Edited by Gerrit De Geest
Chapter 5: Production of legal rules by agencies and bureaucracies
[In: Volume 7, Francesco Parisi (ed) Production of Legal Rules]
The central question posed in the literature on the production of rules by agencies is how the discretion of law-making agencies can be curbed. The political preferences of bureaucracies are most often taken to be exogenously given. These may be the personal political preferences of the director of a single agency or may reflect the influence of interest groups. The influence of interest groups on rule-making agencies has been studied in a descriptive way by Furlong (1997) and Furlong and Kerwin (2005), but is rarely modeled explicitly in this literature. Based on more or less complete answers to the central question, explanations are given for why a legislator may want to delegate law-making power to bureaucracies and how control of this delegated power is to be organized. The key normative question is how much delegation there ought to be. However, answers given to this question barely rely on the positive analyses mentioned herein.
Current positive law-and-economics analyses of how agencies and bureaucracies (the terms are used as synonyms here) produce legal rules differ substantially from approaches applied to other law-making institutions. These analyses do not follow the Public Choice approach commonly used to explain the production of legal rules by legislators, nor do they follow the evolutionary approaches mostly used to explain production of rules by the court system. Theories dealing with bureaucratic law-making do not therefore derive one utility-maximizing position of bureaucratic decision makers (which are hardly ever modeled as multiple individuals); nor do these theories rely on the assumption that inefficient legal rules are challenged more often than efficient ones (so that, on average, legal rules tend to evolve towards an efficient state). Only where the study of production of rules by the legislator concentrates on the consequences of more than one institution needed to pass a law do strong similarities arise with approaches to bureaucratic rule production. There p. 63are no such similarities between theories of bureaucratic rule making and the evolutionary approach.
Comparing the literature on decision-making by agencies and bureaucracies with research on the production of legal rules by the legislator on the one side, and rule production by the judiciary on the other, not only demonstrates how the approaches to very similar problems differ, but also underlines a problem which seems to be purely semantic: bureaucratic rule-making and bureaucratic adjudication are analytically distinct, as are legislative rule-making and judicial adjudication. If bureaucratic decisions are rule-making, Public Choice theory suggests itself for tackling bureaucratic rule making; if they are adjudication, an evolutionary approach seems to be more appropriate. The problem arises in much of the theoretical literature and holds true for many empirical studies. Single bureaucratic decisions or entire classes of bureaucratic decisions are often analyzed without giving any thought as to whether these decisions are rule making or adjudication.
The reason for the reluctance to distinguish between rule-making and adjudication may be that the distinction is not as clear in American administrative law (see, for example, Rugge, 1987) as it would seem to be from the Administrative Procedures Act (5 USC §551 (4)–(7)). This lack of clarity may be a problem in other jurisdictions as well. As a result, very few authors in the field reviewed here mention the distinction at all (for example, Lupia and McCubbins, 1994b). It is therefore often not possible to restrict this review to the literature on bureaucratic rule-making; research with regard to decisions made in single cases has to be referred to in many places. Likewise, some of the literature reviewed here may overlap with the literature discussed in the articles on regulation of contracts and on environmental regulation. Yet the emphasis of this chapter will be different: it deals with the ‘production of legal rules by agencies and bureaucracies’, that is with the rule-making activities of agencies and bureaucracies. I will, however, avoid making too many references to scholarly works on bureaucracies in general, if there is no specific relation to rule-making (for an overview of this literature with special reference to Public Choice arguments, see Benson, 1995).
The chapter is organized as follows: Sections 2 and 3, respectively, review the literature that describes how procedural rules, by framing the production of substantive rules by agencies, facilitate the control of agencies by their principal, that is, the legislator, and writings comparing the different ways in which the legislator is able to control a rule-making agency. The following two sections, 4 and 5, concentrate on the literature breaking down the legislator into its different institutions, including the courts in the analysis of constraints on bureaucratic rule production. Section 4 deals with econometric studies, while Section 5 presents the spatial approach to the topic. Section 6 surveys works that ask normatively whether there ought to be delegation of rule production p. 64to agencies and bureaucracies at all. Section 7 reviews the few inquiries there have been into the interior process of agencies which produce rules, and Section 8 concludes.
2. Curbing Bureaucratic Discretion by Procedural Rules
Two distinct phenomena bring about bureaucratic discretion: one, the inability of legislative institutions to perfectly control the activities of bureaucracies at reasonable costs; the other, the need for cooperation among more than one legislative institution, to sanction an agency or to replace bureaucratic rules by new statutory rules. Although a clear distinction between these phenomena cannot be maintained in reviewing all publications, the first phenomenon is dealt with in this and the next section; the second, in Sections 4 and 5.
As stressed by McCubbins, Noll and Weingast (1987) (‘McNollgast’), the principal–agent relationship between legislators and bureaucracies cannot be dealt with appropriately by mere reference to general principal–agent models of the Laffont-Tirole-Holmström-Milgrom type. The reason is that ex post sanctions, which are imposed after the bureaucracy deviates from the political preferences of the politicians, are far too limited, in view of low probabilities of detection, high monitoring costs, and the costs which legislators themselves must bear to impose sanctions (additional workload, political backlash for poorly working agencies, further deterioration of the agency’s work). Note however, that Mitnick (1975) emphasizes that one has to be careful about the short-and long-run costs and benefits of control here: strict control by the principal may run up more costs than benefits in the short run, but be profitable in the long run because agents become ‘educated’ by control, and strict control may result in the self-selection of ‘good’ agents. In McCubbins, Noll and Weingast (1989), the authors extend the reasons for the inappropriateness of the general principal–agent model to embrace the phenomenon to be discussed extensively in Section 5 of this chapter: the need for coordination among several institutions if control of bureaucratic rule production is to be exercised by the shadow of overriding statutory legislation.
McNollgast argue that the remaining, large problems of compliance may be partly solved by procedural rules. In this respect, procedural rules governing the agency’s soliciting of information and the consequences of not doing so properly are most important: they may ensure that agencies base their decisions on large sets of information which would also be sufficient for legislative control. Soliciting information can only be successful if those who are better informed have enough incentives to disclose their information. This is, of course, the case if the stakes of affected groups in an issue are high or the issue is highly controversial. These variables, however, fail to be within the control of the legislator. The legislator is, thus, forced to concentrate on enacting procedural p. 65rules that modify the costs and the effectiveness of providing information to the bureaucracy.
McNollgast study such procedural rules. Public disclosure requirements facilitate the political intervention of interest groups at early stages in the rule-making process, even earlier than in the legislative process. Thus, political intervention of interest groups may allow ex ante control of bureaucratic rule production. Elaborate rules of procedure result in high costs of rent-seeking and, accordingly, favor those groups that have large resources to spend on representation. On the other hand, budget subsidies for poorly represented groups favor these groups, instead. Next, the discretion of an agency is limited by evidentiary standards, which refer to how evidence must be processed to come to a decision. Burden-of-proof rules may favor or hamper groups which base arguments on unclear cause–effect relationships. Further, ‘ex parte’ rules serve to prohibit direct informal communication between private parties and an agency during the formal rule-making process. These rules variously provide for sanctions against prohibited ex parte communications, which run the gamut up to, at the extreme, imposing a final decision against the infringing group. Ex parte rules enable politicians to participate in the rule-making process, at least as brokers of communication. Also, politicians may exert direct pressure on agencies. For all these procedural rules to enable politicians to exert ex ante control, a judicial remedy for their infringement must be sufficiently likely to occur. This can be achieved by decentralized enforcement of these rules if private parties are able to sue at low costs.
McNollgast not only examine ways in which procedural rules and interest groups interact to curb decision-making by bureaucrats and agencies, but also argue that politicians are not constrained to granting a policy which is currently favorable to some interest groups. McNollgast suggest politicians are able to ensure, for the future, that changes of legal (agency-made) rules are favorable to the interest groups that constitute the original coalition. This ‘autopilot’ for future decisions relieves the enacting coalition of the problem that future political majorities might favor other interest groups. DeShazo and Freeman (2005) extend the argument to competing agencies, which may be used to interfere with an agency’s rule-making in a similar way to lobbying interest groups. The problem of the ‘slack’ of future legislators as opposed to regulatory slack in rule-making is particularly stressed by Horn and Shepsle (1989) and by Steunenberg (1992). They emphasize that there is a trade-off between these two kinds of slack, because constraints on either the agency or future legislators require freedom to act by the other institution. DeShazo and Freeman (2003) draw their readers’ attention to a related problem which one might label ‘committee slack’, that is to the possible deviation of the agencycontrolling committees’ preferences from the legislator’s preferences.
p. 66McNollgast allow, against the thrust of their argument, that the control of agencies exercised by the interest groups may evolve in an unexpected direction, or be impeded altogether, if the interest groups that are engaged in the original deal with the legislator cease to exist. Macey (1992) discounts the importance of this caveat for three reasons, all related to the design of a newly formed agency, and thus strengthening the thrust of McNollgast’s argument: First, when interest groups agree with the legislator to institute a regulatory agency, the agreement is likely to cover institutional rules about which parties may influence the future agency’s decisions and the ways influence can be exerted. For example, the influence of poorly organized groups may be secured by facilitating access for individuals with an interest to agency hearings or judicial review of agency decisions. Levine (1992) criticizes this argument on the basis that, while one may predict that liberalized standing will increase the degree of interference by the courts with regulatory drift, one is unable to predict easily the direction of this interference. This argument apparently neglects the results of evolutionary approaches to the common law (compare the literature following Rubin, 1977 and Priest, 1977), where predictions on the directions of judicial rule-making are certainly possible, even if the court system is not systematically biased in some direction. If the court system is known to be biased, it is obvious that one may predict in what direction judge-made law will evolve.
Second, limiting the jurisdiction of a newly installed agency to the regulation of a small number of industries, or even to parts of a single industry, assures that the agency will recruit its expertise in this specialized field, mainly from the same industry or part of an industry. As a result, opinions held by the agency and industry will usually concur. Accordingly, at least regulation in favor of the regulated industry is stable over time.
Third, when agencies compete for jurisdiction with each other, they seek out the support of constituents and hence will not deviate from the preferences of their constituency. The agency and its constituent interest group(s) thus mutually support each other in order to further their own raison d’être and improve their chances of survival over time. This mutual support, together with the procedural rules mentioned above, serves as an ‘autopilot’ for rule-making by the agency. A similar argument had already been put forward by Benson (1981, 1983a and 1983b), although Benson did not state the argument in terms of curbing regulative drift. Benson focuses on the size of the bureau and the problem of over-regulation. Johnson and Libecap (1994) drive home the argument that the rank-and-file bureaucrats are interested in the duration of their agency to the point that they become an influential interest group, shaping long-run trends in bureaucratic institutions. With particular reference to the third argument (mutual support of regulator and regulated industry), Macey (1992) rejects Shepsle’s (1992) view (also expressed by Horn and Shepsle, 1989 and by Steunenberg, 1992) that more emphasis should be placed on the trade-off between regulatory p. 67and legislative slack. Macey claims that there is no such trade-off, but interest groups with political clout, which participate in the original, political decision to set up an agency, retain their political power through this mutual support and continue to exert a similar influence on the legislator as well as on the agency.
Although both of the McNollgast papers give case studies as empirical examples of their theory (see Basinger, 2003, for another, historical case study), Robinson (1989) casts severe doubts on the empirical relevance of McNollgast’s argument. McNollgast argue that procedural provisions serve mainly as a curb on the discretion of agencies. Robinson points out that while most of the examples cited by McNollgast do not support their theory, the examples introduced by Robinson himself provide even less support. He contends, however, that McNollgast’s approach could be very useful. If the empirical relevance of the argument was shown, it could substantially change and advance the understanding of the legal-administrative process.
In a related study, Rose-Ackermann (1995) compares how procedural rules enable interested (third) parties to control and interfere with bureaucratic rule-making in Germany and the US. In contrast to the situation in the US, in Germany individual citizens have little chance to control rule production by bureaucracies; this control only extends to applications of the rules. Rule-making by bureaucracies in Germany is typically under the influence of well-organized interest groups, in the same informal way as is statutory legislation in both countries. Only in the US do procedural rules give the power to control bureaucratic rule-making to interested parties which are not otherwise politically powerful in the legislative process.
Hamilton and Schroeder (1994) call into question this entire line of reasoning. They ask why agencies would follow procedural rules more strictly than substantive rules. Though they do not enter into a theoretical discussion of this point of view, they provide some empirical evidence that procedural rules are, in fact, circumvented. Agencies leave substantive rules incomplete and fill the gaps with extensive directives, guidance documents and policy memoranda, which fail to be rules in a formal sense and are not subject to formal rule-making procedures that require the participation of interest groups, in particular of those groups that are protected by procedural rules.
DeMesquita and Stephenson (2007) point out the risk that bureaucratic discretion may be over-controlled if oversight is restricted to observable activity by the bureaucracy, but the quality of bureaucratic actions also depends on unobservable activity. In the same vein as the multi-tasking problem modeled by Holmström and Milgrom (1991), they assume a partly benevolent regulator who inefficiently replaces observable with unobservable activity if controlled too tightly.
p. 68Jason MacDonald (2010) varies the approach by investigating the effect on agencies’ behavior of limitation riders, which allow the legislator to forbid agencies to spend money on specific uses for a certain length of time.
3. Police-Patrol Control versus Fire-Alarm Control
An issue closely related to the question of how procedural rules give control over bureaucratic rule-making to third parties is the question of how legislators can best control agencies. The problem posed is whether legislators take a random sample of bureaucratic activity (including its rule-making activity) in the same manner in which a police patrol randomly inspects the streets; or whether legislators look at bureaucratic decisions only if and when someone protests vehemently against a particular decision in the same way in which the fire department takes action only after someone sounds a fire alarm. I refer to this question as the optimal-type-of-control problem.
The leading articles that discuss the optimal-type-of-control problem (for an overview of the consequential literature until 1990, see Ogul and Rockmann, 1990) are McCubbins and Schwartz (1984) and Weingast (1984). These articles argue that, in general, Congress prefers fire-alarm to police-patrol oversight, because the costs of the monitoring activities are partly transferred from Congress itself to interested groups and individuals. However, citizens harmed by violations of legislative goals are not always represented by well-organized groups and, accordingly, are unable to sound alarms loud enough to obtain redress of grievances. The authors answer this criticism by referring to arguments like those in the previous section: disadvantaged groups often have public spokesmen; problems of collective action are overcome by legislation which favors the organization of interest groups, particularly those affected by collective action problems; extensive constituent service by congress(wo)men opens up review procedures.
In the optimal-type-of-control literature, there is no discussion of the social costs of fire-alarm oversight, but one might speculate that it is socially cheaper. With fire-alarm oversight, individuals, who are affected by bureaucratic rule-making and will notice it anyway, detect regulatory slack; with police-patrol oversight, legislators must fund specialized individuals to scrutinize bureaucratic decisions for which they have little inherent regard. The argument might, however, be weakened if one takes into account a problem like dispersed damages in tort law: the adverse effects of regulatory slack may be so dispersed that sounding a fire alarm is too costly for anyone other than specialists hired to search for regulatory slack. This assumption, once again to some extent, harks back to the argument in the previous section.
It is, however, unclear whether asymmetric degrees of organization among opposing interest groups can be completely offset by procedural rules. Hopenhayn and Lohmann (1996) study the results of asymmetric access to p. 69review procedures for a particular kind of fire-alarm oversight, where an agency’s decision can be (partly) controlled before there are adverse effects on social welfare. A fire alarm can be sounded either (i) only if the agency has allowed some regulated activity, even though the activity decreases social welfare or (ii) in addition to this condition, also if the agency prohibits a regulated activity, even though the activity would increase social welfare. If political institutions rely on this kind of fire-alarm oversight, fewer activities will be allowed in case (i) than in case (ii). Hence, there may be adverse social welfare effects due to reliance on fire-alarm oversight, as opposed to some symmetric oversight system, possibly a police-patrol system.
Aberbach (1990) questions the theoretical findings of McCubbins and Schwartz (1984) and Weingast (1984) on factual grounds. They argue that Congress prefers fire-alarm to police-patrol oversight; he offers empirical evidence that Congress has increasingly used police-patrol oversight instead of only fire-alarm oversight. Ogul and Rockmann (1990) explain this finding by pointing to the greater resources employed by Congress (in particular more staff) to scrutinize agencies, where fewer resources are allocated to new legislative activity because congress(wo)men have, in general, fewer opportunities to obtain name recognition through new legislation.
Lupia and McCubbins (1994a and 1994b) develop formal models of both police-patrol and fire-alarm oversight. They explore under what conditions each kind of control is superior to no oversight from the viewpoint of Congressas-principal. The hardly surprising conditions they derive for police-patrol oversight being superior to no oversight are: large differences between an agency’s and legislators’ preferences; and low costs of oversight. Fire-alarm oversight is superior to no oversight when high penalties are imposed for false fire alarms, or where the preferences of legislators and of interested groups or individuals that may raise the fire alarm are close to each other. Figueiredo et al. (1999) develop a variation of the model in which they allow both for fire-alarm monitors with or without own interests and for monitors with information which is better than the principal’s, but still incomplete.
Due to the disparate structure of their two models, Lupia and McCubbins fail to compare the two kinds of oversight with one another. After making exactly this point (Bawn, 1994), Bawn (1997) constructs another formal model to carry out this comparison. This model again mainly restates the arguments of McCubbins and Schwartz (1984) in a formal way. By emphasizing that marginal costs of oversight are lower for members of the committee with jurisdiction over the agency than for other members of the chamber, she adds, however, the insight that the latter will prefer more fire-alarm-like or (as Bawn calls it in a slightly broader sense) statutory control. Bawn provides some empirical evidence for her hypothesis by comparing the likelihood that floor amendments to EPA rules increase statutory control for sponsors that are members of the committee with p. 70jurisdiction and non-members. Balla (2000) casts doubts on these results by controlling for the distance between the sponsor’s and the median committee member’s preferences. In all his regressions, the latter difference in preferences has a statistically significant effect on the likelihood of (co-)sponsorship, while committee membership does not.
Both Lindsay (1976) and Holmström and Milgrom (1991) express fundamental doubts as to whether an optimal degree of control over agencies can be defined merely by taking into account the direct costs of control. They stress that agents that have multiple tasks, which the principal can monitor to different degrees, tend to carry out only tasks for which they will get credit, namely those tasks which the principal can monitor. Such agents neglect the tasks which the principal cannot monitor. By distorting the allocation of the agent’s efforts, additional oversight may thus lead to indirect costs. As there is no common denominator in which to measure the net benefits reaped in the public sector, such as profits, the problem they discuss is particularly relevant to government bureaucracies.
4. Who Controls Bureaucratic Discretion: Econometric Approaches
While the literature discussed so far concentrates on how to overcome bureaucratic discretion by assuming a dichotomy between agencies and the legislature (as a unitary actor), many authors also ask who exactly controls the bureaucracy: whether it is a particular institution among several engaged in the legislative process or none of these, but an independent judiciary. Two strands of literature, only loosely intertwined, must be distinguished: econometric studies (discussed in this section) and theoretical arguments, based on sequential games, which describe strategic choices made between policies, represented by different points in a given policy space (discussed in Section 5.)
Among the econometric approaches surveyed here, some authors ask which of several possible institutions controls an agency. Other authors test whether a particular institution controls an agency’s decisions or not. A typical and prominent example of the first type of approach is Moe (1985) (also compare Moe, 1987). He takes the National Labor Relations Board (NLRB) as an example of an agency which could be influenced by the political preferences of the President, of Congress, or of the courts. His analysis clearly shows that all these institutions are important (changes in political orientation have a significant impact on the decisions of the NLRB), but the President has the strongest impact on the NLRB. The NLRB is not a prominent example of rule production by agencies, because its decisions are nearly all adjudicatory. Notwithstanding this limitation, Moe’s paper may well be labeled as path-breaking: it lays the ground for later econometric studies to inquire which institutions wield the most power over agencies that make rules.
p. 71Beck (1987) employs a model that is less ambitious in comparing the power wielded over agencies by different institutions, but more directly focused on rule production. He studies to what extent the US President is able to influence rule-making within the Federal Reserve. In particular, he tests whether the monetary policy of the Federal Reserve supports a political business cycle favorable to incumbent Presidents. His findings are that the Fed does not actively support such cycles by inducing political monetary cycles, but that it passively accepts such cycles induced by fiscal policy. Accordingly, Beck asserts that the President of the US exerts a certain, but limited degree of power to control the Fed.
Probably the most extensive and detailed econometric study of bureaucratic rule production has been conducted by Magat, Krupnick and Harrington (1986, p. 8). Their book aims at clarifying which factors influence bureaucratic decisions with regard to rule-making. They test fourteen hypotheses derived from different theoretical approaches to bureaucratic rule-making. Unfortunately, few of the results of the study are statistically significant, but most parameters do have the sign to be expected from the theoretical argument. The same caveat holds for the work of Hamilton and Schroeder (1994), who study how much an agency will resort to ‘informal’ rule-making (in the form of directives, guidance documents and policy memoranda issued by an agency) in order to avoid the control of formalized rule-making procedures.
Drawing on studies of the Food and Drug Administration (FDA), Olson (1995 and 1996) does not so much consider which political institutions exert power in controlling agencies; rather she contrasts different approaches to predicting bureaucratic behavior. In particular, she compares capture theory (agencies are controlled by the industries they are supposed to regulate), public interest theory (agencies are the benevolent pursuers of social welfare), and Congressional control theory (legislative institutions, modeled by Olson as a unitary actor, control agencies). She finds that the behavior of one and the same agency is best explained by different theories or even by different combinations of theories for different industries which are regulated. Accordingly, she suggests that unless subsequent empirical studies focus on the micro-level trade-offs which face regulators, it is impossible to get a clear answer as to what are the important incentives in bureaucratic rule production. In order to formulate better explanations or even predictions of bureaucratic behavior, future research must be more sensitive to different regulatory actions and their specific constraints and trade-offs for the regulator.
Hammond and Knott (1996) reject completely econometric analyses of which institutions control the bureaucracy, based on the correlation between, on the one hand, changes in preferences of an institution, and, on the other hand, changes in policy of the agency. As they point out, it is impossible for one institution to control an agency without the cooperation of the other institutions. Thus, p. 72future empirical work on political control should incorporate the interplay of the preferences of the controlling institutions. These authors do not develop, however, an operational measure of the power wielded by institutions over agencies, although they consider such a measure to be indispensable in conducting empirical studies. Comparing the policy orientation of agencies’ monitors to the orientation and the likelihood of re-election of the governmentcontrolling party installing the monitor, Holburn and Vanden Bergh (2006) find empirical support for the hypothesis that procedural rules aim to control agencies in the medium run.
Furlong (1997) and McKay and Yackee (2007) offer empirical studies concentrating on the effectiveness of rent-seekers’ efforts to influence administrative rule-making. McKay and Yackee (2007) also find that the activity of one lobbying group fails to provoke counter-activity by opposing interest groups in a statistically significant way.
5. Who Controls Bureaucratic Discretion: Spatial Approaches
The literature reviewed in Sections 2 and 3 explained the discretion exercised by bureaucrats by highlighting the superior information which agencies possess and explored the ways in which control of the bureaucracy is achieved. The literature discussed in this section takes a game-theoretic approach. This approach is based on preferences in a given policy space to show that bureaucratic discretion exists even without asymmetric information. The basic idea developed in this body of literature is that an agency can be sanctioned, for example, by new legislation that overrules the decisions of an agency or interferes with the resources of the agency, only if more than one of the institutions agree to apply sanctions. If n institutions are needed to sanction an agency and every institution has single-peaked preferences in the policy space, Pareto sets of dimension n – 1 or less result. Within these Pareto sets, the agency is free to act because every change in policy makes at least one institution better off; accordingly, this ‘winning’ institution will be reluctant to sanction the agency. As legislative procedures become more complex than simple acceptance by all relevant institutions, the models are formed as sequential games. In these games, institutions can veto, accept or amend proposals from institutions with the right to take the initiative, according to (a simplified version of) the constitution.
The starting point for this discussion is a paper by Weingast (1981), in which he develops a spatial policy model (of one and two policy dimensions) that explains the bounds of agencies’ decisions on regulation. Based on three examples (airlines, telecommunications, and nuclear power), he studies the interplay between possibly more than one Congressional (sub)committee; interest groups which form the clientele of both the (sub)committee(s) and the agency controlled by the (sub)committee(s); and sometimes (namely if he makes appointments to the agency himself) the President. Weingast shows how p. 73this interplay restricts the range of discretion of the agency. Using the same approach, Weingast and Moran (1983) explain why there is scant evidence of Congressional oversight. They put forward this approach as an alternative to the dominant view in the literature at the time that this lack of evidence is due to major imperfections in the oversight exercised by Congress, due to Congress’s lack of information. This approach argues, alternatively, that Congressional oversight is near perfect; accordingly, agencies do not even attempt to deviate from the preferences of the legislature. Weingast and Moran demonstrate that their theory explains better than an imperfect-oversight theory why Congress was extraordinarily active in its oversight of the Federal Trade Commission (FTC) in 1979–80, after a turnover in the members of the appropriate Senate subcommittee. The FTC temporarily lacked sufficient information on new constraints imposed on its discretion by new Senate subcommittee members, and the FTC deviated involuntarily from the preferences of the subcommittee. Finally, Weingast and Moran test their theory econometrically. Their econometric analysis is challenged by Muris (1986) and successfully defended by Weingast and Moran (1986). Econometric support for their theory is also offered in a paper by Faith et al. (1982). They reject the claim of Katzman (1980) (who relies exclusively on a non-statistical analysis of interviews conducted with members of the FTC) that the decisions of the FTC are independent of Congressional influence and of efforts to maximize the budget of the Commission. Instead, they offer compelling econometric evidence that both the House and the Senate have a substantial impact on the selection of cases by the FTC, though the impact of the Senate seems to vanish after sweeping reforms of the FTC were carried out in 1970.
In ensuing years, the approach of Weingast has been criticized, mainly for its simplifying assumptions on Congressional legislative procedures (see, for example, Moe, 1987). Complicating assumptions, however, were only introduced into the model in the early 1990s. Two papers, by Ferejohn and Shipan (1990) and Gely and Spiller (1990), written independently of each other, incorporate into the model: judicial review of administrative decisions, as well as Presidential veto of Congressional override of both decisions of agencies and judicial review of decisions of agencies. The preferences of agencies are assumed to be identical to those of the President (this flaw is overcome by Steunenberg, 1992). Ferejohn and Shipan allow for an override of a Presidential veto by a two-thirds majority in Congress, which is modeled as a unicameral legislative body. In a different vein, Gely and Spiller abstract from the opportunity to override a Presidential veto but treat the House and the Senate as two different (unitary) agents and extend the model to a two-dimensional policy space. A final important difference between the two models is that judicial control is restricted to replacing the agency’s issuance of new rules by the status quo ante in the paper of Ferejohn and Shipan. In a variation p. 74on the model of Gely and Spiller, the courts act as a kind of super-agency able to replace the agency’s choice of a new rule with whatever policy the courts prefer most (subject to the threat of new legislation).
Ferejohn and Shipan deduce that the next (namely judicial) step in the control of policy-making by agencies restricts the power of agencies, and makes agencies more responsive to the preferences of the current (not the originally enacting) legislature. In a comment on this article, Spitzer (1990) studies some cases in which agencies have preferences which differ from those of the President. In another comment, Rose-Ackermann (1990) expresses the criticism that it is unclear why Congress should allow its own committee to deviate as much from Congress’s own preferences as is assumed in the Ferejohn and Shipan paper and, more closely related to the general approach adopted by Ferejohn and Shipan, that it might substantially increase the agency’s and the court’s discretion and influence to abandon the assumption of costless Congressional action.
The paper by Gely and Spiller yields the same result with respect to judicial control over an agency: the reduction in the discretion of an agency is substantially greater if courts are able to impose the policy they most prefer. Allowing the President to have preferences different from those of an agency forces Gely and Spiller to introduce a second dimension in the policy space. Only then can they demonstrate how an increase in the number of institutions that must cooperate to enact a statute increases the dimension of the Pareto set which defines the discretion of an agency. They apply their model to two cases decided by the Supreme Court. The examples show how the ruling of an agency changes after and due to a change in the preferences of the legislature. Tiller and Spiller (1999) introduce differential costs and benefits of alternative administrative instruments both for the agency making the decision and for the court, or other institution, reviewing it. As a consequence, they can show not only that the choice of the policy is a strategic decision, but also the selection of the instrument to implement it.
Formalizing the presentation of the spatial approach, thus far most often based on a graphical exposition, Hammond and Knott (1996) show that any number of dimensions of the policy space and a complete model of US institutions are tractable with this approach. They set forth a formal model. They include all five institutions most relevant to American legislation: the House, the Senate, the relevant committees of both (some multi-person institutions are even divided into their members) and the President. They include two alternative procedures of legislation: a majority of House and Senate and approval by the President, as well as a two thirds majority of the House and Senate. The judiciary is added in an informal way, which may be more appropriate than the alternative approaches chosen by Ferejohn and Shipan (1990) and Gely and Spiller (1990). The power of the judiciary is reduced to stretch the set of legal policies constraining the agency in the direction preferred by the judiciary. And Hammond and Knott p. 75include in their model the President’s ability or inability to fire an agency director. Complicating the assumptions in the analysis does not provide new insights into how, in theory, bureaucratic discretion changes when additional institutions are included in the model. But they offer the insight that the results of less comprehensive models are not due to their simplifying assumptions. Hammond and Knott, however, stress the conclusion that it is impossible to give a clear answer to the question of who really controls an agency; control is exerted by the interplay of institutions. Thus, they reject empirical studies on the power of controlling institutions if such studies are based on correlations between changes in the preferences of the allegedly controlling institutions and agency policy.
A general problem in the literature is, of course, that the information requirements about the preferences of other agents are as strong as they are for most models of sequential games. One should not overemphasize this problem, though, because spatial approaches to bureaucratic discretion have to be seen as a complement to the literature that bases the explanation of bureaucratic discretion on incomplete information.
While earlier articles which take spatial approaches (this section) or optimal-type-of-control approaches (Section 3) to bureaucratic discretion merely refer to the other approach, only Epstein and O’Halloran (1995, 1999) integrate both spatial and optimal-type-of-control approaches in one formal model. These authors assume that only interest groups and the agency are completely informed about the state of the world and hence know what policy they prefer most. The legislator (modeled as a unitary actor and thus re-simplified as compared to the newer spatial literature) only knows its most preferred policy as a function of the true state of the world. When an agency proposes a policy change to the legislator, the legislator may accept or reject the proposal, but not amend it. From the legislator’s perfect knowledge of the preferences of the other actors, the rule proposed by the agency, and the reaction of the interest groups (which sound a fire alarm or not), the legislator is able to deduce a piece of information about the true state of the world and react accordingly. The most important result derived from this model is that the legislator can deduce the most information from the reaction of an interest group if the legislators’ most preferred policy lies between those of the agency and the interest group. Less information can be deduced if the interest groups’ most preferred policy is at the center. No information can be deduced if the agency’s most preferred policy is at the center. The reaction of more interest groups increases the information deduced by the legislator only if the interests of the additional groups differ more than, and in the same direction as, the preferences of the legislators differ from the most preferred policy of the agency.
p. 76The model of Epstein and O’Halloran (1995, 1999) is seminal in spawning a large lineage of follow-up publications. Sloof (2000) extends the influence of the interest group to sending a signal not only to the agency after delegation, but also to the legislator before the decision to delegate. Eric Posner (2001) combines the asymmetric-information spatial model with the procedural rule (see Section 2) of requiring a cost-benefit analysis. He shows that this may reduce the principal’s lack of information if the signal emanating from the agency’s cost-benefit analysis is precise, that is, if it is more costly, the more the signal favors inefficient action. Huber and Shipan (2002) vary the model to allow for the costs of restricting the agency’s discretionary range. Bendor and Meirowitz (2004) generalize the assumptions to show that many results do not depend on the relatively restrictive framework in Epstein and O’Halloran (1995, 1999) and Huber and Shipan (2002). Holburn and Vanden Bergh (2004) discuss in a formal model and an empirical analysis whether and under which conditions interest groups would aim to influence the agency rather than the legislator or the executive.
Bennedsen and Feldman (2006) introduce interest groups’ efforts to influence the agency’s rule-making. Assuming a single and unitary interest group, they show that due to lobbying at the bureaucratic level, legislators’ willingness to delegate increases if the interest group’s preferences are close enough to the legislator’s and declines otherwise. As a consequence, the interest group’s influence grows with the legislator’s uncertainty about the state of the world if the interest group’s preferences are close enough to the legislator’s. Grajzl (2011) develops a similar model from a property rights perspective in which either the legislator (in case of non-delegation) or the bureaucracy (in case of delegation) reaps benefits from a single interest group willing to transfer resources to the competent rule-making authority. Eom and Sweeny (2009) allow for competing interest groups in a similar model. They model how groups with conflicting interests make proposals for regulations. They find that these proposals become less polarized and eventually converge, if either of the interest groups becomes more effective at influencing the rule-making agency, because more effectiveness implies larger lobbying costs for both groups. They apply their model to incentive-rate proposals made by stakeholders of California energy efficiency programs.
Somewhat deviating from the standard spatial approach, Lehmann (2002) develops a model in which an agency’s monitor may not neglect to act against an agency running astray (type I errors), but may also commit errors of type II, that is raise the alarm although the agency is perfectly on track. They provide conditions under which a court filtering the monitor’s alarm should employ extremely high or extremely low evidentiary standards, to completely avoid either type I or type II errors.
6. Ought there to Be Delegation of Rule-Making Power to Agencies and Bureaucracies?
p. 77While the literature discussed thus far looks at the positive questions of how bureaucratic discretion in rule production can best be controlled by the legislature (Sections 2 and 3) and which particular institution controls agencies (Sections 4 and 5), other authors ask more fundamentally whether there ought to be delegation of rule-making to the bureaucracy from the point of view of the legislature or society. Since answers to the positive questions mentioned above are mostly based on fire-alarm control by the same interest groups which influence the legislation, most authors neglect the problems discussed by the capture theory of regulation. They do not even perceive them as problems, if capture works in favor of the same groups in both legislation and bureaucratic rule-making. This is not the case for the normative question discussed in this section.
Imperfect control of agencies can only be used as an argument against having delegation of rule-making authority from the legislator to the bureaucracy for three reasons: (1) a bureaucrat (agency director) has political preferences that differ from those of the legislator; (2) a bureaucrat (agency director) exploits his superior knowledge for his own non-political advantage, as described by Niskanen (1971 and 1975) (also compare Migué and Belanger, 1974); (3) an agency gets captured by the regulated industry more often than the legislator. The first reason vanishes if the legislator can choose the bureaucrat. The second reason either coincides with the third or does not systematically interfere with the content of rules produced by agencies, except for the case where some rules allow collusion of an agency with the regulated industry or firm and consequential rent-sharing, while other rules do not allow this. An example of such a case is discussed by Laffont and Tirole (1990). They conclude that for the case of price regulation, it might be reasonable not to delegate to the agency the decision between (average) cost-plus pricing and marginal-cost pricing with a transfer to the firm(s) covering the fixed costs.
The third reason is discussed more extensively. Contradicting Page (1981), Wiley (1986) argues (and the argument is reinforced in Wiley, 1988) that there is no general evidence that regulated entities have a disproportionate influence upon regulatory agencies. Accordingly, the decision to delegate regulatory rule-making to an agency ought to be based mainly on the agency’s advantages and disadvantages in dealing with specific rule production as compared to the legislator himself. This argument is refuted by Spitzer (1988), who claims with reference to Fiorina (1977, pp. 15–29) that rule making by agencies is more prone to capture by interest groups than rule-making by the legislator because the latter uses the agency to shirk responsibility for the costs imposed by regulation, while claiming credit for the benefits at the same time (also compare Vaubel, 1991, who applies the same argument to the European legislative institutions; p. 78and Moe, 1989). Wiley (1988) points to the many other reasons for a legislator to delegate specific rule-making to a bureaucracy in order to reduce the importance of the argument of Spitzer (1988). Referring to better responsiveness of the law to specific local conditions, Mashaw (1985) supports this pro-delegation position. Bawn (1995) compares the costs and benefits faced by the legislator from delegation of rule-making in a formal model. She shows that the legislator will balance marginal costs incurred as a result of a loss of control through the delegation with marginal benefits reaped as a result of the better fit between agency decisions and specific cases. Her model uses quite specific functional forms, but can be generalized easily. Cooter (2000, chapter 4) provides simple formal models on the topic in his textbook on the economics of constitutions and constitutional law.
Spulber and Besanko (1992) argue that a problem similar to capture may arise from the way the regulator and the regulated firm or industry interact. If the former is able to commit to some environmental quality standard, that is, if the regulator is a Stackelberg leader, then it is sufficient to choose an agency director with political preferences similar to those of the legislator. However, if the regulator is unable to commit to a specific environmental quality standard, but is likely to adapt the standard to the firm’s or industry’s activities (levels of output and levels of pollution), that is, if the relationship between regulator and regulated industry or firm is Nash, then an optimizing legislator will choose an agency director with preferences deviating from its own and will enact statutory limits to bureaucratic standard setting. The answer which Spulber and Besanko give to the question as to whether or not statutory limits ought to be imposed on regulatory discretion depends upon whether the relationship between the regulator and the regulated firm or industry is Stackelberg or Nash. In addition, their answer to the question in what direction these limits should be erected also depends on the circumstances of the specific case. As stressed by Arvan (1992), the ambiguities can only be cleared up by empirical studies. The results of Spulber and Besanko (1992) are ambiguous even though they abstract from important aspects influencing the degree of ex ante statutory constraints on regulatory agencies, such as judicial review, problems of incomplete information on the technology (legislators have to know the technology up to one parameter which does not influence complementarity or substitutiveness of pollution abatement and production costs), and incomplete knowledge of the preferences of other political actors (Ross, 1992).
If non-delegation is assumed to be the only alternative, the question whether there ought to be a delegation of rule-making power to agencies and bureaucracies is only incompletely answered. In his structural approach to law, Cooter (1996a, also 1994a and 1996b) argues that, among rule-making bodies, agencies should make as few rules as possible because they do not follow social norms to the extent that courts do or at least ought to do. Following social norms p. 79which evolve from efficient incentive structures (that is, social settings in which spillovers to groups not engaged in the social formation of the norms do not exist nor do evolutionary traps resulting from non-convexities exist) is better than rules issued by law-makers such as legislators or regulatory agencies and bureaucracies (that is, constructivist law-makers in a Hayekian sense).
Kaplow (1992), too, deals with the role of the courts in the question of whether to delegate rule-making authority to agencies. Yet he does not see rule production by the courts as an alternative to rule production by agencies. Rather, he emphasizes that non-delegation is only possible if the legislator includes overly general clauses in statutes. Thus, courts are forced to deduce concrete standards from the general clauses and do so necessarily after the regulated firm or industry has acted, whereas rules set by agencies are promulgated before a regulated firm or industry acts. Regulation by ex post standards is inferior to (bureaucratic) regulation by rules if and only if the costs measured per case of formulating ex ante rules, which are readily applicable, outweigh the costs which result from less precise standards. Thus, if and only if the law under review is relevant to frequent behavior of a clearly distinguishable type, bureaucratic regulation through rules will be superior to regulation by standards which courts interpret from statutes.
With the same problem in mind, Fiorina (1982 and 1986) tries to develop a model that shows how legislative decisions to delegate rule interpretation (and thereby the making of new, less abstract rules) to the courts or to an agency depend on the substantive preferences of the legislators. Unfortunately, all his results depend on incorrect graphical representations of the expected utilities of legislators and, in the 1982 article, also on the assumption of bell-shaped utility functions of legislators (compare also Nichols, 1982). Accordingly, the results are not reported here.
Finally, Rubin and Cohen (1985) mention a third alternative to delegation of rule-making to agencies. Law (in particular, regulatory statutes) should be made more concrete and enforced by private firms and not by state agencies for several reasons: the legislator (the state) must merely formulate policies on a general, aggregate level and does not have to deal with specific technologies. These technologies can be decided upon by the private law enforcer who signs a contract to pay some (positive or negative) per-unit tax on the aggregate level of pollution or safety or other regulatory targets. Then, he may require firms to abide by a standard or to pay a fee sufficient to cover the tax. Hence, which technology is used becomes a decision subject to market forces. The private enforcer has stronger (monetary) incentives than an agency. The rules governing the relationship between the firms and the enforcer become more efficient, because evolutionary theory of efficient common law is applicable, since both sides have long-lasting economic stakes in these rules. Rubin and Cohen’s basic idea is to make existing substantive regulation a property right p. 80(entitlement) of the private enforcement agency, and the absence of some regulation a property right of the regulated industry. The property right becomes valuable for the enforcer, since he is paid some money for avoiding (statistical) damages (for example, saved statistical lives) and the enforcer is made liable for (statistical) damages. Property rights thus clearly defined will be reallocated in a Pareto-optimal way by Coasean bargaining.
7. The Rule-Production Process Itself
Little has been written in the law-and-economics literature on how rules are actually produced by agencies and bureaucracies, although taking this approach might substantially clarify the relevance of different arguments with respect to control of bureaucratic rule production by the legislator and its branches. One of the few examples of this approach is the paper by McCubbins and Page (1986). They conclude that decentralized incentives within agencies are too flexible for congress(wo)men who want close control over an agency and, perhaps more importantly, that incentives in an agency are such that agencies do not strive for positive overall results, but for countable ‘beans’ (for example, a ban on dioxin instead of the improvement of public health). If research were to take these insights into account, the models constructed to predict the degree and relevance of bureaucratic discretion would gain in explanatory power.
In developing concise models of bureaucratic rule production, law-and-economics scholars might have to take a close look at general descriptions of the rule-making process, such as the book by Kerwin (1992) and case studies such as those of Fox (1987) and McGarity (1991). And more descriptive studies of this kind might be needed. Kerwin thoroughly describes the legal background and actual process of rule-making at the federal level in the US. Though he does not have a consistent theoretical approach, let alone rely on economic theory, his study is very helpful as a starter and as background knowledge for any scholar not wholly familiar with US institutions. On the basis of three cases, Fox describes how decision-making in the FTC changes with the chairman of the Commission. McGarity studies the effects of having different types of employees working for regulatory agencies. He contrasts the ‘techno-bureaucratic rationality’ for rule-making of one group of employees with the ‘comprehensive analytical rationality’ of another. The former group, professionals who typically have technical or legal training, draft rules without paying much attention to economic trade-offs. The latter group, young economists trained in an antiregulative atmosphere, entered agencies during the 1970s. McGarity uses the term ‘comprehensive’ to stress that the ideal pursued by this second group is an open-ended decision-making process based on the quantitative comparison of all possible routes to a goal, which in turn may be defined as quantitative cost-benefit analysis. Based on a large number of case studies, he discusses different institutional settings within bureaucracies which may promote or hinder p. 81the relevance of cost-benefit analysis to bureaucratic rule-making. Yet McGarity notes that at first, many of the analysts with thorough microeconomic training went too far in replacing the pro-regulation biases of techno-bureaucrats with anti-regulation biases. In his opinion, the different but equally strong biases of the young economists neither furthered their goals nor overcame traditional and still-dominant engineering-and-legal orientations to bureaucratic rule-making. Were case studies such as those of Fox and McGarity to be turned into models of bureaucratic behavior (adding the preference dimension to the constraints thus far discussed in the literature), much would be gained in understanding, predicting, and evaluating rule production by agencies.
An early theoretical approach, which unfortunately did not find many followers, is the study conducted by Breton and Winetrobe (1982). They argue that the lack of economic discussion of what happens inside bureaucracies (both public and private!) is due to a seeming absence of property rights and competition in bureaucratic interactions. Introducing a theory of trust as an instrument which backs inter-temporal contracts, they succeed in establishing an analogy with a pure barter economy. Concentrating on Schumpeterian competition instead of neoclassical price competition, they even construct market-like ‘networks’ within bureaucracies. By combining trust and Schumpeterian competition, they can explain not only many of the alleged inefficiencies in bureaucracies, but also the existence of particularly efficient bureaucratic organizations which replace market exchanges to a greater degree than transaction cost arguments are able to justify. Unfortunately, it seems that it would be difficult to apply this innovative theory to rule production by public bureaucracies. Johnson and Libecap (1994) make further progress (based on more standard, neoclassical assumptions) toward understanding what happens inside bureaucracies but, like Breton and Winetrobe, they do not concentrate on rule-making.
Notwithstanding these problems with positive approaches, some articles and books deal with normative questions of how rules ought to be made by agencies and on what arguments decisions of agencies should be based. For example, Breyer (1985) emphasizes that, of course, it would be most appropriate to base regulation entirely on simple economic models, but that this is often impossible because the resulting policies would lack practicability. Still, economic models ought to be used to organize regulators’ analyses and focus their attention on the relevant questions (McGarity, 1991 is clearly supportive of this point).
Ribstein and Kobayashi (1996a and 1996b) discuss a different kind of rule production by agencies. These authors look at agencies which only draft rules which a legislator may subsequently enact or amend. In particular, they discuss the effects and the efficiency of the National Conference of Commissioners on Uniform State Laws (NCCUSL), which provides fully formulated drafts of statutes to state legislatures. They claim that this procedure may have advantages over enacting federal laws, but that it is clearly worse than free competition p. 82among diverse state laws, which would result in an evolution towards the most efficient laws (for example, other jurisdictions copy provisions out of the corporations code of Delaware because many firms choose to incorporate in that jurisdiction). These authors allow that model codes facilitate competition between jurisdictions and overall law-making by exploiting economies of scale. Yet they point out that private institutions such as the American Bar Association draft model laws more dispassionately. It is claimed that agencies like the NCCUSL are too easily influenced by interest groups. The reviewer comments that the reference which these authors make to capture theory is relevant only if the agencies have a monopoly of the supply of model laws, for if they do not, private institutions are free to provide additional competing drafts.
Closely related to the inquiry into how rules are actually produced by agencies and bureaucracies is the question of how long the process of such rule-making takes. Over the 1990s and 2000s, discussions about the duration of administrative procedure emerged in several countries, independently of whether the procedure is mainly legislative or mainly adjudicatory in the respective countries (see Balla and Wright, 2003, for the US and von Wangenheim, 2005, for Germany). For rule-making by agencies, more empirical insights on the actual duration exist (Balla and Wright, 2003; Kerwin, 1992; Kerwin and Furlong 1992) than for adjudication by agencies (Ziekow et al., 2005), and economic theory is scarce (see Hafner, 1999 and 2000, and von Wangenheim, 2005, with respect to adjudication).
Both the positive and the normative literature on the rule production by agencies and bureaucracies are still incomplete, and further work is necessary and seems to be promising. Substantial progress has been made in the positive literature concerning the constraints of regulatory rule production. Yet it is still particularly difficult to derive a complete microeconomic model which describes or predicts the rule-producing behavior of agencies because the political preferences of agencies and bureaucracies have hardly been discussed outside capture theory. Capture theory suffers from weak econometric support and, if taken seriously, would make obsolete many of the problems posed in the literature based on positive political theory. Bureaucratic preferences obviously are not determined by a single individual but are the result of a decision-making process within a bureaucracy. Models of this process would substantially facilitate understanding and prediction of the rule-making behavior of agencies.
On the normative side, due to the lack of complete positive theories on bureaucratic decision-making constrained by legislative and judicial oversight, arguments often do not seem to be too strong because normative conclusions cannot be based on a clear relationship between institutional settings and their p. 83welfare effects. Hence, much of this literature must be based on ideological premises concerning the behavior of agencies and bureaucracies.
Balla, Steven J. and Wright, John R. (2003), ‘Consensual Rule Making and the Time it Takes to Develop Rules’, in Krause, George A. and Meier, Kenneth J. (eds), Politics, Policy, and Organizations – Frontiers in the Scientific Study of Bureaucracy, Ann Arbor: University of Michigan Press.
Grier, Kevin (1984), ‘The Political Economy of Monetary Policy’, Ph.D. dissertation.
Hafner, Mathias (1999), ‘Das optionale Anzeigeverfahren nach §231a BlmSchG (The optional notification procedure of Sec. 231a of the Federal Emission Control Act)’, in Schmidtchen, Dieter and Schmidt-Trenz, Hans-Jörg (eds), Vom Hoheitsstaat zum Konsensualstaat – Ökonomische Analyse der Flexibilisierung von Genehmigungsverfahren, Baden-Baden: Nomos Verlagsge-sellschaft, 127–147.
Hafner, Mathias (2000), Behördliche Diagnosefehler im Genehmigungsverfahren: Eine ökonomische Analyse immissionsschutzrechtlicher Eröffnungskontrollen und ihrer Reform (Administrative Diagnostic Errors in Licensing Procedures: An Economic Analysis of Disclosure Controls in Emission Control Law and their Reform), Wiesbaden: Gabler.
Hammond, Thomas H. and Knott, Jack H. (1996), ‘Who Controls the Bureaucracy?: Presidential Power, Congressional Dominance, Legal Constraints and Bureaucratic Autonomy in a Model of Multi-Institutional Policy-Making’, 12 Journal of Law, Economics, and Organization, 119–166.
MacKay, Robert J., Miller, James C., III and Yandle, Bruce (1987a), ‘Public Choice and Regulation: An Overview’, in MacKay, Robert J., Miller, James C., III and Yandle, Bruce (eds), Public Choice and Regulation: A View from Inside the Federal Trade Commission, Stanford, CA: Hoover Institute Press, Stanford University, 3–12.
MacKay, Robert J., Miller, James C., III and Yandle, Bruce (eds) (1987b), Public Choice and Regulation: A View from Inside the Federal Trade Commission, Stanford, CA: Hoover Institute Press, Stanford University.
Ribstein, Larry E. and Kobayashi, Bruce H. (1996b), ‘Uniform Laws, Model Laws and Limited Liability Companies’, 66 University of Colorado Law Review, 947–999.
Weingast, Barry R. and Moran, Mark J. (1983), ‘Bureaucratic Discretion or Congressional Control? Regulatory Policymaking by the Federal Trade Commission’, 91 Journal of Political Economy, 765–800. Reprinted in Mackay, Robert J., Miller, James C., III and Yandle, Bruce (eds) (1987), Public Choice and Regulation: A View form Inside the Federal Trade Commission, Stanford, CA: Hoover Institution Press, 30–62.