Abstract
A central issue in climate policy is the question whether longterm targets for greenhouse gas emissions should be adopted. This paper analyzes strategic effects related to the timing of such commitments. Using a twocountry model, we identify a redistributive effect that undermines longterm cooperation when countries are asymmetric and side payments are unavailable. The effect enables countries to shift rents strategically via their R&D efforts under delayed cooperation. In contrast, a complementarity effect stabilizes longterm cooperation, because early commitments in abatement induce countries to invest more in lowcarbon technologies, and create additional knowledge spillovers. Contrasting both effects, we endogenize the timing of climate agreements.
This is a preview of subscription content, access via your institution.
Notes
 1.
We discuss the relevant literature and its relation to our findings in the next subsection.
 2.
We follow the usual modeling approach that whenever countries cooperate, they seek to maximize their joint welfare. Under early cooperation, this implies that they also take into account how their cooperative abatement choices will affect their future, noncooperative R&D efforts.
 3.
 4.
Strategic effects in the context of emissions and R&D policies are analyzed also by Ulph and Ulph (2007).
 5.
 6.
Hence, we model each country as choosing R&D effort directly. Alternatively, we could assume that R&D is carried out by firms, but countries can regulate the firms’ R&D efforts indirectly with a subsidy (see, e.g., Golombek and Hoel 2008). These two modeling approaches are formally equivalent when there is full information (as is the case in our framework), and the subsidy targets directly the firms’ R&D efforts but not the firms’ other choices (such as output or prices).
 7.
For the indices, by convention, let \(i\equiv 2\) if \(i=1\), and \(i\equiv 1\) if \(i=2\).
 8.
Concave benefit functions exhibit \(\partial ^2 B(a_1,a_2)/(\partial a_1 \partial a_2)<0\), which gives rise to what is known as the ‘raising rivals’ costs effect’ in industrial organization. This effect renders the analysis less tractable, while it actually “magnifies the strategic incentive” (Beccherle and Tirole 2011) of delay.
 9.
Recall that in our formulation, investment costs in R&D are included in \(C_i\) as well as the abatement costs.
 10.
The latter condition holds with equality if knowledge is treated as a pure public good.
 11.
The dotted line at the bottom left of the figure indicates that the implementation of the chosen abatement levels under early cooperation is not a decision.
 12.
Nearterm abatement (before R&D levels are chosen) is not explicitly modeled. Underlying this approach is the implicit assumption that nearterm abatement efforts do not interact strongly with the variables of the model. See Beccherle and Tirole (2011).
 13.
E.g., country 1 may raise its R&D effort in the subgame LN in order to induce country 2 to agree to the formation of a late coalition.
 14.
To increase readability, we suppress the functional arguments and the condition \(i\in \{1,2\}\) whenever this does not lead to confusion.
 15.
The outcome would then be defined by the following system: \(\partial C_i/\partial a_i=b_i\) and \(\partial C_i/\partial r_i=0\).
 16.
The outcome would then be defined by the following system: \(\partial C_i/\partial a_i=b_i+\beta _i\) and \(\partial C_i/\partial r_i=0\).
 17.
The outcome would then be defined by the following system: \(\partial C_i/\partial a_i=b_i+\beta _i\) and \(\partial C_i/\partial r_i=0\).
 18.
By continuity, the assumption of symmetric abatement spillovers (\(\beta _1=\beta _2\)) in Lemma 3 can be replaced by the requirement that abatement externalities are not too asymmetric.
 19.
Alternatively, one can reformulate this example by assuming that country 1 benefits also from its own abatement but faces prohibitively high marginal abatement costs (so \(a_1=0\) is always obtained), while country 2 faces prohibitively high marginal R&D costs (so \(r_2=0\) always holds). Under these assumptions, the same results are obtained.
 20.
Formally, this results from \(\partial C_1/\partial r_2 = \partial ^2C_1/(\partial r_1\partial a_1) = \partial ^2C_1/(\partial r_1 \partial r_2) = 0\). See (7), and the expressions for \(\partial r_i^e/\partial a_i\) and \(\partial r_{i}^e/\partial a_i\) given in the proof of Lemma 2. Note also that Lemma 3 does not apply here, as the abatement externalities are asymmetric in this example. This explains why the overall strategic effect under late cooperation can induce higher investments in R&D.
 21.
A more classical example of such asymmetries would be two countries that share a river, with one country being located upstream and the other one downstream.
 22.
From the proof of Lemma 2, we obtain with \(\partial C_2/\partial r_1=0\): \(\partial r^e_1/\partial a_1 = \frac{\partial ^2 C_1}{\partial a_1 \partial r_1} / \frac{\partial ^2 C_1}{\partial r_1^2}>0\), \(\partial r^e_2/\partial a_2 = \frac{\partial ^2 C_2}{\partial a_2 \partial r_2} / \frac{\partial ^2 C_2}{\partial r_2^2}>0\), and \(\partial r^e_1/\partial a_2 = \partial r^e_2/\partial a_1 = 0\).
 23.
To see this, note that countries could neglect the first effect, in which case also the other two effects vanish.
 24.
Note that country 2 could neglect the effect in stage 1, so taking it into account, this country’s welfare is higher. Country 1 benefits from the additional knowledge spillovers, because its own choices of \(a_1\) and \(r_1\) do not have any impact on country 2’s decisions (\(a_1\) enters linearly in \(\Pi _2\), and \(r_1\) does not affect \(\Pi _2\)).
 25.
To see this, note that for high values of \(\beta _1\), under early cooperation the direct effect of this externality upon \(a_1\) becomes very large (which raises \(\Pi _2^e\)), and dominates the strategic effects under early and no cooperation that occur due to the presence of knowledge spillovers. The situation, thus, resembles the case with only a unilateral externality in abatement (Proposition 4).
 26.
A similar specification is used by Harstad (2012).
 27.
This definition is convenient, because welfare often depends on the parameters \(\beta _i\) and \(\rho _i\) only via this expression. The interpretation of it as an ‘aggregate measure of the externalities’ is plausible because the expression depends symmetrically on \(\beta _i\) and \(\rho _i\) and is strictly increasing in these parameters.
 28.
This is because for the given cost functions \(\partial ^2 C_{i}/(\partial a_{i} \partial r_{i})=0\); see the proof of Lemma 1.
 29.
In general, it is not sufficient to compare welfare in these cases in order to derive the equilibrium outcome of the overall game. The problem is that—if late cooperation is expected to fail for a given set of R&D strategies \(r_1\) and \(r_2\)—then a country may change its R&D effort in order to induce the other country to agree to cooperate. However, for the given functional forms, this type of strategy is ineffective, as shown in the proof of Proposition 6.
 30.
The original parameter space is 4dimensional (\(\beta _1,\beta _2,\rho _1,\rho _2\)). It is, thus, remarkable that in cases i–iii, the equilibrium outcome depends only on the relative externalities \(\varepsilon ^r\) and \(\beta ^r\). Only in case iv, the equilibrium outcome (early / no cooperation) depends on the full set of parameters. Hence, to determine the combinations of \(\varepsilon ^r\) and \(\beta ^r\) where country 1 (country 2) is indifferent between early and no cooperation under case iv (\(\Pi _1^e=\Pi _1^n\), resp. \(\Pi _2^e=\Pi _2^n\)), further parameter restrictions must be imposed. To construct the separating curves in Fig. 2, we assumed \(\rho _2=1\), \(\beta _2 \equiv \rho _1\), and then eliminated \(\rho _1\) to obtain two relations \(\varepsilon ^r=\varepsilon ^r(\beta ^r)\), corresponding to \(\Pi _1^e=\Pi _1^n\), resp. \(\Pi _2^e=\Pi _2^n\).
 31.
By symmetry, it suffices to plot the results for \(\beta ^r \ge 1\). Results for \(\beta ^r < 1\) contain no further information.
 32.
Also under this specification, the strategy of country \(i\) to manipulate its R&D effort \(r_i\) to induce country \(i\) to agree to cooperate late when early cooperation has failed, turns out to be ineffective (see the Proof of Lemma 5). This simplifies the characterization of the equilibrium of the overall game (see Proposition 7), because it suffices to compare the payoffs under the different, exogenouslygiven modes of cooperation.
 33.
A ‘Technology Mechanism’ was agreed by the 16’th Conference of the Parties (COP) in Cancun, 2010. “The Technology Mechanism is expected to facilitate the implementation of enhanced action on technology development and transfer in order to support action on mitigation and adaptation to climate change.” http://unfccc.int/ttclear/jsp/TechnologyMechanism.jsp, visited 01/13/2013.
 34.
In the second stage of the early cooperation subgame, countries minimize their individual costs, given \((a_1,a_2)\) determined in the first stage. As a result, reduced cost functions \(C_i(a_1,a_2)\) are obtained that depend only on the abatement levels. The target function in stage 1 is then: \(2baC_1(a_1,a_2)C_2(a_1,a_2)\). This is independent of \(\delta \). The resulting \((a^e_1,a^e_2)\) are, therefore, identical. This symmetry property simplifies the derivation of the early cooperation outcome.
 35.
The parameter restriction \(\delta \in [0,b/2]\) ensures that all nonnegativity constraints are automatically satisfied.
References
Barrett S (1994) Selfenforcing international environmental agreements. Oxford Econ Pap 46:878–894
Barrett S (1997) Heterogeneous international environmental agreements. In: Carraro C (ed) International environmental negotiations: strategic policy issues. Edward Elgar, Cheltenham
Barrett S (2001) International cooperation for sale. Eur Econ Rev 45:1835–1850
Barrett S (2003) Environment and Statecraft. Oxford University Press, Oxford
Barrett S (2006) Climate treaties and “breakthrough” technologies. In: AEA papers and proceedings, vol 96, pp 22–25
Battaglini M, Harstad B (2014) Participation and duration of evironmental agreements. J Polit Econ
Beccherle J, Tirole J (2011) Regional initiatives and the cost of delaying binding climate change agreements. J Public Econ 95:1339–1348
Botteon M, Carraro C (1997) Burden sharing and coalition stability in environmental negotiations with asymmetric countries. In: Carraro C (ed) International environmental negotiations: strategic policy issues. Edward Elgar, Cheltenham
Buchholz W, Konrad K (1994) Global environmental problems and the strategic choice of technology. J Econ 60:299–321
Carraro C, Siniscalco D (1993) Strategies for the international protection of the environment. J Public Econ 52:309–328
Carraro C, Eyckmans J, Finus M (2006) Optimal transfers and participation decisions in international environmental agreements. Rev Int Organ 1:379–396
Eaton J, Kortum S (1999) International technology diffusion: theory and measurement. Int Econ Rev 40:537–570
Finus M (2008) Game theoretic research on the design of international environmental agreements: insights, critical remarks, and future challenges. Int Rev Environ Res Econ 2:29–67
Fischer C, Parry IWH, Pizer WA (2003) Instrument choice for environmental protection when technological innovation is endogenous. J Environ Econ Manag 45:523–545
Folmer H, Mouche Pv, Ragland S (1993) Interconnected games and international environmental problems. Environ Resour Econ 3:313–335
Gerlagh R, Kverndokk S, Rosendahl KE (2009) Optimal timing of climate change policy: interaction between carbon taxes and innovation externalities. Environ Resour Econ 43:369–390
Golombek R, Hoel M (2004) Unilateral emission reductions and crosscountry technology spillovers. BE J Econ Anal Policy 4(2), article 3
Golombek R, Hoel M (2008) Endogenous technology and tradable emission quotas. Resour Energy Econ 30:197–208
Hannesson B (2010) The coalition of the willing: effect of country diversity in an environmental treaty game. Rev Int Organ 5:461–474
Harstad B (2007) Harmonization and side payments in political cooperation. Am Econ Rev 97:871–889
Harstad B (2012) The Dynamics of climate agreements. mimeo
Heal G, Tarui N (2010) Investment and emission control under technology and pollution externalities. Resour Energy Econ 32:1–14
Jaffe AB, Newell RG, Stavins N (2005) A tale of two market failures: technology and environmental policy. Ecol Econ 54:164–174
Kolstad CD (2010) Equity, heterogeneity and international environmental agreements. BE J Econ Anal Policy 10, symposium, art. 3
McGinty M (2007) International environmental agreements among asymmetric nations. Oxford Econ Pap 59:45–62
Requate T (2005) Timing and commitment of environmental policy, adoption of new technology, and repercussions on R&D. Environ Resour Econ 31:175–199
Tinbergen J (1952) On the theory of economic policy. Elsevier, NorthHolland
Ulph A, Ulph D (2007) Climate change: environmental and technology policies in a strategic context. Environ Resour Econ 37:159–180
Acknowledgments
We acknowledge financial support by the German Federal Ministry for Education and Research under the CREWproject (FKZ01LA1121C). We thank the participants in CREW for many fruitful discussions and suggestions. We also like to thank an anonymous referee for his or her helpful comments.
Author information
Affiliations
Corresponding author
Appendix
Appendix
This appendix collects the formal proofs of the lemmas and propositions.
Proof of Lemma 1
Due to the convexity of \(C_i\), it is to show that the righthand side of (5) is nonnegative. In the presence of an abatement externality, we have \(\beta _{i} >0\) so that it remains to be shown that \(\partial a^n_{i}/\partial r_i \ge 0\). To determine the sign of \(\partial a^n_{i}/\partial r_i\), we use the firstorder conditions (4) of stage 2 and insert the functions \(a^n_1(r_1,r_2)\) and \(a^n_2(r_1,r_2)\) that describe the equilibrium outcome:
Applying the Implicit Function Theorem, we obtain (after rearranging):
Because by assumption \(\partial ^2 C_{i}/\partial a_{i}\partial r_{i} \le 0\), the numerator is nonpositive, and due to \(\partial ^2 C_i/\partial a^2_i >0\) the denominator is positive. Hence, \(\partial a_{i}^n/\partial r_i \ge 0\), which proves that the righthand side of (5) is nonnegative. \(\square \)
Proof of Lemma 2
To show that the overall strategic effect in stage 1 of the early cooperation game increases total welfare, note that countries could neglect this effect in the cooperative stage 1. Hence, the effect raises welfare whenever it affects the final outcome, and (by (7)) it affects the final outcome in the presence of R&D spillovers.
To show the second claim, note that the functions \(r^e_1(a_1,a_2)\) and \(r^e_2(a_1,a_2)\) that describe the outcome of R&D competition in stage 2, are implicitly defined by the system \(\partial C_i(a_i, r^e_1(a_1,a_2)\), \(r^e_2(a_1,a_2))/ \partial r_i =0\) for \(i=1,2\) [see (8)]. Applying the Implicit Function Theorem, we obtain (after rearranging):
Given the assumptions \(\partial ^2 C_i/\partial r^2_i>0\), \(\partial ^2 C_i/\partial r_i\partial r_{i} \ge 0\), and \(\partial ^2 C_i/\partial r^2_i > \partial ^2 C_i/\partial r_i\partial r_{i}\), the denominator is always positive. Using \(\partial ^2 C_i/\partial a_i\partial r_i < 0\) and \(\partial ^2 C_i/\partial r_i\partial r_{i} \ge 0\), we find that \({\partial r^e_i}/{\partial a_i} > 0\) and \({\partial r^e_{i}}/{\partial a_i} \le 0\). Using the condition \(\partial ^2 C_i/\partial r^2_i > \partial ^2 C_i/\partial r_i\partial r_{i}\), we find \({\partial r^e_i}/{\partial a_i} > \left {\partial r^e_{i}}/{\partial a_i}\right \). If countries are symmetric, then \({\partial C_{i}}/{\partial r_i} = {\partial C_{i}}/{\partial r_{i}}\). The overall strategic effect is, then, always negative, which [by (8) and (7)] implies higher abatement and higher R&D levels than in the absence of the strategic effect. \(\square \)
Proof of Lemma 3
Using \(\beta \equiv \beta _1=\beta _2\) and inserting the function \(a^l_i(r_1,r_2)\) into (10) yields
Applying the Implicit Function Theorem, we obtain after rearranging:
Using the regularity conditions \(\partial ^2 C_i/\partial a^2_i>0\), \(\partial ^2 C_i/\partial a_i \partial r_i < 0\), \(\partial ^2 C_i/\partial a_i \partial r_{i} \le 0\), and \(\left \partial ^2 C_i/\partial a_i \partial r_i \right \ge \left \partial ^2 C_i/\partial a_i \partial r_{i} \right \), we find that \({\partial a^l_i}/{\partial r_{i}}\) is positive and \({\partial a^l_i}/{\partial r_{i}}\) nonnegative, and \({\partial a^l_i}/{\partial r_{i}} \ge {\partial a^l_i}/{\partial r_{i}}\). Hence, we obtain for the overall strategic effect \(\beta \left( {\partial a^l_{i}}/{\partial r_i}  {\partial a^l_i}/{\partial r_i} \right) \le 0 \), which [by (11)] implies lower R&D efforts than in the absence of the strategic effect. The reduction in R&Dlevels aggravates the problem of underinvestments in R&D (relative to the full cooperation case) and, therefore, reduces total welfare. \(\square \)
Proof of Proposition 1
Without R&D externalities we have \(\partial C_{i}/\partial r_i=0\). Using (2) and (3), the cooperative solution \((a_1^f,a_2^f,r_1^f,r_2^f)\) is with \(i=1,2\) characterized by
According to (8) and (7), early cooperation \((a_1^e,a_2^e,r_1^e,r_2^e)\) is characterized by the system
The result follows immediately because the two systems of optimality conditions coincide. \(\square \)
Proof of Proposition 2
Without abatement externalities we have \(\beta _1=\beta _2=0\). Using (4) and (5), the solution without cooperation, \((a_1^n,a_2^n,r_1^n,r_2^n)\), solves for \(i=1,2\)
Using (10) and (11), we find that the late cooperation outcome \((a_1^l,a_2^l,r_1^l,r_2^l)\) is characterized by the same system of optimality conditions. Hence, the outcome without cooperation and with late cooperation are identical.
Early cooperation satisfies the system (8) and (7), which differs from the other system of equations due to the presence of strategic effects. By ignoring them, in stage 1 countries can assure a total welfare at least as large as under late or no cooperation. \(\square \)
Proof of Proposition 3
The outcome under late cooperation is characterized by the system (10) and (11). According the Lemma 3, the overall strategic effect in stage 1 [see (11)] is welfarereducing if \(\beta _1=\beta _2\). The outcome under early cooperation satisfies the system (8) and (7), which differs from (10) and (11). According to Lemma 2, the overall strategic effect in stage 1 [see (7)] is welfareenhancing. The claim, thus, follows immediately from Lemma 2 and Lemma 3. \(\square \)
Proof of Proposition 4
To show that cooperation in the countries’ choices of abatement targets fails when there is a unilateral abatement externality but no R&D externality, note that the country that generates the abatement externality can achieve its own welfare maximum without any cooperation. Hence, this country can never gain from cooperating.
To show that cooperation fails when there is a unilateral R&D externality but no abatement externality, note first that by Proposition 2 late cooperation in abatement yields no welfare gains. Furthermore, early cooperation fails, because the country that generates the R&D externality has no gains from it. It is assigned a higher abatement level in the cooperative stage (stage 1) in order to trigger additional R&D investments by this country in stage 2. This reduces the welfare of this country that can achieve its own welfare maximum without any cooperation. \(\square \)
Proof of Proposition 5
To show the first claim, note that the country that generates both externalities achieves its individual welfare maximum without any cooperation. Hence, whenever cooperation affects the final outcome, it negatively affects this country’s welfare.
To show that late cooperation fails if each of the unilateral externalities affects a different country, note that for fixed R&D levels, the country that generates the abatement externality suffers from cooperation in stage 2. For any given R&D levels, it attains its individual welfare maximum when ignoring the abatement externality. Hence, it rejects late cooperation.
To show that early cooperation can succeed if both externalities are of comparable strength and each affects a different country, it suffices to demonstrate that there exists an example where this holds. Such an example is provided at the end of Sect. 4.2. \(\square \)
Proof of Lemma 4
Straightforward calculations yield the following results:

1.
No cooperation:
$$\begin{aligned} a_i^n&= 1 \text{ and } r_i^n=1;\nonumber \\ \Pi ^n_i&= 1+\beta _{i}+\rho _{i};\nonumber \\ \Pi ^n&= 2+\beta _1+\beta _2+\rho _1+\rho _2. \end{aligned}$$(16) 
2.
Early cooperation:
$$\begin{aligned} a_i^e&= \frac{3+2 \beta _{i}+\rho _{i}}{3} \quad \text{ and }\quad r_i^e=\frac{3+ \beta _{i}+\rho _{i}/2}{3} < r^f_i;\nonumber \\ \Pi ^e_i&= 1+\beta _{i}+\rho _{i}+\frac{ (2\beta _{i}+\rho _{i})^2}{6} \frac{(2\beta _{i}+\rho _{i})^2}{12};\nonumber \\ \Pi ^e&= 2+\beta _1+\beta _2+\rho _1+\rho _2 + \frac{(2\beta _1+\rho _1)^2}{12} + \frac{(2\beta _2+\rho _2)^2}{12}. \end{aligned}$$(17) 
3.
Late cooperation:
$$\begin{aligned} a_i^l&= 1+ \beta _{i}/2 \text{ and } r_i^l=1=r^n_i;\nonumber \\ \Pi ^l_i&= 1+\beta _{i}+\rho _{i}+\beta _{i}^2/2 \beta _{i}^2/4;\nonumber \\ \Pi ^l&= 2+\beta _1+\beta _2+\rho _1+\rho _2+\beta _1^2/4 +\beta _2^2/4. \end{aligned}$$(18)
A direct comparison yields \(\Pi ^e \ge \Pi ^l \ge \Pi ^n\).
Moreover, \(\Delta \Pi ^{el}_i \equiv \Pi ^e_i\Pi ^l_i = (2\varepsilon _{i}\varepsilon _{i})/12\) so that \(\Pi ^e_i<\Pi ^l_i\) if \(\varepsilon _{i}>2\varepsilon _{i}\). Furthermore, comparing (16) and (18), we find that \(\Delta \Pi ^{ln}_i \equiv \Pi ^l_i\Pi ^n_i = (2\beta ^2_{i}\beta ^2_{i})/4\) so that country \(i\) prefers late cooperation over the noncooperative outcome if \(\beta _{i} < \sqrt{2} \beta _{i}\). Finally \(\Delta \Pi ^{en}_i \equiv \Pi ^e_i\Pi ^n_i =\beta ^2_i(2\beta ^2_{i}/\beta _i^21)/4+\varepsilon _i(2\varepsilon _{i}/\varepsilon _i1)/12\). \(\square \)
Before we state the proof of Proposition 6, we establish the following useful Lemma:
Lemma 6
If \(\Delta \Pi ^{ln}_i(r_1,r_2) \equiv \Pi ^l_i(r_1,r_2)  \Pi ^n_i(r_1,r_2)\) is independent of \(r_{i}\) for \(i=1,2\), then in the subgame LN, late cooperation succeeds iff \(\Pi ^l_i(r_1^l,r_2^l) \ge \Pi ^n_i(r_1^n,r_2^n)\) for \(i=1,2\).
Proof of Lemma 6
If \(\Delta \Pi ^{ln}_i(r_1,r_2)\) does not depend on \(r_{i}\) for \(i=1,2\), then country \(i\)’s decision whether to accept or reject the formation of a late coalition (\(l_i\)) is independent of \(r_{i}\). Hence, if (for any given \(r_{i}\)) country \(i\) prefers the cooperative outcome given \(r_i=r_i^l\) over the noncooperative outcome obtained under \(r_i^n\), and the same holds true for the other country, then late cooperation succeeds, and the choices \((r_1^l,r_2^l)\) are optimal given this expectation. Conversely, if there is at least one country \(i\) that (for any given \(r_{i}\)) prefers the outcome under no cooperation given \(r_i^n\) to the outcome obtained under late cooperation given \(r_i^l\), then this country chooses \(r_i=r_i^n\), planning to reject late cooperation, irrespective of the value of \(r_{i}\) chosen by the other country (which is fixed when countries choose \(l_i\)). Hence, country \(i\) cannot affect country \(i\)’s cooperation decision in the LNsubgame via its own R&D choice, and in anticipation of the failure of cooperation, countries choose \((r_1^n,r_2^n)\) accordingly. A necessary and sufficient condition for late cooperation to succeed is, thus, that \(\Pi ^l_i(r_1^l,r_2^l) \ge \Pi ^n_i(r_1^n,r_2^n)\) holds for \(i=1\) and \(i=2\). \(\square \)
Intuitively, if \(r_i\) affects country \(i\)’s payoff under late cooperation, \(\Pi _{i}^l(r_1,r_2)\), in exactly the same way as it affects \(\Pi _{i}^n(r_1,r_2)\), then a change in \(r_i\) has no effect upon country \(i\)’s decision whether to enter a late coalition or not. In that case, it suffices to compare the payoffs under the no cooperation case (Sect. 3.2) with the payoffs in the late cooperation case (Sect. 3.4) to derive the equilibrium in the subgame LN.
Proof of Proposition 6
By (14), \(r_{i}\) enters linearly in country \(i\)’s payoff. Therefore, both under late and under no cooperation, country \(i\)’s abatement following the R&D stage, \(a_i(r_1,r_2)\), is independent of \(r_{i}\). It follows immediately that \(\Delta \Pi _i^{ln}(r_1,r_2)\) is also independent of \(r_{i}\), so Lemma 6 applies. Hence, to determine the endogenous success and the timing of cooperation in the overall game, it suffices to compare the equilibrium outcomes of the early, late, and no cooperation cases. Now we apply Lemma 4.
If \(\beta ^r \in (\sqrt{2}/2,\sqrt{2})\) and \(\varepsilon ^r \in (1/2,2)\) then it follows from the Lemma that for either country \(i\in \{1,2\}\) we have the ordering \(\Pi _i^e \ge \Pi _i^l \ge \Pi _i^n\). This leads to the unique equilibrium outcome of early cooperation.
If \(\beta ^r \in (\sqrt{2}/2,\sqrt{2})\) and \(\varepsilon ^r \not \in (1/2,2)\) then for either country \(i\in \{1,2\}\) it holds \(\Pi _i^l>\Pi _i^n\), while there exists one country \(j\) for which \(\Pi ^e_j<\Pi ^l_j\). This leads to the late cooperation outcome.
If \(\beta ^r >\sqrt{2}\) and \(\varepsilon ^r >2\) then \(\Pi ^e_1<\Pi ^l_1\) and \(\Pi ^l_1<\Pi ^n_1\), and country 1’s most preferred outcome is \(\Pi ^n_1\). Because country \(1\) can obtain this outcome by rejecting both early and late cooperation, there is no cooperation in any subgame perfect equilibrium. By symmetry the same holds for \(\varepsilon ^r <1/2\) and \(\beta ^r <1/\sqrt{2}\).
If \(\beta ^r >\sqrt{2}\) and \(\varepsilon ^r \in (1/2,2)\) then \(\Pi ^e_2 > \Pi ^l_2>\Pi ^n_2\), while \(\Pi ^e_1> \Pi ^l_1\) and \(\Pi ^l_1<\Pi ^n_1\). Hence, country 2’s preferred outcome is early cooperation. In the case that \(\Pi ^e_1>\Pi ^n_1\), also county 1’s preferred outcome is early cooperation and early cooperation arises in equilibrium. If \(\Pi ^e_1<\Pi ^n_1\), county 1’s preferred outcome is no cooperation, which it can always reach by rejecting any form of cooperation. As a result, the unique equilibrium outcome is no cooperation. In the Proof of Lemma 4, it is shown that \(\Pi ^e_i\Pi ^n_i =\beta ^2_i(2\beta ^2_{i}/\beta _i^21)/4+\varepsilon _i(2\varepsilon _{i}/\varepsilon _i1)/12\) so that the difference is decreasing in \(\beta ^r\). As a result the former case arises for \(\beta ^r>\sqrt{2}\) but sufficiently small (close to \(\sqrt{2}\)), and the latter one for \(\beta ^r\) sufficiently large. By symmetry, the same holds for \(\beta ^r <1/\sqrt{2}\) and \(\varepsilon ^r \in (1/2,2)\). Finally, if \(\varepsilon ^r <1/2\) and \(\beta ^r >\sqrt{2}\) then it follows from Lemma 4 that \(\Pi ^e_1> \Pi ^l_1\), \(\Pi ^l_1<\Pi ^n_1\), \(\Pi ^l_2> \Pi ^e_2\), and \(\Pi ^l_2>\Pi ^n_2\). Again, late cooperation is never obtained as country 1 prefers the noncooperative over the late cooperation outcome. Hence, either early or no cooperation is obtained in equilibrium. \(\square \)
Proof of Lemma 5
Straightforward calculations yields the following results:

1.
No cooperation:
$$\begin{aligned} a_1^n&= \frac{(b+\delta )(3b^2\delta ^2)}{8\gamma }, \quad a_2^n=\frac{(b\delta )(3b^2\delta ^2)}{8\gamma }, \quad r_1^n=\frac{3b^2+2b\delta \delta ^2}{8\gamma }, \quad \nonumber \\ r_2^n&= \frac{3b^22b\delta \delta ^2}{8\gamma };\nonumber \\ \Pi ^n_1&= \frac{27b^4+12b^3\delta 22b^2\delta ^24b\delta ^3+3\delta ^4}{64\gamma }, \quad \nonumber \\ \Pi ^n_2&= \frac{27b^412b^3\delta 22b^2\delta ^2+4b\delta ^3+3\delta ^4}{64\gamma };\nonumber \\ \Pi ^n&= \frac{27b^422b^2\delta ^2+3\delta ^4}{32\gamma }. \end{aligned}$$(19) 
2.
Early cooperation:^{Footnote 34}
$$\begin{aligned}&a^e_i=\frac{216b^3}{125\gamma } =1.728b^3/\gamma , \quad r_i^e=\frac{18b^2}{25\gamma } =0.72b^2/\gamma ;\nonumber \\&\Pi ^e_1=\frac{108 b^3(b+4\delta )}{125\gamma }=108\Pi ^f_1/125, \quad \Pi ^e_2=\frac{108 b^3(b4\delta )}{125\gamma }=108\Pi ^f_2/125;\nonumber \\&\Pi ^e=\frac{216b^4}{125\gamma }=108\Pi ^f/125. \end{aligned}$$(20) 
3.
Late cooperation^{Footnote 35}:
$$\begin{aligned}&a_i^l=b^3/\gamma , \quad r_1^l=\frac{b^2+2b\delta }{2\gamma }, \quad r_2^l=\frac{b^22b\delta }{2\gamma };\nonumber \\&\Pi ^l_1=b^2\frac{3b^2 + 4b\delta  4\delta ^2}{4\gamma }, \quad \Pi ^l_2=b^2\frac{3b^2  4b\delta  4\delta ^2}{4\gamma };\nonumber \\&\Pi ^l=b^2\frac{3b^24\delta ^2}{2\gamma }. \end{aligned}$$(21)
A direct comparison yields \(\Pi ^e \ge \Pi ^l \ge \Pi ^n\).
Given these results, it is straightforward to verify that in the relevant range, country 2 prefers early over late cooperation (\(\Pi ^e_2 \ge \Pi ^l_2\)) iff \(\delta \le \delta _1(b) \equiv (3072 \sqrt{21781})b/250 \approx 0.047b\).
In the subgame LN, for given R&D levels \((r_1,r_2)\), late cooperation succeeds if both countries achieve a higher welfare than in the absence of cooperation, hence, if \(\Delta \Pi ^{ln}_1(r_1,r_2) \ge 0\) and \(\Delta \Pi ^{ln}_2(r_1,r_2) \ge 0\). Under no cooperation (for fixed R&D levels \((r_1,r_2)\)), country \(i\)’s abatement is determined by condition (4): \(\partial C_i / \partial a_i=b_i\). Using \(C_i=\frac{a^2_i}{r}+\gamma r^2_i\), \(b_1=b+\delta \), and \(b_2=b\delta \), this yields
and, hence: \(a=br\). Countries’ payoffs, thus, become:
Under late cooperation (again for fixed R&D levels), country \(i\)’s abatement is determined by condition (10): \(\partial C_i / \partial a_i=b_i+\beta _i\). Using \(\beta _i=b_{i}\), we, thus, obtain \(a_1^l=a_2^l=br\), and \(a=2br\). This yields the following expressions for welfare:
This implies
Hence, for any fixed (\(r_1,r_2\)), country 1 is always willing to cooperate. Country 2 cooperates if \(b^26b\delta +\delta ^2 \ge 0\), which, for the relevant interval \(\delta \in [0,b/2]\), yields the critical value \(\delta _2(b) \equiv b(32\sqrt{2}) \approx 0.17b\). If \(\delta \) is below this value, it holds that \(\Delta \Pi ^{ln}_2(r_1,r_2)>0\). As \(\delta _2(b)\) is independent of \(r_1\) and \(r_2\), late cooperation succeeds in the subgame LN iff \(\delta \le \delta _2(b)\). \(\square \)
Proof of Proposition 7
It is straightforward to show [using (20) and (21)] that country 1 always prefers the early over the late cooperation outcome. The result, thus, follows immediately from Lemma 5. \(\square \)
Rights and permissions
About this article
Cite this article
Schmidt, R.C., Strausz, R. On the Timing of Climate Agreements. Environ Resource Econ 62, 521–547 (2015). https://doi.org/10.1007/s1064001498282
Accepted:
Published:
Issue Date:
Keywords
 Climate treaty
 Abatement
 Longterm cooperation
 Spillover
 Strategic delay
JEL Classification
 D62
 F53
 H23
 Q55