Structural Conditionality in IMF-Supported Programs

Book
Chapter

Chapter 3 Structural Conditionality: Structure and Effectiveness

Author(s):
International Monetary Fund. Independent Evaluation Office
Published Date:
April 2008
  • ShareShare

9. This section describes the different types of structural conditions and reviews their effectiveness at bringing about structural reform. It begins with a classification of structural conditions according to their legal standing, their sectoral distribution, and the degree of structural change that they would bring about if implemented. This is followed by an analysis of compliance with these conditions and of their effectiveness. The section concludes by discussing factors that may affect the design of SC and its impact on reform.

A. Characteristics of Structural Conditions in IMF-Supported Programs

10. Of the 216 arrangements that were approved between 1995 and 2004, 119 were financed through the IMF’s General Resources Account (99 Stand-By-Arrangements (SBAs) and 20 Extended Fund Facilities (EFFs)) and 97 with concessional resources (35 Structural Adjustment Facilities/Enhanced Structural Adjustment Facilities (SAF/ESAFs) and 62 Poverty Reduction and Growth Facilities (PRGFs)); these two groups are henceforth referred to as SBAs and PRGFs respectively. 1 Of the arrangements that were financed with concessional resources, 70 took place with countries that were eligible for debt reduction under the Heavily Indebted Poor Countries’ (HIPC) Initiative. Together, the 216 programs included more than 7,000 structural conditions, of which about 30 percent were prior actions (PAs), 20 percent structural performance criteria (PCs), and 50 percent structural benchmarks (SBs).

11. PAs, PCs, and SBs differ in their legal standing and their timing relative to program discussions by the Executive Board. PAs are measures that a member is expected to adopt prior to the approval of an arrangement or the completion of a review (a condition for credit tranche disbursement); they are set by IMF management. Since 2000, program documents sent to the Board have been required to describe the status of PAs (this was not always done before 2000). PCs are conditions approved by the Board that need to be met before disbursements are made under an arrangement. In case of non-compliance with a PC, the Board needs to grant a waiver before a disbursement can be released. The decision to grant a waiver is usually based on assurances that the program is otherwise on track and/or that remedial actions have been taken. PCs are applied to clearly specified variables or measures, for which timing is considered important and that can be objectively monitored by IMF staff. SBs, on the other hand, are applied to measures that cannot be specified in terms that are objectively monitorable, or to measures where non-implementation of a single component would not be judged sufficient to derail the program. Like PCs, SBs are approved by the Executive Board. Non-compliance with an SB does not automatically lead to an interruption of the program and, therefore, does not require a waiver by the Executive Board. But in response to non-compliance with one or more SBs, management can delay or refuse to submit for the Board’s approval an ongoing review of a program if it assesses that the non-compliance would jeopardize the achievement of program objectives. 2

12. For the purposes of this evaluation, the IEO classified structural conditions into nine sectoral categories, broadly in line with classifications used by IMF staff in internal reviews (see Background Document Chapter III). Four of these sectors—Taxation, Public Expenditure Management (PEM), Financial Sector Reform, and Other Fund Core activities, including issues related to trade, exchange rate, and monetary policy—correspond closely to the definition of the Fund’s core areas provided by the 2002 CG, and are referred to as “Core” in the remainder of this report. 3

13. During the evaluation period (1995–2004) about two-thirds of all structural conditions were concentrated in a few core areas of Fund responsibility, with some 20 percent in each of the following areas: Taxation, PEM, and Financial Sector Reform, and 6 percent in other core areas. About half of the remainder focused on state-owned enterprise (SOE) reform and privatization, and the rest were scattered across a large number of sectors where the World Bank has greater expertise than the IMF, including social policies, civil service reform, and regulatory reform (Figure 1).

(In percent of all conditions)

Figure 1.Distribution of Structural Conditionality by Economic Sector

Source: MONA and IEO staff estimates.

14. The types of conditions used varied across sectors. Although the general pattern was similar in SBAs and PRGFs, there were some significant differences; notably, financial sector conditionality was more prominent in SBAs, while conditions on PEM were more prominent in PRGFs. Within non-core sectors, the conditions affecting SOE reform and privatization tended to take the form of prior actions and performance criteria, while the conditions affecting the other non-core sectors tended to be SBs.

15. The effects of SC depend not only on the number of conditions but also on the degree of structural change that they would bring about if implemented, and on the durability of this change. This evaluation refers to these characteristics as the structural content or structural depth (SD) of conditions. Table 1 presents examples of SC by type (SB, PC, and PA) and classified and scored according to the following three categories of SD: 4

Table 1. Examples of Structural Conditionality, by Type of Condition and Structural Depth (SD)
SD Structural Benchmarks (1) Performance Criteria (2) Prior Actions (3)
High SD (2) (i) Privatization of the electricity generation company Egasa (Peru 2002 SBA) (EBS/02/12 p. 65). (i) Privatize Uganda Development Bank (UDB) (Uganda 2002 PRGF) (EBS/02/163 p. 68). (i) Adopt legislation that allows the effective privatization of five strategic wineries through an open tender process (Moldova 2000 PRGF) (EBS/00/249 p. 38).
(ii) Congressional approval of the Fiscal Responsibility Law (Colombia 2003 SBA) (EBS/02/210 p. 100). (ii) Parliament to pass legislation to close duty-free shops at all land crossing points in border areas to reduce scope for smuggling and tax evasion (Albania 2002 PRGF) (EBS/02/94 p. 63). (ii) Approve legislation to permit private sector participation in construction and operation of roads (Honduras 1999 ESAF) (SM/04/24 p. 26).
(iii) Passage of the bill setting up the autonomous Central Statistical Office by end-September 2001 (Sierra Leone 2001 PRGF) (SM/05/158 p. 41). (iii) Congressional approval of the law for bank resolution under systemic risk (Dominican Republic 2003 SBA) (EBS/03/116 p. 12). (iii) Parliament has adopted a revised VAT law consistent with the MEFP (Macedonia, FYR 2003 SBA) (EBS/03/51 p. 58).
Limited SD (1) (i) Establish amount of road sector arrears and reduce them by C30 billion in first half of 1999 (Ghana 1999 ESAF) (EBS/99/57 p. 55). (i) Congressional approval of a Budget bill for 2004 consistent with the fiscal targets in the program (Dominican Republic 2003 SBA) (EBS/03/116 p. 12). (i) Adoption and passage by parliament of a 2001 budget fully consistent with the economic program described in this memorandum, including associated revisions to the tax code (Georgia 2001 PRGF) (EBS/00/258 p. 56).
(ii) Issuance of a decree to eliminate existing vacancies in the public service with immediate effect, and to close vacancies created by retiring staff, as set out in paragraph 17 of the MEP (Colombia 2003 SBA) (EBS/02/210 p. 99). (ii) Audit of civil servants (Côte d’Ivoire 2002 PRGF) (EBS/02/44 p. 64). (ii) Implementation of a VAT at a single rate of 18 percent (Mali 1999 ESAF) (EBS/99/129 p. 63).
(iii) Adopt joint action program agreed by Treasury and SFSA to monitor budget entities’ bank accounts, ensure that payments arrears problems are properly addressed, and establish a time-bound plan for the progressive reduction and elimination of arrears (Mongolia 2001 PRGF) (EBS/01/166 p. 74). (iii) Raise the maximum wage subject to contributions—used by the CNSS—to ouguiyas 70,000 (Mauritania 2003 PRGF) (EBS/03/94 p. 76). (iii) Establish financial control departments within the Ministry of Finance (Estonia 2000 SBA) (EBS/00/18 p. 74).
Little or No SD (0) (i) Submit to parliament a draft Law on Financial Disclosure, requiring regular reporting by senior elected and appointed officials (Moldova 2000 PRGF) (EBS/00/249 p. 76). (i) Issue a tender for the privatization of SONACOS (Senegal 2003 PRGF) (EBS/03/49 p. 91). (i) Agree on the main elements of the restructuring of state-owned commercial banks (SOCBs) to be defined in consultation with the AsDB and the World Bank (Lao People’s Democratic Republic 2001 PRGF) (EBS/01/53 p. 76).
(ii) Agree between the BROU, the BCU, and the government on a reform plan for BROU (Uruguay 2002 SBA) (SM/05/84 p. 24). (ii) Conduct and publish an economic and environmental analysis, and prepare an action plan with timetable for the cost-effective production, distribution, and sale of fuels and other petroleum products in Ecuador (Ecuador 2003 SBA) (EBS/03/21 p. 11). (ii) Government to submit to parliament a labor law that satisfies the conditions specified in paragraph 24 of the MEFP (to be monitored in consultation with the World Bank) (Croatia 2003 SBA) (EBS/03/3 p. 76).
(iii) Appoint implementation manager (Interim Commissioner-General) (Lesotho 2001 PRGF) (EBS/01/18 p. 53). (iii) Bring KESC to point of sale, as detailed in MEFP, para. 23 (Pakistan 2001 PRGF) (EBS/01/197 p. 83). (iii) Announce that the Cocobod will ensure equal access to its crop financing and warehousing facilities (Ghana 1999 ESAF) (EBS/99/57 p. 55).
Source: IMF Staff Reports and Ex Post Assessments.
  • Little or No (0). This category includes conditions that would not, by themselves, bring about any meaningful economic changes although they may serve as stepping stones for significant reforms. Examples include the preparation or announcement of plans, strategies, or legislation.

  • Limited SD (1). This category includes conditions calling for one-off measures that can be expected to have an immediate and possibly significant effect, but that would need to be followed by other measures in order for this effect to be lasting. It encompasses conditions of two main types: those requiring one-off fiduciary measures, and quasi-macro quantitative conditionality. Examples of the former include the publication, by a given date, of the federal budget or the accounts of public enterprises, or the preparation of specific audits. Examples of quasi-macro quantitative conditionality include changes in controlled prices, limits on the growth of the wage bill, or the reduction of arrears of certain public enterprises.

  • High SD (2). This category includes conditions that, by themselves, would bring about long-lasting changes in the institutional environment. Most of the conditions in this category entail legislative changes (e.g., approval, adoption, or enactment of legislation by a parliament). This category also includes conditions requiring that certain fiduciary measures be taken on a regular and/or permanent basis, even when legislation is not needed (often these measures are implemented through regulation). Examples of such measures are similar in substance to conditions with limited SD, but are expected to have more lasting effects. This category also includes conditions with long-lasting structural impact—e.g., implementing a civil service reform or privatization—that may be grounded in pre-existing legislation but that probably could not be undone without new legislation.

16. A review of the full list of structural conditions in the 43 programs found that fewer than 5 percent of the 1,306 conditions had high structural content (Table 2). More than half of the conditions had limited structural content, and the remaining 43 percent had little or none. 5 Conditions tended to be quite detailed and about one third reached outside the areas of core Fund competency and outside the areas of direct responsibility of the Fund team’s main counterparts. The analysis found no clear differences between the types of conditions that were used in PRGFs and in SBAs. It also found that with minor changes, the same conditions were applied as PCs in some cases and as SBs in others. Compared with SBAs, PRGFs had a higher proportion of conditions with limited structural depth (56 percent against 43 percent), partly reflecting the large number of one-off fiduciary actions linked to HIPC and other donor-led initiatives. The average SD of conditions was greater in core than in non-core sectors—on a scale from 0 to 2 they were 0.68 vs. 0.62 for PAs, and 0.71 vs. 0.60 for PCs—indicating that the Fund was supporting more ambitious sectoral reforms in the sectors of its core competence.

Table 2. Distribution of Structural Conditions by Depth, Program, and Type of Monitoring Instrument in the 43 Case Studies 1
Sector
Number
Sector Description Number of Conditions Share of Total
Depth All Low Limited High All Low Limited High Average
Depth 2
(In percent)
1 Tax Policy/Tax Administration 246 83 148 15 19 34 60 6 0.72
2 PEM 273 97 169 7 21 36 62 3 0.67
3 Financial Sector 293 154 128 11 22 53 44 4 0.51
4 SOE Reform 147 48 96 3 11 33 65 2 0.69
5 Privatization 83 53 21 9 6 64 25 11 0.47
6 Civil Service Reform 42 24 15 3 3 57 36 7 0.50
7 Social Policies 34 20 12 2 3 59 35 6 0.47
8 Other Fund Core 72 25 45 2 6 35 63 3 0.68
9 Other World Bank Core 116 54 56 6 9 47 48 5 0.59
Total All Sectors 1306 558 690 58 100 43 53 4 0.62
By Program GRA 335 167 143 25 26 50 43 7 0.58
PRGF 971 391 547 33 74 40 56 3 0.63
SB 646 310 306 30 49 48 47 5 0.57
By Type PC 260 99 147 14 20 38 57 5 0.67
PA 400 149 237 14 31 37 59 4 0.66
Core vs. Core 884 359 490 35 68 41 55 4 0.63
Non-Core Non-Core 422 199 200 23 32 47 47 5 0.58

Includes all 1306 fully assessed conditions in the 43 case studies from 1999–2003. Conditions not normalized to correct for program length.

Calculated as follows: “Low”=0; “Limited”=1; and “High”=2.

Source: MONA, IMF staff reports and IEO staff calculations.

17. A sectoral analysis conducted for the 43 countries supports this finding. For each program, the study identified the conditions with the greatest structural depth in each sector. This is a good indicator of how ambitious was the overall sectoral reform program, when sectoral strategies entailed a leading (high depth) condition that was complemented by others with lower SD, probably targeted at less critical aspects of the reform. In about 10 percent of the sectoral reforms, the leading condition had high SD, and in only a quarter of cases all conditions had little or no SD. The distribution of the leading conditions according to their SD was similar in PRGFs and SBAs. Again, this indicator points to more ambitious reforms in the core sectors: the leading condition generally had greater SD in core sectors than in non-core sectors. Among the core sectors, 83 percent had at least one condition with high or limited SD, against 61 percent among the non-core sectors.

18. In sum, programs contained a large number of structural conditions dealing with many aspects of policymaking. Often these conditions were quite detailed, even when they covered areas over which the Fund had little expertise and that were outside its core areas of responsibility. While critics may perceive these conditions as intrusive and detracting from government ownership of the reform process, fewer than 5 percent of the structural conditions in fact called for lasting structural reforms and more than 40 percent called only for preparing plans or drafting legislation. Still, almost 60 percent of the structural conditions would have had at least a limited or temporary impact, if complied with, and this share was somewhat greater in the core sectors and in particular in PRGFs. About half of the higher SD conditions were PAs or PCs, and the rest were SBs.

B. Compliance with Structural Conditions

19. More than half of the conditions (54 percent) were complied with on time (Table 3). 6 Compliance rates were significantly higher for conditions in the core sectors, particularly among PRGFs (60 percent compliance in core sectors vs. 39 percent in non-core) and among PCs across all types of programs (60 percent compliance in core sectors vs. 37 percent in non-core). Compliance rates were much lower (at less than one-third) for conditions with high SD—probably because these conditions are in general more difficult to implement—and they differed little between structural conditions with limited SD and those with no SD.

Table 3. Compliance with Structural Conditionality, by Depth and Type of Program
All Programs
Structural Depth All 0 1 2
Sector
Number
Sector Description Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
1 Tax Policy/Tax Administration 189 21 64 70 17 59 109 24 69 10 23 50
2 PEM 174 19 53 67 16 58 101 22 51 6 14 33
3 Financial Sector 209 23 56 111 27 58 90 20 58 8 18 13
4 SOE Reform 77 8 43 24 6 42 52 11 44 1 2 0
5 Privatization 51 6 29 33 8 30 12 3 33 6 14 17
6 Civil Service Reform 32 4 31 18 4 22 11 2 45 3 7 33
7 Social Policies 28 3 43 18 4 44 8 2 50 2 5 0
8 Other Fund Core 53 6 64 20 5 70 31 7 61 2 5 50
9 Other World Bank Core 93 10 54 48 12 58 39 9 49 6 14 50
Total All Sectors 1 906 100 54 409 100 53 453 100 56 44 100 32
Core 625 69 58 268 66 59 331 73 60 26 59 35
Non-Core 281 21 43 141 34 43 122 27 45 18 41 28
PRGF
Sector
Number
Sector Description Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
1 Tax Policy/Tax Administration 162 23 63 58 19 57 99 26 67 5 21 60
2 PEM 153 22 56 58 19 60 90 24 54 5 21 20
3 Financial Sector 135 19 61 64 21 69 64 17 58 7 29 14
4 SOE Reform 58 8 38 16 5 31 41 11 41 1 4 0
5 Privatization 37 5 30 26 9 27 11 3 36 0 0
6 Civil Service Reform 28 4 32 16 5 25 10 3 40 2 8 50
7 Social Policies 13 2 31 9 3 33 4 1 25 0 0
8 Other Fund Core 47 7 70 19 6 68 26 7 73 2 8 50
9 Other World Bank Core 78 11 49 39 13 51 37 10 49 2 8 0
Total All Sectors 711 100 54 305 100 54 382 100 56 24 100 29
Core 497 70 61 199 65 63 279 62 61 19 79 32
Non-Core 214 30 39 106 35 37 103 23 43 5 21 20
GRA
Sector
Number
Sector Description Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
Number Percent of
Total
Compliance
Rate
1 Tax Policy/Tax Administration 27 14 70 12 12 67 10 14 90 5 25 40
2 PEM 21 11 38 9 9 44 11 15 27 1 5 100
3 Financial Sector 74 38 47 47 45 43 26 37 58 1 5 0
4 SOE Reform 19 10 58 8 8 63 11 15 55 0 0
5 Privatization 14 7 29 7 7 43 1 1 0 6 30 17
6 Civil Service Reform 4 2 25 2 2 0 1 1 100 1 5 0
7 Social Policies 15 8 53 9 9 56 4 6 75 2 10 0
8 Other Fund Core 6 3 17 1 1 100 5 7 0 0 0
9 Other World Bank Core 15 8 80 9 9 89 2 3 50 4 20 75
Total All Sectors 195 100 51 104 100 52 71 100 54 20 100 35
Core 128 18 49 69 23 48 52 11 52 7 29 43
Non-Core 67 9 54 35 11 60 19 4 58 13 54 31

Includes 906 fully assessed conditions in the 43 case study countries excluding all PAs. Conditions not normalized to correct for program length.

Source: MONA, IMF staff reports and IEO staff calculations.

20. The average compliance rate varied widely across countries but it was not significantly correlated with the overall success of the corresponding program. While the overall average was about 50 percent, the compliance rate ranged from about 80 percent in some countries (e.g., Brazil, Guinea, and Mozambique) to less than one-third in others (e.g., Croatia, Ecuador, and Tajikistan). A similar degree of variability was observed in the sample of 13 in-depth case studies. On average, PRGFs had higher compliance rates than SBAs. Compliance varied significantly across sectors within individual programs, and in general it was higher in the core sectors.

21. The compliance rate was negatively correlated with the number of sectors covered by a program, but not with the number of conditions in a program. One possible explanation is that the multi-sector coverage of SC taxed the coordination and implementation capacity of the Fund’s main counterparts, even when conditions were not inherently very difficult to implement. Also, monetary and fiscal authorities were generally able to oversee and track compliance with large numbers of conditions in areas under their responsibility, but this was more difficult to manage in non-core areas. This is consistent with the findings of the in-depth case studies, which indicate that compliance was higher in areas under the direct control of the authorities managing the reform programs.

22. About one-quarter of all structural conditions had significant or limited SD and were complied with. This proportion was greater in core sectors, in particular for PRGFs, where it reached one-third. The proportion rises only a little (to about 40 percent) when allowing for conditions that were only partially met or met after a delay. By this criterion, the most “successful” structural conditions were PCs in core sectors in PRGFs, half of which were complied with, though this is still a relatively low rate.

23. It is surprising that compliance rates were so low, especially because almost half the conditions did not require policy or institutional changes. It is also surprising to find very little difference in the compliance rates between PCs and SBs, and especially among conditions with different degrees of structural content, since one might expect compliance rates to be linked to how difficult the conditions were to implement. Such low compliance rates pose a reputational risk to the Fund’s role in furthering structural reforms.

C. How Effective Is Structural Conditionality in Fostering Structural Reform?

24. Fund arrangements are designed to support members’ efforts to achieve medium-term external viability and to foster sustainable growth. SC plays a role in providing assurances to members that resources would be available to them upon compliance with agreed policies, in monitoring the member’s program, and as a mechanism to interrupt the Fund’s support when it becomes apparent that the agreed program goals are unlikely to be achieved.

25. This evaluation examined whether SC was effective in bringing about follow-up structural reforms as well as whether reforms were sustained over time. This is a particularly difficult question because there was no agreed-upon framework to assess results and accountability, and consequently data had not been collected specifically for this purpose. The evaluation used two approaches to address this issue. The first, based on data from the 43 countries, focused largely on reforms in sectors covered by SC; the second, based on the 13 in-depth case studies, looked at the impact of SC at the country level. To avoid intractable causality problems, the analysis asked whether SC had been a step within a broader sectoral reform—i.e., whether compliance with a particular condition was followed by additional reforms in the corresponding sector—but it did not try to establish a direct link. Admittedly, this criterion is a very low threshold for establishing effectiveness, since it links specific conditions to sectoral reforms that may be only marginally (if at all) connected to the corresponding condition, and with reforms that may have taken place without SC. In the 13 in-depth case studies, the evaluation asked about the impact of SC on the country’s overall policy framework, and also attempted to identify the determinants of effectiveness. Each of these approaches has limitations, but together they provide useful insights.

26. The sectoral analysis showed only a weak link between compliance with SC and effectiveness at bringing about reform and ensuring its durability. The analysis was conducted by assessing whether further reforms took place to advance the explicit or implicit objectives in sectors covered by SC. An individual condition was deemed effective if reform continued in the corresponding sector following compliance with that condition. For conditions whose compliance completed a reform, mostly conditions with high SD, the effectiveness test was whether reversals had taken place. Additional reforms took place beyond the program in about 55 percent of the sectors covered by SC. Reform stalled in almost 40 percent of the covered sectors and it backtracked in more than 5 percent. Rates of follow-up were somewhat higher in PRGF than in SBA countries, in core sectors than in non-core sectors, and for PCs than for PAs and SBs. 7

27. The most surprising finding from the sectoral analysis is that the effectiveness figures are almost identical regardless of whether conditions were met, met partially or after a delay, or not met at all. Nor did the analysis find any significant correlation between sectoral average compliance and follow-up reforms (this study’s indicator of the effectiveness of SC), either in individual sectors or at the aggregate level, across core and non-core sectors. The study also examined whether SC with greater SD was more effective in ensuring the continuity of reforms, but again it did not find any significant correlation. 8 In sum, this simple analysis suggests that compliance with SC was not, by itself, a good predictor of lasting sectoral reforms.

28. The 13 in-depth country studies examined whether SC was an effective tool to support economic reform and to strengthen a country’s overall economic framework, even if conditionality was only weakly correlated with reform at the sectoral level. SC might be effective even in these circumstances because Fund arrangements, and SC in particular, are designed, negotiated, and implemented mostly with the central bank and the economic ministries—agencies that are generally in charge of the country’s overall economic framework. This relationship partly explains the higher compliance with SC in the core sectors, which are under the direct control of these authorities. Also, Fund arrangements have often been put in place to support goals, such as catalyzing public or private financing, that are not directly connected with sectoral agendas. To address these issues, the 13 studies examined whether the presence of, and compliance with, SC affected the overall policy framework and the success of the program. Table 4 summarizes the assessment for each of the 13 cases (for a more detailed analysis see Background Document Chapter IV). Box 1 reflects the authorities’ views in these countries. The views of academics and civil society organizations were also sought; these were largely critical, although in some cases these commentators welcomed the PRSP consultation process (Box 2).

Table 4. Progress in Structural Reform in the Areas Supported by Structural Conditionality
Degree of
Progress
Stand-By Arrangements PRGFs
Satisfactory Colombia (2003): High access precautionary type to reduce market uncertainties and address long-term fiscal rules and quasi-fiscal issues. Good progress in spite of congressional opposition to several key measures. (2-year SBA, new administration, completed)

Romania (2001): Addressed successfully major quasi-fiscal issues in SOE, particularly in energy, in the context of EU accession. However, the program was not able to develop more permanent rules for price adjustments. (2-year SBA, new administration, high level of conditionality relative to access, completed)
Armenia (2001): Continuation of a series of post-program actions focusing on banking as well as fiscal and quasi-fiscal structural reforms. Good progress in banking and energy sector reforms. (Program completed)

Pakistan (2001): Program to help debt rescheduling and support reforms in taxation, financial sector, and energy. Good progress on banking, trade, and fiscal responsibility laws. Less so in energy. (Early cancellation at the request of government)
Moderate Croatia (2003): Short precautionary arrangements to avoid reversals in the run-up to elections, and in the context of EU accession. Good progress in labor market reforms, mixed result on state guarantees and energy sector. (14-month precautionary, very low access, program not completed) Mali (1999): Program subject to a terms of trade shock, streamlined midway focusing on public expenditure system in part due to HIPC. Central finances and PEM systems improved markedly. However, little progress on enlarged public sector: e.g., cotton and pension systems. (Program completed)
Limited Ecuador (2003): Ambitious program aimed at BOP support and passage of significant legislation with significant opposition by Congress. Very limited progress on legislation. Most of the measures were fiduciary (e.g., arrears clearance) in public finances. (13-month SBA, program not completed) Ghana (1999); Cameroon (2000); and Madagascar (2001): Program development heavily influenced by external and political shocks that refocused priorities on fiduciary issues of PEM, particularly as a response to HIPC objectives. Limited progress on PEM, revenue collections and SOE management and pricing, particularly in energy. (Only Madagascar program completed)
Little Effect Dominican Republic (2003): Program aimed at containing banking and BOP crisis at the end of the administration. No ownership to tackle structural reforms in taxation and decaying electricity sector. (2-year SBA, program not completed) Kenya (2001): Rather artificial program “pushed from the outside” to address major governance issues requiring legislation. (Program not completed)

Lao P.D.R. (2001): Program where strong differences in strategic directions emerged and became irreconcilable between staff and authorities. Nam Theun project financing was successful. (Program not completed)

29. Overall, the in-depth analysis found mixed results at the country level in the areas covered by SC and in the reform programs at large. Six programs were judged to have been satisfactory or moderately satisfactory (Armenia, Colombia, Croatia, Mali, Pakistan, and Romania), and the other seven (Cameroon, Dominican Republic, Ecuador, Ghana, Kenya, Lao P.D.R., and Madagascar) to have achieved limited or no progress. Experiences also varied across sectors in each country. SC was more successful in supporting reforms in PEM, taxation, and the financial sector than it was in supporting privatization, the reform of the (extended) public sector, or social sector policies. In general, reducing quasi-fiscal transfers and fiscal vulnerabilities in SOEs proved to be especially challenging. These findings are consistent with those of the sectoral analysis based on the 43 cases.

D. The Determinants of Effectiveness

30. What determines whether SC works? While overall SC was not very effective in fostering and sustaining structural reform, the significant variation across cases enables us to draw lessons on design features and country conditions that could enhance effectiveness. This section discusses design features (e.g., sectoral coverage) and country conditions (e.g., ownership of reforms) that may help SC to be more effective.

31. Effectiveness, like compliance, seems to have been higher in the core sectors. This seems to be linked to two interrelated factors. First, these are sectors where Fund staff has solid and widely recognized technical expertise and a good understanding of the implementation constraints faced by the authorities. These factors are conducive to better-designed conditionality that is more likely to be accepted by the authorities and other domestic stakeholders. Second, these sectors are usually under the direct control of the Fund’s main counterparts, suggesting that the authorities agreed to these conditions with a better understanding of how they would be implemented.

32. More generally, the case studies point to ownership of the reform program by a strong economic policymaking team as a critical precondition for success. A country’s economic team can sometimes use Fund SC as a lever to move a desired reform along. For sustaining a reform, however, broader government ownership—at least ownership of the specific SC by the corresponding implementing bodies—seems to have been a necessary condition. 9 Among the case study countries, those that made the most progress were those with the strongest government ownership of the reforms that were supported by conditionality (e.g., Armenia, Colombia, Pakistan, and Romania). Those that made least progress were those where the authorities had little interest in the corresponding reforms (e.g., Dominican Republic and Lao P.D.R.), or where the program largely responded to outside forces and lacked the broad support of the authorities (e.g., Kenya).

Box 1.Structural Conditionality: Views from National Authorities

The evaluation sought the views of the national authorities involved with the IMF-supported program in the countries covered by the 13 in-depth case studies. These authorities included officials at different levels and in different capacities in each country, mainly those that had negotiated the programs.

Views on program design and the process of negotiation differed, but in general, the authorities held more positive views about Fund conditionality in countries where programs had been more successful. Some, for example in Armenia, Colombia, and Romania, viewed structural conditions as deriving from relatively flexible negotiations, in support of an agenda largely developed domestically, although with varying degrees of support from outside of government. In other countries (e.g., Ghana, Madagascar, and Pakistan), however, the authorities took the view that IMF staff had been unnecessarily inflexible on specific policy requirements. In Ghana, for example, the authorities felt that staff was not flexible enough in the face of shocks that called for adjustments in policies. The Pakistani authorities noted, however, that they had perceived a gradual move toward a more consensual approach since the previous arrangement.

Although the authorities in some countries, notably Madagascar, pointed out that conditionality may have been excessive, this was not a matter of concern in Armenia or Romania—two of the countries that had the largest numbers of conditions. The Romanian authorities noted that neither the very large number of structural conditions (46 conditions per program-year) nor their very detailed nature posed a problem.

In many PRGF countries, the authorities saw SC as being imposed by donors (Kenya) or by the Fund (Lao P.D.R. and Madagascar), noting that the conditions were not adapted to the country’s institutional circumstances, implementation capacity, or political constraints. Their comments focused on unrealistic deadlines and on the need to have a meaningful consultation process. In Colombia, Ecuador, Kenya, and Pakistan the authorities explained that staff needed to understand better the political ramifications of conditionality, and that certain conditions may strengthen the opposition’s hand, particularly when requiring legislative changes.

33. Programs with good results had stronger analytical underpinnings in areas subject to SC. They had relatively well specified objectives and medium-term roadmaps that dealt with sequencing and that linked specific conditions to the distortions they were addressing (e.g., Armenia and Colombia). This was not the case in most of the PRGFs studied—which may be an important reason why these arrangements had disappointing results. These PRGFs often lacked appropriate medium-term policy roadmaps, and their outlook was surprisingly short run. 10 They rarely identified priorities and trade-offs for policy changes or specified the appropriate sequencing of reforms beyond a year’s time span, and most of their SC involved little long-lasting institutional change. Also, their program documents often failed to explain why a particular set of conditions was critical and why it represented the best available way to achieve the program objectives.

34. Outside the core sectors, close collaboration with the World Bank is critical to provide the necessary knowledge base for reform. Good collaboration with the Bank took place in the financial sector through participation in the Financial Sector Assessment Programs (FSAPs), which yielded detailed diagnostics and technical recommendations, some of which were incorporated in structural conditions. Coordination in other areas was more problematic, as the two institutions pursued different approaches to reform and conditionality. A comparison of privatization conditionality in the sample of PRGF arrangements with parallel Bank operations illustrates these difficulties (Table 5). The comparison found that PRGFs often had precise conditions (with fixed timetables and short horizons), while the Bank’s parallel operations had no conditionality, or had conditions that were very general in the context of a programmatic type of loan with flexible tranching arrangements. More generally, the Bank was moving toward ex post “supporting success” and emphasizing country ownership of programs rather than setting conditions in advance. Despite these differences, the case studies found productive collaboration at the country level in Armenia, Colombia, Croatia, Pakistan, and Romania. Collaboration was weaker in the Dominican Republic, Lao P.D.R., and Madagascar, where insufficient prior knowledge on structural issues hampered program design.

Table 5. Overlapping IMF-WB Conditionality in Privatization-Related Conditions in PRGF Programs
Country Sector IMF Condition WB Condition in Parallel Operation WB Operation
Madagascar Cotton Completion of the bidding process for the sale of the cotton company HASYMA (PC, 4th review, not met).
Completion of the bidding process for the privatization of HASYMA (PA, 5th review, met).
Ghana Cocoa Offer Produce Buying Co. (PBC) for sale (PC, not met on time; met subsequently as PA for the 1st review). Offer PBC for sale (condition complemented by others conducive to the sale of PBC). ERSO II (5/99–6/03)
Electricity Appoint sales advisor for divestiture of the electricity company (ECG) (PC, met). Borrower to send a ‘Request for Proposals’ to potential advisors in a short list (condition complemented by others conducive to the sale of PBC). ERSO III (7/01–12/02)
Oil Offer the Tema Oil Refinery (TOR) for sale (PC, not met on time; met as a PA for the 5th review).
Cabinet approval of a debt restructuring plan for TOR (PA for the 1st review, met).
Transport Complete divestiture of State Transport Co. and cement company GHACEM (PA for program request, met).
Telecom Select an advisor for the sale of at least 10 percent of the government’s shares in Ghana Telecom (PA for the 2nd review, met).
General Provide the IMF with a divestiture work program, including annual targets for divestiture receipts for 1999–2001 (PA for program request, met).
General Appoint sales advisor for divestiture of Ghana Airways and Ghana Railways (PC, partially met).
General Cabinet approval of a list of government assets valued at not less than US$50 million to be sold in 2001, including estimated dates of sale (PA for the 3rd review, met).
Mali Water, Cotton Launching of the call for bids for the sale of assets of the cotton (CMDT) and water (OHVN) companies, necessary for the installation of a private operator in the OHVN zone (PC, met with delay). Bring to the point of sale all assets belonging to CMDT and OHVN in the OHVN Zone. SAC III (12/01–6/04)
Transport Call for bids for the privatization of the management of Airports of Mali ADM (SB, not met).
Electricity Bids for capital of the energy company (EDM) (PC, met with delay).
Manufacturing Divest government minority shareholdings in canned products (SOMACO), air services (MALITAS), chemical products (SMPC), and machinery (SEMA) firms (PA for program request, met).
Tobacco Sign agreement for privatization of the national society of tobacco and matches SONATAM (PA for program request, met).
Telecom International call for bids to grant at least one cellular telephone operating license to the private sector (SB, met with delay).
General Adoption of a 1999–2002 action plan to reduce the number of non-bank public enterprises in the government’s portfolio from 33 to 18 (PA for the program request, met).
Cameroon Water Resume discussions with the successful bidder for the water company (SNEC) based on the evaluation of the offer conducted with World Bank assistance (SB, met). Publication of the tender document for the privatization of SNEC and two other SOEs. SAC III (6/98–3/04)
Issuance of the invitation to the successful bidders for SNEC and five other SOEs to start negotiations. SAC III (6/98–3/04)
Telecom Electricity Forward to the IMF and the World Bank the Head of State’s notification letter validating the recommendations on privatization of the Inter-ministerial Committee (CIM) with respect to the final offers of the provisional adjudicators for the privatization of the telecom (CAMTEL) and electricity (SONEL) companies (PA, met). Issuance of the invitation to the most successful bidders for CAMTEL and three other SOEs to start negotiations. SAC III (6/98–3/04)
Telecom Issue an invitation to negotiate to the successful bidder of CAMTEL
Pakistan Telecom Bring telecommunications company PTCL to the point of sale through transparent and open public offer for sale (SB, not met). Bring to the point of sale the following firms: PTCL, KESC and five other SOEs. Follow-up actions (FY03–05)
Electricity Bring electricity company KESC to the point of sale (PC, not met).
Cabinet decision on KESC privatization strategy (PA for the 1st review, met). Cabinet approval of a short- and medium-term strategy and timetable for privatization focused on the critical large SOE in power, telecom, banking, oil and gas (not a trigger). SAC (6/01–12/01)

Box 2.Structural Conditionality: Views from Civil Society

Civil society organizations have been at the forefront of the criticism of SC. Briefly, their three main lines of criticism are as follows:

  • Conditions are typically imposed by donors and not owned by countries, and hence a prerequisite for successful development is missing. Eurodad argues that this imposition overwhelms domestic capacities and undermines whatever local ownership exists for those policies (Eurodad, 2006). Wood and Lock-wood (1999), of the Bretton Woods Project, explain that the proliferation of SC and its damaging effect on national ownership has led countries to comply with conditionality only for “tactical reasons” and that after receiving aid, countries reverse the enacted policies. Oxfam has raised concerns about the legitimacy of policies agreed with the IMF on grounds that key sectors of society were not properly consulted or were excluded from program negotiations (Oxfam International, 2004).

  • Donors often advocate policies and conditions based on dogma and ideology rather than on evidence. Critics give trade and domestic price liberalization and privatization as examples. Action Aid (2004) discusses the pervasiveness of conditions aimed at privatization of water and electricity services in lower-income countries, despite the absence of evidence that these policies improve access for poor people, accountability to consumers, or cost effectiveness.

  • Tying aid to conditions causes unpredictability and volatility in aid disbursements, which reduces the usefulness of aid. To address deeply rooted issues, such as poverty or limited access to health services, a country needs a predictable stream of aid on the basis of which it can plan the necessary programs. Lack of compliance with conditionality disrupts these programs. A recent report by the Jubilee Debt Campaign identifies IMF conditions in PRGFs as the main reason for borrowing countries’ delays in securing HIPC debt relief (Jubilee Debt Campaign, 2006).

35. In parallel to their role in supporting stabilization and structural reform, Fund arrangements were used by donors and others as monitoring and signaling mechanisms for other initiatives. For example, they acted as a trigger for HIPC process milestones, served to monitor reforms undertaken toward EU accession and helped to mobilize financing from multilateral development banks, the Paris Club, and official donor agencies. 11 This role was widely acknowledged, although not always explicitly stated in Board documents. In most instances, the presence of SC was important in allowing the Fund arrangement to play these roles, and sometimes these monitoring and signaling needs were what drove the SC agenda and the timing of the arrangement. Some programs were rushed to enable countries to reach the HIPC decision point by an exogenously set deadline. Sometimes conditions were included which, while important for the monitoring role, were not critical to the explicit program objectives. The proliferation of conditions weakened the authorities’ attention to those conditions that called for more critical reforms. In general, Fund arrangements and SC were effective in this role of providing monitoring and signaling.

36. Notwithstanding the successful use of SC as a monitoring and signaling mechanism, their use for this purpose raises important questions. Should this be a legitimate role for Fund arrangements and SC? Should Board documents state that this is a major goal of the arrangement? It appears that there are trade-offs between the role of SC as a signaling or monitoring mechanism and its role in fostering reform. To the extent that such trade-offs are significant, could these other roles be played by other instruments, either new ones or those that already are part of the IMF’s toolkit?

In five cases (Azerbaijan 1996, Pakistan 1997, Yemen 1997, Macedonia 2000, and Sri Lanka 2003), two facilities—the PRGF and the EFF—were used to support the country’s program. In these cases, the program framework was given by the PRGF while the EFF resources were used to “top up” those provided under the PRGF. In this evaluation, these programs were counted only once as PRGFs.

If the authorities disagree with management on the completion of a review, a Board meeting could still be convened on the subject at the request of an Executive Director. However, in these circumstances, management could withhold its recommendation to the Board to complete the review.

The CG define IMF core areas of responsibility as macroeconomic stabilization, namely, fiscal, monetary and exchange rate policies, including the underlying institutional arrangements and related structural measures; and financial systems issues related to the functioning of both domestic and international financial markets.

The definitions of the categories are intended to be comprehensive and exclusive, i.e., each condition falls into one and only one category. The classification was done with a bias toward ascribing to each condition the highest possible SD category, as is illustrated by the examples in Table 1. About a dozen conditions that did not fit clearly into a single category were classified as having Limited SD. See Background Document Chapter III.

Two clarifications about the classification. First, the approval of a specific budget law was classified as having Limited SD. While this usually requires legislative action, one year’s budget, per se, would not necessarily have a lasting impact in that it needs to occur each year. Second, the submission of legislation to parliament was classified as having low SD because by itself it does not bring about any meaningful economic changes. In certain regimes, most laws submitted to parliament are approved, but the corresponding passage of legislation is captured in the analysis of effectiveness or follow-up reforms. In any case, classifying the submission of laws as having high SD increases the share of this category to about 10 percent of the total, and does not change the rest of the analysis in any significant way.

The compliance rates are similar for the 43 countries shown in Table 3 and for the 7,139 conditions in MONA. These numbers do not include PAs because they are met almost by definition, as the database does not include conditions in programs that have gone off track. Including PAs, the compliance rate would rise to about two-thirds; on the other hand, the rate would be less than half if programs that went off track were included (excluding PAs). According to MONA, about half of the structural conditions that were not met on time were eventually met with delay or met only in part. MONA did not distinguish between these two categories until 2001, and it still does not specify the extent of the delay nor of the compliance. In any case, this information is not available for most of the period of this study. These figures compare with a compliance rate of more than 80 percent for quantitative macro conditions (IMF, 2005a, p. 28).

The 55 percent follow-up figure has a significant upward bias since in many cases the specific conditions were not connected with the sectoral reforms that took place subsequently. This bias is not present in regard to reversals, where the figure refers to the specific measures that were taken. The analysis was based on staff reports for the first Article IV consultation after expiration of programs, ex post assessments, and requests for a new arrangement, when relevant.

In fact, sectors that were the object of SC with high SD had a higher than average rate of policy reversals (more than 10 percent, against 6 percent for the whole sample). This may be due to a selectivity bias, i.e., the IMF may be more inclined to ask for conditions with high SD when the authorities’ ownership of the reforms is low or when the underlying situation is particularly difficult.

This study did not test the role of the more generally accepted concept of ownership, which includes support by different constituencies in the population for the policies carried out by the authorities.

The lack of a roadmap in PRGFs was also noted in the IEO evaluation of the IMF’s role in PRSPs and the PRGF (IEO, 2004).

Among the 13 case study countries debt restructuring took place in the Dominican Republic, Ecuador, Kenya, and Pakistan; and Lao P.D.R. was able to mobilize financing for the Nam Theun hydroelectric project. Fund arrangements provided a monitoring framework for the EU accession process, explicitly (e.g., Romania) or implicitly (e.g., Croatia).