Join the Club: How the Domestic NGO Sector Induces Participation in the Covenant of Mayors Program

ABSTRACT How does strength of domestic NGOs influence participation in the Covenants of Mayors program? Launched by the European Commission in 2008, this program invites local and regional authorities to voluntarily commit to implementing EU climate change and energy policies. We focus on the transitional countries of Eastern Europe and Eurasia to examine whether the strength of their domestic NGOs correlates with cities’ decisions to participate in this transnational program. To operationalize NGO strength, we suggest thinking of it as a stock variable that cumulates over time, instead of a single-year, flow variable. With country year as the unit of analysis, we examine the percentage of urban population covered by the Covenant across a panel of 26 transitional economies for the period 2008–2014. We find that the key variable of interest, cumulative NGO strength, is a statistically significant predictor of program participation, even after controlling for domestic and international factors, including the salience of international NGOs and the years since the country began the formal process to join the European Union.

The COM program was launched by the European Commission in 2008. It seeks to involve "local and regional authorities, voluntarily committing to increasing energy efficiency and use of renewable energy sources on their territories" (Covenant of Mayors n.d.a.). The objective is to reduce carbon dioxide (CO 2 ) emissions by 20% (with 1992 as the benchmark) by 2020. For the period 2008-2014, the participation levels in the COM program are impressive: 6,275 jurisdictions covering over 190 million inhabitants have joined. However, the participation levels are uneven across the studied 26 countries, ranging from a low of 1% of urban population in Tajikistan or Azerbaijan to a high of 100% in Bosnia and Hercegovina in 2014 (Covenant of Mayors n.d.b.). How do variations in participation levels correlate with the strength of domestic civil society?
This special issue is rooted in the premise that domestic politics plays a crucial role in the success, especially the adoption, of transnational initiatives: Prevailing explanations largely focus on global processes of diffusion through economic and social networks, as well as the micro-incentives facing individual sub-and non-state actors. While such theories offer important insights, they assume, often implicitly, that participation can be explained as if it occurred outside the domestic contexts in which actors operate. But this assumption stands at odds both with well-established research on the role of domestic interest groups, policies and institutions in shaping international cooperation, and recent case studies that demonstrate how domestic factors condition transnational forces. (Roger, Hale, and Andonova 2017) We build on this core idea to show that even in the context of transitional economies of Eastern Europe and Eurasia with communist legacies, civil society can play an important role in shaping the city-level response to a transnational climate governance initiative.
Why participate in a voluntary program? Scholars studying cities' participation in voluntary climate change initiatives note that cities' participation may not seem rational because their voluntary actions create a nonexcludable global public good, while their own costs might significantly exceed their private benefits (Zahran, Brody, Vedlitz, Grover, and Miller 2008). We examine this puzzle through the lens of club theory (Prakash and Potoski 2006). The costs of joining and implementing COM are not trivial. Participating cities are required to create a baseline inventory of their emissions and to implement an energy plan to meet the emissions reduction targets. To accomplish these goals, they have to establish an administrative infrastructure, including a monitoring process. Finally, the participating cities have to submit their plan to the European Union, along with regular implementation reports.
To examine the role of domestic politics in this transnational governance initiative, we focus on domestic NGOs as drivers of COM adoption. Few would disagree that NGOs are important actors in domestic and global politics. Scholars have invested considerable effort in studying the role of transnational NGOs networks in global politics (Keck and Sikkink 1998). They have paid less attention to the role of domestic NGOs in transnational governance. Why might this be so? For one, domestic NGOs tend to be weak in developing countries. To get their voices heard, these NGOs need support from transnational NGOs via the so-called "boomerang effect" (Keck and Sikkink 1998). Hence, it seems that the real "action" is at the global level, not at the domestic level. This is particularly true in the context of environmental issues-hence the focus on transnational instead of domestic actors.
The emphasis on transnational NGOs and the neglect of domestic NGOs is puzzling because the initial justification for focusing on NGOs as policy actors was their embeddedness in domestic politics, their grassroots character, and their ability to generate social capital that could be deployed to political and economic issues (Putnam 1995). Hence, NGOs represented politics from below and were the true agents for aggregating and then transmitting citizens' preferences on key policy issues. While international NGOs certainly have an important role to play in global and domestic politics, a neglect of domestic NGOs undermines our understanding of how NGOs influence transnational governance and domestic politics.
Theoretically, the transitional economies of Eastern and Central Europe and Eurasia present a "hard case" to test the influence of domestic NGOs, which have been nascent in these countries due to their communist legacies. While international NGOs have been active in several of these countries, scholars debate the extent to which domestic NGOs have been able to exercise policy influence. How might then we theoretically think about the strength and policy influence of domestic NGOs in transitional economies? We suggest that the study of the role of civil society in transitional economies should focus on the civil society's cumulative strength, instead of the measures of their annual strength. This is because building trust in NGOs and recognizing their legitimacy as political actors takes place gradually. Furthermore, the social capital that NGOs' activities accumulate and the capital that is then deployed in economic and political sphere is also not a one-shot process. To study the leverage NGOs exercise on governmental policies, scholars need to focus on the strength they have accumulated over a longer time period. To operationalize this key and novel way of thinking about NGO strength, we examine NGO strength as a stock variable that has cumulated over time, instead of an annual, flow variable. We believe this new way of thinking about NGO power in domestic politics should be extended to other issue areas and other developing countries.
Furthermore, transitional economies also show considerable variation in their wealth levels and in the health of their economies. Hence, it is difficult to claim that postmaterialist values are likely to uniformly encourage participation in the COM program. In addition, most transitional economies have joined the European Union at different points in time (and some might yet join). Because the EU exercises significant influence on the domestic policies of its members and potential members, the levels of socialization and policy harmonization induced by the European Union are likely to vary. This allows some additional room for domestic actors to exercise policy leverage.
Efforts to mitigate global climate change have witnessed remarkable participation levels by subnational actors (Lee 2014;Rabe 2004). The emphasis on local governments, specifically cities, is crucial because "80% of energy consumption and CO 2 emissions is associated with urban activity" (European Commission 2015). In the United States, the absence of federal regulation on climate change has created policy space for subnational actors to step in (Dolšak and Houston 2014;Hsueh and Prakash 2012;Zahran et al. 2008). These scholars find state and local governments to be responding to demand-side factors (such as local air pollution, climate-change risks, and public opinion and concern) and the supply/capacity-side factors (such as socioeconomic-capacity dimensions and institutional characteristics). Even beyond climate change, the role of cities in providing the venues for political struggle is recognized to be important across issue areas (Sassen 2001), the Arab spring being a prominent example. At the peril of exaggeration, one might even suggest that there is an urban bias (Bradshaw 1987;Lipton 1977) in politics, with urban actors playing an important role in shaping policy processes. Superior access to media, both traditional and social, gives urban actors policy visibility and influence. The COM program offers an excellent opportunity to study the extent to which domestic environmental NGOs (which tend to be urban-based in transitional economies) have been able to influence city-level participation in a voluntary environmental program.
Empirically, with a country year as our unit of analysis, we examine the percentage of urban populations covered by COM across a panel of 26 transitional economies in Eastern and Central Europe and Eurasia over the 2008-2014 period. Our key variable of interest is the strength of domestic NGOs. Our model controls for a range of variables, domestic and international, that might influence cities' participation in COM. These include: the number of international NGOs in the country (which allows us to differentiate between the policy influence of domestic and international NGOs), economic growth, nitrogen oxides (NOx) pollution levels, fiscal decentralization, and exports as share of GDP. In addition, we control for the number of years since a country signed an EU accession agreement. To check the robustness of our results, we use several measures of NGO strength (employing the USAID NGO Sustainability Index, as we discuss subsequently) as our key independent variable. Our results with regard to the significant effect of the strength of domestic NGOs as a predictor of the COM participation levels are robust across all these specifications. Finally, we extend our analysis to explore whether NGO strength might also influence some measures of program efficacy. Focusing on procedural efficacy, we find that NGO strength correlates with participating cities formally submitting their Sustainable Energy Action Plans to the European Commission.

The Covenant of Mayors program
Nobel Laureate Elinor Ostrom, noted in her presidential address at the 1997 annual meeting of the American Political Science Association that "the theory of collective action is the central subject of political science" (Ostrom 1998:1; emphasis in the original). Governance pertains to the organization of sustained collective action by creating rules. Challenges to collective action are particularly salient in the context of transnational endeavors. Furthermore, these challenges pose interesting theoretical issues when collective action is sought to be organized at the level of subnational actors and in the realm of voluntary or nonmandatory actions.
This article examines a unique case of transnational collective action aimed at subnational actors, the Covenant of Mayors (COM) program. COM should be understood in the context of the 1997 Kyoto Protocol that established targets for reductions in greenhouse gas (GHG) emissions. There are 192 parties to this regime: 191 states and the European Union. The European Union sought to implement its Kyoto commitment in a phased manner. The EU-15 committed to reducing their collective emissions to 8% below 1990 levels during phase 2, 2008-2012. The member States that joined the EU since 2004 also committed to reduction targets in the range of 6%-8% (5% in Croatia's case). Overall, the European Union committed that by 2020 it will cut its GHG emissions by 20% below 1990 levels. To do so, the European Union had used both mandatory and voluntary tools. It established the European Union Emissions Trading Scheme (EU-ETS), a "cap and trade" scheme that established mandatory national emission quotas for power plants and some industrial facilities and allowed them to buy and sell emission allowances. The European Union began implementing the EU-ETS in 2005. Alongside this, it launched the COM program in 2008.
Governmental agencies as well as intergovernmental organizations such as the United Nations have been in the forefront of establishing voluntary programs. The United States Environmental Protection Agency has been a pioneer in launching a range of programs, over 60 by some counts, including some programs focusing on climate change mitigation. 2 The UN launched the United Nations Global Compact in 2001. Thus, the European Commission's launch of COM should be viewed as part of the broader 2 http://www3.epa.gov/climatechange/EPAactivities/voluntaryprograms.html trend, whereby governmental and quasi-governmental actors have deployed both mandatory and voluntary tools toward specific policy objectives.
Although it is a transnational governance tool, the COM supports the efforts of subnational actors on climate change mitigation, especially via adoption of sustainable energy policies. COM is a voluntary, or a nonmandatory, program. It can be conceptualized as a "green club" (Prakash and Potoski 2006) that induces voluntary contribution to public good provision (global climate change mitigation) by promising excludable reputational and goodwill benefits that have the characteristics of club goods. The promise of appropriating these benefits constitutes the incentive for actors to voluntarily incur the costs of participating in the program.
In terms of branding benefits, the European Union has sought to portray COM as an important communication tool with which cities communicate about their efforts to be pro-environment, and more specifically, proactive on the issue of climate change mitigation. The COM Web site notes that "The European Commission has committed to supporting local authorities involved in the Covenant of Mayors and providing public visibility for them" (Covenant of Mayors n.d.c.). Further, the Web site lists several reasons for joining COM. It notes: Covenant Signatories find multiple reasons to join the movement, such as: Make a public statement of extra commitment to CO 2 reduction; Create or reinforce the dynamic on CO 2 reduction in your territory; Benefit from encouragement and examples of other pioneers; Share the expertise developed in your own territory with others; Make your territory known as a pioneer; Publicise your achievements visible to the Covenant community and beyond via the Covenant website. (Covenant of Mayors n.d.c.) Thus, consistent with the literature on voluntary clubs (van't Veld and Kotchen 2011), COM offers important branding benefits, which hold the promise of reputational payoffs to participants.
Reputational benefits often have significant material payoffs. Firms routinely invest in building their brand reputations, which allows them to charge premiums for their products, strengthen the loyalty of their consumers, and reduce what is called "postpurchase dissonance," especially for premium products. There is a well-established literature on how membership in voluntary environmental programs might affect market capitalization of firms. At the country level, reputations are important as well. Countries that have reputations for corruption tend to attract less foreign direct investment, for example. Firms operating in corrupt countries may fear that their products are tarnished by the country's reputation and may join voluntary programs to offset this reputational damage. Consequently, there is some evidence that joining voluntary programs such as ISO 9001 may allow developing countries that have poor reputations for quality to enhance trade (Potoski and Prakash, 2009). Hence, we do not see reputational benefits of joining a program to be mainly ideational; they have concrete material payoffs.
Which actors will bestow these benefits? We suggest that cities might be more willing to recognize such benefits if there are actors who have the ability and incentives to lobby and persuade them in this regard or create a structural context in which such voluntary initiatives are appreciated and recognized. Participation in COM may also strengthen environmental credentials of city politicians with these vocal constituencies. As we discuss in the next section, scholars note the role of NGOs in creating such reputational benefits for environmental issues.
Why study participation in voluntary programs, especially if there is little evidence that they reduce pollution? This subject has been extensively examined in the voluntary program literature because of a lack of apparent motivation for participation. Actors incur nontrivial costs to join these programs, while the benefits from such participation are not always clear, concentrated on the actors incurring the costs, nor excludable. Even if programs are eventually deemed to be ineffective, it is still important to understand the conditions under which actors are willing to join them and why public policy seeks to promote them. Viewed this way, our article also contributes to the literature on voluntary programs that focuses specifically on the subject of participation in such endeavors.
Voluntary programs typically tend to outline two types of commitments: a focus on internal processes/management systems or outcomes. For example, a program could obligate its participants to adopt specific types of policies, without stipulating the targets these participants need to achieve. The rationale is that if appropriate internal systems are in place, the desired outcomes will follow. Prominent examples include the ISO 14001 program sponsored by the International Organization for Standardization, or the Responsible Care program that is sponsored by trade associations representing the Chemical industry. These programs obligate the participating members to adopt management systems with specific features. However, they do not stipulate specific environmental outcomes program members have to achieve (Prakash and Potoski 2011).
Other voluntary programs stipulate specific environmental outcomes that the participants need to meet as part of their membership obligations. COM imposes both process and outcome obligations. In terms of process requirements, it requires its participants to create a baseline emission inventory (BEI) and implement Sustainable Energy Action Plans (SEAP). BEI requires that the participating city document CO 2 emissions caused by fossil fuel consumption of its large emitters. Sustainable Energy Action Plans outline specific actions cities are committing to implement. This is an important first step (which we have termed as procedural effectiveness subsequently) because entities-be it governmental or nongovernmentalare often oblivious of the pollution they emit (Prakash 2000). The creation of baseline inventories forces them to recognize their carbon footprint, which can create internal pressures for policy action. Cities are expected to update these inventories on a regular basis, which allows them to monitor the progress of specific entities in their jurisdictions in reducing CO 2 emissions.
Most voluntary programs recognize the possibility that some participating actors may not fulfill their obligations-the so-called shirking problem (Prakash and Potoski 2006). How might shirking be curbed? Scholars note that the shirking problem is typically addressed in three ways. Scholars subscribing to club theory see shirking mainly as a problem of institutional design. For them, shirking can be curbed if the program incorporates mechanisms to monitor and sanction. While third-party auditors are the gold standard for monitoring, the actor sponsoring the program does not itself sanction the violators, nor does it make information provided by the auditors available to the public for naming and shaming these firms, making monitoring less effective. Some programs ask participants to report on their achievements and then make this information available to the public. This approach is followed by COM: Cities have to submit implementation plans and performance information to the program secretariat, which makes them publicly available. Then interested stakeholders, particularly the NGOs, can use this information to reward and punish participating actors.
Another perspective suggests that shirking can be curbed through peer and community pressure; after all, obeying rules is the norm of appropriate behavior (March and Olsen 1989). Hence, the community of cities participating in COM will exert normative pressure and socialize each other toward honoring program obligations. The third perspective suggests that actors shirk because they do not have the capabilities to meet their obligations. This might hold for small cities with budgetary pressures or insufficient expertise to draw up climate mitigation plans. To address these challenges, national governments as well as the European Union provide financial and technical support.
Joining green clubs typically imposes nontrivial costs (although the literature notes the presence of "greenwashes" that do not impose any serious obligations on their participants, as noted by Laufer [2003]). The creation of BEI and SEAP requires substantial investment in administrative infrastructure. Given that the launch of COM coincided with the European, if not global, economic crisis and budgetary deficits, we expect city governments to think carefully about the political and economic benefits and costs of participating in the COM club. The role of domestic NGOs is crucial in this regard because they can arguably create political and economic benefits for participation. Indeed, voluntary program literature has noted the important role NGOs can play in persuading actors to join specific clubs (Baron 2009;Lenox and Eesley 2009). To what extent NGOs can be successful in their persuasive efforts is an empirical issue we subsequently examine.
Finally, there is a question about the factors governments consider in making policy choices. Scholars debate whether governments think carefully of costs and benefits of policies prior to making policy choices, an issue that is rooted in the broader debate in the literature on rational choice. This article suggests that politicians think of, among other factors, benefits and costs of joining COM prior to joining it. We are not asserting that they think only in terms of benefits and costs; indeed, as we discuss subsequently, our empirical model takes into account the contributions of "nonrational" mechanisms as well. For example, we control for years since the country signed the European Union accession documents. This variable captures the normative and administrative links with the European Union, which can create a common normative community committed to climate change mitigation. In other words, the EU accession process does not directly alter the benefit and cost calculation a mayor of a specific city faces.

NGOs in transitional economies
Scholars debate the extent to which NGOs exercised policy influence in the communist era (Rose, Mishler, and Haerpfer 1997). In the postcommunist era, the influence of domestic NGOs on environmental politics varied across transitional economies (Badescu and Sum 2004;Mondak and Gearing 1998). Consider the following: The Czech Republic faced historically high emissions of SO 2 and NOx even though Czechoslovakia was a signatory of the longrange transboundary air pollution convention agreements. However, it enacted the Clean Air Act in 1991 with NGOs playing an important role in the policy process (Andonova 2005). Poland also faced high domestic air pollution. Its environmental NGOs, however, could not persuade the government to enact something akin to the Clean Air Act (Dolšak 2013). Thus, the ability of domestic NGOs to influence policymaking varies across transitional countries of Eastern Europe and Eurasia. As the U.S. Agency for International Development (2014:1) notes: While Poland has more than 100,000 registered associations and foundations, the [civil society organizations] sector in Turkmenistan consists of a mere 106 registered organizations. While it is nearly impossible for the government in Latvia to do anything without meaningfully engaging CSOs [Civil Society Organizations], the situation is quite different in Azerbaijan where CSOs cannot organize any event or activity without the knowledge and permission of the executive powers.
Some countries have adopted laws that restrict the ability of NGOs to secure overseas funding, Russia being a prominent case. Even when NGOs are able to freely obtain international funding, they are not always equipped to effectively deploy it to influence public policies. NGOs in these countries frequently lack the capacity to manage resources with the levels of accountability and transparency required by donors. This leads to credibility problems with domestic audiences and international donors. Furthermore, when the domestic economic situation deteriorates and when the culture of philanthropy is not encouraged, the financial viability of NGOs faces considerable challenge.
NGOs' ability to advocate for policy change is also influenced by their low ability to formulate message in the media, lack of knowledge required for coalition formation, and weakness for demanding policy change and monitoring of enacted policies. When NGOs have a poor public image, they face challenges in recruiting volunteers, encouraging domestic donations, and persuading government to work with them.
Given that NGOs are expected to represent public interest, citizen support and trust becomes an important component of their ability to influence the policy process. Of course, developing trust takes time, particularly because most NGOs served under the umbrella of the communist parties in the pre-1990s. The transition from party-controlled organizations to truly grassroots organization was also challenging because many NGOs, aided with the infusion of Western funds, tended to acquire the semblance of professional organizations that were less interested and capable of grassroots action. Yet, although unevenly so, some NGOs have successfully begun to acquire public trust while maintaining the professional image that is sometimes quite appealing to Western donors. In sum, the acquisition of public trust and policy influence is gradual and cumulative. Both the shadow of the future and experiences of the past drive levels of trustworthiness.
Based on this discussion, we suggest that NGO strength in policy process should be conceptualized as a stock variable instead of a flow variable. We are inspired by Gerring, Bond, Barndt, and Moreno (2005), who suggested that the influence of democracy on development should be studied by thinking of democracy as a stock variable. This is because democracy as a method of governance develops over time, building on the resources accumulated in the past. We explore how this idea applies in the context of NGOs; after all, civil society is a pillar of democracy.
There are important theoretical rationales for our approach. One might argue that the stock of NGOs approximates the concept of social capital, which is an institutional feature that shapes the ability of actors to demand policies and accountability from public authorities and to organize for collective action (Ostrom 1990;Putnam, 1995). Social capital is a stock variable because it is accumulated by repeated interactions over a period of time. While Gerring et al. (2005) construct the democracy stock variable by summing up democracy scores over several decades, following Dolšak (2013) we adopt the same theoretical approach for NGOs but for a much shorter duration (since 1997 when the USAID data became available). We find that this cumulative effect of NGO strength can be observed even after a relatively short period of time. This suggests that there might be "shortcuts" in creating valuable political and social resources such as NGOs and civil society.

Variable operationalization
Our dependent variable is the share of the urban population covered by COM. The COM database lists the cities that have joined the program along with their population. Why focus on population share and not on the number of cities that have participated? With population varying dramatically across cities (from a few hundred to millions), counting the number of cities in each country does not capture the political or policy dynamics. Furthermore, NGOs can be expected to lobby several cities to join this program. While several NGOs tend to be located in the capital city, they are able to influence cities in other parts of the country. Viewed this way, unless one has city-level data on NGOs (which are not available, and this is also problematic because NGOs lobby in several cities) and can map how NGOs target specific cities but not others, it will be theoretically difficult to defend an individual city as the unit of analysis. At a more practical level, while the data on urban population are provided by World Development Indicators, we do not have data on the total number of cities in a given country. Consequently, while we can compute the share of urban population covered by COM, we cannot compute the percentage of cities joining this program. Thus, on both conceptual and practical grounds, our dependent variable is the cumulative share of urban population covered by COM.

Key independent variable
We measure the strength of domestic NGOs using a novel index of NGO sustainability developed by the USAID (2014). While this measure reflects the strength of domestic civil society as a whole, we believe it is an appropriate measure to assess the strength exercised by environmental NGOs as well. Furthermore, it is our sense that domestic environmental NGO strength will strongly correlate with overall strength of civil society given that environmental issues are among the key public concerns in this region of the world. In addition, environmental NGOs were among the first to evolve in these countries in the 1980s and 1990s because they were not viewed as articulating a political challenge to the regime.
The index of NGO sustainability reflects several dimensions of NGO strength that should impact their abilities to influence the policy process. It examines the institutional environment-specifically, the ways in which the legal environment allows them to secure resources. Given that NGOs need to function as organizations, it looks at their internal capacities and skills. It also takes into account their financial viability along with local capacities to support NGOs via philanthropy or fund raising. Because NGOs need to work together to influence the policy process, it also incorporates their abilities to form coalitions. Finally, because media coverage is the oxygen for NGO survival, it looks at how the media cover activities of NGOs. The USAID undertakes expert surveys for each country to assess these dimensions. It seeks opinions from local NGOs, academics, think tanks, government officials, media, and business, as well as international donors (USAID 2014:241). We have reversed the original indicators provided by the USAID so that value 1 indicates the lowest level of NGO sustainability and value 7 indicates the highest. While the majority of our models (1-3 and 8-10) use the combined measure, we report specification checks for subcategories of the measure as well (Models 4-7). The data were downloaded from the USAID 2013 report on the NGO Sustainability Index for Central and Eastern Europe and Eurasia.

International controls
Scholars have noted the impact of the EU accession process on domestic policies in EU candidate countries (Gray 2009). We control for European Union influence on these countries. This process begins with the signing of the EU accession agreement. With this, countries begin to harmonize their environmental regulations, standards, and policies. The length of the accession period is important for two reasons. First, it allows the EU's normative influence longer time to infuse the policy processes in the candidate countries. At the same time, this period also allows these countries to build up local capacities. We measure this in terms of years since the signature of the Europe Agreement (for countries of Central and Eastern Europe) and Stabilization and Association Agreement (for countries of the Western Balkans).
An important factor influencing COM adoption is the engagement of domestic actors with the international economy via trade networks. There is a continuing debate on how engagement with the global economy influences domestic environmental policies. Some scholars find that trade engagement leads to stronger environmental policies (Cao and Prakash 2010), while others find the opposite effect (Andonova, Mansfield, and Milner 2007). Therefore, we control for export salience (exports/GDP) by using data from the World Development Indicators database.
While we focus on the role of the domestic NGOs in the participation in COM, we recognize that international NGOs continue to play an important role in the environmental politics of this region. Hence, we control for the count of International NGOs (INGOs) working in these countries by using data from the Yearbook of International Organizations.

Domestic controls
While the EU-ETS requires member countries to contribute to the provision of a global public good, countries might be motivated to voluntarily undertake climate change mitigation policies if they can reduce domestic air pollution (Dolšak 2009). The ability to corner domestic benefits allows political actors to justify the costs of voluntarily adopting mitigation policies. The political benefits of reducing local pollution might extend well beyond the environmental arena. Indeed, politicians can gather support from NGOs working in areas such as public health as well (Dimas 2008). Our model therefore controls for NOx, which is an important source of local air pollution. NOx emissions are measured in kilograms per capita. Data were accessed at the EU Centre for Emission Inventories and Projections.
The prioritization of environmental concerns on policy agendas may be influenced by business cycles. During periods of lackluster economic growth, environmental concerns might take a back seat (Fagan 1994;Slocock 1996). To control for the impact of the economic cycles on cities' incentives to join COM, we control for the annual GDP growth rate. The data are from the World Bank's World Development Indicators database.
COM is aimed at local and regional governments. Arguably, these governments may have the inclinations and abilities to join such programs only if they have the resources and authority to do so. This, in part, may be influenced by the distribution of power between the national and subnational levels of government. The literature highlights several dimensions of decentralization, including political, administrative, and fiscal (Thiessen 2003). COM might require local governments to make changes in the taxation as well as expenditure policies. We employ a measure of fiscal decentralization because local and regional governments' responses to COM are likely to be influenced by their abilities to impose taxes and decide about expenditures. Our model controls for fiscal decentralization (Iimi 2005;Schneider 2003;Thiessen 2003) operationalized as the percent of total government expenditures attributable to local governments (Dziobek, Alves, El Rayess, Gutiérrez Mangas, and Kufta 2001). The data are from the International Monetary Fund's Government Finance Statistics Yearbook. Descriptive statistics and correlations between variables included in the model are available in the Web appendix.

Model, findings, and extensions
We investigate the theoretical claims using an unbalanced panel of 26 countries for the period 2008-2014. We lag our independent variables by one year to address the issue of reverse causality. To address the issue of heteroscedasticity, we use Hubert White robust standard errors clustered by country (Greene 2012). Our model is specified as: where NGO represents the cumulative score for NGO strength, X represents the vector of control variables, and є is the error term. Table 1 presents the results of our analyses. The first model includes the key independent variable of our interest, cumulative NGO strength. Models 2 and 3 control for domestic and international factors respectively. Our analysis suggests that NGO strength, the key variable of interest, is statistically significant even after controlling for the salience of international NGOs. In Model 1, an increase of cumulative NGO strength (which shows considerable variation, from about 27 in Belarus in the beginning of the studied period to over 99 in Estonia or Poland in the last year of the studied period) by 1 point is associated with a 1.2-percentage point increase in the share of urban population covered by COM (p < .001). That is, with an average annual increase of NGO strength of about 4.3 points on the NGO sustainability index, we should expect (on average) about a 5-percentage point increase in the share of urban population covered by COM.
Importantly, the effect of the NGO strength variable is significant, although we have controlled for INGO presence and for years since the country began the formal process to join the EU. Hence, our analysis provides support for Note. Regression for panel data set, random effects. Robust, cluster corrected standard errors reported in parentheses; *.05 < p ≤ 0.1; **.01 < p ≤ .05; ***p ≤ .01 (two-tail test).
the claim that domestic NGOs have influenced COM adoption, beyond what they might exercise via their links with INGOs and via the EU accession process. The EU accession process itself does not have a statistically significant effect on COM coverage. This is an important finding because prior research reports the substantial influence of the accession process on domestic policies of EU candidate countries (Gray 2009). Arguably, countries seek to satisfy the EU's requirements regarding formal laws and regulations at the national level only. This sort of EU environmental normative pressure might be less salient at the subnational level and in the context of a voluntary program. For the EU to exercise direct influence on cities in this regard, it should perhaps offer tangible incentives (especially monetary ones that fiscally strapped cities are likely to appreciate) beyond the reputational and informational goodies, to incur private costs of implementing climate change mitigation policies. We do not find support for the hypothesized impact of the increased fiscal decentralization. The reason may be in that when cities consider joining COM, they count on external financial support. As a matter of fact, cities have received financial assistance to comply with their COM requirements, something beyond the excludable benefits they were expected to gain. Slovenian cities, for example, received support from the Slovenian Ministry of the Environment and Spatial Planning and regional energy agencies for preparation of their sustainable energy action plans. Similarly, Croatian cities received support from German research agencies to prepare their sustainable energy action plans.
Among domestic controls, neither local air pollution (NOx) nor business cycles influence the coverage of urban population under COM. Importantly, neither the presence of international NGOs (INGO) nor the salience of exports influences COM coverage. Some consider INGOs to be important policy actors in transitional economies because domestic NGOs are weak. Arguably, the crackdown on foreign NGOs and on the operations of foreign NGOs in this region might partially explain our finding.

Robustness checks and extensions
We conducted several robustness checks. First, we examined subcomponents of the NGO Index that pertain to NGO advocacy, fiscal viability, public image, and legal environment ( Table 2, Models 4-7). As these models indicate, our results are robust to all operationalizations of NGO strength.
Second, we undertook additional checks. Some scholars suggest that FDI is an important carrier of environmental norms and wealth correlates with environmental preferences. Hence, we controlled for FDI/GDP, per capita GDP, and per capita GDP squared. While FDI/GDP is not significant and per capita GDP is, their exclusion does not influence our substantive results.

Furthermore, Ward and Cao (Forthcoming) in this International
Interactions' special issue suggest the important role of fossil fuel interest groups in opposing COM participation. Building on this insight, we included a variable measuring the importance of coal in a country's total primary energy supply. We drew on data from the International Energy Agency in this regard and found that even though this variable was significant, its exclusion did not change our key results.
Environmental aid is another factor that might influence incentives and capacities of local governments to join COM. We reestimated our model with environmental aid data provided by Tierney, Nielson, Hawkins, Roberts, Findley, Powers, Parks, Wilson, and Hicks (2011). These data are available for the first 3 years of the 6 years we study and for 17 out of 26 countries included in the analysis. We found that although environmental aid is significant, the inclusion of green aid does not influence the significance of the variable of interest-NGO strength.
NGOs and governments are more likely to support COM adoption when there is a clear public concern regarding environmental issues. Furthermore, not all environmental issues might solicit similar levels of concerns. Arguably, citizens might be more concerned about local environmental issues than global issues. In particular, citizens might want their governments to provide local environmental public goods as opposed to global public goods such as reductions in carbon emissions. We estimated the models by including various measures of environmental concern. We operationalized commitment to environmental protection as percent of respondents who agreed with the statement that protection of the environment should be given a priority even if it causes slowdown in economic growth. The data were accessed from the World Values Survey (n.d.). To control for concerns about global warming, we used two variables. The first one measures respondents' awareness of global climate change and the second respondents' view that climate change is caused by human activities. Comparable cross-national data on environmental concern and global climate change awareness are collected mostly at one point in time (Gallup poll or World Values Survey) for a subset of countries (Pelham 2009). As a result, we were only able to analyze their impact for a small subset of our sample (52 observations). In this smaller subset, results were consistent with our earlier models. Might levels of democracy influence assessment of NGO strength? To address this issue, we first included a measure of democracy (Polity IV, scale from -10 to +10, negative numbers indicate authoritarian regimes), and our results hold. Second, we estimated the model without countries with authoritarian regimes (with Polity IV scores below 0). The dropped countries were: Azerbaijan, Belarus, Kazakhstan, and Tajikistan. Again, our results hold. 3 Finally, we extended our analysis to examine the impact of the NGO strength on the "procedural effectiveness" of the COM. Scholars note the challenges in operationalizing the efficacy of voluntary programs (for that matter, intergovernmental regimes as well) because it is difficult to measure the effects in the short run, and it poses important issues of endogeneity between participation and effectiveness (Borck and Coglianese 2009;Lyon and Maxwell 2007;Prakash and Potoski 2011). One way to think of effectiveness is by focusing on procedural efficacy (as opposed to substantive efficacy that focuses on program outputs). Procedural efficacy can be assessed by examining whether the participating locality has undertaken steps to fulfill its program obligations. The COM was launched in 2008. Prior to 2015, there are no systematic data to assess how participation in COM has influenced city-level emissions of greenhouse gases (substantive efficacy). However, there is a possibility to explore how participating cities are fulfilling their obligations (procedural efficacy) under this program and to what extent this is driven by NGO strength, the key variable of our interest. Participating cities are obliged to formally submit their Sustainable Energy Action Plans.

3
Variables for the above specification checks are included in the replication data set.
While 473 cities in the studied countries have joined COM, only 327 have submitted plans during the studied period. As the Table 3 shows, NGO strength continues to have a statistically significant effect on this dimension of procedural efficacy.

Conclusions
The introductory article by Roger et al. (2017) notes the important role of transnational climate governance initiatives in the abatement of greenhouse gas emissions. Challenging the state-centric view, they suggest that the success of such initiatives will depend on mobilization of subnational and nonstate actors. Because there are variations in the extent to which actors are involved in such initiatives, Roger et al. pose the following question: "How do domestic politics and other country-level variables affect sub-and nonstate actors' participation in transnational climate governance?" This article responds to this question by focusing on a voluntary program (COM) sponsored by a supranational organization (European Union) and aimed at subnational units (cities). We examine the participation in this program across transitional economies of Central and Eastern Europe and Eurasia by focusing on the role of an important nonstate actor: domestic NGOs. We find that even in the case of these transitional countries with communist legacies where NGOs are of relative nascent origin and NGOs are considered to be less politically powerful, NGO strength correlates with the percentage of urban population covered under this voluntary program. While our model focuses on the cumulative strength of domestic NGOs and controls for the salience of international NGOs, we recognize that the latter often influence domestics politics through the instrumentalities of the former (Andonova and Tuta 2014). We believe that thinking of NGO strength as a stock variable has a considerable theoretical promise. While Gerring et al. (2005) examined the accumulated democracy stock over several decades, our article adopts the same theoretical approach for NGOs, but for a much shorter duration (since 1997 when the USAID data become available). We find that this cumulative effect of NGO strength can be observed even after a relatively short period of time. This suggests that there might be "shortcuts" in creating valuable political and social resources such as social capital and civil society.
We conceptualize COM as a voluntary club that offers reputational benefits to its participants as an inducement to incur the nontrivial costs of joining COM. Domestic NGOs play a crucial role in the creation of such reputational payoffs via direct advocacy as well as indirectly by structuring the institutional environment in which participation actors function. By their environmental advocacy and lobbying, they seek to persuade city governments to "join the club." The extent to which this club will contribute to climate change mitigation over and above what the cities would have done in any case remains to be seen -an issue future research should closely examine. However, the recruitment into this club, especially during the time of economic crisis, is an important first step. Furthermore, as we report in the article, NGO strength correlates with initial procedural steps such as the creation of Sustainable Energy Action Plans that should eventually lead to lower emissions.
Finally, what policies might be put in place to ensure that participation in voluntary clubs leads to pollution reductions? Arguably, participants may have incentives to garner the reputational benefits without investing substantial amounts in pollution reduction. What might persuade cities to look at the Covenant of Mayors program not merely as a branding exercise but as an opportunity to work toward reducing GHG emissions? The role of financial support from external actors might be crucial here. The subject of effectiveness of climate clubs has assumed added importance because the 2015 Paris Agreement has underscored the important role of voluntary emission reductions. Voluntary clubs such as the Covenant of Mayors can provide a useful template to establish new programs, especially in developing countries.