2011 Review of Conditionality - Content and Application of Conditionality

This paper reviews the design of conditionality in Fund-supported programs from 2002 to end-September 2011, with an emphasis on recent years. It focuses on the content and application of program conditionality—especially structural conditionality—in relation to the 2002 Conditionality Guidelines (the "Guidelines"), the Staff Statement on Principles Underlying the Guidelines on Conditionality, and subsequent revisions to operational guidance on conditionality. The analysis is based on the five key interrelated principles guiding the design of conditionality: national ownership of programs, parsimony in program-related conditions, tailoring to country circumstances, effective coordination with other multilateral institutions, and clarity in the specification of conditions. In particular, the principle of parsimony requires that program-related conditions be critical (or the minimum necessary) to achieve program objectives and goals, critical for monitoring program implementation, or necessary for implementing specific provisions under the Articles of Agreement (the "criticality criterion"). Beyond assessing compliance with these guidelines and principles, the paper also examines the implementation of conditionality

Abstract

This paper reviews the design of conditionality in Fund-supported programs from 2002 to end-September 2011, with an emphasis on recent years. It focuses on the content and application of program conditionality—especially structural conditionality—in relation to the 2002 Conditionality Guidelines (the "Guidelines"), the Staff Statement on Principles Underlying the Guidelines on Conditionality, and subsequent revisions to operational guidance on conditionality. The analysis is based on the five key interrelated principles guiding the design of conditionality: national ownership of programs, parsimony in program-related conditions, tailoring to country circumstances, effective coordination with other multilateral institutions, and clarity in the specification of conditions. In particular, the principle of parsimony requires that program-related conditions be critical (or the minimum necessary) to achieve program objectives and goals, critical for monitoring program implementation, or necessary for implementing specific provisions under the Articles of Agreement (the "criticality criterion"). Beyond assessing compliance with these guidelines and principles, the paper also examines the implementation of conditionality

I. Introduction

1. This paper reviews the design of conditionality in Fund-supported programs from 2002 to end-September 2011, with an emphasis on recent years.1,2 It focuses on the content and application of program conditionality—especially structural conditionality—in relation to the 2002 Conditionality Guidelines (the “Guidelines”), the Staff Statement on Principles Underlying the Guidelines on Conditionality, and subsequent revisions to operational guidance on conditionality.3,4 The analysis is based on the five key interrelated principles guiding the design of conditionality: national ownership of programs, parsimony in program-related conditions, tailoring to country circumstances, effective coordination with other multilateral institutions, and clarity in the specification of conditions.5 In particular, the principle of parsimony requires that program-related conditions be critical (or the minimum necessary) to achieve program objectives and goals, critical for monitoring program implementation, or necessary for implementing specific provisions under the Articles of Agreement (the “criticality criterion”). Beyond assessing compliance with these guidelines and principles, the paper also examines the implementation of conditionality.

2. The analysis builds on earlier reviews of conditionality.6 The main focus of the 2001 Review of Conditionality (RoC) was structural conditionality. It led to efforts to streamline and focus conditionality and to enhance ownership of programs, including through significant revisions to the Guidelines. The 2005 RoC indicated notably that (i) coverage of structural conditionality had shifted towards critical measures, but the number of structural conditions had not decreased substantially; (ii) conditionality could not substitute for program ownership in order to achieve program objectives; (iii) program-related conditions had become clearer; (iv) Fund-supported programs experienced fewer permanent interruptions; and (v) waiver rates had not declined, but implementation, measured as the number of waived PCs that were eventually met, had improved.7 The 2007 IEO Evaluation of Structural Conditionality confirmed the trend in the composition of conditionality toward core areas of Fund expertise (macroeconomic stabilization and financial systemic issues). It also found that the volume of conditionality had not decreased and that some conditionality was not considered critical for achievement of program goals.

3. Following each of these reviews, operational guidance to Fund staff was revised.8 After the 2005 review, revisions highlighted the importance of country ownership and provided updated guidance on the design of conditionality as well as on presentational requirements for Board papers. The 2008 revision implemented Board-endorsed IEO recommendations to strengthen efforts to achieve parsimony by emphasizing criticality and requiring rigorous justification of conditionality. A subsequent revision occurred in early 2010 after the reform of the Fund’s conditionality framework in March 2009 and the introduction of a new architecture of facilities for low-income countries (LICs) in June 2009.9 In particular, structural performance criteria (SPCs) were discontinued in all Fund-supported programs, with a shift to a review-based approach to monitor structural reforms in programs. The aim was to enhance the flexibility of the conditionality framework, to tailor conditionality to countries’ varying characteristics, and to reduce the stigma associated with SPCs and associated waivers, while preserving adequate safeguards for the use of the Fund’s resources.

4. This paper undertakes a quantitative and qualitative assessment of program conditionality. It assesses all forms of conditionality in Fund-supported programs (hereafter also programs) meeting the standard of upper credit tranche (UCT) conditionality.10 In addition to examining SPCs, structural benchmarks (SBs), and quantitative performance criteria (QPCs) as in previous reviews, it more systematically studies prior actions (PAs) and indicative targets (ITs) and incorporates analysis of the depth of structural conditionality.11 The inclusion of PAs and ITs is especially important given the March 2009 revisions to structural conditionality and the consequent questions as to how their use might have changed with the discontinuation of SPCs. ITs are an important element to inform Executive Board’s reviews of program performance and are instrumental to assess the greater emphasis on social protection and the changes in Fund policy regarding public wage bill ceilings in programs.

5. In analyzing conditionality, the paper draws on a wide range of sources.12 This includes statistical analysis based on the Monitoring of Fund Arrangements database; and qualitative assessments based on 23 recent Ex Post Assessments of Longer-Term Program Engagement (EPAs) and Ex Post Evaluations of Exceptional Access (EPEs); 18 case studies representative of the sample of programs during 2006-10; structured interviews with Fund Executive Directors; surveys of country authorities, in-country donors, and Fund mission chiefs and Resident Representatives for countries with programs since 2007; and views from headquarters-based donors, civil society organizations, and other stakeholders.

6. The main findings of the paper are the following:

  • Conditionality became more focused and was generally aligned with program goals. Conditionality was also tailored to country and program characteristics, as well as to initial macroeconomic conditions.

  • Conditionality became more parsimonious. In recent years, the number of conditions per review and per year of ongoing program declined, supported by the discontinuation of SPCs. Structural conditionality, however, increased somewhat in GRA programs in 2011. Conditionality also focused increasingly on areas of Fund core competencies, with fiscal measures taking center stage.

  • ownership was critical to program success. While frontloaded conditionality can be used to support program implementation, experience confirmed that it cannot substitute for ownership. Ownership can be supported by outreach and communication, program flexibility, focused conditionality, and increased clarity. There is some evidence that ownership—while abstract and unobservable—improved on balance, with some possible new issues in 2011.

  • The balance between tailoring and evenhandedness was mostly appropriate. In particular, the design of conditionality tended to match country capacity as well as access granted under the program.

  • Coordination was important for program success. Coordination with other institutions generally supported parsimony and made conditionality more effective by avoiding duplicate measures. Coordination with European institutions on some European programs, however, faced greater challenges: in the midst of a crisis, European Union (EU) partners needed to develop crisis-fighting tools and enhance knowledge of adjustment programs. This experience suggests that coordination with potential institutional stakeholders could be more effective if set ahead of crisis times.

  • Progress in the implementation of program conditionality stalled in 2011. The parsimony of conditionality, together with flexible timelines for program implementation, generally supported program execution through 2010. Difficulties in 2011 likely reflect more challenging economic conditions and adjustment needs.

  • Compliance with the guidelines regarding the clarity of conditionality was uneven. Best practices on clarity need to be applied more broadly in program documents.

7. Going forward, it will be crucial to build on earlier efforts to enhance the effectiveness of conditionality.

  • Focus and parsimony of conditionality may come under pressure in a more challenging global environment. At the time of the financial crisis, large liquidity needs along with a sense of urgency supported focused and parsimonious conditionality. As structural reform and adjustment needs become deeper, notably in the second wave of programs started in 2010, more wide-ranging conditionality may be needed. It will be important to monitor closely conditionality, to ensure that measures remain the minimum necessary to address country-specific issues and ensure the success of Fund-supported programs.

  • Coordination with institutional partners may become more difficult. With more institutions and donors involved in program design, increased financing needs, and relatively constrained financing capacities, efforts to coordinate policies and conditionality with other institutions will be essential, especially in co-financed programs.

  • Communication and transparency on program goals, strategies, and linkages to conditionality could be improved. This could support program ownership and improve coordination with donors by enhancing the understanding of program design and conditionality by all stakeholders. Mandating improved clarity in program documentation and communication could help ensure that conditionality remains focused on critical measures.

  • Surveillance could be leveraged more effectively. A case-study analysis shows that about half of program structural conditionality was identified as needed reforms during pre-program surveillance.

8. The remainder of this paper is structured as follows. Section II examines application of the principle of parsimony in Fund-supported programs, notably in the context of streamlining initiatives by the Fund over the past decade. Section III reviews the recent experience with the other four key principles of conditionality, and Section IV the implementation of conditionality during the review period.

II. Streamlining and the Principle of Parsimony

9. Conditionality in Fund-supported programs aims to help countries solve their balance of payments problems and to safeguard Fund resources. According to the Conditionality Guidelines, program-related conditions should be (i) of critical importance to achieve program goals or to monitor program implementation, or (ii) necessary to implement specific provisions under the Articles of Agreement or policies under them.

10. Initiatives to streamline program conditionality have been central to Fund efforts since at least 2000, following the issuance of the interim guidance note on streamlining structural conditionality.13 That note followed a long process of reflection, which helped lead to the revised conditionality guidelines in 2002.

11. Parsimony is a key principle of streamlined conditionality. It implies a clear focus of program conditions on achievable goals, which “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; and (b) achieving medium-term external viability while fostering sustainable economic growth” (Guidelines, ¶6).

12. Streamlining does not preclude individual programs from having substantial conditionality when needed for the success of the program. Under streamlining, structural conditionality should be both parsimonious and macro-critical (i.e., conditions should be of critical importance for achieving the goals of the member’s program or for monitoring program implementation, and should be limited to the minimum necessary). For example, significant growth-enhancing reforms could be needed to ensure domestic and external stability and sustainability. The Guidelines indeed stress that the criticality criterion applies to all reform measures, whether within or outside the Fund’s core areas of responsibility. Setting conditionality on non-core critical measures, however, requires a strong justification and a more detailed explanation of their criticality, and the Fund is expected to draw on the expertise of other institutions, including the World Bank and donors to the extent possible, for those measures.

13. The analysis that follows examines the focus and parsimony of conditionality. On focus, the underlying assumption is that conditions critical to the success of Fund-supported programs are more likely to be within the Fund’s core area of responsibility. Thus, success in streamlining would suggest greater focus and lower volumes, at least on average, for a given set of conditions. Nonetheless, country needs, as well as global conditions, can change over time, and changes in focus and volume of conditionality may not be solely related to streamlining efforts. For example, during global crises, it is more likely that “innocent bystanders” (or those with strong fundamentals and policies) could be hit by exogenous shocks, resulting in substantial needs for liquidity and financing but limited needs for adjustment or structural reforms. Hence, improved focus and reduced volumes in recent years could have resulted from the underlying global circumstances.

A. Program Goals, Instruments, and Conditionality: Focus and Alignment

14. Program instruments were aligned with key program objectives (Figure 1).14 All programs identified external and macro-economic stability as overarching goals. GRA-supported programs (hereafter GRA programs) centered on external and macro-economic stability; an increasing number of programs included growth objectives (from 40 to 60 percent of programs between 2008 and 2011), reflecting mounting concerns about growth in emerging market (EM) and advanced market (AM) program countries. Almost all PRGT-supported programs (hereafter PRGT programs) aimed also to achieve poverty reduction and economic growth.15 Fiscal and monetary policies were the most frequently used instruments in support of program goals, with some differences across program types.16,17 GRA programs relied more strongly on financial sector policies, while PRGT programs demonstrated their emphasis on growth and poverty reduction by employing a wider set of pro-growth and social sector policies. Program goals, program instruments, and structural conditionality also were more focused for countries with weaker initial conditions (Box 1).

Figure 1.
Figure 1.

Program Goals and Instruments, 2006-2011

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

Program Instruments and Structural Conditionality

Fund-supported programs adjusted goals and strategies for differences in initial macroeconomic conditions. Countries were considered to have weaker initial conditions if any of six key macroeconomic variables were below (or above, as relevant) a threshold derived from the sample of countries with programs. These variables were: real GDP growth, inflation, current account balance, fiscal balance, net international reserves, and medium- and long-term debt. For example, if a country’s growth rate was below the mean minus a VV standard deviation, or a country’s inflation rate was above the mean plus a VV standard deviation, then the country was considered to have weaker initial conditions. Program goals and instruments (or reform strategies) set for these programs appeared to be more focused; for a subsample of programs with weaker initial conditions (indicated by broken lines in Box Figure), the probabilities of finding programs with certain program goals or reform strategies were lower than those for the full sample (indicated by continuous lines).18

The choice of reform strategies subsequently affected the choice and focus of structural conditionality (SC). The analysis compares (1) the probability of having at least one condition set on a sector when a given set of reform strategies has been identified to (2) the probability of having conditionality set on the same sector, irrespective of program strategies. In a number of sectors, structural conditions were set, irrespective of the choice of reform strategies. For example, there was at least one structural condition on the government sector for almost all PRGT programs and one on the financial sector for almost all GRA programs. However, on other sectors, the probability of setting a structural condition increased when relevant reform strategies were chosen. For example, having a specific reform strategy for the monetary sector increased the probability of finding a structural condition on the central bank. Similarly, the probability of finding a structural condition on the public enterprise was higher for programs with reform strategies on the public enterprise. This evidence was present for both PRGT and GRA programs (top panels below), and also for the subsample of programs with weaker initial conditions (bottom-left panel). Having pro-growth and social sector reform strategies in PRGT-supported programs increased the probability of setting a structural condition on public enterprise, reflecting growth strategies via an efficiency gain (bottom-right panel).

15. Structural conditionality was in general broadly aligned with program goals and tailored to country circumstances (Boxes 1 and 2 and BP2). PRGT programs included more wide-ranging structural reform agendas, while capital account challenges, along with concerns about the size of spillovers and systemic risks, in some GRA programs were reflected in higher access levels. Program design took into account political economy considerations (see summary of EPAs/EPEs in BP4) and spillover effects of the global financial crisis (see BP2).

Structural Conditionality Alignment and Tailoring

This box assesses whether conditionality was appropriately aligned and programs properly tailored in a sample of program countries.19 The main conclusions are that (1) program conditionality was generally well aligned with program goals and reform strategies, although there was room for improvement; and (2) programs were generally tailored to country-specific circumstances.

In most cases structural measures were in line with the programs’ stated objectives, though not all objectives were covered by structural measures (Box Table).20 This was equally true for GRA and PRGT countries; for both sets around 80 percent of conditions on average could be linked to specific objectives.21 By contrast, not all objectives were covered by structural reforms: In PRGT countries, about 60 percent of objectives were addressed by structural reforms and 70 percent in GRA countries. Full coverage would not be expected given that certain objectives, like reducing inflation or the deficit, are achieved with QPCs and not structural measures. Nonetheless, there is a large variation across countries, with as little as one third of objectives covered by structural measures in some countries.

Generally, program design and conditionality were focused and tailored to country circumstances. Structural conditionality was typically specific to the country’s policy and institutional setting. When programs were designed in close coordination with other institutions and donors, Fund conditionality was focused on areas within its core area of responsibility. In particular:

  • The particular lending facility chosen for the program depended on the scope and nature of the needed reform agenda. The Fund offers a variety of lending facilities, catering to the specific needs of members. Some allow for more time than others to implement an ambitious structural reform agenda, notably the ECF for PRGT-eligible countries, and the EFF for GRA countries. The FCL and the PLL (formerly PCL) offer support to relatively stronger performers with a commensurate reduction in conditionality. Countries may move from one facility to another, e.g., in 2009, Seychelles from a SBA to an extended arrangement and Armenia from a SBA to an ECF-EFF blend arrangement.

  • Program duration and the timing of conditionality were in line with country characteristics. Among case study countries, PRGT programs tended to have more structural measures than GRA programs in the original program design, reflecting their heavier structural agendas. However, the number of conditions to be implemented in the first three months was lower than in GRA programs (2 compared to 3); this, together with the longer time span on PRGT programs and typically semi-annual reviews, facilitated program implementation in a context of weaker administrative capacities.

  • Past performance and political cycles played a role in designing program conditionality. The 2007 Burkina Faso program followed 10 years of strong program implementation (according to the EPA), which was reflected in lower fiscal adjustment, no reliance on PAs, a number of structural conditions close to the average for the sample, and the need for only small access. The one-year 2006 Uganda PSI was to be replaced by a multi-year PSI later in the year, so as to promote ownership by delaying discussions on substantive medium-term policies until after the elections. In contrast, the 2007 Gambia program, which was approved after several programs with mixed success (attributed by the EPA to a lack of ownership), had three PAs and an above average number of structural measures.

  • Program conditionality usually covered multiple policy areas, but there were instances of strong focus responding to the diagnosis of the problems and priorities. For instance, conditionality for the 2008 Armenia program heavily emphasized tax reforms, the 2008 Iceland one stressed banking reforms, and the January 2006 Uganda one stressed domestic arrears.

Alignment of Objectives and Structural Conditionality

article image
Source: MONA and staff calculations.

16. While EPAs and EPEs generally confirm that conditionality was well-aligned with the key program objectives, there were important exceptions (BP4). In some instances, EPAs and EPEs saw scope to align the reform agenda more consistently with program objectives, and to sequence reforms in order to overcome capacity constraints (Sierra Leone). In other cases, they assessed that the macro-criticality of structural measures could have been demonstrated more clearly in order to ensure focus on core areas of Fund expertise and provide a better sense of direction for policy makers (Benin, Burundi). In a few countries, EPAs and EPEs considered that better diagnostics of underlying institutional and political constraints would have enhanced the effectiveness of conditionality (Belarus, Ukraine, and Kyrgyz Republic). These ex-post exercises also deemed that some programs suffered from lack of appropriate follow-up conditionality, once the initial set of reforms had been completed.

17. Survey results also point to the alignment of program conditionality with program objectives (Figure 2). While Fund staff and country authorities generally agreed that conditionality had been geared toward critical program objectives, responses from donor agencies were more cautious, suggesting room to explain the criticality of Fund conditionality more coherently.

Figure 2.
Figure 2.

Program Objectives and Conditionality

Outcome from questionnaires (percent of respondents)

Statement: The structural conditions in the program only covered reforms that were critical for program objectives, design and implementation

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

uA01fig01
Source: MONA and staff calculations.Probabilities In these charts show the probability of finding a program with corresponding structural conditionalrty In the full sample. Different lines represent conditional probabilities where the conditions are having different reform strategies. The 13 MONA categories for reform strategies are aggregated into 5 categories: Fiscal (Public Expenditures and Fiscal Revenue). Monetary (Monetary stability, stable inflation. Central Bank Reform, and Exchange Rate issues). Financial Sector Reform, Public Enterprises, and Others (Trade Reform. ProKjrowth Regime. Social Outcomes. Economic Governance, and Other).

B. Focus of Structural Conditionality

18. To an increasing extent, structural conditionality has focused on core areas of Fund expertise (Figure 3).22 In principle, conditionality can be applied to areas beyond core Fund expertise if critical for program success. However, the Guidelines state that conditionality that would be critical to program success “will normally consist of macroeconomic variables and structural measures that are within the Fund’s core area of responsibility” (Guidelines, ¶7 (b)). The trend toward conditionality focusing on core Fund competencies picked up in recent years, particularly for GRA programs, reflecting an increasing focus on fiscal-related conditionality.

Figure 3.
Figure 3.

Breakdown of Structural Conditions into Fund Competencies

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

source: MOMA and staff calculations.1 Includes SPCs. SACs. SBs. and priar actions.

19. Fiscal conditionality has dominated the structural reform agenda in Fund-supported programs, with much of it drawn from the Fund’s technical assistance (Figure 4). In PRGT programs, in particular, government-related structural measures accounted for about 58 percent of total structural conditionality during 2008-2011, an increase of more than 7 percentage points compared to the preceding three-year period. Financial sector reforms were the second most important area of structural reforms, notably in GRA programs. As the crisis moved from a financial crisis to a deeper, real-sector crisis, the focus on the financial sector peaked at the onset of the global financial crisis, after which fiscal issues increased in prominence, including in the European Union and Iceland programs (Figure 5).

Figure 4.
Figure 4.

Focus of Structural Conditionality

Distribution of conditions by sector, in percent of total conditions per review

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.
Figure 5.
Figure 5.

Fiscal and Financial Sector Conditionality in GRA-Supported Programs1

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 EU programs include Iceland, Greece, Hungary, Ireland, Latvia, Portugal, and Romania.

20. The breakdown of structural conditionality by sector was determined by the extent of vulnerabilities and the availability of fiscal space (Table 1 and Appendix 2). For instance, general-government or fiscal related conditionality in GRA programs was more likely to be higher with better economic conditions (lower inflation, higher growth), while conditionality in the financial sector was more likely to be important when there are larger exposures (total assets and liabilities) and smaller fiscal deficits. In PRGT programs, more general government conditions were more likely with higher debt levels, lower reserves, and smaller fiscal deficits.

Table 1.

Determinants of the Probability of a High Number of Structural Conditions 1

article image

This table summarizes the results of probit models relating the probability of setting a large number of conditions (defined as a number of conditions belonging to the highest quartile of the distribution) to macroeconomic conditions at program approval. Probit models are run separately on the main categories of structural conditionality. Only statistically significant variables are summarized. Detailed results appear in Appendix 2.

21. Efforts to enhance the focus of programs can be seen in conditions set on public financial management (PFM), macro-social conditionality, and wages. PFM conditionality has typically been sequenced to address the most important vulnerabilities (Box 3). Conditionality on macro-social critical reforms, while remaining parsimonious, has gained in importance in recent years (Box 4). Finally, refocusing conditionality on areas critical to macroeconomic stability also applied to public wage ceilings. While ceilings— specifically, QPCs—on civil service wages were included in programs in the early 2000s, this practice was discontinued in 2007 given concerns about their impact on poverty-reducing spending (Box 5).

Conditionality in the Area of Public Financial Management

Fund conditionality in the area of public financial management (PFM) can be grouped into five main categories. These are: (i) stocktaking and control of arrears; (ii) financial accounting, reporting, and auditing; (iii) expenditure controls; (iv) legal and institutional frameworks; and (v) more advanced PFM reforms.

  • With respect to arrears, conditionality in countries with weak PFM systems has helped to build up the capacity to identify, verify, report, and clear public sector arrears, and helped to create a more efficient framework for preventing arrears and strengthening expenditure controls. These measures were usually associated with the need for fiscal consolidation, resulting from the fiscal crisis that triggered the program.

  • The focus of conditionality related to financial accounting and reporting has been to improve the coverage and quality of fiscal data by strengthening accounting classification, improving the timeliness of fiscal reporting, and enhancing auditing procedures.

  • Once reasonable progress has been made in reporting arrears and improving fiscal information, the emphasis of conditionality has shifted to expenditure controls and the monitoring of fiscal indicators, particularly by (i) strengthening budget execution systems, including commitment controls and active cash management; (ii) developing procurement and payroll systems; (iii) tracking poverty-related expenditure; and (iv) monitoring fiscal indicators, such as the budget deficit and debt.

  • Conditionality has also aimed to strengthen the legal and institutional frameworks, e.g., through revisions to legislation in the areas of the organic budget, procurement, and debt laws, as well as PFM-related regulations. The organizational structure and functions of the ministry of finance have also been the subject of conditionality, with regard to (i) the revision of roles and responsibilities of units and (ii) the creation of units to control and monitor capital investments, public-private partnerships, sub-national debt, and macroeconomic policy developments. In some cases, conditionality has been applied to establishing and strengthening internal and external audit functions.

  • In countries where basic PFM systems were already in place, conditionality has been used to focus on relatively more advanced reforms, such as establishing a treasury single account, implementing and revamping an integrated financial management information system, and improving the integration between cash and debt management functions.

Macro-Social Structural Conditionality in Fund-Supported Programs

The focus on macro-social issues in Fund-supported programs has increased recently. An analysis of all macro-social structural conditions over 2002-2011 showed that, in line with the call for parsimony and focus on macro-critical conditions in programs, structural conditionality directly related to macro-social issues represented a small share of all structural conditions. Macro-social conditionality had been declining since the early 2000s, but started to increase again at the end of the review period. GRA programs have typically included more macro-social conditions, with a greater relative focus on reforms to pensions, social sectors, and labor markets. By contrast and also reflecting labor market characteristics in these countries, civil service reforms are relatively more common in PRGT programs, but conditionality related to pensions or labor markets outside public employment is very limited (Box Figure).

uA01fig02

Macro-Social Conditions, by Type of Fund-Supported Program

Share of total structural conditions

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

What are the short and long-term outcomes? Several Fund-supported programs include measures aimed at lowering fiscal expenditure in the short term and/or improving productivity through structural reforms. While these measures contribute to resolving balance-of-payments difficulties and are likely to result in longer term gains, they may also have short-term costs in terms of employment or income. In some countries, programs included deeps cuts in public sector employment, public sector wages (e.g., Latvia and Greece), and pensions (Greece), reflecting the depth of the fiscal trouble and the need for rapid and strong adjustment. In the case of Latvia, for example, safety measures were introduced, such as extension of employment benefits, public works programs, and higher housing benefits. A close look at these measures suggests that:

  • On labor markets: many measures involved social partners in the decision-making process, helping build ownership for reforms and thereby supporting successful program implementation. While many of these conditions could have short-term costs, they were also likely to improve the functioning of labor markets and make them more inclusive over the longer term.

  • On the civil service and public employment: some conditions could result in lower public employment or lower wages in the short-term. However, measures that aim at enhancing productivity need not have short-term negative effects, and have positive longer-term effects in terms of public sector efficiency. These measures are more common in PRGT programs.

  • On pensions and other social sector reforms: almost half of conditions aimed at improving the social safety net either through higher social spending or through better targeting of benefits. In some instances where programs supported reforms of energy subsidies, conditionality helped mitigate the impact of higher energy prices on the poor.

Wage Bill Ceilings in PRGT Programs23

In 2007, the Executive Board requested that wage bill ceilings in LICs be used only in exceptional cases and with sufficient flexibility in programs. This guidance followed a discussion on the implications of aid inflows for Fund advice and program design in LICs.24 Wage ceilings had come under criticism from both civil society organizations and the IEO, because they could prevent LICs from using donor resources for expanding employment in poverty-reducing sectors, such as health and education. A review by Fund staff found that while wage bill ceilings were used with some flexibility and were tailored to country circumstances, they also created a number of problems: they tended to persist in programs, were difficult to monitor, and distracted governments from efforts to strengthen institutions for better controlling wage spending.25

An analysis of conditionality at program approval showed that the policy on wage bill ceilings for LICs was well followed. There were no QPCs set on wage bills starting in 2007, although there were a few instances of monitoring through ITs. For example, this reflected high wage bills in Djibouti and Moldova compared to similar countries in the region, and the need to reorient spending towards pro-poor spending in Cote d’Ivoire. Structural conditionality related to wage bill ceilings also dropped significantly—the last PRGT program with structural conditionality on wage bills was Moldova with an SPC in 2006 and a PA in 2009 (Box Table).

Use of Wage Ceilings in IMF LIC Program Conditionality

article image

C. Focus of Quantitative Conditionality

22. Quantitative conditionality focused on external and fiscal stability, reflecting the key objectives of Fund-supported programs (Figure 6). Following a focus on debt in the earlier part of the review period, monitoring in GRA programs was shifted toward fiscal consolidation during the global financial crisis in line with the growing fiscal consolidation needs in a number of programs. In PRGT programs, the mix of QPCs remained broadly unchanged with emphasis on external debt, also reflecting the Fund’s involvement in debt-reduction exercises for LICs (BP4). Fiscal and monetary stability were monitored in almost all programs, and all PRGT programs included targets on external debt. The use of reserves targets was, however, limited in countries with no autonomous exchange rate policy (currency unions and currency boards constitute the totality of programs with no reserves target).

Figure 6.
Figure 6.

Quantitative Conditionality in Fund-Supported Programs, 2002-20111

Number of conditions per review

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 External debt includes ceiling on external arrears (including no new arrears), medium and long term external debt, and short term debt. Monetary and fiscal stability includes credit to government, domestic arrears, fiscal deficit, and total domestic credit.

23. Indicative targets have mostly been used to monitor fiscal performance and “social sector” indicators (Figure 7). Assessments of ITs are used to complement QPCs (or QACs) in monitoring quantitative performance. They are mostly used to provide guidance on the expected values of QPCs later in the program (these targets are normally turned into QPCs, with appropriate modifications; these are not reviewed here as they do not provide additional information compared with QPCs). They can also be set on variables that provide important additional information about program performance (such as the primary fiscal deficit, base money, and the change in net credit of the banking system to the public sector), or on variables that are seen as critical but may be difficult to assess with sufficient precision to qualify as QPCs.26 Over the period under review, the largest share of ITs was related to the fiscal deficit in PRGT programs. In GRA programs, there was a shift toward fiscal ITs in the latter part of the period, reflecting concerns about fiscal sustainability. The use of wage ceilings declined both in GRA and PRGT programs (see also Box 4). “Social” ITs—typically floors for priority (such as social or poverty-related) spending—were used to monitor performance in fiscal expenditure in these areas. Following the revamping of facilities for LICs in 2009, which mandated monitoring of poverty-related spending, the use of poverty-related ITs increased in PRGT programs.

Figure 7.
Figure 7.

Quantitative and Social-Related Indicative Targets, 2006-2011

(In percent of total)

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

D. Volume of Conditionality

24. The number of conditions set over the course of a program is often used as measure of parsimony. Parsimony implies containing the number of conditions in a program to the minimum necessary to achieve program objectives, as noted above. This paper counts conditionality on the basis of the year when conditions were reviewed (review-year approach; Box 6).27

Counting Conditions

Conditions can be accounted for either by year of program approval or by year of review.

  • Number of conditions per year of program approval (approval-year approach). Previous assessments and reviews of conditionality as well as the early reviews of crisis programs, relied on counting the total number of conditions per arrangement, identified by the year of program approval.28 The number of conditions is then normalized by year or review. While this approach allows for a comprehensive examination of conditionality per program, the full extent of conditionality can by definition only be known once programs have expired or have been permanently interrupted. It is therefore less suited to evaluate conditionality in recent, ongoing programs. In addition, because the reference is the year of program approval, the effects of policy changes during the lifetime of the program cannot be tracked easily.

  • Number of conditions per year of review (review-year approach). This approach was developed in the 2009 Annual Report on Structural Conditionality, and underpins the 2011 RoC.29 Conditions are accounted for based on the year when they are reviewed (or, for outstanding conditions, based on the year they are/were expected to be reviewed), thereby tracking program developments. Conditions are normalized by the number of reviews per year (providing information on conditionality per review) or by the number of current programs per year, with current programs defined by the number of quarters during which they are under implementation. On the basis of this approach, the numbers of conditions set in ongoing and expired programs are comparable; in addition, the impact of policy changes on conditionality can be identified more effectively.

25. The number of QPCs has been stable, at around five to six per review (Figure 8). Macroeconomic frameworks drive the number of monitored variables and the variance in the number of QPCs by program is limited (the inter-quartile range is between four and six QPCs in GRA programs and between five and seven in PRGT programs). The number of ITs, which are used to facilitate the monitoring of program implementation, was broadly steady overall during 2006-11 (Figure 9).

Figure 8.
Figure 8.

Number of Structural and Quantitative Conditions per Review, 2002-2011

All programs, average and median

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.
Figure 9.
Figure 9.

Number of Indicative Targets per Review, 2006-2011

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 No data for 2006. There was only one GRA program in 2005, with 5 ITs.

26. The discontinuation of SPCs in March 2009 had a sizeable impact on the number of structural conditions per review, notably in GRA programs.30 This impact was further reinforced by the decline in SBs and PAs (both at program approval and during reviews) over most of the period (Figure 10).31 Moreover, the distribution across programs of the number of structural conditions per review narrowed during the latter part of the period, pointing to an effort across the board to further parsimony (Figure 11). The number of structural conditions however stabilized after 2009. Conditions per review would be even lower if arrangements under the FCL and PCL were included, given there is no ex-post conditionality under the FCL and limited ex-post conditionality under the PCL (BP4).32

Figure 10.

Structural and Quantitative Conditionality in Fund-Supported Programs, 2002-2011

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.
Figure 11.
Figure 11.

Distribution of Structural Conditions per Review, 2002-20111

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and Staff calculations.Source: MONA and Staff calculations.1 The white line refers to the median value; grey boxes contain values within the 25th and 75th percentiles; whiskers indicate the dispersion between the lower and upper adjacent values; dots are the outside values. Adjacent values defined as follows: the upper adjacent value (UAV) is the largest data value that is less than or equal to the third quartile (Q3) plus 1.5 times the interquartile range (IQR): UAV = Q3 + [(Q3 - Q1) × 1.5]. The lower adjacent value (LAV) is the smallest data value that is greater than or equal to the first quartile (Q1) minus 1.5 times the IQR: LAV = Q1 - [(Q3 - Q1) × 1.5].

27. The drive towards parsimony may have reached a plateau at the very end of the review period. The number of conditions per review remained comparatively low in both GRA and PRGT programs, but in GRA programs a steady increase in the number of SBs per review was noticeable (Figure 10). The number of reviews with a relatively larger number of conditions tended to increase between 2009 and 2011, likely reflecting the need to tackle deeper-rooted issues, such as competitiveness or large debt burdens (Figure 11). This shift was not limited to a small number of countries, but was rather widespread across all types of GRA programs. As a consequence, the total number of conditions per year picked up modestly starting in 2010, reflecting both an increase in the number of conditions per review (GRA programs) and a greater frequency of reviews per year (GRA and PRGT) (Figure 12).33 These developments may be appropriate giving the changing circumstances in recent programs—especially the greater need for growth-enhancing structural reforms. However, they also point to the importance of continued monitoring of structural conditionality, to ensure that these are the minimum necessary to meet program goals.

uA01fig03

Reviews with more than 4 and 5 conditions

(In percent of total reviews)

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.
Figure 12.
Figure 12.

Total Number of Conditions per Year of Ongoing Programs, 2002-20111

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 Prior actions refer to review prior actions. PRGT includes PSI.

28. More frequent adjustments in structural conditionality point to an ongoing fine-tuning of programs, and maybe a need to better leverage surveillance. Fine-tuning of programs is expected as reviews proceed, especially as new unforeseen circumstances can require policy responses, including adjustments and additions to conditionality and, as the imperative of addressing quickly, a crisis situation may crowd out important structural measures in initial program design. In recent programs with advanced countries (namely, Iceland, Greece and Ireland), adjustments and additions to conditionality were relatively higher during the third and subsequent reviews, compared with other cases (Figure 13). While the changes had been agreed upfront with the authorities in some cases (e.g., Iceland), this could also indicate that conditionality may have become relatively more complex as these programs progressed. In some cases, lags in fully developing the reform agenda could point to some disconnect between Fund surveillance and program design. This suggests surveillance could be better leveraged in program design by relating more systematically program conditionality to the vulnerabilities identified during surveillance exercises, including through enhanced prioritization of recommendations during surveillance (Box 7 and Appendix 3).

Figure 13.
Figure 13.

Number of New Conditions Set , 2008-2011

Average by review number

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

Leveraging Surveillance for Conditionality

This box examines links between structural conditionality in programs and recommendations in previous Fund surveillance. The analysis is based on a review of programs approved since 2008 that had a recent Article IV consultation (less than two years old) that was not combined with a program review or approval. It includes 34 countries, of which 23 (or 68 percent) had GRA programs and 11 (or 32 percent) had PRGT programs (Appendix 3). The analysis maps the conditionality set between program approval and the second review to staff’s recommendations in the latest available Article IV report prior to program approval.

About 48 percent of program conditions were foreshadowed in previous Article IV consultations, with some variation across both countries and time. SBs and SPCs were mapped more closely to Article IV recommendations than PAs. While conditionality in PRGT programs tended to be generally well related to previous surveillance recommendations, conditionality in GRA programs exhibited much larger variation. The need to tackle an unexpected crisis with sometimes unforeseen consequences could explain this wider dispersion. By the second program review, however, the dispersion across program types becomes more similar.

Changes in the global macroeconomic environment also affected the design of conditionality in relation to previous surveillance. The share of conditionality foreseen in previous Article IV reports was the lowest in 2008 for GRA programs, reflecting the unexpected impact of the crisis on countries such as Iceland or Latvia, as well as the need to design programs in an emergency context. In PRGT programs, where the impact of the crisis was delayed, structural conditionality was the least related to past surveillance in 2009. By the end of the period, however, the share of structural conditionality that could be mapped to previous Article IV recommendations tended to increase.

E. Depth of Structural Conditionality

29. The impact of structural conditionality depends not only on numbers of conditions but also on the depth of those conditions, where depth is defined as the degree and durability of structural change that would likely occur if the condition was implemented. As detailed in BP4, the notion of depth refers to the durability of structural change. It does not imply any assessment of (1) the difficulty of implementing conditionality on the part of authorities or (2) the criticality of structural measures agreed upon in a program. The analysis of programs during 2006-11 pointed toward an increase in depth of structural conditionality. There was significant variance among countries and regions, and on average, programs with advanced market economies were structurally deeper than programs with emerging markets. GRA programs with European countries demonstrated the highest degree of structural depth. The structural depth of recent programs with euro area countries (especially Greece) surpassed the average structural depth for other programs. While programs with LICs accounted for the largest number of conditions, structural depth in these programs was lower than that for the other two country groups.

30. Regression analysis indicates that there was a trade-off between the number and the depth of conditions. This suggests that streamlining may have led to a balancing in the burden of conditionality between the number of conditions and their depth (Box 8). In particular, the number of conditions per review correlated negatively with the average depth of conditionality per review in both GRA and PRGT programs during 2006-11.

Number of Conditions and Depth

A mere reduction in the number of conditions would not necessarily satisfy the principle of parsimony.

For example, a given condition can either be relatively easy or very demanding to implement. Measuring the structural burden of conditionality therefore requires taking into account both the number and the depth of conditionality.

The trade-off between the number of conditions and the depth of conditionality is estimated using a fixed-effects panel regression. The total number of conditions per review, for each program, was regressed against the average depth of conditionality. In order to avoid biases related to outliers, the first and last percentiles (in terms of depth of conditionality) were omitted from the regression. The estimated equation was:

Nir=α Depthir+c+ui+wr+εir
A01eq1

Where i is the arrangement, r the review, N the number of total conditions per review (SPC, SB, or PA) and Depth is the weighted average depth of conditionality per review (where the weights for low, medium and high structural depth are set respectively at 1, 2 and 3). u and wr are fixed effects for arrangements and reviews respectively, and sir is the residual.

Results point to a statistically significant tradeoff between the depth of conditionality and the number of conditions set by review, with a one standard deviation increase in the depth of conditionality correlated with a reduction in the number of conditions by 1.2 and .7 in GRA and PRGT programs respectively.

article image

*** and ** refer to a 1 percent and 5 percent significance level respectively. The period covered is 2006-2011 and includes all conditions with test dates up to September 30, 2011.

III. Other Key Principles of Conditionality

A. Ownership

31. Ownership is crucial to successful implementation of Fund-supported programs. According to the Guidelines, the primary responsibility for the selection, design, and implementation of economic and financial policies rests with member countries. Through a collaborative process, the Fund should aim to reach understanding with country authorities on a mutually acceptable path for achieving program goals, taking into account economic priorities, the political economy framework, the specific context for a program request, including the causes of the balance of payments need, and the member country’s capacity to implement reforms within the specified time horizon. The Fund also should encourage members to seek to broaden and deepen support for sound policies to enhance likelihood of implementation.

32. Streamlined and tailored programs, through parsimonious conditionality, support ownership. EPAs and EPEs noted that lack of ownership was sometimes exacerbated by the complexity of measures and capacity constraints in public administration. They also suggest that consideration of both political and capacity constraints was essential to program effectiveness (BP4 and Box 2).When slower-than-expected implementation of structural reforms reflects capacity constraints and/or limited ownership in implementing some of the “stronger” structural measures, a more prioritized approach in defining the reform may be required.

33. The Fund’s outreach to both the authorities and a broader group of stakeholders can help build ownership. Some EPEs and EPAs suggested that extensive outreach efforts, in combination with focused conditionality and flexible implementation, helped achieve a higher degree of ownership. Moreover, the review of country cases indicates that outreach served to facilitate some program negotiations (see also Box 9). While active interactions with civil society organizations were frequent, they were not the rule, reflecting both the fact that the main counterparts in program negotiations are the country authorities and possibly political sensitivity on issues related to outreach in some countries (Figure 14, and BP4). The recording of outreach activities in support of ownership in program documents was uneven, possibly reflecting staffs efforts to produce concise staff reports. In particular, three out of the nine GRA case studies did not report outreach activities, although there was evidence that outreach was undertaken.

Ownership, Program Implementation, and the Use of PAs

PAs can be used when program implementation concerns lead the Fund to look for frontloaded implementation of measures that are particularly critical for program success. In that context, PAs are both a screening device used to ensure that the minimum necessary level of ownership is present for program success and a tool to support implementation of conditionality.

Simple regression analysis suggests that the use of PAs at program approval was related to future implementation issues in GRA programs, pointing to the role of PAs as a screening device. A higher number of PAs at program approval was associated with a lower implementation ratio of structural conditionality over the length of the program (see Table below). This finding was only statistically significant for GRA programs with more than 2 PAs. Indeed, conditionality was much more focused when the number of PAs was limited, and largely related to fiscal measures such as adoption of budgets in line with the program agreement (Box Figure). In PRGT programs, the relation failed to be significant, suggesting that PAs were used regardless of potential implementation issues and possibly to advance important structural reforms in countries with large reform needs.

uA01fig04

Distribution of the Number of Prior Actions at Program Approval, 2002-2011

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

Number of Board Prior Actions and Subsequent Program Implementation, 2002-2011

article image
Source: MONA and staff calculations.

Includes a dummm for Romania (2004 program) p-values in brackets.

p<0.1,

p<0.05,

p<0.01

The use of PAs during program implementation (or reviews) underlined their role as implementation tools. In particular, probit analysis indicated that the probability for a given review to include a PA was related to the implementation record of previous reviews.34 In GRA programs, a better track record of current and past implementation lowered the likelihood that a PA was set for a given review. An improvement in program implementation by one percent in the current review reduced the likelihood of having a PA at that review by 0.6 percent. Concurrently, an improvement in implementation in the previous review by one percent reduced the likelihood of a PA by 0.5 percent (see Table to right).

PAs in PRGT programs were determined by the track record of the current review, while implementation in the previous review failed to have an impact. An improvement in the implementation ratio by 1 percent during the current review reduced the likelihood of having a PA at that review by 0.89 percent.

Probit Results for Prior Actions and Implementation 1

article image
Source: MONA and staff calculations.

The table reports marginal effects. P(y=1) denotes a review with at least one prior action

Figure 14.
Figure 14.

Ownership - Results of the Survey of Fund Staff and Country Authorities

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

34. Fund policies to support ownership of programs seemed to have brought some success. There are indications that ownership may have improved over 2002-2010 for GRA programs, as evidenced by less use of a large number of PAs at program approval (Box 9). Supporting this view, survey results indicated that Fund staff has confidence in country ownership. The MEFPs, which form the basis of program documents and describe the authorities’ economic policies, were designed with the active contribution of authorities. Drafting, however, remained mostly done by Fund staff, suggesting room for further improvement towards ownership (Figure 14).

35. The survey results, however, point to some difficulties in conveying Fund efforts to enhance ownership (BP4). For instance, the discontinuation of SPCs, which was designed partly to accommodate countries’ specific timelines in the implementation of structural conditionality, had a limited impact on the sense of ownership, as conveyed by Fund staff. Similarly, stigma attached to programs appeared to have not been significantly reduced.

36. In cases of limited ownership, specific instruments can enhance program implementation. Staff-monitored programs, for instance, can be used to assess the potential for a future Fund-supported program with UCT-quality conditionality, thereby gauging both implementation capacity and the degree of ownership by the authorities. PAs at program approval could also be used as screening devices for the degree of ownership. EPAs/EPEs suggested that a flexible approach to program conditionality, with regard to implementation timelines, was found to enhance ownership, while the discontinuation of SPC did not have a noticeable impact (Appendix 4).

37. However, conditionality cannot substitute for ownership. For example, a higher number of PAs did not systematically result in more successful program implementation (Box 9). In some programs—for example, with Ukraine—program success was reduced by weak ownership despite many PAs (BP4). Lessons from recent GRA programs also showcase the importance of ownership and the difficulty of gauging ownership through simple measures of implementation (Box 10).

Ownership in Recent Programs: Lessons from Iceland, Latvia, and Belarus

Recent programs have underscored the critical importance of country ownership in helping to deliver successful outcomes. Countries such as Iceland and Latvia where programs goals enjoyed strong local support have succeeded in overcoming the crisis despite changing circumstances, whereas Belarus was left with serious residual problems in the wake of a program with weak program ownership at the highest levels.

The SBAs for Iceland and Latvia came in the wake of external and financial-sector crises. Iceland’s program was brought on by the failure of three large commercial banks in October 2008, whose combined assets represented almost 10 times its GDP. The trigger in Latvia’s case was the collapse of its largest deposit-taking institution. These failures induced deep recessions, strains in the balance sheets of the private and public sector, and (in Iceland) extreme disruptions in the foreign exchange market. Iceland imposed capital controls quickly after the collapse of the banks, and formal foreign exchange rules were adopted by the authorities at end-November 2008, while Latvia allowed its quasi-currency board to operate. Against this backdrop, the Iceland program aimed at: (i) preventing a further sharp depreciation of the krona; (ii) ensuring medium-term fiscal sustainability; and (iii) developing a comprehensive financial sector restructuring strategy, with conditionality targeted in these areas. In Latvia’s case the focus was to correct the liquidity crisis and safeguard long-term external stability while maintaining the exchange rate peg, to which the national authorities and European institutions were firmly committed. This required measures to stabilize the financial sector, fiscal tightening to reduce financing needs and spur real depreciation, and structural reforms to bolster competitiveness.

Belarus’s program request, by contrast, was brought on by a current account shock induced by a severe deterioration in the terms of trade. The program was based on strong macroeconomic policy adjustment, including exchange rate realignment, wage restraint, and demand management to reign in external imbalances, while a more flexible exchange rate regime was developed to act as a shock absorber.

The Icelandic authorities took considerable ownership of the policy priorities early on, and conditionality was in line with their adjustment and reform plans. Measures to meet Iceland’s fiscal consolidation targets, reflected in fiscal performance criteria, were subject to extensive consultation with the government’s social partners, with the aim of preserving the key elements of Iceland’s Nordic welfare state model. The program supported the authorities’ implementation of capital controls in late 2008 to help contain pressures on the krona, the authorities’ initial plan for liberalizing these restrictions (published in August 2009), and through structural conditionality for the fourth and fifth reviews, the development of the authorities’ capital account liberalization strategy approved in March 2011. The bulk of structural conditionality under the program addressed various aspects of financial sector reform. In the early stages of the program this included the need for recapitalizations, a review of supervisory strength, and strategic plans for new banks and asset recovery, but in later reviews included conditionality on the domestic debt restructuring process.

In Latvia, structural and especially fiscal policy became the key adjustment tools, as the authorities and European institutions targeting a reduction of pre-crisis imbalance while being firmly committed to the fixed exchange rate. The rationale for maintaining the peg was to avoid severe balance sheet effects and support the authorities’ plans for euro adoption, when other options (such as immediate euro adoption) were off the table. This strategy inherently entailed significant sacrifices by all stakeholders. The recession proved much deeper than originally expected and fiscal targets were missed in the first half of 2009. The program was redesigned to accommodate these fiscal slippages, and to create additional space for social safety net measures. Nevertheless, additional fiscal measures (the bulk in 2009-10) were required. They were primarily structured as expenditure cuts reflecting the government’s aversion to substantial further tax increases, except in 2011.

Although Belarus met most of the performance criteria under the program, fiscal adjustment was stymied by quasi-fiscal activity. Belarus’ balanced-budget program targets were met despite major revenue losses, but these efforts were undermined by quasi-fiscal activity in the form of bank lending under government programs (LGP). LGP has traditionally played an important role in Belarus in directing funds to particular projects and sectors, and enjoys leadership at the highest levels. The authorities opted to increase LGP by almost 50 percent, compounding demand pressures and exacerbating external imbalances. The authorities adhered to the letter of the program by circumventing the original conditionality on LGP. The conditionality was adapted as the extent of LGP growth became apparent, but the lack of comprehensive data on LGP also hampered the program’s effectiveness.

By 2011 Iceland and Latvia were growing again and on their way to overcoming the challenges that necessitated the original programs. In Iceland, the value of the krona has stabilized, and the authorities have been able to begin capital account liberalization under their strategy. Significantly progress was also made to place public finances on a sustainable path, although unemployment remains high. A new, significantly smaller, banking system has emerged, with work to address remaining legacy vulnerabilities progressing. Latvia has also largely stabilized its banking sector, and was on course to bring its budget deficit down to 4.5 percent of GDP, putting the country within sight of the 3 percent of GDP Maastricht criterion and the program exit strategy of euro adoption. Strong program ownership has been key to Latvia’s strategy, which has reduced pre-crisis imbalances, while sustaining the peg. While this has stabilized the economy, it was accompanied by significant reductions in output and employment, with unemployment still exceeding 14 percent.

Although Belarus avoided a full-blown balance of payment crisis, its program was less successful at resolving the original vulnerabilities. The current account deficit widened in 2009 and 2010 (one of the few crisis programs where this was the case), and the overvaluation of the real exchange rate persisted. A lack of ownership had contributed to the shortcomings of the program, and limited progress at addressing vulnerabilities (see also BP4).

B. Tailoring

38. In tailoring conditionality, programs need to strike a balance between consideration of country circumstances and evenhandedness among countries.35

Program design needs to accommodate a broad spectrum of initial conditions. For instance, countries with wide-ranging structural reform agendas may require deeper structural changes than countries suffering from short-term liquidity shocks. Large-scale disbursements may justify a different type of conditionality than more limited disbursements. The Fund takes a liberal view to requests for transactions within the first credit tranche, provided that the member itself is making reasonable efforts to solve its problem. Conditionality should also be adapted to countries’ capacity and track record.

Capacity

39. Conditionality needs to reflect implementation capacity. In countries with low implementation capacity, program design entails tradeoffs between an ambitious structural reform agenda that ultimately would help the country to develop capacity, and a realistic set of conditions, possibly divided into small, achievable modules. Within the capacity-determined framework, conditionality could thus be somewhat less complex for low-capacity countries, and more ambitious for countries where stronger institutions allow for a higher degree of implementation.

40. Case study analysis suggests that capacity considerations did affect the design and sequencing of structural conditionality. The 2006 Sierra Leone program presented an example of a carefully paced structural agenda reflecting the country’s post-conflict nature. Similarly the structural agenda in the 2008 Togo program took into account the country’s capacity constraints after its long political crisis and donor disengagement, with no measures for the first six months of the program other than two PAs.

41. The burden of conditionality was tailored to the challenges facing low-capacity countries.36 The correlation between country capacity at the time of program approval and the total number of conditions per program was positive and statistically significant (Figure 15).37 The depth of conditionality was also adjusted to the capacity of countries: conditions of high structural depth tended to be set more frequently in countries with a higher Country Policy and Institutional Assessment (CPIA) rating.

Figure 15.
Figure 15.

Number of Structural Conditions and CPIA Rating, 2005-20091

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Sources: MONA; World Bank; and staff calculations.1 Includes SBs, SPCs, and PAs. The positive relation between the number of conditions and the County Policy and Institutional Assessment (CPIA) rating is significant at the 1 percent level for programs with CPIA rating equal or lower than 4.

Access

42. Reflecting the need to safeguard Fund resources, conditionality was tailored to the access level under an arrangement. With differences across types of programs, access was related to the number of PAs for Board approval in PRGT programs and to the number of deep conditions in GRA programs (Figure 16). These results may be linked to capacity issues in PRGT countries, leading to a focus on measures that can be achieved prior to program approval. In GRA programs where capacity was usually stronger, higher access was correlated with deeper structural change, as higher BOP needs translated into both a higher access and a greater need for structural reforms to bring down large macro-imbalances.

Figure 16.
Figure 16.

Structural Conditionality and Access in GRA and PRGT Programs, 2002-20111

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 For PRGT programs, the sample is limited to programs with at least one prior action at program approval, and non-zero access levels.

43. The level of access impacted on the focus of conditionality (as defined in Section II). GRA and PRGT programs in the upper quartile of the access distribution for each group had structural conditionality that was more focused than average on financial sector reforms, reflecting more mature economic systems with greater financial difficulties. On the quantitative side, the monitoring of arrangements involving higher access was focused on fiscal issues. This was true for high exceptional access GRA-supported arrangements (access over 1000 percent of quota), as well as for PRGT-supported arrangements in the upper half of the access distribution, reflecting the need to ensure fiscal sustainability in a context of rising external debt (Figure 17).

Figure 17.
Figure 17.

Focus of Conditionality and Access, 2008-2011

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 Includes review PAs, SBs, and SPCs. High exceptional access denotes access over 1,000 percent of Quota.

44. Tailoring also applied to the specification of debt limits in programs with LICs. These programs typically included external debt limits, seeking to prevent the build-up of unsustainable debt, while allowing for adequate external financing. With the new guidelines on debt limits in place since December 2009, the Fund moved away from a single design for concessionality requirements toward a menu of options, taking better account of differences among LICs regarding their debt vulnerability and their macroeconomic and public financial management capacity (BP4).

45. According to survey results, conditionality was generally viewed as both tailored and flexible. While Fund staff were more convinced of the appropriateness of program design and flexibility, both country authorities and donors broadly shared the positive assessment (Figure 18).

Figure 18.
Figure 18.

Tailoring and Flexibility of Fund Programs - Survey Results

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: Survey of IMF staff and stakeholders.

46. However, some programs were affected by weaknesses in the design of conditionality. As noted in the EPAs and EPEs, inappropriate diagnostics of underlying problems reduced the effectiveness of some programs. In a few cases, critical conditionality or conditionality following-up on initial reforms did not form part of the program.

C. Coordination

47. Fund collaboration with partner institutions is essential for program success. The Guidelines underscore the need for coordination with other multilateral institutions— especially the World Bank—in designing and monitoring conditionality. While coordination has also been important with other international organizations—for example, bilateral and multilateral donors—the need for coordination with new partners, such as regional institutions, has grown in recent years, including to reduce the risks of duplication and overstretched capacity. EPAs and EPEs pointed to the essential role of coordination with the World Bank and other institutions in macro-critical areas of structural reform beyond the areas of core responsibility of the Fund, such as decentralization and energy sector reform.38

48. Prominent examples of Fund coordination with other institutions include (BP4):

  • Sub-Saharan Africa: The survey of Fund stakeholders and views from Fund country teams suggest that the coordination process with multilateral and regional institutions—notably with the World Bank—was effective (BP4). Coordination focused in areas of joint or overlapping responsibilities, supporting the streamlining of Fund conditionality and avoiding the duplication of conditionality across different organizations. Responsibilities were shared according to the “lead agency” principle, with the Fund focusing on macro-critical conditionality. Coordination was supported by the Joint-Management Action Plan and World Bank-Fund Collaboration (JMAP), but informal and formal coordination tended to be equally important. However, as noted in some of the EPAs and EPEs, on some occasions, collaboration with the World Bank met with challenges.

  • Recent programs with European member countries: Coordination with the EU supported the design and monitoring of conditionality. The complementarities of Fund and EU expertise proved useful in various program areas. In fiscal policy, the Fund’s expertise was crucial in operationalizing the fiscal adjustment path and tackling fiscal risks, while European Commission (EC) expertise focused on specific fiscal areas, such as health care and pension reforms. In euro area countries, however, unlike under conventional programs, additional stakeholders added a layer of complexity in conditionality design, often resulting in extended periods of discussion and decision making. Discussions about critical issues (such as private sector involvement) also weighed on coordination as well as on program design. In addition, the large and growing number of structural reforms identified by the EU is increasingly at odds with the principle of parsimony (Box 11).

    Design of Conditionality in Co-financed Euro Area Programs

    The three euro area (EA) member countries—for Greece, Ireland, and Portugal—are supported by the Fund, the European Commission (EC), and the European Central Bank (ECB), the so called “Troika.”

    Monitoring is ensured through a set of unified macroeconomic parameters and structural measures. Conditionality for Fund Board reviews is based on a standard quarterly framework of QPCs and SBs. EC conditionality is based on an overall assessment of progress against the EC structural agenda and macroeconomic targets. Joint Troika review missions and assessments support coincidental disbursements from the Fund and the EU.

    Coordination is important for these programs. It has developed in the spirit of the Guidelines and was generally guided by the core expertise of each institution (BP4).39 The Fund mostly focuses on short-term macro-critical policies, while the EC covers comprehensive medium-term structural reforms. Fund conditions mainly address the fiscal and financial sectors, but also include SBs related to competitiveness in the Portugal and Greece program. The EC’s structural agenda aims at reforming areas deemed critical for medium-term recovery and is geared toward achieving compliance with EU directives; reforms bear on public administration, health, labor market, the judicial system, and sectoral competition. Within this division of labor, some overlaps between institutions exist (for instance, the EC/ECB memoranda of understanding include fiscal and financial chapters). On the quantitative side, the EC targets are designed to be consistent with the Fund’s, even though the definition of the targets may differ (e.g., fiscal targets are defined on a cash basis for the Fund, but in percent of GDP for the EC). On the structural side, EC measures are usually much more precise and detailed than the Fund’s. While this allows for a more effective division of labor, it may create ex-post inconsistency issues (e.g., Fund conditions for Greece provide for a —comprehensive pension reform that reduces the projected increase in public spending on pensions over the period 2010-60 to 2% percent of GDP,” while the EC has a set of precise measures defining the new pension system). While the assessment of the Fund’s parsimonious conditions is straightforward, the large set of EC measures calls for a broad-based assessment of implementation prior to authorizing a disbursement.

    Over time, both the Fund and the EC have increasingly ventured into areas of structural reforms initially devoted to the other institution. EC-supported measures have been more and more focused on fiscal issues, while the Fund introduced “competitiveness” conditionality in the Portugal and Greece programs. While the exact number of EC measures is difficult to establish (as they are broken down into sub-measures), it has been far less parsimonious than the Fund standard, and became even less so over time (Box Figure). Moreover, whereas EC measures set for the Greece program were initially clearly identified as conditionality, the identification became less clear-cut in following reviews or other EA programs, as some measures included both very detailed sub-actions and broad overall objectives.

    uA01fig05

    EC structural conditionality in Euro Area programs

    Number of measures at Board approval

    Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

49. Country authorities, Fund staff, and donor agencies viewed coordination as critical to program implementation. More than 80 percent of respondents noted intensive coordination with donors and effective coordination with the World Bank (Figure 19).

Figure 19.
Figure 19.

Fund Coordination with Aid Partners - Survey Results

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: Surveyof IMF staff and stakeholders.

D. Clarity

50. Clarity in Fund conditionality enables stakeholders to better comprehend the key elements of programs. In particular, the operational guidance notes on conditionality call for clear justification of program conditionality, as well as of the link between program goals, program instruments, and the corresponding structural conditionality. Program documents should discuss possible concerns related to implementation constraints and provide strong justification and more detailed explanation of the critical importance of conditions outside Fund core areas of expertise. Discussions with country authorities on policy options should be conveyed clearly where possible. Moreover, program conditions should be transparently distinguished from other elements of the authorities’ program in program documents and staff reports.40 Except when otherwise indicated, the following assessment of clarity is based on the review of the country case studies (see BP4 for details).

51. Country case studies indicated that the implementation of the Guidelines with respect to clarity in conditionality has been uneven. In particular, while most staff reports reviewed provided a clear justification of the criticality of conditionality and presented implementation risks, the specific guidance on the presentation of critical structural measures and of a discussion of policy options was largely ignored (Table 2).

Table 2.

Presentation of Conditionality

In percent of case study Staff Reports meeting guidance

article image
Source: Staff calculations, based on a sample of 18 countries.

52. Staff reports for many of the case study countries presented the proposed conditionality in a comprehensive way. The 2009 Ghana program contained a comprehensive table with measures and their macroeconomic rationale. The 2008 Togo one presented a complete set of past, immediate, and envisaged structural reforms, while the 2008 Pakistan program contained a box explaining how structural measures link to the achievement of program objectives. However, some reports could have better set out the rationale and criticality of the proposed conditionality.

53. Reports did well in justifying the macro relevance of structural conditionality, but could have done better in linking conditionality clearly to program goals. The July 2008 revision to the operational guidance note called for a clearer explanation of the choice of conditionality, in particular its macro-criticality. Much progress was made with explaining the macro criticality. Of the 19 case study programs approved in or after 2008, almost all did so, usually referring to “macro rationale”—in most cases in a structural conditionality table. However, no staff report followed the guidance note’s approach of introducing structural conditionality with an explanation of program goals, strategies/instruments, and structural conditionality. Only the documents for Armenia (2008 and 2009) grouped the structural conditionality by strategy. Hence, the intention of the 2008 guidance note of clearly presenting the mapping of structural conditionality to objectives and strategies was only partially realized. A few programs stood out by succinctly articulating program objectives and strategies in their executive summaries—2008 Hungary, 2008 Seychelles, and 2008 Togo. In combination with a table linking those to structural conditionality that approach could qualify as best practice.41

54. Programs include non-core structural conditions that could be better justified. The analysis of case study countries suggests that a number of structural measures fell outside the Fund’s core areas of expertise (Table 3).42 Overall roughly 40 out of 200 measures in 26 programs were non-core. Most of these had to do with state-owned enterprises, social programs, or civil service reform. One third of GRA programs had such measures, while about two thirds of PRGT programs had non-core conditionality. In one third of cases the justification for the measure was quite detailed; in the rest it was succinct and not different from the other measures. In only three cases was a partner institution mentioned that would provide technical assistance with the implementation. In sum, programs could do better in identifying measures that are not core, but nonetheless macro-critical, and therefore included in conditionality.

Table 3.

Non-Core Conditionality in Case Studies

article image
Source: staff calculations, based on a sample of 18 countries.

55. Most reviewed staff reports presented the implementation risks well. Nearly 90 percent of staff reports provided a discussion of the implementation risks in a broad sense, with many also providing details about the risks for individual measures. For example, Hungary’s request for an SBA provided considerable detail about the potential risks, in terms of the domestic political situation and the external environment. Capacity constraints can present implementation risks, mitigated by a slow start up and careful pacing of the structural agenda (2008 Togo and 2006 Sierra Leone).

56. The recent European programs have usually been successful at distinguishing between Fund conditionality and other elements of the authorities’ program. In the case studies, staff reports for the Hungary and Greece programs clearly outlined the framework for cooperation between the EU and the Fund (e.g., Box 1 in the Hungary staff report), and separated Fund conditions from those set by other organizations, allowing for a clear identification of program-related conditionality. By contrast, in Iceland it was not initially clear to the public that the requirement to compensate for losses of foreign depositors was not set by the Fund.

57. The reviewed staff reports contained little information on discussions of policy options. Only a quarter of the case studies contained any record of a discussion between the staff and the authorities on the policy options considered. Survey results suggested broad satisfaction from country authorities and staff that policy options and implementation plans were being discussed, suggesting that the issue is that discussions of policy options were not being recorded in the program documents, as per the guidance (Figure 20).

Figure 20.
Figure 20.

Policy Options - Survey Results

During program discussions, Fund staff provided a sufficient range of policy options and implementation plans.

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

IV. Implementation of Conditionality

58. A key goal of the initiatives on conditionality over the past decade was to improve overall program implementation. Notably, the emphasis put on ownership, tailoring, and parsimony in the conditionality guidelines and guidance notes was expected to support successful implementation of programs. The discontinuation of SPCs also increased program flexibility, with expected positive effects on implementation. This section examines program implementation by assessing two sets of information: information on temporary or long-lasting suspensions of programs and information on the implementation of SPCs, SBs, and QPCs (i.e., according to MONA classification, whether met, partially met, met with delay, waived, or not met).43

A. Program Interruptions

59. Program performance can be partly assessed by considering how often program interruptions were temporary or long lasting. Programs fall “off-track” when implementation is not satisfactory and agreement on corrective actions is not reached. In this paper, programs are considered to be off-track when the time elapsed between two reviews (or between the Board date and the first review) was larger than twice the planned time between reviews, i.e. more than six months for quarterly programs, and more than one year for semi-annual programs. An interruption is considered temporary when an off-track period was followed by a review that brought the program back on track. Interruptions are long lasting when a program went off-track for more than 12 months, and no review subsequently takes place before the program lapses.

60. Delays in completing reviews were infrequent. During 2002-end-September 2011, the average share of off-track programs hovered around 15 percent. Long-lasting program interruptions accounted for about 6 percent of the sample. By contrast, about 60 percent of the programs in the sample experienced no off-track period.

61. After declining earlier in the 2000s, program interruptions have become more frequent since the start of the global financial crisis (Figure 21). Overall implementation weakened progressively between 2009 and 2011 in PRGT programs, while deterioration was notable for GRA programs in 2011. Cases of off-track PSIs have been exceptional, with only one case recorded since the instrument was created (Uganda).

Figure 21.
Figure 21.

Program Interruptions By Year of Review, 2002-2011

On-track and off-track programs, in percent of total

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

62. While interruptions have become more frequent, the average length of interruption has declined since 2007 (Figure 22). In PRGT programs, off-track periods amounted to about 10 percent of total program periods over 2002-2007, a proportion that fell significantly after 2007. Improvement was also observed for GRA programs. The improved record will need to be confirmed once those programs have been completed.

Figure 22.
Figure 22.

Average Duration of Off-Track Periods

By year of program approval, 2002-2011 (In percent of program duration)

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.

B. Implementation of Conditionality

63. Implementation of conditionality improved over most of the review period (Figure 23).44 The implementation of QPCs outperformed that of structural measures in both GRA and PRGT programs, with more even and generally better implementation in PRGT programs. Implementation of structural conditionality was consistently better for SBs than for SPCs, and the discontinuation of SPCs does not seem to have affected the implementation record of structural conditionality, which remained broadly unchanged in 2010.45 However, implementation deteriorated in 2011. In GRA programs, this worsening reflects an increase in the frequency of programs with at least one non-implemented SB. By contrast, the proportion of PRGT programs with at least one non-implemented SB has remained stable, while the average number of SBs that were not met at the time of the review increased. In both cases, these developments reflect more difficult program implementation conditions. The implementation of structural benchmarks was generally better in core Fund areas.

Figure 23.
Figure 23.

Implementation of Program Conditionality1

Implementation ratio

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 Excludes prior actions, which are always implemented. PRGT programs include PSIs. The implementation ratio is a weighted average of assessed conditions, whereby a weight of 2 is assigned to fully met conditions, a weight of 1 to partially met conditions and conditions met with delay, and a weight of zero to not met or waived conditions.

64. Improvements in implementation were not limited to strong-performing countries, although progress in relatively poor performers was sometimes uneven. In general, countries with a good track record in the previous review tend to have better implementation (Table 4). Nevertheless, in both GRA and PRGT programs, the implementation rate of structural conditionality generally improved at a similar pace for the average of the sample, the median, and the bottom quartile, highlighting across-the-board progress in program implementation.

Table 4.

Track Record and Implementation1

article image
Source: MONA and staff calculations.

Panel probit relating the probability of a 100 percent implementation of conditionality to the implementation ratio at the previous review.

and ** refer to a 1 and 5 percent significance level, respectively.

65. Flexible timelines for program reviews have enhanced implementation. In the majority of cases, flexibility was not required, as most structural conditions were implemented as scheduled. Nevertheless, accommodating delays in the implementation of structural conditionality was helpful in ensuring that agreed reforms were implemented, as about 50 percent of measures that were reset for a future date were eventually met (Table 5).46

Table 5.

Flexibility in the Timeline and Implementation Record

Status of implementation of structural conditions, 2002-2011

article image
Source: MONA and staff calculations.

66. Implementation was supported by parsimony and not impaired by deeper structural conditionality (Table 6). A panel regression of the implementation ratio against the number of conditions (by review) showed that a larger number of conditions in a given review tended to have a negative correlation with overall implementation. However, implementation appeared to be unrelated to the depth of conditionality, pointing to the importance of other determinants—such as capacity or ownership. In particular, the combination of low capacity and a large number of conditions was linked to worse implementation (Table 6, last column).

Table 6.

Determinants of Implementation by Review, 2006-2011

Panel data estimate1

article image
Source: MONA and staff calculations.

Excludes prior actions at Board approval. t-statistics in parentheses;

p<0.1,

p<0.05,

p<0.01

67. In programs facing implementation challenges, PAs provided limited support for program execution. PAs ensure that some critical structural measures are implemented by the time of program approval or review. Econometric analysis confirmed that programs with lower implementation ratios faced a higher probability of having a PA at the current or next review. However, a higher number of PAs did not necessarily translate into improved overall implementation, suggesting that the effect of PAs on program execution was limited (Box 8).

Appendix 1. Institutional Classification of Structural Conditionality

68. The institutional classification of structural conditionality maps structural conditionality to areas of Fund core responsibility. It is based on three categories: core areas of Fund expertise; areas where Fund expertise is shared with World Bank (or other multilateral agency) expertise; and non-core areas. The classification is based on the economic classification of structural conditionality that was redesigned following the MIP following the IEO Evaluation of Structural Conditionality. More specifically:

  • Areas of core Fund expertise include (1) fiscal measures such as on revenue, revenue administration, budget preparation, debt-management, fiscal transparency, and inter-governmental relations; (2) central bank measures such as central bank operations and reforms, central bank auditing, transparency, and financial controls; (3) measures related to the financial sector in advanced economies; and (4) measures related to exchange rate systems and controls.

  • Areas of shared expertise include (1) public expenditure (including auditing, accounting, and financial controls); (2) civil service and public employment reforms and wages, as well as non-public sector labor markets; (3) pension and other social sector reforms; (4) public enterprise reform and pricing; (5) financial sector reforms in low-income and emerging markets; and (6) PRSP development and implementation.

  • Non-core structural conditionality includes (1) international trade policy (excluding custom reform); (2) economic statistics; and (3) other structural measures (e.g. natural resources and agricultural policies).

69. The current classification is similar in essence to the institutional classification developed for the 2005 review of the conditionality guidelines. The reclassification of conditionality only goes back to 2002, making comparisons in the breakdown of conditions difficult. However, trends can be compared, and point to a further focus of structural conditionality on areas of core Fund expertise.

Appendix 2. Determinants of Structural Conditionality

70. This appendix explores the determinants of the sectoral composition of structural conditionality, focusing on macroeconomic determinants. Fund-supported programs were strongly focused, and showed limited dispersion in terms of general focus: for instance, even when the share of general government-related structural conditionality was in the lowest (or first) quartile of the distribution, it still accounted for 40 percent of total conditionality in PRGT programs and almost 30 percent in GRA programs (Appendix 2 Figure). This appendix thus focuses on explaining relatively heavy reliance on specific types of structural conditionality, based on a probit model. The probability for a program to set a number of conditions in the upper quartile of the distribution by sector is estimated against a series of variables describing the macroeconomic situation in the year preceding program approval. The models are estimated separately for GRA and PRGT programs (see Appendix 2 Table).

Appendix 2 Figure.
Appendix 2 Figure.

Dispersion in the Focus of Structural Conditionality, 2002-20111

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: MONA and staff calculations.1 The graphs displays the share of conditions (in percent of total structural conditionality) by sector, for programs that belonged to the first and third quartile of the distribution in terms of the corresponding sector.
Appendix 2 Table.

Probit Analysis: Determinants of a Large Number of Conditions 1

article image
Source: MONA and staff calculations.

Large number of conditions is defined as a number of conditions within the last quartile.

and ** denote significance at the 10 and 5 percent significance level, respectively.

71. In GRA programs, fiscal conditionality was heavier with favorable initial conditions, while financial sector conditionality was heavier with larger vulnerabilities. The probability of a large number of fiscal conditions increased with lower inflation or higher growth, suggesting that the focus on structural reform in the fiscal area was stronger with more overall policy space. By contrast, conditionality in the financial sector was more likely to be important with a wider fiscal deficit, larger financial exposure through assets and liabilities, and larger external exposure through trade. The focus on public employment was also stronger when vulnerabilities (lower growth, larger openness) were more important.

72. In PRGT programs, the focus of conditionality responded to initial macroeconomic weaknesses as well as to policy space. In particular, more fiscal conditions were likely with larger initial debt and lower reserves, although the focus on fiscal conditions tended to be more important when more policy space in terms of fiscal deficit was available. Central bank conditionality was also set with regards to vulnerabilities (the probability of setting a large number of conditions increased with inflation, lower reserves) and policy space (more conditions were set when the financial exposure was lower).

Appendix 3. Leveraging Surveillance for Conditionality

73. This appendix looks at how surveillance anticipated and fed into structural conditionality of subsequent programs. Surveillance recommendations seem to have been a basis from which conditions were drawn in about half of the programs. But programs typically demand crisis measures. This means that less pressing issues get crowded out, which may be brought in during program reviews. This suggests surveillance could be better leveraged in program design by relating more systematically program conditionality to the vulnerabilities identified during surveillance exercises, including through enhanced prioritization of recommendations during surveillance.

74. Program design generally draws on the Fund’s long-standing relationship with program countries built through its regular surveillance consultations. Surveillance supports long-term and in-depth engagement of the Fund with country authorities, allowing discussions of the strength and vulnerabilities in both the macro-economic and structural policies of a member country.

75. How did recent program leverage the conclusions of past surveillance? When a country is faced with a crisis that would require structural reforms, it would be expected that program structural conditionality be informed by past surveillance. However, depending on the nature of the crisis (e.g., whether it stems from an exogenous and unexpected shock such as the global financial crisis, or whether it is the result of homegrown policy decisions), the contribution of surveillance to program design should vary. This section draws on the analysis of a sample of 34 countries to answer this question.

76. The analysis is based on the review of programs approved since 2008 which had a recent Article IV consultation (less than two years old) that was not combined with a program review or approval. Article IV consultations combined with program approvals or reviews make it difficult to disentangle the contributions of surveillance from that of program design. The resulting list includes 34 countries, of which 23 (or 68 percent) with GRA programs and 11 (or 32 percent) with PRGT programs.47 The analysis maps the conditionality set between program approval and the second review to staffs recommendations in the latest available Article IV concluded the year preceding program approval.

77. About 48 percent of conditions set at program approval were foreshadowed in previous Article IV consultations. The share of foreshadowed conditions set up to the second review declines somewhat, to 45 percent, reflecting changing conditions in program countries and the need to adjust conditionality. SBs and SPCs were mapped more closely to Article IV recommendation than PAs: 50 percent of SBs and SPCs were foreshadowed in previous surveillance reports at program approval, and about 45 by the second review, whereas the average was 37 percent for PAs. Unforeseen PAs often include passage of a specific budget or emergency measures in the banking sector.

78. There was some dispersion in links between conditionality and surveillance recommendations both across countries and over time. While conditionality in PRGT programs tended to be generally well related to previous surveillance recommendations, conditionality in GRA programs exhibited a much larger variation (Appendix 3 Figure 1). This reflects more intensive Fund engagement with PRGT-eligible countries, which often have a history of multiple programs. As a result, structural issues tend to be well identified and surveillance and program agendas are more connected, as reflected in the large number of countries that have a combined Article IV and program approval or review. It also reflects the relatively better insulation of PRGT-eligible countries from a sudden capital account crisis, and therefore the more limited need to adjust to unexpected macroeconomic shocks. By contrast, most GRA program countries in the sample had not had a Fund-supported program in the recent past, and some had never had any Fund-supported program. The need to tackle an unexpected crisis with sometimes unforeseen consequences explains the wider dispersion of structural conditions that could be mapped to previous Article IV conclusions. By the time of the second review however, the dispersion across program types becomes more similar (Appendix 3 Figure 2).

Appendix 3 Figure 1.
Appendix 3 Figure 1.

Distribution of structural conditions set a Board approval

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Note: Staff’s calculations using MONA.
Appendix 3 Figure 2.
Appendix 3 Figure 2.

Distribution of structural conditions set up to the second review

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Note: Staff’s calculations using MONA.

79. Changes in the global macroeconomic environment also affected the link between conditionality design and previous surveillance. The share of conditionality foreseen in previous Article IV reports was the lowest in 2008 for GRA programs, reflecting the unexpected impact of the crisis on countries such as Iceland or Latvia. For PRGT-eligible countries, where the impact of the crisis was delayed, program structural conditionality was the least related to past surveillance in 2009. By the end of the period, however, the share of structural conditionality that could be mapped to previous Article IV recommendations tended to increase (Appendix 3 Figure 3).

Appendix 3 Figure 3.
Appendix 3 Figure 3.

Leveraging Surveillance: Share of Conditionality Foreseen in Article IV Consultations, 2008-2011

Citation: Policy Papers 2012, 045; 10.5089/9781498340380.007.A001

Source: Staff calculations, based on the 34 countries with programs approved between 2003 and 2011, for which there was at least one standalone Article IV consultation in thetwo years preceding program approval.1 R0 refers to program approval and R2to the second review.

80. Individual country cases show that the main issues identified in surveillance were generally picked up in program design. In the Hungary program, an example of great parsimony, two of the three measures had to do with bank resolution, in line with the surveillance agenda, which had stressed the importance of strengthening bank supervision and regulation. In Costa Rica, the need for an improved bank intervention and resolution framework had been identified in previous Article IV reports and determined most conditionality. The 2009 Article IV report for Greece laid the basis for the fiscal measures making up the bulk of Fund conditionality.

81. When the imperatives of stabilization prevailed, program conditionality appeared sometimes disjointed from the main recommendations of surveillance. For instance, seven of the Gambia’s fifteen measures had to do with reporting audited fiscal and monetary data, which had not been anticipated. In Pakistan and Sri Lanka, the imperatives of addressing stabilization crowded out program measures linked to structural surveillance recommendations. For example, in Sri Lanka the reforms of the state-owned banks were not in the program but the recapitalization of a private bank was.

82. In a number of cases, prominent or recurrent surveillance recommendations were not incorporated in the program, but subsequently added during the reviews. In Iceland, the reform of the Housing Financing Fund was incorporated during the reviews (together with local government fiscal reforms, another surveillance recommendation). In Greece, reviews added labor market reform to conditionality. In Pakistan, tax administration-related conditionality was also brought in during the course of the program; and the Hungary program added pension reform conditionality at later stages. Togo’s initial program was fiscally oriented, but financial sector reforms were added later.

83. In order to enhance the leverage of surveillance into program design, surveillance recommendations could be better prioritized. Article IV Staff Appraisals typically cover an array of measures and reforms, and the implicit suggestion is that they are all urgent. Some degree of prioritization, taking into account technical and political feasibility, might be helpful in setting benchmarks for policy makers, instilling some urgency, and preparing the ground should a program be requested. Conversely, program documents could note the surveillance issues to be addressed once the immediate crisis has abated.

Appendix 3 Table.

Leveraging of Surveillance: List of Countries

article image

Appendix 4. Discontinuation of SPCs and Ownership

84. One of the aims of the discontinuation of SPCs was to reduce the stigma attached to programs, with a view to enhancing ownership through improved flexibility for the authorities in the timing of structural reforms. Evidence from the implementation of structural conditionality suggests that discontinuing SPCs—which were imposing tight constraints in terms of the timing of implementation—did not worsen the overall implementation of conditionality, which tended to improve in 2009 and 2010. This suggests that increased flexibility in the timeline supported a better implementation record for structural conditionality.

85. However, Fund survey respondents disagreed with the notion that more flexible structural conditionality had enhanced ownership or lessened the stigma (BP4). Less than 30 percent of mission chiefs and resident representatives (out of a total of 84 respondents) found that phasing out SPCs had increased ownership or reduced stigma. Responses also confirmed that the discontinuation of SPCs had not entailed worse implementation of program conditionality. A few interlocutors during an outreach mission felt that the discontinuation of SPCs had reduced the leverage of the Fund in supporting structural reforms.

Appendix 5. Guidance on the Principle of Clarity

86. The Operational Guidance Notes emphasize the need for clarity in the specification of conditions. They also set out what is required to be presented in program Board papers. The 2006 Guidance Note stipulated that:

  • The criticality of program goals and conditionality should be clearly exposed;

  • The link between structural conditions and program objectives should be brought out;

  • Where possible, discussions on policy options with authorities should be conveyed;

  • Conditions outside Fund core areas of expertise should be discussed in detail;

  • Fund conditions should be identified (in a table) separate from the authorities’ agenda;

  • Concerns related to implementation should be discussed; and

  • Bank-Fund collaboration should be noted.

87. Following up on the IEO Evaluation of Structural Conditionality, the 2008 Guidance Note added that:

  • At the time of approval of a new arrangement, the staff report should lay out the links between the program goals and reform strategies, and the corresponding structural conditionality, e.g., by presenting them in a separate text table;

  • Board documents should transparently report when expertise in a critical reform area is not available within other multilateral institutions; and

  • The authorities’ reform efforts should be clearly distinguished from conditionality.

88. Finally, the 2009 Guidance Note requested that:

  • Staff reports indicate what is expected in the subsequent review(s) and specify critical future actions as structural benchmarks; and

  • Staff reports for program reviews provide an update related to the objectives of the structural reform agenda, as well as the strategies adopted to achieve them.

1

Prepared by a staff team led by Ranil Salgado and Amina Lahreche and comprising Lars Engstrom, Katrin Elborgh-Woytek, Robert Gregory, Jean-Baptiste Le Hen, Anton Op de Beke, Mika Saito, and Sarah Sanya under the guidance of Dominique Desruelle (all SPR). Valuable contributions were also provided by Sailendra Pattanayak and Mario Pessoa (FAD); and Mali Chivakul, Gavin Gray, Sergi Lanau, and Nathan Porter (SPR). This is the first of four background papers for the 2011 Review of Conditionality and the Design of Fund-Supported Programs, consistent with the Concept Note for the Review and the subsequent Board discussion on February 14, 2011. The second background paper is titled “Design of Fund-Supported Programs” (hereafter, referred to as BP2), the third “Outcomes of Fund-supported Programs” (BP3), and the fourth “Technical Appendices” (BP4).

2

For 2011-12, this paper replaces the Annual Reports on Structural Conditionality, given it provides an update of the application of structural conditionality as proposed in the Implementation Plan in Response to Board-Endorsed Recommendations Arising _ from the IEO Evaluation of Structural Conditionality in IMF-Supported Programs, 2008 (hereafter, MIP for the IEO Evaluation of Structural Conditionality).

3

In 2000-02, the Fund carried out a comprehensive review of its program conditionality, culminating in the adoption of a set of conditionality guidelines by the Executive Board in 2002 (Guidelines on Conditionality, September 2002). This was the first major revision of conditionality since 1979.

4

Quantitative conditionality and some elements of structural conditionality are also covered in BP2.

5

Revised Staff Statement on Principles Underlying the Guidelines on Conditionality, and Operational Guidance Note on Conditionality - December 2009 Revisions.

6

Conditionality in Fund-Supported Programs - Overview. 2001 (hereafter referred to as 2001 RoC); Review of the 2002 Conditionality Guidelines. 2005 (hereafter, 2005 RoC); and Independent Evaluation Office (IEO), Structural Conditionality in IMF-Supported Programs, 2007 (hereafter, IEO Evaluation of Structural Conditionality).

7

The IMF distinguishes between waivers of nonobservance and waivers of applicability as set out in Decision No. 12254-(00/77), July 27, 2000, as amended by Decision No. 14354-(09/79), July 23, 2009, effective January 7, 2010 (http://www.inrf.org/extenial/pubs/ft/sd/index.asp?decision=12254-(00/77).

8

An operational guidance note to staff was issued in 2003, and subsequently revised in 2006, 2008, and 2010 (Operational Guidance Note on Conditionality—December 2009 Revisions, January 2010).

9

The 2009 reform of structural conditionality under the PRGT also provided flexibility for structural benchmarks in ECF arrangements to be linked to program reviews rather than a specific target date.

10

The paper thus covers programs under Stand-by Arrangements (SBA) including precautionary SBAs, such as High Access Precautionary Arrangements (HAPAs), and the Extended Fund Facility (EFF) in the General Resource Account (GRA) and programs under the Policy Support Instrument (PSI), the Stand-by Credit Facility (SCF), and the Extended Credit Facility (ECF) in the Poverty Reduction and Growth Trust (PRGT), along with the high access component of the Exogenous Shocks Facility (ESF-HAC; which was discontinued with the introduction of the SCF) and Poverty Reduction and Growth Facility (which was replaced by the ECF) for low-income countries. Ex-ante conditionality in arrangements under the Flexible Credit Lines (FCLs) and Precautionary Credit Lines (PCLs) (which has been replaced by Precautionary and Liquidity Line)—and ex-post conditionality for PCL arrangements—are discussed in BP4. This review excludes assessment of financing under the Emergency Natural Disaster Assistance (ENDA) facility, the Emergency Post-Conflict Assistance (EPCA) facility, the rapid access component of the ESF (ESF-RAC), and the Rapid Credit Facility (RCF), as well as under staff-monitored programs. ENDA and EPCA have been replaced by the Rapid Financing Instrument (RFI).

11

Quantitative conditionality in PSIs takes the form of quantitative assessment criteria (QACs).

12

See BP4 for more information and summaries.

13

Streamlining structural conditionality in Fund-supported programs-Interim Guidance Note, 2000.

14

In implementing the MIP for the IEO Evaluation of Structural Conditionality, information on program objectives and reform strategies (referred to in this paper also as program instruments) was introduced into the MONA database in August 2008. The sample includes 59 LICs and 31 AMs and EMs, covering the period 2005-2011. Only programs approved after 2005 are included in the sample since data on program goals and reform strategies are only available for those programs. Program goals or objectives are classified into four categories in MONA: macroeconomic stability, external stability, economic growth, and poverty reduction.

15

Unless otherwise mentioned, the PRGT programs include PSI-supported programs.

16

Program strategies or instruments are specific strategies that are taken in Fund-supported programs to achieve the program goals. The 13 MONA categories for program strategies are aggregated into 10 categories: Fiscal (Public Expenditures and Fiscal Revenue), Monetary (Monetary Stability, Stable Inflation, and Central Bank Reform), Exchange Rate Issues, Financial Sector Reform, Trade Reform, Pro-Growth Regime, Social Outcomes, Public Enterprises, Economic Governance, and Other.

17

Fiscal policy strategies group together the following MONA categories: (i) revenue measures, excluding trade; (ii) revenue administration, including customs; (iii) expenditure measures, including arrears clearance; (iv) combined expenditure and revenue measures; (v) debt management; (vi) expenditure auditing, accounting, and financial controls; (vii) fiscal transparency (publication, parliamentary oversight); (viii) budget preparation (e.g., submission or approval); and (ix) inter-governmental relations.

18

The variation for PRGT programs is limited given that almost all these programs included all four objectives.

19

This box is based on a review of original program documents for 18 case study countries (BP4). The total number of program documents reviewed was 26, as some countries had multiple programs.

20

Alignment is assessed in terms of the ability to map program conditions into objectives (identified from staff report and LOI/MEFP) and objectives into conditions. For objectives, both program goals and reform strategies were identified, and if both were available, the latter were used for the mapping.

21

In about 30 percent of GRA cases and 50 percent of PRGT ones, program measures were summarized in a table describing their macroeconomic rationale or criticality.

22

See Appendix 1 for definition of areas of Fund expertise (core, non-core, and shared).

23

The Box 5 Table is based on staff reports. ITs set in 2006 and beyond are recorded in MONA.

26

The analysis in this paper covers ITs set during 2006-11.

27

Unless otherwise specified, this background paper examined all conditions set by the Executive Board or by Management up to September 30, 2011—whether reviewed or not—and include PAs (set by Management), QPCs, SBs, ITs, and the now discontinued SPCs. Conditions that have been cancelled were omitted, since these have been eliminated by the Board. Modified conditions were also excluded as they would otherwise be double-counted. Finally, to avoid multiple counting, continuous SBs and SPCs were only counted when the status of implementation was recorded in MONA.

28

See, for example, the 2005 RoC, the 2007 IEO Evaluation of Structural Conditionality, and the Application of Structural Conditionality—2008 Annual Report.

29

Application of Structural Conditionality—2009 Annual Report.

30

Both the 2005 RoC and the 2007 IEO Evaluation of Structural Conditionality found that the number of structural conditions had remained broadly stable following the issuance of the 2002 guidelines. However, more recent reviews of both GRA and PRGT programs noted that the number of conditions per program had declined. See Review of Recent Crisis Programs and Creating Policy Space - Responsive Design and Streamlined Conditionality in Recent Low-Income Country Programs.

31

PAs at program approval (vs. program review) are considered separately in the figures in this paper.

32

The one PCL arrangement for FYR Macedonia included only two ITs (in addition to standard continuous PCs).

33

Conditions per year are defined as the total number of conditions reviewed in a given year, normalized by the number of programs that are ongoing during the same year (on a quarterly basis, but annualized). In any given year, the volume of conditionality is affected by the frequency of reviews. Frequency itself depended on the initial schedule of reviews (quarterly or semi-annual) and on the implementation record of programs, as delays in reviews or off-track periods will affect the conditionality that is expected to be implemented in the year.

34

Implementation ratios for both the current and previous reviews were used as explanatory variables (PAs were excluded from the analysis). Implementation ratios are defined in Section IV of this paper.

35

The balance between tailoring and evenhandedness—especially for quantitative conditionality—is also analyzed in BP2, which concludes that the design of Fund-supported programs has generally been evenhanded.

36

A probit regression showed that the probability of setting at least one structural condition of high depth in a given review was positively and significantly related to the Country Policy and Institutional Assessment (CPIA) rating of the country. More specifically, an increase by one in the CPIA rating increased by 20 percent the likelihood of a deep condition being set.

37

Capacity was proxied by the World Bank CPIA rating for macro-economic management. The analysis covered 2005-2009, the years for which CPIA data were available for a subset of 54 countries.

38

BP2 also highlights the important role played by the Vienna Initiative for programs in Central and Eastern Europe.

39

The Guidelines do not provide explicitly for coordination with regional institutions. However, they provide clear guidance regarding coordination with the World Bank that can be transposed to coordination with other institutions.

40

See Appendix 5 for more details on the principle of clarity in the guidance notes.

41

Identifying program objectives can be complicated by the use of imprecise terminology, such as “challenges”, “pillars”, “focal areas”,” main elements”, or “components” instead of objectives or targets.

42

Non-core conditionality here does not distinguish non-core conditionality from conditionality in areas of shared expertise with the World Bank.

43

MONA only tracks waivers of observance for performance criteria.

44

Overall implementation is measured as the weighted average of the implementation status of conditionality, where conditions that are not met or waived are valued at zero, conditions that are partially met or met with a delay are assigned a value of 0.5, and conditions that are fully met are assigned a value of 1. Total implementation refers to the implementation of both quantitative and structural measures (excluding PAs). By design, the implementation ratio varies between 0 and 1.

45

Data for SPCs in 2009 and 2010 for PRGT programs were based on a very small sample and were not representative of a trend.

46

Note that database did not allow the tracking of conditions that were equal in essence but were modified slightly to adjust for program specificities. This may bias downwards the figures for implementation.

47

See the list of countries and programs in Appendix 3 Table.