اقتصاد سیاسی آماده سازی پروژه: تجزیه و تحلیل تجربی از پروژه های بانک جهانی

Using a novel application of stochastic frontier analysis to overcome data limitations, this paper finds substantially shorter project preparation periods for World Bank loans to countries that are geopolitically important (especially to the U.S.). Accelerated preparation is one explanation for how the World Bank might increase the number of loans to a recipient member country within a fixed time frame, for example in response to that country siding with powerful donor countries on important UN votes or while that country occupies an elected seat on the UN Security Council or the World Bank Executive Board. This channel of donor influence has important implications for institutional reform and provides a new angle to examine the cost of favoritism and the impact of project preparation.

Over the past decade, the political economy of international financial institutions (IFIs) has emerged as an important area both for policy and for empirical research. A growing number of empirical studies have reinforced anecdotal reports of the powerful donor countries (notably, the U.S.) intervening to overturn the technocratic decisions of these international organizations. This has been particularly well documented for the IMF where links have been found between access to Fund resources, on the one hand, and UN voting patterns and United Nations Security Council (UNSC) temporary membership, on the other hand (Andersen et al., 2006a, Barro and Lee, 2005, Dreher and Jensen, 2007, Dreher et al., 2009b, Stone, 2002, Stone, 2004, Stone, 2011 and Thacker, 1999). Similar patterns, including links with trade and bilateral aid flows, have also been found for the World Bank (Andersen et al., 2006b, Dreher et al., 2009a, Fleck and Kilby, 2006, Frey and Schneider, 1986, Kilby, 2009b, Kilby, 2013 and Weck-Hannemann and Schneider, 1991) and the Asian Development Bank (Kilby, 2006, Kilby, 2011a and Lim and Vreeland, 2013). This paper is part of a project that builds on this literature to examine donor influence in IFIs at different stages in the resource transfer process. A better understanding of donor influence at each stage is critical to develop a complete picture of how donors impact the efficacy of IFIs. It is also essential for the design of appropriate policy reforms. The present paper focuses on the “upstream” process at the World Bank, the length of time between project identification by World Bank staff and project approval by the World Bank Executive Directors (EDs). This topic is important for a number of reasons. It may elucidate the mechanism by which politics influence the number of projects (Dreher et al., 2009a) or the volume of lending (Andersen et al., 2006a and Kaja and Werker, 2010). “Quality at entry” (the quality of preparation) has been identified repeatedly as an important determinant of project success (e.g., Kilby, 1994, Limodio, 2011 and Smets et al., forthcoming). Rushing a project to the World Bank's board for approval could undermine quality by limiting consideration of alternatives and local needs during the identification process, leaving insufficient time to develop a full project plan, and creating a disincentive for a critical appraisal. If we can identify which projects were rushed, a more precise measure of the “cost of favoritism” (in terms of reduce aid effectiveness) is possible (Dreher et al., in press). To date, no one has tackled this issue because project identification dates (at the World Bank and elsewhere) are not publicly available. I side-step this problem by using a stochastic frontier model (SFM) to estimate the identification date from sequentially issued project identification numbers. This methodology, developed for studying productive efficiency at the firm level and since adapted to analyze total factor productivity at the national level, allows me to use project identification numbers, loan approval dates, and project/country characteristics to explore what determines the duration of project preparation. Duration in this context is akin to cost where the most “efficient” projects – those with the shortest duration – define the frontier. The methodology is analogous to duration analysis (in this case modeled with an exponential distribution) which simultaneously estimates the starting date from a variable that is a noisy measure of that date. The advantage of this approach is that, because it explicitly models the data generating process, the marginal effects of explanatory variables have a clear interpretation in terms of their impact on preparation duration, a feature not shared by a linear regression. The analysis finds that several political factors have a significant impact on the length of World Bank project preparation. When recipient countries vote with the U.S. in the UN General Assembly (UNGA) on measures the U.S. considers important, occupy one of the non-permanent seats in UNSC, or have their own national representing them on the World Bank Executive Board, the length of project preparation is reduced. This fits with Dreher et al.'s (2009a) finding that the number of World Bank projects approved per year is higher while a country holds a non-permanent seat on the UNSC. It is also consistent with Kaja and Werker's (2010) result that loan amounts from the World Bank's less concessional window, the International Bank for Reconstruction and Development (IBRD), are higher when a country's national is serving as a World Bank Executive Director. The next section (Section 2) presents a brief survey of the relevant portions of the literature on the political economy of IFI lending. Section 3 describes data on project identification numbers and explains how to incorporate them in a stochastic frontier analysis (SFA). Section 4 describes the remaining data and presents estimation results. Section 5 explores the robustness of these results. Section 6 concludes.

This paper uses sequentially issued World Bank project identification numbers to estimate a stochastic frontier model that explores the determinants of the duration of project preparation. I find robust evidence linking shorter preparation periods to a range of political variables that reflect donor interests and recipient country political or organizational leverage. The project preparation process is accelerated (and thus receipt of World Bank funds expedited) for countries more closely aligned with the U.S. on UN votes that the U.S. considers important. When a country temporarily holds a geopolitically important position, such as a seat on the UNSC, World Bank projects are delivered more quick. A similar “benefit” accrues from having a seat at the table in the World Bank Executive Board.24 This evidence of powerful countries using the World Bank as an instrument of their own foreign policy – doling out rewards to friends and punishment to enemies – adds to an already impressive literature. It validates the findings of Dreher et al. (2009b) and Kaja and Werker (2010) by demonstrating at least one channel through which the World Bank delivers more projects or more funding to favored countries. These findings are also interesting in light of earlier work on donor influence at different stages of the project cycle. Kilby, 2011a and Kilby, 2013 looks at “downstream” donor influence in the Asian Development Bank and the World Bank, i.e., donor influence over disbursement after loan approval. Because the Board has no direct involvement in post-approval disbursement decisions at these institutions, the measured influence is purely informal. Conversely, donor influence over commitment decisions (loan approval) is not purely informal. This allows a comparison of the relative importance of formal and informal influence and sheds light on the likely effectiveness of proposed formal governance reforms. Kilby (2009a) likewise focuses downstream to explore a narrower but equally important issue, the role of informal U.S. influence in the failure of structural adjustment. Research on the role of donor influence “upstream” during project identification and preparation is likely to be just as fruitful. This work provides important information to guide IFI reforms, specifically efforts to increase representation and limit excessive donor influence that distorts the allocation of resources and undermines the credibility of the institution. It also has interesting things to say about the role of World Bank staff versus borrowing countries governments in the process of selecting and preparing projects, that is, borrower ownership. That characteristics of a borrowing government – its bureaucratic quality and political institutions – have little impact on the duration of preparation is telling. The finding that donor interests impact both IDA and IBRD preparation times raises two important questions for future research. First, how do donors influence the allocation of IDA credits given the apparently rule-based IDA allocation system? Second, why do borrowing governments want more IBRD loans given that such loans are much less concessional than IDA credits? Indeed, these questions of mechanics and incentives apply broadly to other research that finds evidence of donor influence at the World Bank.25 The IDA allocation process is complex and historically not transparent. It draws on Country Performance and Institutional Assessment (CPIA) scores as well as Annual Review of Portfolio Performance (ARPP) ratings to set Country Performance Ratings (CPRs) that feed into the process determining multi-year “indicative allocations” of IDA funding for each eligible country (AFD (Agence Française de Développement), 2005 and Kanbur, 2005). It may be that U.S. influence impacts CPIA scores themselves (which are only publicly available from 2005 on). As noted in IEG (2009), these scores are somewhat subjective and align more closely with independent measures for non-IDA countries—where allocation is not determined by a CPIA-based formula.26 Alternatively, donor pressure could influence the weighting of subcomponents of the CPIA in IDA allocation calculations. Prior to 2009, these weights were not made public (IEG, 2009). ARPP ratings are generated internally by the World Bank and are not public. Again, CPRs are subject to adjustment (IEG, 2009). Finally, the gap between indicative allocations and actual IDA allocations by country is one third on average (AFD, 2005, 36). Although IDA credits are far more concessional and their allocation subject to intense debate, IBRD funds may be particularly useful to governments that are financially vulnerable or experiencing financial crises (Dreher et al., forthcoming, Humphrey and Michaelowa, forthcoming and Kilby, 2012) or that have shortened time horizons due to uncertain tenure. This is an important question closely linked to the long-term viability of multilateral development banks and there are a number of promising related avenues awaiting future research. From an empirical vantage, understanding of role of donor influence in IFIs has two interrelated benefits. First, it allows us to assess the impact of donor influence on IFI effectiveness, that is, the cost of favoritism (Dreher et al., in press). Second, by identifying exogenous sources of variation in IFI behavior, it provides empirical strategies to measure the impact of IFIs that solve vexing endogeneity problems. Building on the results of this paper in just this way, Kilby (2012) explores the impact of exogenous variation in preparation duration on World Bank project performance. Avoiding endogeneity problems noted in early studies (e.g., Denizer et al., 2013), that paper finds a significant positive effect of World Bank preparation on project outcomes. It also reinforces the message of Dreher et al. (in press) that political favoritism can lead to a rushed preparation process that undermines the development effectiveness of aid in some circumstances. Thus, understanding the impact of project preparation has important policy implications for donors and, viewed from the broader perspective, provides another way to assess the contribution of aid to economic development.