who specializes in learning c in order to trymore often to implement first-best allocations. Inthis way the government can save on rents left to thefirm and on distortions in the high-cost state. Sup-pose, then, that when the firm has low costs, theregulator learns it with probability λ, and learns it ina way that is verifiable to the government. Withprobability 1– λ the regulator finds nothing. Whencosts are high, the regulator can never learn any-thing.7 The timing of interaction is as follows. First,nature determines the firm’s costs, and the firmlearns it. Second, the government announces both awage to the regulator and regulatory contracts to thefirm, both potentially contingent on the regulator’sreport. Third, if c = cl, the regulator learns it withprobability λ (and the firm learns whether theregulator learned costs or not). Fourth, the firm andthe regulator decide whether to collude. Fifth, theregulator sends a report to the government r ∈{cl,∅}. This means that the regulator, when seeingthat c = cl, can reveal it, but can also claim to beuninformed, both in that situation and when he reallyis uninformed. After getting the report, r, wages andcontracts are made available to regulator and firm.

Note that when costs are low and the regulatorlearns this is the case, a truthful report to thegovernment will induce the latter to offer the fullinformation contract to the firm and not accept anyclaims that costs are higher. Thus, the firm’s infor-mational rent evaporates. Hence, the firm has anincentive to pay the regulator a bribe up to the valueof the informational rent in order to buy his silence.The value of the informational rent is q( ph )∆c. It isclear that if the government wants to compete withthe firm and push the regulator towards honestreporting, a payment conditional on r = cl should bepromised. Denote this payment w and assume thatthe reservation wage of regulators is zero. Assumethat every dollar that the firm may offer the regula-tor in side payments costs the firm 1 + ψ. For theregulator to want to report that costs are low, thewage must satisfy,

The benefit of preventing collusion fully is to gain fullinformation more often. The cost is the incentive

payments to the regulator and the potential distor-tions to the contract to be offered by the firm. Wenow assume that deterring collusion is worthwhileand focus on the consequences for equilibriumallocations.

If a regulator is colluding with the firm, he willalways claim to know nothing, and the probability ofa low state is simply γ. Under a collusion-proofcontract, though, the regulator behaves honestly.So, the probability that costs are low when theregulator claims to be uninformed is

We assume that the government does not careabout the income of regulators from a welfare pointof view, except in that it detracts from consumers’income. Recall that when the government obtainsno information, it will offer the second-best contractcharacterized earlier, although taking into accountthat the regulator claiming to be uninformed affectsthe perceived probability of the low state. Thesecond-best contract specifies marginal cost pricingfor the low-cost firm but leaves rents in it, and theserents depend on the price chosen for the high-costfirm. No rents are left to the firm in the high-costsituation. Then, expected social welfare, as seen bythe government, can be written as

EV = γλ{s (cl) – F – w} + (1 – γλ) {µ [s (cl) – Tl (µ)]+ (1 – µ) [s (ph, µ) – Th (µ)]},

where Ti (µ) (i = l,h) denotes that transfers are beingdetermined in the second-best solution correspond-ing to the case when the government has receivedr = ∅ and hence holds beliefs µ on the low-coststate. The maximization of this problem yields,

This expression contains a familiar term, (γ/(1 –γ))∆c, which is the distortion introduced in theoptimal contract seen before, when a regulator wasnot available. But the extra term,.

1)(ψ+∆

>cpqw h

7 I model regulatory learning as Armstrong and Sappington (2005) do. Assuming that the regulator can also become informedin the high-cost situation complicates notation without adding insight.

.1

)1()1()1(

)1()|(λγ−λ−γ

=γ−+λ−γ

λ−γ=∅==≡µ rccP l

.1

11)1(1

cccp hh ∆

ψ+

−γ−

γλ−∆

γ−γ

+=

c∆

ψ+

−γ−

γλ−

111

)1(

210


implies that the distortion to the high-cost price whena regulator is available will be smaller, even if theregulator is corruptible. The parameter ψ might beseen as a ‘capture’ parameter. When it is zero, thefirm has no difficulty sharing the information rent withthe regulator. So, a dollar from the firm has maximumpurchasing power. In that situation, the term

goes to zero, and the distortion-saving power ofhaving a regulator with learning capabilities disap-pears. The price set for the high-cost firm con-verges to that when the government has no accessto a regulator. When ψ is positive, however, thedistortion savings increase. In the limit, when ψ goesto infinity, the regulator becomes essentially incor-ruptible because even the smallest bribe is infinitelycostly to the firm. In that case, we obtain

The last expression indicates that, relative to thecase with no regulator, the distortion is diminished inproportion to the amount of information brought bythe regulator (measured by the probability λ that heis informed). Note that the informational quality ofthe regulator also affects the solution. When theregulator is never informed (λ = 0), things areexactly as when the regulator does not exist, and thesolution reverts back to the original one. If theregulator is always informed (λ = 1), the savings onpricing distortion are maximal. Note that the idealcombination of a regulator that is perfectly informedand incorruptible (λ = 1, ψ → ∞) eliminates alldistortion, yielding ph = ch. The reason is that if theregulator is effectively incorruptible, and alwaysalerts the government if costs are low, the problemreverts back to one where the government is per-fectly informed.

At this point it may be useful to provide a brief non-technical summary of the model and findings. Themodel relies on a few basic assumptions. Theregulated firm has superior information about itsproduction costs, and the government wants thefirm to produce as much as would be convenient forconsumers. The firm, however, cannot be forced toproduce when doing so would trigger losses, so the

remuneration allowed by the government must besensitive to production costs. Thus, the governmentwill want to appoint a monitor (the regulator) tomitigate the informational advantage of the firm.The danger is then that the firm may tempt theregulator not to disclose information that wouldreduce the approved compensation. Thus, regula-tors and firms may collude to keep the money owingfrom consumers even when the prevailing low costswould justify a reduction in compensation. Thescope for such capture of the regulator by the firmdepends on the amount of information that theregulator may obtain, and on how easy the environ-ment makes it to bribe regulators.

(iii) Discussion

The model just reviewed offers a precise frame-work to understand how asymmetric information isthe source of regulatory discretion, making capturepossible. The model also considers optimal re-sponses by the political principal, and how these willaffect implemented allocations. This emphasis maystrike some readers as pertaining to a normative,rather than positive, approach. It may also be arguedthat the offer by the government of regulatorywages that are contingent on the regulators’ reportsis implausible. In the current model, the inability tooffer contingent wages would eliminate the possibil-ity of combating collusion.

The literature has long considered alternative waysto use wages to fight capture. Becker and Stigler(1974) consider wages that are not directly contin-gent on the regulator’s actions. However, the wagesare higher than the regulator’s alternative earningsand make the latter eager to keep his job. Incombination with above-market wages, the govern-ment could then audit the reports of regulators andfire those who are found to have misrepresented thefirm’s private information. This introduces a cost toregulators of engaging in wrongdoing, and a judi-cious choice of wages and auditing frequencies cancontribute to mitigate capture.8 A large literaturehas used similar ideas in connection with crimedeterrence and corruption control (on crime, see, forexample, Becker (1968), and on corruption see, forinstance, Besley and McLaren (1993) and Ades andDi Tella (1997)). In real life, the scope for using

ψ+

−1

11

.)1(1)1(1

cccccp hhh ∆λ−γ−

γ+=∆

γ−γλ

−∆γ−

γ+=

8 On the role of external supervisors in collusive hierarchical agency, see Kofman and Lawarree (1993).

211

E. Dal Bó

wages and monitoring to deter regulatory favours islimited. The stakes are so large that, fines beinginfeasible and monitoring being imperfect, the re-quired wages would be astronomical. Thus, morecomplex institutional responses may be required.One possibility is the separation of regulatory pow-ers so as to lower the stakes in collusion (Laffontand Martimort, 1999). Another possibility is to levelthe playing field by incorporating the differentstakeholders into the bureaucratic design of regula-tory decision-making (McCubbins et al., 1987).

(iv) An Alternative Model that does not Relyon Private Information

Bernheim and Whinston (1986) introduced the ‘com-mon agency’ model of influence, which has a morepositive angle. In this set-up, multiple parties offer‘price lists’ for different favours or allocations thatthe official, or common agent, may set. Thoseparties compete, offering such schedules reflectingthe prices they are willing to pay for different policychoices, and hence this model is built upon a ‘menu’auction. This model allows a more precise char-acterization of the nature of influence in situationswhere there is competition for policy favours, asoriginally studied by Peltzman (1976) and Becker(1983). In the common-agency set-up there is noasymmetric information, and the origin of the discre-tion of the official remains unmodelled. The modelfeatures multiple equilibria which may be ineffi-cient.9 Grossman and Helpman (1994) extended thecommon-agency model to analyse trade policy de-termination, while sticking with quasi-linear pay-offfunctions. We comment later on empirical investi-gations inspired by this model. Dixit et al. (1997)further extended the common agency set-up toconsider general pay-off functions, which allows usto analyse situations where income effects matter.

IV. ON THE INSTRUMENTS ANDTARGETS OF INFLUENCE

The capture models reviewed so far assume thatfirms will provide incentives to regulators, and that

regulators can be thought of as single person. Twoimportant questions are whether the provision ofincentives is the only, or even the most relevant,instrument of influence, and what happens when thetarget of influence is a collective body. This sectionthen begins by discussing the provision of informa-tion as opposed (or in addition) to incentives. Arelated matter is that the type of incentives used inthe models of section III can, broadly speaking, beunderstood as ‘bribes’. So an additional questionarises as to what form of incentives can be under-stood that way, and whether other forms of incen-tives may play a role. The second part of this sectionthen discusses alternative forms of influence that donot exclusively rely on the provision of incentivesthat can be construed as bribes. Finally, the third partof the section briefly discusses work studying influ-ence over collective bodies.

(i) Are Incentives the Only Instrument ofInfluence?

A literature in political science and economics hasanalysed the provision of information, rather thanincentives, as a form of exerting influence. Calvert(1985) proposes a particular model of information-processing where the known bias of an advisor (orinterest group) cannot be fully ‘undone’ by thereceiver of information (say, the regulator).10 Thismodel yields that the decision-maker may in factprefer to use biased advisors, much as when regu-lators receive information from firms and otherplayers of the regulatory game. The benefit ofhaving a biased advisor is that it can become a morecredible source in extreme situations where a mis-take would be particularly costly. Austen Smith andWright (1992) offer a model where there are two,rather than one, sources of information. In their set-up, two interest groups may make costly searchesfor information and then offer messages to a deci-sion-maker. Although the model is intended to cap-ture informational lobbying to win over a legislator,the set-up can help explain situations when firmsand consumer advocates offer information to try toaffect a regulator’s decision. For example, thesestakeholders may commission costly consultants to

9 Equilibria that are ‘truthful’ are also efficient. Truthful equilibria involve truthful strategies in that the schedules of prices offeredby interest groups must reflect the shape of those groups’ pay-off functions. This class of equilibria coincides with the set ofcoalition-proof equilibria.

10 The bias does not merely affect the expected levels of advice, but also the shape of the probability distribution over possibleadvice messages.

212


provide industry-specific studies. In this model,messages need not be truthful, but the regulator hasthe possibility, at a cost, of verifying the informationhe has received (possibly by commissioning furtherconsulting). For some parameter values it is shownthat none, one, or both interest groups may acquireinformation with positive probability and then submitmessages to the regulator. Moreover, there areregions of the parameter space where the regulatorends up making perfect decisions.

Lohmann (1993) tackles the (very different) ques-tion of whether mass political action can conveyinformation. This set-up is different from the previ-ous ones in that signalling is costly. In her model,engaging in the activity that may ultimately conveyinformation is costly in itself; in other words, the costis the message. Although the connection with regu-lation is tenuous, the basic message of her model isuseful: by engaging in costly actions (such as com-missioning consulting that may be in itself irrelevant)an interest group such as the firm may signal itsprivate information about the appropriateness of aparticular policy.

(ii) What are the Incentives Provided byIndustry?

As suggested by the model reviewed in section III,one possibility is for firms to offer outright bribes.Another possibility is money payments with politicaluse, such as campaign contributions to politicianswith electoral goals. In the case of regulators with-out electoral ambitions, a much talked about form ofincentive is the promise, tacit or explicit, of futurelucrative employment in industry. The twin factsthat regulators have often had industry jobs before,and sometimes end up in industry after their tenure,is referred to as the ‘revolving doors’ phenomenon.In section V I discuss some theoretical contributionsregarding this phenomenon and I review evidenceabout it in section VI.

Another possibility is that firms may attempt toinfluence regulators through negative incentives,such as threats, tacit or explicit, of lowering theirutility. This could be attained through direct physicalpunishment. Although perhaps not so relevant foreconomies where the rule of law is well established,this form of incentive to officials is more widespreadin some emerging economies (see Dal Bó and Di

Tella (2003) for some examples). Firms may per-haps more often be able to spread negative rumoursabout the competence of a regulator, which maydamage the latter’s career. Also, they may destabilizea regulator’s grip on his job by open confrontation,which may raise the political costs for governmentof backing up the regulator. Indeed, Hilton (1972)argued that in real life the main objective of regula-tors is to minimize complaints by firms, or keepingregulated firms from ‘squawking’. However, whenit is known that regulators may be influenced throughpotential complaints from firms, the observation ofsuch behaviour may simply indicate that the regula-tor is performing his duties well. Thus, the power ofsquawk to damage the regulator is not a logicallyforegone conclusion. Leaver (2002) offers a modelwhere regulators may take more or less aggressivestances toward firms, and both positions may beright or wrong depending on the state of nature (say,cost levels facing the firm). Because regulators donot exactly know the state of nature, they may makemistakes and hurt firms unnecessarily. Regulatorscome in two types: smart and dumb. The latter aremore likely to make mistakes, and the market willlearn about mistakes when firms squawk. Whenregulators’ concerns about reputation are strong,there is a Bayesian perfect equilibrium where twothings happen: firms complain against mistakentough policies, but not against generous ones (evenif mistaken). Less able regulators, fearing that firmcomplaints may damage their reputation, are gener-ous to the firm some of the time, even when havinginformation indicating that a tough policy is calledfor. The overall message is that the possibility thatfirms may complain publicly about tough regulatorydecisions buys firms some regulatory slack.

The incentive-based models reviewed so far as-sume that a special interest such as the regulatedfirm will use either bribes or some form of coerciveinducement. However, there is no reason to sup-pose that firms will restrict themselves to using onlypositive or only negative incentives to influenceregulators, if both instruments are available. Dal Bóet al. (2006) offer a model where an interest groupoffers both bribes and threats to officials in itsattempts to influence public policy. Three mainimplications follow from this model.

(i) When firms have a way of imposing disutilityon regulators, they will save on rewards by

213

E. Dal Bó

simultaneously threatening potential retaliations,private or political, against the regulator.

(ii) Relative to a world in which firms cannot exertcoercive pressure, the pay-off to regulators islower. If pay-offs determine competence, com-petence in regulation will be lower when coer-cive pressure is feasible.

(iii) When retaliations are feasible the cost of influ-ence is smaller. Hence, even when the gain inprofits from a price favour may not be as large,influence may be affordable.

Therefore, regulatory capture may be more preva-lent in countries where regulators are less protectedfrom violence or from firm-originated rumours, aswould be the case when the rule of law is weaker orwhen regulatory job stability is lower. If the termlength of regulatory positions is a proxy for relativestability, one would expect longer-term regulators tofeel more insulated against firm retaliations thatdamage the market value of regulator’s reputations.The section on empirics presents evidence on theconnection between term lengths and regulatoryoutcomes. Implications (ii) and (iii) together implythat countries with more prevalent capture will haveless competent regulators.

(iii) Influence over Collective Bodies

The literature on influence over collective bodies isrelevant for regulation both when regulatory agen-cies are run by boards and when legislatures affectregulatory policy. This literature has been mainlyassociated with concerns about the capture oflegislators. Denzau and Munger (1986) attack theproblem by focusing on a single legislator who facesa problem analogous to that studied by Peltzman(1976): on the one hand, a special interest may offerrewards such as campaign contributions, but on theother, constituents may punish a legislator who sellsout. The authors then indicate that a special interestwill typically face legislators with varying ‘prices’,and that the purchaser of legislative favours will tryto assemble a majority of minimum-cost legislators.Snyder (1991) was probably the first fully to modelthe vote-purchasing decision of an outsider whileconsidering a multi-member legislature. In equilib-rium, an optimizing principal will direct its resources

to legislators who are not ‘friendly’ enough to votefor the principal’s project, but who are at the sametime not unfriendly enough to become too expen-sive. At the same time, the outside interest will limitwhat it asks for in order to lower the costs ofinfluence. In his model, passing a bill will typicallyrequire bribing entire blocks of legislators, suggest-ing that influencing larger bodies will be typicallymore expensive. Neeman (1999) studies a multi-agent contracting problem under uncertainty andapplies it to voting. He shows that in very largevoting bodies a member will attach a small probabil-ity to being pivotal and sell her vote cheap. Dal Bó(2000) seeks to unveil vulnerabilities of collectivebodies to outside influence that will operate evenunder certainty, and then studies potential remedies.He considers the possibility that a lobby may makeoffers that are not only contingent on the way anindividual committee member acts, but also on theway other members do. This flexibility tends toallow a lobby acting unopposed to capture commit-tees of any size at no cost. The main protectionsagainst this vulnerability are committee membersthat can enforce cooperation among themselves,committee members that are accountable to outsideparties, or secret voting.

Bennedsen and Feldmann (2002) extend the litera-ture on informational lobbying to the case where thetarget is not an individual but a collective body, suchas a legislature. In their model, an interest groupwants a particular public good provided (say, envi-ronmental protection). The legislature is made up ofrepresentatives coming from districts that havedifferent valuations for the public good. They showthat a lobby may have an incentive to generateinformation on what districts would benefit the mostfrom the public good. This information will allow anagenda-setter a more finely tuned coalition-buildingstrategy that relies on high-valuation districts. Thisminimum-cost coalition made of high-valuation dis-tricts leads to the approval of higher levels of thepublic good desired by the lobby. When party cohe-sion is very strong in a legislature and legislators votein large blocks, the heterogeneity of districts cannotbe easily exploited by an agenda-setter seeking toform a minimum-cost coalition. Thus, party cohe-sion in parliamentary democracies (as in Europe)dilutes the incentives to direct informational lobby-ing efforts at legislators. As a result, informational

214


lobbying should target regulatory bodies more heav-ily in countries displaying high party cohesion inlegislatures.

V. REVOLVING DOORS

The fact that many regulators come from industry,or end up there, has long been thought to be a sourceof bias in regulatory decisions. In keeping with therest of the paper, this section uses a ‘public utility’lens, but the topic of the revolving doors is moregeneral and the models reviewed here can certainlybe thought to have implications for other areas ofregulation. For example, defence procurement offi-cials may go to work in the defence industry,government health policy-makers may eventuallytake jobs with private health companies, tax officialsmay become corporate tax advisers, etc. The list ofpossible moves from the public to the private sector(and vice versa) is, indeed, long.

The channels through which industry employmentmay affect regulatory performance are multiple. Animportant distinction is whether such employment isheld before or after regulatory involvement. Comingfrom industry may induce regulators to make pro-industry decisions because of the regulator havingbeen ‘socialized’ in an industry environment. This inturn may yield different cases. In one extreme, wemight find fairly irreflexive, partisan pro-industrytypes; on the other, well-meaning individuals whotend to see the concerns of industry as more legiti-mate, salient, or relevant to general welfare, be-cause those are the concerns they are most familiarwith. An example of the latter case might be aperson with industry background who worries aboutthe fact that low prices may discourage investment,and in turn hurt future consumers.11 The possibilityof post-regulatory employment is different: regula-tors may bias their decisions in order to enhancetheir chance of future employment in industry. Anexplicit quid pro quo may exist, whereby lenientregulation is rewarded with future employment inindustry. At other times, firms may mainly hireformer regulators because the latter possess valu-able skills and not because such hiring is part of areward scheme. Still, given skills, firms may prefer

employees that seem to have industry interestsmore at heart. Then, regulators may try to signaltheir appeal to industry by being lenient to it. In thislast situation, a pro-industry regulatory bias is moreof an instance of the ‘collateral damage’ of a freecirculation of human capital.

The set-ups explored so far illuminate a way inwhich posterior employment in industry may affectregulatory decisions: such a possibility acts like acovert bribe and induces lenient decisions towardindustry. Having explored models in sections II andIII that can help rationalize the negative side of therevolving doors, I now review more optimistic views.The remainder of this section explains in detail thearguments in two papers pushing the view that therevolving doors may not only allow industry a largerpool of human capital; the revolving doors may eveninduce better regulatory outcomes.

The first of these two papers is by Che (1995). Heconsiders a hierarchical agency model à la Tirole(1986) and provides three different treatments. Inthe first, it is assumed that regulators may invest inhuman capital during their tenure. Two forms ofcapital are available: ‘technical’ capital, or exper-tise, and ‘lobbying’ capital, or contacts. The firsttype of capital reduces the marginal cost of monitor-ing the firm, while the second does not. Whenindustry future employment is associated to firmswanting to obtain technical expertise, then regu-lators will want to obtain such expertise. This willin turn lower a regulator’s costs of monitoring thefirm. Under the optimal contract placed by theoverall principal (the government, say), this lowercost of monitoring translates into better regulatoryoutcomes and lower informational rents left to thefirm. Thus, opening the revolving doors would inthis case encourage the acquisition of usefulexpertise by regulators. However, if firms em-ploy former regulators because of their contactsand potential lobbying usefulness, then opening therevolving doors may have the opposite effect,namely that of discouraging the acquisition ofuseful expertise. An important empirical questionis then what is the primary drive behind firmsemploying former regulators: expertise or lobby-ing potential?

11 Thus, one could argue that very harsh regulation may reduce investment and affect capacity to serve future consumers. Navarro(1982) studies the cost of capital and the rankings of Public Utility Commissions (PUCs) from the point of view of stockholders.Firms under the watch of harsh PUCs appear to face higher cost of capital.

215

E. Dal Bó

A second treatment in Che (1995) considers thepossibility that regulators may want to be aggressivein order to signal their intellectual quality. More ableindividuals, when acting as regulators, will moreoften obtain information about the firm that can beused to justify harsher regulation. When firms wantto employ more able individuals, and ability is notreadily observable, the firm may want to employformer regulators that were harsh because thisprovides a positive indication of ability. A crucialassumption here is that regulators cannot separatetheir manifestations of intellectual fire-power fromtheir use of discretion. In real life, a regulator maybe able to convey both great intellectual qualifica-tions and a disposition towards leniency. Also, incomplex situations, such as those surrounding tariffreviews, a person may convey great intellectualability by finding apparently good arguments that getthe firm to more favourable ground.

The third treatment by Che (1995) is when he allowsfor explicit collusion between the firm and theregulator. The key assumption is that such collusionmay not always succeed. As we saw in section III,collusion happens because the regulator may shieldthe firm from the tightest regulatory policy that thestate of nature justifies. For the regulator to find outabout whether the firm has a stake in hiding informa-tion, the regulator must monitor the firm. Thus, moreintense monitoring by the regulator increases thechance that the latter will sign a profitable sidecontract with the firm. Open revolving doors can beseen to facilitate such transactions by giving the firma currency with which to buy off the regulator.Therefore, revolving doors increase the regulator’sincentives to monitor the firm. But because some ofthe time collusion fails, the extra monitoring some-times translates into better regulatory outcomes.

The second paper offering a revisionist view on therevolving doors is by Salant (1995). His focus is onthe classic hold-up problem facing a party that hassunk a relationship-specific investment. Supposethe regulated firm has invested to create productioncapacity. Regulators that aim at maximizing con-sumer surplus may then push down prices so muchas to expropriate the firm’s sunk investment, insteadof allowing for capital recovery. Foreseeing this,firms will under-invest. In a game where both thefirm and the regulator are infinitely lived, there is a(sub-game perfect Nash) equilibrium where both

players ‘cooperate’: firms invest, and regulatorsallow for capital recovery. In real life, however, noplayer is infinitely lived, but rather made up ofsubsequent generations of players. When genera-tions of regulators and firms overlap (with youngregulators facing old firm managers and vice versa),some degree of cooperation is possible, even if eachgeneration is finitely lived (see also Cremer, 1986).Salant (1995) shows that the amount of cooperationbetween overlapping generations of regulators andfirms can be enhanced by the revolving door. Thepossibility of future consulting income ‘on the otherside of the fence’, depends on players having be-haved cooperatively during their tenures as manag-ers and regulators. This endows the ongoing rela-tionship between firms and regulators with an extrainstrument to punish noncooperative behaviour, andthus allows the system to sustain investment andcapital recovery. One important associated risk ofthe revolving doors in this model is that it givesplayers ‘too much’ of an incentive to cooperate, inthe sense that they would find it worthwhile to over-accumulate capacity. Projects that are not benefi-cial to consumers can be sustained in equilibrium.An important assumption behind the result that therevolving doors cannot damage welfare is thatprojects not satisfying a break-even constraint willnot be approved.

VI. EVIDENCE

The evidence on the determinants of regulatorycapture is still well short of abundant. Evidenceinvolving corruption at the national level was estab-lished relatively early, and studies have shown thatlack of market competition and closeness to tradeare determinants of corruption (see, for instance,Ades and Di Tella, 1999). The related questions atthe regulatory agency level, such as whether regu-lators with overlapping jurisdictions or budgets tiedto outcomes fare better or worse, have, to myknowledge, not been tackled by empirical research.The emphasis of theory on asymmetric informationand on the use of wages as a capture deterrent alsoseem unmatched by empirical research. An excep-tion is Di Tella and Schargrodsky (2003), whoinvestigate the use of wages and monitoring inhospital procurement. Reinikka and Svensson (2005)also find that more information through a newspapercampaign in Uganda gave schools more of a chance

216


to claim funds to which they were entitled, reducingthe probability that the funds would be captured bylocal officials. Also, some work has been done onthe role of transparency in inducing fiscal responsi-bility. Lowry and Alt (2001) offer a model andevidence that more-restrictive fiscal institutions mayenhance transparency by allowing investors to ex-tract better signals regarding the government’sfiscal responsibility. Alt and Lassen (2006) usesurvey data on the perceived transparency of budg-etary information for OECD countries and find thatlower-transparency countries have a political fiscalcycle, while high-transparency countries do not.Apart from this relatively scant evidence on the roleof transparency, none of which is related to indus-trial or utility regulation, there is virtually no evidenceof how (or whether) asymmetric information fos-ters regulatory capture.

In the remainder of this section I review availableevidence of four types. The first pertains to thepotential impact of capture on policy outcomes. Thesecond type specifically concerns the revolvingdoors. The third relates to whether regulator char-acteristics matter. Some of this evidence comes outin work investigating the effects of revolving doorsand is covered then. The fourth type of evidence ison whether regulators appear to reflect pressurefrom citizens. To get at this issue I survey evidenceon whether the mechanisms to elect regulators playa role. If they do, then under certain institutional set-ups regulators may be more accountable to citizens.This is important in order to ascertain the extent towhich models such as Peltzman’s are accurate whenportraying regulation as the outcome of a balance ofpressures in which consumers play a role.

(i) Capture and Regulatory Outcomes

Relating the amount of capture to regulatory out-comes is difficult, mainly because measuring cap-ture is tricky. One way of doing this is usingnationwide measures of corruption, which may becorrelated to regulatory capture. Dal Bó and Rossi(2004) test a simple model for why countries whereregulators are more easily capturable should havemore inefficient utilities. When regulators are morelikely to be vulnerable to influence and approve pricehikes, firm managers do not have incentives to trytheir best when coordinating and supervising the useof production factors. This will induce more mana-

gerial shirking and lobbying instead of effort exer-tion, and the way to meet service obligation targetsis to use more inputs per unit of output. Dal Bó andRossi (2004) study a panel of 80 electricity distribu-tion firms in 13 Latin American countries for theperiod 1994–2001. They find that firms are moreinefficient in countries and times displaying highercorruption.

Another way of analysing the connection betweencapture and outcomes is looking at whether influ-ence in the form of campaign contributions topoliticians matters. Because the focus of this articleis on regulation, I do not cover the large literaturestudying whether campaign contributions affectdimensions of legislative behaviour such as roll-callvoting. Rather, I comment on the (unfortunatelyscarce) evidence regarding the link between politi-cal influence and regulatory outcomes. We may seelegislators as regulators themselves, or, given, forexample, the Congressional oversight on regulatoryagencies, one may conjecture that campaign contri-butions to legislators may affect the inclination of thelatter to exert pressure over agencies.

De Figueiredo and Edwards (2005) analyse whole-sale price determination by state regulatory com-missions in telecommunications. They focus on theprice that incumbents are allowed to charge en-trants to local networks in densely populated areas.The authors have a panel of price decisions made atthe state level in the United States, corresponding tothree electoral cycles (1997–8, 1999–2000, and2001–2). The main explanatory variable they con-sider is the campaign contributions to candidates tothe state legislature made by incumbent firms rela-tive to entrant ones. Although limited by the numberof regulatory decisions, the authors find evidenceconsistent with the idea that campaign contributionsaffect price regulatory decisions. When campaigncontributions by incumbent firms are relatively higher,so are the prices that incumbents are allowed tocharge entrants.

Taking regulation in a broad sense, we may considertrade policy as a regulatory outcome. Hansen andPark (1995) study the determinants of theInternational Trade Administration (ITA) decisionswhen US domestic firms report unfair practices byforeign competitors. In implementing US laws, theITA eventually faces a confrontation with foreign

217

E. Dal Bó

governments and firms, creating a cost to simplyyield to domestic pressure. The authors use over athousand decisions by the ITA (which may end upin the granting of protection or not) between 1980and 1990. They find that protective decisions aresignificantly associated with the industry’s PoliticalAction Committee (PAC) contributions to tradeoversight committee members. Further work ontrade policy has tested the predictions of Grossmanand Helpman’s (1994) menu auction, commonagency model. Goldberg and Maggi (1999), Gawandeand Bandyopadhyay (2000), and Eicher and Osang(2002) offer econometric studies of the determina-tion of imports, trade protection, and lobbying ex-penditures. They show that the pattern of non-tariffprotection and lobbying expenditures through indus-try PACs contributing to legislators fits the predic-tions of the Grossman and Helpman model. Sectorswith more elastic import demands are less pro-tected, the import penetration matters differently fororganized and unorganized industries, and industrylobbies seem to contribute in accordance to thestrength of opposing lobbies. For instance, upstreamfirms seeking protection lobby harder when theirpotential opposition (downstream firms) is moreconcentrated. The work studying how well theGrossman and Helpman (1994) set-up fits tradepolicy data tends to show that the model does a goodjob at accounting for the patterns of protection. Ibelieve it is noteworthy that a model that altogetherabstracts from asymmetric information explainstrade policy well. It is unclear, however, to whatextent the inclusion of ‘transparency’ variablescould improve explanatory power.

(ii) Revolving Doors and Regulators’ PersonalCharacteristics

Interest in the empirics of revolving doors emergedfirst in political science. Gormley (1979) presents astudy of the voting patterns of the seven membersof the Federal Communications Commission (FCC)between 1974 and 1976. He focuses on the idea thatindustry background may affect the behaviour ofregulators. Thus, his work lies at the intersection ofthe concern for the revolving doors and the roleplayed by personal characteristics. Given that, in astrict sense, only two of the seven commissioners inGormley’s study were former broadcasters, theanalysis should be seen as a quantitative case study.Gormley finds that former broadcasters do tend to

vote in support of the broadcasting industry moreoften. For instance, they voted to renew broadcastlicenses 88.7 per cent of the time, while othercommissioners only supported that decision 69.2 percent of the time. However, Gormley also finds thatparty allegiance matters as much, and for somedecisions, more. Republicans tend to vote consist-ently in the interests of industry. Thus, Gormley’sconclusion is that although industry backgroundseems to matter, it is not clear that it has a verystrong effect once one considers the role of politicalaffiliations. On a similar note, Navarro (1982) re-ports that liberal PUCs tend to create a more ‘ratesuppressing’ climate (i.e. to bring consumer pricesdown). Beyond the incentives connection high-lighted by the capture model reviewed in section III,the evidence seems to indicate that the personalcharacteristics of regulators, as well as the endow-ments and capabilities of regulatory agencies, playa potentially crucial role. In this connection, profes-sionalism appears to have non-innocuous effects.Berry (1979) presents evidence compatible with theidea that regulatory commissions enjoying higherdegrees of professionalism (as related to budgetsand recruitment policies) tend to reduce prices toconsumers relative to large users. This is especiallythe case when the value of power consumption ishigher and hence more salient to consumers.

Cohen (1986) revisits the issue of whether industryemployment may affect decisions in the FCC, usingdata from 1955 to 1974. He analyses both theproblem of previous and posterior industry employ-ment. In addition, he looks at effects at differentpoints along agency tenure. He finds that commis-sioners with previous industry experience are moresupportive of industry interests all along their ca-reers as regulators. As with Gormley’s study, it isdifficult to tell to what extent industry backgroundreally matters, because it is highly correlated withthe party affiliation of commissioners. In fact, noDemocratic administration appointed a commis-sioner with industry background during the sampleperiod. Commissioners that take industry employ-ment after leaving office are less supportive ofindustry in average, which may be surprising. Oneinterpretation is that such difference supports Che’s(1995) model: more-able regulators planning to ob-tain post-agency employment in industry signal theirquality by being tougher. However, commissionerswho take industry employment after leaving the

218


FCC increase their support for industry interests bynearly 11 per cent during their last year in office.This may suggest two things. Maybe explicit quidpro quos are realized close to the end of tenure,when the terms of such exchanges can be madeconcrete. Alternatively, consider a commissionernear the end of his term who finds out that alterna-tive employment opportunities are dwindling andrealizes he has been tough. That person may welldecide unilaterally to improve his profile as ‘employ-able’ by voting in support of industry.

One interesting connection is between the revolv-ing-doors phenomenon and the choice of term lengthsin regulatory commissions. If terms are very short,then banning former regulators from taking post-agency employment in industry may be difficult tosustain in practice. High-ability individuals maysimply decide to forgo agency employment in ordernot to lose their chance of obtaining industry jobs.On the other hand, the combination of short termsand the possibility of moving on to industry may leadregulators to care too much about their reputationswith the market. Leaver (2002) studies the effect ofregulatory term limits on the incidence of utility ratereviews in the United States.12 She analyses a panelof 99 electric utilities serving 39 states from 1982 to1990, which are under the regulation of term-limitedPublic Utility Commissioners appointed at the statelevel. The main finding is that reductions in termlimits appear associated with a decrease in thepropensity of regulators to file for rate reviews insituations of falling operating costs (which is whenrate reviews would be bad for firms).

(iii) Advocate Groups, Regulator Selection,and Consumer Power

One important issue that remains to be covered isthe extent to which regulatory outcomes are de-pendent on pressure stemming from consumers.Two main possibilities for consumer control areexamined. One is the creation of consumer advo-

cate groups; the other is the direct election ofregulators.

The oil shocks in the 1970s ended a long period ofdeclining costs and faced consumers with a sce-nario where firms filed for rate reviews and typicallyobtained permission to raise prices. Many statelegislatures in the United States then created inde-pendent consumer advocates in order to protectconsumers from what was perceived as one-sidedlobbying. These groups were officially funded, hadthe right to be automatically involved in agencyhearings, and could bring in information that coulddiminish informational asymmetries favouring firms.Holburn and Spiller (2002) present evidence sug-gesting that the creation of those consumer advo-cates had an important impact on electricity prices.Holburn and Vanden Bergh (2006) investigate theemergence of independent consumer advocates inthe USA. They find that Democratic state legisla-tures were more likely to fund consumer advocategroups. This seems especially true in states andperiods in which Democrats saw their grasp onpower as transitory. Creating consumer advocategroups appears to have been a way to insulateconsumer protection against probable adverse po-litical developments. This expectation by Demo-crats seems to have played a role especially in thecreation of advocate groups representing residentialconsumers, arguably the most likely to becomeunrepresented when the legislative majority changedcolours.

A different way in which consumers may haveregulation going their way is by controlling regula-tors through the vote. It is not obvious whether theeffects operate through the provision of differentincentives or through the selection of a differenttype of regulator. Still, most work in this area findsevidence consistent with the idea that selectionmethods matter. Most of this work has been doneusing cross-sections of utilities, which is potentiallyproblematic because of omitted variables problems.

12 She raises the point that minimal squawk theory is at odds with classic capture theory. According to Stigler (1971), the longera regulator stays in office, the cosier the relationship with firms gets. Presumably, very long terms allow this, so one could observea positive correlation between regulatory commissioners facing long terms and policies being more generous to firms (Hagermanand Ratchford (1978) do find evidence in that direction, using a cross-section of 79 electric utilities in the USA). In contrast, ifregulators care more about protecting their reputations with the market, when term limits are shorter these concerns may be madestronger, because regulators will be on the job market sooner. Then it should be shorter terms in office that give incentives to dofavours for firms, in order to stop firms from squawking.

219

E. Dal Bó

Hagerman and Ratchford (1978) use a cross-sec-tion of 79 electric utilities to investigate determinantsof allowed rates of return and find no evidence thatthe method used to appoint regulators matters.13

Boyes and McDowell (1989), however, find thatwhen regulators appointed by the executive requirethe approval of the legislature, the price of electricitytends to be lower, suggesting that legislatures chan-nel some consumer pressure on executive appoint-ees. Smart (1994) investigates pressures to keepbasic rates down in telecommunications followingthe divestiture of AT&T. Maintaining low basicrates became difficult with the introduction of com-petition in other service areas. She argues thatallowing for basic rate increases would be hardestfor regulators running in specific elections becauseof the salience of the issue to those voting in them.On the other hand, an appointee of the governorwould have more political room to let prices go up.After all, the governor is elected on a platform inwhich telecom prices is only one of many issues, andthat very issue may not be salient to the majority ofvoters. (An argument along these lines would belater formalized by Besley and Coate (2003).) Anintermediate institutional case would be that ofgovernor appointees that require legislative confir-mation, especially when this confirmation must takeplace in a divided legislature. Using a cross-sectionof telecommunication companies, she finds evi-dence that appointed regulators are associated withhigher prices for users in small cities, although thiseffect is undone when the regulator must be con-firmed by the legislature. Kwoka (2002) uses across-section of 543 US electric utilities and findsthat elected commissioners are associated withlower prices.

A different type of evidence is provided by Fields etal. (1997), who perform an event study and showthat a substantial drop in the market value of lifeinsurance companies followed the passage of Propo-sition 103 in California. This proposition altered themechanism for selecting life-insurance regulators ina way that was presumed to allow for more con-sumer influence. It is unclear, however, whether thedetected effect reflected a more pro-consumerstance of regulators or, alternatively, more unpre-dictable or incompetent regulation.

Besley and Coate (2003) attack the question ofwhether selection methods matter using a panel ofpass-through decisions regarding fossil-fuel costs.Their panel covers 42 states in the United Statesfrom 1960 to 1997. They find that elected regulatorsallow half as much pass-through relative to ap-pointed regulators. An attractive feature of thisstudy is that by interacting the selection method withcost changes it is possible to identify the role of theselection method even when including fixed effectsby state in a sample where no states ever switch theselection method.

It has been noted that the determination of regula-tory outcomes is not solely controlled by regulatorsand, possibly, the legislators that supervise them. Infact, the courts may affect outcomes because firms,regulators, and legislators interact under the shadowof an eventual court intervention (see, for instance,Gely and Spiller, 1990; Spiller and Tiller, 1997,1999). As a result, the method of selecting judgescould also affect regulatory outcomes. Guerriero(2003) provides a panel-data study exploiting statesthat switched judge-selection methods. His evi-dence indicates that states that elect their judgestend to allow lower pass-through of costs intoelectricity prices. One problem with the work onselection methods is that it highlights a relativedifference only. When elected regulators choose‘populist’ prices, they may have in mind the righttrade-off between present and future consumers,and more generous pricing by appointed regulatorsmay be a giveaway to industry. It is also possible thatpopulist pricing is too costly from the social point ofview because it may lower investment and hinderfuture service. Analysing regulatory outcomes fromthe point of view of some social efficiency benchmarkappears to be a large pending task for future research.

VII. SUMMARY AND PATHS FORFUTURE RESEARCH

I have attempted two main tasks in this review. Onewas to present in an accessible way the maintheoretical frameworks that academic economistsuse to think about regulatory capture. The other onewas to review available evidence on what factors

13 For other early studies, see Harris and Navarro (1983), Costello (1984), Primeaux and Mann (1986), and Atkinson and Nowell(1994).

220


may affect capture and what impact capture mayhave on regulatory outcomes. The main lessonsfrom prevailing theories are that capture is possiblebecause firms have private information that is hardfor citizens or their political representatives to ob-tain. This implies that the emergence of regulatoryagencies should be associated to the production ofuseful, industry-specific, information. However,regulatory agencies can be captured. An example iswhen firms induce regulators to hide informationthat could be used to offer consumers a better deal.

The theory we reviewed in section III prescribesresponses that tie regulators’ pay to their informa-tional performance. Such use of wage incentivesappears impractical, given the difficulty of contract-ing on regulatory outcomes and on the informationproduced by agencies. Alternatives based on theuse of above-market wages and monitoring to in-duce regulators to want to keep their posts alsoappear limited when taking into account the size ofthe stakes involved. These responses may thenhave to be complemented with more-involved insti-tutional building: the creation of bureaucratic proce-dures that allow various stakeholders to share infor-mation, the creation of legislative committees thatspecialize in monitoring the regulator, and possiblythe creation of consumer advocate groups. Theseelements are all present, to some extent and withregional variations, in the current regulatory com-plex of the United States.

Other theories emphasize that the provision ofpositive incentives (such as bribes or future industryemployment) may not be the only way to influenceregulators. The provision of ‘trouble’ may alsowork. The provision of trouble, for instance in theform of actions that trigger a reputational damage(or a direct utility loss) for the regulator, has thepotential negatively to affect the pool of talent thatcan be attracted to regulation. This will be especiallyimportant if less talented regulators are more gullibleand therefore more vulnerable to informational in-fluence from political and industry sources. Nega-tive incentives can play a larger role when regula-tory employment is weak in terms of stability or timehorizons. In such situations regulators may havemore to lose from harming powerful firm interests.The view that collusion will be easier to sustain inlong-run relationships calls for term limits on agency

employment. However, term limits that are too shortmay make regulators hostages to an extreme neednot to ‘rock the boat’. The optimal tenure lengthshould balance the two opposing concerns that havebeen studied.

In addition to incentives, interest groups may alsooffer information with the hopes of ensuring morefavourable treatment. One possibility that is, to myknowledge, uninvestigated empirically is that firmsmay out-consult regulators. Thus, regulators maycome to view the world the way firms do, notbecause they have been captured through incen-tives, but because they have been convinced. Moreevidence on persuasion and on whether bad fundingof regulatory agencies makes it more likely would bevaluable.

The empirical evidence on the causes and conse-quences of regulatory capture is scarce. Interestingwork has been done on the role of monitoring atcurbing graft and procurement abuses, but directtests are still lacking of how asymmetric informationraises the probability of regulatory capture. More-over, common-agency models that abstract fromasymmetric-information problems have been shownto fit the data on trade policy quite well. We choseas our main model in section III one that putsasymmetric information at its centre, because itwould seem that capture would be very unlikelywere wrongdoing immediately visible to citizens.(The only exception would be when citizens arepowerless to punish governments that are known tobe ineffective.) However, empirical research hasnot yet fully confirmed this presumption.

The empirical literature on capture, especially in itsearly years, devoted more attention to the role of theindividual characteristics of regulators and the ef-fects of the revolving doors. Those with industrybackground have been shown to be more lenienttowards industry, but not particularly so once onecontrols for political affiliations. Posterior industryemployment does not seem to have a robust effecton regulatory decisions, except in the last year ofregulatory tenures. Term limits in regulatory bodies,however, seem to make regulation more lenient,perhaps because regulators are more concernedabout their reputations with the market. An impor-tant open question in connection with the revolving

221

E. Dal Bó

doors is whether firms employ former regulatorsmainly because of their technical expertise, or be-cause of their lobbying potential.

The personal characteristics of regulators may beresponsible for the observation that the way inwhich regulators are appointed seems to make adifference to regulatory outcomes. Much in thesame way in which Supreme Court judges revealthe inclination of the presidents that appoint them,regulators may owe their stance to the governors orlegislatures that place them in their jobs.14 Recentwork has shown that elected regulators tend to havea more pro-consumer stance. This may reflect boththe fact that regulators are responsive to the prevail-ing pressures (and the electorate is a powerful one),or that pro-consumer regulators self-select intorunning for office. More work is needed to disentan-gle the selection and incentives effects stemmingfrom different methods for selecting regulators.Whether having elected regulators is a good ideamay depend on whether the main concern is captureby firms or pandering to populist views on utilitypricing. Beyond ideology, evidence is extremelyscarce on whether the talent, professionalism, andfunding of regulators matter, providing another pos-sibility for future empirical research.15

A final, important issue to consider is that of thecosts of capture. Is capture a big problem? On theone hand, capture could have large distributiveconsequences when it transfers income from, say,consumers to firms. This could be enough of areason for the political principal to want to curbcapture. Moreover, capture may cause net wealthlosses. Measuring these is another area in whichmore research should be done. Lacking a directestimate of how much wealth is destroyed bycaptured regulation, I would like to offer an indirectindication that capture can be very costly to society.

I will do this by relying on two studies that offer amicro view of the connection between corruptionand firm behaviour.16 The first study, by Kwhajaand Mian (2005) does not involve regulated firms,but shows that when captured officials grant rentsto firms the costs to society can be large. Theseauthors study distortions in the allocation of loans bypublic banks to firms with political connections inPakistan. The authors estimate that the politicallyinduced distortions generate social costs rangingfrom 0.3 to 1.9 per cent of GDP—figures that reachmacroeconomic relevance. The second study, byDal Bó and Rossi (2004), uses a measure of corrup-tion that proxies the corruptibility of all public offi-cials and not just regulators, but contains evidenceon the impact of corruption on regulated utilities.They find that if the median Latin American countryin terms of corruption in their sample (Brazil) had thecorruption level of the least corrupt country (CostaRica), then electricity distributors would use around12 per cent fewer employees and would incur 23 percent less operation and maintenance expenditures.This microeconomic evidence is of limited scope interms of external validity: it covers a small subset ofthe firms operating in the countries under study.Clearly, more studies should be performed to com-plete our picture on the microeconomic conse-quences of corruption. However, the magnitude ofthe effects is too large to be dismissed. Accordingto Dal Bó and Rossi (2004), the damage done bycorruption to the technical efficiency of regulatedutilities is similar in magnitude to that done byoperating in an environment characterized by poorlaw and order, and it is orders of magnitude largerthan the damage done by macroeconomic instabil-ity. Large efforts in both academia and policy circleshave gone into improving macroeconomic policy,and for good reason. The evidence suggests that weshould probably do more to understand and curbregulatory capture.

14 Alt and Lassen (2005) find evidence that the selection method for state Supreme Court members in the United States mattersto the corruption levels prevailing in the state. They also provide an indication that the selection effect is likely to be driving theresults.

15 See Domah et al. (2002) for a study of determinants of the size of electricity regulatory agencies across countries. Their evidencesuggests that smaller countries face a disadvantage at staffing regulatory agencies.

16 Svensson (2005) includes a review of the costs of corruption from the macroeconomic perspective. On procurement, see DiTella and Schargrodsky (2003), who show how prices for hospital supplies in Buenos Aires went down by about 15 per cent undera crackdown on corruption. Reinikka and Svensson (2004) show that official graft in Uganda in the mid-1990s absorbed around80 per cent of the public funds from a central government grant intended to finance schools.

222


.1

ccp hh ∆γ−

γ+=

APPENDIX

(i) Optimal Contracts under Asymmetric Information

The four constraints that the contract offered by the government must respect are:

Πl(pl, Tl) ≥ Πl(ph, Th), incentive compatibility constraint of low type (IC low);Πh(ph, Th) ≥ Πh(pl, Tl), incentive compatibility constraint of high type (IC high);Πl(pl, Tl) ≥ 0, participation constraint of low type (PC low);Πh(ph, Th) ≥ 0, participation constraint of high type (PC high).

The government wants to maximize

γ[s(pl) – Tl] + (1 – γ) [s(ph) – Th].

PC high implies Th ≥ F – q(ph) (ph – ch), and IC low means Tl ≥ F – q(pl) (pl – cl) + q(ph)∆c. Raising rentsin the high state only makes it more costly to deter the low-cost firm from claiming costs are high. Thegovernment will set Th = F – q (ph) (ph – ch). On the other hand, the government has a clear incentive tolower Tl as much as possible while satisfying IC low. Hence, the government will set Tl = F – q(pl) (pl –cl) + q (ph)∆c. From the last expression we see that, given q'(.) < 0, setting pl > cl could save the governmentsome transfers. However, departing from marginal cost is a more expensive way of maximizing s (pl) –Tl than setting pl = cl and then satisfying IC low by making direct transfers. The reason is that departuresfrom marginal cost pricing destroy surplus. Hence, the government will set pl = cl and Tl = F + q (ph)∆c.The problem for the government then becomes to maximize

γ[s(cl) – F – q(ph)∆c] + (1 – γ) [s(ph) – F + q(ph) (ph – ch)].

The first-order condition for this problem is

–γq'∆c – (1 – γ) q(ph) + (1 – γ) q'(ph – ch) + (1 – γ) q(ph) = 0

which then yields

(ii) Collusion-proof Optimal Contracts

The government wants to maximize

EV = γλ{s(cl) – F – w} + (1 – γλ) { µ[s (cl) – Tl (µ)] + (1 – µ) [s (ph) – Th (µ)]},

where, following the logic used before, Tl (µ) = F + q (ph)∆c, and Th (µ) = F – q (ph)(ph – ch). Consideringthat wages will be set at the minimum compatible with collusion proofness, the objective of the governmentbecomes to choose ph to maximize

The first-order condition is

.)])(()()[1(

])()([)1(

1)()(

−+−µ−++∆−−µ

γλ−+

ψ+∆

−−γλ=hhhh

hlhl cppqFps

cpqFcscpqFcsEV

223

E. Dal Bó

REFERENCES

Ades, A., and Di Tella, R. (1997), ‘National Champions and Corruption: Some Unpleasant Interventionist Arithmetic’,Economic Journal, 107(443), 1023–42.

— — (1999), ‘Rents, Competition and Corruption’, American Economic Review, 89(4), 982–94.Alt, J., and Lassen, D. (2005), ‘Political and Judicial Checks on Corruption: Evidence from American State Governments’,

Harvard Government Department, mimeo. — — (2006), ‘Transparency, Political Polarization, and Political Budget Cycles in OECD Countries’, forthcoming

in American Journal of Political Science.Armstrong, M., and Sappington, D. (2005), ‘Recent Development in the Theory of Regulation’, UCL, mimeo (prepared

for M. Armstrong and R. Porter (eds), Handbook of Industrial Economics, Vol. III).Atkinson, S., and Nowell, C. (1994), ‘Explaining Regulatory Commission Behavior in the Electric Utility Industry’,

Southern Economic Journal, 60(3), 634–43.Austen Smith, D., and Wright, J. (1992), ‘Competitive Lobbying for a Legislator’s Vote’, Social Choice and Welfare,

9, 229–57.Baron, D., and Myerson, R. (1982), ‘Regulating a Monopolist With Unknown Costs’, Econometrica, 50, 911–30.Becker, G. (1968), ‘Crime and Punishment: An Economic Approach’, Journal of Political Economy, 76(2), 169–217. — (1983), ‘A Theory of Competition Among Pressure Groups for Political Influence’, Quarterly Journal of

Economics, 98(3), 371–400. — Stigler, G. (1974), ‘Law Enforcement, Malfeasance, and the Compensation of Enforcers’, Journal of Legal

Studies, 3, 1–19.Bennedsen, M., and Feldmann, S. (2002), ‘Lobbying Legislatures’, Journal of Political Economy, 110(4), 919–46.Berg, S., and Tschirhart, J. (1988), Natural Monopoly Regulation: Principles and Practice, Cambridge, Cambridge

University Press.Bernheim, D., and Whinston, M. (1986), ‘Menu Auctions, Resource Allocation and Economic Influence’, Quarterly

Journal of Economics, 101(1), 1–31.Berry, W. (1979), ‘Utility Regulation in the States: The Policy Effects of Professionalism and Salience to the Consumer’,

American Journal of Political Science, 23(2), 263–77.Besley, T., and Coate, S. (2003), ‘Elected versus Appointed Regulators: Theory and Evidence’, Journal of the

European Economic Association, 1(5), 1176–206. — McLaren, J. (1993), ‘Taxes and Bribery: The Role of Wage Incentives’, The Economic Journal, 103(416), 119–

41.Boyes, W., and McDowell, J. (1989), ‘The Selection of Public Utility Commissioners: A Reexamination of the Importance

of Institutional Setting’, Public Choice, 61(1), 1–13.Calvert, R. (1985), ‘The Value of Biased Information: A Rational Choice Model of Political Advice’, Journal of Politics,

47, 530–55.

0)]}())((')()[1(])('[){1(1

)('=+−+−µ−+∆−µγλ−+

ψ+∆

γλ− hhhhhhh pqcppqpqcpqcpq

,0))(1()1(1

)(' =

−γ−+∆µγλ−−ψ+

∆γλ− hhh cpccpq

.1

11)1(1

cccp hh ∆

ψ+

−γ−

γλ+∆

γ−γ

+=

which becomes

which then, using the fact that 1 – µ = (1–γ)/(1–γλ) yields

224


Che, Y.-K. (1995), ‘Revolving Doors and The Optimal Tolerance for Agency Collusion’, Rand Journal of Economics,26(3), 378–97.

Cohen, J. (1986), ‘The Dynamics of the “Revolving Door” on the FCC’, American Journal of Political Science, 30(4),689–708.

Costello, K. (1984), ‘Electing Regulators: The Case of Public Utility Commissioners’, Yale Journal on Regulation, 2,83–105.

Cremer, J. (1986), ‘Cooperation in Ongoing Organizations’, Quarterly Journal of Economics, 101(1), 33–50.Dal Bó, E. (2000), ‘Bribing Voters’, Department of Economics Working Paper 39, University of Oxford. — Di Tella, R. (2003), ‘Capture by Threat’, Journal of Political Economy, 111(5), 1123–54. — Rossi, M. (2004), ‘Corruption and Inefficiency: Theory and Evidence From Electric Utilities’, University of

California, Berkeley, mimeo. — Dal Bó, P., and Di Tella, R. (2006), ‘Plata o Plomo?: Bribe and Punishment in a Theory of Political Influence’,

American Political Science Review, 100(1), 41–53.De Figueiredo, R., and Edwards, G. (2005), ‘Does Private Money Buy Public Policy?: Campaign Contributions and

Regulatory Outcomes in Telecommunications’, University of California, Berkeley, mimeo.Demsetz, H. (1968), ‘Why Regulate Utilities’, Journal of Law and Economics, 11, 55–6.Denzau, A., and Munger, M. (1986), ‘Legislators and Interest Groups: How Unorganized Interests get Represented’,

American Political Science Review, 80(1), 89–106.Di Tella, R., and Schargrodsky, E. (2003), ‘The Role of Wages and Auditing During a Crackdown on Corruption in the

City of Buenos Aires’, Journal of Law and Economics, 46(1), 269–92.Dixit, A., Grossman, G., and Helpman, E. (1997), ‘Common Agency and Coordination: General Theory and Application

to Government Policy Making’, Journal of Political Economy, 105, 752–69.Domah, P., Pollit, M., and Stern, J. (2002), ‘Modelling the Costs of Electricity Regulation: Evidence of Human Resource

Constraints in Developing Countries’, Department of Applied Economics Working Paper 229, University ofCambridge.

Eicher, T., and Osang, T. (2002), ‘Protection for Sale: An Empirical Investigation: Comment’, American EconomicReview, 92(5), 1702–10.

Fields, J., Klein, L., and Sfiridis, J. (1997), ‘A Market Based Evaluation of the Election versus Appointment of RegulatoryCommissioners’, Public Choice, 92(3–4), 337–51.

Gawande, K., and Bandyopadhyay, U. (2000), ‘Is Protection for Sale? Evidence on the Grossman–Helpman Theoryof Endogenous Protection’, Review of Economics and Statistics, 82(1), 139–52.

Gely, R., and Spiller, P. (1990), ‘A Rational Choice Theory of Supreme Court Statutory Decisions with Applicationsto the State Farm and Grove City Cases’, Journal of Law, Economics, and Organization, 6(2), 263–300.

Goldberg, P., and Maggi, G. (1999), ‘Protection for Sale: An Empirical Investigation’, American Economic Review,89(5), 1135–55.

Gormley, W. (1979), ‘A Test of the Revolving Door Hypothesis at the FCC’, American Journal of Political Science,23(4), 665–83.

Grossman, G., and Helpman, E. (1994), ‘Protection for Sale’, American Economic Review, 84, 833–50.Guerriero, C. (2003), ‘Dependent Controllers and Regulation Policy: Theory and Evidence’, Cambridge University,

mimeo.Hagerman, R., and Ratchford, B. (1978), ‘Some Determinants of Allowed Rates of Return on Equity to Electric Utilities’,

Bell Journal of Economics, 9(1), 46–55.Hansen, W., and Park, K. O. (1995), ‘Nation-state and Pluralistic Decision-making in Trade Policy: The Case of the

International Trade Administration’, International Studies Quarterly, 39(2), 181–211.Harris, M., and Navarro, P. (1983), ‘Does Electing Public Utility Commissioners Bring Lower Electric Rates?’, Public

Utilities Fortnightly, 112, 23–88.Hilton, G. (1972), ‘The Basic Behavior of Regulatory Commissions’, American Economic Review, 62, 47–54.Holburn, G., and Spiller, P. (2002), ‘Interest Group Representation in Administrative Institutions: The Impact of

Consumer Advocates and Elected Commissioners on Regulatory Policy in the United States’, University ofCalifornia Energy Institute, mimeo.

— Vanden Bergh, R. (2006), ‘Consumer Capture of Regulatory Institutions: The Creation of Public UtilityConsumer Advocates in the United States’, Public Choice, 126, 45–73.

Joskow, P. (2005), ‘Regulation of Natural Monopolies’, MIT, mimeo (prepared for A. M. Polinsky and S. Shavell (eds),Handbook of Law and Economics, Elsevier).

225

E. Dal Bó

Khwaja, A., and Mian, A. (2005), ‘Do Lenders Favor Politically Connected Firms? Rent Provision in an EmergingFinancial Market’, Quarterly Journal of Economics, 120(4), 1371–411.

Kofman, F., and Lawarree, J. (1993), ‘Collusion in Hierarchical Agency’, Econometrica, 61(3), 629–56.Kwoka, J. (2002), ‘Governance Alternatives and Pricing in the US Electric Power Industry’, Journal of Law, Economics

and Organization, 18(1), 278–94.Laffont, J.-J., and Martimort, D. (1999), ‘Separation of Regulators against Collusive Behavior’, Rand Journal of

Economics, 30(2), 232–62. — Tirole, J. (1993), A Theory of Incentives in Procurement and Regulation, Cambridge, MA, MIT Press.Leaver, C. (2002), ‘Bureaucratic Minimal Squawk Behaviour: Theory and Evidence from Regulatory Policy’, University

of Oxford, mimeo.Lohmann, S. (1993), ‘A Signaling Model of Informative and Manipulative Political Action’, American Political Science

Review, 88, 319–33.Lowry, R., and Alt, J. (2001), ‘A Visible Hand? Bond Markets, Political Parties, Balanced Budget Laws and State

Government Debt’, Economics and Politics, 13(1), 49–72.McCraw, T. (1975), ‘Regulation in America, a Review Article’, The Business History Review, 49(2), 159–83.McCubbins, M., Noll, R., and Weingast, B. (1987), ‘Administrative Procedures as Instruments of Political Control’,

Journal of Law, Economics and Organization, 3(2), 243–77.Myerson, R. (1979), ‘Incentive Compatibility and the Bargaining Problem’, Econometrica, 47, 61–74.Navarro, P. (1982), ‘Public Utility Commission Regulation: Performance, Determinants, and Energy Policy Impacts’,

The Energy Journal, 3(2), 119–39.Neeman, Z. (1999), ‘The Freedom to Contract and the Free Rider Problem’, Journal of Law, Economics, and

Organization, 15(3), 685–703.Olson, M. (1965), The Logic of Collective Action, Cambridge, MA, Harvard University Press.Peltzman, S. (1976), ‘Toward a More General Theory of Regulation’, Journal of Law and Economics, 19, 211–48.Posner, R. (1974), ‘Theories of Economic Regulation’, Bell Journal of Economics and Management Science, 5(2), 335–

58.Primeaux, W., and Mann, P. (1986), ‘Regulator Selection Methods and Electricity Prices’, Land Economics, 62(1), 1–

13.Reinikka, R., and Svensson, J. (2004), ‘Local Capture: Evidence From a Central Government Transfer Program in

Uganda’, Quarterly Journal of Economics, 119(2), 679–705. — — (2005), ‘Fighting Corruption to Improve Schooling: Evidence from a Newspaper Campaign in Uganda’,

Journal of the European Economic Association, 3(2–3), 259–67.Salant, D. (1995), ‘Behind the Revolving Door: A New View of Public Utility Regulation’, Rand Journal of Economics,

26(3), 362–77.Smart, S. (1994), ‘The Consequences of Appointment Methods and Party Control for Telecommunications Pricing’,

Journal of Economics and Management Strategy, 3(2), 301–23.Snyder, J. (1991), ‘On Buying Legislatures’, Economics and Politics, 3, 93–109.Spiller, P. (1990), ‘Politicians, Interest Groups, and Regulators: A Multi-principals Theory of Regulation, or “Let Them

Be Bribed”’, Journal of Law and Economics, 33(1), 65–101. — Tiller, E. (1997), ‘Decision Costs and the Strategic Design of Administrative Process and Judicial Review’,

Journal of Legal Studies, 26(2), 347–70. — — (1999), ‘Strategic Instruments: Legal Structure and Political Games in Administrative Law’, Journal of Law,

Economics and Organization, 15(2), 349–77.Stigler, G. (1971), ‘The Theory of Economic Regulation’, Bell Journal of Economics and Management Science, 2, 3–

21.Svensson, J. (2005), ‘Eight Questions About Corruption’, Journal of Economic Perspectives, 19(3), 19–42.Tirole, J. (1986), ‘Hierarchies and Bureaucracies: On the Role of Collusion in Organizations’, Journal of Law, Economics

and Organization, 2(2), 181–214.

regulatory capture: a review - berkeley haas

Documents