تعدیل ساختاری دقیقا چه چیزی را تعدیل کرده است؟ ارتباط سیاست و رشد با وام های مکرر صندوق بین المللی پول و تعدیل بانک جهانی

Analysis of adjustment loans often overlooks their repetition to the same country. Repetition changes the nature of the selection problem. None of the top 20 recipients of repeated adjustment lending over 1980–99 were able to achieve reasonable growth and contain all policy distortions. About half of the adjustment loan recipients show severe macroeconomic distortions regardless of cumulative adjustment loans. Probit regressions for an extreme macroeconomic imbalance indicator and its components fail to show robust effects of adjustment lending or time spent under IMF programs. An instrumental variables regression for estimating the causal effect of repeated adjustment lending on policies fails to show any positive effect on policies or growth.

On February 5, 1980, World Bank President Robert McNamara sought and received approval from the World Bank Board to launch a new instrument: the structural adjustment loan (SAL). The proposal followed a year of discussion with the operations chief Ernest Stern, with the outline of the SAL emerging on a flight the two took together to the Bank-Fund Annual Meeting in Belgrade in late September 1979. The loans would provide finance over a period of several years in return for reforms in trade protection and price incentives for efficient resource use.1 The introduction of the new instrument came against the backdrop of the second oil shock in 1979. It was intended as a preventative instrument so that the “current account deficits of many developing countries do not become so large as to jeopardize seriously the implementation of current investment programs.” Although the IMF had always been making “adjustment loans” in the form of standbys, the IMF also in the 1980s expanded the number and maturity of adjustment loans it was making. The dual rationale from the SALs from the beginning was to maintain growth and to facilitate balance of payments adjustment. The “specific objective” of the SALs were to help countries “reduce their current account deficit to more manageable proportions by supporting programs of adjustment … to strengthen their balance of payments, while maintaining their growth and developmental momentum.”2 As the 1981 World Development Report said, successful adjustment implies “a minimum sacrifice of income growth.”3 This emphasis on growth continued. In June 1983, for example, the World Bank and IMF published excerpts of speeches by their respective heads under the overall heading: “Adjustment and growth: how the Fund and the Bank are responding to current difficulties.“4 In 1986, the World Bank president A.W. Clausen gave a speech entitled “Adjustment with growth in the developing world: a challenge for the international community”.5 In 1987, the World Bank and IMF published a volume entitled “Growth-oriented adjustment programs” with an introduction discussing the “fundamental complementarity” of “adjustment and economic growth.”6 Because the SALs were supposed to facilitate balance of payments correction, the structural adjustment loans were intended to end after a period of several years of adjustment. As the initial McNamara document put it, structural adjustment lending entailed “an association with a borrower in a program of structural change over 3 to 5 years which will require financial support.”7 A flavor of the early structural adjustment package is given in 1981 in the first of what would turn out to be 26 structural adjustment loans to Cote d'Ivoire: The loan would be in support of the Government's program of structural adjustment. The reforms envisaged by the program are designed to improve the level of public savings and the efficiency in the use of public resources; restructure the agricultural planning system and associated development institutions so that an expanded, well-designed investment program yielding high returns can be mounted in the sector; reflect the costs of providing public services to the sector; assure that rational prices and world market conditions would guide decisions to invest and produce; restructure public enterprise, management, financing and accountability to ensure efficient market oriented operations; and restructure incentives, to promote efficient export-oriented industrial investments.8 This statement already contains the main features of what would characterize adjustment lending for the next two decades for the IMF and World Bank: fiscal adjustment, getting the prices right, trade liberalization, and, in general, a movement towards free markets and away from state intervention. The IMF had long been doing conditional stand-by loans, but it also expanded the number and types of adjustment loans in the 1980s. IMF adjustment loans, which often served as a prerequisite for World Bank adjustment loans, stressed macroeconomic stabilization—especially fiscal adjustment and inflation stabilization. Exchange rate devaluation was also a key element in IMF loans. IMF and World Bank conditionality has evolved over time, but there is a common element of macroadjustment and getting prices right that has remained constant from the beginning. One way to evaluate an initiative like adjustment lending is to compare results to objectives. This kind of evaluation is informative because it measures success against the ex ante benchmarks imposed by the policy-making institutions themselves and against the expectations they created. This kind of monitoring of policy-making institutions has some normative value in that it has strong incentives for the institutions; it elicits strong effort from the institutions because it does not allow them to blame poor outcomes on unobservable shocks or on their particular choice of control variables. The conclusions reached by this kind of evaluation are not particularly favorable: “there is a long legacy of failed adjustment lending where there was no domestic constituency for reform ... donors have not been sufficiently selective with policy-based lending” (World Bank, 1998, p. 48). Or, as a more recent study, World Bank study of African cases puts it, “that the 10 countries in our sample all received large amounts of aid, including conditional loans, yet ended up with vastly different policies suggests that aid is not a primary determinant of policy” (Devarajan et al., 2001, p. 2). This follows the World Bank's (1994)Adjustment in Africa report that found limited and uneven policy improvement in countries undergoing “structural adjustment.” The World Bank's (2002, pp. 110–111) most recent statement about structural adjustment is the guarded statement that it “can contribute, and often has contributed, to growth,” but “the performance of adjustment operations has been mixed, especially during the 1980s.” Another way to evaluate success is the counterfactual methodology—how the intervention changed the outcome compared to what would have happened without the intervention. Countries that received adjustment loans did so because they were having poor macroeconomic and growth outcomes, and so it would not be surprising if we found a negative association between these outcomes and adjustment loans without correcting for selection bias. To use a medical analogy, we would expect hospital patients to be sicker than the average person on the street, but this does not imply that hospitals cause sickness. The vast literature on evaluating IMF and World Bank adjustment loans has made much of the selection bias problem.9 This has variously been addressed by using Heckman-type selection techniques, before and after analysis, or control group methodology. For example, in an earlier research, the World Bank (1992, p. 2) found that after controlling for selection bias, adjustment lending meant “the middle-income countries enjoyed growth four percentage points higher than would otherwise have occurred and the low-income group had growth two percentage points higher.” This early study concludes “adjustment lending is also associated with improved policies.” However, the results from a wide range of independent researchers, World Bank and IMF studies have been all over the map, with positive, zero, or negative effects of adjustment lending on growth, and with similarly mixed evidence of AL on policies (see the survey by Killick et al., 1998). Two recent studies (Przeworski and Vreeland, 2000 and Barro and Lee, 2002) find a significantly negative effect of IMF lending on growth. These studies have almost universally treated adjustment loans as independent events, not using the information contained in the frequent repetition of adjustment loans to the same country. The repetition of adjustment loans changes—even if it does not eliminate—the nature of the selection bias. To return to the medical analogy, if a patient is readmitted to the hospital after the first treatment, this suggests that the first treatment was not effective. The alternative, more favorable, explanation for why adjustment loans were repeated is that adjustment was a multistage process that required multiple loans to be completed. In the medical analogy, the patient needed multiple doses of medicine to fully cure the illness. Under this interpretation, we would expect to see a gradual improvement in performance with each successive adjustment loans, or at least an improvement after a certain threshold in adjustment lending was passed. Selection bias could still operate with repetition if adjustment loans were repeatedly initiated in countries that failed to correct the macroeconomic problems and poor growth under earlier adjustment loans. It could be that governments failed to follow through with the conditions of each loan (the patients did not take their medicine) and so additional programs became necessary. If this is the explanation, then the question then becomes why the IMF and World Bank kept giving new adjustment lending resources to countries that had such a poor track record of compliance with the conditions. Again, the interpretation is not particularly favorable to the effectiveness of adjustment lending as a way to induce “adjustment with growth.”

The big stylized facts of adjustment lending suggest that structural adjustment did not succeed in adjusting macroeconomic policy and growth outcomes very much. Structural adjustment loans were repeated many times to the same country, which itself is suggestive of limited effect of the earlier adjustment loans. There were some successes, but also some big disasters. The principal finding is that, taken together or separately, the prevalence of one or more extreme macroeconomic distortions did not diminish as adjustment lending accumulated. There is no evidence in any of the statistical exercises that per capita growth improved with increased intensity of structural adjustment lending. These findings are robust to controlling for endogeneity of adjustment lending and initial policy distortions in the cross-section sample. There are many possible caveats to the findings. Only in the last section do I attempt to address the causality problem, and the instruments there may well be imperfect. In the earlier sections, I provide econometric and other types of descriptive statistics in an attempt to place bounds on what the counterfactual would have to be to generate a positive impact for adjustment lending. The emphasis on repetition of structural adjustment loans is a new contribution to the literature, but this focus may miss some cases of success that only took a small number of adjustment loans. There are also caveats that go in the other direction. I have limited myself to easily quantifiable macroeconomic indicators. Structural adjustment lending also sought to privatize state enterprises, reform inefficient and loss-prone financial systems, remove the penalty imposed on agriculture, improve the efficiency of tax collection and public spending, reform and downsize the civil service, control corruption, and improve many other areas. If anything, progress on these less-quantifiable reforms has been slower than on the macroeconomic indicators, according to complaints in many World Bank and IMF reports. The findings of this paper are reminiscent of results on foreign aid—that foreign aid was not very selective in rewarding good policies and did not on average increase growth (Boone, 1995, World Bank, 1998 and Burnside and Dollar, 2000). The same seems to be true of adjustment lending. Putting external conditions on governments' behavior through structural adjustment loans has not proven to be very effective in achieving widespread policy improvements or in raising growth potential. If the original objective was “adjustment with growth”, there is not much evidence that structural adjustment lending generated either adjustment or growth.