The Administrative Broker

Abstract

Although traditional models of bureaucratic politics have relied on the old assumption that information is expensive, information is prevalent nowadays; the monopoly of bureaucratic expertise has been undermined as interest groups have significantly developed and are professionalized. As a result, what is really important in current bureaucratic politics is not just neutral expertise, but the political capacity to affect the behaviors of information sources. Through mediating conflicts of interest and minimizing unnecessary contingencies, agencies can persuade their stakeholders not to provide information to legislators and, therefore, indirectly affect legislators’ decisions on delegation and oversight. Different from traditional principal–agent theories, this article suggests the “administrative broker” model in which politically influential agencies can block information leakage to legislators and enhance their own discretion. Moreover, the administrative brokers occasionally transform traditionally hostile principal–agent relations into more favorable ones.

Keywords

brokerage capacity bureaucratic politics discretion contingency control

Introduction

During the 1960s and the 1970s, legislators were incapable of systematic policymaking and inevitably provided agencies with significant discretionary status (Huntington, 1965; Ogul, 1976; Scher, 1963). Scher (1963, p. 532) argued that public agencies were “impenetrable mazes” for inexpert legislators. Similarly, Ogul (1976, p. 14) said, “The more technical and complex the subject matter is perceived to be, the less the likelihood of oversight. Only a few members of the Congress are experts in any area of bureaucratic operations.” Because collecting information was very costly during this period, inexpert legislators delegated broad discretion to agencies. Supported by this “abdication principle,” agencies were believed to enjoy significant discretion relying on information asymmetry between agencies and legislators (Epstein & O’Halloran, 1994; Gailmard, 2009; Rourke, 1984). However, this view has been challenged by “congressional dominance” studies since the 1980s. They have claimed that legislators can overcome information asymmetry thanks to information provided by interest groups (Epstein & O’Halloran, 1995; McCubbins, Noll, & Weingast, 1989; McCubbins & Schwartz, 1984). This congressional dominance view has been highly supported, as interest groups have been more professionalized, communication systems have been further advanced and, as a result, interest groups are able to provide information to legislators more easily (Bosso & Collins, 2002; Hula, 1999). In other words, legislators have become capable of delegating only limited policymaking authority to agencies, as interest groups could have informed legislators of policy environments and bureaucratic behaviors (Boehmke, Gailmard, & Patty, 2005; Epstein & O’Halloran, 1995). It implies that prevalent information has allowed legislators to overcome information asymmetry, which threatens discretionary agencies. However, interest groups’ signaling is not always concrete. Rather, politically capable agencies have affected the signaling behaviors of interest groups and have prevented information leaks to Congress.

Although many studies have assumed that bureaucrats are neutral professionals, they are highly political (Carpenter, 2001; Moe, 2006). According to Aberbach, Putnam, and Rockman, agencies “continuously interact with organized interest groups and mediate among established interests” (1981, p. 93) and “engage in anticipatory conflict management, seeking consensus among the relevant participants before a proposal is actually put forward” (p. 13). In other words, agencies possess not only neutral expertise, but also political capability to affect their stakeholders’ behaviors by brokering conflicting interests. This political capacity is highly meaningful in policy implementation, because it is inevitably political, evolutionary, and ad hoc (Majone & Wildavsky, 1979). Although implementation translates a policy into action, a policy is not a fully articulated plan. Rather, it is “only a collection of words” prior to implementation (Bardach, 1980, p. 139). Thus, policy implementation varies depending on contingencies derived from interactions between governmental actors, interest groups, and regulated organizations. In this sense, adverse reaction, resistance, and noncompliance of interest groups increase unanticipated contingencies as well as reduce the probability of desired action (Clune, 1983). Therefore, if agencies can control contingencies with ex ante compromises, they can provide additional benefits to all the related stakeholders. Then, if interest groups are content with the bureaucratic brokerage, they have little incentive to provide information to legislators. This implies that politically capable agencies are able to intervene in the relations between their principal and informants and, as a result, interrupt signaling to the legislature. As a consequence, agencies with the political capacities can affect (not only interest groups’ behaviors, but also) legislators’ decisions on bureaucratic discretion and oversight.

It is not novel, of course, to point out the issues of administrative brokerage and political power of agencies. Although traditional bureaucratic politics studies have touched on the brokerage role of public agencies (Aberbach et al., 1981; Susskind & Ozawa, 1983) and coalition building activities (Eisner, Worsham, & Ringquist, 1996; Ringquist, 1995), they have generally neglected the political impact of administrative brokerage—in particular, the political influence of broker agencies on their principal. Moreover, though several other studies have discussed the informal political power of agencies (e.g., Fritschler & Hoefler, 1989), they have been more concerned about how to keep the “deviant” bureaucrats in tune with Congress: the causal mechanism between administrative brokerage and congressional delegation/oversight has been rarely discussed. In this respect, this study examines the political capacity of agencies with regard to interest groups’ signaling and bureaucratic discretion, and clarifies the causal mechanism between the administrative brokerage and congressional delegation/oversight.

In this article, I suggest the “administrative broker” model to show how agencies affect their stakeholders politically and what happens to bureaucratic discretion as a result of their political influence. The formal model for this analysis is based on the venue choice of interest groups. To illustrate, if interest groups can reach a satisfactory agreement in the “bureaucratic venue” prior to signaling in the “legislative venue,” they do not have to provide information to legislators, which results in expensive access costs. In this case, agencies could maintain significant discretion. Otherwise, interest groups would provide information to legislators and agencies’ discretion would decrease. This political capacity to mediate interest groups and to derive compromises is called brokerage capacity in this article. The basic result of the model is that agencies with high brokerage capacity are likely to have more discretion. To put it another way, information asymmetry preserved by political brokerage compels legislators to delegate more discretion.

Although this result is based on information asymmetry, the “administrative broker” model is different from traditional principal–agent theories on several points. The relations between agencies and Congress are dependent not only on expertise, but also on political “brokerage capacity.” In recent decades, information has become cheaper and more prevalent. Thus, what is really important in legislature–agency relations is not information itself, but the political influence on information sources. In addition, brokerage capacity occasionally transforms traditionally hostile principal–agent relations (between the legislature and agencies) into more favorable ones. Thanks to the contingency control benefits from ex ante bureaucratic bargaining, their relations can be cooperative under limited conditions.

This article specifies brokerage capacity and models its impact on bureaucratic discretion and the relations between Congress and agencies. Subsequently, the propositions derived from the “administrative broker” model will be examined by empirical examples of the Environmental Protection Agency (EPA) and administrative laws.

The Model Setup

Brokerage Capacity Specification

Public agencies are surrounded by diverse political pressures. Interest groups frequently claim their own “property rights” against extant policymaking or implementation (Majone & Wildavsky, 1979). In this process, interest groups are likely to incur many contingencies through resistance, lawsuits, and violations. These unexpected contingencies can cause serious administrative delays, uncertainties, and mistakes as well as be damaging to all stakeholders. However, if agencies can provide an accessible policy venue and acceptable solutions to conflicting stakeholders, they can minimize the detrimental contingencies.¹ As a result of the benefit of brokerage, if interest groups are satisfied with the compromising solutions in the bureaucratic venue, they do not have much incentive to provide information to Congress. Then, information asymmetry would persist and agencies could maintain their discretionary authorities.

In terms of the venue choice of interest groups, a bureaucratic venue should be accessible and furnish the chance to negotiate with their competitors on public policies (e.g., public meetings, advisory committees, and cooperative programs with private actors). Otherwise, interest groups would leave the bureaucratic venue and go to some other transaction-cost-effective venues such as the legislature or courts. Therefore, reducing transaction costs is a preliminary condition for mediating interest groups. However, it is not sufficient for brokering compromises with interest groups. Even when interest groups can communicate with little transaction cost in the bureaucratic venue, they frequently cannot reach agreements because of inseparable property rights in public policies. For example, on the issue of the “wardrobe malfunction” at the Super Bowl half-time show in 2002, diverse interest groups, such as broadcasting firms, parents’ groups, and artist groups, claimed their own “property rights” regarding broadcasting content. However, no group clearly separated its rights in broadcasting and viewing TV programs. Thus, it is also important for agencies to suggest “acceptable” policy alternatives. If agencies fail to suggest acceptable solutions to conflicting interest groups, they leave the bureaucratic venue. Then, detrimental contingencies may occur. Therefore, if bureaucratic agencies broker compromises with interest groups with accessible venues and agreeable solutions, only then can they minimize negative contingencies.

In this respect, bureaucratic brokerage should be distinguished from regulatory capture. As Salisbury (1990, p. 213) said, “In a destabilized world of fragmented interests and multidimensional challenges from externality groups it becomes impossible for policy makers to identify which interests, if any, they can succumb to without grave political risk.” Given that the present interest groups have been highly pluralized and each interest group has sufficient resources and expertise, if public agencies favor specific interest groups in a hidden room, other dissatisfied opposing groups would move to other policy venues to overturn the agencies’ decisions. Because of this reason, since the 1970s, it has become highly difficult for industries to capture an agency (Wilson, 1989). For example, the EPA, which was slightly captured by industries in the early 1980s, attempted to suspend the ban on the disposal of liquid waste in landfills (47 FR 8304) on February 19, 1982, without much discussion with environmental groups. Immediately after EPA’s announcement, however, environmental groups began their lobbying and protests, complaining about the lack of administrative brokerage. This resulted in congressional attention to the issue and a House oversight hearing on March 10, 1982, where the agency was pressured by pro-environment legislators to reconsider the suspension. Moreover, legislators prepared several bills to limit EPA’s discretionary authority regarding the issue. Because of congressional intervention, EPA reversed its decision and reinstituted the stricter regulation several days later.² Therefore, it is necessary for agencies to decipher the precise property rights of conflicting stakeholders and to suggest acceptable policy alternatives, not to be just captured by specific interests. If agencies are captured without compromising negotiations, alienated groups are likely to leave the bureaucratic venue, informing legislators of administrative environments and bureaucratic behaviors.

In sum, brokerage capacity is the capability to establish favorable environments for interest groups to negotiate as well as to make authoritative and acceptable suggestions. If public agencies have sufficient brokerage capacity, additional contingency control gains are available in policy implementation, and as a result, agencies can maintain information superiority over their principals.

Formal Structure

In this game, there are four players: two policy venues and two interest groups. The two policy venues are Congress (denoted by L) and an agency (denoted by A).³ Without loss of generality, L’s ideal point is x_L = 0 and A’s is x_A > 0. Moreover, there are two different types of interest groups named Fromleft (denoted by FL) and Fromright (denoted by FR), where x_FL < x_L = 0 < x_FR. All the players’ actions are based on a compact and convex policy space, which is denoted by X ⊂ ℜ¹. Similar to the traditional bureaucratic politics model, the policy outcome (denoted by x, such that x ∈ X ) is determined by policy (denoted by P) and the state of the world (denoted by ϵ~U[-E,E]), that is, x = P− ϵ. Moreover, the state of the world is observable only by A, FL, and FR, which are assumed to possess “expertise.” The observed current state of the world is denoted by ϵ^*.

The sequence of the game is as follows:

1. Bargaining in the bureaucratic venue:

A makes a take-it-or-leave-it (TIOLI) policy offer p_T. Then, FL and FR either accept or reject the offer.

2. Signaling in the legislative venue:

If bureaucratic bargaining is successful, FL and FR cannot signal.

If bureaucratic bargaining fails, FL and FR send their signals, θ_FL and θ_FR, to Congress.

Each interest group should pay congressional transaction cost TC_L ≥ 0, unless θ _j = ∅, j ∈ {FL, FR}.

3. L decides a discretionary window, $D = [\underline{p}, \bar{p}]$ .

4. Implementation: bureaucratic action is realized; that is, P_A is implemented by A.

In the sequence of the game, steps 1 and 2 represent interest groups’ sequential venue choices. It is assumed that bureaucratic agencies can bargain with interest groups (step 1) prior to their access to Congress (step 2), because A can move quickly in response to specific social problems by their experience and standard operating procedures. This assumption does not imply that agencies can constrain interest groups. Rather, this game sequence is related with interest groups’ incentive to send costly signals. In reality, interest groups can contact legislators to provide information even during negotiations with agencies. However, if interest groups are sufficiently satisfied with the administrative brokerage, they have no incentive to send costly signals to Congress during or after negotiations in the bureaucratic venue. The game sequence represents this situation. If bureaucratic bargaining fails, FL and FR have a chance to signal in the legislative venue. Thus, step 2 is a threat to A during their bargaining in step 1. Because bureaucratic bargaining in step 1 is frequently informal and secretly performed, L cannot know P_T, regardless of bargaining results.

Players’ utility function u_i(x, c(·)|x_i), i ∈ {L, A, FL, FR} is given by u_i: X × ℜ¹ × X → ℜ¹ and is twice differentiable and negative quadratic in |x–x_i|, where c(·) denotes the contingency function. In this article, contingencies represent all kinds of unexpected administrative inefficiency (e.g., delay, slowness, interruption) caused by interest groups’ behaviors against policy implementation such as resistance, violation, and litigation.⁴ If agencies are capable of making robust contracts with interest groups prior to policy implementation, they can enhance the quality of the policy outcome, limiting unnecessary contingencies. Specifically, c(b) is strictly increasing with $ċ (1 - β b)$ , where b denotes constant brokerage capacity such that b ∈ [-∞, 1) and default contingency, denoted by ċ, is positive. When brokerage in the bureaucratic venue is successful at step 1 and the brokered contract is implemented as promised (i.e., P_A = P_T at step 4), β is equal to 1. Otherwise, β is equal to 0.

Because property rights in public policies are often inseparable, interest groups frequently incur negative contingencies, which are detrimental to all stakeholders because they reduce administrative efficiency. However, if agencies have sufficient brokerage capacity, they can minimize the probability of interest groups becoming discontent with suggested solutions. In the game sequence above, if bureaucratic agencies make successful compromises under b ≥ 0 at step 1, and implement P_A = P_T as promised at step 4, all the game players can acquire additional utilities from reducing contingencies. To simplify, it is assumed that the additional utility from contingency control is the same for all the four game players and independent from policy outcome. However, even when agencies have some ability to define property rights effectively—that is, to suggest “acceptable” policy options to minimize contingencies—if the transaction costs of the bureaucratic venue are significant, it is nearly impossible for interest groups to make a successful compromise in the bureaucratic venue. Negative brokerage capacity (b < 0 ) represents this situation. In other words, negative b stands for the case that bureaucratic brokerage is meaningless because of the high transaction costs of the bureaucratic venue. In terms of the discretionary window, $D = [\underline{p}, \bar{p}]$ at step 3, Congress has few concerns about the lower bound of the discretionary window, because x_A > x_L = 0. Thus, it can be assumed that P_– = – ∞ and that equilibrium discretionary window is represented only by upper bound P^–. This assumption has been used by many bureaucratic politics studies especially regarding bureaucratic discretion (e.g., Huber & McCarty, 2004). Because A has no incentive to implement policies that are lower than what an informed Congress would have wanted, only the location of the upper bound is meaningful in this game setting. Moreover, following traditional delegation models, it is assumed that x_A is not ideologically extreme in terms of x_L —specifically, $x_{A} < \sqrt{\frac{2}{3} E^{3}}$ . It implies that the legislature prefers full delegation to any tight control, if it knows only ϵ^*∈ [-E, E]. In addition, $x_{FL} < - \frac{E}{2}$ and $x_{FR} > \frac{E}{2}$ is assumed to make the model parsimonious. The opposite case (i.e., $x_{FL} > - \frac{E}{2}$ or $x_{FR} < \frac{E}{2}$ ) will be examined later when explaining the benchmark model. Based on these game settings, perfect Bayesian equilibria (PBE) will be characterized.

Results

In this section, two different models are proposed. One is the benchmark, which represents traditional bureaucratic politics arguments, especially congressional dominance and bureaucratic drift models. Compared with the benchmark, the “administrative broker” model is suggested to analyze the relations between brokerage capacity and discretion. Then, the model is examined in terms of the Pareto-improving solution of bureaucratic brokerage.⁵

Benchmark (b < 0)

Before discussing bureaucratic brokerage, it is necessary to examine the negative brokerage capacity (b < 0) case—meaningless bureaucratic brokerage—as a benchmark. There are two purposes for this examination. One is that this benchmark supplies an overview of traditional bureaucratic politics arguments. Thus, this benchmark provides a convenient comparison with the administrative broker model, which will be suggested in the next section. Moreover, the equilibrium outcome of the benchmark denotes the “credible threats” or “disagreement points” of interest groups when bureaucratic bargaining is available under b ≥ 0.

Congressional dominance theorists have argued that the legislature can control bureaucratic agencies based on the signals from interest groups. Several signaling models support this view that complete information can be delivered by two opposite informants (Krishna & Morgan, 2001a, 2001b). However, interest groups cannot always provide complete information about current state of the world (ϵ^*). Rather, their signals are sometimes mere babbling, partly because of the congressional transaction costs and interest groups’ ideal points. If P_A = x_A + ϵ^*is considered sufficient by interest groups given the congressional transaction cost (TC_L), they would be reluctant to provide information to Congress. In this case, agencies can maintain information asymmetry.

Suppose that L believes that ϵ^* = θ_FR − x_FR = θ_FL − x_FL, only if θ_FR − x_FR = θ_FL -x_FL and that if θ_FR − x_FR ≠ θ_FL − x_FL, L cannot infer anything from this signals and believes only ϵ^*∈[-E, E]. In this case, if θ_FR = x_FR + ϵ^* and θ_FL = x_FL + ϵ^*, L can acquire complete information about ϵ^* in equilibrium. In other words, complete information provision necessitates the cooperation of the two interest groups. So, only if u_j (0, ċ) − TC_L > u_j(x_A, ċ) where j∈{FL, FR}, interest groups can provide complete information about ϵ^* in cooperation. This case represents the congressional dominance argument, which holds that the legislature can overcome informational asymmetry thanks to help from interest groups. Otherwise, at least one interest group does not have any incentive to provide information to L under the assumption that $x_{FL} < - \frac{E}{2}$ and $x_{FR} > \frac{E}{2}$ . In this case, the legislature knows only ϵ^*∈[-E, E]. This case represents the bureaucratic drift argument: agencies have enjoyed remarkable discretion based on information asymmetry. Even when interest groups have expertise, if the congressional transaction cost is not negligible, information asymmetry can persist and agencies can acquire much discretion. Then, a benchmark’s equilibrium discretionary upper bound (denoted by $\bar{p}$ ^* _b _<0) and policy outcome (denoted by x^*_b<0) are:

({\bar{P}}_{b < 0}^{*}, x_{b < 0}^{*}) = {\begin{matrix} (ε, * x_{L}) i f u_{j} (0, ċ) - T C_{L} > u_{j} (x_{A}, ċ), j \in {F L, F R} \\ (x_{A} + E, x_{A}) otherwise \end{matrix}

This result simplifies traditional bureaucratic politics arguments. If interest groups have sufficient incentive to provide information to the legislature (partly) because of low congressional transaction costs, there can be separating PBE. Then, Congress can strictly control agencies. In contrast, if the congressional transaction cost is high given the interest groups’ ideal points, only pooling PBE is available. Then, bureaucratic drift might be prevalent based on information asymmetry. These outcomes represent two extreme cases in signaling under the assumption that, $x_{FL} < - \frac{E}{2}$ and $x_{FR} > \frac{E}{2}$ . In contrast, if $x_{FR} > \frac{E}{2}$ or $x_{FL} > - \frac{E}{2}$ , partitional equilibria in thesignaling are available, which implies somewhat “moderately” credible threats. In this case, the legislature can incompletely update its belief about ϵ^*, that is, ϵ^*∈[E₁, E₂], where |E₁ − E₂| < 2E, depending on congressional transaction costs and interest groups’ ideal points. Expecting this result, agencies can suggest P_T, considering the “moderate” threats in step 1. Thus, the basic result is the same even when allowing $x_{FL} > - \frac{E}{2}$ or $x_{FR} < \frac{E}{2}$ . However, this argument is omittedin this article to simplify the examination of brokerage capacity. Only two cases of separating and pooling equilibria suggested above will be considered hereafter.

This benchmark assumes that bureaucratic brokerage is not available because of negative brokerage capacity. Thus, the equilibrium outcomes of the benchmark, $\bar{p}$ ^*_b<0 and x^*_b<0, are a “credible threat” of interest groups against A at the bureaucratic bargaining of step 1. In other words, the results show the reserved outcomes for interest groups even when they do not accept P_T at step 1. Therefore, successful brokerage is possible only when A can suggest a policy alternative, which ensures at least u_j(x^*_b<0, ċ), j∈ {FL, FR} for both interest groups.

Administrative Broker (b ≥ 0)

In contrast, suppose that bureaucratic bargaining is now available with b ≥ 0. In this case, A can broker interest groups by suggesting a TIOLI offer between ϵ^*≤ P_T ≤ x_A + ϵ^*so as to block information leakage to L. In other words, A limits interest groups’ incentives to reach the legislative venue. Because there is a trade-off between u_A and u_FL (sometimes, and u_FR) with respect to P_T, A’s TIOLI offer is dependent on how much FL can acquire from x^*_b<0. In other words, A suggests distinctive P_T to interest groups depending on the “credible threats” of the benchmark. Specifically, A’s TIOLI offer should ensure at least the value of u_FL(x^*_b<0, ċ) − TC_L to FL. Thus, suppose that EP_FL(b) = u^-1_FL[u_FL(x^*_b<0, ċ) − TC_L, c(b)]+ ϵ^*. In words, EP_FL is the indifferent policy option to FL, compared with the benchmark outcome, under the relieved contingency level of c(b) ≤ ċ. In this case, A will suggest P^*_T = min(x_A + ϵ^*, EP_FL). Because P^*_T is weakly dominant for FL, FR, and A, compared with the benchmark, there could be successful bargaining in equilibrium with this suggestion. In this case, if the bureaucratic bargaining of step 1 is binding, successful bureaucratic bargaining under P^*_T = min(x_A + ϵ^*, EP_FL) and babbling messages are the unique PBE.

However, in this case, A’s commitment problem arises. In other words, bureaucratic brokerage is not binding and it is probable that A deviates at the implementation stage (step 4) to P_A ≠ P_T. Because public policy is implemented at the end of the game, interest groups cannot reverse A’s implementation after step 4. In real policy implementation, agencies sometimes withdraw their proposed rules without much procedural burden such as notice and comment. Moreover, when agencies already initiate administrative programs, it is much harder for stakeholders to turn down the policy implementation because of the significant fixed cost already invested in the program. In other words, when bureaucratic agencies make fait accomplis betraying their compromised public policies, it is difficult for interest groups and political principals to overturn the agencies’ decisions. Therefore, A’s brokerage is successful only when the commitment problem is already resolved. In other words, it should be impossible for A to deviate profitably to P_A ≠ P_T for successful brokerage. Otherwise, interest groups would not participate in the bureaucratic brokerage.

From this perspective, L’s decision on the discretionary window is important. If L provides broader discretion, it is highly probable that serious commitment problems occur and that bureaucratic bargaining cannot be successful. Although L cannot know ϵ^* and P^*_T, the actor has the information on ċ, b, and TC_L. Therefore, the legislature can infer P^*_T − ϵ^*. In words, L can conjecture the policy outcome of the bureaucratic bargaining, although L cannot know P^*_T itself.From the assumption above $x_{A} < \sqrt{\frac{2}{3} E^{3}}$ , the legislature prefers $\bar{p}$ = x_A + E to $\bar{p}$ = 0, if the actor knows only ϵ^*∈[-E, E]. Because Congress can acquire (at least) additional contingency control gains through bureaucratic brokerage, the actor would like to minimize A’s commitment problem at step 3. Thus, L has the incentive to eliminate A’s commitment problem by deciding the strictest discretion, even as allowing A to implement P^*_T . Under this situation, $\bar{p}$ = P^*_T − ϵ^* + E is a weakly dominant strategy of L. Even in this case, it is probable that A deviates from P^*_T to min(x_A + ϵ^*, P^*_T − ϵ^* + E) after successful bargaining. In other words, only when u_A(P^*_T − ϵ^*, c) > u_A[min(x_A, P^*_T -2ϵ^* + E), c] is A’s commitment ensured. Otherwise, interest groups (especially, FL) would not accept P^*_T and bureaucratic bargaining would fail. Then, the equilibrium would be the same as the benchmark, $\bar{p}$ ^*_b<0. Because of this commitment problem, A cannot always broker interest groups, though non-negative EP_FL always exists under b ≥ 0. One way to eliminate the commitment problem is high brokerage capacity. Because players’ utility functions are negative quadratic in terms of policy outcomes, A has more incentive to deviate from P_A = P^*_T, when low brokerage capacity makes P^*_T distant from x_A. As an extreme example, significantly high b makes x_A + ϵ^* < EP_FL. In this case, P^*_T will be decided at x_A + ϵ^*, where A has no incentive to deviate. As a consequence, the probability of successful bargaining is weakly increasing in b.

Lemma 1: There is a successful brokerage PBE, only if u_A(P*_T − ϵ*, c) ≥ u_A[min(x_A, P*_T-2ϵ* + E),ċ] and _{b ≥ 0}, where u_A(P*_T − ϵ*, c) − u_A[min(x_A, P*_T-2ϵ* + E),ċ] is weakly increasing in b and decreasing in TC_L.

Lemma 2: P^–*_b≥0 = P*_T − ϵ* + E is weakly higher than $\bar{p}$ ^*_b<0 and is weakly increasing in b,

where $\bar{p}$ ^*_b≥0 denotes the equilibrium discretionary upper bound under b ≥ 0.

In sum, $\bar{p}$ ^*_b≥0 = P^*_T − ϵ^* + E, only if u_A(P^*_T − ϵ^*, c) ≥ u_A[min(x_A, P^*_T-2ϵ^*+ E),ċ]. By the definition of EP_FL, $\bar{p}$ ^*_b≥0 is also weakly higher than $\bar{p}$ ^*_b<0 and weakly increases in b. Moreover as seen above, higher b increases the probability of successful bargaining. In other words, P– always weakly increases in b. Therefore, Propositions 1-1 and 1-2 are derived from above-mentioned lemmas:

Proposition 1-1: Bureaucratic discretion is weakly increasing in brokerage capacity.

Proposition 1-2: A certain level of brokerage capacity is necessary for successful bureaucratic brokerage because of commitment problem. More brokerage capacity is necessary for successful brokerage when congressional transaction costs increase.

Pareto-Improving Brokerage

The administrative broker model shows that agencies can acquire additional distributive gains from information asymmetry, similar to other bureaucratic politics models. However, the model is different from the traditional arguments in the sense that “administrative brokers” can provide Pareto-improving solutions through brokerage, especially when the congressional transaction cost is insignificant. From the benchmark, x^*_b<0 is Pareto-optimal. This is the basic assumption of traditional bureaucratic politics that all the public policies between the ideal points of Congress and an agency are Pareto-optimal and the gain of one player implies loss to the other player. Therefore, the benefits of “principal” and “agent” are likely to be traded off. The benchmark reflects the assumption that the loss of Congress is dependent on the gain of the agency. However, bureaucratic brokerage occasionally yields Pareto-improving public policies through bureaucratic brokerage. This does not imply that agencies can be fully accountable through brokerage. Rather, additional utilities from contingency control to the legislature can be greater than the distributive loss from bureaucratic drifts under some limited conditions.

Basically, when the congressional transaction cost is not significant, the legislature can acquire more benefits from bureaucratic brokerage, even compared with $u_{L} (x_{b < 0}^{*}, ċ | {\bar{p}}_{b < 0}^{*})$ . Suppose that TC_L = 0 and b > 0. In this case, $u_{L} [p_{T}^{*} - ε^{*}, c (b) | {\bar{p}}_{b > 0}^{*}]$ is always strictly greater than $u_{L} (x_{b < 0}^{*}, ċ | {\bar{p}}_{b < 0}^{*})$ ,if there is no commitment problem, because u_i is negatively quadratic with respect to x, that is, $u_{L} [p_{T}^{*} - ε^{*}, c (b) | {\bar{p}}_{b > 0}^{*}] - u_{L} (x_{b < 0}^{*}, ċ | {\bar{p}}_{b < 0}^{*}) > u_{FL} (p_{T}^{*} - ε^{*}, c | {\bar{p}}_{b > 0}^{*}) - u_{FL} (x_{b < 0}^{*}, ċ | {\bar{p}}_{b < 0}^{*})$ . In other words, Congress strictly prefers bureaucratic brokerage at any b > 0 to brokerage failure under TC_L = 0, if there is no commitment problem. However, as TC_L increases, P^*_T is closer to x_A + ϵ^* and $u_{L} [p_{T}^{*} - ε^{*}, c (b) | {\bar{p}}_{b > 0}^{*}] - u_{L} (x_{b < 0}^{*}, ċ | {\bar{p}}_{b < 0}^{*})$ is decreasing, because the congressional distributive gain is decreasing whereas contingency control gain is static. Therefore, Pareto-improving brokerage is available only under low congressional transaction costs. However, low transaction cost is a necessity, not a sufficiency, for Pareto-improving brokerage, because of the commitment problem. Minimal congressional transaction costs force the agency to decide P^*_T around ϵ^*. This would increase the distributive gain for Congress. However, this also provides the agency with sufficient incentive to disobey the compromise solution. Thus, Pareto-improving brokerage necessitates high brokerage capacity, which allows the agency to set up P_T sufficiently close to x_A + ϵ^* (see Proposition 1-2). Otherwise, Pareto-improving brokerage may not be available under low transaction costs, because of the commitment problem.

In sum, bureaucratic brokerage weakly increases all stakeholders’ utilities compared with the benchmark case, if the congressional transaction cost is minimal and brokerage capacity is sufficiently high, despite some distributive losses of L and FL compared with the benchmark. These relations between brokerage capacity and discretion and the threshold of Pareto-improving brokerage are summarized in Figure 1.

Lemma 3: $u_{L} (p_{T}^{*} - ε^{*}, c (b) | {\bar{p}}_{b > 0}^{*}) > u_{L} (x_{b < 0}^{*}, ċ | {\bar{p}}_{b < 0}^{*})$ is more probable as TC_L is smaller, under sufficiently high b such that u_A(P^*_T − ϵ^*, c) > u_A[min(x_A, P^*_T-2ϵ^* + E),ċ].

Proposition 2: Bureaucratic brokerage can improve the utilities of Congress and an agency simultaneously compared with the benchmark only under low congressional transaction costs and high brokerage capacity.

Figure 1.

Brokerage capacity and bureaucratic discretion.

Congressional Transaction Cost, Pareto-Improving Brokerage, and Oversight

Congressional Transaction Cost Reduction and Endogenous Brokerage Capacity

The above-mentioned game assumes that brokerage capacity is exogenous. However, from a substantive viewpoint, agencies can enhance their brokerage capacities endogenously through meetings and negotiations. This section briefly examines the endogenous aspect of brokerage capacity especially in terms of congressional transaction costs. From Figure 1, the marginal increase rate of the discretionary window with respect to brokerage capacity increases as TC_L decreases. Because A’s distributive gain is highly dependent on the discretionary window, it can be inferred that A has a significant incentive to increase its brokerage capacity as TC_L decreases—unless the endogenous cost for the capacity is exceptionally huge. Moreover, serious commitment problems in brokerage are more possible as congressional transaction costs decrease (Lemma 1). It means that $\bar{p}$ can be minimized to ϵ^* because of brokerage failure (and subsequent information leakage) if TC_L decreases. In this case, if A can increase its brokerage capacity endogenously, the agency can enjoy remarkable discretionary “jump” and distributive gains by overcoming the commitment problem. As a consequence, it can be intuitively inferred that A would invest more resources for its brokerage capacity as TC_L decreases, if the actor is allowed to enhance its capacity endogenously.

This intuition is implicative regarding Pareto-improving brokerage, especially under the current political situation where congressional transaction cost has declined steeply since the 1980s. Interest groups have significantly developed and coalesced to share congressional transaction costs (Baumgartner, Berry, Hojnacki, Kimball, & Leech, 2009; Hula, 1999; Wright, 1996). Moreover, the growth of communication systems and mass media has also decreased the cost. Given these political environments, it can be expected that bureaucratic agencies have enhanced their brokerage capacity remarkably in recent decades, as congressional transaction costs have declined. From Lemma 3, Pareto-improving brokerage is available only under low transaction costs and high brokerage capacity. If brokerage capacity is only constant, the decline of congressional transaction costs can incur serious commitment problems in brokerage and Pareto-improving brokerage may not be highly available even under low congressional transaction cost. However, brokerage capacity is not constant; agencies have developed various ways to interact with policy stakeholders and encourage public participation, such as negotiated rulemakings, advisory committees, administrative dispute resolutions, and public meetings. Thus, the decline of congressional transaction costs may increase brokerage capacity. As a result, both conditions of Pareto-improving brokerage (low TC_L and high b from Proposition 2) may be highly achievable in current political situations.

Proposition 3: Substantively, the bureaucratic efforts for brokerage capacity have increased endogenously, as congressional transaction costs have decreased. As a result, Pareto-improving brokerage has become more achievable in recent decades.

Brokerage Capacity and Congressional Oversight

Because traditional principal–agent theories have emphasized only information asymmetry, the cheapest oversight methods to “acquire information” have been highlighted. However, they have neglected the fact that some oversight methods cause significant contingency control loss. From this perspective, Pareto-improving brokerage is implicative in selecting oversight methods.

Although there are many oversight methods, they can be divided into two types: fire alarm and police patrol. Fire alarm mechanisms have been highlighted since the 1980s. It has been argued that fire alarms are more efficient than police patrols because they enfranchise interest groups to monitor agency actions (McCubbins & Schwartz, 1984) and prohibit faits accomplis by incurring administrative slowness (McCubbins et al., 1989). When legislators enact more “fire alarm” statues, interest groups monitor agency behaviors and inform legislators of bureaucratic drifts more conveniently. However, fire alarm mechanisms impose administrative burdens on agencies, which frequently cause administrative delay, rulemaking ossification, and unexpected contingencies (Asimow, 1994; McGarity, 1992). Thus, fire alarm oversight negatively affects bureaucratic brokerage capacity while decreasing congressional transaction costs. Regarding the game model, fire alarm oversight decreases not only TC_L but also b. This means that fire alarm oversight may incur serious commitment problems in bureaucratic brokerage (see Proposition 2).

If bureaucratic brokerage is Pareto-improving, Congress is likely to prefer allowing agencies to maintain successful brokerage. Thus, the legislature may be reluctant to decrease brokerage capacity through fire alarm mechanisms, if bureaucratic brokerage is Pareto-improving. Rather, the legislature would be more likely to use police patrols, unless the cost of the method is overly high. If Congress hold hearings or collects more data from staffs, it is much easier for interest groups to provide complex information to individual legislators, who already have some background information (Aberbach, 1990). In this sense, police patrol oversight also contributes to reducing congressional transaction costs by internalizing oversight costs, and can cause some commitment problems in reducing TC_L. However, the problems are less serious compared with fire alarm oversight, because police patrols are less likely to impair brokerage capacity without imposing significant administrative burdens on agencies. In contrast, when bureaucratic brokerage is not Pareto-improving, Congress may enact additional procedures to increase distributive gains, because the legislature can enhance its utility by preventing bureaucratic brokerage.⁶ In this case, police patrol oversight would be less likely to be selected.

In a substantive context, given Proposition 3 (i.e., the congressional transaction cost has decreased remarkably, owing to interest group society growth and information system development and as a result, agencies might increase their brokerage capacity endogenously and Pareto-improving brokerage is highly probable), it can be inferred that Congress might have relaxed fire alarm oversight recently because of the increasing probability of Pareto-improving brokerage.

Proposition 4: Fire alarm oversight has been restrained as the congressional transaction cost has been reduced in recent decades. Relatively speaking, it is probable that Congress would like to utilize the police patrol method.

Empirical Examination

The propositions suggested above have at least two different substantive implications. The first one is that the agencies with high brokerage capacity are likely to have more discretion. As a general consequence, agencies tend to have tried to enhance their brokerage capacities. The administrative broker model shows a detailed causal mechanism between brokerage capacity and discretion. The second one is that Congress has relaxed fire alarm oversight methods as agencies increase their brokerage capacity, because agency brokerage is Pareto-improving under low congressional transaction costs. In this section, these two hypotheses are empirically examined.

Brokerage Capacity and Discretion

Since the 1970s and 1980s, the American political landscape has become highly open and fragmented (Baumgartner et al., 2009; Hula, 1999; Salisbury, 1990); diverse interest groups have developed and acquired professional knowledge (Kaufman, 1981; Kearney & Sinha, 1988). This political transition has ended the era of information monopoly by bureaucrats. Vigorous signaling behaviors of interest groups have contributed to collapsing information asymmetry between Congress and agencies.⁷ Despite this trend, many agencies have enjoyed significant discretion, promoting brokerage capacity endogenously. Their investment in brokerage capacity has become greater, as the congressional transaction cost has dropped since the late 1980s.

For example, federal advisory committees as a typical method for agencies to negotiate with important policy stakeholders did not develop much until the late 1980s. In 1974, there were 3,626 meetings and 22,702 members participating in the advisory committees. This number was somewhat constant in the 1970s and 1980s.⁸ However, as congressional transaction costs significantly declined in the 1980s and 1990s, federal agencies tried to enhance their brokerage capacity. As a result, there were 6,614 meetings and 62,521 members of advisory committees in 2008, showing significant growth compared with the 1970s and 1980s. Moreover, federal agencies have used diverse mediating methods such as negotiated rulemaking and alternative dispute resolution (ADR).⁹ Since it took root in the 1970s, ADR has experienced rapid growth (Senger, 2004). Almost all the federal agencies have used several ADR methods such as mediation, settlement judges, and conflict coaching. A presidential report said, “Twenty years ago, ADR was just the beginning of an idea. Today, ADR has come of age. It is a framework for management of conflict for both the public and private sectors” (Federal Interagency Alternative Dispute Resolution Working Group, 2007, p. 123). Despite its recent decline, negotiated rulemaking also emerged in the 1980s and the 1990s as one of the important ways for federal agencies to mediate between opposing stakeholders in rulemaking. Based on this effort, several agencies have maintained broad discretion, despite reduced congressional transaction costs and prevalent information.

One example is the EPA. Since the 1970s and 1980s, environmental issues have been highlighted and environmental groups have developed significantly. These political situations threatened the expertise monopoly of the EPA in the 1980s; the EPA’s scientific reviews were frequently criticized by policy stakeholders (Applegate, 1992; Jasanoff, 1990). Moreover, many environmental groups challenged the EPA’s actions in courts and provided administrative information to legislators through hearing participation and lobbying. To illustrate, former EPA Administrator William Ruckelshaus (March 1983 to February 1985) told in a conference in 1984 that more than 80% of the EPA’s rules had been challenged and agency officials felt that their autonomy had been seriously limited by political principals.¹⁰ In other words, the obstacle of information asymmetry was being collapsed by the growth of professional interest groups. Because the EPA had only minimal brokerage capacity in this period, however, the agency could not effectively mediate between interest groups: almost all environmental interest groups had selected Congress rather than the EPA as the policymaking arena. As a result, the agency’s discretion became limited by several public laws. One typical example was the Hazardous and Solid Waste Amendments of 1984 (HSWA, P.L. 98-616), which contained severe limitations on the EPA’s discretion. Rep. James J. Florio (D-NJ), the sponsor of the bill, commented about the law: “Instead of authorizing EPA to regulate the disposal of chemical wastes, Congress has prescribed the limits. Instead of relying on EPA to meet deadlines, Congress has established self-enforcing standards . . . Instead of allowing EPA to establish technical standards of safety, Congress has set minimum requirements” (Florio, 1986, p. 351). This trend continued until the early 1990s when the Clean Air Act Amendments of 1990 (P.L. 101-549) was enacted.

However, the agency began to develop brokerage capacity in this period. The EPA reduced its transaction cost by increasing advisory committees and public meetings. For instance, although there were only 27 advisory committee (7 committees) meetings in 1981, the number increased markedly in the 1980s and 1990s to 120 meetings (22 committees) in 1991 and, 206 meetings (27 committees) in 2008. Moreover, the agency suggested more “agreeable” policy options to interest groups through formal and informal discussions. For instance, the EPA used negotiated rulemakings to derive more compromises on several controversial issues, such as emergency pesticide exemptions, residential woodstoves, coke oven batteries, and wood furniture coating in the 1980s and 1990s. Furthermore, the agency started several cooperative programs, such as the Common Sense Initiative and Project XL. These brokerage efforts of the EPA have been successful in holding interest groups in the bureaucratic venue. Langbein and Kerwin (2000) showed that interest groups express greater satisfaction with the EPA’s negotiated rulemakings compared with conventional rulemakings. Similarly, Lubell (2000) also suggested a survey that the EPA’s collaborative programs, such as the National Estuary Program, have yielded higher levels of attitudinal support or general satisfaction from environmental stakeholders. As a result, interest groups of environmental issues began to remain in the bureaucratic venue of the EPA, rather than moving to other venues. To illustrate, the number of environmental protests has decreased significantly (Agnone, 2007; Olzak & Soule, 2009). The number of American environmental protests reported in the New York Times peaked to 40 cases in the 1970s, and remained active in the 1980s. Then, the trend was slowly reversed and the number of environmental protests has significantly decreased to less than 10 cases in the 1990s (Agnone, 2007). Moreover, the increasing rate of environmental litigation has also slowed down since the 1990s (Gerrard, 2000; Karkkainen, 2002). A Government Accounting Office (GAO) report shows that the number of litigations against the EPA have generally decreased since the mid-1990s (United States Government Accountability Office, 2011). Similarly, the growth rate of reported judicial opinions in environmental citizen suits has decreased since the 1990s, whereas the rate had steeply increased in the 1970s and 1980s (May, 2003).

Consequently, since the early 1990s, Congress has become unable to enact significant restrictions on the EPA (Klyza & Sousa, 2008; Percival, 2007), although the environmental policy area has broadened in recent decades and new issues have emerged continuously.¹¹ In this respect, Foreman (2002, p. 160) said that the EPA’s brokerage efforts have been successful as “political cover against a hostile Congress.” To illustrate, brownfield cleanup issues emerged in the mid-1990s. Then, the EPA initiated the Brownfield Action Agenda in 1995, accentuating negotiations and brokerage in dealing with brownfield sites rather than enforcement based on expertise. This brokerage encouraged professional interest groups to remain in the bureaucratic venue rather than depart to the legislative venue for information provision. Thus, no serious restriction on the EPA regarding brownfield issues has been enacted by Congress. Rather, the Small Business Liability Relief and Brownfields Revitalization Act (P.L. 107-118), which provides the EPA with the statutory authority for brownfield management, was legislated in 2002. Similarly, although a myriad of pollutants and toxics have been found and examined, no public laws containing serious discretion restrictions on the EPA (regarding other important environmental issues such as air and water pollution) have been enacted for about 20 years.¹²

Administrative Laws and Rulemaking

The brokerage capacity affects not only bureaucratic discretion, but also procedural burdens on agencies (especially as the congressional transaction cost has decreased). Since the enactment of the Administrative Procedure Act of 1946 (APA; P.L. 79-404), several procedural burdens have been placed on agency rulemaking. For example, Congress enacted the National Environmental Policy Act of 1969 (P.L. 91-190) that requires all federal agencies to prepare Environmental Assessments and Environmental Impact Statements. In addition, the legislature passed the Regulatory Flexibility Act of 1980 (P.L. 96-354), which directs federal agencies to analyze the potential impact of regulations on small business.¹³ Similarly, the Unfunded Mandates Reform Act of 1995 (P.L. 104-4) requires agencies to assess the effects of federal regulatory actions on state, local, and tribal governments and the private sector. Although these procedural laws promote serious administrative delays and rulemaking ossifications, they enfranchise diverse policy stakeholders to sound “fire alarms” more conveniently.

Because congressional transaction costs were significantly higher before the 1980s, agencies had little incentive to enhance their brokerage capacity. Due to the limited brokerage capacity, bureaucratic brokerage was not likely to be Pareto-improving in this period. Consequently, Congress was likely to “set up” more fire alarms, to acquire more information from interest groups, despite impairing bureaucratic brokerage capacity. However, as congressional transaction cost has dropped since the 1980s, agencies have had more incentive to enhance their brokerage capacity endogenously to block information leakage. These changes transformed traditional principal–agent relations into more favorable ones. Several agencies such as the EPA adroitly dealt with controversial issues through informal/formal negotiations and increased administrative efficiency. As a result, Congress had more incentive to relax fire alarm oversight to relieve procedural burdens on agencies, which can impair bureaucratic brokerage capacity. In this respect, there were several procedural reforms to relax procedural burdens on agencies in the 1990s. Bureaucratic agencies initiated the use of “interim final rules” and “direct final rules,” which avoided burdensome “notice-and-comment” procedures—the most important administrative procedure imposed by the APA. Although these exceptional procedures are recommended by the Administrative Conference of the United States, not enacted by the legislature, Congress has tacitly allowed these rulemaking procedures. Then, bureaucratic agencies have increased their usage of the methods significantly (Asimow, 1999; O’Connell, 2008). In addition, as Pareto-improving brokerage became available, legislators provided agencies with significant brokerage authorities. For example, Congress enacted and reauthorized several laws such as the Administrative Dispute Resolution Act of 1990 (P.L. 101-552) and the Negotiated Rulemaking Act of 1990 (P.L. 101-648). These laws formally authorized agencies to broker interest groups in rulemaking and adjudication.¹⁴

Political Implications and Conclusion

In the arena of public administration, there are multiple political actors involved in policymaking and policy implementation process and their interactions with public agencies have become more complex. As a consequence, it has not been easy to observe all the intricate interactions and to examine the effects of the interactions inductively. Therefore, a deductive framework is necessary for consistent empirical examination of public agencies and their interactions with policy stakeholders (Gill, 1995). This formal model of “administrative broker” offers a parsimonious framework in which to explain the influence of “political” agencies.

This model is implicative in the sense that brokerage capacity is a “political power,” not a neutral capability, and that this capacity is meaningful under the current pluralistic society and prevalent information. Traditional bureaucratic politics literature has mainly focused on bureaucratic expertise and ideological conflicts between agencies and political principals (Epstein & O’Halloran, 1999; Huber & Shipan, 2002). Although several studies have emphasized bureaucratic capacity (Huber & McCarty, 2004; Ting, 2009), the capacities are also organizational and neutral, rather than political. Given this tradition, the only meaningful principle developed regarding bureaucratic politics was the ally principle, which implies that a bureaucratic agency can acquire much discretion only when its ideology is similar to that of their political principal. However, the ally principle is robust only under the limited situations where brokerage capacity is minimal. The principle is vulnerable and even meaningless if agencies have political “brokerage capacity” to affect political environments and actors, especially when the agencies can achieve Pareto-improving results with high brokerage capacity. In the case of the EPA as seen above, although the agency tends to be definitely “liberal” (Clinton & Lewis, 2008), its discretion was not much restrained much even during the mid-1990s and early 2000s, when Republicans were dominant in the Congress. As several studies have asserted, bureaucratic agencies are by nature political (Bardach, 1980; Carpenter, 2001; Majone & Wildavsky, 1979). However, this does not only mean that agencies have their own ideological stances. Rather, agencies have their own “political power” to affect policy stakeholders through brokerage.

The “political” brokerage capacity has yielded variations in the relationships between the legislature and agencies. Simply put, agencies with high brokerage capacity tend to acquire significant discretion, even in the era of prevalent information. They can block information leakage and maintain information asymmetry, despite the emergence of a professional and pluralistic interest group society. Moreover, high brokerage capacity even allows agencies to transform the relations with the legislature into more favorable ones: Pareto-improving relations between Congress and agencies are available. Furthermore, as congressional transaction costs have declined in recent decades, these implications of the brokerage capacity have been increasing.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Jongkon Lee is a PhD candidate in the Department of Political Science at the University of California, Berkeley, CA, USA. He is interested in bureaucratic politics and principal–agent theory.

References

Aberbach

J. D.

(1990). Keeping a watchful eye: The politics of congressional oversight. Washington, DC: The Brookings Institution.

Aberbach

J. D.

(2002). What happened to the watchful eye? Congress & the Presidency, 29(1), 3-23.

Aberbach

J. D.

Putnam

R. D.

Rockman

B. A.

(1981). Bureaucrats and politicians in western democracies. Cambridge, MA: Harvard University Press.

Adelman

D. E.

(2010). A cautiously pessimistic appraisal of trends in toxics regulation. Washington University Journal of Law & Policy, 32, 377-442.

Agnone

J.-J. M.

(2007). Amplifying public opinion: The poilcy impact of the U.S. environmental movement. Social Forces, 85, 1593-1620.

Applegate

J. S.

(1992). Worst things first: Risk, information, and regulatory structure in toxic substances control. Yale Journal on Regulation, 9, 277-353.

Asimow

(1994). On pressing McNollgast to the limits: The problem of regulatory costs. Law and Contemporary Problems, 57(1), 127-137.

Asimow

(1999). Interim-final rules: Making haste slowly. Administrative Law Review, 51, 703-755.

Balla

S. J.

Deering

C. J.

(2001). Police patrols and fire alarms: An examination of the legislative preference for oversight. Washington, DC: George Washington University.

10.

Bardach

(1980). On designing implementable programs. In Majone

Quade

E. S.

(Eds.), Pitfalls of analysis (pp. 138-158). New York, NY: Wiley.

11.

Baumgartner

F. R.

Berry

J. M.

Hojnacki

Kimball

D. C.

Leech

B. L.

(2009). Lobbying and policy change: Who wins, who loses, and why. Chicago, IL: University of Chicago Press.

12.

Boehmke

F. J.

Gailmard

Patty

J. W.

(2005). Whose ear to bend? Information sources and venue choice in policy-making. Quarterly Journal of Political Science, 1(2), 139-169.

13.

Bosso

C. J.

Collins

M. T.

(2002). Just another tool? How environmental groups use the internet. In Cigler

A. J.

Loomis

B. A.

(Eds.), Interest group politics (6th ed., pp. 95-114). Washington DC: CQ Press.

14.

Carpenter

D. P.

(2001). The forging of bureaucratic autonomy: Reputations, networks, and policy innovation in executive agencies, 1862-1928. Princeton, NJ: Princeton University Press.

15.

Clinton

J. D.

Lewis

D. E.

(2008). Expert opinion, agency characteristics, and agency preferences. Political Analysis, 16(1), 3-20.

16.

Clune

W. H. I.

(1983). A political model of implementation and implications of the model for public policy, research, and the changing roles of law and lawyers. Iowa Law Review, 69(1), 47-125.

17.

Coase

R. H.

(1960). The problem of social cost. The Journal of Law and Economics, 3(1), 1-44.

18.

Eisner

M. A.

Worsham

Ringquist

(1996). Crossing the organizational void: The limits of agency theory in the analysis of regulatory control. Governance, 9, 407-428.

19.

Epstein

O’Halloran

(1994). Administrative procedures, information, and agency discretion. American Journal of Political Science, 38, 697-722.

20.

Epstein

O’Halloran

(1995). A theory of strategic oversight: Congress, lobbyists, and the bureaucracy. Journal of Law, Economics, & Organization, 11, 227-255.

21.

Epstein

O’Halloran

(1999). Delegating powers: A transaction cost politics approach to policy making under separate powers. New York, NY: Cambridge University Press.

22.

Federal Interagency Alternative Dispute Resolution Working Group. (2007). Report for the president on the use and results of alternative dispute resolution in the executive branch of the federal government. Washington, DC.

23.

Florio

J. J.

(1986). Congress as reluctant regulator: Hazardous waste policy in the 1980’s. Yale Journal on Regulation, 3, 351-382.

24.

Foreman

C. H.

(2002). The civic sustainability of reform. In Kettl

D. F.

(Ed.), Environmental governance: A report on the next generation of environmental policy (pp. 146-176). Washington, DC: Brookings Institution.

25.

Fritschler

A. L.

Hoefler

J. M.

(1989). Smoking and politics: Policy making and the federal bureaucracy. Upper Saddle River, NJ: Prentice-Hall.

26.

Gailmard

(2009). Discretion rather than rules: Choice of instruments to control bureaucratic policy making. Political Analysis, 17(1), 25-44.

27.

Gerrard

M. B.

(2000). Trends in the supply and demand for environmental lawyers. Columbia Journal of Environmental Law, 25(1), 1-15.

28.

Gill

(1995). Formal models of legislative/administrative interaction: A survey of the subfield. Public Administration Review, 55(1), 99-106.

29.

Grossman

G. M.

Helpman

(2001). Special interest politics. Cambridge, MA: MIT Press.

30.

Huber

J. D.

McCarty

(2004). Bureaucratic capacity, delegation, and political reform. The American Political Science Review, 98, 481-494.

31.

Huber

J. D.

Shipan

C. R.

(2002). Deliberate discretion? The institutional foundations of bureaucratic autonomy. New York, NY: Cambridge University Press.

32.

Hula

K. W.

(1999). Lobbying together: Interest group coalitions in legislative politics. Washington DC: Georgetown University Press.

33.

Huntington

(1965). Congressional response to the twentieth century. In Truman

D. B.

(Ed.), The congress and America’s future (pp. 5-31). New York, NY: Prentice-Hall.

34.

Jasanoff

(1990). The fifth branch: Science advisers as policymakers. Cambridge, MA: Harvard University Press.

35.

Karkkainen

B. C.

(2002). Environmental lawyering in the age of collaboration. Wisconsin Law Review, 2002, 555-574.

36.

Kaufman

(1981). Fear of bureaucracy: A raging pandemic. Public Administration Review, 41(1), 1-9.

37.

Kearney

R. C.

Sinha

(1988). Professionalism and bureaucratic responsiveness: Conflict or compatibility? Public Administration Review, 48, 571-579.

38.

Klyza

C. M.

Sousa

D. J.

(2008). American environmental policy, 1990-2006: Beyond gridlock. Cambridge, MA: MIT Press.

39.

Krishna

Morgan

(2001a). Asymmetric information and legislative rules: Some amendments. American Journal of Political Science, 95, 435-452.

40.

Krishna

Morgan

(2001b). A model of expertise. Quarterly Journal of Economics, 116, 747-775.

41.

Langbein

L. I.

Kerwin

C. M.

(2000). Regulatory negotiation versus conventional rule making: Claims, counterclaims, and empirical evidence. Journal of Public Administration Research and Theory, 10, 599-632.

42.

Lubell

(2000). Attitudinal support for environmental governance: Do institutions matter? Proceedings of the Annual Meeting of American Political Science Association. Washington, DC.

43.

Majone

Wildavsky

(1979). Implementation as evolution. In Pressman

J. L.

Wildavsky

(Eds.), Implementation (2nd ed., pp. 177-194). Berkeley: University of California Press.

44.

May

J. R.

(2003). Now more than ever: Trends in environmental citizen suits at 30. Widener Law Review, 10(1), 1-47.

45.

McCubbins

M. D.

Noll

R. G.

Weingast

B. R.

(1989). Structure and process, politics and policy: Administrative arrangements and the political control of agencies. Virginia Law Review, 75, 431-482.

46.

McCubbins

M. D.

Schwartz

(1984). Congressional oversight overlooked: Police patrols versus fire alarms. American Journal of Political Science, 28, 165-179.

47.

McGarity

T. O.

(1992). Some thoughts on deossifying the rulemaking process. Duke Law Journal, 41, 1385-1462.

48.

Moe

T. M.

(2006). Political control and the power of the agent. Journal of Law, Economics, and Organization, 22(1), 1-29.

49.

O’Connell

A. J.

(2008). Political cycles of rulemaking: An empirical portrait of the modern administrative state. Virginia Law Review, 94, 889-986.

50.

Ogul

M. S.

(1976). Congress oversees the bureaucracy: Studies in legislative supervision. Pittsburgh, PA: University of Pittsburgh Press.

51.

Olzak

Soule

S. A.

(2009). Cross-cutting influences of environmental protest and legislation. Social Forces, 88, 201-225.

52.

Parker

D. C. W.

Dull

(2009). Divided we quarrel: The politics of congressional investigations, 1947-2004. Legislative Studies Quarterly, 34, 319-345.

53.

Percival

R. V.

(2007). Environmental law in the twenty-first century. Virginia Environmental Law Journal, 25(1), 1-35.

54.

Ringquist

E. J.

(1995). Political control and policy impact in EPA’s office of water quality. American Journal of Political Science, 39, 336-363.

55.

Rose

C. M.

(2005). Environmental law grows up (more or less), and what science can do to help Lewis & Clark Law Review, 9, 273-294.

56.

Rourke

F. E.

(1984). Bureaucracy, politics, and public policy (3rd ed.). Boston, MA: Little, Brown.

57.

Salisbury

R. H.

(1990). The paradox of interest groups in Washington—More groups, less clout. In King

(Ed.), The new American political system (2nd ed., pp. 203-229). Washington, DC: American Enterprise Institute.

58.

Scher

(1963). Conditions for legislative control. The Journal of Politics, 25, 526-551.

59.

Senger

J. M.

(2004). Federal dispute resolutoin: Using ADR with the United States government. San Francisco, CA: Jossey-Bass.

60.

Sloof

(2000). Interest group lobbying and the delegation of policy authority. Economics & Politics, 12, 247-274.

61.

Stewart

R. B.

(2001). A new generation of environmental regulation? Capital University Law Review, 29(1), 21-182.

62.

Stigler

G. J.

(1966). The theory of price. New York, NY: Macmillan.

63.

Susskind

McMahon

(1985). The theory and practice of negotiated rulemaking. Yale Journal on Regulation, 3(1), 133-165.

64.

Susskind

Ozawa

(1983). Mediated negotiation in the public sector. American Behavioral Scientist, 27, 255-279.

65.

Tarlock

A. D.

(2004). Is there a there there in environmental law? Journal of Land Use & Environmental Law, 19, 213-254.

66.

Ting

M. M.

(2009). Organizational capacity. New York, NY: Columbia University

67.

United States Government Accountability Office. (2011). Environmental litigation: Cases against EPA and associated costs over time. Washington, DC.

68.

Wilson

J. Q.

(1989). Bureaucracy: What government agencies do and why they do it. New York, NY: Basic Books.

69.

Wright

J. R.

(1996). Interest groups and congress: Lobbying, contributions, and influence. Boston, MA: Allyn and Bacon.