Controlling coalitions: Social lending at the multilateral development banks

Strand (2003a, b) offers a sophisticated and compelling exception.

These conventional alternatives are rarely specified explicitly. We employ our preference specification and aggregate preference for the great powers model in a way that we believe best replicates the implicit model used by most analysts.

Moravcsik (1998) persuasively argues that “The Commission was successful only when it was pursuing the policies that were emerging from Franco-German discussions” (205). Moravcsik’s “intergovernmental” approach to understanding the EU is a conceptual predecessor to our focus on coalition politics within MDB executive boards. While Moravcsik focuses empirical attention on the power and preferences of the three largest member states of the EU, we demonstrate that accurate predictions of MDB agent behavior require analysis of the power and preferences of all member states.

See Stone’s (2002) explanation for changing Fund policy toward Russia during Clinton Administration. Other IO scholars use similar evidence to explain IO behavior or individual IO decisions. See for example Keohane (1984: chapters 8–10); Thacker (1999); or Barnett (2003).

The United States, which controls more than 15 percent of voting shares, can unilaterally block institutional change and thus provides the closest approximation to an independent principal of the World Bank. Although the United States can block major institutional changes, however, it does not have the authority to unilaterally bring them about. For any policy change or project approval, the United States requires coalition partners representing at least an additional 35 percent of the voting shares.

Interview with Jan Piercy, June 2005, Washington DC.

Interview with anonymous official, U.S. Department of State, March 2009, Washington DC; Interview with Cinnamon Dornsife, Former U.S. Executive Director as the ASDB, Washington DC June 2005.

We confirmed that at least 92 of the 98 projects were approved by searching the online database of the World Bank website. It is entirely possible that the other 6 were also approved, but the documentation for these 6 projects was missing or incomplete. All data on voting history of U.S. EDs from 2004 to the present is published by the US Department of the Treasury Office of International Affairs.

In this paper we assume that formal institutions are efficacious. In previous work one of the authors explains why this assumption is useful and under-utilized for analyzing the politics of IOs (Tierney 2008). However, specifying the general conditions under which this assumption is valid constitutes a critical future step in this research program. For an effort to specify these conditions and to reveal counter-intuitive formal results see Aghion and Tirole (1997) and Frieden (2004).

We realize that such formalism flies in the face of consensus among IO specialists and IR theorists who typically champion the role of norms and informal power within IOs; but the utility of any assumption should be judged by its explanatory power when it confronts actual empirical evidence.

As the former United States ED to the Asian Development Bank, Cinnamon Dornsife, explains, “The borrowing countries always vote as a bloc. In fact, I can only remember a single instance of a borrowing country voting against a project. The Taiwanese ED voted against a project for China…and when this happened, the collective inhaling of breath was audible. It was a rare exception.” Dornsife Interview, Washington DC, June 2005.

In a recent paper Copelovitch (2009) improves upon existing studies by modeling the G-5 as the collective principal of the IMF since these states maintain “de facto control over the Executive Board.” While a focus on the top 5 is theoretically preferable to the hegemonic model so prevalent in the literature, a logical extension of Copelovich’s approach would consider all member states rather than just five.

The other governments in this shared constituency were Afghanistan, Bangladesh, Bhutan, India, and Tajikistan. Like Laos, 61 of the 64 member governments share a Director with other states. Hence, it is unwise to generalize from the United States case where the Director is responsible to only one government.

In most MDBs, including the World Bank, the EDs who represent plural constituencies must vote all their shares in one direction or the other. This leaves EDs in a quandary when their constituencies are split. One might assume that such ED agents simply vote their own preference or that of their home country in such a situation, and they certainly have the authority to do so. However, interviews with participants suggest that EDs from plural constituencies work hard to represent all their member governments. Former United States ED at the World Bank, Jan Piercy, explains why a Dutch national is often chosen to represent their constituency on the board. “They would say, ‘we are not instructed by the authorities at home. We represent all the countries in our constituency.’ And they did.” The point is made even clearer by a veteran ED from New Zealand who has served on both the ASDB and the World Bank Boards. At the ASDB, EDs are allowed to split their votes based on the expressed preferences of different member governments. As John Austin explains, “When I was at the ASDB it was a bit easier. I could play the role of budget hawk when voting the shares of New Zealand or Australia by voting “no,” then I’d turn around and cast “yes” votes for Tonga, Samoa and the rest of the borrowers that I represented. However, at the World Bank I either find a consensus among our members, or else I might have to abstain on the vote.” Interview with John Austin, ED at the World Bank from New Zealand. Washington DC, June 2005. Despite these assurances by participants in coalition politics of the Board, Kaja and Werker (2009) provide systematic evidence that individual EDs do, on average, bring home a “bonus” of loan dollars when they are selected to represent their constituency of governments on the Board.

Interview with former United States ED at a regional development bank. Washington, DC. June 2005.

Interviews with MDB staff, senior managers, and both current and former EDs suggest that coalitional politics on the Bank Boards are very important and can often be intense. The press coverage surrounding former World Bank President Paul Wolfowitz in the spring of 2007 reveals that authority ultimately resides with the Board and that coalition politics on the Board are consequential.

Many of these same recipients were also democratizing their political institutions during this period. Schools, hospitals, and social safety nets are tangible goods that can benefit large numbers of voters. To the extent that citizens are able to hold their leaders accountable, these leaders are likely to pursue goods that are widely available, rather than goods that are narrowly targetable. Interview with Senior World Bank economist, July 2005.

Interviews with senior World Bank staff members, July 2004 and July 2005, Washington, D.C.

We did use this more primitive procedure in an early version of the paper and our statistical results were substantively similar as—though slightly weaker than—those reported below.

See, in particular, Noel and Therien (1995); also, Imbeau (1988); Stokke (1989); Lumsdaine (1993).

Online appendices are available as electronic supplementary material to the online version of this article.

In pooled time-series–cross-sectional regressions using panel-corrected standard errors.

Recall that the SPI was already set with 1996 as a baseline year allowing time-series variance. This temporal variance was preserved in the combined social index.

As discussed above, we developed four measures of states’ preferences for multilateral social lending and for each principal structure (hegemonic, great powers and all member states) we estimated models with each of the four proxy measures of preferences. For reasons of clarity and space, we report the results only for the hybrid measure of preferences. As explained in the previous section, this measure most accurately captures the preferences of both donor and recipient member states.

Both the SPI and combined measures are relative measures, which compare the USA to other countries in 1996. The relative index scores allow the USA’s position to vary over time.

Or if the powerful actors can and do completely ignore the demands of the smaller states.

Unidimensionality assumptions are more problematic if there is extensive trading or logrolling of policies between issue areas, but in the case of MDBs, voting in the Executive Board occurs on discrete loans, so omnibus packages do not occur. While trading may take place at the margins, formal institutions mitigate against it.

For discussion of the analytic point see Laver and Schofield (1990). The literature contains numerous illustrations of the shifting nature of voting coalitions on MDB executive boards. See Rich (1994); Udall (1998); and Upton (2000).

See Laver and Schofield (1990: 98–101).

We developed a program to perform these calculations, written for SAS, which is available online.

These values have been modified slightly and rounded to the nearest one hundredth for presentational purposes.