نظارت و عملکرد: در مورد پروژه های بانک جهانی

This paper explores the impact of donor supervision on development project performance using data from World Bank-funded projects. Maximum likelihood estimation of a restricted ordered probit function finds that early supervision has a positive impact on performance. Given the size of World Bank-funded projects, gains from increasing supervision far outweigh the costs. The results provide evidence of an institutional bias toward lending at the expense of increased development impact.

Even the best laid plans can go astray. Nowhere is this aphorism more apt than in the developing world where carefully planned development projects all too frequently fail to achieve their goals. In its 1993 Annual Review, the World Bank's Operations Evaluation Department judged nearly 40% of recently completed projects to be failures.1 While 86% of World Bank projects completed before 1980 were judged satisfactory, only 72% of projects completed since 1980 have achieved such performance. Simultaneously, several researchers have noted a divergence between anticipated and actual economic rates of return.2 The sources of problems are myriad and, though some are attributable to imperfect plans, many arise during the implementation phase of development projects — policy reforms that never take place, delays, cost overruns, substandard construction, coordination failures and “slippage.” This paper examines aid donor attempts to combat these problems via project supervision, a combination of monitoring and advising. Using data from World Bank-funded projects, I estimate the impact of World Bank supervision on project performance. Better understanding of the role and impact of supervision is of both practical and theoretical interest. International development projects account for a very sizable investment in developing countries with the World Bank alone supervising a portfolio of 1850 projects representing some US$300 billion of investment.3 Improved project supervision may increase the return on this investment and is one of the only instruments available to donors during project implementation. Aid performance is also central in maintaining support for aid funding in the face of donor fatigue in the post-Cold War era. From a more theoretical perspective, examining donor allocation decisions also sheds light on donor objectives. This is the approach used in the extensive literature on donor allocation of aid among recipient countries.4 In a parallel fashion, examining the allocation of administrative resources within an aid agency sheds light on its objective function.5 Estimates in this paper show that, at current levels of supervision, the marginal benefit of World Bank supervision greatly exceeds its nominal marginal cost. This suggests that World Bank practices are more complex than simply maximizing expected performance on a project by project basis. Specifically, the current low level of supervision indicates a management bias toward new lending at the expense of development effectiveness. The econometric analysis draws on data from 1426 projects completed between 1981 and 1991. Project performance, in theory a continuous variable, is measured by a discrete rating generated annually by World Bank project managers; supervision is the number of World Bank staff weeks devoted to supervising a project in a given year. The project performance equation relates lagged annual supervision to annual changes in performance. Taking changes in performance ratings as an indicator of changes in the latent performance variable, the result is a restricted ordered probit model estimated by maximum likelihood. Coefficient estimates show that supervision has a positive impact on project performance. Supervision is most effective early in the project and in smaller projects. Translating these results into dollar figures, the marginal benefit of supervision is much greater than the marginal cost at current supervision levels. The paper is organized as follows. Section 1 explains the concept of project performance used in this paper and develops the performance equation. Section 2 turns to the data and discusses data reliability issues. Section 3 presents the results of the performance equation estimation and translates these into dollar figures. Section 4 concludes.

This paper measures the impact of World Bank supervision on the performance of World Bank-funded projects and finds that supervision does improve performance. Estimates indicate that the benefits from supervision far outweigh the costs. Supervision is most effective early in the implementation period and in projects with smaller loans but has a relatively homogeneous impact across regions and sectors. While a single additional week of supervision is not likely to change project performance dramatically, larger and more persistent increases can have a sizable impact. The average expected economic rate of return appears to be a few percentage points higher with the current level of supervision than with minimal supervision; substantial increases above the current level of supervision are predicted to have a similar effect. When translated into dollar figures, these changes in the economic rate of return indicate large gains — gains that appear to justify increased supervision given its relatively low marginal cost. Comparing the marginal benefit and marginal cost of supervision raises a number of problems, the most important of which is the impact of political constraints on the World Bank's administrative budget. Given the constant pressure from the United States and other donors to limit administrative spending and staff size, the budget is best regarded as fixed. From the point of view of World Bank management, the opportunity cost of more supervision is less project preparation.34 Assuming that additional preparation results in more lending, more supervision means less new lending. The relevant comparison is, then, between the marginal benefit of supervision and the marginal benefit of preparation. Using data on preparation and calculations parallel to those for supervision, we can impute an approximate marginal benefit of US$100,000 per staff week for preparation as compared to US$250,000 per staff week for supervision.35 Thus, even acknowledging the constraints it faces, the World Bank appears to supervise projects much less than it should to maximize the development impact of its activities. Scholars of international development assistance have advanced numerous theories to explain the World Bank's apparent bias against supervision and toward lending. Some attribute it to institutional incentives set-up by McNamara during his push to grow the institution. Others point to the World Bank's role in promoting international capital and specifically the interests of American banks. The desire to maintain a net flow of funds from the World Bank to the developing world may also be a factor. Indeed, the premium placed on quantity over quality has been noted in many development agencies.36 Supervision levels have increased since the highly critical Wapenhans report and significant restructuring during Wolfenson's presidency (Wapenhans, 1992). While increasing average annual project supervision by a few staff weeks and placing more emphasis on project outcome are certainly steps in the right direction, in light of the results presented in this paper, larger adjustments may be in order. Until the World Bank undertakes more radical reforms, its bias toward new lending will likely continue to reduce the institution's development effectiveness.