The Review of International Organizations

, Volume 5, Issue 4, pp 475–496

Enforcing international environmental cooperation: Technological standards can help


    • Department of Political ScienceColumbia University

DOI: 10.1007/s11558-010-9086-1

Cite this article as:
Urpelainen, J. Rev Int Organ (2010) 5: 475. doi:10.1007/s11558-010-9086-1


Market instruments, such as emissions trading or pollution taxes, are less costly than “command and control” regulation. Yet technological standards are common in international environmental agreements and now figure prominently among proposals to mitigate global warming. I show that technological standards can be combined with market instruments to create collective enforcement power. They allow states to internationally enforce technology installation, so the payoff to free riding decreases. A notable feature of the argument is that technological standards and market instruments are complements, while previous research has treated them as substitutes. Empirically, technological standards are most valuable if international cooperation is difficult to enforce and the rate of technological change in concerned industries is slow.


Environmental policyInternational cooperationTechnological standardsInstitutional designGame theory

JEL Codes


1 Introduction

According to environmental economists, market instruments, such as emissions trading, minimize the cost of pollution abatement (Baumol and Oates 1988). Defined as “regulations that encourage behavior through market signals rather than through explicit directives regarding pollution control levels or methods,” they assign a price or quota on pollution without specifying how exactly emissions should be reduced (Stavins 2003). This flexibility allows producers and consumers to exploit their private information and creates a strong incentive for innovation.

Given the superiority of market instruments, it is surprising that technological standards—explicit regulations that mandate the installation of certain equipment for pollution abatement—figure prominently in international environmental agreements. They constrain the range of alternatives for pollution abatement, so they increase the cost of international environmental cooperation. Yet from the 1972 Helsinki Convention for the Protection of the Baltic Sea to the 2001 Stockholm Convention on Persistent Organic Pollutants, states have prescribed or proscribed the use specific technologies. Most importantly, the idea of “sectoral agreements” based on technological standards has recently gained currency in the global warming debate (Barrett 2008).

I provide a new rationale for technological standards. My argument is based on two key observations. First, technology installation largely precedes actual pollution abatement in time: companies invest in equipment because they anticipate legal restrictions on pollution. Second, technology standards are easier to monitor than fully flexible installation: if states contract on a specific technological standard, it is easier for them to verify that any given state has complied with the technology-installation requirement.

Based on these observations, I demonstrate that while technological standards increase the total cost of cooperation, they also reduce the temptation to “free ride” (Barrett 2003; Mitchell and Keilbach 2001). Without technological standards, a state can first avoid the immediate cost of technology installation and subsequently free ride while others provide the public good. Under technological standards, however, a failure to comply with the technological standard is easier to observe. Thus, other states can suspend cooperation, and the defector loses the benefit of free riding. While flexible installation minimizes the cost of cooperation, technological standards improve enforceability by reducing the incentive to free ride.

This simple argument has not been recognized in previous research. Mitchell (1994) and Ausubel and Victor (1992) argue that in some cases, such as the marine pollution regime, technological standards are easier to monitor than market instruments. In contrast, the present argument applies even though market instruments are also perfectly observable. The key difference is that in my model, technology installation is not a substitute for pollution abatement. Even if states impose a technological standard, they must subsequently coordinate efforts to abate pollution. Consequently, my argument can explain technological standards in cases where the standard monitoring rationale is invalid.

Sawa (2008) claims that it is possible to use trade sanctions to enforce technological standards without violating the rules of the World Trade Organization. Conversely, the theory that I propose does not depend on elements outside the issue area. Barrett (2006) shows that they can help states coordinate efforts to develop breakthrough technologies. However, this rationale only applies if superior market instruments are somehow unavailable. My theory is consistent with, but qualitatively different from, each argument.

The contribution is notable for several reasons. First, it explains why states often use a policy mix in international environmental cooperation. Previous research has treated various policies, such as technological standards and market instruments, as alternatives (Barrett 2006; Mitchell 1994). By contrast, the present theory predicts that they go hand in hand. Second, it contributes to the broader debate on the design of international institutions (Keohane 1984; Koremenos et al. 2001). By showing how technological standards and market instruments create collective enforcement power, it provides a new perspective to commitment problems in international politics. Finally, the analysis shows how sequential cooperation influences the design and effectiveness of international institutions. While the extant literature usually examines international cooperation in an unchanging strategic environment (Fearon 1998; Gilligan 2004), the present argument shows that this restrictive assumption is not innocuous.

I first discuss the role of technological standards in international environmental cooperation. I then present and solve the model. Discussion and a summary of empirical implications follow. The Mathematical Appendix presents several extensions of the model.

2 Motivation

Environmental economists offer several reasons why market instruments are superior to “command and control” regulation. First, individuals have private information that is unavailable to the government, so flexibility admits a greater range of options for pollution abatement (Aidt 1998; Barrett 2003; Baumol and Oates 1988; Kruger et al. 2007; Oates 1999; Stern 2006; Weitzman 1974). Second, technological standards are a barrier to technological change (Fischer and Newell 2008; Kverndokk et al. 2004; Weyant and Olavson 1999). If individuals must choose a specific technology, they have fewer incentives to invest in research and development.

In domestic politics, hidden transfers to special interests could motivate technological standards or other inefficient policy instruments (Buchanan and Tullock 1975; Keohane et al. 1998; Maloney and McCormick 1982). For example, the government could protect domestic companies from foreign competition (Kono 2006). However, this argument does not apply to international environmental agreements. Even if governments collude with special interests at home, they have incentives to minimize the cost of international environmental cooperation.

According to the conventional wisdom, technological standards should only be used if market instruments are unavailable. For instance, conservation areas could be efficient in biodiversity preservation if threatened species cannot be otherwise protected. However, in most cases, states should have a common incentive to minimize the cost of environmental protection.


Against this unfavorable backdrop, it is interesting that technological standards have historically been used in many international environmental agreements. Article II of the 1988 Protocol on Nitrogen Oxides to the 1978 Convention on Long-Range Transboundary Air Pollution (LRTAP) requires that the parties “[a]pply national emissions standards to major new stationary sources and/or source categories, and to substantially modified stationary sources in major source categories, based on the best available technologies which are economically feasible.” These technological standards appear completely unnecessary because the protocol also specifies targets for each state. The 2001 Stockholm Convention on Persistent Organic Pollutants provides a list of chemicals to be eliminated or restricted, with a limited number of exempted uses. This regulation dampens the incentive to develop new uses, whereas a simple fee could have been used to avoid this distortion.

Many international agreements to control water pollution also include technological standards. The marine pollution regime, consisting of the 1973 International Convention for the Prevention of Pollution from Ships and a modifying 1978 Protocol, prescribed a binding technological standard (Mitchell 1994). This technological standard did reduce pollution from ships but it also suppressed incentives to develop less costly technologies. The 1974 and 1992 Helsinki Conventions for the protection of the Baltic Sea combine nonbinding quantitative targets with binding technological standards for land sources of effluents, although some states have chosen to opt out (Greene 1998). The 1972 Oslo Convention for the protection of the North Sea, along with the related 1974 and 1992 Paris Conventions, also combine binding technological standards with nonbinding targets for several pollutants (Skjaerseth 1998). Both regimes have been effective, but technological standards increase the cost.

A particularly important recent example is the energy and climate legislation that the European Union adopted in December 2008. It specifies inter alia a legally binding target for renewable energy sources to cover one fifth of the total energy mix by 2020.1 While the overall target seems broad, it is constraining compared to a simple quota on greenhouse emissions. For example, the energy industry must produce one fifth of its output from renewable energy production. The legislation also promotes technological benchmarking. Sawa (2008, 12) writes that “in the EU ... sectoral approaches are being considered in the form of benchmarking, as an effective method for allocating allowances among the actors in the [emissions trading scheme].”

The interest in technological standards is shared by many states outside Europe. In May 2008, the Japanese government submitted a proposal for a sectoral approach to international climate policy after the Kyoto Protocol expires in 2012.2 Among other things, it emphasizes the importance of technological benchmarks. According to Meckling and Chung (2009), the United States has also recently promoted a sectoral approach that builds on technological standards to global warming.


The extant literature provides four main rationales for technological standards. First, technological standards could permit enforcement through trade sanctions without violating international trade law (Barrett 2008; Sawa 2008). Under international trade law, states cannot impose tariffs on polluting imports to prevent “unfair” competitive advantage, but specific technological standards are admissible in certain circumstances. However, this argument only applies if states cannot reconcile international trade law and environmental protection.

Second a related argument emphasizes competitiveness concerns and “carbon leakage,” whereby polluting industries move to states without carbon constraints (Biermann and Brohm 2005; Esty 2001). Some states might be unwilling to accept binding commitments, but they could nevertheless consider technological standards. This would alleviate competitiveness concerns in states that are ready to accept binding commitments. However, it remains unclear why states cannot simply use market instruments to achieve the desired distribution of gains.

Third, market instruments are sometimes difficult to monitor. By choosing a technological standard, states can monitor compliance (Ausubel and Victor 1992; Mitchell 1994). However, this explanation is only applicable to situations in which technological standards are easier to monitor than other instruments. For example, in the case of global warming, technological standards and market instruments could be equally difficult to monitor.

Finally, a technological standard could help states coordinate research efforts. Barrett (2006) shows that while technological standards are often as difficult to enforce as are emissions targets, a technological standard could facilitate equilibrium selection under “economies of scale.” A weakness of this explanation is that it does not consider the possibility of committing to efficient subsidies or other superior incentive schemes for technology development.

3 The Model

In the model, states i = A, B engage in international environmental cooperation. To protect the environment, they must abate transboundary pollution. To reduce the cost of pollution abatement, a representative company in each state must install clean technology. The states can either prescribe a technological standard or allow the industry to select the least costly clean technology.

To facilitate exposition, I model the choice between a technological standard and flexible installation as the choice between two separate games. In the game of technological standards, states choose between a technological standard and non-cooperation. In the game of flexible installation, states choose between flexible installation and non-cooperation. Since the model is symmetric, I investigate the expected payoff to the states in each game and assume that the game producing a higher payoff will be played. Thus, all I need to do is to solve the two games separately and then compare the expected payoffs. For convenience, all model notation is summarized in Table 1.
Table 1

Model notation




Technological standard, intention to cooperate


Compliance, clean technology installation


Pollution taxation




Cost of pollution taxation


Cost of clean technology installation


Extra cost of unilateral pollution taxation


Extra benefit from free riding

In the model, I use several simplifying assumptions that warrant discussion. While these simplifying assumptions abstract away from many standard issues in international environmental cooperation, they allow a sharp focus on the key issue of technology installation here.

First, I do not model international environmental cooperation as a repeated game. While the main results would hold if play was repeated, this would complicate exposition without essential insight. Indeed, one interpretation of my model is that it represents, in reduced form, the decisions to install technology at some time τ and impose a pollution tax subsequently at time τ + 1. Such repetition might be particularly relevant for environmental issues with unusually long time horizons, such as global warming (Asheim et al. 2006; Barrett 2008). For example, states could agree on a technological standard at time τ and suspend cooperation for T periods if one of them fails to impose the technological standard on domestic companies.

Second, I assume that the game is played by two states. The model could be extended to multilateral cooperation as well, but it is better to present the simplest possible model that allows me to easily communicate the main result. If I were to assume N symmetric states instead, all main results would continue to hold, though it should be noted that as the number of players increases, enforcement in general would become more difficult (Barrett 2003).3 Additionally, if I were to allow asymmetries between multiple states, complicated issues of coalition formation would emerge (Barrett 2003; Carraro and Siniscalco 1993).

Third, I do not investigate the microanalytics of technology installation. I consider a representative company without political power that decides on technology installation before the state requires abatement. In reality, companies continuously update production technologies, both before and after abatement is a legal requirement. For the results, it is not necessary that technology installation ceases as abatement begins; nor it is necessary that every company in a polluting industry successfully selects an optimal technology. As long as sufficiently many companies incur the cost of technology installation ex ante, so that the cost–benefit ratio of cooperation is contingent on previous technology installation, the equilibrium analysis is applicable.

Finally, I model the pollution tax as a binary choice. In reality, states may choose among various levels of pollution taxation. However, adding such a continuous choice would not affect the main result: technological standards sometimes reduce the incentive to defect because verification is easier. Indeed, one interpretation of the binary choice is that states first collectively agree on a given pollution tax, and each state subsequently selects between that pollution tax and the optimal defection.

3.1 Technological Standards

The first option that I consider is an international environmental agreement with technological standards. In this game, the sequence of moves is as follows.
  1. 1.

    Each state i simultaneously decides on a technological standard si ∈ {0, 1}.

  2. 2.

    In each state i, a representative company decides on compliance ei ∈ {0, 1}.

  3. 3.

    Each state i simultaneously decides on a pollution tax ti ∈ {0, 1}.

First, both states decide on imposing a technological standard. If state i does so, si = 1, it promises to use a “best available technology.”

Second, a representative company decides on compliance. If it complies, ei = 1, it can always produce. If it does not comply, ei = 0, it can only produce when there is no technological standard, si = 0. I assume production profits π exceed the cost of compliance k, so that π > k > 0. Thus, foreshadowing the equilibrium analysis, it will comply if and only if si = 1. Realistically, I assume that if the representative company complies, the state also incurs a cost \(\overline{C}>0\).

Finally, each state i decides on a binary pollution tax, ti ∈ {0, 1}. It represents a broader class of market instruments. The payoff from pollution taxes tA, tB to state i is given in Table 2. The collective environmental benefit is normalized to ti + tj. The domestic abatement cost is cx.
Table 2

Payoffs from pollution taxation




No tax



2 − cx, 2 − cx

1 − cx − β, 1 + α

No tax

1 + α, 1 − cx − β


Parameter x tells how “stringent” the international environmental agreement must be to solve the problem at hand. If stringency x is low, relatively inexpensive changes are enough to abate pollution, so the international environmental agreement is easy to comply with. If stringency x is high, costly adjustments are necessary to abate pollution, so compliance is a major problem. In reality, the stringency of an international environmental agreement is obviously endogenous, but for tractability I first assume it is exogenous and only endogenize it below.

For realism, I include competitiveness and distributional concerns in the model. If state i unilaterally imposes a pollution tax, it pays an extra cost β. Substantively, it can be thought of as the competitiveness effect of pollution taxation or even intrinsic concerns with distributional issues. I assume this cost is so high that unilateral pollution taxation is not profitable, \(\beta>1-\overline{c}x\). This assumption is not necessarily realistic, as pollution abatement also produces domestic benefits, but it permits a sharp focus on the international dimensions of the problem.

This formulation is in reduced form, so it does not shed light on why exactly competitiveness or distributional concerns emerge. However, the scholarly literature consistently indicates that they are a key issue in international environmental policy (Bechtel and Tosun 2009; Desombre 1995; Neumayer 2001). Additionally, this assumption is not necessary for the key results, as long as there is some cost of “unilateral cooperation” and benefits to free riding.

Conversely, if state j unilaterally imposes a pollution tax, the other state i obtains an additional benefit α. Again, competitiveness effects and distributional consequences are plausible substantive interpretations. As I show below, this benefit must be low enough for international environmental cooperation to succeed. To reduce notation, I do not write α or β as a function of stringency x. This simplification does not change the results.

The unit cost c depends on technology installation. If the representative company in state i has not installed clean technology, ei = 0, the unit cost c is prohibitively high, \(\overline{c}>\frac{1}{x}\). If state i has installed clean technology, ei = 1, the unit cost is \(\underline{c}\), where \(\overline{c}>\underline{c}>0\). Intuitively, the only benefit of technology installation is to reduce the cost of pollution abatement.

The game form is common knowledge. The pollution taxes tA, tB are public. This is important because it implies that the market instrument is easy to monitor. Additionally, state i can observe both the imposition of technological standards, sA, sB, and compliance, eA, eB. Intuitively, a technological standard mandates the installation of a specific technology, so compliance is easy to verify.

3.2 Flexible Installation

If states design an international environmental agreement without technological standards, the sequence of moves is as follows.
  1. 1.

    Each state i privately communicates a cooperation intention to its representative company, \(\tilde{s}_{i}\in\{0,1\}\).

  2. 2.

    In each state i, a representative company decides on installation \(\tilde{e}_{i}\in\{0,1\}\).

  3. 3.

    Each state i simultaneously decides on a pollution tax \(\tilde{t}_{i}\in\{0,1\}\).

This game is mostly identical to the game of technological standards. However, there are two critical differences. First, instead of publicly imposing a technological standard, each state i privately communicates an intention, or lack thereof, to cooperate to the representative company. Since this is “cheap talk,” the value of \(\tilde{s}_{i}\) is only observable to state i and the representative company therein.

How is such communication to be interpreted? In the absence of verifiable technological standards, the state can either (i) tell domestic firms that it intends to cooperate or (ii) inform them that it has no real intention to comply with the international environmental agreement in focus. Since state i has no incentive to mislead the representative company, I assume throughout that the latter believes this announcement is sincere, thus conditioning the installation decision on this communication. Intuitively, the representative company is a “price taker” that does not strategically try to prevent cooperation by not installing. This standard microeconomic assumption is plausible as long as the industry in focus is sufficiently competitive.

Second, the representative company in each state i decides on installation. Installation, \(\tilde{e}_{i}=1\), allows production in all circumstances. Without installation, \(\tilde{e}_{i}=0\), the representative company can only produce if there is no pollution tax, \(\tilde{t}_i=0\). Profits are again π while the cost is \(\tilde{k}\), where \(\pi>k>\tilde{k}>0\). The cost to state i is \(\underline{C}\), where \(\overline{C}>\underline{C}\).

Given that the representative company accepts the prior announcement by state i as sincere, in equilibrium it will install if and only if \(\tilde{s}_i=1\). Intuitively, it expects cooperation only if the state communicates an intention to comply with the international environmental agreement, and technology installation is only optimal if cooperation is forthcoming. The subsequent cost of pollution abatement to each state i is again that given in Table 2.

Although I do not focus on the microanalytics of flexible installation, the following formalism captures the basic idea. Suppose there is a very large number of available technologies, θ1,...,θn, and nature privately reveals to the representative company in each state i an optimal technology, \(\theta_j^i\), that can be installed at cost \(\tilde{k}\), as opposed to the higher cost of a technological standard k. If the representative company expects cooperation, it has an incentive to choose the least costly alternative \(\theta_j^i\). Otherwise it has no incentive to install at all.

Since the number of available technologies is very large, I assume that the states cannot verify the installation decision of the representative companies. Since the states have given the representative companies free hands to reduce technology in every conceivable fashion, they cannot hold the manager accountable for the absence of any given piece of equipment, such as a scrubber. After all, the manager can always say that she has found an alternate solution that will have the same effect at a lower cost. Since states do not have the expertise to evaluate the functionality of new technologies, the cost of verifying that the manager’s proposed solution works would be prohibitively high.

In sum, the key difference to the game of technological standards is that the first two moves are not observable by the foreign state. Indeed, it is as if the representative company was deciding on installation simultaneously with the choice of pollution taxes, \(\tilde{t}_A,\tilde{t}_B\).

Of course, it is unrealistic to assume that the installation decisions are not observable at all. It would be plausible to assume that even flexible installation without technological standards is observable with some positive but small probability. However, this modification would not qualify any of the results. As long as the probability is small enough, each state would believe that installation failure would be detected with virtual certainty and the other state would not cooperate. Thus, the equilibrium incentives that I identify below would remain unchanged.4 Similarly, all results would hold even if the technological standard was imperfectly observable, as long as the probability of detection was high enough.

4 Equilibria

I model the international environmental agreement as a collective choice between the two games introduced above. To begin with, note that in each game, a subgame-perfect Nash equilibrium exists such that no technology is being installed, tA = tB = 0, and no cooperation is forthcoming, tA = tB = 0. These equilibria give each state i a payoff normalized to zero, as shown in Table 2. Thus, the purpose of international cooperation is to avoid this inefficient outcome.

If the states decide to play the game of technological standards, the question is if a cooperation equilibrium exists with the following play on the equilibrium path, so that neither state (or representative company) has an incentive to deviate:
  1. 1.

    Each state i imposes a technological standard, si = 1.

  2. 2.

    A representative company in each state i complies, ei = 1.

  3. 3.

    Each state i imposes a pollution tax, ti = 1.

The intuition is straightforward. The two states incorporate a technological standard in an international environmental agreement. If this equilibrium exists, the states can implement an international environmental agreement with technological standards.
If the states decide to play the game of flexible installation, an equilibrium that meets the following criteria on the equilibrium path must exist or cooperation fails:
  1. 1.

    Each state i privately announces cooperation, \(\tilde{s}_{i}=1\).

  2. 2.

    A representative company in each state i installs the clean technology, \(\tilde{e}_{i}=1\).

  3. 3.

    Each state i imposes a pollution tax, \(\tilde{t}_{i}=1\).

The most important feature of this agreement is that states cannot condition pollution taxation on technology installation. Since flexible installation without technological standards is not public, states must decide on the pollution tax based on prior information. An example is the Montreal Protocol on Substances that Deplete the Ozone Layer, for it mandates national reductions in pollution but does not mandate any specific equipment for pollution abatement.

Given that pollution taxation is prohibitively costly without technology installation, cooperation is impossible without technology installation. Since flexible installation is less costly to state i than compliance with technological standards, it would be collectively optimal for the states to play the game of flexible installation. Thus, if the cooperation equilibrium exists in the flexible installation game, states will play it.

5 Analysis

An international environmental agreement is in equilibrium if neither state i has an incentive to deviate. To compare technological standards and flexible installation, I find the most stringent agreement x that admits cooperation in the two games. It turns out that if the cost of imposing a technological standard is not excessive, technological standards allow greater stringency.

To begin with, refer to Table 2 to see that pollution taxation is never possible in the cooperation subgame unless
$$ 2-\underline{c}x\geq1+\alpha. $$
The left side is the payoff from cooperation while the right side is the payoff from defection. I assume throughout that this condition holds, so that the benefits exceed the costs if state i has installed clean technology. One plausible interpretation is that with clean technology, the domestic benefits of pollution abatement are substantial enough that the incentive to free ride is suppressed (Vogel 2000). Similarly, it is easy to verify that if state i does not install technology, \(e_{i},\tilde{e}_i=0\), the prohibitive cost \(\overline{c}>\frac{1}{x}\) prevents pollution taxation, \(t_{i},\tilde{t}_i=0\). Intuitively, pollution taxation is too costly if state i has not adopted the clean technology.

Flexible Installation

Consider the flexibility game. If condition 1 holds, should state i inform the industry that pollution taxation is forthcoming? If it does, the total payoff is \(2-\underline{c}x-\underline{C}\), as both representative companies install and states cooperate in equilibrium. If state i deviates by not announcing cooperation, the representative company will not install, so the total payoff to state i simply 1 + α. Notably, state j is tricked into unilateral cooperation because technology installation cannot be verified. Consequently, technology installation and international cooperation is possible if and only if
$$ 1\geq\underline{c}x+\underline{C}+\alpha. $$
The left side is the marginal payoff from installation and cooperation. The right side is the marginal cost of installation and cooperation relative to free riding. Importantly, if state i intends to free ride, it does not install clean technology. This dual defection is optimal because clean technology is not valuable if state i intends to free ride.

This expression reveals a major problem with flexible agreements. Suppose state i intends to free ride. First, it avoids the cost of technology installation, \(\underline{C}\). Second, it continues to benefit from cooperation by state j, so the environmental cost of defection is only 1 instead of 2. This incentive to defect is amplified by the distributional gains α. The resulting “double dividend” weakens the enforceability of flexible agreements.

Technological standards

Consider now technological standards. Again, condition 1 must hold in the cooperation subgame. Suppose it does. Should state i impose the technological standard? The technological standard allows verification, so failure to impose a technological standard prevents international environmental cooperation, and the payoff from the subgame is zero, as shown above. Thus, technology installation and subsequent cooperation is profitable if and only if
$$ 2\geq\underline{c}x+\overline{C}. $$
Why is this expression so different from condition 2? The reason is that state i cannot first avoid technology installation and then also exploit state j by free riding. Failure to impose a technological standard prevents international environmental cooperation, so state i forgoes both the environmental benefits 2 and the distributional gain α. The technological standards dissipate the double dividend from free riding on a flexible agreement.

6 Results

The equilibrium analysis provides a full characterization of the advantages and disadvantages of the two agreements. In this section, I present and prove three propositions that summarize the analytical results. To begin with, if cooperation is enforceable, the payoff from the flexible installation is always higher:
$$ 2-\underline{c}x-\underline{C}\geq2-\underline{c}x-\overline{C}. $$
This condition captures the conventional wisdom on cost minimization through market instruments.

Proposition 1

If an international environmental agreement is enforceable under flexible installation, it produces a higher payoff than under technological standards.

In the absence of enforcement concerns, the conventional wisdom on cost minimization is valid. The flexible agreement capitalizes on private information and innovation, so both states benefit from cost minimization.

However, it is not clear that both agreements are enforceable. Conditions 2 and 3 contain the value of mutual cooperation, \(2-\underline{c}x\), so enforceability can be investigated simply by comparing the marginal payoffs from defection, \(1+\underline{C}+\alpha\) and \(\overline{C}\). If the former value is high, it is difficult to enforce the flexible agreement. If the latter value is high, it is difficult to enforce the technological standards.

Let us now fix 1 + α and consider the cost difference \(\overline{C}-\underline{C}\). With 1 + α > 0, it is clear that if \(\overline{C}-\underline{C}\) is small enough, enforceability favors technological standards.

Proposition 2

If the cost difference \(\overline{C}-\underline{C}\) is lower than the payoff from free riding, 1 + α, it is easier to enforce an international environmental agreement under flexible installation than under technological standards.

What are the implications for international environmental cooperation? If the cost difference \(\overline{C}-\underline{C}\) is high, technological standards are neither efficient nor enforceable. But in other cases, the cost difference \(\overline{C}-\underline{C}\) is low.

Investigating Eq. 2, the stringency of a flexible agreement is bounded from above by some x* such that
$$ 1-\underline{c}x^{*}=\underline{C}+\alpha. $$
Given x*, each state i is exactly indifferent between cooperation (left side) and defection (right side). For low enough cost difference \(\overline{C}-\underline{C}\), it immediately follows that
$$ 2-\underline{c}x^*>\overline{C}. $$
Now the highest feasible stringency x**, implicitly defined by \(2-\underline{c}x^{**}=\overline{C}\), must exceed x*. Consequently, technological standards improve the efficacy of international environmental cooperation.

Proposition 3

If the cost difference \(\overline{C}-\underline{C}\) is lower than the payoff from free riding, 1 + α, technological standards admit higher stringency, x** > x*.

Counter to intuition and the conventional wisdom, costly technological standards enable the use of market instruments. If we plausibly interpret stringency as the pollution tax rate, technological standards allow a higher tax, so market instruments and technological standards are “strategic complements” (Topkis 1998). One is not useful without the other, so a policy mix is needed for successful cooperation.

To conclude the analysis, it is useful to note that technological standards may have another surprising advantage over flexible agreements. Suppose states have chosen flexible installation and the exogenous stringency is x*, the highest enforceable one, so that condition 5 holds. After technology installation, they could clearly enforce a more stringent agreement, x** > x*, as the cost of technology installation is already sunk and thus irrelevant.

This is not a problem if stringency is truly exogenous, so that x* cannot be changed. However, it may be more realistic to assume that x* is actually endogenous and initially chosen exactly because it was the highest enforceable stringency. In these circumstances, both states might prefer to substitute a higher stringency x** for x*, so as to reap additional environmental benefits, after the cost of technology installation is no longer relevant. However, if such renegotiation is anticipated, each state i expects the final stringency to be x**, regardless of what they initially agreed on. By assumption, x** was too high to be enforceable, so the flexible installation game does not have a cooperative equilibrium when renegotiation is allowed.5

By contrast, this is never a problem under technological standards. If state i fails to impose the technological standard, it cannot cooperate in any case because the cost is prohibitive. Thus, the expectation of renegotiation from x* to x** cannot remove the incentive to impose a technological standard. After all, such a deviation would prevent any cooperation at all.

7 Implications

The efficiency-enforceability tradeoff stems from the impossibility of writing a treaty that mandates technology installation without sacrificing flexibility. In every state, individual companies hold private information, so states cannot contract on efficient technology installation. Technological standards increase the cost of installation, but they also allow states to credibly commit to reducing the pollution intensity of the economy.


The result depends crucially on the sequential nature of international environmental cooperation. Most theories of international environmental cooperation do not distinguish between technology installation and pollution abatement, but in reality, major investments in new technology are an integral part of pollution abatement (Barrett 2009; Stern 2006). For example, the cost of retrofitting old coal plants for carbon capture and sequestration is prohibitively high (Lackner 2003). If technology installation must come before actual pollution abatement, states can build collective enforcement power by conditioning cooperation on prior technology installation. Since successful international environmental cooperation requires that states first install clean technology and then abate pollution, the second part can be made contingent on the first part. To the degree that other international cooperation problems share this feature, such sequencing could be a promising strategy to improve the enforceability of international agreements.

The examples that I have used are consistent with this sequencing. Article II of the 1988 Protocol to the LRTAP explicitly requires that technological standards based on best available technologies be implemented “no later than two years after the date of entry into force,” whereas the first quantitative restrictions are applied “at the latest by 31 December 1994,” or full 4 years later. The European renewable energy legislation requires that member states submit national plans by July 30, 2010, and contains interim targets already for the 2011–2012 period.6 By contrast, emissions trading begins only in 2013 after the lax requirements for implementing the Kyoto Protocol expire in 2012.7 Finally, although the North Sea and Baltic Sea agreements do not contain binding national targets, it is notable that they require that technological standards be immediately applied upon entry into force.

One substantive interpretation of the “rigid agreement” with technological standards that I have characterized are interim technology commitments. Instead of fully specifying contractual obligations in the long run, states could first agree to develop and install new technologies based on available benchmarks. This reduces the cost of international environmental cooperation in the future and allows states to easily renegotiate the commitments after observing compliance.

For example, the argument could be particularly relevant for efforts to mitigate global warming. In December 2009, major emitters failed to successfully conclude Copenhagen negotiations.8 Among other issues, they disagreed on the pace of emissions reductions and the monitoring mechanism. Particularly pivotal was China who both rejected the ambitious commitments proposed by the industrialized countries and adamantly opposed the idea that commitments accepted by developing countries be monitored by an international body.9

While my model cannot directly capture the problem of emissions monitoring—assumed to be perfectly observable for simplicity—the argument may nevertheless shed light on this issue. If emissions are monitorable, but such monitoring depends on establishing an international body with verification powers, it may be that China finds the accompanying sovereignty cost too high. If my argument is valid, technological standards could alleviate monitoring concerns and pave the way towards gradually deeper emissions reductions. They are intrinsically easier to monitor than broader commitments, so a powerful international body with intrusive verification capabilities is not necessary. And although technological standards are not cost-effective in the conventional economic sense, they reduce the cost of future emissions reductions by facilitating the installation of clean technologies. This is particularly important because global warming is a problem with an unusually long time horizon (Barrett 2009; Stern 2006).

Of course, the enforcement problem that I have identified does not exist in a vacuum. In multilateral environmental cooperation, for example, a major problem is that if one country fails to cooperate, others may not be able to implement a reciprocal punishment, because the collective cost of suspending cooperation is high (Barrett 1999). My model may nevertheless prove useful, as the incentive to defect in the first place depends on how verifiable the commitments are (Ausubel and Victor 1992).

Policy mix

The combination of technological standards and pollution taxation is an essential feature of the analysis. The extant literature on technological standards also highlights enforcement advantages, but there are few results on the optimal policy mix under imperfect enforcement. In my model, the choice of technological standards increases the highest feasible pollution tax, so the efficiency-feasibility tradeoff has new dimensions that previous research has ignored. Economic theory emphasizes the advantages of market instruments while enforcement concerns recommend technological standards, but the present analysis shows that technological standards are often a precondition for using market instruments. This complementarity relationship could explain why international environmental agreements often contain this specific policy mix. As formulated in Japan’s proposal for sectoral agreements to mitigate global warming, technological standards are not intended to “replace quantified national emission reduction targets.”10

The empirical record of technological standards offers ample support to this argument. Of the agreements surveyed above, most combine technological standards with quotas or other market instruments. The main policy instrument of the Convention on Long-Range Transboundary Air Pollution are quantified targets (Murdoch et al. 1997), and the Helsinki-Oslo-Paris Conventions to reduce marine pollution also combine targets with technological standards (Greene 1998; Skjaerseth 1998). Similarly, the European climate and energy package combines technological standards with an overall target for emissions reductions. The notable exception in the set of agreements that I have reviewed, the marine pollution regime, does not use market instruments simply because they are too difficult to monitor (Mitchell 1994). Another notable exception, the Kyoto Protocol, has not managed to secure compliance (Barrett 2008).

The idea of a policy mix is related to monitoring difficulties as follows (Ausubel and Victor 1992; Mitchell 1994). The conventional formulation has emphasized monitoring complications that market instruments cause, but my theory excludes the issue, as both the technological standard si and the pollution tax ti are perfectly observable in the model. Instead, the present argument builds on the notion that flexibility in installation—an indirect consequence of expected pollution taxation—complicates verification. The effect is qualitatively similar, as states must choose technological standards to avoid the problem, but the empirical implications are profoundly different. Where my theory applies, states should use a policy mix that is not available if market instruments are too difficult to monitor. This reasoning shows that the theory is complementary to, but qualitatively different from, the standard monitoring rationale for technological standards. The diverging empirical implications also admit falsification in comparative analysis.

To be sure, technological standards are not the only solution to the time-inconsistency problem that flexible installation creates. For example, states could create a verification agency that measures the pollution intensity of production already before quantitative targets kick in. Similarly, states create a hybrid system, whereby companies are allowed to choose among a range of alternative technological solutions. Alternately, states could adopt gradually more stringent quantitative targets, so as to ensure that the incentive to defect at any given time is relatively low. All three solutions carry certain costs, but if the costs are low enough, they may prove a useful alternative or complement to technological standards. Even in that case, my model is useful because it lays out the basic problem and introduces technological standards as one possible, indeed empirically common, solution.


The renegotiation problem is another conspicuous feature of the theory. In standard international cooperation theory, renegotiation usually refers to the difficulty of enforcing overly ruthless punishments, such as the canonical “grim trigger” strategy in infinitely repeated games (Axelrod and Keohane 1985; Barrett 1994). In contrast, here renegotiation is a contracting problem “on the equilibrium path.” If two states are to enforce an international environmental agreement of stringency x*, they should not suddenly decide to upgrade it to a higher stringency x**. The expectation of such upgrading implies that a free rider has an incentive not to install technology, so the equilibrium implodes. To my knowledge, this form of collective time inconsistency has not been recognized in international cooperation theory.

8 Conclusion

The tradeoff between cost-effectiveness and political feasibility is central to the design of international environmental agreements. To enforce compliance with a treaty that purports to solve an international environmental problem, I have argued, states sometimes gain from imposing technological standards despite powerful economic arguments to the contrary. This claim follows from an examination of the incentives to defect under various international environmental agreements. My key analytical result is that these incentives are usually stronger if states do not impose a technological standard.

Beyond international environmental affairs, the analysis can inform the study of international cooperation and institutions. First, it provides a fresh perspective to enforcement. So far, scholars have mostly focused on direct strategies such as monitoring systems or dispute resolution mechanisms and sanctions (Dai 2002; Downs et al 1996; Koremenos et al. 2001). The present analysis shows that seemingly unrelated design choices, such as provisions for technology installation, can strongly influence the enforceability of international commitments. Given that international law generally imposes rather soft constraints on state behavior (Goldsmith and Posner 2005), it is not implausible that these ancillary factors could often be more important than the actual enforcement mechanisms.

Second, the analysis sheds light on sequential cooperation. The standard assumption of a static strategic environment, modeled as an infinitely repeated game, cannot capture sequencing effects. In addition to the conventional distinction between bargaining and enforcement (Fearon 1998), the present analysis shows that cooperation itself often has a sequential logic. As it turned out, this sequence is crucial for effective institutional design. For instance, such sequencing could shed light on the problem of credible commitment to trade liberalization (Fernandez and Portes 1998; Milner and Kubota 2005). If the benefits of international trade cooperation require such reforms as privatization or regulatory overhaul, sequencing could help industrialized countries enforce the conditions of preferential trading agreements with developing countries.

Finally, the renegotiation problem that I have identified is new to international cooperation theory. In addition to incredible punishments “off the equilibrium path,” the temptation to renegotiate international agreements for mutual gain could compromise their credibility. According to my theoretical analysis, commitments that are initially unenforceable could become enforceable over time. If this transformation increases the initial incentive to defect or free ride, states must collectively commit to retaining an international agreement despite the existence of a superior alternative.


“EU Leaders Reach New Climate Deal.” BBC News December 12, 2008.


See the Mathematical Appendix for a formal proof.


See the Mathematical Appendix for a formal proof.


See the Mathematical Appendix for a formal proof.


“EU Renewable Energy Policy.” EurActiv February 15, 2010.


“The EU Climate and Energy Package.” European Commission February 23, 2010.


“Obama Negotiates ’Copenhagen Accord’ With Senate Climate Fight in Mind.” New York Times December 21, 2009.


“Copenhagen Climate Deal Shows New World Order May Be Led by U.S., China.” Washington Post December 20, 2009.


Article II, paragraph 3.



I have presented this research at the 2009 Annual Convention of the International Studies Association (New York, NY, February 15–18) and at Columbia University. I thank Mark Axelrod, Jorg Balsiger, Scott Barrett, Jennifer Kavanagh, Chris Marcoux, Thania Sanchez, Detlef Sprinz, Laurence Tai, and the seminar audiences for comments.

Copyright information

© Springer Science + Business Media, LLC 2010