The Unintended Consequences of Democracy Promotion: International Organizations and Democratic Backsliding

IOs and Democratic Outcomes

The literature linking IOs to democracy emphasizes these organizations’ impacts on elections and executives. Because executives serve as the main intermediaries between states and IOs (Moravcsik, 1994), democracy promotion by IOs is heavily concentrated on executives. Scholars find IOs support democracy via material incentives that induce incumbents to undertake institutional reforms (Donno, 2013). IOs use economic sanctions to disincentivize antidemocratic behavior and provide economic assistance to member states, which helps domestic leaders deter rebellions. IOs also socialize rulers into accepting democracy and influence their international status either by helping them build a reputation for being committed to democracy (Poast & Urpelainen, 2018) or by shaming those who violate electoral norms (Donno, 2013). Finally, IOs serve as commitment devices, helping democratic leaders gain domestic elites’ support and deterring political losers from overthrowing the new regime (Pevehouse, 2005).

Research on IOs and democracy also emphasizes elections. Approaches to international democracy promotion developed according to the transitions paradigm, which assumes any state moving away from authoritarianism is transitioning toward democracy, stresses the importance of elections as the first step toward eventual democratic consolidation, and largely disregards the impact of structural and institutional factors on a state’s democratic prospects (O’Donnell & Schmitter, 1986). As such, democracy promotion heavily emphasizes free and fair elections, viewing these as a critical prerequisite for long-term democratic success (Hadenius & Teorell, 2007; Howard & Roessler, 2006; Schedler, 2002), yet provides much less support for other institutions, such as political parties (Carothers, 2002).

This electoral focus has driven IOs to promote elections even in countries that lack other democratic institutions; electoral assistance is the most common type of democracy assistance from IOs. Initially, this assistance centered primarily on election monitoring, but it has more recently expanded to also include an emphasis on improving election laws, updating voter registries, and assisting with other electoral logistics (Donno, 2013; Johnstone & Snyder, 2016; Poast & Urpelainen, 2018).

The emphasis placed on executives and elections is evidenced by IOs’ democracy promoting strategies. For example, the OAS is largely limited to election monitoring and training and consultation for new democratic leaders (Cooper & Legler, 2006). Similarly, EU democracy promotion in post–communist central and eastern Europe favors executives and their bureaucracies over other democratic institutions and is undertaken in a top-down manner without support from political groups or civil society (Bugaric, 2015; Grabbe, 2001; Pridham, 2007).

Although elections and executive compliance with democratic institutions are important (Dahl, 1971; Huntington, 1991; Przeworski et al., 2000), they are insufficient to guarantee long-term democratic quality and survival. Election monitoring’s limits have been identified (Hyde, 2011; Simpser & Donno, 2012), and elections alone are insufficient to promote ongoing democratic progress (Flores & Nooruddin, 2016). Research finds organizations for managing mass participation in politics and performing key representative functions are equally, if not more important (Almond & Verba, 1963; Grzymala-Busse, 2007; Herman, 2015; Mainwaring, 1999; O’Donnell & Schmitter, 1986). Such institutions include political parties, which aggregate mass interests and serve as intermediaries between citizens and the state (Carothers, 2006; Huntington, 1968), and civil society organizations. Horizontal checks on government power and institutions that constrain elected officials’ power are also critical. This includes independent judiciaries, which can block executives seeking to increase their power (Gibler & Randazzo, 2011), and intragovernmental checks and balances, such as legislatures (Graham et al., 2017). IOs devote much less attention to these other critical institutions.

To summarize, IOs associated with democracy unintentionally make democratic backsliding more likely in new democracies by focusing on elites and elections at the expense of other democratic institutions. This makes new democracies more likely to experience negative changes in their levels of democracy.

IOs also more directly contribute to backsliding in two interrelated ways: by increasing relative executive power and constraining the domestic policy space, which further impedes democratic institutional development.

Increasing Executive Power

IOs promote democracy by altering elites’ incentives to comply with or maintain democratic institutions (Donno, 2013; Pevehouse, 2005; Poast & Urpelainen, 2018), or by providing technical assistance to leaders in new democracies (Poast & Urpelainen, 2018). Authors that approach this topic primarily from the domestic point of view also emphasize the importance of domestic elites, and especially executives, as the critical intermediaries or “gatekeepers” between international actors and states (Moravcsik, 1994; Tolstrup, 2013; Yilmaz, 2002). As such, IO membership can increase executive power relative to other domestic institutions.

I argue IOs provide both incentives and opportunities for executives, who are rational actors concerned with maintaining office, to manipulate institutions to their advantage; this in turn makes backsliding more likely. The intense emphasis international actors place on free and fair elections has rendered elections a condition for receiving international material benefits (Hyde, 2011). Therefore, executives have incentives to refrain from manipulating elections to maintain power; instead, they resort to more subtle tactics, such as “strategically manipulating” elections by interfering with nonelectoral institutions (Beaulieu & Hyde, 2009).

IOs facilitate this by increasing executives’ relative domestic power in several ways. First, IOs provide democracy-related financial assistance to states. However, this financial assistance, which is often received, controlled, and allocated by executives, increases government power over economic activity, which is found to be problematic for democracy (Friedman, 1958). Foreign aid also weakens governmental accountability and stunts civil society development. Democracy first developed when elites demanded increased rights and government accountability in return for taxes (North, 1990; Tilly, 1990). However, aid from international sources undermines this accountability mechanism, making rulers more responsive to international donors than to citizens (Brautigam, 1992; Karl, 1997; Moore, 1998). Executive control over incoming financial resources also increases recipient governments’ ability to use patronage to maintain power and garner electoral support (Ahmed, 2012; Jablonski, 2014), and has been linked to increased corruption (Knack, 2001).

Second, IO membership redistributes power between domestic governmental institutions, often to the executive’s advantage. As IO members, executives either represent their state within the organization or they have the power to directly appoint representatives. As a result, international cooperation increases executive power over domestic policy and allows them to influence or even ratify international decisions; this is often done “with relatively little ministerial, legislative, judicial, or public oversight” (Moravcsik, 1994, p. 7). The financial aid states receive from IOs further contributes to this power asymmetry. Foreign aid provides resources outside the national budget, allowing executives to pursue projects without legislative approval (Brautigam, 2000), which shifts “the balance of power within countries in ways that distort the constitutionally established system of checks and balances” (quoted in Knack, 2004, p. 253).

Finally, in certain IOs, membership conditionality requires countries to develop an extensive bureaucracy to facilitate preparations for membership. To this end, IOs offer technical assistance to states for constructing regulatory, economic, and other bureaucratic offices. A strong bureaucracy and state capable of providing public goods is important for democracy (Fortin, 2012; Grzymala-Busse, 2007). However, bureaucracies are unaccountable to voters and are often closely linked to, and thus serve as an additional resource for executives.

The more IOs a state is in, the more opportunities executives have to receive financial resources and make international-level decisions. However, the extent to which IO membership increases executive power depends on the IO’s type. While minimalist IOs have limited effects on domestic processes, structured and interventionist ones can alter state-level outcomes, and are more likely to have institutions that increase executives’ international decision-making powers. As such, membership in more institutionalized IOs is particularly likely to increase executive power by creating venues for executives to influence domestic policies. This is especially the case in political and economic IOs, which focus on substantive policy issues.

Although strong executives are not incompatible with democracy, they are problematic when institutional checks are underdeveloped. Below I discuss how the limits IOs place on states’ domestic policy options inadvertently stunt institutional development.

Limiting the Domestic Policy Space

The domestic policy space is the universe of policy alternatives political actors can feasibly debate, adopt varying positions on, and implement. A meaningful policy space is critical for institutional development, conditions the extent to which leaders can govern effectively, and influences the strategies politicians adopt to appeal to voters. In states with large policy spaces, institutions that aggregate mass interests and check executive power develop more fully, governments are better able to provide public goods, and, therefore, elections are contested along ideological lines. However, when the policy space is limited, institutions remain underdeveloped, and elections occur in an ideological void.

Scholars have studied how economic interdependence affects state policies, finding increased globalization, influence from international financial markets, and transnational capital induce economic policy convergence both across countries (Andrews, 1994; Cerny, 1994; Mosley, 2000; Pierson, 1995; Rodrik, 1997; F. Ross, 2000; Simmons & Elkins, 2004) and across parties within a country (Mishra, 1999). International financial pressures, which favor lower taxes, less economic regulation, and smaller governments, limit states’ abilities to determine domestic economic policy (Haggard & Maxfield, 1996).

IOs also often specify policy requirements for prospective and current member states. These conditions include policy changes and economic or monetary reforms (Pevehouse, 2005); the upper limit the EU imposes on members’ budget deficits is a salient example (Mosley, 2000). In addition to inducing cross-national economic policy convergence (Cao, 2009), IO membership has been shown to influence national science (Finnemore, 1993), mass education (Meyer et al., 1992), and environmental policies (Frank, 1997).

The domestic policy space is critical for developing representative institutions and institutional checks on executive power in new democracies. First, an expansive policy space creates opportunities for parties to develop. In developed systems, parties compete by situating themselves along salient, politicized cleavages that give voters options based on policy-relevant issues (Lipset & Rokkan, 1967). However, IO requirements limit policy alternatives for political parties to adopt and debate since advocating policies that conflict with IO requirements is not electorally viable. The inability to debate substantive policy and ideological alternatives stunts party system development in new democracies because these parties cannot develop coherent and distinct policy platforms to appeal to voters. This, in turn, impedes democratic quality and consolidation, which depends in part on the extent to which parties structure political conflict (Dix, 1992; Mainwaring, 1998).

Scholars find economic globalization has limited the range of economic policy appeals political actors can make. As voters are aware that governments’ economic options are constrained (Hellwig, 2014), it is difficult for parties to differentiate themselves based on economic appeals. Instead, they turn to populism or campaign on valence issues, or uncontroversial issues “on which all parties declare the same objective but dispute each other’s competence in achieving the desired policy” (Kitschelt et al., 1999, p. 137), such as support for anti-corruption, nationalism, or rule of law. This is true for all types of parties, not just new or niche ones (Ward et al., 2015).

The IMF’s impact on party development in Latin America illustrates these issues. In response to the economic recession in the 1970s and 1980s, many Latin American governments borrowed heavily from international financial actors, inducing a decade-long debt crisis. Unable to repay their debts, governments turned to the IMF for loans. The IMF heavily conditioned these loans, requiring states to cut regulatory, redistributive, and social welfare policies. These neoliberal reforms included labor market liberalization, which induced employment insecurity, as well as anti-inflationary and other austerity programs that decreased public spending. Ironically, historically labor-based parties were required to implement these reforms. As a result, voters became increasingly detached from political parties, creating opportunities for individual politicians to appeal directly to voters without parties as intermediaries. In other words, IMF conditionality severely weakened party systems in Latin America, which created a space for populism to emerge as a political alternative (Pastor, 1987; Roberts, 2007). Intense austerity policies imposed by the EU and the IMF in Greece more recently have had similar effects (Vasilopoulou et al., 2014).

External policy requirements also infringe on one of the primary roles of legislatures: proposing, drafting, and implementing legislation. This stunts the institutionalization of the legislature in new democracies and contributes to power asymmetries between the executive and the legislature, which further hinders democratic progress, in two ways. First, strong legislatures create an impetus for party system development. In states where legislatures are weak, parties lack a forum in which to develop and mature. Another consequence is that legislatures, which play a critical role in providing horizontal accountability, are unable to check executive power. Indeed, in addition to the judiciary, legislatures are the main institutional check on executive power (Fish, 2006).

Finally, the domestic policy space is linked to state fiscal capacity and, in particular, a state’s ability to govern effectively and provide public goods. State capacity, or a state’s ability to “maintain the rule of law, to protect the rights of citizens, and to regulate economic transactions” (Fortin, 2012, p. 904), is another important condition for long-term democratic success (Bäck & Hadenius, 2008). IOs restrict the economic and fiscal policy options available to states by limiting tariffs, public spending, and government deficits. These fiscal limits, in addition to restricting government control over economic policies, decrease government resources.

A state’s fiscal space, or its ability to collect revenue for public goods provision, affects politicians’ electoral strategies. When states have a large fiscal space, politicians campaign based on evidence of effective public goods provision. However, external policy requirements limit the range of policies governments can credibly advocate and diminish the state’s ability to collect revenue. When politicians are unable to use ideological positions and evidence of past economic success to appeal to voters, they turn instead to populism or clientelism; this style of politics impedes democratic progress in young democracies (Flores & Nooruddin, 2016).

The more IOs with policy requirements a state is a member of, the more policy constraints it faces. These policy requirements limit states’ policy space, which stunts institutional development and diminishes the government’s capacity to provide public goods; one consequence is that elections are contested in an ideological void. These effects are particularly pernicious in new democracies where institutions are often underdeveloped and have been neglected by democracy promoting IOs, and also where executives are already relatively strong due to IO membership. I argue this combination of strong executives surrounded by weak institutions makes democratic backsliding more likely.

To summarize, I predict increased membership in IOs relevant to democracy is associated with democratic backsliding or negative changes in levels of democracy among new democracies. The EU provides a useful illustration. The EU is a democratically committed, political and economic, interventionist IO widely viewed as unrivaled in its ability to promote democracy (Dimitrova & Pridham, 2004). Nevertheless, some of its newly democratic members, including Hungary, Poland, and the Czech Republic, have shown evidence of backsliding. To join the EU, these states had to meet a series of democratic criteria and undergo a lengthy accession process. This process favored executives and the bureaucracy over other domestic institutions (Follesdal & Hix, 2006; Grabbe, 2001), and created institutions from above without support from political groups or civil society (Bugaric, 2015). EU membership also increases domestic executive power by giving executives supranational decision-making power in the European Council (Tallberg, 2008). The EU also imposes significant policy constraints on its candidate and member states. For example, the EU’s direct influence over monetary policies gives it tremendous influence over national policies regarding budgetary issues (Katsanidou & Otjes, 2015). I argue this combination of factors exacerbated by the EU—strong executives surrounded by weak democratic institutions at the domestic level—has contributed to backsliding in its member states.

Dependent Variable

I use the Liberal Democracy index from V-Dem to construct the dependent variable. I depart from existing work that focuses on the relationship between IOs and levels of democracy and consider instead the changes in levels of democracy. This choice is driven by my theory: I predict IOs make subsequent backsliding, which is conceptually a decrease in a state’s level of democracy, more likely. I do not argue IOs are negatively associated with overall levels of democracy; more democratic states are undoubtedly more likely to be in the IOs I consider. Rather, I argue these IOs make new democracies more susceptible to democratic erosion.

As backsliding is an incremental process (Waldner & Lust, 2018), 1-year changes are unlikely to capture this phenomenon. Therefore, the dependent variable is the 5-year change in a state’s democracy index. A negative value on the dependent variable indicates a state has experienced backsliding between times t – 5 and t, while positive values indicate democratic progress.

Independent Variables

The main independent variables are counts of state membership in certain types of IOs. Broadly, I focus on three groups of IOs linked to democratic outcomes: democratically committed, political and economic, and structured and interventionist IOs.

Democratically committed IOs “reference democracy, human rights, or the rule of law in their constitutive documents” (von Borzyskowski & Vabulas, 2018, p. 5); there are 56 IOs that meet these criteria. ² Because democratically committed IOs with near or entirely universal membership are less likely to actively require member states to maintain democratic standards, I create a second group that includes only democratically committed IOs with non-universal membership. ³

Next, political and economic organizations are theorized to promote democratic consolidation by providing newly democratic leaders with technical assistance for public goods provision (Poast & Urpelainen, 2015). Poast and Urpelainen (2015) create a classification of 494 organizations that expands Ingram et al.’s (2005) categorization of IO functionality from four groups to six mutually exclusive groups—Political, General-Economic, Commodity-Economic, Resource-Economic, Technical, and Unclassified—and covers the years 1946 to 2000. Using Version 3.0 of the IO membership dataset (Pevehouse et al., 2016), I use Poast and Urpelainen’s (2015) coding strategy to categorize the functionality of democratically committed IOs created after 2000.

Finally, only certain types of IOs influence state behavior. Therefore, I consider structured and interventionist IOs, which are particularly likely to affect domestic outcomes. To identify these, I use Boehmer et al.’s (2004) three-level coding scheme. ⁴

These groups are not mutually exclusive. Some democratically committed IOs, such as the European Bank for Reconstruction and Development, are interventionist, while others, such as the Francophone Agency, are minimalist. Similarly, certain political or economic organizations, such as the International Labor Organization, are explicitly committed to democracy, whereas others, such as the IMF, are not. Table 1 provides additional examples to summarize these groups.

OPEN IN VIEWER

Table 1.

Categories of International Organizations.

International Organization	Democratic	Political or economic	Structured	Interventionist
ECOWAS	✓	✓
Council of Europe	✓	✓	✓
Mercosur	✓	✓	✓
OAS	✓	✓	✓
IMF		✓		✓
World Bank		✓		✓
EBRD	✓	✓		✓
EU	✓	✓		✓

OAS = Organization of American States; IMF = International Monetary Fund; EU = European Union; ECOWAS = Economic Community of West African States; EBRD = European Bank for Reconstruction and Development.

IO membership entails ongoing policy commitments states must meet and opportunities for executives to increase their power; these commitments and opportunities continue beyond the accession date. Therefore, the main independent variables are states’ membership counts in different types of IOs. I create a series of variables measuring the number of democratically committed IOs with universal membership, democratically committed IOs with non-universal membership, democratically committed political or economic IOs, political or economic IOs, and structured or interventionist IOs a state is a member of each year. ⁵ As the dependent variable is the 5-year change in a state’s democracy level, I lag these counts by 5 years. Therefore, an observation consists of the number of IOs a state was a member of 5 years prior, and the subsequent 5-year change in its democracy level.

Control Variables

I control for state and regional characteristics relevant to democratic outcomes, including GDP (gross domestic product) per capita and GDP growth per capita (Feenstra et al., 2015), because development and democracy are highly correlated (Lipset, 1959). I also include a measure of the effective number of political parties (Cruz et al., 2016), predicting higher levels will be associated with democratic progress (Mainwaring & Scully, 1995). I control for state age, logged to account for diminishing returns (Correlates of War Project, 2016), and the number of years since a state transitioned to democracy to ensure that the number of IO memberships is not simply a proxy for time as a democracy (Boix et al., 2012). To control for diffusion effects, I include the average change in a region’s liberal democracy scores over the relevant time period (K. S. Gleditsch & Ward, 2006; Simmons et al., 2006). ⁶

Drawing on theories of democracy, I include an indicator for internal conflict, which can negatively affect democracy (Hegre, 2014). ⁷ Although the relationship between oil and democracy is debated (Brooks & Kurtz, 2016; M. L. Ross, 2015), research suggests the two are related; therefore, I control for a country’s annual oil income per capita (M. L. Ross & Mahdavi, 2015). I also control for ethnic fractionalization, which is often linked to regime outcomes (Alesina et al., 2003). ⁸

As the dependent variable is the change in a state’s level of democracy, I account for a state’s democracy level at the beginning of the observation. Any observed relationship between IOs and subsequent backsliding may result from states artificially increasing their levels of democracy prior to joining an IO, and then regressing to the mean following accession. Alternatively, states struggling with democracy may join additional IOs to safeguard democracy, yet continue backsliding. Therefore, I include a variable measuring changes in the liberal democracy index in the 5 years prior to an observation.

Each independent variable is lagged by 5 years to guard against endogeneity, and all observations are at the country-year level and include data on new democracies from 1946 to 2015. ⁹ All variables are standardized to facilitate interpretation.

Robustness Checks

These results are robust to alternative specifications. First, since the time over which backsliding occurs can vary, I estimate models that consider the 3-, 7-, and 10-year changes in a state’s level of democracy, finding similar results. ¹¹

The main models include observations from the Cold War period to capture the third wave of democratization. However, democracy promotion expanded significantly after the end of the Cold War, and the third wave of autocratization began around 1993 (Luhrmann & Lindberg, 2019). As such, the theory is particularly relevant to the post–Cold War era. I account for this in several ways. First, the fixed effects models I estimate ¹² include fixed effects by year. I also re-estimate the main models on two subsets of the data: one for observations from 1946 to 1989, and another for 1990 to 2015. ¹³ Some but not all of the coefficients on the main independent variables in the 1945–1989 models are statistically significant. Although this is likely due in part to missing data, the dampened results are unsurprising because international democracy promotion became much more common after the Cold War. This is reflected in the results for the 1990–2015 models, which are consistent with those from Table 2.

Globalization and increased international interactions in general, rather than IOs specifically, may contribute to backsliding. I test for this by controlling for two proxies for globalization: trade openness and foreign direct investment inflow (World Bank, 2017). I also subset the data to just consider the post–Cold War period, which marks a significant shift in the extent to which international actors influence domestic democratic outcomes, and find similar results. ¹⁴

Although my focus is on backsliding in democracies specifically, anocracies, or states between democracy and autocracy, may experience similar processes. Therefore, I re-estimate the main models using data on both developing democracies and anocracies, defined as country-years with a Polity score between –5 and 5. The results of these models are similar to those in Table 2. ¹⁵

There is also the possibility of reverse causality: Are certain new democracies joining IOs to guard against future democratic erosion, but backsliding nonetheless? I contend that this is unlikely. First, research finds that all new democracies join IOs at high rates (Mansfield & Pevehouse, 2008). Furthermore, it is unlikely that current executives, who strongly influence whether or not a state joins an IO, can accurately anticipate if future executives will induce backsliding. Nevertheless, to test for this and any other selection bias, I estimate two-stage selection models (Heckman, 1979) and still find evidence linking democratically committed and political and economic IOs to backsliding. ¹⁶

Finally, there may be an unobservable third variable predicting both IO membership and backsliding. An ideal experiment would include cases that received treatment (IO membership) and ones that are similar on all dimensions except for treatment status. Unfortunately, this scenario is not observable: All new democracies are members of IOs associated with democracy. Therefore, I use the generalized synthetic control (GSC) method. This method combines the synthetic control method (Abadie et al., 2015) and interactive fixed effects models (Bai, 2009), both of which have been used to estimate average treatment effects in time-series cross-sectional data. The GSC method uses pre-treatment characteristics of treated units to construct synthetic counterfactuals and then estimates an average treatment effect. The GSC method allows for heterogeneous treatment effects across units, and accounts for variations in treatment time (Xu, 2017).

As virtually all new democracies are IO members, it is difficult to identify a binary treatment for groups of IOs. Therefore, I focus on a specific organization: the EU. The EU is a democratically committed, political and economic, and interventionist IO, and is widely viewed as unparalleled in its democracy promoting abilities; therefore, it is a particularly hard test for the theory. Using the GSC method, I find EU accession has a negative and significant treatment effect on states’ democratic trajectories. ¹⁷ Although encouraging, this test raises an additional question: Are organizations like the EU driving the results in Table 2? Although this is unlikely because the main models include varying intercepts for region, I re-estimate the main models, excluding the EU altogether, and find similar results. ¹⁸

Executive Power

I measure executive power using the horizontal accountability index from the V-Dem dataset. This variable measures institutional checks that prevent abuses of executive power, including the legislature, judiciary, ombudsmen, and prosecutors and comptrollers general (Coppedge et al., 2018). Lower values indicate less institutional constraints on the executive, or higher relative executive power. Because I am interested in how IOs change executive power in democracies, the dependent variable is the 5-year change along this index. My theory predicts increased IO membership corresponds with a negative change along this index, or increased executive power.

IOs increase executive power by giving executives control over domestic outcomes decided within an IO. International opportunities for executives to make decisions unilaterally are most likely within IOs that are sufficiently institutionalized to influence domestic policy, and these opportunities increase as states join more organizations. Therefore, I estimate models where the main independent variables indicate the number of structured or interventionist IOs a state was a member of 5 years prior. In addition, I consider political and economic IOs, which are more likely to make decisions that affect substantive domestic policy outcomes. Finally, I combine these two characteristics and consider structured or interventionist, economic or political organizations.

I account for domestic institutional characteristics that influence executive power. I control for the number of years the chief executive has been in office, whether the executive’s party has an absolute majority in houses with lawmaking powers (Cruz et al., 2016), and a state’s level of executive power at t – 5 using the horizontal accountability index. I also include an indicator for whether or not a state was involved in either an internal or external conflict, as executives often use conflict as a justification for increasing their own power (Allansson et al., 2017; N. P. Gleditsch et al., 2002). Next, I control for oil income per capita. One of the proposed mechanisms linking natural resources to autocracy is increased incentives for politicians to dismantle state institutions to extract rents; as such, oil income may be associated with less institutional checks on the executive (M. L. Ross, 2015; M. L. Ross & Mahdavi, 2015). I also control for GDP per capita and GDP growth per capita (Feenstra et al., 2015), a state’s democracy level (Coppedge et al., 2018), and state age (Correlates of War Project, 2016). Finally, I include the lagged dependent variable as a predictor. ¹⁹ The universe of cases is new democracies from 1946 to 2015, and I estimate multilevel models with varying intercepts by country and region; the results are reported in Table 3.

OPEN IN VIEWER

Table 3.

Number of IO Memberships and Checks on Executive Power.

Independent variable	Struc/Interven IO	Pol/Econ IO	Pol/Econ, Struc/Interven IO
Num. IO memberships 5 years prior	−0.16 ** (0.08)	−0.16 *** (0.05)	−0.18 *** (0.05)
Exec. yrs. in office	0.11 *** (0.02)	0.11 *** (0.02)	0.12 *** (0.02)
Exec. controls legis.	0.09 *** (0.02)	0.09 *** (0.02)	0.09 *** (0.02)
Govt. fractionalization	0.01(0.02)	0.01(0.02)	0.01(0.02)
Horizontal account. index	−1.84 *** (0.07)	−1.83 *** (0.07)	−1.84 *** (0.07)
GDP pc (log)	0.54 *** (0.07)	0.58 *** (0.07)	0.59 *** (0.07)
GDP growth pc	0.02(0.02)	0.02(0.02)	0.01(0.02)
Lib. dem. index	−0.20 *** (0.06)	−0.20 *** (0.06)	−0.19 *** (0.06)
State age (log)	0.02(0.05)	0.02(0.05)	0.02(0.05)
Oil income pc	0.37(0.23)	0.30(0.23)	0.31(0.23)
Conflict (any)	0.02(0.02)	0.01(0.02)	0.01(0.02)
Lagged DV	−0.01(0.01)	−0.01(0.01)	−0.01(0.01)
AIC	2,466.57	2,462.66	2,460.92
BIC	2,552.31	2,548.40	2,546.66
Log. lik.	−1,217.28	−1,215.33	−1,214.46
Observations	1,570	1,570	1,570
Num. countries	83	83	83
Num. regions	17	17	17
Variance: Countries	1.05	1.00	1.00
Variance: Regions	0.05	0.05	0.06
Variance: Residual	0.21	0.21	0.21

Dependent variable: 5-year change in Horizontal Accountability index. IO = international organization; GDP = gross domestic product; DV = dependent variable; AIC = Akaike information criterion; BIC = Bayesian information criterion.

**

p < .01. ***p < .001.

I find support for Hypothesis 2: IO membership is linked to subsequent decreased checks on executive power. These results are robust to models that control for globalization and ones that use 3-, 7-, and 10-year changes in executive power as the outcome. As these are linear OLS models, the coefficient on the main independent variable is the movement along the horizontal accountability index over 5 years associated with joining one additional IO.

Domestic Policy Space

The domestic policy space is the range of possible policies political actors can adopt and implement. As we only observe implemented policies, it is difficult to directly measure the policy space; therefore, I rely on an observable implication. In mature democracies, parties distinguish themselves by taking distinct policy positions (Lipset & Rokkan, 1967). However, this is more difficult when the policy space is constrained. Instead in these cases, parties define themselves in terms of populism or broad-based appeals, such as support for anti-corruption, nationalism, or rule of law (Kitschelt et al., 1999; Ward et al., 2015). Thus, one observable implication of a limited policy space is ideological convergence across parties.

The dependent variable is the 5-year change in the distinctness of party platforms in a country, which captures the extent to which publicly available party manifestos are distinct in terms of content or ideology across parties within the same country. In other words, it measures ideological convergence across parties within a state. This variable is coded on a 5-point scale, where 0 indicates few, if any, parties have platforms and/or none have platforms distinct from other parties, and 5 indicates all or nearly all have publicly available and distinct manifestos (Coppedge et al., 2018). I argue higher values indicate states where parties have a more expansive domestic policy space. Therefore, a negative change along this index indicates convergence across parties over time, or a diminished policy space.

For the main independent variables, I consider structured and interventionist IOs, which are especially likely to require and have the tools to enforce policy changes within their member states, and political and economic organizations, which are more likely to impose requirements with substantive policy implications for their member states. I also combine these two and consider structured or interventionist, political and economic IOs.

I control for factors likely to influence party platforms and ideological development, including GDP and GDP growth per capita (Feenstra et al., 2015), a state’s electoral system type and government structure (Cruz et al., 2016), urban population as a percentage of the total population (World Bank, 2017), state age (Correlates of War Project, 2016), and democracy level (Coppedge et al., 2018). I also include the party distinctness variable and the lagged dependent variable as controls. ²⁰ I estimate multilevel models with varying intercepts by country and region, focusing on new democracies between 1946 and 2015. The results can be found in Table 4.

OPEN IN VIEWER

Table 4.

Number of IO Memberships and Domestic Policy Space, Distinct Party Platforms.

Independent variable	Struc/Interven IO	Pol/Econ IO	Pol/Econ, Struc/Interven IO
Num. IO memberships	−0.15 ** (0.07)	−0.19 *** (0.05)	−0.03(0.05)
GDP pc (logged)	0.42 *** (0.07)	0.47 *** (0.07)	0.40 *** (0.07)
GDP growth pc	0.05 *** (0.02)	0.05 *** (0.02)	0.05 *** (0.02)
Lib. dem. index	−0.05(0.03)	−0.03(0.03)	−0.07 ** (0.03)
Electoral system type	0.19 *** (0.04)	0.19 *** (0.04)	0.19 *** (0.04)
Pres. vs. parliamentary	0.02(0.04)	0.02(0.04)	0.01(0.04)
Urban. level	0.21 ** (0.09)	0.28 *** (0.09)	0.18 ** (0.08)
Distinct party platforms	−2.52 *** (0.05)	−2.53 *** (0.05)	−2.52 *** (0.05)
State age (log)	0.08 * (0.05)	0.10 ** (0.05)	0.06(0.05)
Lagged DV	0.03 ** (0.01)	0.03 ** (0.01)	0.04 *** (0.01)
AIC	3,021.72	3,011.62	3,026.91
BIC	3,098.67	3,088.57	3,103.86
Log. lik.	−1,496.86	−1,491.81	−1,499.46
Observations	1,802	1,802	1,802
Num. countries	85	85	85
Num. regions	17	17	17
Variance: Countries	1.23	1.19	1.24
Variance: Regions	0.84	0.69	0.85
Variance: Residual	0.24	0.24	0.24

Dependent variable: 5-year change in distinctness of party platforms. IO = international organization; GDP = gross domestic product; DV = dependent variable; AIC = Akaike information criterion; BIC = Bayesian information criterion.

*

p < .05. **p < .01. ***p < .001.

I find evidence that IOs limit states’ domestic policy space: Membership in structured or interventionist and political or economic IOs is associated with subsequent decreases in the dependent variable, or party platform convergence. As with previous models, the coefficients can be interpreted as changes in the distinctness of party platforms for each additional IO membership. This suggests these externally imposed policies are meaningfully limiting the domestic policy space and restricting the ideological options available to political parties in emerging democracies.