Competing for Attention: Lobbying Time-Constrained Politicians

1. Introduction

Popular opinion is that political contributions corrupt policymaking. The majority of Americans believe that “money buys results in Congress,” and that money biases legislation away from the needs of average constituents and in favor of deep-pocketed special interests (Lessig 2011, p. 88). Theoretical models of politics largely support these popular views, with classic models depicting lobbying as little more than a quid pro quo exchange of political contributions for policy (e.g., Tullock 1980; Grossman and Helpman 1994). Insider accounts of policymaking, however, describe overtly corrupt behavior as the rare exception, with most policymakers trying to do the right thing, while facing severe constraints on their time and limits to their expertise (e.g., Bauer, Dexter, and de Sola Pool 1963; Hansen 1991; Schram 1995; Baumgartner et al. 2009; Baumgartner and Jones 2015). We present a model of the political process consistent with the insider accounts of policymaking. Ours is a model of informational lobbying, where political contributions help special interest groups (IGs) capture the attention of a time constrained policymaker, and are not provided in the direct exchange for policy favors. We show how political contributions lead the policymaker to focus his efforts on the most beneficial policies. By allocating his limited attention based on political contributions, the policymaker can guarantee better policy outcomes than in the absence of contributions.

We develop a model of lobbying that approximates the political process described in a number of insider accounts, which emphasize how severely constrained policymakers are in their ability to learn about and implement policy proposals (e.g., Bauer et al. 1963; Hansen 1991; Baumgartner and Jones 2015). In our model, a policymaker faces a number of policy proposals and must choose which to implement. The proposals may involve earmark funding for projects within the policymaker's home district. Due to budget constraints, the policymaker may not be able to fund all of the projects he believes are beneficial.2 Alternatively, the proposals may involve introducing legislation reforming policy on alternative issues. Drafting, introducing, and promoting legislation is time-consuming, preventing the policymaker from introducing legislation in all areas that may benefit his constituents. Although the proposals may not by nature be mutually exclusive alternatives,3 resource and time constraints necessitate that the policymaker prioritize proposals and choose only some on which to focus his efforts.

Before choosing which proposals to implement, the policymaker can assess the costs and benefits of alternative options by meeting with IGs and experts, reading government reports or independent research, holding legislative hearings, or asking staff or government agencies to research the political feasibility and constituent benefits of different options.4 If the policymaker carefully assesses all alternatives, then he can be certain to implement the most beneficial options. But, this is generally not feasible. Time and resource constraints prevent the policymaker from carefully assessing all proposals, just as they limit the number of proposals he can implement. Faced with binding time and resource constraints, the policymaker must prioritize proposals, first choosing which to review and then choosing which to implement.

The main contribution of this paper is to analyze in detail one method the policymaker may use to choose which proposals to review: a Contest for Attention. In a contest for attention, IGs representing each policy proposal pay political contributions in competition for the policymaker's limited attention, with the policymaker reviewing the proposals whose IGs paid the highest contributions. The policymaker sells political access to the highest bidders. This is the type of exchange between policymakers and IGs that the general public and campaign finance reform advocates view as corrupt and detrimental for the average citizen.

Our main result stands in contrast to the popular idea that selling access is necessarily detrimental for policy outcomes. We show that a policymaker who sells his limited attention to the highest bidder will become fully informed about the quality of all proposals even though he can directly review only some of them. Allocating attention to the highest contributors is not detrimental for policy, but rather leads to a fully informed policymaker and the first best policy.5

The reason for this is as follows. Each IG's equilibrium contribution is strictly increasing in the quality of its proposal. IGs involved with high-quality proposals are willing to pay more to capture the policymaker's attention (leading him to review their proposals) than IGs with lower quality proposals. This means the policymaker directly observes the proposal quality of those who win attention (the highest contributors), and indirectly infers the proposal quality of others from their contributions. As a result, the policymaker is fully informed about the quality of all proposals, and implements policy as if he directly observed each proposal's quality. Although in equilibrium the IGs who submit the highest payments see their proposals implemented, this is not because they submitted the highest payments. Rather, it is because the groups with the highest quality proposals submit the highest payments, and in equilibrium the policymaker implements the highest quality proposals. We show that this insight is robust to a number of alternative assumptions and generalizations, which we discuss in the paper and explore in detail in an online appendix in the Supporting Information.

The analysis also considers equilibrium contributions, an important consideration as the policymaker may benefit from both collecting contributions and implementing good policy. We show that giving attention to the highest contributors not only leads to the first best policy, it can also result in the first best level of contributions.6 This means that a policymaker may not need to sacrifice policy in order to maximize contributions. The result stands in contrast to other models of lobbying in which the trade-off between implementing good policy and collecting campaign contributions plays a central role (e.g., Prat 2002; Coate 2004; Cotton 2009). This is the case in our model when IGs share the same valuation for having their proposals implemented, when the policymaker reviews no more proposals than he is able to implement, and when attention is allocated through an all-pay contest for attention. Here, a policymaker who cares about both policy and contributions can be no better off than when he allocates attention through a contest for attention before implementing the proposals he believes are best. There exists no other mechanism by which the policymaker can choose which proposals to implement that will result in higher payoffs. In other settings, the contest for attention still leads to the first best policy, but does not necessarily maximize contributions compared to the selling policy directly. There, using a contest for attention is preferable to any other method of implementing policy when the policymaker cares enough about policy relative to payments.

The environment with asymmetric IGs gives insight into the effects of special interest wealth on policy outcomes. In that setting, contributions are maximized when the policymaker allocates attention through a handicapped contest for attention. Such a contest is biased against wealthy IGs, requiring them to pay higher contributions for the same contest “score” as a lower paying, less wealthy group. In the payment-maximizing full-information equilibrium, rich IGs tend to contribute more than poor IGs, but they are no more likely to have their proposals implemented. Despite higher payments from rich interest groups, the contest for attention still leads to a fully informed policymaker who implements the first best policy.

Although the analysis is motivated by lobbying and policymaking, the underlying framework and the implications of competing for attention are applicable to a number of settings beyond politics. Consider the following examples. An employer wants to hire the highest ability applicants and grants interviews to the most persistent applicants, those who have undertaken the greatest costs to gain the employer's attention.7 Similarly, a venture capitalist may listen to pitches from the entrepreneurs who have done the most to gain his attention. A bachelorette may accept a date from the suitor who is most persistent or makes the biggest fool out of himself in order to capture her attention, or she may accept a date from the suitor who offers to take her out to the most expensive restaurant. In each of these situations, a decision maker must allocate a limited number of “prizes” among agents, and prefers to award the prizes to the highest quality agents (where quality is orthogonal to an agent's willingness to pay for a prize). The decision maker may review some (but not all) agents to learn their quality before choosing how to allocate prizes.8 Agents make payments or undertake costly actions (or submit bids) observable to the decision maker, who reviews the agents who submit the highest payments. The model suggests that by awarding attention to the highest bidder, the decision maker may be better able to identify and award the prizes to the most qualified agents.

2. Relationship with the Literature

The only other paper to consider a contest for attention is Cotton (2009), which brings a highly stylized version of the contest for attention into a more traditional model of lobbying, and uses the model to study campaign finance reform policies. A policymaker chooses whether to sell policy or sell political access (which allowed disclosure of private evidence) through an all-pay auction. When the policymaker sells policy, the game is similar to a traditional all-pay auction for policy (e.g., Hillman and Riley 1989). When the policymaker sells access, the game becomes a highly stylized version of the divisible resource framework that we consider in Section 4. and the online appendix. Both Cotton (2009) and the current paper make the point that a policymaker who sells attention (i.e., political access) can become informed about a policy by observing payments made by IGs as they vie for attention. Cotton (2009) makes this point with a very simple version of the contest for attention in the process of deriving results about contribution limits and taxes. The focus there is on campaign finance reform, not the contest for attention itself. The current paper, on the other hand, considers contests for attention in detail. We generalize the contest for attention, show that the main insight extends to a variety of environments, and derive a number of important new results that were not possible in the stylized version of the contest for attention considered previously.9

A handful of other papers have also modeled the connection between political contributions and access to policymakers (i.e., attention). In Austen-Smith (1998) and Cotton (2012), IGs must pay a price set by a policymaker in order to engage in informational lobbying. Both consider how a policymaker may offer more access to wealthy IGs, who are willing to pay more.10 These papers focus on how a policymaker restricts access in order to extract rent from the political process. In contrast, we focus on whether a time-constrained policymaker can sell attention in such a way that he is able to learn about and implement the best policy, despite not having the capacity to review all policy options.11 Groll and Ellis (2014) model a market for political access where commercial lobbying firms connect citizens to politicians. In the political access models of Austen-Smith (1995) and Lohmann (1995), contributions lend credibility to unverifiable information presented by IGs. This is in contrast to our framework, as well as the other political access papers discussed above, where information is verifiable and may be observed by a policymaker who devotes attention to an issue. In Levy and Razin (2013), IGs compete in an all-pay auction, and the winner has its proposal considered by policymakers. What is meant by consideration differs greatly between their framework and ours. In their framework, there is no uncertainty about proposals, and consideration means being put up for a vote in the legislature, facing off against the status quo policy. In our framework, consideration means a policymaker learns about the costs and benefits of a proposal before deciding whether to move forward with it. Their work focuses on how a status quo policy evolves over time. Ours focuses on how competing for attention improves the policymaker's ability to learn about the best policy.

The majority of the literature models lobbying as a system of quasi-bribery, where contributions are given in exchange for policy outcomes (e.g., Tullock 1980; Grossman and Helpman 1994), or as a system of information provision in the absence of political contributions (e.g., Austen-Smith and Wright 1992; Cotton and Dellis Forthcoming). Bennedsen and Feldmann (2006) and Dahm and Porteiro (2008) consider both information provision and the exchange of money for policy.12 Although the theoretical literature focuses on the use of political contributions to buy policy, this role of contributions is supported by neither insider accounts of the lobbying process nor empirical evidence.13 Rather, empirical accounts of lobbying suggest that contributions help IGs capture the attention of policymakers, assuring that their policies receive full consideration. For example, Langbein (1986), Ansolabehere, Snyder, and Tripathi (2002), and Hall and Wayman (1990) present evidence that IGs provide political contributions in order to secure access to policymakers. See also the excellent descriptions of the policymaking process in Bauer et al. (1963); Hansen (1991), and Baumgartner et al. (2009), and the surveys by Herndon (1982); Schram (1995), and Makinson (2003).

There is a substantial literature on the efficient allocation of resources. Esteban and Ray (2006) show that both high wealth and economic desirability increase lobbying by special interests, and show that this can lead to greater misallocation when these factors are not observable. Cotton (2013) and Fullerton and McAfee (1999) consider environments in which a decision maker wants to allocate a limited resource based on some characteristic orthogonal to applicant value. Most related to our current paper is Fullerton and McAfee (1999), where a contest designer must select a subset of agents to participate in a research tournament. They show that using a contest to allocate entry into the research tournament ensures that only the most promising researchers enter the tournament. In our setting and in Fullerton and McAfee, higher quality agents (e.g., those advocating on behalf of higher quality policies) are willing to pay more to “participate” in the next stage of the game, whether the next stage is our paper's review process or Fullerton and McAfee's research tournament.

All-pay auctions are typically used to directly allocate prizes (e.g., Hillman and Riley 1989; Baye, Kovenock, and de Vries 1993). In our framework the all-pay auction is used to allocate the policymaker's attention. When we allow for observable agent asymmetries, the policymaker's preferred method for choosing which proposals to review involves a handicapped all-pay auction. Such an auction fully adjusts IG payments to account for known heterogeneity. There exists a growing literature on handicapping contests. Siegel (2014) develops a general all-pay contest framework with handicaps in an environment with complete information. Kirkegaard (2012) considers the use of handicaps in a model with private information about valuations. Eso and Szentes (2007) show that a version of a handicap auction can maximize revenue in a game in which the auctioneer chooses how much information to reveal to the bidders about the value of the good. The present paper assumes that bidders reveal information to the auctioneer (not the other way around), and the auctioneer's goal is to collect as much information as possible (rather than revenue maximization). We show how a handicapped auction can be used to maximize the revelation of information, not only to maximize revenue. None of these other models use a contest to elicit information about bidder types.

3. Model with Symmetric IGs

We begin with a relatively simple version of the model. There are n symmetric policy proposals that differ only in quality. The policymaker reviews one of the proposals before choosing one proposal to implement. This symmetric game with one review slot and one implementation slot provides a simple setting in which to develop intuition for our main results. After presenting the results for the game with one review slot and one implementation slot, we show that the results continue to hold when the policymaker reviews and implements multiple proposals. In later sections, we allow for observable asymmetries between the proposals (specifically, preference intensity of the IGs advocating in favor of the proposals).

3.1. Setting

A policymaker is faced with a set of policy proposals , and must choose which of these n proposals to implement. The proposals may represent different earmark funding requests for projects within the policymaker's district, or reform proposals on different issues. Let denote the quality of proposal and the underlying state of the world. Each is the independent realization of a random variable distributed on according to continuously differentiable distribution F with density f.14 Distribution F is common knowledge. The policymaker is ex ante uncertain about the realized quality of each proposal.

The policymaker has the resources to implement one of the proposals. Let () indicate that the policymaker implements (does not implement) proposal i, where denotes the full policy outcome.

Before implementing a proposal, the policymaker may review one of the proposals to directly observe its quality. Let () indicate that the policymaker reviews (does not review) proposal i, where . A review may involve meeting with IGs or lobbyists to gain a better understanding of the proposal, asking staff or government agencies to conduct research, or holding legislative hearings. When the policymaker reviews proposal i, he perfectly observes . Let represent the information directly observed by the policymaker by reviewing a proposal, where when he reviews i and when he does not review i.

Each proposal is supported by an independent IG, an advocate on behalf of its proposal. refers to the IG associated with issue i. Each IG, an expert on its respective project or policy area, privately observes at the beginning of the game. Before the policymaker decides which proposal to review, the IGs may independently provide payments to the policymaker in an effort to capture his attention. Such payments may be monetary political contributions or in-kind transfers. Let denote any “contribution” made by IG_i, where .

We adopt an all-pay auction to model the allocation of the “review slot” based on payments. The IGs provide payments, and the policymaker reviews the proposal associated with the IG that provided the highest payment. In the “contest for attention” for a single review slot, if for all , and ties are broken randomly.

The policymaker's payoff is made up of contribution utility and policy utility :

(1)

Contribution utility is strictly increasing in individual contributions (and total payments ). Policy utility is strictly increasing in the quality of the implemented proposal, . All else equal, the policymaker prefers to implement the highest quality proposal. Therefore, quality may represent net benefit to the policymaker's constituents or effects on the policymaker's chances of reelection. Parameter represents how much the policymaker cares about policy relative to contributions.

An IG's payoff depends on whether its proposal is implemented and its payment. For each ,

(2)

Parameter v represents the relative value IGs put on policy outcomes relative to payments; in later sections, we allow this parameter to differ across IGs. An IG is an advocate in that its willingness to pay to have its policy implemented is independent of proposal quality.15 The game takes place as follows. First, each IG_i privately observes its own proposal's quality , and then chooses contribution . Second, the policymaker reviews the proposal associated with the highest payment, directly learning that proposal's quality. He updates his beliefs about the quality of the other proposals accounting for the payments made by the IGs. Third, the policymaker chooses one proposal to implement.

3.2. Preliminaries

The analysis solves for the perfect Bayesian equilibria of the game. A complete description of equilibrium must define the following:

• (1) Equilibrium contribution strategy for each IG, where describes the equilibrium value of when IG_i observes that its proposal is quality .
• (2) Equilibrium implementation strategy, , used by the policymaker to determine which proposal to implement given c and σ.
• (3) The policymaker's posterior beliefs, given c and σ. We denote these beliefs by μ.

These components constitute an equilibrium if (1) no IG has an incentive to deviate from given the payment strategies of the other IGs and the implementation strategy of the policymaker, (2) the policymaker's implementation strategy is sequentially rational given his beliefs μ, and (3) the policymaker's posterior beliefs μ are consistent with and σ. Function μ denotes the policymaker's updated beliefs about q, where is the probability the policymaker puts on state given contribution profile c and review outcome σ. Similarly, is the posterior probability proposal i is quality .

We impose an assumption on the behavior of IGs when indifferent between multiple contributions.

ASSUMPTION 1.An IG that is indifferent between multiple contributions chooses the contribution that maximizes the expected probability that its proposal is implemented, given the equilibrium strategies of the other IGs and the policymaker.

This assumption plays a role in the analysis only in cases where IGs are indifferent between contributing according to equilibrium function and contributing other amounts. Any equilibrium found under Assumption 1 will also be a perfect Bayesian equilibrium in the absence of the assumption. In the online appendix, we consider implications of both this assumption and the alternative assumption that an indifferent IG minimizes payments.

3.2.1. Full-information equilibria

The primary insight of our paper shows how the policymaker becomes fully informed about the quality of all proposals when he sells attention (i.e., the review slot) to the highest bidder. We say that the contest for attention leads to the policymaker being “fully informed” if there exists an equilibrium of the contest for attention in which his equilibrium beliefs, μ, put probability 1 on the true state of the world: . We begin by providing a sufficient condition on the IGs' equilibrium payment strategies to guarantee that the policymaker is fully informed.

LEMMA 1.If for each and all possible , then the policymaker is fully informed in equilibrium. All Proofs are relegated to the Appendix.

When the policymaker reviews a proposal, he perfectly observes , and therefore has correct beliefs about its quality. Lemma 1 shows that in equilibrium the policymaker will also have correct beliefs about the quality of the proposals he does not review. Strictly monotonic payment functions mean there is a one-to-one mapping between the quality of each proposal and the payment made by its IG. In equilibrium, when the policymaker observes payment , he expects the payment was generated by the IG's equilibrium contribution function . Because is strictly increasing, it is invertible; we define . The policymaker therefore updates his beliefs about upon observing , expecting that . In equilibrium, these beliefs are correct.

We refer to any perfect Bayesian equilibrium for which the condition in Lemma 1 is satisfied as a full-information equilibrium. The analysis limits attention to such equilibria.

Before proceeding with the analysis, it is helpful to determine the policymaker's sequentially rational policy choice in any full-information equilibrium. In the final stage of the game, the policymaker implements the proposal with the highest expected quality given his beliefs μ.16 Denote the policymaker's posterior beliefs about by .

LEMMA 2.If the policymaker is fully informed, then in equilibrium he implements the highest quality proposal with probability 1.

3.2.2. Payment-maximizing full-information equilibrium

Lemma 2 shows that in any full-information equilibrium, the policymaker implements the highest quality proposal. This means that an IG with proposal quality expects equilibrium benefit , where

(3)

Parameter v is the value to an IG of having its proposal implemented, and is the probability that proposal i is the highest quality proposal. Expression 3 gives the expected benefit to an IG when the policymaker certainly implements the highest quality proposal. It implies an individual rationality constraint for each IG_i:

(4)

No IG will ever pay more than its expected benefit from lobbying, and in any full-information equilibrium, its expected benefit equals .

LEMMA 3.There does not exist a full-information equilibrium in which for any and .

The maximum feasible payment from IG_i equals . We define a payment-maximizing full-information equilibrium as a full-information equilibrium in which all IGs pay their maximum feasible amounts: for each and .

3.3. Solving the Contest for Attention

The preliminary analysis determined that in any full-information equilibrium, the policymaker's posterior beliefs are always correct, and he implements the first best policy. Here, we begin with a full characterization of the set of symmetric full-information equilibria of the contest for attention game. In a symmetric full-information equilibrium, all IGs play the same strategy , where for all possible .

To derive the equilibrium payment function, we consider IG_i's best response when all other IGs contribute according to . Because the other IGs contribute according to the equilibrium payment function, the policymaker has correct beliefs about their proposal qualities, regardless of whether they win attention. If they do win attention, the policymaker observes directly. If they do not win attention, the policymaker (correctly) believes , where . We derive equilibrium payment functions such that IG_i has no incentive to deviate from contributing according to the strategy when it expects other IGs to do the same.

No incentive to over-contribute—We first explain why IG_i prefers to any higher contribution. If IG_i provides the equilibrium contribution, it submits the highest contribution, wins attention, and has its proposal implemented if and only if it has the highest quality proposal.17 Contributing introduces the possibility that IG_i wins attention even when it does not have the highest quality proposal. This does not increase the probability that proposal i is implemented, however, because winning attention guarantees that the policymaker reviews and directly learns . Since the policymaker's beliefs about the quality of the other proposals are accurate, the policymaker continues to implement proposal i if and only if it is the highest quality proposal. Therefore, overbidding requires a larger payment without providing an expected benefit to the IG.

No incentive to under-contribute—The equilibrium contribution function must also be such that no IG wants to under-contribute compared to the equilibrium payment. If IG_i deviates downward, its costs decrease (a benefit). But, it also decreases the probability that IG_i submits the highest payment and has its policy implemented (a cost). To rule out such deviations, a decrease in contribution must lead to a sufficiently large decrease in the probability of winning attention. This will be the case when the slope of the equilibrium contribution function is sufficiently low.18

We will derive a condition on the equilibrium contribution function such that each IG_i prefers to any lower amount when the other IGs contribute according to the equilibrium function. Notice that for any contribution , IG_i wins attention and has its proposal implemented if and only if it provides the highest contribution. If another IG provides a higher contribution, then IG_i will not win attention. The policymaker will have expectations that , which will be less than the observed quality of the IG that contributes according to the equilibrium function and wins attention. The policymaker does not observe the true quality of , and will therefore implement the other proposal regardless of whether that proposal was actually higher quality than i. If, on the other hand, IG_i provides the highest contribution, then IG_i wins attention. Even before the policymaker reviews proposal i, he expects that it is highest quality proposal based on the contributions. After he reviews the proposal, he learns that i is even higher quality than he expected, and will implement it over the other options.

The expected payoff to IG_i from any is

(5)

Given the definition of B from 3, the expected payoff can be rewritten as

(6)

For to constitute an equilibrium, IG_i must prefer contributing to any lower value. Contributing the equilibrium amount results in expected payoff , which must be larger than 6 for all . This will be true for all possible when

(7)

Individual rationality further requires that .

Summary of equilibrium requirements—These conditions and the requirement that a full-information equilibrium requires for all are the foundation of the first proposition.

PROPOSITION 1.For any function such that for all and , there exists a full-information equilibrium in which for each . No other symmetric full-information equilibria exist.

Notice the implications of this result. First, it establishes that the policymaker may become fully informed about the quality of all proposals, even if he only reviews one of them. In equilibrium, the fully informed policymaker is guaranteed to implement the highest quality proposal. Selling attention to the highest contributor improves policy outcomes. Second, it shows that there are in fact many full-information equilibria.

Payment-maximizing equilibrium—Next, we describe the full-information equilibrium that maximizes contributions from all IGs, and the payments received by the policymaker. Such an equilibrium corresponds with the highest level of competition between the IGs, an equilibrium characteristic that may serve as a focal point and lead to coordination on such an equilibrium.

COROLLARY 1.There exists an equilibrium in which for each . This is the payment-maximizing full-information equilibrium.

In the payment-maximizing full-information equilibrium, each IG contributes . The policymaker extracts all rent from the policymaking process. From an ex ante perspective, the expected individual contribution is , and the sum of total expected contributions equals

The finding that total IG payments equal v is remarkable. Even if the policymaker explicitly sold proposal implementation (rather than attention) to the highest bidder or by making a take-it-or-leave-it price to one of the IGs, he would not be able to collect more than v in expected payments from the IGs. There does not exist a method for deciding which proposal to implement that results in higher payments than the payment-maximizing equilibrium of the contest for attention. And, unlike directly selling proposal implementation to the IGs, the contest for attention has the added benefit of guaranteeing the first best policy.

PROPOSITION 2.The payment-maximizing full-information equilibrium results in the first best policy and the first best level of contributions for the policymaker. There exists no other method for implementing policy that leads to higher expected payoffs for the policymaker.

This results implies that a policymaker who cares about both the quality of the implemented proposal and collecting payments can do no better than first allocating attention through a contest and then implementing the proposal he believes best. Such an allocation method guarantees both the first best policy outcome and the maximum expected payments.

This means there does not necessarily exist a trade-off between implementing the best policy and maximizing political contributions. This result is in contrast to other papers on lobbying, where there is a trade-off between implementing policy that is good for constituents, and implementing policy that encourages contributions from special interests (e.g., Prat 2002; Coate 2004; Bennedsen and Feldmann 2006; Cotton 2009).

This result, however, is not as strong in Section 3.5., where we consider IG asymmetries. There, we show that the contest for attention continues to guarantee the first best proposal implementation even if IGs differ in their valuations. However, the contest for attention will no longer guarantee the highest possible total payments from IGs, although it will maximize payments compared to any other method that guarantees the first best policy.

3.4. Implementing Multiple Proposals

The above results extend to a setting where the policymaker reviews the k proposals associated with the highest IG payments, and then implements the m proposals he believes are highest quality, where . Here, we focus on the case where the number of review slots is no greater than the number of proposals the policymaker can emphasize, including the extreme case where the policymaker reviews only a single proposal before implementing multiple proposals.19 By allocating as few as one review slot through a contest for attention, the policymaker becomes fully informed about the quality of all proposals and is able to implement the m highest quality alternatives.

There are a number of differences between the analysis in the case where that we consider above, and the analysis in this section. First, the ex ante expected benefit to IG_i when the policymaker certainly implements the m highest quality proposals is now

(8)

the value of having its proposal implemented times the probability that fewer than m other proposals are higher quality.

Second, when there is the possibility that in equilibrium the policymaker implements a proposal without first reviewing it. This is consistent with the idea that a legislator may vote on issues he has not had time to study in detail.20 This means the policymaker will implement some proposals without first reviewing them, thereby altering the incentives IGs have to over-contribute and signal higher quality proposals than they actually have. If it is sufficiently low-cost to inflate the policymaker's beliefs about , then the IG will prefer to over-contribute compared to equilibrium. This means that must not be too low; it must be sufficiently costly to signal a marginally higher quality. Otherwise, IGs would have the incentive to at least marginally inflate their payments in an effort to provide the impression (when they are not reviewed) that their proposals are higher quality than they truly are.

The incentives IGs have to under-contribute compared to equilibrium are similar to the previous section. The equilibrium requires that for all , which is the multiple-proposal equivalent to 7 from the previous section. Proposition 3 establishes that the constraint that be increasing at a sufficiently hight rate never contradicts the constraint that it must be increasing at no greater rate than B.

PROPOSITION 3.In the contest for attention game where , there exists a function such that

• (1) for all , and
• (2) for all functions such that for all and , there exists a full information equilibrium in which for each .

The equilibrium in which is the unique payment-maximizing full-information equilibrium.

In the payment-maximizing equilibrium, total expected payments equal , implying that once again the policymaker is able to fully extract all of the rent from the policymaking process. Again, a policymaker who cares about both the quality of the implemented proposal and collecting IG payments will never expect higher utility than when he allocates attention to the highest bidder before implementing the proposals he believes best.

PROPOSITION 4.Consider the contest for attention game with k review slots and m implemented proposals, where . The payment-maximizing full-information equilibrium results in the first best policy and the first best level contributions for the policymaker. There exists no other method for implementing policy that leads to higher expected payoffs for the policymaker.

4. Model with Asymmetric IGs

In reality, IGs differ in their benefit from having their proposal implemented, or opportunity costs of providing political contributions. In this section, we allow for such observable asymmetries between the IGs. Except for the following changes, the underlying setting remains identical to the previous sections.

The IGs' payoffs depend on whether their proposal is implemented and their payment: . Parameter represents the relative value IG_i puts on policy outcomes relative to payments. A higher may be interpreted as either IG_i having more at stake from their proposal being implemented, or as i being more wealthy or having lower costs of funds.22 Without loss of generality, we rank-order IGs according to their value, such that .23

The policymaker reviews k and then implements m proposals, where . The expected benefit to IG_i if the policymaker always implements the m highest quality proposals can be defined as , which differs from as defined in 8 only in that v is replaced by in its argument. Where clear, we refer to by . Given that , it follows that for all i and ; the benefit to from participating in a full-information equilibrium is lower than the benefit to any other IG.

In the previous section, the contest for attention was symmetric, giving the same weight to all IGs' payments, and awarding attention to the k IGs that provided the highest payments. If the policymaker uses such a symmetric contest to allocate attention when IGs are asymmetric, he will still become fully informed and implement the first best policy in equilibrium, as long as the differences between the IGs are not too large. While a symmetric contest for attention results in the first best policy, it is not optimal for a policymaker who also cares about contributions.24 For the remainder of this section, we allow for a more general asymmetric (i.e., “handicapped”) contest for attention, which requires greater payments from some IGs than from others for the same equilibrium probability of winning attention. We show how such a contest is optimal for the policymaker compared to any other method that guarantees the first best policy.

We generalize the contest for attention, modeling it as an asymmetric all-pay contest defined as a set of score functions , one for each IG. When IG_i provides payment , its “score” is given by , where and for all . The proposals associated with the highest scores in the contests receive attention. That is, the policymaker reviews proposal i if fewer than k other proposals have .

A policymaker may require higher payments from IGs with higher than from IGs with lower , in exchange for the same equilibrium probability of winning attention. The revenue-maximizing contest for attention involves score functions that fully adjust for observable IG asymmetries. Such a contest involves for all . In equilibrium, each IG provides a payment , and an IG wins attention and has its policy implemented if and only if it has one of the k highest . In equilibrium, the policymaker implements the first best policy, and collects the maximum possible revenue compared to any other mechanism that guarantees the first best policy. We provide a detailed analysis in the online appendix. Proposition 6 summarizes this result.

In equilibrium, IGs with higher tend to pay more than IGs with lower , implying higher contributions from more wealthy groups or groups with more to gain from having their proposal implemented. This does not, however, translate into either the attention or implementation decisions being biased in favor of more wealthy IGs. In equilibrium, more wealthy groups pay more, but the contest accounts for observable differences between IGs, and in equilibrium only the highest quality IGs (regardless of wealth) gain attention and have their proposals implemented.

When IGs differ only in proposal quality, a policymaker who cares about both collecting payments and proposal quality can be no better off than when he sells attention to the highest bidders before implementing the proposals he believes best. This procedure for implementing policy results in the first best total payment (equal to in expectation) and the first best policy outcome. It no longer guarantees the first best total payment in the environment with asymmetric IGs. In this section, the payment-maximizing full-information equilibrium guarantees the first best policy outcome, and guarantees the highest possible payments compared to any other full-information equilibrium. But it no longer guarantees the highest payments compared to certain methods of selling policy.25 The payment-maximizing full-information equilibrium describes the most profitable outcome associated with any mechanism that guarantees the first best policy outcome. A contest for attention that results in this outcome is preferred to any other method of awarding policy, as long as the policymaker cares enough about policy relative to payments (i.e., as long as λ in his payoff function is not too small).

5. Other Considerations

We summarize a number of extensions, which illustrate the generality of our results. Detailed consideration of these extensions is provided in the online appendix.

Awarding attention through a winner-pay auction— We have thus far modeled the contest for attention as an all-pay auction. The policymaker may still be fully informed and guaranteed to implement the first best policy if he uses a winner-pay auction to award attention. However, with a winner pay contest, no payment-maximizing full-information equilibrium exists. This means that potential revenue is higher when the contest for attention takes the form of an all-pay contest rather than a winner-pay contest.

Other methods for allocating attention— The online appendix discusses alternative ways that the policymaker may sell access or choose policy. We also consider an alternative version of the game in which IGs can send cheap talk messages instead of payments prior to the allocation of attention.

Noisy quality signals— The model assumes that IGs perfectly observe the quality of their own proposals. It is possible, however, that IGs only observe an imperfect signal of their proposal's quality and are not perfectly aware of how the policymaker will perceive their proposal's quality following a review. In this case, the policymaker learns about the private signal of all IGs from their contributions, but he only becomes fully informed about the true quality of the proposals he actually reviews.

Allocating a divisible resource— We consider an alternative setting in which the policymaker chooses how to allocate a divisible resource when a project's optimal allocation is increasing in its own quality and decreasing in the quality of the other proposals. It is a generalized version of the competition for attention that appeared in Cotton (2009). We show in this alternative setting that a contest for attention leads the policymaker to become fully informed about the optimal allocation.

We reserve the detailed consideration of other extensions for future research. These include the following:

IG budget constraints—Assuming that IGs face binding budget constraints changes the equilibrium outcomes. When all IGs face the same budget constraint, the constraint may represent the implementation of campaign contribution limits as in Cotton (2009). An asymmetric constraint may represent cash or borrowing constraints which may differ across IGs. With budget constraints, an IG's equilibrium payment strategy is no longer strictly increasing in proposal quality, but rather strictly increasing in quality up to a point, before jumping to the maximum budget. This implies that the policymaker will fail to become fully informed about the quality of all proposals when more than k IGs have high enough realizations of quality that they contribute their entire budget.

Unobservable IG asymmetries—When IG differences in v are unobservable, a policymaker will be uncertain whether high contributions are due to IGs having high v or high proposal quality. He will only learn the true quality of proposals he reviews, and will remain less informed about the quality of the other proposals. This adds noise to the policymaking process, which we anticipate is similar to the extension with noisy IG quality signals.

IG have opposing positions on the same policy space—The model requires IGs to have some uncertainty about the proposal quality of the other IGs. The model would apply in a situation where the policymaker is deciding whether to eliminate steel tariffs, where the steel industry knows how many jobs the tariffs save in the steel industry, and manufacturers know how many manufacturing jobs the tariffs costs, but the opposing interests have uncertainty about the effect of the tariffs on jobs in the other industry. Such a situation is consistent with our framework. However, our framework is not consistent with a situation where the opposing interests are fully informed about the policymaker's optimal policy on their issue. That is a different model that may be explored in future research.

6. Conclusion

The paper shows how a policymaker who sells political access to IGs based on contributions may become fully informed about the quality of all policy proposals, and be guaranteed to implement the best policy. This is the case when the policymaker allocates his limited attention through a “contest for attention” before implementing the proposals he believes are best. This result stands in contrast to popular intuition that assumes the exchange of political contributions for access must be detrimental for policy. We find exactly the opposite. Selling political access to the highest bidders can guarantee the first best policy.

We establish that this main insight holds across a variety of assumptions and extensions to the model. The result holds regardless of how constrained the policymaker is, whether IGs are symmetric or asymmetric, whether the contest for attention involves an all-pay or winner-pay contest, and whether the policy proposals involve the allocation of a nondivisible or divisible resource. Our analysis suggests that a contest for attention can result in a fully informed policymaker and the first best policy in a wide variety of settings.

The paper goes on to establish that in some settings a contest for attention may result in both the first best policy and the first best level of total contributions. When this is the case, a policymaker does not need to sacrifice political contributions in order to implement the best policy. This result is in contrast to numerous models of lobbying in which a policymaker can bias policy in favor of IGs in order to increase political contributions. With a contest for attention, it is possible for a policymaker to maximize both contribution and policy utility. When this is the case, it is impossible for a policymaker to be better off in terms of either contributions or policy than when he allocates attention to the highest bidder before implementing the proposals he believes best.

Additional insights are gained when the analysis allows IGs to differ in terms of wealth or policy valuation, as well as proposal quality. In this setting, the payment-maximizing contest for attention is handicapped to fully account for interest group asymmetries. A rich interest group must pay more than an otherwise similar poor group for the same equilibrium probability of access (e.g., big oil must pay more than a local community nonprofit). The result gives insight into the popular intuition that the rich have better access to politicians, and that policy must therefore be biased in favor of the rich. This is not true in our analysis. Although we find that rich interest groups tend to contribute more than poor interest groups, this does not imply that the rich have a higher probability of winning attention or having their proposals implemented. In equilibrium, the policymaker devotes attention to and implements the highest quality proposals. Under our assumptions about policymaker behavior, policy is independent of interest group wealth.

Finally, the competition for attention framework may provide insight into other settings beyond political lobbying. A contest for attention may improve allocation decisions in other environments in which the primary goal is to choose the highest-quality or most qualified options, including the selection of candidates for jobs, scholarships, admissions, or marriage, and the selection of investment opportunities. We leave further consideration of these and other applications to future research.