2.1 IMF decision-making
International organizations like the IMF can be considered as sets of “nested principal-agent relationships” (Nielson and Tierney
2003, p. 250). From this perspective, the IMF is part of a delegation chain starting with voters in member countries, the ‘ultimate principal’ (see also Vaubel
2006). The chain runs via national parliaments, governments, their representatives in the IMF’s executive board, and ends with the IMF’s staff. There are two main reasons for why IMF decisions may reflect agent preferences that diverge from the preferences of voters in member countries that are affected by these decisions.
First, the governments of major shareholders have substantially more influence on the IMF than the governments of the countries that usually receive IMF loan programs. Empirical evidence for the disproportional influence of the US and other “G5” governments abounds (for reviews of this literature see Dreher and Lang
2019 and Vreeland
2019). The delegates of these governments have the largest formal voting power in the Executive Board, but even beyond formal votes they have a considerable impact on IMF policies through so-called “informal governance” (Stone
2008). Various channels of influence allow the US and other G5 governments to influence IMF decision-making in a way that it reflects their political (e.g., Dreher et al.
2018), geostrategic (e.g., Reynaud and Vauday
2009), and economic (e.g., Copelovitch
2010) interests. The governments of the countries that receive most IMF programs, on the other hand, tend to lack significant formal voting power, individual representatives in the Board, and substantial informal channels of influence (Kaja and Werker
2010).
Second, it is well documented that the IMF’s policy decisions also reflect the particular interests of its staff. Due to high costs of information and control, and the ability of agents to exploit preference heterogeneity among multiple and collective principals, there is substantial ‘agency slack’ in international organizations like the IMF (Copelovitch
2010; Hawkins et al.
2006; Nielson and Tierney
2003; Vaubel
2006). This increases the ability of staff to pursue their own interests. Multiple studies observe IMF behavior that reflects staff interests like maximizing budgets, responsibilities, and autonomy, and find that IMF officials are able to push for longer programs, larger loans and more far-reaching conditionality than what is economically optimal (Barnett and Finnemore
2004; Copelovitch
2010; Lang and Presbitero
2018; Vaubel
2006). A second strand of this research shows that staff’s ideological beliefs and policy preferences are also reflected in the IMF’s policy decisions (Barro and Lee
2005; Chwieroth
2007a; Nelson
2014). These studies, inter alia, identify links between staff preferences for market-liberal policies and corresponding reforms in program countries.
In sum, major shareholder governments exploit their influence on the IMF to further their own political and economic interests, while staff shape the IMF’s policy decisions in accordance with their material interests and ideological preferences. As will be discussed next, these preferences are often unlikely to align with preferences of voters in program countries when it comes to policy reforms with distributional implications.
2.2 Divergent priorities and distributional implications
Which policy preferences of major IMF shareholders and IMF staff can have distributional consequences in program countries? The existing literature suggests that the Fund’s major shareholders have an economic interest in guarantees of debt repayments and cuts of public spending in program countries as this helps prevent financial losses for creditors from their country (e.g., Copelovitch
2010; Gould
2003).
2 Furthermore, to increase trade with and opportunities for investments in these countries they also have an interest in other countries liberalizing their trade and financial policies (Woods
2006).
3 In addition, multinational firms based in major shareholder countries have a commercial interest in less regulated labor markets, lower taxes, and privatizations in developing countries to produce more cheaply. Major shareholder governments will represent these interests if lobbied or convinced of beneficial effects for their economies. Consistent with this argument, large shareholder governments have been shown to influence the World Bank in accordance with commercial interests of multinational firms based in their countries (Dreher et al.
2019; Malik and Stone
2017). Similarly, IMF programs were found to be associated with subsequently rising flows of foreign direct investment from the United States and have been shown to benefit US commercial banks (Biglaiser and DeRouen
2010; Gould
2006).
For the IMF bureaucracy, the gradual expansion of the scope of IMF conditionality into policy areas where reforms are more ‘structural’ has often been linked to the bureaucratic incentive to expand the organization’s mission (Barnett and Finnemore
2004; Dreher and Lang
2019; Kentikelenis et al.
2016; Reinhart and Trebesch
2016). In policy areas like labor-market regulation IMF conditions go beyond setting quantitative benchmarks and instead include structural reforms (Reinsberg et al.
2019). This gives IMF staff more direct and detailed influence on policies (Babb and Buira
2005; Kentikelenis and Babb
2019). It is consistent with this explanation that IMF staff also played an important role in strengthening the IMF’s focus on reforms in the area of social policy (Vetterlein and Moschella
2013). Furthermore, scholars have identified a strong tendency among IMF staff to favor market-liberal policies over government intervention in market processes and outcomes. As the IMF’s internal structure, hiring patterns, and organizational culture are typically described as stable, hierarchical, and monolithic (Momani
2005), scholars consider the market-liberal ideological preferences of its staff as highly stable over time (Chwieroth
2007a; Nelson
2014). As a result, there is substantial evidence suggesting that policies stipulating reduced public spending as well as trade and financial liberalization are associated to these ideological preferences of IMF staff (Barnett and Finnemore
1999; Chwieroth
2007a; Nelson
2014). It is furthermore worthwhile to add that preferences of IMF staff and major shareholders are not independent of each other. There is evidence suggesting that the United States played an important role in shaping the political orientation of the IMF bureaucracy (Kentikelenis and Babb
2019; Momani
2004).
In line with these arguments, studies show that IMF conditionality reflects these preferences and find conditions in these three areas –
cuts of public spending,
trade and financial liberalization,
labor-market reforms – to be frequently included. According to Stone (
2008, p. 600) “there is almost always some limit on public debt or government spending.” According to Kentikelenis et al. (
2016), more than 70% of programs include conditions on trade and financial liberalization and about 50% set labor-market conditions.
But do these conditions lead to reforms in program countries? Even though not all IMF-mandated reforms are complied with, program countries implement many of the IMF’s conditions and their impact is measurable (Rickard and Caraway
2019; Stubbs et al.
2020). Through the threat to withhold loan disbursements, IMF conditionality rises costs for domestic political actors, e.g., parliaments, to block reforms under a program. As disbursement are in practice often withheld (Dreher
2006), this threat is credible. Furthermore, unpopular reforms become more likely as governments can use the IMF as a “scapegoat” and can “dilute accountability by blaming IMF conditionality” (Smith and Vreeland
2006). The IMF itself states that conditionality helps “strengthen […] the hand of reformers” (IMF
2007, p. 8).
Empirically, IMF programs are indeed associated with policy reforms in the mentioned policy areas. They were found to come along with trade and capital account liberalization, (e.g., Chwieroth
2007b; Mukherjee and Singer
2010), cuts in the social sector and public wages (Kentikelenis et al.
2015; Nooruddin and Simmons
2006; Rickard and Caraway
2019; Stubbs et al.
2017), and less-regulated labor markets (Blanton et al.
2015; Caraway et al.
2012; Lee and Woo
2020). The latter includes minimum wage reductions, dismissals in the public sector, pension cuts, the legalization of nonpermanent labor, and the privatization of state-owned enterprises.
Which distributional effects should be expected from these reforms? The remainder of this paper examines the hypothesis that IMF programs
increase inequality. This expectation is in line with existing evidence on the type of reforms that the IMF supports in the three discussed areas of public spending, trade and financial liberalization, and labor-market regulation: To the extent that IMF programs reduce public spending they can increase inequality by reducing the extent of redistribution and by affecting the distribution of gross income. Pension cuts or freezes, which are frequently included in IMF programs, may also increase inequality. Cuts in public wages could both increase and reduce gross inequality, depending on employment effects and the relation between public wages and median income. As regards the liberalization of trade and the capital account, most recent studies find inequality-increasing effects for capital account openness, FDI inflows, and composite measures of financial liberalization (de Haan and Sturm
2017; Furceri and Loungani
2018; Lang and Mendes Tavares
2018). The evidence on the effect of trade also points to inequality-increasing effects for many countries (e.g., Antràs et al.
2017; Autor et al.
2014; Goldberg and Pavcnik
2007). More generally, the fact that IMF programs restrict government expenditure during periods of economic liberalization limits the opportunities to ‘embed liberalism.’ As IMF conditionality often combines liberalization and austerity, vulnerable segments of society may lack the “compensations” for distributional risks that result from increasing openness (Rodrik
1998; Walter
2010)
. Typical IMF labor conditions like minimum wage reductions and weakening collective labor rights are, according to the literature, also likely to lead to higher gross inequality (Autor et al.
2016; Kerrissey
2015). Inequality may also rise if layoffs in the public sector and privatizations of state-owned enterprises increase unemployment.
In sum, the implementation of typical IMF conditions concerning social spending, liberalization and labor-market reform runs the risk of increasing inequality. In many countries, these reforms can mean a substantial departure from pre-program policy paths. Compared to a counterfactual scenario without IMF influence on national economic policies, inequality could thus rise if countries enter IMF programs.
The subsequent part of this paper tests the empirical implications of the theoretical argument. At the core of the analysis is the test of the overarching hypothesis that IMF programs increase inequality. In addition to standard measures of inequality, new global data of absolute income growth for different income deciles of affected countries are considered. This allows testing whether inequality rises because the poor lose or because the rich gain in absolute terms. To investigate channels, heterogeneous effects are examined and the links between IMF conditionality and inequality are analyzed.