What's the damage? Environmental regulation with policy‐motivated bureaucrats

• DOI: http://doi.org/10.1111/jpet.12299
• Author: Jörg Lingens,Achim Voss
• Date: 01 August 2018
• Published date: 01 August 2018

Received: 24 February2017 Accepted: 19 March 2018

DOI: 10.1111/jpet.12299

ARTICLE

What's the damage? Environmental regulation

with policy-motivated bureaucrats

Achim Voss1Jörg Lingens2

1Departmentof Economics, University of

Hamburg

2MuensterSchool of Business and Economics,

Universityof Muenster

AchimVoss, Department of Economics, Uni-

versityof Hamburg, Von-Melle-Park 5, 20146

Hamburg,Germany (achim.voss@wiso.uni-

hamburg.de).

JörgLingens, Muenster School of Business

andEconomics, University of Muenster, Am

Stadtgraben9, 48143 Muenster, Germany

(joerg.lingens@wiwi.uni-muenster.de).

In choosing environmental policy, governments rely on information

provided by bureaucrats, who may have a political motivation of

their own. We analyze the ensuing principal–agent relationship and

derive the government's optimal contract. We find that a regula-

tory agent who is more environmentalist than the government is

rewarded for truthfully stating that the environmental impact of the

regulated economic activity is low (and vice versa). The bureaucrat

has a stronger influence on policy if there is greater uncertainty

about the environmental impact, or if the policy choice has a strong

weight in his utility function. For some impact values, the bureaucrat

is permitted to set his own preferred policy, which is a form of opti-

mal delegation.

1INTRODUCTION

In the Pigouvian model of environmental regulation, the government maximizes welfare by equalizing marginal dam-

age costs and marginal abatement costs of pollution. Real-world environmental regulation, however,is hampered by

imperfect information. For instance, firms'information about their abatement cost is usually better than the regula-

tor's. Optimal mechanisms for overcoming this problem have been analyzed in the literature by,for example, Kwerel

(1977), Spulber (1988), Laffont and Tirole (1993), Lewis (1996), and Hoel and Karp (2002).

The underlying assumption in this literatureis that the government has perfect information about damage costs and

is able to implement the necessary regulatory policy to internalize them. These assumptions, however, seem to be at

odds with real-world policy making. It requires considerable expertise to estimate damage and abatement costs, and

to choose and implement the right policy instruments.

For these reasons, a government consisting of elected politicians usually relies on a specialized bureaucracy that

bundles the required expertiseand is able to implement the policy. As an example, consider estimates of the social costs

of carbon. Official estimates are used in government cost-benefit analysis and are necessary for all kinds of regulatory

policy (see, e.g., Pizer et al., 2014). Their calculation is delegated to environmental agencies likethe Environmental Pro-

tection Agency in the United States or the Umweltbundesamt in Germany.Decisions of the government are thus based

on the information provided by the agency,such that it has considerable influence on policy. In models of environmen-

tal regulation, this relationship between the government and the bureaucracyis treated as a black box. Our paper aims

at understanding the agency problems in this context and their influence on policy.

Journal of Public Economic Theory.2018;20:613–633. wileyonlinelibrary.com/journal/jpet c

2018 Wiley Periodicals,Inc. 613

614 VOSS AND LINGENS

Wemodel the interaction between a government and a regulatory bureaucrat who can be understood as the head of

an environmental agency.The government wants to maximize aggregate welfare.1To do so, it instructs the bureaucrat

to set an environmental policy,for which we use the example of an energy tax. The bureaucrat's objective is similar to

the government's, but his preferences concerning the balance between goods production and environmental quality

may differ from those of the government,implying a different preferred policy.2

The bureau's expertise in judging the marginal environmentaldamage costs of energy usage, in short referred to as

impact, transforms the problem of uncertainty into one of asymmetric information. The bureaucrat is tempted to mis-

representthe truth to get closer tohis own preferred policy. The government has to takethe bureaucracy's preferences

as given because such agencies usually are permanent organizations whose specialized staff cannot be replaced at will.

But the government can implement a mechanism that induces the bureaucrat to tell the truth and implement a policy

that is closer to the government's objective. In order to achieve this deviationfrom the bureaucrat's preferred policy,

the government can offer various benefits—for example, budgets, office endowments, or promotions. For simplicity,

we model them as reward payments.

In the language of information economics, the government is the uninformed principal and offers a tax and reward

schedule—a contract—toits environmental bureaucrat (who is the informed agent). The government is uncertain about

the environmental impact; from its perspective, it is drawn from a randomdistribution. The agent, by contrast, knows

the realization of this draw. Thus, even though the impact is a characteristicof the economy in the model, due to the

asymmetric information it becomes a characteristicof the agent and it takes the role of the agent's private-information

parameter or type.3

The government cannot make the policy or reward conditional on whether the agent reveals the true environ-

mental impact, but it can design the contract in an incentive-compatible way.That is, the policy and the reward must

depend on the announced impact in a way that makesrevealing the truth optimal for the agent. Incentive compatibility

implies that the government must compensate the bureaucrat for sacrificing his ability to conceal the truth. Accord-

ingly, the bureaucratearns an information rent. This rent is the utility increase that the agent obtains from accepting

the contract—which implies a compromise policy and some reward—compared to turning down the contract, forgoing

any reward, and implementing the policy that he prefers.4

The regulatory agent is policy-motivated, which means that he cares about environmental damage, irrespective of

whether or not he accepts the contract. Technicallyspeaking, his participation constraint is type-dependent. Hence,

the agent may haveincentives to overstate the impact, in order to receive higher compensation because a high impact

reduces his contract utility,or to understate the impact, in order to receive higher compensation because a low impact

would reduce his utility without the contract, such that he would need a higher compensation for accepting it. The

government's optimal contract has to take these countervailing incentives (Lewis & Sappington, 1989) into account. To

derive it, we apply the method proposed by Maggi and Rodríguez-Clare (1995), based on optimal control theory with

a nonnegativity constraint on the agent's rent. Technically, our model is also close to Martimort and Semenov (2008),

who model lobby influence on a governmentunder uncertainty about the government's policy preferences.

Jullien (2000) offers a general analysis of type-dependent participation constraints. Jullien'sanalysisisparticularly

relevant if the type-dependent participation constraint could either lead to the exclusionof agents or to bunching. The

1Throughout the paper,we do not explicitly model a government or elections. Instead, we assume that the government acts as a principal representing the

citizens'interest.

2This could be due to selection effects, or future career prospects in the regulated industry.On the one hand, young eco-activists may choose fields of study

that, intentionally or as a side effect, qualify for governmental services in this field, leading to a “greener” bureaucracy.On the other hand, environmental

regulatorsmay sometimes be biased toward industry, possibly through being too familiar with those they should regulate (cf.Muehlenbachs, Staubli, & Cohen,

2016).

3By contrast, the agent's policy preferences are assumed to be common knowledge and thus they are not a type in this sense. That the state of the world

becomes the agent's type can also be the case if the state results from the agent's effort, as in the model of Roger (2016). In Roger's model, however,the

agent does not directly care about the value of that parameter,but only about her own effort and payments, such that an additional mechanism of audit and

punishmentis used for an optimal contract.

4Theassumption that the regulator can refuse the government's offer and implement his preferred policy represents an extreme case. This allows to analyze

the influence of a powerful regulatory agency,and how its expertise augments this influence. (Additionally, it frees us from the need to model an additional

stageoutside the contract).