Ratification of multilateral environmental agreements: Civil society access to international institutions

ABSTRACT

The stage in which countries formally decide on whether to participate in (i.e., ratify) international agreements is crucial to global governance efforts. The reason is that, by and large, international agreements with greater participation are more likely to contribute to effective problem solving. We study the role procedural design characteristics of agreements play in such decisions. Specifically, we examine whether treaties’ provisions allowing non-state actors to participate in treaty making, which is widely regarded as an important procedural aspect of governance, increases the likelihood of ratification. Our empirical testing relies on a new time-series-cross-sectional dataset that includes information on the ratification behaviour of 154 countries with respect to 178 multilateral environmental agreements in 1950–2011. We find that treaty provisions allowing for greater non-state actor access to the meetings of the parties indeed increase the likelihood of treaty ratification. The result is robust to controlling for the effects of various other treaty design characteristics and country characteristics on ratification behaviour. The main policy implication is that, despite occasional debate over drawbacks of involvement of non-state actors, the latter tends to support global environmental governance efforts and should be further enhanced.

KEYWORDS:

• Multilateral environmental agreements
• treaty design characteristics
• civil society
• Conference of the Parties (CoPs)
• ratification

Acknowledgements

We thank Ronald B. Mitchell for providing us with data on treaty termination dates, Michelle Cohen, Eleanor Hughes, and Katherine Woolbright for coding, and Mike Hudecheck for data management. This article was written in the context of the National Center for Competence in Research (NCCR) ‘Democracy in the twenty-first century’.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Typically, governments sign an agreement at the end of its negotiation phase when the participating governments have agreed on the adoption of a final text. For most agreements and countries, this government signature does not implicate consent that the country is bound by the agreement but it only creates the obligation to refrain from actions against the agreement’s objective or purpose. A country’s consent to be constrained by an agreement is usually expressed via the ratification of this treaty by the national parliament (Vienna Convention on the Law of Treaties (1969)).

2 Existing research has also examined the impact of political and economic globalization as well as political regimes on states’ ratification behavior (e.g., Spilker & Koubi, 2016; Bernauer et al., 2010; Von Stein, 2008; Roberts et al., 2004).

3 For the purpose of this article, the term ‘nongovernmental organizations’ is used interchangeably with ‘transnational actors’ (TNAs), and includes nongovernmental organizations (NGOs), civil society organizations, social movements, philanthropic foundations, business associations, and multinational corporations (Tallberg et al., 2013, p. 1).

4 A Conference of the Parties (CoP) is the supreme decision-making body of an international agreement/treaty. It consists of representatives from state governments that have signed/ratified the particular treaty.

5 For instance, regarding global environmental governance, Green (2010) examines data on delegation to TNAs in more than 150 environmental treaties over the past century, and reports that the proportion of policy functions delegated to TNAs has grown dramatically over the past 25 years.

6 Bäckstrand (2006), however, attributes the emergence of this participatory norm of global environmental governance to the United Nations Conference on Environment and Development (UNCED) in Rio de Janeiro in 1992.

7 Several scholars are skeptical about the ability of TNAs to increase the democratic legitimacy of international institutions. This is mainly because TNAs are not accountable to voters, represent narrow and parochial interests, and their operational and decision-making procedures are often undemocratic and nontransparent (Bernauer & Betzold, 2012; Mitchell, 2011; Piewitt et al., 2010; Steffek & Ferretti, 2009; Kissling & Steffek, 2008; Moravcsik, 2004; Nanz & Steffek, 2004; Scholte, 2004). Dellmuth and Tallberg (2015) and Agné et al. (2015), however, report that citizen as well as stakeholder organization perceptions of IO legitimacy do not appear to be influenced by the representation of their interests in IOs.

8 A plot with the distribution of the dependent variable over time is reported on the online appendix.

9 We identify sovereign states using the Correlates of War data (2011).

10 In some cases, it is not entirely clear whether certain treaties are only in principle or also de facto open to all countries globally. We show in the robustness section that our results hold for several subsamples of treaties that are considered to be global.

11 Keeping an MEA-country observation in the data after ratification has occurred would indicate that the respective country ratifies the particular treaty in each subsequent year. This would bias our findings (McGrath, 2015).

12 The results remain unchanged if we cluster the standard errors by treaty or treaty-country dyad, or if we use a complementary log-log regression instead of the logit link function (see online appendix).

13 Identical codes were assigned in 94% of the related coding decisions.

14 To illustrate, the Montreal Protocol (Art.11.5) grants TNAs access: ‘(…) Any body or agency, whether national or international, governmental or non-governmental, qualified in fields relating to the protection of the ozone layer which has informed the secretariat of its wish to be represented at a meeting of the Parties as an observer may be admitted unless at least one third of the Parties present object. (…)’. In Contrast the Convention for the Prohibition of Fishing with Long Driftnets in the South Pacific, does not grant any non-members access to its meetings of the parties: ‘Parties to the Protocols shall be invited to any such meeting [of the Parties] and to participate in a manner to be determined by the Parties to the Convention’. (Art.7(2))

15 Calculating the intensity variable in this way implies that treaties receive a positive intensity value if range, depth, or permanence is coded ‘unspecified’. An alternative way of dealing with such unspecified values would be to assign missing values to the intensity variable if at least one of the variables range, depth, or permanence is coded ‘unspecified’. We show in the appendix that the results of model 2 (Table 2) hardly differ depending on how the intensity-variable is coded.

16 Descriptive statistics and information on missing values are reported on the online appendix.

17 We thank an anonymous reviewer for the suggestion to emphasize this surprising fact.

18 Spilker and Koubi (2016) report these variables from 1950 to 2000 and emphasize their time invariance. Thus, we carry the values observed in 2000 forward to all years from 2001 to 2011. We discuss in the robustness section that our results remain unchanged if we only study observations from 1950 to 2000.

19 For 1950–1965, IGO membership is only reported at five-year intervals. Since these consecutive observations are relatively similar, we assume a linear trend and interpolate values for the unobserved years in between observations as well as for the six years following the last observation (2006–2011).

20 This variable is taken from Spilker and Koubi (2016) who relied on Hathaway (2008). The latter describes countries as of 2007. We assume that these domestic rules are time invariant from 1950 to 2011. We address this assumption by re-estimating the main model after omitting countries that experience a regime change and therefore are more likely than other countries to have experienced constitutional changes (see the online appendix).

21 We coded whether a treaty addresses the following subjects: energy, freshwater, habitat, nature, ocean, pollution, species, or weapons. This information was downloaded on December 3 2015 from the IEA Database website (Agreements by Subject: http://iea.uoregon.edu/page.php?query=list_subject.php).

22 The six models shown in Table 2 differ in the explanatory variables used to measure TNA access, while including the same additional control variables. Models 4–6 capture only one dimension of TNA access, that is, its range, depth, or permanence. Model 3 includes measures of range, depth, and permanence together but, in order to make this feasible, only includes Treaties granting some sort of TNA access, thus the smaller number of observations and treaties included (42 vs. 178). Models 1 and 2 use one variable each that measures whether or with which intensity a treaty grants TNAs access, therefore aggregating range, depth, and permanence of TNA access.

23 We change dummy variables from zero to one and other variables from their 10th to their 90th percentile. Regarding the intensity-variable, we calculated percentiles based on positive values only. The remaining variables are held constant at their median, except for the time variables, which are set to 1960 and the public good variables, which are set to indicate that no public good is addressed. Confidence intervals summarize 20,000 simulations.

24 The reported changes are differences in probabilities and have no units.

25 Based on the expected probabilities plotted in Figure 3 for 1950–1954 (see also the online appendix).

26 Low and high intensity is defined by the smallest positive and the largest observed intensity-value. The bands indicate 95-percent confidence intervals based on 20,000 simulated expected values.

27 Given the high correlation between depth, range, and permanence, the regression estimates in models 4–6 should not be taken at face value without either controlling for the other two dimensions directly (model 3) or aggregating the information into one single measure (models 1 and 2).

28 See online appendix for results.