Introduction
The past few decades have witnessed growing societal contestation of international organisations (IOs). While IOs historically have enjoyed latent popular support, developments since the early 1990s suggest that the time of passive acceptance are over.Footnote 1 Social movements have engaged in campaigns, demonstrations, and protests, criticising IOs for their policies and decision-making modes.Footnote 2 Electorates in several European countries have rejected government-negotiated treaties when given the opportunity in popular referenda.Footnote 3 Dissatisfaction with elite-driven international cooperation has led to the establishment of alternative arenas for public debate, such as the World Social Forum.Footnote 4
These developments have generated increasing scholarly interest in the legitimacy of IOs. Informed by normative political theories, students of International Relations (IR) have debated the existence of a democratic deficit in global governance, and offered alternative assessments of the legitimacy of IOs. Several scholars have explored the democratic qualities of IOs, and found IOs wanting in terms of participation, accountability, and transparency.Footnote 5 Others have emphasised the need to evaluate IOs based on their contribution to problem-solving, and found less cause for concern.Footnote 6
This wave of research has addressed the legitimacy of IOs in its normative sense – IOs' right to rule based on their conformity to certain philosophical values and principles. By contrast, we have little systematic knowledge about the legitimacy of IOs in the sociological sense – the acceptance of IOs' right to rule by states and societies (here termed social legitimacy, for short). To date, existing empirical research on the social legitimacy of IOs is restricted to a well-developed literature on public opinion in the European Union (EU),Footnote 7 a specific literature on cosmopolitan values and public support for IOs,Footnote 8 a growing literature on legitimation and delegitimation strategies in relation to IOs,Footnote 9 and select contributions on topics such as legitimacy in the United Nations (UN) Security Council,Footnote 10 public opinion toward IOs in developing countries,Footnote 11 and the role of antipathy toward powerful states in shaping IOs' perceived legitimacy.Footnote 12
Better understanding of the social legitimacy of IOs is imperative, since we can expect it to influence IOs' potential to make a difference. Legitimacy is often seen as central for the effectiveness of political institutions in general.Footnote 13 Institutions that are perceived as legitimate can more easily attract the resources required for their persistence and secure compliance with the principles they establish. According to some observers, this pertains particularly to IOs, since IOs typically do not have recourse to the option of coercion as a means of social control.Footnote 14 Absent social legitimacy, IOs may experience greater problems gaining state support for ambitious policy goals, securing national ratification of negotiated agreements, and achieving effective compliance with IO rules and norms. Hence, ‘[t]he perception of legitimacy matters, because, in a democratic era, multilateral institutions will only thrive if they are viewed as legitimate by democratic publics.’Footnote 15
This article asks what drives citizens to support or oppose IOs. Its broader purpose is to advance an agenda of empirical research on the social legitimacy of IOs. To this end, we isolate and assess three generic theoretical explanations of the social legitimacy of IOs. The first explanation (interest representation) suggests that citizens form opinions of IO legitimacy based on IOs' procedures for channelling and representing popular interests. The second explanation (institutional performance) posits that citizen perceptions of IO legitimacy are anchored in evaluations of IOs' contributions to general and individual welfare. The third explanation (confidence extrapolation) submits that citizens are little influenced by input- or output-related properties of IOs when forming opinions about these organisations; instead, attitudes toward IOs are derived from citizens' experiences of domestic political institutions.
We test hypotheses derived from these theories through a statistical analysis of data on the social legitimacy of the UN. We focus on the UN for three reasons. First, we have access to more encompassing public opinion data on the UN than on any other global organisation. Second, the UN probably constitutes the most well-known IO in world politics, making it reasonable to assume that respondents are able to form and express opinions toward the UN. Third, the UN occupies a central position in global governance as the largest IO in terms of membership and policy scope, making it a particularly important case for an assessment of social legitimacy in international cooperation. Conceiving of social legitimacy as an attitudinal phenomenon, we construct a dataset on public confidence in the UN based on World Values Survey (WVS) and the European Values Study (EVS) data for 26 countries from 1999 to 2004.
Our multilevel analysis provides clear empirical support for two explanations: institutional performance and confidence extrapolation. By contrast, the evidence for a link between interest representation and UN legitimacy is weak. These results indicate that the social legitimacy of the UN primarily is anchored in the organisation's capacity to deliver, as well as in citizens' general confidence in political institutions. While many have called for IOs to become more inclusive and democratic in order to strengthen their perceived legitimacy, the evidence from the UN suggests that many citizens still base their assessments of IOs on the benefits they generate for states and societies – next to domestic experiences that IOs often have little to do with and can do little to change.
This article proceeds as follows. The next section outlines the three alternative explanations of the social legitimacy of IOs. The third section introduces the dataset and formulates a set of statistical models that test the developed hypotheses. The fourth section presents the empirical results. The conclusion summarises the findings and outlines implications for the understanding of global governance.
The social legitimacy of IOs: Theories and hypotheses
Legitimacy has two alternative conceptual meanings: a normative and a sociological.Footnote 16 While normative legitimacy refers to an institution's right to rule, based on its conformance to certain values and principles, sociological legitimacy refers to the acceptance of an institution within a given audience. Inquiries into the normative and sociological legitimacy of political institutions are thus guided by different questions. Studies of normative legitimacy typically ask ‘[b]y what normative standards should we evaluate institutions' right to rule, and how do individual institutions fare when measured against these standards?’ while inquiries into social legitimacy ask ‘[t]o what extent are institutions accepted by the people, and what explains variation in their perceived legitimacy?’
When we refer to the social legitimacy of IOs, we speak of their legitimacy in the sociological sense. We perceive of legitimacy as a social property of the relationship between an IO and its public, here conceptualised as the citizens of its member states.Footnote 17 The social legitimacy of an IO says little about the actual rightness or goodness of the organisation; it refers exclusively to the public's acceptance of and support for that organisation. Consequently, it is not a constant, but may vary over time and across subsets of the public. Neither is the social legitimacy of an IO necessarily based on a single logic, but may be shaped by multiple sources that make citizens more or less supportive of an organisation.
In line with existing literature, we distinguish between input- and output-legitimation of IOs.Footnote 18 In the logic of input-oriented legitimation, IOs generate societal acceptance as a result of procedures that allow for participation and representation by and of the people. In the logic of output-oriented legitimation, IOs gain acceptance by governing effectively and generating benefits for the people. In the following, we refer to these explanations as interest representation and institutional performance. Existing studies of both discourse and attitudes grant support to the assumption that we can distinguish analytically and empirically between input- and output-related sources of social legitimacy.Footnote 19
In both logics, we expect that citizens may form opinions about IOs based on both private and collective dimensions of interest representation and institutional performance. This presupposes a broad understanding of legitimacy that does not preclude sources of societal acceptance based on instrumental considerations. While some researchers restrict legitimacy to the acceptance of an institution's right to rule irrespective of its consequences for private interests, we operate with a conceptualisation that allows for instrumental concerns to play a causal role in the formation of legitimacy beliefs.Footnote 20 Hence, input-oriented considerations may be based on whether citizens themselves are well-represented or whether IO decision-making in general provides opportunities for citizens to be represented. Likewise, output-oriented considerations may include both whether IOs contribute to citizens' individual welfare and whether they promote collective welfare.
In addition to these standard logics of legitimation, we assess a third explanation that recently has gained prominence in research on the social legitimacy of the EU, and whose explanatory potential remains to be explored beyond this empirical realm.Footnote 21 According to the logic of extrapolation, citizens form legitimacy beliefs about IOs based on heuristics derived from domestic political institutions. In other words, citizens' confidence in IOs is based on cues obtained in domestic political arenas and may be unrelated to citizens' evaluations of IOs. The hypotheses we develop are tailored for an assessment of these logics in the context of a single IO – the UN.
Interest representation
The first explanation is based on the premise that citizens form opinions about the legitimacy of IOs based on the organisations' institutional arrangements for channelling and representing citizen demands. It assumes that popular conceptions of legitimacy are informed by input-oriented standards – participation and representation – that are central to domestic democracies.Footnote 22
This explanation of the social legitimacy of IOs features prominently in existing scholarship, where many contributions point to growing societal contestation of IOs, rooted in discontent with existing modes of citizen involvement in policymaking.Footnote 23 Common examples are popular protests against organisations such as the EU, the World Bank, and the UN. While IOs for long profited from what Leon Lindberg and Stuart Scheingold termed a ‘permissive consensus’ – a situation where citizens in general took little interest in international cooperation, but supported its broad goals – this appears to have become an increasingly inaccurate description.Footnote 24 In the context of the EU, there is evidence that citizens' evaluations of the organisation's democratic procedures and institutions positively affect their attitudes toward the EU, and that an exclusive focus on the effectiveness of the EU would not be sufficient to satisfy citizen demands.Footnote 25
The overall expectation derived from this explanation is that popular perceptions of IO legitimacy will be shaped by existing procedures for citizen input and representation. Generally speaking, we would expect IOs with more inclusive decision-making systems to be perceived as more legitimate. In the context of a single IO, such as the UN, we would expect citizens who are relatively better represented in the decision-making system to be more favourably disposed toward that IO. This logic yields three specific expectations.
First, IO procedures for citizen representation through member governments may influence public perceptions of legitimacy. This form of indirect representation remains the predominant mode of citizen representation in most IOs, and it is reasonable to expect that citizens oftentimes conceive of the national government as their primary channel of representation. Yet, while many intergovernmental bodies are based on the principle of state equality, some IOs grant an exclusive group of member states a disproportionate say in decision-making.Footnote 26 The World Trade Organization (WTO) is famous for its club model of multilateral cooperation, the International Monetary Fund (IMF) for differentiated voting power according to economic strength, and the UN for the exclusive powers of the permanent members of the Security Council. Such institutionalised disparities have been the objects of popular criticism and can be expected to influence public perceptions of IO legitimacy. Typically, we would expect those states and citizens who are disfavoured by such arrangements to find them least acceptable, and vice versa.Footnote 27 Hence, we hypothesise that:
Hypothesis 1. The better-represented citizens are in an IO through their national government, the more likely they are to perceive of that IO as legitimate.
Second, IOs' procedures for citizen representation through civil society organisations (CSOs) may shape the social legitimacy of IOs. Bernauer and Gampfer, for example, find that civil society involvement positively affects public support for global environmental governance.Footnote 28 According to many accounts, recent years have witnessed the emergence of a global participatory norm, prescribing civil society involvement as a remedy for democratic deficits in global governance.Footnote 29 In this vein, former UN Secretary-General Boutros Boutros-Ghali characterised CSOs as ‘a basic form of popular representation in the present-day world’.Footnote 30 Furthermore, research shows that IOs have increasingly opened up to CSOs over recent decades, but that significant variation remains.Footnote 31 Some IOs provide for more civil society involvement than others, and where CSOs enjoy access, some citizens tend to be better represented than others. Some types of organisations (for example, business associations) may enjoy greater access and be better mobilised than others (for example, consumer protection associations), just as some countries (for example, democracies) may provide a more fertile ground for CSOs to develop and go international than others (for example, autocracies). Such differences are likely to influence citizen evaluations of IO legitimacy. Functioning as a ‘transmission belt’ between citizens and IOs,Footnote 32 CSO both alert IOs to citizen preferences and inform citizens of IO activities. Hence, we expect that:
Hypothesis 2. The better represented citizens are in an IO through CSOs, the more likely they are to perceive of that IO as legitimate.
Third, citizens' general attitudes toward democracy as a system of governance may affect popular assessments of IO legitimacy. The creation and empowerment of IOs is a case of delegation to non-majoritarian institutions, equal to the delegation of power to non-elected expert institutions in national democracies, such as agencies, courts, and central banks.Footnote 33 If citizens greatly value democracy as a system in itself, with its emphasis on popular participation and majoritarian rule, we could expect them to be critical of IOs. Indeed, the shifting of issues from the domain of national democratic contestation to international technocratic governance is a common component of popular criticism against IOs.Footnote 34 We hypothesise:
Hypothesis 3. The more citizens value democracy as a system of rule, the less likely they are to perceive of IOs as legitimate.
Institutional performance
The premise of the second explanation is that citizens form opinions about the legitimacy of IOs based on their institutional performance. It assumes that popular conceptions of legitimacy are informed by output-oriented standards – effectiveness and benefits – that contribute to general and individual welfare. This explanation suggests that substantive outcomes are more powerful in shaping citizens' views toward IOs than inclusiveness and representation.
The notion that institutional output can account for public confidence is a prominent explanation in the study of national political institutions. In this vein, Newton and Norris suggest that: ‘[g]overnment institutions that perform well are likely to elicit the confidence of citizens; those that perform badly or ineffectively generate feelings of distrust and low confidence.’Footnote 35 This line of argument is often invoked in the extensive literature on public attitudes toward European integration,Footnote 36 but rarely in present scholarship about IOs other than the EU. One exception is Ecker-Ehrhardt, who finds support for an output-related logic of legitimation in relation to the EU, UN, WTO, IMF, and G8.Footnote 37 Instead, it is the conventional view that output legitimation dominated in the period of permissive consensus, when citizens enjoyed the fruits of international cooperation and gave their latent support, but has fallen out of fashion, as citizen awareness and contestation of IOs has grown.Footnote 38
The overall expectation derived from this explanation is that the social legitimacy of IOs will be shaped by the organisations' contribution to general and individual welfare. In terms of general welfare, we would expect IOs that perform well to be more likely to enjoy citizens' support than IOs that perform badly. In terms of individual welfare, we would expect citizen support to vary depending on whether the institutional output from a specific IO on balance is advantageous or disadvantageous.Footnote 39 Such output-based legitimacy may be rooted in, and explored through, either subjective or objective benefits from IOs, depending on whether we focus on citizens' beliefs about interest satisfaction or objective facts about their needs and interests.Footnote 40 This logic generates three specific expectations.
First, since IOs nowadays engage in a broad range of executive tasks, they are likely to be evaluated by citizens according to how well these functions are performed. The delegation of authority to IOs is typically motivated by the capacity of IOs to address cross-border problems better than individual states. Yet perceptions of how well IOs succeed may vary across citizens, influencing their acceptance of and support for IOs. We hypothesise that:
Hypothesis 4. The higher the perceived problem-solving effectiveness of an IO, the more likely citizens are to perceive of that IO as legitimate.
Second, IOs generate costs and benefits that tend to be unequally distributed across member states and segments of their populations, and which we can expect will influence the perceived legitimacy of IOs. Such costs and benefits may be economic, as when regulatory policy produces varying distributional consequences, or when IOs engage in redistributive policy, such as development aid. In some IOs and countries, member state contributions to the budget are highly politicised as well, as in the case of the US contribution to the UN budget and net payers' contributions to the EU budget. The costs and benefits from cooperation can also be political. While states may value membership of an IO because of the general advantages of collective decision-making, not all states are likely to enjoy the same political benefits from IO policy.Footnote 41 Moreover, specific decisions or actions may be targeted at individual member states, generating domestic contention of IO policy. We hypothesise:
Hypothesis 5. The more advantageous the distribution of costs and benefits from IO output, the more likely citizens are to perceive of that IO as legitimate.
Confidence extrapolation
The premise of the third explanation is that citizens form legitimacy beliefs vis-à-vis IOs based on their general predispositions rather than their perceptions of input- and output-related properties of these organisations. As a consequence, confidence in IOs may be unrelated to IOs themselves but rather derived from confidence in national political institutions.
The general notion that different forms of confidence in political institutions are inextricably related, regardless of whether institutions are situated at the domestic or the international level, is well represented in existing literature. Previous works in Comparative Politics find that positive experiences can give rise to virtuous circles of high trust in political institutions, just as negative experiences can lead to vicious circles of low trust.Footnote 42 For instance, when the performance of representative government is seen as effective, this strengthens the perceived legitimacy of the legislature, the judiciary, and the executive.Footnote 43 Conversely, corruption in a political system tends to undermine citizen confidence in political institutions in general.Footnote 44
Recently, this logic has been put to a test in research on the legitimacy of the EU, and been found to have extensive explanatory power.Footnote 45 According to this literature, legitimacy beliefs vis-à-vis the EU are, at least in part, a function of a ‘trust syndrome’,Footnote 46 whereby the legitimacy of the EU can be predicted to a significant degree by the legitimacy of domestic political institutions. This argument is based on the premise that citizens know little about IOs, their procedures, and their performance, and therefore form legitimacy beliefs about these organisations based on their general experiences of political institutions at the domestic level, which are used as heuristics or cues. In this vein, Armingeon and Ceka explain the decline in support for the EU during the recent economic recession, not with the performance of the EU, but with the declining levels of trust in national governments.Footnote 47 The general implication is that IOs can do little on their own to raise their legitimacy in the eyes of citizens. Drawing on this body of literature, we hypothesise:
Hypothesis 6. The higher citizens' confidence in national political institutions, the more likely citizens are to perceive of IOs as legitimate.
Research design
To empirically study the social legitimacy of IOs, we require a method for measuring the acceptance of the organisations within a given audience. Existing research suggest three alternative methods of inquiry: assessments of public opinion (audience beliefs), mapping of public participation and protest (audience behaviour), and analysis of discursive practices in the public realm (audience statements).Footnote 48 In this article, we are interested in examining the theoretical mechanisms that may explain variation in popular support for IOs at the individual level, and therefore rely on assessments of public opinion. Recent Gallup polls in 72 countries indicate that citizen awareness of IOs is high, with as much as 84 per cent reporting familiarity with the UN.Footnote 49
To operationalise the hypotheses, we compile a dataset based on individual-level data from the third wave of the EVS and the fourth wave of the WVS. Since some of the hypotheses predict an effect of contextual factors, we add country-level characteristics for the specific survey years to create a dataset with individuals nested in 26 countries that vary widely in terms of economic development, degree of democratisation, and experience with military and colonial conflicts.Footnote 50
The dependent variable, UN confidence, builds on responses to a question about how much confidence citizens have in the UN: a great deal of confidence, quite a lot of confidence, not very much confidence, or no confidence at all.Footnote 51 The more citizens know about the particular operations and policies of the UN, the more likely it is that responses to this question capture citizens' specific support for the UN. Yet, since the particular activities of the UN may not always be visible and known to citizens, this question may also capture citizens' diffuse support for the IO, based on general impressions. The confidence measure is a conventional indicator of the social legitimacy of political institutions in Comparative Politics.Footnote 52
Since variation in citizen confidence in the UN across countries and groups of individuals has been thoroughly described elsewhere,Footnote 53 we limit ourselves to depict how the variable UN confidence varies across the countries in the dataset. Figure 1 reveals that in four countries, more than 80 per cent of the respondents indicated having a great deal or quite a lot of confidence in the UN (Albania, Bangladesh, Tanzania, and Uganda). By contrast, less than 50 per cent of the respondents in Argentina, Guatemala, and Indonesia had a great deal or quite a lot of confidence in the UN. The mean percentage of respondents expressing a great deal or quite a lot of confidence in the UN across the 26 countries in our dataset is 65.80 per cent, with a standard deviation of 12.44 and a range of 50.15 per cent.
UN confidence in 26 countries
Note: Number of individuals: 21,860
Next, we turn to the measurement of the independent variables. To begin with, we assume that citizens can be represented at the UN level through two main venues: first, through national governments, and second, through CSOs. To capture the degree of representation through national governments predicted in hypothesis 1, we draw on the UN Security Council membership database to code the variable Government representation 2 if a country was a permanent member of the Security Council in the year in which the survey data for that particular country was collected, 1 if it was a temporary member, and 0 otherwise.Footnote 54 We assume respondents to be aware of their country's permanent or temporary membership in the Security Council. While the exclusive status of the five permanent members is common knowledge, temporary memberships generate extensive attention as well, as contestants invest considerable resources into election campaigns and, when successful, try to make the most of the visibility and authority of the office.Footnote 55
To measure representation through CSOs, which we expect to have a positive effect on confidence levels in hypothesis 2, we draw on information from the civil society database of the UN.Footnote 56 The variable Civil society representation measures the number of CSOs accredited to the UN from a particular country in the year in which the survey data for that country was collected. A few countries have an exceptionally high number of CSOs accredited to the UN, notably, the United States (292 CSOs) and India (222 CSOs). We address the potential problem that the inclusion of these two particular countries biases the regression results by including the variable Civil society representation in its logarithmic form to treat deviations above and below the average value of this variable in the same way. Furthermore, we replicate all analyses, excluding the United States and India.
Third, to test hypothesis 3, we create a measure of citizens' Democracy support through an index used in previous studies.Footnote 57 To do so, we sum up the respondents' support for the statement ‘[d]emocracy may have problems but it's better than any other form of government’ and ‘[h]aving a democratic political system in the home country.’ This yields a 0-6 pro-democratic scale, from which we subtract another scale that we obtain by adding respondents' support for the statements ‘[h]aving a strong leader who does not have to bother with parliament and elections’ and ‘[h]aving the army rule.’ The resulting index ranges from –6 (maximum autocratic support) to 6 (maximum democratic support).
Turning to institutional performance, we operationalise hypotheses 4 and 5 through two indicators. First, to measure citizens' perceptions of UN problem-solving, we use responses to a question about who should decide on specific problems: the UN only, the national government under the guidance of the UN, or the national government only. Respondents could express their opinion with regard to five policy areas: human rights, international peacekeeping, protection of the environment, aid to developing countries, and refugees. By adding up the responses on each of these dimensions, we obtain an index, Perceived UN problem-solving, ranging from 0 (minimum support for UN policy output) to 10 (maximum support for UN policy output). This measure rests on the assumption that respondents who wish a greater involvement of the UN in policymaking are more content with UN output.Footnote 58
Second, we create a measure of the costs that a country incurs due to UN membership, assuming that citizens do cost-benefit calculations when forming opinions about UN policy output. The variable Net contributions measures the yearly average of a country's net contributions per capita to the regular UN budget. To code this variable, we take figures from the official UN record of member state contributions.Footnote 59 The amount of member state contributions to the regular budget is calculated on the basis of the country's ability to contribute, whereby factors such as gross national income and state debt are considered.Footnote 60 As wealthier countries contribute more than their poorer counterparts, the variable is skewed to the right. Again, to treat deviations above and below the average value of this variable in the same way, Net contributions enters the regression models in its logarithmic form.
Finally, we operationalise hypothesis 6 on confidence extrapolation through one indicator. To explore whether citizens form attitudes toward the UN based on their attitudes toward domestic political institutions, we create a variable on the basis of a question about confidence in various national state institutions. The variable Government confidence summarises citizen confidence in the national parliament and civil services in an index ranging from 0 (minimum confidence) to 6 (maximum confidence).
We also include several control variables in the analysis. At the individual level, we seek to capture citizens' sense of global belonging by using the responses to two questions about the geographical domains respondents feel that they belong to first and second.Footnote 61 The variable Geographical identity is coded 2 if the respondent indicated a belonging to the world or the continent as a first choice, 1 if the respondent indicated a belonging to the world or the continent as a second choice, and 0 in the case of a national, regional, or local identity.Footnote 62 Furthermore, we measure Income on the basis of a five-point ordinal scale. By calculating the income quintiles of respondents' households, we seek to capture the relative income of households. This is based on the notion that relative rather than absolute income should influence respondents' attitudes toward political institutions.Footnote 63 To illustrate, households that score 1 are below the poverty line, whereas households that score 5 have an annual income of 75,000 USD or more. We construct the variable Education as an index ranging from 0 to 4, where respondents score 0 if they have inadequately completed elementary education, 1 if they have completed elementary education, 2 if they have completed secondary school of a vocational or technical type, 3 if they have completed secondary school of a university-preparatory type, and 4 if they have entered university or have a university degree. Age enters the analysis as a continuous variable. Furthermore, we measure Cognitive mobilisation, the level of awareness of an institution, by using the frequency of a respondent's discussion of politics with friends. Those who possess a greater store of political information (the ‘cognitively mobilised’) should be more able to take positive information about the UN into account when forming an opinion about the organisation and may hence be more likely to support the UN.Footnote 64 Gender is as a dummy variable equal to 1 for women and 0 for men. Finally, we include a dummy variable measuring generalised social trust, as previous research suggests that social trust may be related to confidence in the UN.Footnote 65 The variable Social trust is coded 1 if respondents indicate that ‘most people can be trusted’ and 0 if they indicate that one ‘can't be too careful’.Footnote 66
At the country-year level, we include a variable Conflict location that measures domestic armed conflict. It is coded 0 if a country is not listed as a conflict location, 1 if it is listed as a location of a minor armed conflict, 2 in the case of an intermediate armed conflict, and 3 in the case of a war.Footnote 67 Citizens from conflict zones might perceive the UN differently than citizens from countries that are not plagued by armed conflicts or wars, depending on their perception of the success of UN peacekeeping interventions. Countries listed as conflict zones in the dataset are Algeria, India, Macedonia, the Philippines, and Uganda. Given that most countries score zero on this variable, we replicate all regression models by excluding this variable in order to check if the results hold when we do not control for Conflict location. Moreover, we use the Corruption Perception Index to capture the quality of domestic institutions.Footnote 68 The idea behind this variable is that people who experience corruption in the domestic context will be more sceptical of IOs as well. On the other hand, high levels of corruption may raise hopes among citizens that IOs can help combat this problem, which would increase their support for the UN. Last, we introduce a variable Colonial past that assumes a value of 1 if a country previously has been colonised by a Western overseas colonial power.Footnote 69 Citizens in countries with a colonial legacy might view the UN as a vehicle of Western hegemony, which would decrease their confidence in the organisation.Footnote 70 At the same time, citizens from countries with a colonial past could be more supportive of the UN, as they might perceive the equal voting rights in the General Assembly as an opportunity to have influence on a par with that of former colonial powers. More than half of the countries in the data set have a colonial origin, among them Argentina, Bangladesh, and Zimbabwe.
Empirical analysis
The empirical analysis consists of a series of multilevel models that distinguish between two levels: the individual and the country level. Given that our data are hierarchical, the main advantage of multilevel models compared to single-level regression models is that multilevel models adjust for the correlation of the error components at the two levels.Footnote 71 As the dependent variable UN confidence takes on four ordered values, we estimate the models using ordered logistic regression. This type of regression analysis estimates a latent variable as a linear function of the independent variables and a set of cut points. Since the indicator for UN confidence takes on four values, there are three cut points. The aim is to observe the probability that the estimated linear function is within the range of these cut points estimated for the dependent variable. In other words, we will observe at which point on the latent variable the categories change, for example from ‘no confidence’ to ‘not very much confidence’. We add a random intercept, as responses are likely to vary across national contexts. The main model is written as follows:Footnote 72
$$logit\left \{Pr(y_{ij} > s\mid X_{ij},\zeta_{1j} )\right \}=\beta _{2}(Government\ representation_{2j})+ \beta_{3}(Civil\ society\ representation_{3j})+\beta_{4}(Democracy\ support_{4ij})+\beta_{5}(Perceived\ U.N.\ problem-solving_{5ij})+ \beta_{6}(Net\ contributions_{6j})+\beta_{7}(Government\ confidence_{7ij})+\beta_{w}X_{ij}+\beta_{z}Z_{j}+\zeta_{1j}+\varepsilon_{ij}$$
, where Pr(y ij >s|x ij , ζ1j ) is the cumulative probability that respondent i living in country j has a level of confidence in the UN that is higher than the threshold s; w are vectors for the individual-level controls; z are vectors of country-level controls; ζ1j is the intercept of the cumulative logits that varies over country j; and εij is the error term that is separate for country-level j. The error term is normalised to π2/3.Footnote 73
Results
We begin the empirical analysis by establishing the extent to which variation in UN confidence is clustered among countries. An analysis of variance (ANOVA) model of our sample decomposes the variance in the dependent variable between individual and country levels. This way we establish the intra-class correlation ρ, that is, how much of the total variation in UN confidence that can be explained by differences at the country level. Table 1 shows that the random-intercept variance is estimated as 0.265, implying that about 8 per cent of the differences in the level of confidence in the UN can be explained by the fact that respondent live in different countries.Footnote 74
Base model predicting UN confidence
Notes: **p<.01. Robust standard errors in parentheses. Number of individuals: 21,860. Number of country-years: 26.
Table 2 presents the results. Models 1 and 2 test each of the three sets of hypotheses by including the control variables in two steps. Taken together, the results grant extensive support to the explanations of institutional performance and confidence extrapolation, and less support for interest representation.
Multivariate models predicting UN confidence
Notes: *‘p<.05, **p<.01. Figures are unstandardised coefficients. Robust standard errors in parentheses. Number of individuals: 21,860. Number of country-years: 26.
Beginning with interest representation, Government representation and UN confidence are negatively related, contrary to the expectation in hypothesis 1. This result suggests that citizens do not respond to their government's membership status in the UN Security Council with more confidence in the UN. Hypothesis 2 is corroborated by the results, as the coefficient of Civil society representation is positive and statistically significant at the 1 per cent level. It seems that citizens respond to better representation in the UN through their CSOs with greater confidence in the organisation. Contrary to the expectation in hypothesis 3, citizens appear to have greater confidence in the UN, the more they value democracy as a system of governance. This might be explained by citizens associating the UN with the representation of all global peoples, or by citizens not perceiving technocratic governance as a threat to democracy, since majoritarian and non-majoritarian institutions typically coexist within domestic democracies.
Turning to institutional performance, citizen perceptions of UN problem-solving and UN confidence are positively related, as expected in hypothesis 4. To illustrate the magnitude of this association, we calculate odds ratios on the basis of model 2. The results from model 2 reveal that a one-unit increase in UN problem-solving raises the odds of expressing a great deal of confidence in the UN, compared to the combined lower categories, by a factor of 1.069, or 7 per cent, holding all other covariates fixed. With regard to hypothesis 5, Net contributions and UN confidence are negatively related in model 1, as expected. However, the coefficient turns insignificant in model 2. We suspect that this finding in model 2 is a statistical artifact, resulting from the high correlation between Net contributions and Corruption perception (r=0.854). When replicating model 2 by excluding the other two country controls and keeping Corruption perception, the coefficient of Net contributions remains insignificant. By contrast, when estimating model 2 by excluding only Corruption perception but keeping the other two country controls, the coefficient of Net contributions is significant at the 1 per cent level and negative. Likewise, Net contributions is negatively significant across a range of other model specifications (see the discussion below on robustness checks). We therefore choose to interpret this result as support for hypothesis 5. In sum, these results underline the importance of the institutional performance of the UN for the social legitimacy of the organisation.
Finally, we find support for the logic of extrapolation. In line with hypothesis 6, confidence in the national government is positively related to confidence in the UN. For a one-unit increase on the index of Government confidence in model 2, the odds of more confidence in the UN increase by as much as 71 per cent. This result corroborates the logic of extrapolation, suggesting that many citizens' confidence in IOs is systematically associated with their confidence in national political institutions.
With regard to the control variables at the individual level, there is mixed support for Geographical identity as a factor influencing UN confidence. Geographical identity is significant in model 1, but turns insignificant when we add country-level controls in model 2. This result suggests that there is no clear-cut, positive relationship between a person's sense of global belonging and her/his confidence in the UN, as suggested in previous research.Footnote 75 As for the other individual-level controls, there is no robust evidence that they are systematically related to UN confidence.
Among the country-level controls, the coefficient for Conflict location is insignificant, indicating that armed conflicts do not matter for citizens' confidence in the UN. We note that the UN only had peace-keeping operations in two of the countries listed as conflict zones (Guatemala and Uganda), which may explain why the presence of conflicts does not shape perceptions of the organisation.Footnote 76 Corruption perception is negatively significant, indicating that people who experience corrupt institutions in the domestic context are sceptical of the UN as well. Last, Colonial heritage is negatively related to UN confidence. This may be because citizens in countries with a colonial legacy perceive the UN an organisation dominated by old colonial powers (France, UK) and new powers with worldwide political interests (US, China).
Robustness checks
To test whether our findings hold across different model specifications, we conducted a series of robustness checks. First, we sought to address several concerns regarding the measurement of CSO representation. We replicated the models in Table 2 by including a logged per capita measure of the number of CSOs represented in the UN. We also ran all models using a non-logarithmised measure of CSO representation. In both replication models, the alternative measures of CSO representation are positively associated with confidence in the UN, corroborating the results presented in Table 2. Furthermore, we were concerned that the results for the variable CSO representation may be driven by two outlier countries that host by far the greatest number of CSOs, India, and the US. Consequently, we replicated all models excluding the observations for India and the US. However, the magnitude and significance levels of the coefficient of CSO representation and the other coefficients in the replication model remain largely unchanged.
Second, we assessed whether the direct effects we found differed among the cognitively mobilised. Previous literature on public attitudes toward the EU has found that cognitively mobilised citizens differ in the way they acquire information from their environment about the organisation, which in turn shapes their opinion toward the EU. The cognitively mobilised talk more frequently about politics and are more likely to work and live in environments that facilitate the acquisition and maintenance of political knowledge.Footnote 77 In light of this, some of the direct effects we find may in fact be conditioned by the variable Cognitive mobilisation. In particular, context information about civil society or government representation may affect UN confidence only or especially among the cognitively mobilised. To test the robustness of the direct effects, we replicated model 2 by adding product terms for the variables that operationalise the hypotheses and Cognitive mobilisation. We added these product terms separately in the regression models and plotted them to see whether the effects are statistically significant at all levels of Cognitive mobilisation and whether the magnitude of the effects changes.Footnote 78 However, none of the direct effects we present in Table 2 is mediated by Cognitive mobilisation.
Third, and related, we assessed whether the effects of some variables differed among those citizens that have a sense of global belonging. Previous research has theorised a connection between cosmopolitan identities and attitudes toward international organisations.Footnote 79 In the context of our study, citizens identifying with a global community may be more likely to be exposed to information about the UN and more positively predisposed toward that information. In other words, some of the direct effects we find may be affected by how citizens score on the variable Geographical identity. However, we do not find evidence of interaction effects between the explanatory variables and Geographical identity. These results strengthen our confidence in the findings about the direct effects presented in Table 2.
Fourth, the fact that some citizens did not give a substantive answer when asked about their confidence in the UN raises the issue of non-attitudes. If these missing responses are unevenly distributed, we might face a problem of sample selection that could cause biased estimates. To check whether our regression results are biased for this reason, we replicated all models using two equations: one equation considering effects on the outcome variable UN confidence, and another equation considering a portion of the sample whose outcome is observed and mechanisms determining the selection process.Footnote 80 The magnitude and significance levels of the coefficients yielded by these models are similar to the estimates presented in Table 2. The only difference is that the coefficient of Geographical identity, which is significant and positive in model 1 but not in model 2, turns insignificant in the selection models, corroborating our conclusion that a citizen's sense of global belonging may not matter for that citizen's confidence in the UN.
Finally, we were concerned that some of the country-level variables may have biased the estimates presented in Table 2 because they are highly correlated with each other. Specifically, we were concerned about two variables. First, Colonial heritage, which is moderately highly correlated with Corruption perception (r=−0.578) and Net contributions (r=−0.561), and second, Corruption perception, which is highly correlated with Net contributions (r=0.854). We ran all models presented in Table 2 by including all country variables separately. In addition, we estimated all models without Conflict location to check whether its skewed distribution biases our estimates. These robustness checks do not change the interpretation of the findings from Table 2 reported above.
Conclusion
Although social legitimacy is central to IOs, systematic assessments of the sources of public support in global governance have been in short supply. This article is an attempt to remedy this situation. Its central finding is clear empirical support for institutional performance and trust extrapolation as explanations of the UN's perceived legitimacy. When forming opinions about the legitimacy of the UN, citizens are influenced by the real or anticipated policy output of the organisation. In addition, they draw on experiences from national political institutions and extend these attitudes toward the UN. By contrast, citizen perceptions of UN legitimacy do not appear to be influenced by the representation of their interests in the organisation.
These findings highlight a gap between what much of existing scholarship presumes that citizens care about and what actually determines their evaluations of an IO such as the UN. Following public protests against several multilateral economic organisations and growing dissatisfaction with elite-driven cooperation, it has been conventionally assumed that citizens' perceptions of IO legitimacy primarily are rooted in evaluations of their potential for participation and representation. As Steven Bernstein concludes: ‘[t]he dominant answer to what legitimacy requires in global governance is democracy’.Footnote 81
Yet our findings suggest that IOs' systems for channelling and representing citizen interests may not be as decisive for popular legitimacy as commonly thought. Instead, the UN case indicates that the social legitimacy of IOs may be based primarily on the organisations' capacity to solve societal problems and generate benefits. While popular awareness and contention of IOs is higher today than three or four decades ago,Footnote 82 securing better citizen representation is not necessarily the panacea to faltering social legitimacy. While IOs have become increasingly open to civil society actors over recent decades,Footnote 83 the consequence of this transformation for IO legitimacy may primarily be indirect, through its effects on IOs' problem-solving capacity.
In addition, our findings suggest that perceptions of IO legitimacy are partly unrelated to the institutions and actions of these organisations. While performing well helps, it is no guarantee for popular legitimacy, since citizens partly base their assessments of IOs on their experiences from domestic political institutions. For many citizens, IOs are complex and distant organisations, whose legitimacy is best approached via heuristics and cues derived from the more familiar national political context. These findings from the UN mirror those in recent research on the EU, suggesting that trust extrapolation may have extensive scope as an explanation of the social legitimacy of IOs. This impression is reinforced by the fact that both IOs are among those most visible to citizens. If people use domestic heuristics when forming opinions about these relatively well-known IOs, they are even more likely to do so in relation to IOs that are less visible and known. This would mean that international policymakers can do little to boost the popular legitimacy of IOs, since the determinants of confidence largely are out of their hands.
The contribution of this article should be judged against its limitations, of which three should be noted, since they suggest an agenda for future research. In all cases, the principal barrier to progress so far has been the availability of systematic and comparable data. First, our empirical exploration has been isolated to the UN. Expanding the scope to other IOs would allow us to assess whether the sources of social legitimacy vary across IOs that perform different functions, are active in alternative world regions, and are vested with varying levels of political authority. Second, the analyses have been limited to the time period 1999–2004 because this is the only time period for which data suitable to operationalise all hypotheses is available. This time period includes the global war on terror during which the UN may have been exceptionally visible owing to its involvement in debates about the US's interventions in Iraq and Afghanistan. People may have been paying attention to the UN more than usual in this time period, and this implies that they may have based their evaluations of the UN to an unusual degree on output considerations. Future research is needed to explore whether our findings hold for a more extended time period. Third, our inquiry has focused on the citizens of member states as the audience whose acceptance and support IOs must win to be legitimate. While this may be the most intuitive choice, it is not the only way to conceptualise an IO's audience.Footnote 84 If we distinguish between elites and citizens, we may explore whether legitimacy conceptions vary systematically across the state-society divide. Moving toward a comparative, empirical study of the social legitimacy of IOs presents a promising agenda for future research and a central task for IR scholars concerned with the conditions for effective and legitimate global governance.
Appendix A
Countries included in the dataset and country-level variables
Summary statistics
Correlations between independent variables
Appendix B
World Values Survey Questions and Assigned Codes
UN CONFIDENCE: ‘I am going to name a number of organizations. For each one, could you tell me how much confidence you have in them: is it a great deal of confidence (3), quite a lot of confidence (2), not very much confidence (1) or none at all (0)? United Nations’
DEMOCRACY SUPPORT: ‘I'm going to read off some things that people sometimes say about a democratic political system. Could you tell me if you agree strongly (3), agree (2), disagree (1) or disagree (0) strongly, after I read each one of them? Democracy may have problems but it's better than any other form of government’
‘I'm going to describe various types of political systems and ask what you think about each as a way of governing this country. For each one, would you say it is a very good (3), fairly good (2), fairly bad (1) or very bad (0) way of governing this country? Having a democratic political system; Having a strong leader who does not have to bother with parliament and election; Having the army rule’
UN PROBLEM-SOLVING: ‘Some people believe that certain kinds of problems could be better handled by the United Nations than by the various national governments. Others think that these problems should be left entirely to the respective national governments; while others think they would be handled best by the national governments working together with co-ordination by the United Nations. I'm going to mention some problems. For each one, would you tell me whether you think that policies in this area should be decided by the national governments (0), by the United Nations (2), or by the national governments with UN coordination (1)? Human rights; International peacekeeping; Protection of the environment; Aid to developing countries; Refugees’
GOVERNMENT CONFIDENCE: ‘I am going to name a number of organizations. For each one, could you tell me how much confidence you have in them: is it a great deal of confidence (3), quite a lot of confidence (2), not very much confidence (1) or none at all (0)? Parliament; The civil services’
GEOGRAPHICAL IDENTITY: ‘To which of these geographical groups would you say you belong first of all? Locality (1), region (2), country (3), continent (4), world (5)?’
COGNITIVE MOBILISATION: ‘When you get together with your friends, would you say you discuss political matters frequently, occasionally or never? Frequently (2), occasionally (1) or never (0)?’
SOCIAL TRUST: ‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Most people can be trusted (1), can't be too careful (0)’
