Review of the 2002 Conditionality Guidelines - Selected Issues

Review of the 2002 Conditionality Guidelines - Selected Issues


Review of the 2002 Conditionality Guidelines - Selected Issues


1. This paper examines selected issues by way of background for the review of the 2002 Guidelines on Conditionality. It examines program implementation, program goals and strategies, the breadth of coverage of conditionality, some issues in the use of the tools of conditionality, prior actions, conditionality in capital account crisis cases, and the process of program development. Annexes explain how the conditionality data used in various chapters were derived and provide a list of programs included in the analysis.

2. This review examines developments since end-2000 and, where relevant, compares them to earlier developments. The terms “pre-“ and “post-guidelines” are used hereafter to refer to this cutoff date of end-2000. The cutoff date is chosen because an Interim Guidance Note on Streamlining Structural Conditionality, which foreshadowed the main changes in the conditionality guidelines of September 2002, was issued in September 2000. Moreover, the Poverty Reduction and Growth Facility (PRGF), which shares some of the objectives of the new guidelines, replaced the Enhanced Structural Adjustment Facility (ESAF) in September 1999. Because the new policy was thus, essentially, phased in over several years, and also because data for every arrangement are assigned to the year in which it was approved, even if the arrangement stretched into later years (see Annex I), a clear cutoff point is somewhat artificial.

3. The review comes at an early stage of experience with the new guidelines and caution must be exercised in interpreting its results, given the small samples involved and the inherent difficulty of comparing program-related conditions. Even though the review draws in part on case studies, no substitute exists for cross-country analysis for identifying broad trends, and at this stage such analysis is hampered by small sample sizes— typically, about 10 arrangements a year under each of the GRA and PRGF. Moreover, the analysis must be cognizant of the fact that program-related conditions can be of quite different import. Nevertheless, there is no good way of controlling for these differences without creating new—and in many respects more problematic—weaknesses in the analysis. Hence, in what follows conditions are weighted equally, regardless of whether they entail small or sweeping policy changes.

I. Program Implementation1

A. Introduction

4. Implementation of Fund-supported programs was expected to improve with the new guidelines. The guidelines explicitly put greater emphasis on parsimony of conditions, national ownership of policies, and implementation capacity, all of which were expected to improve the prospects for successful program implementation—and ultimately lead to better economic outcomes. While it is still too early to assess economic outcomes, a preliminary assessment can be made on implementation of Fund-supported programs since the advent of the new guidelines.

5. It is useful to distinguish between implementation of the overall policy program and implementation of specific structural conditionality. More consistent implementation of policy programs by country authorities should be evidenced by more completed reviews and thus more programs staying on track, and also by better implementation of conditions. At the same time, the focus on criticality of individual conditions was expected to lead to a decline in the number of waivers of missed conditions.

6. The assessment of program implementation is based on two main inputs: an analysis of program reviews and an analysis of implementation of individual conditions. For the latter, information on individual conditions is obtained from the Monitoring Fund Arrangements (MONA) database (see Annex I), which tracks Fund arrangements based on the initial request and completed program reviews. Thus, implementation of conditionality is not recorded if a program goes off track and later scheduled reviews remain uncompleted. In other words, this data source only allows the analysis of implementation conditional on programs remaining on track, or having been brought back on track after a temporary interruption. By contrast, the analysis of program reviews—including uncompleted ones— captures developments in all arrangements through to expiration.

7. Only a few years have passed since the streamlining initiative began and the results therefore have to remain tentative. Only 60 new arrangements were approved during 2001–03, 27 supported from the General Resources Account (GRA) and 33 under the Poverty Reduction and Growth Facility (PRGF). Of the latter, only four had expired by end- June 2004, the cutoff date used for much of the following analysis, and seven by end- December 2004.2 The average number of structural conditions per arrangement (not adjusted for program length) is 14 structural benchmarks (SBs), five structural performance criteria (PCs), and nine prior actions (PAs). Reflecting the small sample sizes, the observed changes since end-2000 are often not statistically significant.3

B. Program Interruptions

8. A fundamental question in assessing program implementation is how much time programs spend on track and off track. A natural metric is completion of program reviews. Completion of a review implies that the overall program remains broadly on track, despite possible slippages that may have occurred. Conversely, failure to complete a scheduled review generally indicates that overall implementation is not sufficient to ensure achievement of the program objectives.4

9. It is useful to distinguish between temporary and permanent program interruptions:

  • A temporary interruption is defined as a delay in completing a scheduled review of at least six months, provided that the review is eventually completed. Previous rephasing and rescheduling of reviews are taken into account, in the sense that a delay is assessed relative to the most recent review schedule. To make ESAF and PRGF arrangements comparable, each ESAF-supported program is treated as one three-year program by combining the individual annual arrangements. The approval of the 2nd and 3rd annual arrangements are treated as reviews, and are generally presumed to be scheduled six months after the mid-term review.5

  • A program interruption is considered permanent if the program went off track and was not brought back on track before the arrangement expired. This means that scheduled program reviews were left uncompleted at the end of the arrangement.6

10. In line with expectations, more programs have been completed successfully since 2000. The share of expired programs that were permanently interrupted has declined significantly from its peak in 1998–2000 (Figure 1, left panel).7 This trend is particularly pronounced for arrangements in the General Resources Account (GRA). While permanent interruptions have also declined among PRGF (and ESAF) arrangements, the corresponding sample size is small, as only seven arrangements approved in the post-guidelines period have expired as of end-December, 2004. Nevertheless, an improvement is also visible among PRGF arrangements, at least relative to 1998–2000.

Figure 1:
Figure 1:

Permanent Program Interruptions

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

11. Similarly, fewer programs remain completely unimplemented. Some programs go off track shortly after approval, that is before completing the first review, and remain off track until the program expires. As expected from the greater focus on ownership and implementation capacity, the share of such completely unimplemented programs has also declined, although for GRA arrangements it was somewhat above the share in 1995–97 (Figure 1, right panel).

12. Notwithstanding these improvements, temporary program interruptions and review delays have continued as previously. Temporary interruptions of at least six months are frequent and occur in more than half of PRGF arrangements and roughly one third of GRA-supported arrangements (Figure 2, left panel). Moreover, no recent improvements are evident (and temporary interruptions in PRGF-supported programs even increased slightly in the post-guidelines period). Average delays in completing reviews also remained stable recently at about three months (Figure 2, right panel). Compared to 1995–97, temporary interruptions are more frequent and delays have increased.

Figure 2:
Figure 2:

Temporary Program Interruptions and Review Delays

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

13. Overall, full program implementation in line with the initially envisaged timing is still the exception rather than the rule. Almost two thirds of all expired post-guidelines arrangements were off track for a significant period of time—i.e., they were interrupted for at least six months (Table 1). And only about one fourth of arrangements had all reviews completed broadly on time.

Table 1.

Program Interruptions

article image
Source: MONA, Staff Reports

Only expired arrangements as of December 31, 2004.

Only reviews scheduled before 7/1/2003. Reviews still pending as of 31/12/2004 are assumed to never be completed.

C. Implementation of Structural Conditionality

14. A key question is whether the decline in permanent program interruptions reflects better underlying implementation, or the granting of waivers and completion of reviews when implementation has been weak. The Fund’s decisions on reviews and waivers are of course constrained by the respective policies, but—just as the initial program design requires judgments on the consistency of policies with program goals—require the use of judgment in cases where there have been implementation slippages. A look at implementation of individual conditions may help answer the above question.8

15. Implementation of individual conditions was clearly expected to improve with the new guidelines. First, better ownership should make it more likely that the overall programs—and thus individual conditions—are implemented as envisaged. Second, the focus on criticality was expected to reduce waivers for missed conditions. And third, more flexibility in modalities—e.g. through more outcomes-based conditionality—aimed at reducing the need for waivers in cases where acceptable alternative measures are taken. As a result, measured implementation while programs are on track was expected to improve. The following sections take a look at all three types of structural conditions (performance criteria (PCs), structural benchmarks (SBs), and prior actions (PAs)) separately, with a focus on PCs and SBs. Prior actions for initial program approvals and completed reviews have by definition a high implementation score.

Structural Performance Criteria

16. Despite expectations, the rate of waivers for nonobservance of structural PCs has not declined.9 Table 2 shows summary indicators of implementation of structural conditions.10 For PRGF arrangements, waiver rates have remained broadly stable, while waiver rates appear to have risen for GRA-supported arrangements (Figure 3).11 For comparison, PC waiver rates for quantitative conditions—which are much lower than for structural conditions—have actually declined slightly over time in GRA-supported arrangements, while remaining broadly flat in PRGF arrangements (Figure 3, right panel). The failure of the waiver rate for structural PCs to decline raises the question whether the Fund has actually increased its focus on criticality or if too much flexibility is provided ex post and the Fund is accommodating slippages that threaten the achievement of the program goals. Ultimately, economic outcomes will provide an answer, but a closer look at the waived PCs can already provide important insights at this time.

Table 2.

Implementation of Structural Conditionality

article image
Source: MONA, Staff Reports

Conditional on programs being on track, i.e. excluding PCs with test dates after the last completed review

Based on all SBs. A restriction to SBs with test dates prior to the last completed review is not possible because test dates for SBMs are not consistently recorded in MONA.

Percent of conditions not fully met by the specified test date.

Figure 3:
Figure 3:

Waiver Rates for PCs

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

17. Waivers for structural PCs can be decomposed into three sub-categories, based on whether the measure was replaced by an alternative (or slimmed down) measure, whether it was implemented late (including alternative delayed measures), or whether the measure was not implemented at all during the period of the Fund-supported program (i.e. the measure “lapsed”). This breakdown is presented in Figure 4.12

Figure 4:
Figure 4:

Composition of Structural PC Waiver Rate (in percent of all structural PCs)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

18. Delayed measures account for the bulk of waived structural PCs:13

  • Some of these delayed measures are implemented quickly—indeed, as prior actions for a program review and thus for the granting of a waiver. While MONA does not permit precise identification of missed PCs that become prior actions, a comparison of PCs recorded as “met with delay” with prior actions recorded in MONA suggests that a growing proportion of PCs (from 3 percent in 1995–97 to 7 percent in 1998-00 and 14 percent in 2001–03) is waived but then converted into prior actions to be implemented before purchases or disbursements can proceed.14 Indeed, these rough estimates suggest that the increase in measures that become prior actions may account for all or most of the increase in delayed measures. Such a development would be consistent with an increased focus on criticality in setting PCs.

  • At the same time, some measures can take a long time to be implemented, with waivers granted on the basis that action is in train to ensure implementation at a later date.15 This is consistent with the fact that structural reforms are seldom short-term measures and that their specific timing is often less important than whether they are implemented.

19. Alternative measures play a more, and increasingly, important role in GRA- than PRGF-supported programs. These are measures that are implemented on a modified basis acceptable to the Fund. Though they seem to be largely absent from PRGF-supported programs, they contributed about 10 percent of conditions in stand-by and extended arrangements over the 2001–03 period, well above their share in 1995–97. Examples of alternative measures include an increase in the electricity price by 15 percent together with efficiency improvements in the sector, together equivalent in cash flow terms to the condition’s requirement of a 20 percent electricity price increase (Serbia), the establishment of salary coefficients and associated wage ceilings in most rather than all ministries (Croatia), and a reduction in the specific number of branches of two banks slightly below the numerical target set in the Fund condition (Turkey).

20. Lapsed measures have declined in both GRA- and PRGF-supported programs, suggesting that Fund-supported programs have become more focused on criticality. Lapsed rates in GRA-supported programs have declined from 26 percent over 1995–97 to 15 percent over 2001–03; the corresponding figures for PRGF-supported programs are 20 percent and 14 percent respectively. Of course, this lapsed rate does not give a picture of conditions the arrangement could have “done without:” it is calculated for all arrangements, including programs that went and stayed off track, and some structural PCs that are recorded as lapsed may indeed have been the cause of a permanent program interruption. Restricting the sample to expired and completed programs captures more precisely those structural PCs that arrangements were able to “do without,” and in this sample the lapsed rate declines for GRA-supported programs from 31 percent to 17 percent over the same period.16 A similar calculation cannot yet be done for PRGF-supported programs, since the sample of expired post-guidelines PRGF arrangements is too small. But the lapsed rate in expired and completed PRGF-supported programs has probably declined alongside, or even more than, the unadjusted lapsed rate shown in Figure 4, since the inclusion of continuing programs in the post-guidelines period imparts an upward bias to the lapsed rate.17 Increasing insistence on implementation—despite the higher waiver rate—could also help explain why delays in completing reviews and temporary program interruptions have not declined.

Structural Benchmarks, Prior Actions, and Overall Implementation

21. The trends described above for structural PCs also broadly apply to SBs. While SBs do not require waivers when missed, an “implicit waiver rate” can be calculated for SBs as the share of conditions that was not fully met on time—analogously to the treatment of PCs. The same holds for the “lapsed rate.” Figure 5 clearly shows that differences between measured implementation of PCs and SBs are minimal in the post-guidelines period. The observed trends are also broadly similar, which is to be expected given that the focus on ownership and criticality applies to SBs as well as PCs. Implicit waiver rates for SBs have also not declined significantly. For GRA-supported programs, they remained very stable across the subperiods, while in PRGF-supported programs an increase in 1998–00 was followed by a subsequent reversal, leaving the rate at the level observed in 1995–97. The decline in the lapsed rate for SBs, however, is only evident in GRA-supported programs, and only between the 1995–97 and the 1998–00 periods. The lapsed rate for SBs is broadly stable in PRGF-supported programs, although as noted above its most recent measurement is likely to be biased upward.

Figure 5:
Figure 5:

Implementation of Structural PCs versus SBs

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

22. As expected, implementation is much better for prior actions than for other forms of conditionality. Implied waiver rates are very low since all prior actions usually have to be done before a review is completed or a new arrangement approved. Moreover, implementation is measured as having improved significantly over time. This improvement is likely related to the increased formality of prior actions following their inclusion under the misreporting policy in 2000. Additional uncertainty also remains regarding the data quality and especially consistency over time.18 Nevertheless, arrangements are occasionally approved and reviews completed despite the non-observance of prior actions. This explains why prior actions still do not register full implementation.

23. An implementation index can be constructed to summarize the implementation record of specific program conditions.19 The implementation index is constructed as follows: all “lapsed” conditions get the value 0; all conditions that were partially met, met with a delay, or for which some alternative action has been taken get the value 1; and all conditions that were fully met on time get the value 2. Figure 6 shows the results for PCs and SBs combined.20 Overall implementation in GRA-supported programs is almost unchanged compared to the three years preceding the new guidelines: the increase in the waiver rate was offset by the decline in the lapsed rate.21 PRGF-supported programs, however, displayed a significant improvement—admittedly compared to a relatively low level in 1998–00—driven by the reduction in the lapsed rate for PCs and the reversal of the rise in the implicit waiver rate for structural benchmarks. (The observed increase of about 0.1 in the implementation index for PRGF arrangements is equivalent to 5 percent of conditions changing from “lapsed” to “fully met.”)

Figure 6:
Figure 6:

Implementation Index, excluding PAs

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

D. Regional Differences

24. Regional differences in program implementation can be examined by grouping countries by area departments, although the results should be treated with caution. Since for most of the sample period (1995–2003) the European II (EU2) department still existed, this distribution is maintained. Of course, the characteristics of countries that turn to the Fund for support vary widely across area departments—and even within an area department over time—and observed differences in implementation may well be due to differences in this country mix. In addition, the samples obtained by classifying arrangements by department as well as by three-year period are often small.

25. The decline in the frequency of permanent program interruptions is robust across regions. In almost all regions, permanent program interruptions have declined since the advent of the guidelines (Figure 7, upper panel).22

Figure 7:
Figure 7:

Implementation by Region

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

26. Regional differences regarding average delays in completing reviews have increased (Figure 7, middle panel).23 While most regions showed some moderate shortening or remained stable, average delays increased in EU2 and especially in APD. To some extent, the increase in these two regions can be explained by a shift in the composition of arrangements away from GRA-supported arrangements, which generally show shorter delays, toward PRGF arrangements.

27. The implementation index improved significantly in the African department (AFR), while it declined in APD (Figure 7, lower panel).24 Other regions remained broadly unchanged, with the Middle Eastern department (MED) also registering some improvement. This suggests that the general post-guidelines increase in the implementation index of PRGF-supported programs is primarily driven by African countries. Other regions now dominated by PRGF arrangements (EU2—where the implementation index was already high—and APD) do not show similar recent improvements. In APD the decline in the implementation index has accompanied a marked shift in the composition of the group of program countries, from predominantly emerging markets to exclusively low-income countries, which experienced significant difficulties during their programs.25

E. Conclusions

28. While the decline in permanent program interruptions is in line with the expectations of the 2000-02 conditionality review, the analysis of individual conditions shows a mixed picture. Permanent interruptions have become less frequent and this development is evident broadly across types of arrangements and regions, providing some confirmation that the result is robust and that a significant change has taken place since the advent of the guidelines. The evidence on implementation of individual conditions, however, shows a different picture. On the positive side, more conditions are implemented, as evidenced by the declining lapsed rate. However, waiver rates have not declined as expected across Fund-supported programs.

29. A full assessment of these developments must await availability of outcomes data, but the current evidence is consistent with an increasing focus on criticality. Generally, the analysis has been hampered by the fact that only a few years have passed since the adoption of the guidelines. Consequently sample sizes are often small and some of the results may not prove robust in the long run. Nevertheless, some clear trends have emerged regarding certain measures of implementation, although without an unambiguous interpretation. On the one hand, fewer permanent program interruptions could be the result of better country ownership of the policy programs, and the high waiver rate may be an indication the persistence of difficulties in formulating the precise timing and modalities of measures. On the other hand, the same developments could also mean that criticality is compromised and judgments stretched too far in the direction of accommodating slippages that threaten the achievement of the program objectives. The decline in the lapsed rate suggests that the Fund is actually “giving up” less on conditions it sets, and thus points to the former interpretation, and to a greater focus on criticality in setting conditions. Ultimately, economic results will reveal if program outcomes have improved or not.

II. Goals and Strategies in Fund-Supported Programs26

A. Introduction

35. Implementation of the guidelines’ principles of parsimony and criticality requires the specification of clear and explicit program goals, and can only be helped by the identification of clear and explicit economic strategies. The guidelines, as part of the effort to focus program conditionality, indicate that “Fund-supported programs should be directed primarily toward the following macroeconomic goals: (a) solving the member’s balance of payments problem without recourse to measures destructive of national or international prosperity; (b) achieving medium-term external viability while fostering sustainable economic growth” (guidelines, ¶6).27 To these goals, the PRGF Instrument adds “fostering durable growth, leading to higher living standards and a reduction in poverty.”28 Measures critical to program goals can only be identified if goals are clearly and explicitly defined, and clear program goals should also encourage greater focus on identifying the strategies by which those goals could best be achieved, and serve to center program design on a set of conditions critical for their attainment. Parsimony would follow, as conditions external to the stated strategies would be unnecessary.

36. This chapter examines how clearly goals and strategies are articulated in staff reports for requests for use of Fund resources (UFR), whether they are in line with the guidelines, and the process by which they affect conditionality. It looks at whether and how program goals and strategies for their achievement are specified in staff reports for arrangements approved since 2001, what these goals and strategies are, and whether their identification has led to focused conditionality since the adoption of the new conditionality guidelines (and the creation of the PRGF). The articulation considered is that of the staff; the way the authorities set out program goals and strategies in letters of intent, memoranda of economic and financial policies, and poverty reduction strategy papers is beyond the scope of this chapter.

37. At the outset, it is worth noting that goals and strategies are distinct elements; still, identifying either of these elements in staff reports is not straightforward. Staff reports should ideally indicate whether the program is focused on any or all of the macroeconomic goals outlined in the conditionality guidelines. Beyond that, some discussion of the strategy contemplated toward that goal is warranted. For example, in a PRGF-supported program with the stated goal of higher growth, structural measures might focus on removing obstacles to private sector development or reducing crowding-out caused by excessive domestic financing of the government’s imbalances. In practice, however, the statements about program goals and strategies made in staff reports reflect imperfectly the underlying process by which critical reforms are identified. Indeed, exploring how decisions are made about the conditionality set in a Fund-supported program inevitably requires drawing conclusions on general intentions from frequently imprecise statements made in staff reports. As a result, the risk exists that presentational issues may be mistaken for real problems in the formulation of conditionality. To minimize this risk, statements are interpreted liberally. It should be noted, however, that a weak presentation can itself be problematic if it makes program proposals difficult to understand or interpret.

38. The following rules are thus followed in classifying the goals and strategies that can be identified in staff reports. Any statement focusing attention on at least one of the macroeconomic goals in the guidelines is deemed to represent consistency with the guidelines, as drawing finer distinctions would second-guess the prioritization among different goals. Further, any area described in program requests as a focus of discussions or a priority for the near- to medium-term agenda is interpreted as a strategy. Ideally, conditionality would also in practice be focused in those areas defined here as strategies. As some statements cite only strategies, it seems reasonable to impute an underlying goal:29 for instance, citing fiscal sustainability as a program focus reasonably leads into macroeconomic sustainability as an overarching goal. While this liberal imputation of strategies may err on the side of attributing more process than may actually have occurred, a more rigorous interpretation of goals statements might reject strategies that underlay program discussions, even though they were not explicitly presented as such in staff reports.

B. Goals

39. Almost all requests for new arrangements include some statement of goals (Table 3).30 Staff reports in almost all cases do include some statement of the focus of either the authorities’ program or of the ultimate objective of Fund conditionality in requests for arrangements. In only about one in thirteen staff reports is no such clarifying statement made. About two-thirds of staff reports specify the goals of the Fund-supported program, which are either explicitly or implicitly expressed as a subset of the authorities’ overall goals. Another twenty-five percent of staff reports cite only the authorities’ goals, without specifying whether the program supported by the Fund encompasses all or only a subset of these goals. Finally, in the remaining five percent, the staff report is not clear whether the goals described are those of the Fund-supported program or the wider goals of the authorities.

Table 3.

Frequency of Stated Goals and Strategies in Staff Reports for Requests for Fund Arrangements

article image

Programs often cite more than one goal or strategy.

Includes PRGF/EFF blended arrangements.

Includes arrangements under the EFF.

40. Separating the goals of the Fund-supported program from the authorities’ wider set of goals is not merely a semantic difference. Circumscribing goals focuses discussions in critical areas. Thus, distinguishing program goals from the larger goals of the authorities leaves the authorities wider policy space. It is a good sign that two thirds of staff reports state clearly the program goals that the Fund is supporting. Presentational issues—in particular, the wish to emphasize the authorities’ ownership of the program—may account for the remainder of staff reports where such goals are not distinguished. Program requests are clearer, however, when this distinction is drawn. To be sure, in some cases program goals may be implicit from the staff report in its entirety, though a clear statement remains preferable.

41. In general, program goals in UFR requests that indicate an overarching goal, as distinct from an area in which to focus conditionality, fall into five categories: external stability, macroeconomic stability, growth, employment, and poverty reduction. While external stability is contained in the concept of macroeconomic stability, the reverse is not true; in some cases, measures to ensure stability in the domestic economy may be crucial for program success while, in other cases, external shocks may have caused a dislocation in an otherwise well-functioning economy and external stability is a sufficient goal.

42. Much of this falls easily within the guidelines, though growth is to some extent a gray area. As noted in section II.A, macroeconomic and external stability are appropriate goals for any Fund-supported program, and poverty-reducing growth is an explicit goal of the PRGF. While growth is likely to be a goal of the authorities in any country, and indeed, is often a crucial component in improving external sustainability, the guidelines indicate that conditionality in a GRA-supported program should aim at external sustainability while fostering sustainable growth. That is, external sustainability should be pursued with the help of pro-growth measures; but measures that would be aimed at increasing economic growth, but not having a substantial impact on the sustainability of the balance of payments (or where the latter is not at issue), would not be appropriate.31 By focusing conditionality on areas closely linked to external viability—and such conditionality in general will support improved long-term growth—the guidelines aim at focusing the Fund’s attention on issues central to its mandate (and where its expertise is greatest), while leaving maximum policy space to the authorities in other areas.

43. Poverty reduction (in GRA-supported programs) and employment are other objectives that do not obviously delineate appropriate areas for Fund conditionality. While poverty reduction is clearly a goal of PRGF-supported programs, the only guidance in this regard for GRA-supported programs is the proposition in the staff statement that sustainable growth should be (inter alia) equitable (¶6). Poverty reduction can thus be seen as an issue to be kept in mind in the design of GRA-supported programs, rather than an intended focus of them. Employment promotion is not mentioned in the guidelines as an appropriate goal and focus for conditionality for Fund-supported programs; it is presumably a feature or consequence either of growth or of poverty reduction, since measures beneficial in these areas are likely also to increase employment.

44. In line with the guidelines, a large majority of requests for use of Fund resources in the GRA cite macroeconomic stability as a goal, while most requests for PRGF-supported programs cite stability, growth, or both. About one-half of GRA-supported programs and three-fifths of PRGF-supported programs cite macroeconomic stability as a goal, with one-third of the former and only one in ten of the latter specifying external stability.32, 33 Growth is cited as a goal in about half of PRGF-supported programs.

45. Many requests for use of Fund resources in the GRA also cite growth as a program goal. Indeed, the share of GRA-supported arrangements citing growth is as high as that for PRGF arrangements—about half. In all but three GRA-supported programs (Brazil, Croatia, and Peru), growth is cited along with external or macroeconomic sustainability as the program goal—as would be expected since growth is often a critical element in improving debt sustainability—and thus the overall goals are in line with the guidelines. Moreover, inclusion of growth as an independent program goal does not seem to lead to differences in the formulation of conditionality, suggesting that citing growth as a separate goal is likely to be a statement of solidarity of the Fund-supported program with the authorities’ wider goals, rather than representing a failure to focus specifically on areas critical to external viability.34

46. Some GRA-supported programs also cite poverty reduction or employment as goals, though without an effect on the conditionality being set. Poverty reduction is cited in a few GRA-supported programs, all in the Western Hemisphere department (WHD), and employment in a slightly wider group. Conditionality in these cases does not differ in any significant way from conditionality in most other GRA-supported programs, and, for the poverty reduction cases, is less focused on fiscal reforms than conditionality in PRGF arrangements in WHD. It therefore seems likely that inclusion of poverty reduction or employment in goals statements signals the concern that both the authorities and the Fund place on poverty reduction as a long-term objective, rather than a focus for Fund conditionality.

C. Strategies

47. About two-thirds of requests for arrangements cite at least one strategy to achieve program goals. Strategy here refers to an area of structural reforms where the program’s conditionality will be focused, such as the financial sector, public expenditure management, or governance. Strategies differ from program goals by constituting intermediate steps between what programs expect to attain (an improved external position, higher growth) and the measures selected to get there.35 This leaves about one program in three where no specific strategy for reaching program goals is mentioned. Conditionality in those cases is not significantly more scattered than in others. The lack of an explicitly specified focus for conditionality may thus be merely an oversight. However, the absence of a presented strategy deprives the Board of a full justification for program conditionality.36

48. Not surprisingly, most programs cite more than one strategy. Consistent with the fact that countries rarely face obstacles of only one sort in the path toward external stability or higher growth, only about one-third of programs that enumerate any strategy cite one focal strategy (perhaps unsurprisingly, in most of these the focus is fiscal). Indeed, most programs cite two or three strategies, and two programs (Brazil’s 2001 Stand-By Arrangement and Tajikistan’s 2002 PRGF arrangement) cite four strategies each. Multiple strategies raise concerns about whether such far-ranging tasks can be reconciled with parsimony and focus. However, conditionality for Brazil and Tajikistan was more focused than implied by the statements on economic strategies.

49. The strategies laid out in Fund documents to achieve program goals vary widely, with areas of concentration lying both in and out of the Fund’s main areas of expertise. Statements of program strategies can be fairly easily grouped into a small number of categories. Three-quarters of strategies fall into five areas: fiscal policy, monetary policy, financial sector reform, governance, and private sector growth. The first three are also areas of concentration of conditionality (see Chapter III); the last two are explored in more detail below. The remaining areas cited are exchange rate policy, inflation, public expenditure management, privatization, central bank reform, trade policy, and the vague categories of “institutional reform” and unspecified “structural agendas.”

50. Of these, fiscal reform is the most commonly cited strategy, particularly among GRA-supported programs, and this focus carries over into conditionality. Of those programs citing strategies, only two GRA-supported programs in the period analyzed (for Bulgaria and Romania) did not cite fiscal reforms as a strategy.37 Fiscal reform strategies extend beyond deficit reduction commitments buttressed by quantitative conditionality. In all GRA- supported programs citing fiscal policy as a program focus, conditionality included at least some measures in revenue enhancement, expenditure control, or other fiscal areas. A smaller share of PRGF arrangements cite fiscal strategies (14 of 25 indicating any strategy), and again in all cases structural conditionality included fiscal reforms.

51. PRGF-supported programs frequently cite improvements in the business environment and other “pro-growth” reforms—more frequently than do GRA-supported programs. Almost half of PRGF arrangements cite this strategy, while only about one in five GRA arrangements go the same route. It is not clear, however, how citing this goal affects program conditionality. Indeed Figure 8, which summarizes the distribution of the number of growth-related conditions in programs that do and do not cite a growth strategy, suggests that a growth strategy does little to predict conditionality in this area (defined according to the classification presented in Chapter III).38 For example, Tanzania’s 2003 PRGF arrangement request cites improvement of the business climate as a strategy, but includes only a benchmark on business licensing in that area. The implicit assumption may be that the stability afforded by quantitative conditionality and progress in structural reforms (particularly in governance and the financial sector) will lead to an improved climate for growth, and indeed the recent study of program design confirmed the importance of macroeconomic stability for growth.39 But, as noted in Chapter III, it is also possible that the increasing concentration on the Fund’s core areas has begun to usher in a lesser focus on growth.

Figure 8:
Figure 8:

Distribution of Growth-Related Conditionality

(Number of conditions per program year)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

52. Some PRGF-supported programs cite governance as a focus, but seem to include no relevant conditionality, and the opposite is also quite common (Figure 9). Governance-related reforms are widespread in PRGF-supported programs (about two-thirds of all programs include at least one condition in the area), but far fewer programs cite governance as a strategy—five out of 25 PRGF-supported programs citing strategies.40 Programs citing governance as a strategy tend to have slightly more conditions in this area than programs that do not do so, but some appear to include no conditions in the area.41 Conversely, programs that do not cite governance as a strategy still tend to have conditions in this area. In sum, this strategy is sometimes cited but not followed, or alternately, followed but not cited. This disconnect may point to something of a gap between, on the one hand, the widespread perception of the need to improve governance in many PRGF-eligible countries, and, on the other hand, the difficulty of identifying measures that will both make a real difference and will be monitorable by Fund staff.

Figure 9:
Figure 9:

Distribution of Governance Conditionality

(Number of conditions per program year)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

53. Financial sector reforms are also frequently cited as a focus, though not as frequently as they are included in conditionality. As discussed in Chapter III, financial sector measures have become very common in GRA-supported programs, although they are cited as a strategy in only 25 percent of programs indicating strategies. Despite their less-developed financial systems, the percentage of requests for PRGF arrangements citing financial sector reforms as a focus is only slightly below the share of GRA-supported programs. Conditionality in this area generally tracks stated strategies, in both GRA- and PRGF-supported programs. However, consistent with the increasing focus on conditionality in the area of vulnerability, more programs seem to have conditions in financial sector reform than include it in statements of strategy: arrangements as varied as Nicaragua’s PRGF arrangement and Ukraine’s SBA include conditionality in the financial sector, but do not cite it as a specific focus. Figure 10 confirms the widespread nature of conditionality in the financial sector, with very few programs including no conditionality in this area, even among those that do not cite this strategy.

Figure 10:
Figure 10:

Distribution of Financial Sector Conditionality

(Number of conditions per program year)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

54. The inclusion of conditions in areas not cited as program strategies raises the question of how conditions were chosen. Such cases do not seem to identify strategies to attain program goals, and then determine the critical reforms. Strategy statements presumably represent the broad focus and most crucial areas for conditionality, and do not describe all conditions. Clearly, some reforms outside of a program’s broad-stroke strategy may nevertheless be critical to program goals. At the same time, programs in which strategies are laid out and addressed, but extensive conditionality is included in other areas, raise questions whether strategies were properly presented, or whether the program had a broader focus than stated.

55. Moreover, program documents do not generally explain how strategies are chosen. In the request for Lesotho’s 2001 PRGF-supported arrangement (Country Report 01/79, 6/12/01,, the program goal of supporting Lesotho’s return to external sustainability is laid out, accompanied by a box explaining the connection between Lesotho’s untenable fiscal position and its worsening current-account deficit (see Box 1). This analysis, explaining how particular strategies have been chosen and why, is not often presented. In many cases, such analysis may not be necessary: the causes of a particular balance of payments imbalance may be clear and explained in background sections of staff reports. But in many cases, particularly in countries with PRGF arrangements, the move toward more parsimonious conditionality is likely to raise the issue of why particular strategies have been chosen.

Justifying Strategies: The Case of Lesotho

Lesotho’s 2001 request for a PRGF-supported arrangement included a box explaining the relationship between the fiscal and external balances in the country. The text, shown below, was accompanied by a chart showing the high correlation between the (declining) fiscal surplus and the worsening current-account deficit in the country.

Lesotho is a small, open economy; there is a close correspondence between the fiscal balance and the external current account balance (see figure below). In the 1990s, movements in the current account balance (excluding flows on account of the Lesotho Highlands Water Project), in percent of GDP, mirrored those in the fiscal balance. The relationship between the two appeared to have weakened in the late 1990s. However, this reflected the fact that the deterioration in the fiscal balance in this period stemmed mainly from noncash expenditures (issue of bonds in 1999 to the bank that took over the deposit liabilities and good assets of the Lesotho Bank) and higher debt-service payments on behalf of the Muela hydroelectric power project. When these two factors are taken into account, the close relationship between the two variables is maintained.

The strong relationship between the fiscal balance and the external current account balance reflects the important role of the government in the economy, the high proportion of traded goods in the economy, and the high degree of capital mobility. Government expenditure is a major component of domestic demand, the bulk of which is satisfied through imports, while monetary policy has been unable to sterilize fiscal operations. One of the objectives of the program is to develop the central bank’s capacity to influence liquidity conditions and thus the balance of payments.

This box clearly lays out the main external vulnerability facing the member country, and ties the focus of the Fund-supported program, in this case, fiscal reforms and capacity-building at the central bank, to this goal.

D. Conclusion

56. Almost all program requests state clearly the program’s goals, and these are generally in line with the guidelines. Among PRGF-supported programs, almost all requests cite program goals as growth and poverty reduction, while almost all GRA-supported programs cite macroeconomic or external stability. Moreover, approximately two-thirds of requests clearly state the goals of the program supported by the Fund, which are likely to be a subset of the broader goals of the authorities. While some GRA-supported programs include growth as a separate goal (independent from external sustainability), and some cite poverty reduction and employment goals, this does not seem to affect conditionality.

57. Most programs also indicate a strategy or strategies the authorities will follow to attain the stated goals, and conditionality often tracks these strategies, but sometimes conditionality spreads well beyond those areas, or not into those areas at all. These strategies, intended as foci for conditionality, are often in areas of Fund expertise, and conditionality tends to be focused in those areas. Strikingly, those frequently cited strategies that are outside the Fund’s core expertise—“pro-growth” reforms and governance—seem to predict conditionality less well than do core strategies. A disconnect may exist between what is considered important and what can be specified in terms suitable for inclusion in conditionality. Additionally, conditionality quite often extends into areas not cited as strategies, raising the question whether all the relevant measures are truly critical. Program requests do not generally explain why a strategy has been chosen, thus making it more difficult to evaluate the focus of conditionality.

III. Breadth of Coverage of Structural Conditionality42

A. Introduction

58. Focused coverage of conditionality is a key feature of the 2002 conditionality guidelines: conditions should be limited to measures critical to the success of the program.43 This chapter assesses developments in the coverage of structural conditionality since the guidelines were adopted. To this end, structural conditionality is classified into an institutional and an economic classification.44 The first classification serves to examine developments in conditionality that pertain to the Fund’s technical expertise. The second classification helps to identify links between the choice of conditionality and initial economic conditions.

59. Any classification inevitably involves an element of arbitrariness. Some structural reform measures may fit more than one category. For instance, measures that change the tax structure may be needed for economic stabilization, but are also likely to affect economic efficiency. As another example, reforms in the forestry sector aim at removing bottlenecks that affect the development of this sector; yet, in some member countries, the importance of this sector for fiscal revenues provides these reforms with an additional motivation.

60. Equally important, the analysis of sectoral coverage suffers from the same methodological limitations as all counts of numbers of conditions. Counting conditions is a crude metric, not least because it weights equally conditions of quite different import. The proposed classifications are a methodological construction used to examine the trend changes over time in the share of conditionality in different categories. They do not, however, address the limitations that arise from the lack of link between the number of conditions and the quality of reforms.

61. This chapter is organized as follows. A review of the distribution of conditions based on an institutional classification is presented in Section B. The following section discusses a classification based on the economic aspects of structural conditionality. Section D examines the links between conditionality and initial economic conditions. Some concluding remarks follow.

B. Institutional Classification


62. The proposed institutional classification attempts to map structural conditionality into the Fund’s areas of core responsibility. The implicit assumption is that reforms related to the Fund’s mandate and technical expertise are likely to be more critical to macroeconomic stability and external sustainability, which are the primary goals of Fund-supported programs. Indeed, recognizing this likelihood, the guidelines state that “conditions will normally consist of macroeconomic variables and structural measures that are within the Fund’s core areas of responsibility.” Notwithstanding the above, it is important to highlight that the guidelines do not preclude setting conditionality in other areas, and indeed require it if those areas are critical to the success of the Fund-supported program. Conditionality might be necessary, for instance, in areas aimed at reducing financial vulnerabilities, and in noncore areas that are judged crucial for achieving specific program objectives (e.g., energy sector reforms might be required to address a key source of fiscal imbalances and ensure a sustainable allocation of resources in the economy).

63. More precisely, structural policies are mapped into three categories: core areas of Fund expertise, reforms where expertise is shared with the World Bank, and structural policies in noncore Fund areas. The content of each category can be described as follows.

  • Core reform areas refer to measures linked to fiscal, monetary, and exchange rate policy. This category covers all aspects of fiscal policy, from tax policy and administration to expenditure control and management.45 Policies aimed at improving the transmission mechanism of monetary policy and strengthening the signaling role of exchange rates are also included.

  • Shared reforms areas include measures that are linked to the Fund’s mandate, but in which both the World Bank and the Fund have complementary technical expertise. Included in this category are financial sector reforms and reforms that strengthen the prospects for private sector development (from trade liberalization to policies aimed at supporting an enabling business environment). The degree to which these are shared between both institutions varies; there is more involvement by the Fund in financial sector issues than there is in reforms aimed at strengthening private sector activity. Governance-related reforms are also classified within this category, though the Fund’s role is more limited than that of the Bank.

  • Noncore reforms are outside the Fund’s areas of expertise and relate to measures where the World Bank usually takes a leading role. These structural reforms (and corresponding conditionality) include poverty-related policies and measures that deal with weak public enterprises and civil service reform. Measures that address structural weaknesses in specific economic sectors (e.g., agricultural markets) are also classified within this category.

64. The classification along the above lines reveals that there has been a marked shift toward shared and core areas among Fund-supported programs over the past few years (Figure 11 and Table 4). The shift in GRA-supported programs has been out of noncore areas of expertise and has led to a reduction of 17 percentage points in the share of this category in Fund conditionality since its peak in 1995-97. This declining trend began already in the late 1990s; initially from noncore areas toward shared areas and, more recently, toward core areas. The shift among PRGF arrangements was similar to that of GRA-supported arrangements—a decline of 18 percentage points, concentrated in this case in the 2001-03 period (13 percentage points).

Figure 11:
Figure 11:

Institutional Classification of Structural Conditionality

(In percent of the total number of conditions)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

Table 4.

Institutional Classification of Structural Conditionality 1/

(excluding prior actions)

article image
Source: MONA database and staff estimates.

The number of conditions is adjusted by program duration.


65. The implementation record of Fund-supported programs reveals a minor improvement in recent years among PRGF-supported programs and a largely unchanged implementation record in GRA-supported programs—a record which is only marginally affected by the shifts in the coverage of conditionality. Shifts from sectors with traditionally weak implementation records to sectors with stronger implementation could strengthen aggregate implementation substantially (and vice versa for the opposite shifts), and the core and shared areas conditionality has shifted into traditionally have a stronger implementation record than noncore areas, suggesting that composition effects may be important. To understand better the forces behind the changes in the implementation index,46 this index is decomposed into a design effect and a composition effect.47 The former reflects genuine improvements in implementation after controlling for the effects that arise as a result of shifts in the sectoral coverage of conditionality—the composition effects. As is evident from Table 5, GRAsupported programs have registered a very small deterioration in their design (or “pure”) implementation record (-0.02), but this deterioration is counterbalanced by an equally small, but positive (0.01), composition effect that arises from the shifts in the sectoral coverage of conditionality. Among PRGF-supported programs the implementation record shows some improvement in the 2001-03 period. Composition effects explain about a third of this improvement, and the rest can be attributed to genuine improvements in implementation; still, both the aggregate changes as well as the individual effects are small. In sum, though neither the composition nor the design effects are large, the decomposition into these two categories provides a more accurate depiction of implementation, and it will be important to follow these effects in future, as more experience accumulates.

Table 5.

Decomposition of Aggregate Changes in the Implementation Index

(Changes between 1998-2000 and 2001-2003, institutional classification; excludes prior actions)

article image
Source: MONA database and staff estimates.

C. Economic Classification


66. The economic classification serves to assess the choice of structural conditionality. Although data limitations preclude an assessment based on economic outcomes since the introduction of the guidelines, the proposed classification allows an examination of the links between conditionality and initial economic conditions.

67. The economic classification is largely the same as in Macroeconomic and Structural Policies in Fund-Supported Programs (11/24/04,; measures that strengthen economic management, those that address vulnerabilities, and supply-side conditions—in turn, the latter is broken down into three supply-side sub-categories. The main components of this economic classification can be described as follows:

  • Reforms that underpin a medium-term framework for economic management. These policies are designed to enhance stabilization and to strengthen the functioning of fiscal, monetary, and exchange rate policies. In particular, the aim is to ensure that adjustment efforts are effective, credible, and sustainable. Reforms in the fiscal area include measures that improve the tax structure—including widening the tax base—and tax administration, as well as policies that improve public expenditure management. Deepening markets for government securities and expanding the menu of instruments available to the monetary authorities are also included, as these provide a more stable environment for conducting monetary policy. Measures that strengthen the functioning and depth of foreign exchange markets are also part of this category.

  • Policies that address economic and financial vulnerabilities. These policies are directed at tackling unsustainable public or external debt dynamics, reducing the vulnerability of domestic balance sheets to sharp swings in exchange or interest rates, and other financial sector weaknesses. Strengthening prudential regulations and financial sector supervisory capabilities form an important element of this category and are a central aspect of many new programs.

  • Measures that enhance economic flexibility and efficiency, which are important to strengthen growth, increase adaptability to shocks, and improve the delivery of public services. The economic flexibility category includes measures that increase the ability to adapt to new conditions—including trade liberalization reforms and policies that affect resource allocation across most economic sectors, such as pricing policies that affect factor markets (labor and capital), the institutional aspects of these markets, and pricing policies with economy-wide effects (e.g., energy sector prices). The private sector efficiency category refers to impediments to investment and growth and reforms that affect individual sectors, such as marketing policies in agricultural markets. Privatization of state enterprises and utilities also falls into this category, though these measures can have other objectives as well, such as controlling quasi-fiscal costs.48 The public sector efficiency category relates to improvements in the delivery of public services. These three reform areas are also referred to as supply-side conditions.

68. As noted in Macroeconomic and Structural Policies in Fund-Supported Programs, the distribution of structural measures in GRA- and PRGF-supported programs in the 1995-2000 period mimicked the goals of these programs. In GRA-supported programs, reforms were evenly split across all categories, while PRGF-supported programs were slightly more focused on economic efficiency and flexibility—albeit marginally so, as the difference was not statistically significant except among transition economies. This alignment with the growth goals of PRGF programs is consistent with the correlation identified in Macroeconomic and Structural Policies in Fund-Supported Programs between growth- and efficiency-related structural measures and the growth performance of Fund-supported programs approved in the period 1995-2000.

69. The data for programs approved in 2001-03 suggest that the GRA-supported programs shifted sharply away from flexibility- and efficiency-related measures since the guidelines came into effect (Figure 12 and Table 6). This shift has been primarily toward vulnerability conditions, though there is also an increased focus on economic management. This shift appears to be in line with the growing understanding of the importance of addressing financial and other balance sheet vulnerabilities in members that use Fund resources under the GRA.

Figure 12:
Figure 12:

Economic Classification of Structural Conditionality

(In percent of the total number of conditions)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

Table 6.

Economic Classification of Structural Conditionality 1/

(excluding prior actions)

article image
Source: MONA database.

The number of conditions is adjusted by program duration.

70. In PRGF-supported programs the shift has been toward economic management and away from supply-side reforms. A possible explanation for this shift is that the improved growth performance of recent years has resulted in a focus on economic management and less need for additional growth-related reforms. Some evidence in this direction is presented later in this section—namely, high growth has been accompanied by a lesser focus on some supply-side reforms, and this correlation has strengthened in the 2001-03 period. A second plausible explanation is that “low-hanging fruits” with quick growth returns have already been harvested, thus leaving fewer reforms to be tackled. At the same time, the guidelines, in particular through their emphasis on conditions that are critical, may be limiting conditionality on measures that yield benefits over longer periods of time. Even if the near- and medium-term costs of these reform gaps might be low (judgment must await availability of outcomes data), this shift may have longer-term consequences. This will require watching in coming years. It is also important to note, however, that the state of economic knowledge on measures that sustain growth is limited. For example, while there is broad agreement that institutions are of critical importance for growth, it is less clear what are the policies that serve to strengthen a country’s institutional framework, particularly given the different experiences and institutional arrangements that characterize high-growth countries. Perhaps a combination of these explanations is at play, suggesting that the decline in supply-side conditionality warrants being kept under scrutiny.

71. The shifts described are present—with only minor differences—both among transition and non-transition economies. Specifically, there is a marked shift toward vulnerability-related conditions among GRA-supported programs, though the shift among transition countries is weaker (Figure 13). Indeed, these countries end up having a slightly larger share of economic management, rather than vulnerability-related, conditionality. The differences between transition and non-transition economies in PRGF-supported programs are also small: a few differences in levels exist, but these do not affect the nature of the shifts observed in recent years—namely, toward economic management and away from private sector efficiency.

Figure 13:
Figure 13:

Economic Classification in Transition and Non-Transition Economies

(In percent of the total number of conditions)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

1/ Only an average of five PRGF arrangements exists among transition economies in each 3-year period.

72. The shift in the distribution of conditionality also plays an important part in the lack of decline in the number of conditions among GRA-supported programs. In particular, the distribution of vulnerability-related conditionality has registered significant increases as depicted in the box-and-whiskers chart (Figure 14). Vulnerability-related conditionality has also increased among programs with relatively low vulnerability-related conditionality—the number of conditions in the lower 25th percentile (i.e., the whiskers to the left of the median in the figure) have doubled in the last 3-year period compared with the previous 3-year period, while the median has increased by much less, suggesting an increase in the dispersion of vulnerability-related conditionality.

Figure 14:
Figure 14:

Distribution of Vulnerability-Related Conditionality

(Normalized per program year; GRA-supported programs only)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

The boxes in these charts represent the range of the 25-75 percentiles in the number of conditions and the vertical line in each box represents the median values. The whiskers (lines extending beyond each end of a box) reflect the low and high adjacent values. These adjacent values are calculated as a multiple of the difference between the upper and lower quartiles, P75 and P25, also known as the interquartile range (or IQR = P75 - P25). Specifically, the adjacent values are the highest existing value not greater than P75 + 1.5 * IQR and the lowest value not less than P25 - 1.5 * IQR (Tukey, 1977). Outliers in each category are represented by dots and reflect observations that are beyond the calculated adjacent values.

73. The evidence suggests that vulnerabilities have played a large role in conditionality among GRA-supported programs. Vulnerability-related measures are on an upward trend vis-à-vis all other measures, and the number of measures is increasing in complementary reform areas among “vulnerability” programs. Table 7 presents data on the number of conditions for all GRA-supported programs and for two groups of GRA-supported programs: “non-vulnerability” and “vulnerability” programs—that is, programs that have a below-average number of vulnerability conditions and those that have an above-average number of these conditions.49 At an aggregate level, vulnerability-related conditionality increased by 1.6 conditions per program year in 2001-03 relative to the previous three-year period, while all other reform areas registered a reduction of 1.7 conditions per program year. In “vulnerability” programs, the total number of conditions has increased by as much as 35 percent—perhaps in part on account of the extent of detail that can characterize such programs (see Box 2). These programs are, moreover, the sole source of increase in conditions outside the vulnerability category (3.4 additional conditions), a large share of which (2.4 conditions) is related to economic management (including fiscal) reforms, many of which are linked to efforts to address debt dynamics. Indeed, while conditions among “non-vulnerability” programs declined by 19 percent, even among these programs there was an increase in the number of vulnerability conditions (0.8 additional conditions per program year, compared to a decline of 4.6 conditions in all other reform areas). In sum, vulnerability-related conditionality and vulnerability programs are an important reason why the number of structural conditions among GRA- supported programs did not decline during the 2001-03 period.50

Table 7.

Developments in Vulnerability-Related Conditionality

(Normalized per program year; GRA-supported programs)

article image
Source: MONA database.

Structural Benchmarks in Brazil’s Financial Sector (2002 Stand-By Arrangement)

By end-September 2002

Develop a streamlined set of indicators from the financial data currently reported to the Off-Site Banking Supervision Department of the central bank to assess the economic and financial soundness of banks.

Implement a systematic assessment of credit risk level of the largest borrowers in the financial system.

Develop a monitoring system to report significant changes in key accounting data such as credit levels and composition, credit portfolio stress risks, and risk diversification.

By end-December 2002

Further progress in the auctioning of the four remaining federalized state banks.

Completion of a review of differences between Brazilian and internationally-accepted practices for the use of independent external bank auditors, in order to evaluate the appropriateness of implementing such international practices.

Completion of the updating of existing regulations regarding licensing of financial institutions, including the adoption of a multistage approach to the licensing process; the introduction of a requirement that firms present detailed operating plans before licensing, including information on corporate organization and structure, internal controls, and corporate governance; and the requirement that the licensing process include consideration of the impact of the proposed new financial institution on market concentration and competition.

Develop a streamlined set of indicators from the financial data currently reported to the Off-Site Banking Supervision Department of the central bank to assess the economic and financial soundness of nonbanking financial institutions.

By end-May 2003

Progress towards the passage of a new Bankruptcy Law aimed at protecting creditor rights and a quicker restructuring of distressed businesses.

By end-June 2003

Progress towards sale of the four remaining federalized state banks.

By end-September 2003

Start of operations of the new primary and secondary dealer system and the development of the sale of public debt over the internet through the Tesouro Direto program.

By end-March 2004

Implement the Provisional Measure to allow workers to pledge a fraction of their future wages to the repayment of consumer loans and expand the arrangement to include retirees in the public pension system via a National Social Security Institute (INSS) ruling.

Introduce legislation to Congress to expand the scope of portfolio adjustments that are free from the bank debit tax (CPMF).

By end-June 2004

Enable loan applicants and banks to have access to the central bank's centralized credit rating system (SCR, Cadastro Positivo), with a view to make available to competing banks information on borrowers credit profiles.

Implement the bankruptcy law (after congressional approval) by providing training as outlined in paragraph 11 (later reset for end-December).

74. Moreover, reflecting in part the increased focus on vulnerability, both GRA- and PRGF-supported programs show evidence of greater concentration in the coverage of conditionality. This is represented in Table 8 by the average number of distinct reform categories in which conditionality is being set. The average number of supply-side categories with conditionality has declined, in GRA arrangements, from a peak of 3.6 categories (out of a total of 10 possible supply-side categories) to 2.8 in the 2001-03 period and in PRGF arrangements, from 5.2 to 2.9 categories over the same period. In sum, there are fewer and more focused reform areas within each individual policy program—a development that points to some “streamlining” in GRA-supported programs even if the number of conditions is not declining.

Table 8.

Concentration of Conditionality in Distinct Reform Categories 1/

(Average number of categories with conditionality)

article image
Source: MONA database.

A total of 14 different categories; 4 related to macroeconomic and financial sector reforms, and 10 related to supply-side conditions.


75. As was the case with the institutional classification, implementation appears to have improved, albeit marginally, in some of the areas of increased conditionality among PRGF-supported programs, but remains broadly unchanged in GRA-supported programs. Specifically, the implementation of economic management and vulnerability measures in GRA-supported programs remains as in the pre-guidelines period (Table 6), but shows some improvement in PRGF-supported programs.51 In contrast, among reforms aimed at strengthening private-sector efficiency, the implementation record has weakened. Although the reasons behind this deterioration are unclear, the deterioration has been particularly strong in reform areas outside the Fund’s areas of expertise and where it is more difficult to define conditions given the numerous unknowns involved, in part because these reforms are both politically as well as technically demanding. For example, implementation of public enterprise reforms has deteriorated sharply, perhaps because the Fund is countenancing greater deviations from original plans in either timing or design, though many of these reforms do eventually get implemented.

76. The decomposition of the implementation index into its composition and design effects provides similar conclusions as those described for the institutional classification; namely, as is evident from Table 9, GRA-supported programs have registered a very small deterioration in their design (or “pure”) implementation record (-0.03) and this deterioration is counterbalanced by an equally small, but positive (0.03), effect that arises from the shifts in the sectoral coverage of conditionality. Among PRGF-supported programs composition effects explain about a third of the improvement, and the rest can be attributed to genuine improvements in implementation; still, both the aggregate changes as well as the individual effects are small.

Table 9.

Decomposition of Aggregate Changes in the Implementation Index

(Changes between 1998-2000 and 2001-2003, economic classification; excludes prior actions)

article image
Source: MONA database and staff estimates.

D. Conditionality and Initial Economic Conditions

77. Conditionality is in some key respects aligned with initial economic conditions. The regression results (Table 10) contain two interactive dummy-based variables related to initial economic conditions; the first variable represents initial economic conditions in the preguidelines period and the second those of Fund-supported programs in the post-guidelines period. Controls are added for area department, program type, and macroeconomic variables.52 The results indicate that:

  • More conditions are now specified to address vulnerabilities when the initial external debt levels are high. This relationship is not observed in pre-guidelines programs, perhaps denoting the recent emphasis on financial and balance sheet vulnerabilities.

  • High debt levels do not affect the prevalence of supply-side conditions, such as economic efficiency conditions. While the latter is surprising in that growth can be a crucial component of a country’s efforts to maintain (or reduce) debt to sustainable levels, some of the same reasons that underlie the decline in growth-related conditionality in PRGF- supported programs (see ¶70) may be at play—including perhaps the difficulty of identifying actions that can be confidently predicted to make a real difference to growth.53

  • Since the guidelines were introduced, a strong fiscal performance prior to the program year has resulted in fewer structural conditions in the economic management category. This could reflect increased selectivity in conditionality. In contrast, high inflation in the pre-program period does not affect the number of economic management conditions (either in the pre- or post-guidelines period).

  • Pre-program growth rates affect the choice of conditionality. Specifically, poor growth performance is correlated with more conditions directed toward private sector efficiency. Interestingly, the link between pre-program growth rates and efficiency conditions appears to be quantitatively stronger in the post-guidelines period (the coefficient increased from -0.15 to -0.53). The link between growth and growth-related conditionality is also consistent with the decline in growth-related conditions in PRGF countries, as their growth performance has improved in recent years.

Table 10.

Number of Structural Measures and Initial Economic Conditions: Regression Results 1/

article image
Sources: International Monetary Fund, MONA; and IMF staff estimates.Note:

= significant at 10% level,

= significant at 5% level,

= significant at 1% level.

Regressions based on a dataset of programs approved during the period 1995-2003. Controls for initial economic conditions are included; these controls are instrumented using WEO actuals regressed on MONA projections (current account, fiscal balance, inflation are among the list of instrumented macroeconomic indicators).

Equal to 1 if a PRGF-supported program.

F-statistic for null hypothesis that all explanatory variables (other than the constant) are jointly equal to zero.

E. Conclusions

78. The distribution of conditionality has shifted over the past few years. The shift has been out of noncore reforms and toward areas of greater Fund expertise. The shift has also been toward economic management and vulnerability categories. The shift toward economic management has been more pronounced for PRGF-supported programs, while GRA-supported programs have witnessed a greater shift towards vulnerability-related conditionality. In addition, conditionality has been set on fewer reform categories than in earlier periods.

79. Supply-side conditionality in PRGF-supported programs has declined. Several explanations are possible. For example, higher growth has typically resulted in fewer efficiency-related conditions; this correlation has become quantitatively stronger over time. Another explanation is that fewer reforms with quick growth returns remain, limiting the agenda to be tackled in Fund-supported programs. The shift in conditionality may have also been exacerbated by the new guidelines; the focus on parsimony, criticality, and ownership might result in less attention to growth issues that stretch beyond the program period. A more benign interpretation, however, relates the decline in supply-side conditionality to the lack of understanding about what measures might serve to promote growth. All of these factors are likely at play. Still, the decline is striking and warrants being kept under close scrutiny.

80. Composition effects have had an impact on the numbers of conditions in Fund- supported programs, but this has not been at the expense of program focus. Indeed, both vulnerability-related conditionality and programs have been a source of the lack of decline in the numbers of conditions in GRA-supported programs, thus highlighting the role of composition effects. Still, as noted above, conditionality has been set in fewer reform categories.

81. Finally, a stronger relation of structural conditions to the economic situation at the start of the program is observed in the post-guidelines period. Specifically, initial debt levels (higher debt results in more vulnerability conditions), fiscal performance prior to a program (weak fiscal positions result in more economic management conditions), and pre- program growth rates (high growth leads to fewer supply-side conditions) influence the number and distribution of conditions in Fund-supported programs. Moreover, these links have either emerged or become quantitatively more important in the post-guidelines period.

IV. Use of the Tools of Conditionality54

A. Types of Conditions—Structural Performance Criteria and Structural Benchmarks

82. The implementation of structural conditionality is monitored on the basis of prior actions, structural performance criteria, and structural benchmarks. This section considers the choice between structural performance criteria and structural benchmarks.

83. The guidelines specify rules for the use of structural performance criteria and benchmarks in order to ensure that usage of both is consistent with the basic principles of parsimony, criticality, and ownership. The stipulations for the two types of conditions are as follows:55

  • Performance criteria (PCs) must by themselves be so critical that purchases or disbursements under the arrangement should be interrupted in case of nonobservance, and must be objectively monitorable.

  • Structural benchmarks (SBs) are used in instances where a condition cannot be specified in terms that are objectively monitorable, or where its non-implementation would not, by itself, warrant an interruption of purchases or disbursements under the arrangement—that is, as clear markers in the assessment of progress in the implementation of critical structural reforms in the context of a program review.

Recent Developments

84. The combined number of structural PCs and structural benchmarks has followed different trends in PRGF and GRA-supported arrangements. While the total number of these conditions has remained fairly stable among PRGF arrangements since 2000 at about 11-12 conditions per annum, the figure has risen since 2002 among GRA-supported arrangements to some 15½ conditions per annum, from about 13½ in 2000-02.

85. These divergent trends are explained largely by the behavior of structural PCs, which have remained broadly stable in PRGF-supported programs since 2000, but have risen in GRA-supported arrangements since 2002.56 This latter increase is in part traceable to measures intended to shore up weak financial systems in a number of Latin American countries (Bolivia, Dominican Republic, Box 3), and was supported by Executive Directors.

Countries with Extensive Structural Performance Criteria in 2003-04

While the number of performance criteria in GRA-supported arrangements was fairly stable over the 2000-02 period, it rose sharply in 2003, with only a slight decline in 2004. Four countries with the highest number of structural performance criteria in 2003 are considered below to illustrate some of the possible reasons for this rise.

Conditionality in Bolivian programs has increased sharply in the current program. In the final year of the Bolivian PRGF arrangement in 2001, conditions were pared down substantially, and this action was criticized by the Executive Board. Since then the Bolivian banks have suffered multiple deposit runs, weakening the financial system.

In response, the SBA-supported program, which began in 2003, focused on weaknesses in the financial sector and on fiscal discipline, with particular emphasis on lowering the fiscal deficit to ensure debt sustainability. To address financial sector vulnerabilities, the program set structural PCs on regulations for bank resolution, clarifying the roles of the various institutions with oversight over the financial sector, adopting an action plan to strengthen weak banks, and submitting to Congress a draft bankruptcy law. On fiscal policy, structural PCs were set on the submission to Congress of a tax procedures code and on the issuance of associated regulations, both of which aim to develop a personal income tax in Bolivia. This conditionality was justified by reference to the difficulty of the situation, and the Board shared the view that it was appropriate. Indeed, some concerns were expressed that the passage of the tax procedures code was a benchmark and not a PC, and that future structural conditionality was light. The implementation of conditionality was comparable to the average among programs, with delays in the implementation of the structural PCs and some structural benchmarks not implemented as yet.

Ecuador is another country where performance criteria have increased over the past two programs. The 2000 program was based on supporting the official dollarization of the economy, addressing weaknesses in Ecuador’s banking system through strengthening supervisory regulations and raising capital adequacy ratios, achieving a more flexible labor market, and facilitating the privatization of major state enterprises.

Following a succession of weak governments, a political outsider was elected in late 2002 on an anticorruption platform. This change in government kindled interest in a new Fund-supported program based on strengthening the fiscal balance and reducing rigidities in fiscal policy, cleaning up the banking system, and modernizing state enterprises. Nine structural PCs were set for the first year of the program in the areas of civil service reform, customs administration, cost-effective production and sale of fuels, the foreign transfer of the management of the electricity distribution companies, and passage of tax reform law, in addition to banking measures on liquidating banks and returning blocked deposits to depositors.

Although the structural conditions in this program were wide-ranging, the Board was quite positive on the breadth of conditionality, emphasizing that all the measures had fiscal implications, and that the program faced significant political risks. The concerns voiced about the specifics of conditionality were limited to the likelihood of achieving necessary and substantive civil service reform in a fractured political environment.

In the event, most of the structural PCs were missed, requiring corrective measures and a conversion of some of the conditions into prior actions. Implementation of structural reform remained weak, partly associated with the government’s unpopularity and limited consensus in policy making. The arrangement expired in April 2004, with only a third of the committed financing having been disbursed.

Until 2003, the Dominican Republic (DR) had enjoyed rapid growth and a stable macroeconomic environment. It had not had a Fund-supported program (except under emergency assistance) for 10 years. The global slowdown in 2002 and 2003 heavily affected DR because of its high exposure to the world economy. Rather than initiate bank reform, the authorities chose to ease monetary policy, which may have contributed to the failure of the third largest private bank. Concurrently, two other banks were affected by accounting malpractices and mismanagement. The exchange rate depreciated by over 40 percent in the first seven months of 2003, and inflation rose to nearly 30 percent.

The authorities turned for support to the Fund, which approved an arrangement based on restoring macroeconomic stability, maintaining confidence in the banking system, ensuring debt sustainability, and resuming strong growth. To achieve these objectives, structural PCs were set on approving laws on the deposit insurance fund, lender of last resort facility, and bank resolution under systemic risk, approval of a budget bill, and unification of the foreign exchange market. As in the other cases, the original program was well supported at the Board—especially the use of prior actions to demonstrate commitment in the face of implementation risks.

While progress has been made on the resolution of the three problem banks, banking supervision and adherence to the legal framework for bank resolution has been weak. In late 2003, in response to continued exchange rate depreciation, the government intervened, leading to a parallel foreign exchange market. Four of the five structural PCs set for the first review were waived (none of the benchmarks were met either, though one was implemented with delay), and the program was permanently interrupted after the first review on account of insufficient fiscal adjustment.

Paraguay fell into its worst recession in decades in 2002, having had no economic growth on average since 1996. The severe recession was triggered by the regional crisis and a poor harvest. In 2003, a new government came into office, and became the catalyst for structural reforms that are being supported by a Fund arrangement. Structural conditionality was wide-ranging, despite the fact that the 2003 funding crisis had been resolved by the time the arrangement was approved. The original program set eight structural performance criteria in the areas of banking system legislation, fiscal adjustment (including pensions), improved financial relations between the central bank and the government, liberalization of fuel and utility prices, and independent audits of public enterprises. The Board was once again generally supportive of this conditionality, although a few concerns were expressed that the proposed conditionality would not be met on time because of the requirement for congressional approval of many conditions. In addition, the precedent and usefulness of including as a condition the requirement for all banks to be credit-rated was questioned. In fact, all of the structural conditions have been met except for two related to the passage by the (opposition-controlled) Congress of new comprehensive banking system legislation, and growth has resumed.

86. Structural benchmarks have declined slightly in PRGF arrangements and have remained fairly flat in GRA-supported arrangements since 2000. These developments contrast somewhat with the Board’s recommendation in March 2001, that, in light of the fact that “..the boundaries of conditionality had become blurred, due to the increasing use of structural benchmarks..... Directors saw a need for structural benchmarks to be used more sparingly.” 57

Rationale for Choosing SBs over Structural PCs

87. The guidelines provide two reasons for choosing SBs over structural PCs. First, SBs are appropriate if a measure cannot be objectively monitored because its definition lacks precision. For these conditions, Board review is needed in order to establish whether implementation warrants the tranche to be released. Second, SBs are also chosen over structural PCs to monitor small steps that are key components of a broader reform. Although too small to interrupt the arrangement by themselves, these steps are markers of progress in a critical area.

88. In countries with large numbers of structural benchmarks, SBs are often clustered around specific actions. For example, in Sierra Leone, benchmarks have been set on developing a National Revenue Authority (4) and making inroads into developing the financial sector (7). In the Democratic Republic of Congo, a number of benchmarks were set up to help improve fiscal transparency. In Nicaragua, a quarter of the SBs in the first two years of the program focused on tax reform, while a third of the SBs were aimed at strengthening bank supervision. Similarly, in Armenia, over half of the benchmarks in the first-year program were associated with tax and customs reform, with a further 25 percent aimed at bank supervision and liquidation. At the commencement of Albania’s most recent program almost half of the ten SBs were set on tax and customs reform.

89. Structural benchmarks are also often used for reforms that have a long gestation period. In the early stages of a reform process, when the focus is often on preparation rather than implementation, specific measures are identified as clear markers, and therefore are set as SBs. Not surprisingly, SBs are overrepresented in conditions that require a lengthy period to be adopted, with heavy concentration in social security reforms, economic statistics, and governance and institutional issues.58

90. The large number of SBs raises the question whether their use conforms to the guidelines in all cases, or whether they sometimes signal less critical or less urgent measures than other conditions. Early indications suggest that the latter rationale is sometimes used. In some cases, a condition initially set as a PC was subsequently converted into a SB when the condition proved to be too ambitious. In Lesotho, filling the position of Accountant General was made a PC in 2002, but later became a structural benchmark because of the difficulty in finding the appropriate candidate (although the measure was eventually implemented). Conversely, sometimes measures are converted from SBs into structural PCs when they fail to be implemented. In Ethiopia, the consolidation of the federal and regional budgets was converted from a SB to a PC because of a delay in implementation and a growing realization of the importance of the measure (which was also a trigger for the completion point under the Heavily Indebted Poor Countries (HIPC) Initiative). The recording and resolution of inter-enterprise arrears in Albania and the submission to Congress of a bankruptcy law in Bolivia were switched from benchmarks to structural PCs for similar reasons. These measures did not obviously qualify as “small steps” in a larger process of reform when they were originally formulated and were clearly monitorable because they were later converted into structural PCs. These actions suggest that the Fund does sometimes make judgments about the criticality of structural PCs versus SBs.59

91. Evidence that a distinction is made between structural PCs and benchmarks in terms of criticality also comes from the widespread presence of isolated benchmarks, many of which appear monitorable. Almost 30 percent of benchmarks set in arrangements approved in 2003—after the new guidelines went into effect—are the sole condition in a particular sectoral area as defined by the list in Chapter III, and many of these conditions are fairly precise. For example, in Colombia, a benchmark was set on the issuance of a decree to eliminate existing vacancies in the public service and to close vacancies created by retiring staff, while no other condition was put in the area of civil service reform. Similarly, in Macedonia, the submission of a new Tobacco Law to Parliament was a structural benchmark.

92. While only small differences exist in the relative numbers of structural PCs and SBs across area departments, a number of country-specific differences are observed. Over the 2001-03 period, there are some 3-4 performance criteria for every 10 structural benchmarks across area departments, although in some country cases, the choice in favor of SBs or structural PCs is very marked.

93. In some arrangements, structural PCs are absent, or virtually so. For example, in Benin, no new structural PCs have been added since the first review (Table 11). The lack of new performance criteria was justified by the completion of first-generation reforms at the start of the program, and therefore remaining structural measures were deemed less critical. 60 This rationale of course does suggest that the Fund is making a distinction between the criticality of structural PCs and SBs. In the capital account crisis programs of Brazil and Indonesia (since the first review), and Peru’s 2002 program based on enhancing market confidence, structural PCs were also absent. In most of these cases, the absence of structural PCs reflects the authorities’ concerns about the potential for adverse market reactions in the event of waivers—which may be deemed particularly visible signs of implementation slippages—and the greater flexibility in policy choices that is often afforded by benchmarks.

Table 11.

Selected Cases of Choice of Structural Conditions

(number of conditions)

article image

No structural PCs after the first review.

94. The implementation record of these countries over the recent period is fairly strong. While Indonesia and Peru’s implementation indices (calculated as set out in ¶22) were roughly at the median level, Benin’s implementation index was at the 70th percentile, while Brazil’s 2002 program implementation index was at the 80th percentile.

95. Some arrangements do not include structural benchmarks, because of concerns about track record. Through mid-2004, the current Guyanese program had on average five structural PCs per review, while the Djibouti program (which began in 1999) had on average four structural PCs per review. In the Guyanese program, none of the structural PCs were met on time, although most were met with a delay. In the Djibouti program, about half of the structural PCs were met on time and the program was eventually permanently interrupted.

96. All in all, it seems that, similar to prior actions, structural PCs are often used when the authorities’ ownership of the program is thought to be weak, and that benchmarks are often the preferred form of conditionality when implementation has been relatively strong. Indeed, structural PCs appear sometimes to be used “strategically”, signaling higher criticality or the need to overcome ownership difficulties. This rationale stands in contrast to the guidelines, which discourage this approach. As the notion of criticality becomes more diffuse, the likelihood rises that measures that are less than critical will be included in conditionality. Moreover, raising the status of structural PCs risks encouraging the Fund to try to substitute conditionality for ownership.

B. Timetables for Implementing Structural Conditions

97. As discussed in Chapter I, a large share of structural conditions that are waived are implemented with delays, suggesting scope for more realistic timelines (Box 4). At the same time, an ambitious timetable for implementing conditions can sometimes help focus efforts and expedite implementation.

Is the Timetable for Structural Performance Criteria Realistic?

In recent years some arrangements stand out as having a large number of structural performance criteria, many of which required waivers because implementation took longer than planned. These cases raise the question of how far timeliness in carrying out the actions is critical. While in some cases, it could be argued that the waivers were the result of technical delays, this possibility should ideally be taken into account in setting the timetable for meeting conditions.

In Bolivia, five structural PCs were set at the beginning of the 2003 stand-by arrangement to address the looming fiscal crisis, but three required waivers. Two of the waived conditions were adopted within a month of the set date, and alternative actions were found to replace the condition on revenue measures. Delays were thus relatively minor.

In Guyana, all eight structural PCs set for the first review were waived. All conditions except one (the implementation of profit sharing in the sugar company) were subsequently met, although the maximum implementation delay was 8 months. The timing of the initial conditions in hindsight was too ambitious—possibly related to an underestimation of capacity constraints—and a more gradual timetable would have been appropriate.

In Azerbaijan and Lesotho, about five structural PCs per review were set, compared to an estimate of about two for all PRGF arrangements, and about 50 percent of conditions have been waived in each country. In Azerbaijan, each review was delayed, with a maximum lag of over one year. Indeed, three of the four conditions set for the second review were not met until one year after the target date. For Lesotho, some measures were also implemented at least one year after the original deadline, although in this case the reviews were not held up on account of these delays. In these cases it could be argued that weak ownership (Azerbaijan) and overambitious timetables in a few cases (Lesotho) also contributed to the high waiver rates.

98. Since delayed structural measures are a major component of the high waiver rate in both GRA- and PRGF-supported programs, the need for waivers could be reduced by using floating tranches. Such financing is disbursed when a condition or set of conditions is successfully completed. The guidelines envisaged that floating tranches would be used to give the authorities more flexibility in the timing of structural reforms, but, as yet, this type of conditionality has not been used. An analogous concept is the floating completion point of the Highly Indebted Poor Country (HIPC) Initiative.

99. One difficulty in applying floating tranches is that few measures satisfy their requirements. Measures must be immediately need-increasing so that Fund financing is required to fill the financing gap, but need-reducing over the longer term so that they warrant greater Fund exposure. Trade liberalization and civil service reform may include measures with these features, but these reforms are not widespread in Fund-supported programs, and have not been associated with floating tranches. Recent examples of measures that meet the criteria for floating tranches include reducing the unweighted average MFN tariff in Bulgaria, and reducing the maximum tariff rate in Kyrgyz Republic. In these cases, the condition is immediately need-creating through attracting more imports, but over time should make the economies more competitive.

100. In considering the use of floating tranches, lessons can be learnt from the experience with floating completion point triggers set in association with the HIPC Initiative.61 Experience using these conditions in the HIPC Initiative suggests that the related reforms have been accomplished at a fairly slow pace, associated with the modest financial benefits of completing conditions early. A notable reason for the weak incentive for early achievement of the completion point among HIPC countries is that the multilateral agencies have consistently provided interim relief to these countries between the decision and completion points and no debt service is paid to bilateral creditors unless the parties have negotiated a recent rescheduling agreement. As a result, the added short-term financial benefit for the HIPC country in reaching the completion point early is small. The HIPC experience thus suggests that floating tranches of moderate size may not capture the benefits associated with standard conditionality, namely of helping to focus the authorities’ efforts and, in some cases, to strengthen weak ownership or overcome divided ownership.

C. The Scope of Reviews

101. Program reviews have largely succeeded in limiting themselves to the issues identified at the outset of the program unless country circumstances have changed. While the guidelines treat the introduction of new conditionality during the program flexibly, this flexibility must not be misused, or the Fund would be accused of “moving the goalposts.” In general, it appears that when conditionality has changed substantially during the course of a program, this change is related to new economic or political circumstances in the country, or to a recognition that the program’s intermediate objectives could not be met without adding specific actions or conditions (e.g., revenue targets without tax policy changes, or state-owned enterprise reform without specific plans and conditions). At the same time, there are arguably some cases where the Fund has exceeded its mandate in this area (Box 5).

Changes in the Structure of Conditions during Fund-Supported Programs

The conditionality guidelines emphasize the need to specify the Fund-supported program as fully and transparently as possible ex ante, although they permit the modification of conditions in order to ensure that program objectives remain achievable, and the modification of program objectives strictly “as new information becomes available” (e.g., if additional adjustment is needed owing to changes in economic conditions). The following four examples capture a variety of ways in which conditionality was changed during the program. The rationale for the changes is diverse: mid-course corrections when it becomes apparent that the conditionality in the original program is not likely to achieve its objectives, responding to political changes, and addressing the enormous need for reform in war-torn countries; in the latter case, possibly stepping beyond the boundaries set in the original program.


The Armenian economy weakened following the political assassinations in late 1999, evidenced by a significant deterioration in the fiscal position and a slowdown in growth. Reflecting these conditions, the original Fund-supported program emphasized the need to reduce the fiscal deficit partly through the achievement of ambitious tax revenue targets. However, no structural conditionality was placed on tax and customs policy through the third review.

The absence of measures may have contributed to the decline in tax revenues over the 2000-02 period. By the third review, measures on the development of codes of conduct and a number of legislative changes were made conditions of the Fund-supported program. Notwithstanding these efforts, capacity improvements in both tax and customs agencies fell short of expectations.


In the original program (2001), it was recognized that reforming the banking and state-owned enterprise (SOE) sectors was essential to accelerate the transition toward a market-based economy and to promote private sector activity. However, during the early stages of the program, this strategy was not buttressed by structural conditionality on the SOEs. Over time it was realized that structural measures on SOEs would facilitate the reform of these enterprises. In response, conditions were introduced in the program to provide enforcement power to the Committee overseeing the SOE reform plan, and to implement guidelines on SOE debt resolution.

In the event, program conditionality was inadequate to address structural weaknesses in the state- owned sector, partially because too much emphasis was placed on capital injections rather than divestiture as a tool for inducing reform, and because the resolution of nonperforming loans stalled, possibly owing to the delayed implementation of conditionality on SOEs.


Madagascar’s original Fund-supported program (2001) focused on raising tax revenues and fostering privatization. However, a crisis triggered by the presidential elections later that year brought the economy to a standstill, seriously weakening the financial position of public enterprises, including those slated for privatization. In response, measures concentrating on the condition of the SOEs were introduced; the accounts of the oil company were examined and the electricity company’s debts to private sector suppliers were offset through eliminating the latter’s tax obligations. All of these measures had Fund conditionality attached. This change in direction was motivated by the fact that quasi-fiscal costs were large.

Democratic Republic of Congo

The Democratic Republic of Congo (DRC) has been subject to civil war since mid-1998 with displacement of populations, growing numbers of refugees, disabled persons, and destruction of infrastructure, including hospitals and schools. In early 2002, the recently appointed pro-reform government started to address the country’s main challenges and requested financial support through a new PRGF arrangement.

The intensity of the structural conditionality set by the Fund at the outset of the program reflected the devastated state of the economy. The program initially included eight prior actions, three structural PCs, and four structural benchmarks, set in the areas of public enterprise reform, governance, central bank audit, and the liquidation of insolvent banks. The letter of intent was broader in scope, with additional discussion of the need to mobilize more revenues, control and track expenditures, and introduce a new labor code. Some of these measures were taken up in the second year of the program; for example, a benchmark was placed on the effective implementation of new expenditure procedures.

However, in the third year of the program conditionality changed considerably, in ways that are difficult to explain by reference to changes in circumstances. Many new conditions were added including civil service reform, the settlement of cross arrears between the government and the private sector, and the regulation of the diamond company.

V. Prior Actions62

102. Prior actions play a dual role in program conditionality. They are used to ensure upfront implementation of key reforms in cases where this is critical to the success of the program. As such, they are used, on the one hand, for measures where there is a straightforward need for early implementation, and, on the other, in cases where there is some doubt about the likelihood of implementation if the actions are not taken up front.63 The use of prior actions in the latter role helps to boost the program “delivery rate,” or, in principle, might also work as a “test” for those countries whose program ownership or reform commitment may be in doubt.

103. The main role of prior actions—to ensure upfront implementation of key reforms in cases where this is critical for the success of the program—is supported by their relatively large number in countries emerging from conflicts.64 In these countries, which have typically tended to have PRGF-supported programs, the average number of prior actions was a little over five per program year during 2001–03, compared with the average of some 3½ prior actions for that period in PRGF-supported programs.65 The larger number of prior actions is consistent with the upfront structural measures required to stabilize these economies or impart momentum to a large reform agenda.

104. In capital account crisis cases, experience with prior actions has been mixed.66, 67

  • Among the large access capital account crisis cases, Argentina and Turkey had a divergent usage of prior actions.68 In Turkey, at the time of augmentation of the 1999 Fund- supported program in May 2001—after the onset of the financial crisis—nine prior actions were set. By contrast, in the January 2003 Argentina program, no prior actions were set. The absence of prior actions (and indeed limited structural conditionality overall) in Argentina’s January 2003 program—which was a transitional arrangement to preserve macroeconomic stability through the changeover to a new government in May 2003—was driven to a large extent by the political environment prevailing in Argentina at that time, which constrained the government’s ability to act decisively on the structural side. The follow-up September 2003 stand-by arrangement had no prior actions and only a limited structural reform agenda, which was sequenced to give time to build consensus. Uruguay’s 2002 arrangement had three prior actions for approval and around 5½ prior actions per program year subsequently.

  • Among the countries with lower access, the 2003 programs in Bolivia and the Dominican Republic had two and eight prior actions for program approval, respectively. The small number of prior actions in Bolivia contrasts with the weak track record in its previous program (implementation index of 1.0). In the Dominican Republic, where there was no recent track record, prior actions were needed for upfront implementation of some key financial sector measures.69

105. More prior actions were typically established in arrangements for countries with a weak track record (Figure 15, upper panel).70, 71 Recent arrangements with Ecuador (2003), Ghana (2003), Guyana (2002), and Romania (2001), to name a few, are examples of programs with a relatively large number of prior actions and a relatively weak track record. The case of Romania is particularly interesting (Box 6)—the 2001 program had about 24 prior actions per program year. Five previous programs had gone off track, and measures subject to prior actions served a vital role in establishing credibility and overcoming vested interests. Despite the establishment of a successful track record in the 2001 program, the 2004 Fund-supported program with Romania has continued to use many prior actions because they have proved to be effective in fostering progress in structural reform.

Figure 15:
Figure 15:

Structural Conditionality and Track Record

(normalized per program year)

Citation: Policy Papers 2005, 029; 10.5089/9781498331739.007.A001

Source: MONA database and staff calculations.

Structural Conditionality in Romania1

  • Fund-supported programs in Romania since 1997 have consistently relied on a large number of prior actions (Table). In the 2001 stand-by arrangement with Romania—the first arrangement with the country to be successfully completed—prior actions comprised half of the total number of structural conditions. The current 2004 SBA, at least so far, is keeping up this trend.

  • Prior actions have proved effective in supporting reforms in Romania. Given the slow pace of structural reforms and the authorities’ weak ownership record in the various programs during 1990-1996, the strategy in subsequent programs was to entrench structural reforms through conditionality, largely in the form of prior actions, on bank restructuring and enterprise privatization.2 As programs in the late 1990s went off track as well,3 the Fund required the completion of more prior actions to preserve program credibility, and the 2001 program entailed strong and upfront structural conditionality.4 These prior actions, indeed, contributed to both a higher “delivery rate” of program policy measures and a much improved macroeconomic environment; they also greatly aided the authorities in getting past vested interest groups and other obstacles to structural reforms. Indeed, the authorities supported progressively tougher upfront conditionality in each successive program as a means of facilitating the implementation of reforms and boosting their credibility.5 While using more prior actions in the 2001 program was a reasonable response to the weak track record, prior actions have continued to be extensively used in Romania even after the successful completion of the 2001 program, as seen in the 2004 Fund- supported program.

  • Not all the prior actions were in the Fund’s core areas of responsibility, although an argument can be made for the macro-criticality of each. Of the 48 prior actions in the 2001 program, 7 were on bank restructuring, 7 were on privatization, 12 were on energy sector reforms, and the remainder on fiscal measures, incomes and wage policy measures. In the 2004 Fund-supported program, there have been 27 prior actions up to the first review—6 in enterprise privatization, 7 in energy sector reforms, and the remainder in the fiscal sector, wage policy, and governance issues. Quasi-fiscal activities were a major source of macroeconomic instability and thus critical, justifying the conditionality. The original letters of intent of the 2001 and 2004 programs are generally clear as to what constitutes program-related conditions, and conditionality elaborated in the programs remained broadly within the terms set out in the original memorandum of economic and financial policies.6

  • Fund conditionality in noncore structural areas overlapped with World Bank conditionality. The Bank has focused its conditionality on privatization, private sector development, poverty reduction, institution building, and governance. The Fund-supported programs have included conditionality, based on the technical expertise of the Bank, in some of the Bank’s lead areas when these areas were considered to be crucial for macroeconomic stabilization, such as reforms in areas with significant fiscal and quasi-fiscal impact.

Romania: Structural Conditionality

article image

Conditionality up to the first review.

1 Draws from Romania—Ex Post Assessment of Longer-Term Program Engagement (Country Report No. 04/113).2 There were three SBAs between 1991 to 1996, all of which went off track due to piecemeal reforms, lack of ownership, and questionable conditionality (Romania’s EPA suggests that the emphasis on macro-based conditionality, as opposed to addressing vested interests with frontloaded structural conditionality, in the earlier programs postponed the emergence of sustained growth).3 Despite the authorities’ sizable initial stabilization efforts, the 1997 and 1999 programs were not successful owing to, among other things, insufficient measures to tackle deep-rooted structural reforms, which resulted in a persistent build-up of quasi-fiscal deficits.4 The Fund repeatedly entered into financial arrangements with Romania (as opposed to a staff monitored program for first establishing a track record) because in most cases, the Fund’s approval of new arrangements was in response to severe and immediate balance of payments pressures.5IMF Reviews Romania’s Performance Under Past Fund-Supported Programs,” PIN No. 04/44, 4/23/04).6 Note that the 2004 arrangement is still ongoing.

106. Prior actions complemented structural performance criteria in countries with relatively weak track records, particularly in GRA-supported programs (Figure 15, lower panel). The number of structural performance criteria in the PRGF-supported programs of members with weak track records has been broadly comparable to the average in PRGF-supported programs, but the number of structural performance criteria in the Fund-

107. Regression analysis points to more prior actions in countries with a weak track record. The analysis, using data for all countries that had at least two arrangements during the period 1995 to 2003, relates the number of prior actions to the score on the structural implementation index in the country’s previous Fund-supported program, after controlling for various other factors (Appendix, Section A). The results show that programs in countries with a low implementation score in the previous program tend to have more prior actions in their program. The analysis also reveals a persistent effect in the use of prior actions in an individual country’s programs.

108. Furthermore, regression analysis indicates that the ratio of structural performance criteria to structural benchmarks is positively and significantly related to the number of prior actions. The estimation, based on the same sample, shows that after controlling for lagged movements in the implementation index, the “package” of prior actions and structural performance criteria appears to be considered together in a program (Appendix, Section B).

109. A greater number of prior actions, however, does not necessarily translate into better program implementation. Previous research has found that Fund-supported programs with more prior actions are neither more nor less likely to suffer from a program stoppage.72 Further analysis including the 2000-04 period suggests that more prior actions for program approval do not bring the subsequent implementation of structural conditionality to the Fund-wide average level (Appendix, Section C). This could be interpreted to mean that prior actions do not work effectively in screening out potential weak performers. Alternatively, it could also mean that prior actions may be used to compensate for the low ex ante probability of program implementation, raising the implementation rate through the completion of prior actions to some minimum acceptable implementation standard.

110. Staff-monitored programs could play an important supplemental role in the screening process. Sometimes, where past implementation has been weak, a staff-monitored program (SMP) is employed to establish a favorable track record.73 Satisfactory completion of the SMP is a necessary, though not sufficient, condition in the subsequent consideration of a request for access to Fund resources or resumption of an arrangement when the program has gone off track. During 2000 to September 2004, 29 countries had SMPs, of which only about a third satisfactorily completed them, moving on to a subsequent Fund arrangement (Box 7). Implementation of the Fund financial arrangement by members that completed an SMP has been, on average, at least as good as the average implementation of all Fund- supported programs.

Prior Actions and Staff-Monitored Programs

  • Prior actions can potentially screen out cases where the likelihood of program implementation is low. In cases where the track record for policy implementation has been weak, typically prior actions are set for the approval of a Fund-supported program. However, empirical analysis indicates that more prior actions do not always succeed in screening out countries with weak ownership.

  • Staff-monitored programs (SMPs) can play a role in the screening process. SMPs are designed to help members establish a track record of consistent policy implementation for a Fund financial arrangement or Fund emergency post-conflict assistance, or to re- establish a track record when an arrangement has gone off track.1 While not a sufficient condition, it is necessary for countries with SMPs to complete them satisfactorily in order to continue to a Fund arrangement.

  • A third of members with SMPs in recent years successfully moved on to Fund- supported programs; their program implementation has been at least as good as the average performance of Fund-supported programs. Of the 29 SMPs approved between 2000 and September 2004, 10 had subsequent Fund-supported financial arrangements. On average, the implementation index (excluding prior actions) of Fund-supported arrangements obtained subsequent to an SMP was about 1.5 compared with the Fund-wide program average of around 1.4 during 2000 to 2003 (on a scale of 0 to 2, where 0 represents conditionality not implemented and 2 represents conditionality fully implemented).

SMPs with Subsequent Arrangements 2000-2004

article image
1 The guidance to staff on SMPs was revised following the Board discussion of Signaling Assessments of Members’ Policies (PIN No. 03/12, 2/13/03).

111. The above discussion suggests that caution should be used in placing heavy emphasis on prior actions in Fund-supported programs. More prior actions may be effective as a way of ensuring that implementation rates for programs are raised, and they “screen in” programs that, as a result, are likely to have an acceptable implementation standard. In some countries (e.g., Romania), this higher implementation rate is the rationale for prior actions in the Fund-supported programs, as they have been used effectively to overcome vested interests. Indeed, some recent ex-post assessments (EPAs) have promulgated the use of prior actions as a tool to achieve better implementation.74 However, implementation rates are not lifted to the Fund-wide average level simply by adding more prior actions. Faltering implementation of the subsequent program, following the heavy use of prior actions, suggests that prior actions do not work very well as tests of ownership, and draws attention to the possibility that—even with improved implementation of program- related conditions—follow-up measures may not be undertaken in the reform process (i.e., the actions may be implemented to the letter, but not in spirit), leaving implementation of the broader policy program still unsatisfactory.

VI. Conditionality in Capital Account Crisis Cases87

119. Experience with structural conditionality in Fund-supported programs in countries with capital account crises has been varied as regards the number of conditions, the choice of conditionality tools, and the sectoral distribution of conditions. In the post-guidelines period (2001-03), the Fund-supported programs that were classified as capital account crises included Argentina (January 2003), Bolivia (2003), Dominican Republic (2003), Turkey (May 2001, when the 1999 arrangement was augmented after the financial crisis), and Uruguay (2002).88

A. Overall Findings

120. Total conditionality at the time of the approval of the arrangement in these countries is generally below the Fund-wide average for GRA-supported programs. The total number of conditions averages about 14¾ , compared with the overall average of 16¾ conditions for GRA-supported programs.

121. Capital account crisis cases do not tend to have especially large numbers of prior actions; indeed, some have none. However, cases with few prior actions tend to have more structural performance criteria, and the number of structural performance criteria and prior actions (combined) at the time of program approval in the capital account crisis cases is higher than average.

Conditionality at Approval

article image

Refers to prior actions for the 6th and 7th reviews and benchmarks through June 2001 that were set at the time of the augmentation in May 2001 of the 1999 program

Mean for approvals in GRA-supported programs from 2001 to 2003.

122. For the program as a whole, the capital account crisis countries tend to have a higher than average number of structural conditions— 28 conditions per program year compared with 19 conditions across GRA-supported programs between 2001 and 2003. The higher than average number of conditions applies across all the different types of conditions (with the only exception of Bolivia). In terms of access to Fund resources, the picture is somewhat more mixed. In Turkey especially, very high access was accompanied by a large number of conditions. This tendency is less evident in Argentina and Uruguay.

  • Lastly, consistent with the nature of these crises, the share of conditions to resolve economic vulnerabilities is typically higher than average. About 45 percent of the structural conditionality for program approval focused on economic vulnerability issues, compared with the overall average for GRA- supported programs of 35 percent.89 Conditions on economic management were more numerous than average in Bolivia, Dominican Republic, and Uruguay. In Turkey there was less focus than elsewhere on economic management issues, perhaps because the May 2001 augmentation occurred in the middle of the 1999 program after the onset of the banking crisis, so that additional conditionality focused particularly on banking sector vulnerabilities. Argentina had a higher than average number of conditions in public sector efficiency because of the key program strategy to implement federal-provincial bilateral pacts.

Cumulative Program Structural Conditionality

(Normalized per program year)

article image

Bolded programs refer to the crisis programs. For Turkey, both programs are bolded since the crisis occurred in late 2000 and early 2001, thus conditionality in both programs addressed some crisis issues

Amount approved as a percent of member quota.

As of the eighth review in July 2004.

As of the second review in March 2004.

As of the fourth review in February 2004.

As of the first review in January 2004, after which the program stopped.

As of the third review in June 2004.

Sectoral Distribution of Conditionality at Approval

article image

Refers to prior actions for the 6th and 7th reviews and benchmarks through June 2001 that were set at the time of the augmentation in May 2001 of the 1999 program.

B. Country Experiences90


123. Structural conditionality in the “transitional” stand-by arrangement was limited by domestic political factors and the short duration of the arrangement. While far-reaching structural reforms were needed for lasting economic recovery, the focus of the eight-month SBA was on maintaining macroeconomic stability through the May 2003 presidential elections, providing time for the incoming government to develop a medium-term reform agenda. Accordingly, structural conditionality in the January 2003 transitional program was set only through May 15, 2003 and limited to some initial fiscal reforms, preparations for the banking strategy, and maintaining a measure of legal certainty. The September 2003 three- year program aimed at deepening and broadening structural reforms essential for restoring sustained growth, including (i) reforms of the system of intergovernmental relations and tax policy and administration; (ii) a comprehensive and sustainable sovereign debt restructuring; (iii) reforms of the banking system; and (iv) measures to improve the framework for restructuring corporate debt, including that of the utility companies.

124. Progress on the structural measures in the January 2003 transitional program was limited and there were policy reversals and setbacks. In particular, the fiscal structural agenda was delayed, little progress was made with bank restructuring, and there were setbacks with respect to improving the legal certainty and creditor rights. The weak implementation (the score on the implementation index was only 1.0) seemed to reflect the prevailing political environment and insufficient political consensus.91

125. Although there was a structural agenda set forth in the September 2003 program, the program had no prior actions despite the weak track record with respect to the January 2003 structural reforms. The program laid out the principles of the structural agenda in the banking and fiscal areas; but many of the precise modalities of implementation and timing were not specified and were left to be formulated at a later stage, mainly because of the lack of implementation capacity and the need for time to build necessary domestic consensus for reforms. Structural performance criteria and benchmarks, covering only the first six months of the program, were set at the time of approval of the arrangement in the areas of fiscal and banking sector reforms, amid continuing uncertainty in political support for key structural reforms.

126. At the completion of the second review in March 2004, the implementation index for the structural conditionality had risen to 1.5, following the advances made in key areas such as the banking system and, to some extent, the utilities sector. However, subsequent delays were experienced in structural reform implementation; no review has been completed since March 2004.


127. In contrast to Argentina, Turkey is an example of a capital account crisis program where structural conditionality was extensively used, reflecting the need to address both a currency crisis and a banking sector crisis. After the May 2001 augmentation of the 1999 arrangement (in response to the crisis in early-2001), the focus of structural conditionality was on three key areas: banking, fiscal transparency, and policies to enhance the role of the private sector in the economy. The structural conditionality underlying the 2002 program was mainly in the areas of banking, public sector and fiscal reform, and private sector development, including privatization.

128. The large number of structural conditions in the 1999 and 2002 Fund-supported programs was motivated by Turkey’s deep-rooted problems and the lack of confidence stemming from the structural weaknesses in the economy. A number of prior actions were set to achieve upfront implementation of key measures and perhaps also partly because of the relatively weak track record (see below). However, the 2002 program progressively relied far more on structural benchmarks. Some of the conditionality, particularly in the banking sector, may have been perceived to be micromanaging policies, although the detailed nature of the conditionality was aimed at providing assurances of the seriousness of reforms.

129. Policy implementation of the 1999 program, particularly prior to May 2001, was uneven. Since May 2001, however, considerable progress has been achieved, except in the areas of privatization and private sector development (in which slow progress was partly due to weak market conditions). The average overall implementation index (including prior actions) for structural conditionality for the 1999 and 2002 programs was 1.5, slightly above the median across Fund-supported programs. In the financial sector, overall observance of conditions was high, as 80 percent of financial sector conditions were fully met. However, the implementation index of conditionality excluding prior actions was weaker—the average implementation of structural performance criteria and benchmarks for the 1999 and 2002 programs was around 1.2, below the Fund-wide average of 1.4. In the 2002 program too, shortfalls were mainly in privatization and enhancement of the private sector.


130. There was a substantial increase in the number of conditions in Uruguay’s 2002 arrangement—there were around 23 conditions per program year, compared with less than ten in the 2000 program. Perhaps reflecting Uruguay’s relatively weak program track record, there were also more prior actions set for the approval of the 2002 arrangement and for the completion of subsequent program reviews. The increase in the number of structural conditions in the 2002 program reflects the financial crisis that developed in the country and the continuing pressures on the banking system that Uruguay faced following the deepening of the crisis in Argentina. Aside from the banking sector, the reform agenda envisaged the continuation of efforts to open up to the private sector areas previously reserved for the public sector, in order to foster productivity increases and economic growth in the medium term. Conditionality focused on fiscal consolidation, banking system issues, intensifying reforms to the public banks, and deregulating the economy.

131. The score on the implementation index of the 2002 Uruguayan program including prior actions was 1.4 in the 2002 program as of the fourth review (comparable with an index value of 1.5 in the 2000 program). Without prior actions, the implementation index on the structural conditions was weaker at about 1.1. Progress in the 2002 program lagged in restructuring of the banking system, strengthening of public banks, deregulating the economy, and improving information disclosure in state enterprises and banks. Additionally, the government’s inability to shore up political support for reforms in the context of the approaching presidential elections (October 2004) and persistent differences among the political parties on the appropriate scope and pace of reforms also constrained implementation.

VII. The Process of Program Development92

A. Introduction

132. The new conditionality guidelines emphasize that national ownership is key to successful implementation of Fund-supported programs. The guidelines stress that the member has the primary responsibility for selecting, designing, and implementing policies. They also clearly point out that the need for ownership implies selectivity: approval of the use of Fund resources depends on the Fund’s assessment that the member is sufficiently committed to implement the program.93

133. The guidelines underscore the role of the process of negotiation and program development in strengthening ownership. An effective and inclusive process of engagement between the staff and the authorities can promote and strengthen ownership, and also provides a basis for the staff to make a judgment on the likelihood of implementation.

134. This chapter examines the actual process employed in selected programs and discusses its effectiveness in supporting program implementation. Ten cases from the 2001-2004 period are reviewed: Brazil, Bulgaria, Burkina Faso, Ecuador, Georgia, Guinea, Guyana, Romania, Sri Lanka, and Tanzania. These cases were chosen to represent a mix of programs with strong and weak implementation. While these selected cases cannot be taken to reflect the full range of experiences in Fund-supported programs, a case study approach is the only way in which process can be analyzed, and the commonalities and even lack of commonalities across the diverse cases examined may yield insights. The information is drawn from staff reports, mission briefs, back-to-office reports, memoranda and documents related to the internal review process, staff papers, Independent Evaluation Office (IEO) Reports, Ex Post Assessment (EPA) Reports, interviews with mission members, other input from area departments, and input from the authorities, who saw the draft case studies in an earlier stage of preparation.

135. The remainder of this chapter is organized as follows. Section B discusses the linkages between process, ownership, and program implementation. Section C presents the review of the process in the ten selected cases. The final section concludes.

B. Ownership, Implementation and the Role of Process 94

136. National ownership, along with countries’ capacity to implement reforms, is an important determinant of successful implementation of policy reforms. The staff statement accompanying the guidelines defines national ownership as “a willing assumption of responsibility for a program of policies, by country officials who have the responsibility to formulate and carry out those policies, based on understanding that the program is achievable and is in the country’s best interest.”95 In the absence of confidence that the reforms are beneficial for the country, the authorities are likely to face conflicts of interests, resulting in poor implementation. Moreover, even if conditions are formally met, without genuine ownership implementation may be only nominal (e.g., laws may be adopted, but not enforced). Genuine ownership is more likely to result in consistent implementation of reform programs over time.

137. Ownership, while crucial, is also abstract and unobservable. The “willing assumption of responsibility” described in the staff statement is unobservable before the program is implemented (ex ante). In fact, ownership remains unobservable even after program implementation (ex post), since implementation may not always be based on the genuine belief that the reforms are in the country’s best interest. Ownership is a judgment call, and judgments on the depth and breath of national ownership are difficult especially when the authorities have an incentive to express commitment to reforms, even without genuine ownership.

138. An effective and inclusive process of program development can play an important role in promoting ownership and strengthening program implementation. The overall process is aimed at ensuring that programs receiving Fund support are well designed and formulated through a mutually acceptable process led by the member. The process of dialogue and negotiation should ensure that local knowledge is fully brought to bear on program design, and should strengthen program ownership by providing the authorities greater control of the agenda and building a relationship of trust between the staff and the authorities. In addition, engaging a wide range of country officials and other stakeholders at an early stage could develop a critical mass of ownership, which can be crucial in sustaining program implementation.

139. The process also matters because it produces conditionality that can promote and strengthen ownership. Conditionality, if well designed and developed through an effective process, can serve as a valuable commitment device that complements and enhances ownership, particularly in structural reforms. For example, conditionality can reinforce ownership in cases where the commitment to reforms is less than complete, for instance due to divided ownership or vested interests, or because parliamentary support is not a foregone conclusion. In these cases, conditionality can “strengthen the reformers’ hands,” enhance the credibility of the government’s program, and marshal support for it.

140. An effective process should also help the staff make judgments on ownership before program approval, although these judgments are “inherently subjective and difficult” (staff statement, ¶3):

  • As noted above, ownership is unobservable. The staff is expected to make a judgment based on all observable indicators during the negotiation (and may complement this judgment by seeking implementation of prior actions or a staff-monitored program).

  • Ownership is also dynamic; it may change during program implementation. Consistent implementation of a program may involve painful adjustments in the short run, and opposition from entrenched interest groups or a lack of protection of vulnerable groups can seriously weaken ownership. Conversely, when programs begin to show benefits, and if their costs turn out to be less drastic than originally feared, ownership may strengthen.

  • Responsible officials vary from one case to another, and are not always easy to identify. Ideally, all agents with decision and implementation authority (including the required parliamentary majorities, where relevant) would be committed to the program; but if they are not, additional difficulties arise in judging whether those officials who are committed will be able to carry the day.

  • Relatedly, concluding that the program is both politically and technically feasible requires difficult judgments. The staff is not always equipped with enough capacity to make political assessments, may underestimate the political problems, and may rely too heavily on the reformists’ assurance that opposition can be overcome. Another difficulty is the assessment of administrative capacity, and related judgments on the usefulness of technical assistance and sequencing of the program. A limited amount of time available in a crisis makes it more difficult to reach good judgments on both political and technical counts.

141. While process is important, ownership and implementation are also affected by various exogenous factors. An economic crisis may lead to political succession, which may produce a new government with a strong commitment to reforms. Prospects for European Union accession are another good example of a factor that can create momentum for reforms. Conversely, when the authorities do not implement programmed policies despite their commitment to do so, ineffective process is only one of several possible causes. Political conflicts, changing ownership, changing circumstances, or bottlenecks in administrative capacity may have made an alternative policy preferable from their point of view.

142. What is expected from an effective process? The guidelines and their supporting documents suggest a number of elements: 96

  • conditionality should be formulated through a mutually acceptable collaborative process led by the member (staff statement, ¶3);

  • program documents will be prepared by the authorities, with the cooperation and assistance of the Fund staff (guidelines, ¶10);97

  • the staff will endeavor to reach understandings with the authorities on a mutually acceptable means of achieving the program goals (staff statement, ¶4);

  • the staff will pay due regard to the domestic social and political objectives, the economic priorities, and the circumstances of the member (guidelines, ¶4);

  • involvement of country officials who have the responsibility to formulate and to carry out a program of policies (staff statement, ¶3);

  • the staff will encourage members to seek to broaden and deepen the base of support for sound policies (guidelines, ¶3);98

  • the staff will be guided by the principle that the member has primary responsibility for the selection, design, and implementation of its economic and financial policies (guidelines, ¶3), and be flexible in program design when possible to meet the authorities’ preferences (operational guidance, ¶6); and the staff will stand ready to advise the authorities on a range of available policy options and implementation plans, and to provide technical assistance as appropriate, particularly in cases where the member’s administrative capacity is weak (staff statement, ¶4);

  • collaboration with other multilateral institutions (guidelines, ¶8).99

C. Review of the Process

143 The ten case studies are reported in Table 12. The table summarizes the evidence available on the various elements of process set out above.

Table 12.

The Process of Program Development--Selected Programs

article image
article image
article image
article image

The 1998 EFF arrangement, which ended in 2001, is included because it laid the basis for the 2002 SBA.

A lack of a complete understanding is taken as evidenced by (i) the authorities indicating a lack of agreement with the program at the time of approval, or (ii) departures from the agreed program during implementation with an apparent lack of awareness by the authorities that these departures would endanger the arrangement.

Evidence that specific circumstances were missed is taken to be present when, with hindsight, a key element is found to be missing from the program.