Chapter 7: Fiscal Reforms in IMF-Supported Programs

International Monetary Fund. Independent Evaluation Office
Published Date:
November 2003
  • ShareShare
Show Summary Details

Fiscal adjustment in IMF-supported programs typically includes an agenda of fiscal reforms, and in this chapter we focus on the experience with such reforms on the basis of the sample of 15 programs. We then turn to the process of learning from the past and the role of surveillance in monitoring reform and its link with program design.

Fiscal Reforms in Programs: An Overview

Each program typically includes a number of reform measures in the fiscal area. The 15 programs studied for this evaluation identified 153 specific fiscal-related reform measures, of which 101 were subject to conditionality (divided into 79 structural benchmarks and 22 performance criteria).1 In this chapter we present an overview of these measures in order to identify the relative emphasis placed on different reforms. We also provide an assessment of the success in implementation in different areas based on the assessments reported by staff in program documents.

The universe of reforms can be divided into nine categories: tax policy; tax administration; wage bill and civil service reforms; social sector spending; other spending issues; public enterprise reform, privatization, and private sector development; social security and pensions; organizational reform; and pricing policy of public utilities. Box 7.1 describes the typical reform measures in each category.2

Box 7.1.Public Finance Reform Areas1

The 153 fiscal reform measures reported in the staff reports for the 15 IMF-supported programs are divided into 9 reform areas:


1. Tax policy.(i) Introduction of the VAT (Jordan and Tanzania) or modifications to the VAT such as widening the base (Algeria, Bulgaria, the Philippines, and Ukraine), or rate increases (Ecuador and Senegal); (ii) introduction or expansion of other consumption taxes including excises and taxation of petroleum products (Ecuador, Egypt, Pakistan, the Philippines, and Romania); (iii) reduction of taxes on international trade (Algeria, Bulgaria, Egypt, Jordan, Pakistan, Peru, the Philippines, and Ukraine); and (iv) income tax reform (Ecuador, Egypt, Jordan, Pakistan, the Philippines, and Ukraine).

2. Tax administration.(i) Measures aimed at large tax payers (Bulgaria, Peru, and the Philippines); (ii) improving identification of tax payers (Bulgaria and Pakistan); (iii) strengthening enforcement of collections (Bulgaria, Peru, Romania, and Tanzania); and (iv) personnel training (Jordan).


3. Wage bill and civil service reforms.(i) Wage bill controls (Algeria); (ii) limiting wage increases (Algeria, Peru, Romania, Tanzania, and Uruguay); (iii) limits or cuts in employment (Costa Rica, Egypt, the Philippines, Tanzania, and Ukraine); (iv) legislative action to change civil service statutes (Bulgaria, Costa Rica, and Venezuela); and (v) formulation of reform proposals (Pakistan and Ukraine).

4. Social sector spending.(i) Reform of social sector subsidies (Algeria); (ii) improved targeting (Algeria, Bulgaria, and Ukraine); (iii) improvement or introduction of social safety net (Algeria and Venezuela); and (iv) increase and/or rationalization of welfare spending (Bulgaria, Pakistan, Peru, and Ukraine).

5. Other spending issues.(i) Rationalizing public investment (Algeria); and (ii) reducing spending (Uruguay).


6. Public enterprise reform, privatization, and private sector development.(i) Restructuring public enterprises (Algeria, Jordan, Senegal, Uruguay, and Venezuela); (ii) privatization (Bulgaria, Egypt, Jordan, Pakistan, Peru, Romania, Senegal, Tanzania, and Ukraine); and (iii) encouraging private sector entry to areas dominated by the state (Costa Rica, Jordan, and the Philippines).

7. Social security and pensions. (i) Ensuring the viability of pension systems (Bulgaria, Peru, Senegal, Ukraine, Uruguay, and Venezuela).

8. Organizational reform

(i) Transparency in government accounts or budgeting (Bulgaria and Pakistan); (ii) improved coverage of budget including extrabudgetary funds (Bulgaria and Ukraine); (iii) reduced earmarking (Bulgaria); (iv) improved public expenditure management such as budgeting procedures (including multiyear), controls, and audit (Bulgaria, Jordan, Pakistan, Peru, the Philippines, Ukraine, and Senegal); and (v) creation or revamping of institutions including to manage debt or natural resource–related revenue (Venezuela).

9. Pricing policy

(i) Decontrol or raising of energy-related prices with a fiscal impact (Ecuador, Egypt, the Philippines, Senegal, and Venezuela).

1 The examples in this box are illustrative and not meant to cover all 153 reform measures.

Table 7.1 presents in summary form the frequency of occurrence of the different types of reform measures in the 15 programs as well as the frequency of occurrence of those supported by conditionality. The areas supported by conditionality follow a pattern similar to the overall universe of measures. Tax policy, public enterprise reform, and privatization are the areas of largest emphasis for conditionality, followed by organizational reform, wage bill and civil service reform, and tax administration. Social sector and other spending reforms are typically little emphasized in conditionality.

Table 7.1Distribution of Areas of Fiscal Reforms and Those Supported by Conditionality(In percent)
Areas of Fiscal

Measures Subject

to Conditionality
Tax policy2625
Tax administration149
Wage bill and civil service reforms1210
Social sector spending75
Other spending issues23
Public enterprise reform, privatization, and private sector development1925
Social security and pensions44
Organizational reform1112
Pricing policy57
Source: IEO staff calculations, based on program documents.
Source: IEO staff calculations, based on program documents.

The data also suggest that programs tend to emphasize revenue-related reforms over those related to spending, with a focus on tax policy relative to tax administration. Quasi-fiscal issues, particularly public enterprises, receive more coverage than some core fiscal issues, such as spending reform. The emphasis on quasi-fiscal issues may be attributed to efforts aimed at redefining the overall role of government, which is particularly evident in the sample of transition economies. It may also reflect that earlier adjustment efforts focused on bringing extrabudgetary activity into the central budget (e.g., extrabudgetary funds, public enterprises, and implicit and explicit guarantees in lieu of explicit subsidies). The stress on revenue and the limited attention paid to reallocating or reforming nonsocial spending may also be the result of the short horizon of programs and concentration of IMF expertise.

A number of programs incorporated measures that reduced the short-term deficit but did not reduce fiscal vulnerabilities or improve sustainability. Examples include across-the-board cuts that usually spare the wage bill (e.g., the Philippines) and increasing tax rates on a narrow base, such as raising already very high social security contributions (Romania).

Tax reform focuses much more on introducing or expanding VAT or increasing VAT rates as well as reducing trade tariffs, with relatively less attention paid to income and property taxes.3 Less attention is also given to reducing tax exemptions and evasion of income taxes and customs duties. For example, in the Tanzania program, the reduction in import duties was not accompanied by equivalent efforts to reduce tax evasion in the ports.

The Ecuador program provides a dramatic example of what can be achieved when a determined effort is made to reduce evasion broadly, rather than relying on a VAT rate increase. This effort started prior to the Ecuador program and yielded significantly higher revenue than those envisaged in the program. In fact, these unprogrammed increases in revenue owing to improved tax collection were significantly higher than those expected from the programmed increase in VAT rates (Box 7.2).

Box 7.2.Ecuador: What a Determined Head of Tax Administration Can Do

The success of the Ecuador program (SBA 2000) in increasing revenue was due to both an increase in the VAT from 10 percent to 12 percent, but, more important, to radical improvements in tax administration that started in 1997.

In the 1993–97 period, tax revenue averaged only 6.7 percent of GDP, including 3.5 percent of GDP from VAT. A myriad of loopholes, exemptions, sophisticated evasion, and tax erosion prevailed in the tax system. The country did not have a modern functioning tax administration. The Tax Collection Department of the Ministry of Finance lacked a tax accounting system. It relied on outdated tax forms and did not have any information cross-checking system.

After 1998, tax collections increased dramatically—rising by 80 percent from 1998 to 2001, mainly because of sharp increases in value-added and income tax collections. The collection efficiency of the VAT increased from 42 percent to 68 percent. Of the 4 percentage points of GDP improvements in VAT collections, about one-fourth can be attributed to the increase in the VAT rate, with most of the increase reflecting improved tax administration.

What had changed so abruptly? In mid-1997, the Internal Revenue Service (SRI) was created as an autonomous government agency. The first year was dedicated basic reforms to the old tax collection depart-inherited from the Ministry of Finance, but progress was limited. Following the nomination of a new head in September 1998, a massive process of reform started. An important share of the personnel was dismissed and new staff was hired, and incentives and compensation were improved due to the autonomous nature of the agency complemented by improvements in technology and training. The agency started a forceful process to control evasion, such as surprise visits to enterprises to check invoices and the vigorous implementation of penalties, including closures of enterprises. The overall process was supported by technical assistance from the Inter-American Development Bank.

The lesson is that institutional changes accompanied by determined enforcement can improve collection by amounts significantly higher than increases in tax rates. These changes take time and need to be encouraged by continuous efforts during noncrisis periods. However, legal changes per se do not suffice if, due to political interference, heads of administrations are inhibited from using the available legal tools.

Tax Collections

(In percent of GDP)
Total tax collection6.77.410.012.313.2
Income tax (personal plus corporate)
Value-added tax3.
Excise consumption taxes0.
(In percent)
Collection efficiency of VAT23542455768
Memorandum item:
Nominal VAT rate1010101212
Sources: SRI and Banco Central del Ecuador. In January 2003, a new official GDP series starting in 1993 was introduced, with upward adjustments in GDP figures by about 15 percent to 20 percent. Expressing tax collection as a share of this new GDP series reduces the level of tax collection as a share of GDP. However, it does not change the significant trend toward improvements in tax collection as a share of GDP.

In 1999–2000 includes the 1 percent capital transactions tax. Between January and April 1999, income taxes were abolished and replaced by the 1 percent tax.

The ratio of actual VAT collections over GDP times the legal rate.

Sources: SRI and Banco Central del Ecuador. In January 2003, a new official GDP series starting in 1993 was introduced, with upward adjustments in GDP figures by about 15 percent to 20 percent. Expressing tax collection as a share of this new GDP series reduces the level of tax collection as a share of GDP. However, it does not change the significant trend toward improvements in tax collection as a share of GDP.

In 1999–2000 includes the 1 percent capital transactions tax. Between January and April 1999, income taxes were abolished and replaced by the 1 percent tax.

The ratio of actual VAT collections over GDP times the legal rate.

Often, tax administration reform has focused on the technology side (information systems, manuals for training, and the like) rather than on politically demanding action, including steps within the purview of the executive branch or legislation that would empower tax administrators to collect tax arrears, forcefully pursue tax evasion, and be better protected from political influence. There appears to be significant scope for reorienting efforts to a more vigorous attack on tax evasion and exemptions parallel with efforts to increase VAT rates or broadening the base.

On the spending side, conditionality has concentrated on short-term quantitative targets to reduce public employment, or cap public sector wage increases, or across-the-board spending cuts. The benefits are usually short-lived because of the easily reversible nature of these measures compared with the reorientation of public spending and civil service reform geared to improve efficiency and link pay to productivity. Except for PRGF-supported programs, there is relatively little emphasis on improving propoor public spending beyond vague statements concerning better targeting.

The internal review process usually addresses several of the areas of weakness identified earlier, such as the need to look also at income taxes, spending reallocations, and, perhaps most important, the need for determined actions by the executive in the areas of reducing tax exemptions, limiting tax incentives, and taking concrete actions against tax evasion and tax arrears. But again, these comments come mainly during the review of program implementation, rather than at an earlier stage when they would have more impact on program design.

In summary, the overall picture that emerges is one of heavy emphasis on the revenue side relative to spending reform. On the revenue side, the accent has been on increasing the yield from VAT/consumption taxes. This may reflect the need for measures that quickly yield revenue increases. However, other measures that could also provide important revenue in the short run, such as forceful efforts at collecting tax arrears and reducing tax evasion and exemptions, have received relatively less attention. On the spending side, most measures aim at capping the public sector wage bill through quantitative targets. Less emphasis has been given to reallocating public spending and launching durable civil service reforms. This emphasis may again reflect the mismatch between the quantitative targets and the short length of programs on the one hand, and the time required to complete institutionally and politically difficult reforms on the other hand. Many of these conclusions have also emerged from past staff assessments of cross-country experience in fiscal reform (Abed and others (1998), Mackenzie and others (1997), and Schadler and others (1995a and 1995b)).

Progress in Implementing Reforms

This section presents an assessment of the extent to which programs have been effective in implementing the reform agenda discussed in the previous section. For this purpose, we tracked each of the 153 reform measures described earlier.4 Progress was classified into three categories: “significant,” “partial,” and “little” on the basis of staff’s own evaluation of progress as reported to the Board in program review documents covering the life of the arrangement.5 An index was constructed to measure performance in implementation in each area by assigning a weight of zero to cases of “little progress,” 0.5 to “partial progress,” and 1.0 to “significant progress.”6

Our assessment focuses on reform measures highlighted in IMF-supported programs as reflected in documentation presented to the Board. This only provides a partial view of the total efforts of the IMF in promoting reform of public finances and strengthening fiscal systems. A more complete picture would need to consider medium-term efforts of technical assistance (TA) to address key problems of fiscal systems. We have not done so here because this evaluation concentrates on fiscal adjustment under specific IMF-supported programs. IMF technical assistance will be the subject of a separate forthcoming evaluation by the IEO.7

The results are summarized in Figure 7.1 for several reform areas (Appendix 6 provides the detailed results).8 Overall, the index of implementation ranges between 30 percent and 60 percent—indicating a mixed picture and a sense of partial success at best. Tax policy and tax administration, social sector, and public enterprise reform seem to fare better. Social security and wage bill or civil service reform tend to perform worse.

Figure 7.1.Index Indicating Implementation Progress in 153 Fiscal Reform Measures in 15 IMF-Supported Programs

(In percent)

Source: IEO staff calculations based on program documents.

Areas of strength and weakness

The index provides a measure of “average” performance. To get some idea of variation we have also looked at the distribution of programs according to progress achieved (Figure 7.2). The highest degree of success in terms of significant progress in implementing reform is achieved in the social sector area. Approximately 40 percent of reforms in this area were implemented with significant progress. However, even in this relatively successful area there was significant variability. An example of success includes the introduction of a public works scheme and unemployment compensation under the Algeria program. The Ukraine program illustrates partial progress owing to delays and incomplete implementation of plans to improve the efficiency of health and education spending and targeting of allowances. The finding that implementation of reform in this area was relatively successful is not inconsistent with the earlier findings (Chapter 6) that specific social sector reforms were addressed in less than half of the programs. This means that when these reforms were indeed addressed, their implementation was good in relation to other reform areas.

Figure 7.2.Progress in Implementing Fiscal Reforms in 15 IMF-Supported Programs

(In percent)

Source: IEO staff calculations based on program documents.

In the middle of the performance scale, we identified three areas in which about 30 percent of reform measures showed significant progress. These are organizational reform (including public expenditure management), public enterprise reform (including privatization), and tax administration. However, even across these areas, there was considerable variation.

In organizational reform, the elimination of earmarking under the Bulgaria program is an example of success. Limited progress with organizational reform is illustrated by the failure to establish a debt redemption fund under the Venezuela program or to include extrabudgetary funds in the budget under the Ukraine program. We also consider progress in the Senegal program to be limited, despite having met a basic objective of the program (preparation of documents proposing how to improve public expenditure management).9 Partial progress with organizational change under the Philippines program is reflected in movement to a three-year budgeting framework but only limited results in reviewing the devolution of funds and responsibilities to local government units.

Within public enterprise reform, significant progress includes accelerating privatization under the Romania program. Partial progress includes the unfinished preparation of a plan to restructure public electricity utilities under the Venezuela program.

Regarding tax administration, examples of limited progress include lack of improvement and follow-up on measures to increase penalties for tax evasion and close loopholes in the Peru program; the failure to collect revenue in cash and abstain from netting out operations under the Ukraine program; limited progress in meeting structural benchmarks to strengthen tax administration and taxpayer registration under the Pakistan program.

The proportion of substantial implementation of structural reforms was lowest—under 20 percent—in areas such as civil service and/or wage bill reform and social security (including pensions), as well as tax policy. Successes include limits on the wage bill under the Tanzania program and pension reform under the Peru program to cover unfunded liabilities and issue pension bonds. Partial progress is exemplified by the Costa Rica program that met targets for reducing public sector employment but not those for approval of a Public Employment Law; and by the submission under the Uruguay program of a law to reform special pension funds for some, but not all, groups. On the wage bill, limited progress was achieved, for example, under the Egypt program, which failed to achieve the targeted 2 percent annual reduction in employment or under the Peru program, which failed to contain wage increases to an average 12 percent. An example of limited progress on social security reform comes from the Venezuela program, which failed to result in measures to improve the finances of the IVSS (Social Security Institute).

Why institutional reforms have often been so intractable in IMF-supported programs—examples from the case studies

The previous section showed that significant progress in fiscal reform areas has been limited—in no area did it exceed 40 percent of cases. Insufficient institutional reforms in areas such as tax administration, reallocation of spending, public expenditure management, and civil service reform results in insufficient progress in improving the long-term equity and efficiency of public finances and the flexibility of fiscal systems in response to shocks.

The case studies bring out some of the reasons progress in these areas has been limited. Often it is due to an excessive emphasis in meeting short-term quantitative targets rather than focusing on critical institutional changes that might extend beyond the end of the program. It is largely the result of a mis-match of time frames, for example, the short horizon of programs relative to the time needed to complete these institutional reforms.

As pointed out by Tanzi (2000a), many developing countries now face important second-generation fiscal reforms that focus more on improving institutions than reforming policies. These institutional changes require significant time compared with first-generation policy reforms. The IMF needs to address the resulting mismatch of time frames since the benefits from the first-generation reforms need to be sustained through second-generation reforms. Such reforms may need to be broken down into several steps: some of them can be started at the outset of the program with enough determination from the executive branch; others will require time to the extent they call for legislation and improvements in the implementation capacity of agencies. We elaborate below with some specific examples.10

Examples on the revenue side

During the 1998 Philippines program, tax collection deteriorated owing to governance problems that remained unresolved for the duration of the program—reversing earlier painfully acquired progress.11 Moreover, the inability to reduce tax evasion remains critical today, as noted by the December 2002 Post-Program Monitoring Mission.

Increases in tax rates of “easy to collect taxes” may not be effective when such rates are already high and the tax base is low. For example, in the Romania program, there were diminishing returns to raising already high social security taxes imposed on a low base (see Box 7.3). That lack of flexibility could have been prevented if long-term reforms to widen the tax base and reduce evasion had been pursued more forcefully over time.

Box 7.3.The Romania Program: Diminishing Returns to Raising Tax Rates

In order to meet quantitative targets in the Romania program, the statutory rate of social security contributions was increased from 35 percent in 1997 to 43 percent in 1998, reflecting both the deterioration in the finances of the public pension system and the establishment of the health social insurance fund. Budget revenue from social security contributions consequently surged from 7 percent of GDP in 1997 to 8.9 percent of GDP in 1998. However, the compliance rate was low—about 53 percent. Contribution arrears of large state-owned companies ballooned, with an increasing number of private companies following suit. As a result, total arrears to social security funds went up from 2.4 percent of GDP in 1997 to 3.4 percent of GDP in 1998.

Such circumstances do not warrant a further increase in contributions. It is thus rather surprising that the 1999 SBA relied on a hike of the statutory social contribution rate to the outstandingly high level of 60 percent. As a result, the compliance rate worsened in 1999 to about 44 percent, while arrears to social security funds increased to 3.8 percent of GDP. The private sector accounted for the bulk of the increase in contribution arrears, perhaps because state enterprises were closely monitored under the program.

Program projections implicitly incorporated a significant reduction in compliance rates. The revenue yield of social security contributions in 1999 was conservatively targeted at the same level as in 1998 (8.9 percent of GDP), including on account of the negative impact of the envisaged wage discipline upon the tax base. The actual yield was 10.7 percent of GDP. This revenue performance is partly explained by the fact that wage discipline was actually looser as compared with the targets of the program.

There are also occasions when total revenue might fall if tax reform that (rightly) reduces trade taxes and excessively high statutory income and corporate tax rates is not accompanied by measures to improve collection and reduce exemptions. Reductions in tax rates are institutionally easy—they are stroke-of-the-pen reforms with few losers. In contrast, improving collection requires politically demanding decisions and the development of strong independent revenue-collection agencies. For example, during the implementation of the Tanzania program, tax evasion increased in the ports as trade expanded and important tax exemptions were granted to importers of petroleum (Box 7.4).

Box 7.4.Tax Reform in the 1996 Tanzania ESAF

The policy challenge

Tanzanian taxation prior to the 1996 program was characterized by far-reaching discretionary powers accorded the Minister of Finance, substantial statutory exemptions, including investment incentives, and widespread tax evasion. In 1994 discretionary tax exemptions amounted to the equivalent of over 20 percent of total recurrent revenue. The revenue losses and associated inequities were compounded by tax evasion.

The program, therefore, focused on improving tax administration through support and equipment to the newly created Tanzania Revenue Authority, curbing the discretionary powers of the Ministry of Finance in granting exemptions, diversifying the tax base, and establishing a tax appeals system. In parallel, energy sector fuel pricing and importation were liberalized.

Why revenue failed to increase

While the program’s macroeconomic and trade reforms were relatively successful, progress on fiscal reforms was limited, with a serious gap between tax policy and implementation. The 1996 program aimed at revenue increases of 2 percent of GDP by 1999 while in practice total revenue fell by 2 percent of GDP. Some observers attribute the poor revenue performance to severe weather shocks (El Niño), the negative impact of the Asian crisis, and the contagion effects of the Great Lakes crisis. However, the economy grew at close to 4 percent a year during the program, higher than for many countries in the region. Thus shocks are not a sufficient explanation. More importantly:

  • The program overestimated the speed at which institutional capacities could be strengthened, and VAT revenue projections were too optimistic. Tax evasion continued to be a serious problem. Adjusting tax legislation was important in modernizing tax administration, but much more attention should have been paid to capacity building. Lacking technical and managerial capacity, the Tanzania Revenue Authority was unable to implement the new policies expeditiously or to resist political pressure. Lack of technical competence and inadequate data on potential taxpayers led to poor tax assessments, inefficient coverage, and thus to revenue loss.

  • While discretionary exemptions were largely eliminated at the government level, statutory exemptions for religious foundations, nongovernmental organizations, and other institutions remained substantial. In an earlier bid to attract investors, the government provided broad tax incentives to firms in mining and tourism. This minimized the revenue contribution of these growth sectors. Moreover, legal provisions for exemptions, most recently in the statutory provisions of the VAT, result in pressure to use them in ways not intended.

The policy sequencing also contributed to the revenue decline. In retrospect, tariffs were lowered too quickly before compensatory tax broadening measures, including strengthened administration, were in place. Increased corruption in the ports and customs administration were major contributors to the revenue decline. In the case of oil sector liberalization, the freeing of import licensing before setting up an industry regulator led to a situation of significant fuel smuggling. In addition, contrary to assumptions, lower tariff rates did not automatically increase tax compliance. Also, the tax base was eroded due to the failure of several inefficient industrial enterprises.

Many reforms to improve revenue performance (both quantitative and qualitative) require different time spans and are subject to different constraints such as: (1) lack of support of the executive to en-courage tax agencies to collect tax arrears and improve collections from well-known sources of tax evasion owing to lack of political will; (2) lack of legislation to empower tax agencies which hinders effectiveness even though the executive is willing to support the actions of these agencies; (3) implementation capacity of the tax agencies may be inadequate even if (1) and (2) are not problems. Such capacity can only be improved through training and technical assistance, which require long lead times. A clear road map is needed to guide actions in these areas over time and could be provided through surveillance. Where decisions under the control of the executive branch are the bottleneck, this can be taken up directly in program conditionality. When the constraint is the lack of legislation to empower tax agencies or implementation capacity that requires time to develop, surveillance should aim at evolving an agreed time frame for reform. This approach would allow conditionality in program situations to be more effectively focused on critical areas and would, therefore, be fully compatible with present streamlining initiatives.

Examples on the spending side

Programs often aim to contain the wage bill by capping public sector wages and/or reducing public employment by specified levels. However, progress in this area has been elusive. Short-term declines in real wages are usually followed by pressure for reversals, such as in the Romania and Ecuador programs. Although both the Tanzania and Costa Rica programs were able to achieve reductions in public sector employment, such progress is easily reversed after the program. Long-term civil service reform is therefore critical, but it is also problematic. Efforts to pass public employment legislation under a program have proven difficult. Attempts to pass legislation in the Costa Rica and Bulgaria programs were unsuccessful. Civil service reform initiatives require long preparation and consensus building. They should be encouraged in the context of longer-term programs such as EFFs and/or integrated, in close collaboration with the World Bank, into a longerterm framework of reforms specified under the broader road map discussed above.

Learning from the Past and the Role of Surveillance in Monitoring Progress

As argued in the previous chapter, programs often have too short a time frame to tackle major public finance reforms—particularly when programs are associated with crisis. Sustained reform, particularly in complex institutional areas calling for important political decisions, is better addressed in noncrisis years. The role of surveillance in setting a clear road map of structural reform and monitoring over time could greatly encourage this process.

In this chapter, we summarize our findings regarding learning from past experience and the role of surveillance in monitoring progress in structural reforms in the fiscal area and their links to programs. Specifically, we consider the extent to which (1) programs build on past reform efforts and try to learn from such efforts, (2) surveillance follows up and encourages reform, and (3) programs build on surveillance to address major public finance distortions (Appendix 8, Table A8.1 provides details). For each of these three areas, we explore a subset of questions as follows:

Learning from the past

  • To what extent do program documents analyze and evaluate past fiscal performance?

  • To what extent do program documents specifically analyze and evaluate fiscal performance under the previous arrangement? Does self-standing surveillance (not associated with a program request or program review) tend to perform better in this area?

Monitoring of fiscal reforms under surveillance

  • To what extent has surveillance flagged the need to accelerate fiscal reform in areas where implementation was lacking?

Links between surveillance and programs

  • Were most major fiscal reform issues flagged during surveillance incorporated into the program?

  • Were most problem areas taken up in programs identified by earlier surveillance?

To address these questions, we reviewed surveillance activity over the three years prior to the program. This involved an analysis of 33 preprogram surveillance documents associated with the sample of 15 programs studied (Appendix 8, Table A8.2).

To obtain a quantitative estimate of the effectiveness of the IMF in each of these areas, the evaluation team’s assessment in response to each question was classified into three categories: poor, mixed, and good performance. We then again constructed an index of performance by assigning a weight of 1, 0.5, and 0, respectively, for “good,” “mixed,” and “poor” performance.

Figure 7.3 summarizes the resulting average performance in each of the five areas. “Learning from the past” appears as an area of generally poor results. Program requests are only partly successful in evaluating past fiscal performance—with an index of success of about 50 percent. The results are worse (35 percent success) when documents are judged on a more pointed question: how well they analyze performance and policy failures under the previous arrangement. Overall, programs tend to focus on performance during the previous year and rather independently of previous arrangements. Few efforts are made to analyze the factors behind past policy failures.

Figure 7.3.Index of Performance: Learning, Follow-Up, and Links Between Programs and Surveillance

(In percent)

Source: IEO staff calculations based on program documents.

Efforts during surveillance to flag the need to accelerate reforms are also limited, with an index of success of about 40 percent.

Finally, Figure 7.3 shows a sharp asymmetric link between the issues identified under surveillance and those taken up by the subsequent IMF-supported program. Problem areas flagged under previous surveillance are typically incorporated fairly well in programs and this is the area of best performance (80 percent). On the other hand, programs include many reform areas that were not flagged early on by surveillance. In fact, this is by far the worst area of performance (performance is good in only 15 percent of cases). Although unexpected developments and shocks may call for programs to include fiscal reforms not previously flagged in surveillance, we would expect this to be the exception rather than the rule. Moreover, introducing issues which have not previously been flagged as a concern may reduce country ownership and suggest that the IMF is using its leverage to push for reforms that are not essential (since they were not previously flagged).

Inevitably, the average value of the index masks significant variation. To explore this variability, Figure 7.4 shows the distribution of cases behind the averages. We also give examples of specific cases to provide a better sense of such variability and identify best practice.

Figure 7.4.Distribution of Cases According to Performance

(In percent)

Source: IEO staff calculations based on program documents.

Learning from the past

Program requests show a satisfactory analysis of past fiscal performance in only 40 percent of cases and in one-third of cases that analysis is poor. These results deteriorate when programs are judged by how well they examine performance under the last arrangement; only one-quarter perform well and almost two-thirds perform poorly.

There are, however, good examples, where program requests perform well on both counts, such as the Algeria, Philippines, and Senegal programs (see Box 7.5).

Box 7.5.Good Examples of Learning from the Past

The Algeria report has three well-focused chapters that evaluate past economic performance. This includes reform implementation during 1989–91 (including the past two IMF arrangements), and developments from 1992 to 1994. The report also discusses major fiscal distortions not raised before. It proposes new policy recommendations to en-courage reform (on government investment spending, wage policy, and various revenue measures). For example, the report proposes that government investment should be limited to priority projects and proposes transferring investment financing responsibility from the Treasury to the enterprises and banking system.

The Philippines program-request document, which was prepared jointly with an Article IV surveillance report, thoroughly discusses performance under the previous EFF. The assessment includes main goals, achievements, and policy failures for the overall program and fiscal policy. Box 2 on “The Extended Arrangement in Retrospect” provides a brief and clear summary of the main areas of progress as well as lack of progress in fiscal policy and reforms. Box 4 on the “Comprehensive Tax Reform Package” discusses the main elements of the reform and the main implementation issues.

The Senegal report does a good job overall, although there is scope for more specificity and a more analytical look at the past. The executive summary and the chapter “Performance Under the Previous ESAF-Supported Program and Recent Developments” comprehensively evaluates fiscal performance under the previous arrangement. The document includes a summary of selected policy performance indicators with the main achievements in fiscal policy during the three years of the previous arrangement. There is a thorough discussion of why some reform implementation was behind schedule (energy and privatization). The main fiscal achievements under the previous arrangement are clearly addressed. However, fiscal targets and objectives for the previous ESAF are not made explicit and are discussed only for the previous year.

An attempt was made to assess whether more recent program-request documents make stronger efforts to evaluate past fiscal performance than was done in the 15 programs originally studied. Eight of the 15 countries had more recent programs, namely, Algeria, Bulgaria, Jordan, Pakistan, Peru, Romania, Tanzania, and Uruguay.12 These 8 programs were then examined under the same criteria, namely the extent to which they evaluated and analyzed past fiscal performance. The results show no significant improvement with respect to the earlier results—only 4 of the 8 cases (the Algeria, Tanzania, Jordan, and Romania program-request documents) were judged as successful in this area.

In order to assess whether performance is better for self-standing surveillance reports—which potentially have the opportunity to analyze progress and learn lessons without program distractions and operational pressures—we looked specifically at six free-standing Article IV documents that preceded the program being studied.13 Although the sample is small, the results are revealing. Only the Romania 1998 Article IV conducted an in-depth examination of the main fiscal issues of the prior 18 months and the reasons why the previous arrangement went off track.

Monitoring of reforms under surveillance

In only one-quarter of cases was surveillance forceful in flagging the need for reform where implementation was lacking. In 40 percent of cases this effort was weak.

Ukraine is one of the better cases. Both Article IV consultations (for 1995 and 1997) thoroughly identified and analyzed reasons for failure in past reform implementation, as well as remaining implementation risks. The 1995 report identifies and discusses four areas most affected by slippages in implementation of the 1994/95 stabilization program (SM/95/320). Separate sections discuss the problem of external arrears and the social safety net. Implementation issues are explicitly analyzed and specific measures recommended. The 1997 consultation identifies the main risks to the program and singles out risks to fiscal policy and the budget.

The 1995 Egypt Article IV report was candid in focusing on areas of disagreement between the staff and the authorities on such issues as wage bill reduction, public sector employment cuts, civil service reform, privatization, and social safety net issues where the staff pressed for improved targeting of social transfers and less reliance on generalized subsidies.

The surveillance process in Bulgaria is a good example of improvement over time. The 1995 Article IV is weak on recommendations. Vague statements were made such as: “the staff underscored the importance of slowing wage increases as much as was feasible” [but with no target] or “the staff recommended that the authorities focus on expenditure rationalization rather than spending cuts” [with no specifics]. In contrast, the 1997 Bulgaria Article IV is clear on recommendations and evaluation of progress (or lack of it), with structural reforms. The report contains a template with the status of conditions for completion of the first review under the SBA. It discusses special efforts to accelerate banking system reform, and follow-up on the efforts of the new government to implement legal wage limits. Nevertheless, the report has very little on expenditure control, with the only reference being: “the authorities are also taking steps to improve expenditure control by including in the budget provisions to limit commitments of spending agencies to 90 percent of allocated funds.”

On the other hand, preprogram surveillance in Peru was not forceful in encouraging reforms to reduce tax evasion, contain the growth of the public sector wage bill, and increase social spending. Another example is the 1997 Article IV for Ecuador—the only surveillance exercise in the 1996–99 period. The report failed to highlight the dramatic deterioration of the banking sector and the need for urgent actions particularly in the supervision area. Instead the recommendations are buried in the middle of the report rather than being flagged up front in the summary. Little attention was given to documenting and addressing the massive tax evasion taking place most of the references being focused on the need to increase VAT rates.

Links between surveillance and IMF-supported programs

Programs successfully include issues identified during surveillance. On the other hand, surveillance fails to identify many of the reforms that subsequent programs found necessary to incorporate.

Programs incorporate the main issues flagged during surveillance in about 80 percent of cases. For example, the Tanzania program reiterates many of the issues raised during surveillance, notably the need to strengthen the implementation capacity of public sector institutions. During surveillance, inadequate institutional capacities were held responsible for poor tax administration, tax evasion, and inability to formulate and implement policies. Similarly, the Uruguay program request explicitly targets areas flagged during surveillance, including the need to continue improving tax administration and tax compliance. It also clearly and forcefully recalls measures raised in earlier surveillance reports to strengthen public sector banks, restrain wages, and complement social security system reform.

Preprogram surveillance fails to identify problem areas dealt with by programs in almost 90 percent of cases. Only in the Pakistan and Philippines programs were almost all issues under the program previously identified in surveillance.

In the case of Bulgaria, preprogram surveillance failed to flag important measures to enhance fiscal transparency (such as explicitly incorporating into the budget quasi-fiscal costs of restructuring and liquidating state-owned enterprises and any support provided to them), and the consolidation of the largest extra-budgetary funds into the budget. Surveillance also did not flag tax administration measures to enhance tax collection (such as the development of a tax collection strategy, including enforcement, audit, and fraud investigation), measures to improve expenditure controls, and, finally, measures to rationalize and increase the cost-effectiveness in the supply of public health services, all of which were included in the subsequent program.

Surveillance prior to the Peru program did not address the need to rationalize employment in the public sector and reform the public sector wage structure.


Focusing on the unfinished reform agenda and reducing vulnerabilities to future crises will require strong follow-up during surveillance as well as continuity in successive programs. At present, surveillance does not forcefully flag policy inaction—many times it is insufficiently candid in language. Many program-request documents are insufficiently linked to past outcomes and past reform attempts. Although based on a very small sample, self-standing surveillance does not seem to yield better results. This is a missed opportunity because we would expect that surveillance not associated with a program request or review would have a genuine opportunity to take a more strategic perspective on both assessing whether fiscal reforms over time add cumulatively to better fiscal systems, and spelling out clearly what the remaining fiscal agenda for the future should be.

Surveillance should play a much more forceful role in providing a medium-term road map of structural reforms to be followed up over time, with or without programs. Progress and reasons for inaction should be reported candidly. That road map would then provide guidance for the specific reform priorities to be taken up in successive programs—this being particularly important in repeat users of IMF resources. Such an analysis would also provide the broader strategic overview of fiscal reform priorities as well as the success or failures of past efforts that would help inform choices about the priorities for reform implementation in any future programs.

Some qualifications regarding the universe of reforms are necessary. First, for programs possessing extensive reform agendas, such as the Bulgaria and Ukraine programs, we narrowed down the number of reform measures to a subset of reforms representing the major areas of emphasis. Second, for programs possessing obvious groupings of intricate and interrelated reform measures, we collapsed various measures into one all-encompassing measure; for example, the Uruguay program requires various measures relating to the publishing of fiscal data, other reports, and studies. These have been consolidated into an umbrella “transparency and disclosure” reform measure.

Due to the small sample size of the measures supported by performance criteria, we collapse the structural benchmarks and performance criteria into one single group referred to as measures supported by “conditionality.”

Property taxes are usually levied by local governments while the IMF focuses on the central government. However, to the extent that local governments receive significant transfers, the central government has leverage to press for a more aggressive use of property taxes.

One caveat to our findings: This analysis has been handicapped by lack of consistency in following up and/or reporting progress with reform. This limitation has introduced an element of subjectivity in interpreting progress in implementing the agreed structural measures.

Significant progress” indicates that by the end of the program most of the agreed reform was enacted. For example, the first review of the Bulgaria program reports that the largest extrabudgetary funds were incorporated into the budget, as envisaged. “Partial progress” indicates that the agreed agenda remains to be implemented but there was noticeable movement in a positive direction. For example, the Pakistan program called for improved spending monitoring based on a variety of transparency, governance, and accounting measures. However at the end of the program, the review stated that “reconciliation of especially provincial spending remains too slow, resulting in large amounts of spending remaining unclassified for too long, thus hampering proper expenditure management and prioritization.” On the other hand, the program resulted in improved fiscal transparency, such as publishing reconciled public accounts. “Little progress” suggests change that is barely perceptible, if at all. For example, the Egypt program envisaged phasing in the extension of the input crediting mechanism to capital goods under the general sales tax from January 1997. However, this reform was delayed more than once owing to lack of parliamentary approval.

The index can be interpreted in one of two ways. For example, an index value of 50 percent could indicate that on average about half of the reform measures were successfully implemented. Alternatively, it could also be interpreted as all reform measures showing only partial progress in implementation.

See Appendix 7 for a summary of the IMF’s fiscal TA in the 15 countries.

The discussion excludes the “other spending” area given that it is only covered in three programs and the sample size is thus too small to draw general conclusions. Pricing policy is also excluded as it is not a core fiscal area.

This is the one case where we are more critical than the assessment of the staff due to the undemanding measures required.

In considering these examples, we would like to reiterate that this report has focused on adjustment and policy reforms under specific IMF-supported programs and did not explore the links between past levels of TA and programs. In particular, FAD TA has been crucial for tracking HIPC spending and ROSC initiatives, areas not focused on in this evaluation.

The problems encountered in improving the tax structure and strengthening tax administration over a long series of IMF-supported programs are discussed in more depth in a detailed case study of the Philippines prepared as part of the evaluation of prolonged use of IMF resources. See IEO (2002), chapter 10, pp. 163–65.

The specific program requests examined are Algeria EFF 1995, Bulgaria SBA 2002, Jordan SBA 2002, Pakistan PRGF 2001, Peru SBA 2002, Romania SBA 2001, Tanzania PRGF 2000, and Uruguay SBA 2002.

These six Article IV documents are a subset of the 11 self-standing Article IV documents included in the sample. This subset of Article IV includes Algeria 1992, Costa Rica 1994, Ecuador 1997, Romania 1998, Tanzania 1995, and Venezuela 1993.

    Other Resources Citing This Publication