2011 Review of Conditionality - Design of Fund-Supported Programs

Design of Fund-supported programs aims to address country specific needs while remaining even-handed and consistent with Fund policy. This paper examines the extent to which program design and conditionality have been appropriate in pursuing these goals, by seeking to answer several questions: has program design been consistent and evenhanded; has it addressed country specific needs and objectives appropriately; has it been based on reasonably good macroeconomic projections; and has it been flexible in the face of evolving country circumstances. The description and analysis focuses on the period between 2006 and September 2011, with some attention to the 2002-05 period.


Design of Fund-supported programs aims to address country specific needs while remaining even-handed and consistent with Fund policy. This paper examines the extent to which program design and conditionality have been appropriate in pursuing these goals, by seeking to answer several questions: has program design been consistent and evenhanded; has it addressed country specific needs and objectives appropriately; has it been based on reasonably good macroeconomic projections; and has it been flexible in the face of evolving country circumstances. The description and analysis focuses on the period between 2006 and September 2011, with some attention to the 2002-05 period.

I. Overview

1. Design of Fund-supported programs aims to address country specific needs while remaining even-handed and consistent with Fund policy.1 This paper examines the extent to which program design and conditionality have been appropriate in pursuing these goals, by seeking to answer several questions: has program design been consistent and evenhanded; has it addressed country specific needs and objectives appropriately; has it been based on reasonably good macroeconomic projections; and has it been flexible in the face of evolving country circumstances. The description and analysis focuses on the period between 2006 and September 2011, with some attention to the 2002-05 period.2

2. To examine these questions, the paper relies primarily on descriptive and regression analysis, particularly of programs’ macroeconomic adjustment and access. It begins with a regression analysis of the extent to which factors like initial macroeconomic conditions and country characteristics explain macroeconomic adjustment and levels of access in program design. The results permit an examination first of the degree of comparability across countries and second of the economic appropriateness of these factors. This regression analysis is complemented by a description of adjustment patterns and structural reform in key sectors.3 Third, the paper analyzes the accuracy and possible bias of the macroeconomic projections in program countries, which are critical to both initial program design and subsequent modifications. Fourth, the paper examines data on trends in the flexibility of program implementation.4

3. The paper finds that program design appeared—in general—tailored to country needs, even-handed, and flexible. The positive findings broadly hold, looking at all programs together, comparing among programs, comparing to past periods, and assessing the economic logic of the program design. They hold for the substantial majority of programs supported under both the Fund’s General Resources Account and the Poverty Reduction and Growth Trust (hereafter GRA programs and PRGT programs, respectively), with variations as specified below.

  • As examined in section II, a limited number of initial conditions can explain programs’ macroeconomic adjustment and access to Fund financing, supporting a conclusion of even-handedness in program design.5 The fit of regressions explaining program design was generally good for GRA programs; the weaker fit for PRGT programs does not conclusively confirm even-handedness but suggests consistency, considering the greater heterogeneity of PRGT countries.

  • As discussed in section III, the extent of adjustment by sector and access levels broadly made economic sense, considering relevant initial country conditions and characteristics. This finding supports the conclusion that programs are generally well-tailored to country needs and characteristics, with reasonable balance of adjustment and financing. In particular, adjustment and access adapted appropriately to the exceptional conditions associated with the global financial crisis, with increased access. Programs addressing capital account crises also had higher access, reflecting the lessons of the Asian financial crisis.

  • Moreover, section IV shows that the macroeconomic projections underpinning program design in the 2006-11 sample were generally unbiased, and did not exhibit a systematic optimistic projection bias, contrary to studies based on programs in earlier periods.

  • Finally, section V finds that flexibility has increased compared to the past, especially in terms of modifications of conditionality and augmentations of access, and has helped maintain high implementation rates despite the global recession.

4. However, some of the recent programs faced program design challenges. Program design faced myriad difficulties in these so-called “wave 2” crisis programs, approved after August 2009. In particular, policy space has narrowed in many countries, making the inherent trade-off between adjustment and financing more difficult. Some of these programs posed a number of exceptional design challenges and risks, including: high public debt, sustained loss of market access, low growth, and competitiveness issues. Membership in a currency union, macro-financial linkages, spillovers, and systemic risk created further challenges in some cases. The appropriateness of the resulting strong programmed policy adjustment also needs to be continually reassessed.6 In addition, membership in the Euro Area (EA) was found to be associated with higher access (taking into account other measurable factors). Large systemic risks justified the access levels; however, the reference to systemic risk in justifying high levels of access could have benefitted from more in-depth analysis at program inception. While relatively few in number, these programs nevertheless represent a large proportion of access to Fund financing, also creating financial risks to the institution. It is too early to assess the final outcomes of these programs, although the first Greece program already encountered significant difficulties.7

5. Other important caveats to the analysis also lead to caution in interpreting the paper’s encouraging findings. Projection errors have grown for GRA cases: the shocks associated with the global financial crisis help explain this outcome, but room for improvement exists. The findings are founded on an overview of programs in the aggregate, and certain programs have exhibited specific design flaws, as highlighted by Ex Post Assessments of Longer Term Program Engagement (EPAs) and Ex Post Evaluations of Exceptional Access (EPEs) and indicated below, although no systematic design flaws were apparent.8 Two findings suggest it is worthwhile to examine whether there is further scope in certain programs for relaxing macroeconomic adjustment: (i) the fiscal adjustment in programs in fragile states tended to be as large as or larger than other PRGT cases, controlling for other factors; and (ii) while program design in capital account crisis cases has drawn on the lessons of the past, there may be some room in certain cases for further relaxing initial fiscal targets, since stabilization in the medium term was better than expected despite narrowly missing fiscal targets in the initial program period.

II. Comparability and Even-Handedness of Fund-Supported Programs

6. The design of Fund-supported programs appeared generally consistent and even-handed, although certain results raise questions. Based on regression results, programmed adjustment and access took into account relevant initial conditions and country characteristics in a fairly consistent manner, implying comparable, even-handed treatment.9 The generally good fit of regressions explaining program design, especially for GRA programs supported this conclusion, as did the positive responses on the surveys of stakeholders (see BP4 for survey results). These regressions used a methodology that automatically and objectively selected the best explanatory model using an ample list of candidate regressors.10 In particular, program design was also largely comparable across regions,11 and there was no evidence that powerful Fund stakeholders exerted a systematic influence.12 However, the power of the dummy variable for membership in the Euro Area in explaining variations in access warrants further examination (below).

7. Stakeholders responding to surveys perceived programs to have equivalent conditionality, though country authorities often could not judge. Few respondents disagreed with the proposition that programs have equivalent conditionality, but around half of country authorities and donors responded that they did not know, suggesting room for improvement in communications on evenhandedness.

8. As noted above, the design of GRA programs was highly consistent with countries’ initial conditions and characteristics, which strongly supports a conclusions of even-handedness (Figure 2.1). While the fit for the regressions was not as high for reserve adjustment and the number of structural conditions, it was still fairly good. The specific initial conditions and country characteristics explaining program design generally appeared appropriate. The absence of some variables one might have expected to be significant constitutes an additional caveat, however (e.g., institutional capacity measures for the number of conditions or successor program for GRA access).

Figure 2.1.
Figure 2.1.

Summary of Factors Affecting Adjustment and Access

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.Note: This figure summarizes how much of the variations in each aspect of the Fund-supported programs (in horizontal axis) can be explained by variables identified as macroeconomic conditions and country- and program-characteristics. Variables such as fiscal balance, current account balance, inflation, and growth are considered macroeconomic conditions; currency union and fragile state dummies are country characteristics; and crisis-program and subsequent program dummies are program characteristics (more details can be found in Appendix Tables I.1 and I.2).

9. PRGT program design reflected fairly consistent relationships to conditions and characteristics, but it is harder to draw strong conclusions on even-handedness. The fit of the regressions for programmed adjustment and access for PRGT programs was moderately good, but perhaps insufficient to conclude strongly whether or not PRGT program design was even-handed in all cases. Importantly, the fit was stronger for access. The specific macroeconomic initial conditions and country characteristics that explained program design appeared largely appropriate.

10. Importantly, the more moderate fit of the regressions on PRGT program design likely reflects heterogeneity rather than clear evidence of inconsistent design. In comparison to GRA countries, the heterogeneity of PRGT countries, programs, and objectives likely plays a role in the lower fit of some regressions, so that the results are not surprising. In particular, the method used favors more parsimonious models over larger models with better fit. In any case, an average fit of 0.5 is considered good for this type of method and cannot constitute evidence of inconsistency. In fact, to the extent that country heterogeneity could not be captured in the regression, lower R2 values could possibly also imply that programs were tailored to country characteristics that were not readily quantifiable or common to several countries.

11. Membership in the Euro Area was a significant variable in explaining the level of access to Fund financing in GRA programs. Its inclusion explained much of the variation represented by the high levels of access for these programs. On the other hand, membership in the Euro Area was not statistically significant in explaining adjustment, although the “wave 2” countries, which include the Euro Area programs, were associated with somewhat higher fiscal adjustment.

12. The extraordinarily high access in the Euro Area programs was viewed as necessary given the systemic risks from their crises and large debt rollover needs. As explained in program documents, the spillovers from these crises posed specific, broader systemic risks that justified higher access, including through risks to European and global financial systems. These specific spillovers indeed seemed to be a main driver behind program design, especially since general financial interconnectedness was not found to be an important explanatory variable.13 In addition to contagion and systemic risks, the Euro Area programs faced serious design challenges, related to membership in a currency union (eliminating nominal exchange rate devaluation), competitiveness problems, and sustained total loss of market access. The large-scale co-financing from the European Union also added complexity to the programs. It is useful to examine briefly the evolution of the role of spillovers and systemic risks in program design.

13. The 2008 Hungary program was among the first cases in which reducing risks of regional spillovers was cited as an important objective. The Stand By Arrangement (SBA) request made an explicit reference to the need to reduce the scope of financial spillovers to other countries, given the exposure of the largest Hungarian bank to Central, Eastern, and Southeastern Europe (CESE) and of several Euro Area banks to Hungary via their subsidiaries. In fact, the large exposure of foreign banks to Hungary helped set a precedent of cross-border banking supervision and resolution framework at the European Union level.

14. Cross-border spillovers and systemic risks were invoked in justifying high levels of access for Ireland, Greece and Portugal, despite concerns over debt sustainability. The proposed exceptional access in Ireland (19.5 billion SDRs or 2,322 percent of quota at program approval14), Greece (26.4 billion SDRs or 3,212 percent of quota at program approval), and Portugal (23.7 billion SDRs or 2,306 percent of quota) entailed substantial risks to the Fund, both in terms of the stock of outstanding credit and the projected debt service, that would last for an extended period. However, the crises—if uncontained—could have spread rapidly in other countries through financial exposures both to and from peripheral Europe. The analysis of systemic risk, while supported by subsequent analysis and experience, was not fully developed or quantified in the initial Greece program document. Deeper analysis of risks in Euro Area programs over time has facilitated better informed decisions. As risks evolved and became more prominent during the implementation of the program, the analysis and presentation of risks was expanded in subsequent program documents. This increased coverage should be emulated in future exceptional access cases at program inception. It is also important to note that historically, access levels have periodically diverged from trends as the Fund has strived to meet new challenges—such as the massive liquidity needs in the Asian capital account crises or public debt challenges in Brazil and Turkey in 2000s, although the risks to debt sustainability were perhaps smaller in these cases (Figure 2.2).

Figure 2.2.
Figure 2.2.

Access through the Ages: Exceptional Access in GRA-supported Programs, 1995-2011

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: MONA

III. Adjustment and Access in the Design of Fund-Supported Programs15

15. Macroeconomic adjustment and access in Fund-supported programs generally were tailored to country needs and objectives in an economically appropriate manner.16 As examined also in BP1 and BP3, program design aimed to achieve economic stabilization through a combination of adjustment and financing, including access to Fund resources. Promoting growth and poverty reduction was also a goal in nearly all PRGT programs and about half of GRA cases.17 Pursuing stabilization and sustainability while minimizing output losses from adjustment appears as a consistent theme in the following assessment.

16. The section examines the economic reasoning behind adjustment and access using descriptive and regression analysis. First, it examines initially programmed adjustment in three sectors in turn—fiscal, monetary, and external—looking at its magnitude, composition, timing, and outturns. The descriptive analysis focuses on cases with imbalances (e.g., fiscal, inflation, and current account) in the top quartile of all programs, supplemented by the regression analysis presented in section II.18 Subsequently, the section reviews the factors explaining the levels of access to Fund financing, as well as the extent of structural conditionality and its linkages to adjustment. To conclude, the section assesses the resulting macroeconomic policy mixes and adjustment/financing balances for different policy challenges (such as large fiscal deficits, public debt, capital account crises, or fragile states), considering how policies interact with and reinforce each other.

A. Fiscal Adjustment

17. Fiscal adjustment in Fund-supported programs appeared generally appropriate considering countries’ initial conditions, program characteristics, and objectives. Moreover, fiscal adjustment appeared to be more flexible in its magnitude and timing than in the past (Box 1 and BP3).

18. The magnitude of fiscal adjustment was generally restrained, largely out of concern for output effects, particularly during the global economic crisis (Figure 3.1 and Table 3.1). In PRGT cases, fiscal adjustment was larger both for countries with stronger institutions (rule of law), presumably given their higher capacity. Surprisingly, however, this effect was offset to a degree for fragile states, which regression results also show had higher fiscal adjustment, controlling for other factors (including their low rule of law scores)—these offsetting results should be interpreted carefully.19 Generally, initial fiscal and debt positions were strong enough to allow for limited adjustment in most cases, while preserving sustainability and generating resources for priority spending. In particular, fiscal adjustment was smaller during 2006-11 for PRGT programs and more back loaded for GRA cases compared to previous periods (Box 1). The magnitude of adjustment picked up for wave 2 programs, however.

Figure 3.1.
Figure 3.1.

Programmed and Actual Macroeconomic Adjustment

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.Note: PRGT recent crisis programs are defined as those approved after August 2008.
Table 3.1.

Regression Results Explaining Programmed Fiscal Adjustment

article image
Source: IMF staff estimationsNotes: All variables are statistically significant at the 10 percent level or higher. See Appendix Tables I.1-3 for descriptions.
Appendix Table I.1.

Regressions Explaining Planned Macroeconomic Adjustment in GRA Programs

article image
Source: IMF staff estimations.Notes: ***, **, * denote 1, 5, and 10 percent significance levels, respectively. The set of regressors included in each regression is determined using the iterative Bayesian Model Averaging (iBMA) methodology. Grey shading indicates variables excluded from the set of candidate regressors in iBMA. The number of observations is 56 for all regressions.
Appendix Table I.2.

Regressions Explaining Planned Macroeconomic Adjustment in PRGT Programs

article image
Source: IMF staff estimations.Notes: ***, **, * denote 1, 5, and 10 percent significance levels, respectively. The set of regressors included in each regression is determined using the iterative Bayesian Model Averaging (iBMA) methodology. Grey shading indicates variables excluded from the set of candidate regressors in iBMA. The number of observations is 85 for all regressions. The exception is the access regression, which excludes PSI programs and only has 72 observations, because access in PSIs is zero by definition.
Appendix Table I.3.

Descriptive Statistics and Sources of Variables Used in Regressions

article image
article image

Given reporting lags, the last available year is used throughout for those programs for which the appropriate observations was not yet available. Variables cover IMF programs from 2002-11, except those for which data was missing for some variables. Those were for GRA programs Antigua and Barbuda (2010), Bosnia and Herzegovina (2002), Iraq (2005, 2007), Kosovo (2010), Serbia (2009, 2011), Seychelles (2008, 2009), and St. Kitts and Nevis (2011). The excluded PRGT programs are Afghanistan (2006), Cote d’Ivoire (2002), Lesotho (2010),.Grenada (2006), Sao Tome and Principe (2005, 2009), and Sri Lanka (2003).


Access is not used as an explanatory variable in explaining programmed adjustment because of the risk of simultaneity bias. However, including it would not change results in most regressions given that BMA only rarely selects it as an effective regressor in explaining programmed adjustments.


Programmed adjustment between t-1 and t+3 when program was approved. While the actual duration of programs varies, it is mostly 2-3 years. Here we approximate adjustment under the program using a uniform three year period in order to achieve comparable time periods of adjustment across countries. Data are taken from the IMF World Economic Outlook vintage closest in time after program approval.


This dummy takes the value of 1 for the following sample programs: Albania (2002, 2006), Armenia (2005, 2008, 2009, 2010), Belarus (2009), Bosnia and Herzegovina (2009), Croatia (2003, 2004), Macedonia FYR (2005), Moldova (2010), Tajikistan (2002, 2009), and Ukraine (2004, 2008, 2010).


Takes the value of 1 if the country has experienced a “war” (as defined by the World Bank Violent Conflict database) within five years prior to program approval.


Takes the value of 1 if a country is in a fragile situation as classified by the World Bank. The World Bank defines a fragile situation as a country having either a harmonized CPIA rating of less than 3.2 or a UN or regional peace-keeping or peace-building mission having been present in the previous three years before program approval.


Defined as (Imports+Exports)/GDP in the year prior to program approval.


Used as a proxy for public sector institutions for PRGT/PSI countries.


Used as a proxy for institutions for GRA countries.


Although some judgment had to be applied for some special cases, capital account crises were identified mainly in a rules-based manner. The narrow capital account crisis definition includes program countries that (i) experienced a fall in net total private capital flows above 1 percent of GDP at time t or t-1 (where Iraq 2010 is excluded on the basis of being a transition program) and (ii) experienced a fall in net portfolio capital inflows above 3 percent at time t or t-1 (where Ireland is taken to satisfy the criterion as a special case, given its large portfolio capital outflows at t-2). It identifies 9 programs: Greece (2010), Hungary (2008), Iceland (2008), Ireland (2010), Jamaica (2010), Latvia (2008), Portugal (2011), and Ukraine (2008, 2010).


The broad capital account crisis definition includes programs that only meet the first criterion in note 10. It thus identifies 16 programs in addition to those identified by the narrow definition. These 16 programs are: Antigua and Barbuda (2010), Armenia (2009), Belarus (2009), Bosnia and Herzegovina (2009), Costa Rica (2009), Dominican Republic (2009), El Salvador (2009, 2010), Gabon (2007), Guatemala (2009), Maldives (2009), Pakistan (2008), Paraguay (2006), Romania (2009), Serbia (2009), and Sri Lanka (2009). Of these Antigua and Barbuda (2010) and Serbia (2009) could not be included in the final estimation sample because of data limitations on other variables.


Takes the value of 1 if a program was approved in September 2008 or thereafter. It is excluded in structural burden regressions given its close collinearity with the dummy for the abolition of structural PCs.


Takes the value of 1 if a program was approved in September 2008 or thereafter, but before September 2009.


Takes the value of 1 if a program was approved in September 2009 or thereafter.


Takes the value of 1 if another program for the country was approved within 1 year of expiry of the original program. It is also 1, if an existing program was cancelled at the time a new


At program approval.


Takes the value of 1 if the HIPC decision point was reached sometime during the program period. Hypothesis is that anticipation of debt relief may have influenced the adjustment contemplated in program design.


Takes the value of 1 if the HIPC completion point was reached sometime during the program period. Hypothesis is that anticipation of debt relief may have influenced the adjustment contemplated in program design.


Takes the value of 1 for programs approved after structural performance criteria were abandoned in March 2009.


Percentage of times the program country voted with the U.S. in the U.N. General assembly during the 3 years prior to program approval on decisions classified as “important” to the U.S. by the U.S. Department of State.


Percentage of times the program country voted with the respective country in the U.N. General assembly during the 3 years prior to program approval. All U.N. General Assembly votes are considered in the calculation.


Fraction of IMF senior management level economists that have the program country’s nationality.


Fraction of IMF non senior management level economists that have the program country’s nationality.


Trade is exports plus imports.

  • The size of fiscal adjustment generally remained sufficient to at least stabilize public debt at sustainable levels. As examined in more detail in BP3, debt dynamics for most PRGT and GRA cases improved under programs to achieve at least debt stabilization. Programs with very high initial debt in the Euro Area and Caribbean were exceptions to this finding, however. (In some cases, e.g., Belarus and Mongolia, EPEs found that the specific measures of fiscal balance used could have been better suited to the country’s policy challenges.)

  • High debt programs involved relatively more fiscal adjustment than other programs, although sustainability concerns may still arise in some cases.20 Regression analysis shows that the magnitude of adjustment is explained to a large extent by the high initial fiscal deficits in high debt GRA cases in year t-1 (Figure 3.2).This pattern is evident, to a lesser extent, for PRGT programs with high debt ratios as well. In both high-debt PRGT and GRA programs, fiscal adjustment, coupled with debt relief under the HIPC and MDRI initiatives in many PRGT cases, was usually programmed to contribute to reductions in debt ratios. However, under the debt projections in one high debt GRA case (Greece), debt ratios remained at high levels, albeit on a downward trend after 2013, even with macroeconomic assumptions that turned out to be more positive than the outturn (BP3 and Appendix IV).

    Figure 3.2.
    Figure 3.2.

    Programmed Adjustment in High Debt Cases for GRA and PRGT Programs

    Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

    Source: WEO, MONA, and Staff calculations.Note: Due to data availability only 46 PRGT programs and 29 GRA programs could be taken into account in establishing the high debt grouping. The very large change in fiscal balances in high debt PRGT countries is driven to a large extent by significant debt relief operations.

  • Programs with less fiscal adjustment seemed to emphasize deeper structural reform agendas. Programs with extensive structural reform agendas adjusted less in the initial program years (Appendix Figure III.1). This pattern implied that program design takes into account political constraints by avoiding both large fiscal adjustment and extensive structural conditionality. However, it is unclear whether this tendency was appropriate in all cases; in certain cases, structural reforms needed to support fiscal consolidation were found lacking (e.g., see EPE for Ukraine).

Figure III.1.
Figure III.1.

Program Objectives and Structural Conditionality

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.

19. The composition of fiscal adjustment generally appeared tailored to country conditions and usually safeguarded priority spending. In PRGT programs with high initial fiscal deficits, adjustment relied partly on expenditure restraint, while in other PRGT cases, fiscal balances and expenditures were stable in GDP terms, with increases in revenue and grants playing a larger role in the initial program years (Figure 3.3 and Appendix III). This approach seemed appropriate given that initial expenditure ratios were not particularly high, and it helped programs meet their objectives of protecting priority social and capital spending. In GRA programs with high initial fiscal deficits, most of the fiscal adjustment fell on spending, as spending levels were relatively high. (The sustainability of this adjustment merits close monitoring since previous experience suggested that revenue enhancement has been important for sustained adjustment in emerging economies.) Both social spending on health and education and capital expenditures were fairly well protected in both program types (see BP3)—capital expenditures actually increased in PRGT programs both as a share of GDP and overall spending (Figure 3.4). The social consequences of expenditure and revenue policies were increasingly taken into account, although there was much room for further work (Boxes 2 and 3; see also BP3).21

Figure 3.3.
Figure 3.3.

Fiscal Adjustment and Initial Conditions

(means in percent of GDP unless otherwise indicated)

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.
Figure 3.4.
Figure 3.4.

Capital Expenditure in GRA and PRGT Programs

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.Note: Due to data availability, only 40 PRGT programs and 19 GRA programs were taken into account for this chart.

20. The timing of fiscal adjustment was often back-loaded, to accommodate concerns over growth in the initial program years.

  • In GRA programs—most of which were initiated following the onset of the global economic crisis—the overall balance was programmed to improve by just 1 percent on average in the first two program years.

  • In PRGT cases, expenditure was planned to increase early in programs, supported by grants and debt relief, and to remain above its initial ratio to GDP in all but the third year after program initiation.

21. Strong outturns compared to targets in early program years indicated that most fiscal adjustment targets were not over-ambitious. Partly driven by faster than expected economic growth, performance against program targets was generally good in the first two program years (see also Appendix III).

  • GRA programs initially adjusted faster than projected and about as much as programmed towards the end of the projection horizon. In contrast, capital account crises missed targets throughout the program horizon.

  • The size of adjustment was an important determinant of performance in GRA cases. Regression analysis suggests that programs requiring larger fiscal adjustment in a given year were less likely to meet that year’s quantitative performance criteria (QPCs).22 Consequently, it may be detrimental to program performance to count on extraordinarily large adjustment efforts.23

  • PRGT programs also adjusted faster than programmed in the initial program years (except for Policy Support Instrument arrangements which adjusted less than initially expected) but subsequently adjusted less than initially programmed.

Flexibility in Fiscal Adjustment

Compared to the 2002-05 sample, GRA programs in 2006-11 planned more fiscal adjustment in the face of higher deficits and inflation rates (see Appendix I for details). Fiscal deficits in the first year of programs were substantially higher on average in the 2006-11 sample than during 2002-05 (Box Figure 1). Programmed fiscal adjustment thus had to be larger. Initial inflation rates were also higher (in year t), but were projected to fall by a similar magnitude over the program horizon as in 2002-05, thus targeting slightly higher rates in the medium term.

Box Figure 1.
Box Figure 1.

Programmed Adjustment in GRA Programs: 2006-11 vs. 2002-05

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.

But adjustment was more back loaded in current GRA programs than the earlier period, in light of the larger growth declines during the global crisis. Both fiscal and inflation adjustment in the initial program periods were smaller in the 2006-11 sample in spite of larger initial deficits and falling revenues (Box Figure 2). This back loaded adjustment was especially pronounced in capital account crises which only adjusted moderately after two years of strongly expansionary policy.

Box Figure 2.
Box Figure 2.

Composition of Programmed Fiscal Adjustment Between t-1 and t+1

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.

PRGT programs in the 2006-11 targeted less adjustment and higher deficits and inflation (Box Figure 3). Programs faced lower initial fiscal deficits in the 2006-11 programs, and adjustment was significantly less than before. In the initial periods, fiscal adjustment relied solely on revenue and grants, while expenditure remained expansionary, allowing for counter-cyclical policies (Box Figure 2). Monetary policy targeted inflation rates of about 4 percent in 2006-11 programs, slightly higher than in 2002-05.

Box Figure 3.
Box Figure 3.

Programmed Adjustment in PRGT Programs: 2006-11 vs. 2002-05

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.

Conditionality on Expenditure Measures and Protecting the Poor

Expenditure policy measures potentially affecting the poor focused mainly on price increases. The most common spending measures required an adjustment of utility tariffs or increases of domestic petroleum prices. These were deemed necessary to avoid excessive volatility of taxes and revenues, curtail untargeted subsidies, eliminate quasi-fiscal deficits, and put loss-making public enterprises on a sounder footing. Occasionally, price increases were preceded or accompanied by automatic price adjustment mechanisms that would smooth out periodic adjustments, dampen price volatility, and depoliticize price increases.

Between 2006 and 2010, nine of the 18 countries in the case study sample were subject to program conditionality affecting prices of products consumed by the poor. While the Uganda (2006), Pakistan (2006), and Ghana (2009) programs predicated specific tariff increases, Moldova (2006 and 2010), Sierra Leone (2006), The Gambia (2007), Togo (2008), and the Dominican Republic (2009) implemented measures related to a utilities tariff adjustment mechanism or the definition of institutional responsibilities for price setting. The Gambia, Ghana, and Togo programs included conditionality related to automatic fuel pricing.

In five out of these nine countries, price adjustment conditionality was accompanied by an analysis of their distributive impact. Technical assistance (TA) explicitly considered adverse impacts of such measures on the poor and how these could be mitigated. Since 2002, the IMF’s Fiscal Affairs Department carried out 15 missions, including in Ghana and Moldova. In the Gambia, Sierra Leone, and Togo, TA focused on issues related to the design and implementation of a price smoothing mechanism where explicit account was taken of its effects on the poor, and recommendations for offsetting measures were provided.

Where analysis of distributive impacts of measures was not available, Fund staff sought support from the World Bank and included conditionality aimed at strengthening social protection. In cases where distributive impact analysis or other relevant TA was not carried out, conditionality in the area of price increases was generally accompanied by a benchmark requiring the expansion of existing social safety nets and social benefits to the poor. In the case of Pakistan, this occurred with the help of the World Bank. The Uganda program did not include price increases as explicit conditionality; rather the authorities expressed their desire to increase electricity tariffs in the Memorandum of Economic and Financial Policies. This was coupled with a floor on expenditures from the Poverty Action Fund as an indicative target in the program.

Expenditure Conditionality and the Poor

article image
Sources: SPR MONA database and FAD.

Measures include for example, introducing targeted social assistance system; strengthening targeting of social safety nets; increasing coverage of conditional cash transfer programs, etc.

Conditionality on Tax Policy Measures and Equity Considerations

Although technical assistance (TA) often provides advice on the equity dimension of tax policy reforms, equity considerations appeared to be less prominent in program conditionality on tax policy. Between 2006 and 2010, the number of tax policy conditions in Fund programs increased tenfold to reach contained 102. Motivated by a desire to enhance revenue while minimal distortions or negative effects on growth, these tax policy conditions tend to focus on reforms concerning a VAT or other indirect taxes, as well as streamlining tax expenditures related to incentives and exemptions for businesses (Box Table 1).

Box Table 1.

Policy Conditionality for All Program Countries, 2006-2010

article image

While considering the equity dimension of tax policy reforms, TA typically advises that equity objectives are best addressed through the use of targeted expenditure policies rather tax policy. This allows tax policy to focus on efficient revenue raising to finance these expenditures. Furthermore, it is not the case that efficient tax policy inevitably involves compromising equity objectives. The assumption that VAT is a regressive revenue raiser is subject to controversy. Eliminating tax expenditures to broaden tax bases and improving the efficiency of tax administration improve horizontal equity by leveling the playing field between different sectors and taxpayers.

Program conditionality on tax policy appeared to give less prominence to equity considerations than TA. A review of the seven countries in the sample with tax policy conditions highlights how tax policy structural benchmarks center on introducing modern value-added tax (VAT) systems and broadening the tax base through the elimination of incentives, exemptions, and other forms of tax expenditures. With the exception of Seychelles (and to a certain extent Greece), the equity implications of tax policy structural benchmarks were not explicitly addressed in the design of tax measures. In the Seychelles, a flat-rate personal income tax (PIT) on salaries, interest and dividend income was extended untaxed non residents and expatriate workers, while reducing the PIT rate previously applied to resident wage earners. In Greece, for example, revenue-raising has targeted higher-income segments, with temporary surcharges on highly profitable entities, the presumptive taxing of professionals and increases in luxury goods taxes. Tax administration reforms there were also crucial to ensuring effective revenue effort and improving fairness. Moreover, some other programs contain explicit expenditure measures for protecting low-income households from the adverse impacts of macroeconomic adjustment (e.g., Grenada and Pakistan).

In the future, program design could benefit from offering to the authorities to analyze more directly the impact of tax policy measures on low-income households or the progressivity of revenue collection. Structural conditionality could explicitly refer to certain essential items for exemption from a VAT or to tax-free thresholds for both direct and indirect taxes. In addition, including well-designed natural resource and property tax reforms as program measures would allow developing countries to increase revenue collections without adversely affecting the equity of the tax system or impairing growth.

B. Monetary Adjustment

22. Monetary policy adjustment in Fund-supported programs during 2006-11 generally appeared well adapted to country needs and program characteristics.

  • The magnitude of monetary policy adjustment can be well explained by relevant initial macroeconomic conditions, particularly initial inflation (Table 3.2). Programs with higher initial inflation consistently involved larger inflation reduction for both GRA and PRGT programs. Regression analysis also indicates that programs in currency unions targeted more initial inflation reduction. Crisis programs involved less inflation reduction in light of already subdued price pressures.

    Table 3.2.

    Regression Results Explaining Programmed Inflation Reduction

    article image
    Source: IMF staff estimations.Notes: All variables are statistically significant at the 10 percent level or higher. See Appendix Tables I.1-3 for descriptions.

  • Programs targeted broadly reasonable inflation levels. For both GRA and PRGT cases, countries with lower initial inflation generally targeted around 5 percent, while those with higher initial inflation on average targeted slightly higher rates. These levels are consistent with the literature on optimal inflation levels for middle and low income countries (e.g., Habermeier et al. (2009) and Berg et al. (2010), respectively).24

  • In high inflation cases, PRGT programs relied on initially tight monetary and fiscal policy to rein in inflation, while GRA programs leaned more on monetary policy. For PRGT cases, concerns over output losses were evident in regression results that programs during the global crisis targeted less inflation adjustment.

  • The timing of targeted disinflation was generally fairly gradual, except for high inflation cases. For high inflation cases among GRA programs, disinflation involved a fairly sharp reduction in money growth in year t+1 and an initially contractionary fiscal policy, whereas in PRGT programs with high inflation, adjustment in year t was programmed to lead to sharp falls in inflation (Figure 3.5).

Figure 3.5.
Figure 3.5.

Monetary Adjustment and Initial Conditions

(means in percent of GDP unless otherwise noted)

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.Note: Programs are grouped according to initial conditions. Programs with a high fiscal deficit, a high current account deficit or high inflation are those in the upper quartile of the sample distribution with respect to each measure.

23. Monetary policy conditionality evolved during the review period to adapt to countries’ characteristics and policy preferences. An increasing number of (primarily GRA) program countries implemented inflation targeting, to which conditionality adapted in flexible and innovative ways (Box 4). Structural conditionality in monetary policy often focused on clarifying the relationship between policy objectives and tools (Appendix Table II.1).

Table II.1.

Number of Monetary Policy Reforms in GRA and PRGT Programs (based on a review of the case studies)

article image
Source: IMF Staff estimations.

24. In PRGT countries, monetary adjustment played a somewhat limited role in macroeconomic adjustment policies. This pattern tended to reflect a mixture of several factors: weak monetary policy transmission mechanisms, inefficient monetary policy frameworks, lack of key pre-requisites (including infrastructure) for market operations, and lack of competition in money and foreign exchange markets. These limitations, the evolution in financial intermediation, and other structural changes lead to instability in money demand (IMF Sub-Saharan Africa REO April 2008). Nevertheless, in light of the weaknesses of other monetary policy anchors, many programs continued targeting the monetary base as a key anchor (Figure 3.6).

Figure 3.6.
Figure 3.6.

Monetary Policy Implementation in PRGT Programs

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.

25. Inflation outturns came fairly close to targets for both GRA and PRGT cases in the first program year, indicating that these targets were not unreasonable. Money growth diverged more often from program targets, however; implementation of a “pure” monetary targeting regime has proven increasingly complex, given instability in money demand and the increasing incidence of price and capital flow shocks in recent years.

  • Both GRA and PRGT programs initially performed fairly well relative to inflation targets, especially during the first year following the beginning of a program. Despite diverging later from initial program targets, average inflation still fell to around 6 percent for GRA cases and 7 percent in PRGT cases (roughly 1½ to 2 percent above initial targets). Higher than programmed inflation outcomes for PRGT programs could be partly due inflationary impact of the surge in commodity and fuel prices.

  • In GRA capital account crisis cases, money growth slowed more than targeted, as sudden stops, capital outflows, and increased risks in financial markets had a higher impact than projected.

  • In GRA programs both fixed and flexible exchange rate regimes seemed to be successful in controlling inflation and supporting macroeconomic adjustment.

Program Conditionality, Monetary Policy Frameworks, and Inflation Targeting

Program conditionality has been adapting to the growing diversity in monetary policy frameworks, seen most notably in cases of inflation targeting. The number of program countries with such a framework increased from two to 10 compared to the 2005 RoC, including one PRGT case (Ghana). This movement added to the diversity of monetary policy frameworks in Fund-supported programs. In 2006-11, about 60 percent of program countries had flexible exchange rate regimes, necessitating a monetary policy anchor. In programs without inflation-targeting, conditionality specified a ceiling on net domestic assets and a floor on net international reserves, either as performance criteria or indicative targets. In inflation targeting regimes, conditionality has shifted to a combination of consultations in the event of deviation from targeted inflation and reserves targets (Appendix II).

Programs incorporating inflation-targeting have sought to strengthen the monetary policy foundation. Structural conditionality and technical assistance have stressed: an independent central bank with a mandate to follow price stability as the main goal of the monetary policy; accountability for inflation performance; a “forward-looking assessment of inflation pressures” to inform and communicate policy actions; and finally, transparent and well communicated monetary policy strategies and implementation. Countries adhering to inflation targeting framework seemed to perform better in monetary and inflation adjustment (Box Figure 1). There is more experience, however, of such a framework in successfully reducing inflation from moderate levels than from high levels.

Box Figure 1.
Box Figure 1.

Money Growth and Inflation under Inflation Targeting

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.

C. External Adjustment

26. Adjustments in the external sector appeared well designed to support country needs and objectives.

  • The magnitude of external sector adjustment provided for sustainability and buffers while seeking to minimize output losses (Figures 3.1 and 3.7). GRA programs targeted modest and gradual improvements in both the current account balance and reserves, reflecting the objectives of restoring external sustainability and rebuilding buffers. PRGT programs targeted stability in the current account and reserve levels over three years, with modest deterioration in the initial years, reflecting a greater preoccupation with supporting output. BP3 found that the adjustment in current account balances in most program countries was sufficient to reduce external debt or stabilize it at sustainable levels; however, in a few cases, external debt was projected to remain at high levels.

    Figure 3.7.
    Figure 3.7.

    Planned and Actual Reserve Coverage Adjustment

    Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

    Source: WEO, MONA, and Staff calculations.

  • Generally, the initial conditions that explained the magnitude of external adjustment were reasonable (Tables 3.3 and 3.4).25

    Table 3.3.

    Regression Results Explaining Programmed Current Account Adjustment

    article image
    Source: IMF staff estimations.Notes: All variables are statistically significant at the 10 percent level or higher. See Appendix Tables I.1-3
    Table 3.4.

    Regression Results Explaining Programmed Reserves Adjustment

    article image
    Source: IMF staff estimations.Notes: All variables are statistically significant at the 10 percent level or higher. See Appendix Tables I.1-3 for descriptions.
    • ○ For the current account, a lower initial or declining balance led to larger adjustment for both GRA and PRGT countries, explaining the bulk of variation among programs. PRGT programs with high external debt levels also targeted more current account adjustment.

    • ○ For international reserves, a capital account crisis led to higher programmed accumulation in GRA countries, while reaching the HIPC completion point led to higher programmed accumulation in PRGT countries.26

27. The composition of external adjustment between demand management and exchange rate tools reflected both initial conditions and national policy priorities. This combination was intended to promote ownership. Adjustment in PRGT programs with a high initial current account deficit tended to rely mainly on tighter monetary policy, while in GRA cases, it relied on a mix of fiscal and monetary policies. Many program and non-program countries increased exchange rate flexibility during the turbulence of the global financial crisis, with low initial reserve levels playing a large role in these decisions. However, there was no evidence that having a program affected that decision (Box 5). Discussions of program design in the Latvia and Iceland programs illustrated how a flexible, pragmatic approach—in which different policy options to achieve sustainability were developed for the authorities to consider—can strengthen ownership.

28. Considering timing and outturns, planned external adjustment appeared reasonably ambitious.

  • GRA programs envisaged smooth current account adjustment. The outturns were broadly consistent with the projections. Reserve accumulation was also gradual but exceeded targets throughout programs. GRA programs facing high current account deficits tended to delay inflation adjustment (in many cases due a weakening exchange rate).

  • PRGT programs largely exceeded targets for planned reserve accumulation, with slight underperformance in current account adjustment later in the program. The outturn of reserve accumulation under PRGT programs exceeded programmed levels on average and in specific cases, including high initial current account balance and inflation—this could be partly explained by the impact of SDR allocation on LICs. The magnitude of the allocation was significant for LICs, and about half of all LIC programs did not envisage a need to spend and absorb it, leading to reserve accumulation (for details see BP4).

Exchange Rate Flexibility and Fund-Supported Programs

During 2006-11, countries—both with and without programs—increasingly moved toward more flexible exchange rate regimes (Box Table 1). Among program countries:

Box Table 1.

Factors in Moving to More Flexible Exchange Rate Regimes, 2006-2011

article image
  • About half considered exchange rate policies as a strategy for achieving program objectives.

  • About a quarter moved to more flexible arrangements, notably during the global crisis.

  • There was explicit conditionality related to exchange rate policies and restrictions in 16 programs (8 GRA and 8 PRGT) launched during 2006-11 (including removing multiple currency practices, abolishing market rationing and removal of current account restrictions).

  • Countries with currency boards (both with and without programs) did not change their exchange rate regimes.

  • Regression analysis concluded that program countries have not moved to more flexible exchange rate regimes more often than their non-program peers (Appendix III Table III.2). This result did not support a commonly held view that Fund-supported programs influenced countries’ decisions to move toward more flexible exchange rate regimes.

    Table III.2.

    Probit Results for Fund Effects on Change in an Exchange Rate Regime

    article image
    Note: Dependent Variable is 1 denoting a change in the exchange rate regime from t-1 to tNote: A Hard Peg is defined solely as those exchange rate regimes classified as a currency board or no separate legal tender; A soft peg is defined as those exchange rate regimes classified as crawling peg, crawling band or a peg

  • The shift towards greater exchange rate flexibility under programs was supported by reforms in the institutional framework and market infrastructure, including rules-based and transparent foreign exchange intervention policies; enhancing central bank liquidity management; and stronger supervision of commercial banks’ risk management.

The initial level of reserves and competing macroeconomic objectives appeared to be important factors in a country’s decision to increase exchange rate flexibility (Box Table 1). (This analysis addresses program documents that had discussions on exchange rate policies.) In countries with high international reserves, program design tended to envisage a drawdown in reserves to limit exchange rate movements (in both fixed and flexible exchange rate regimes), with the goals of dampening inflationary pressures, calming market sentiment, and preserving financial stability. However, in countries with fixed exchange rate regimes and low reserves, programs envisioned a move towards more flexibility, particularly when multiple macroeconomic and monetary policy objectives were becoming increasingly incompatible. Countries with floating exchange rates and low reserves relied more heavily on exchange rate movements in order to avoid reserve losses and absorb large exogenous shocks.

Box Table 2.

Evolution of Exchange Rate Regimes

article image
Source: AEEAER database 2011.Note: Data for 2011 based on Jan-Sept.

D. Access to Fund Financing27

29. Access levels under GRA programs can largely be explained by macroeconomic variables and membership in the Euro Area (Table 3.5). More specifically, financing needs, program strength, and capacity to repay the Fund are also key determinants of access, but since consistent cross-country data on those are not available and they may be endogenous to program design, they were not included in the set of candidate regressors. An initially higher current account balance was associated with lower access levels, while programs with a capital account crisis had higher access. More Fund credit outstanding was associated with higher access, but immediate successor programs were not.28 Membership in the Euro Area has a substantial and statistically significant explanatory power for variations in access levels, as already discussed in Section II on evenhandedness. The 2009 change in the Fund policy that raised access limits appeared crucial to satisfying higher financing needs during the crisis. The regression results suggest that access was higher in wave 1 (i.e., crisis programs before August 2009) and Euro Area programs, but not for non-Euro Area wave 2 programs. Thus, if the earlier access limits had remained in place, they would arguably have constrained access for peak-of-the-crisis programs to less than that warranted by macroeconomic conditions (see BP3 for further analysis).

Table 3.5.

Regression Results Explaining Access Levels

article image
Source: IMF staff estimations.Notes: All variables are statistically significant at the 10 percent level or higher. See Appendix Tables I.1-3 for descriptions.

30. Access under PRGT programs was fairly well explained by macroeconomic and structural variables. The initial conditions for growth, international liabilities, and cross-border bank claims helped explain access levels. Successor programs had less access, as expected, while countries in transition received higher access. Programs approved from 2006 onwards received higher access after controlling for other factors. This is likely due both to higher crisis-related financing needs and the revision of PRGF access guidelines in 2009.

31. Access under GRA programs was tailored to projected adjustment, while it was evenly distributed in PRGT programs (Figure 3.8). GRA programs with higher fiscal adjustment (linked to high initial fiscal deficits or high public debt) on average received higher access. In PRGT programs, access was generally fairly evenly distributed across programs facing different challenges, reflecting a consistent application of access norms for PRGT cases. Also, the more even distribution of access in PRGT programs reflects their longer-term growth and poverty reduction objectives to a degree, in contrast to the strong medium-term stabilization objective of GRA programs.

Figure 3.8.
Figure 3.8.

Access and Adjustment

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source :WEO, MONA, and Staff calculations.Note: Access denotes average access per quota; fiscal (current account) adjustment denotes fiscal (current account) balance in t+3 minus fiscal balance in t-1; monetary adjustment denotes broad money growth in t-1 minus broad money growth in t+3. The sub-samples of countries in the county groupings in this chart differ slightly from those in the remainder of the paper as PSIs have been drapped forth is exercise (dueto zero access)

E. Structural Reforms

32. Structural reforms have appeared reasonably well tailored to country needs and planned macroeconomic adjustment (see also BP1). This section reviews the determinants of the number of non-financial structural conditions in programs and examines their links to planned macroeconomic adjustments (the following section concentrates on financial sector reforms).

33. The variations in the number of structural conditions can be best explained by ownership and the onset of the global financial crisis (Table 3.6). Ownership can be proxied by the number of prior actions at Board approval of the program, as in Wei and Zhang (2005). The falling number of structural conditions following 2006 can be explained by a combination of Fund policy shifts and increased focus on near-term stability following the global financial crisis (interestingly, the specific policy measures of the abolition of structural PCs and the move to review-based conditionality in 2009 did not appear to have a statistically significant effect on the number of conditions in these regressions). In addition, the only other significant explanatory variable for PRGT cases was a higher current account balance, which was associated with an increase in the extent of structural conditionality— presumably, less urgency for macroeconomic adjustment left more policy space for structural reform. As examined in BP1, the areas targeted by structural conditions were increasingly linked to core macroeconomic objectives, such as growth and public financial management.

Table 3.6.

Regression Results Explaining Programmed Structural Burden

article image
Source: IMF staff estimations.Notes: All variables are statistically significant at the 10 percent level or higher. See Appendix Tables I.1-3 for descriptions.

F. Financial Sector Reforms29

34. The design of financial sector conditionality appeared consistent with the Fund’s core mandate and tailored to evolving country needs. In keeping with the Fund’s mandate, financial sector reforms have played an important role in structural conditionality, especially in programs addressing capital account crises (Figure 3.9). However, rising non-performing loan ratios in GRA cases raise concerns that banking sector weaknesses may not have been fully addressed (Figure 3.10), even though banking crises were avoided in most program countries. The integration of macro-financial linkages posed heightened risks in the crisis period that will require continuing efforts to integrate them fully and systematically into program design (Box 6).

Figure 3.9.
Figure 3.9.

Financial Sector Conditionality in GRA Programs

(percent of total structural conditions in initial program year)

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: WEO, MONA, and Staff calculations.
Figure 3.10.
Figure 3.10.

Non-Performing Loans

(in percent of total loans, bands represent 25th and 75th percentiles of the distribution)

Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

Source: Crisis Program Review 2012.
  • Program conditionality in the financial sector focused on areas related to the Fund’s core mandate to support financial sector soundness. Nearly all of financial sector conditionality relate directly to this mandate, together accounting for at least two-thirds of total conditionality (Appendix Figure III.3). The next largest category, financial infrastructure and market development, focused on contributions to growth and supported soundness.

    Figure III.3.
    Figure III.3.

    Classification of Financial Sector Conditionality

    Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

    Source: WEO, MONA, and Staff calculations.

  • The relative emphasis of conditionality was reasonably tailored to different types of programs. In particular, the share of infrastructure and market development was larger in PRGT programs, consistent with their emphasis on medium-term growth. Bank restructuring, privatization, and resolution figured prominently in transition economy and capital account crises programs.

  • Financial sector conditionality adapted rapidly to the global financial crisis (Figure 3.11). When the global crisis hit, the focus of reform in GRA cases initially shifted to curbing liquidity pressures and financial stress.30 Subsequently, reforms included measures to address solvency concerns and balance sheet weaknesses and then bank resolution and intervention. A number of program countries also appeared to benefit from reform progress made prior to the crisis in strengthening supervision and prudential regulation, addressing systemic risks, and reducing balance sheet vulnerabilities—an observation that merits more rigorous analysis (Review of Recent Crisis Programs). Nevertheless, in at least two cases (Benin and Georgia), EPAs suggested that program conditionality could have been more focused on financial sector soundness (although in the former case, it was explained by the fact that supervision is the responsibility of a regional institution and a Financial Sector Assessment Program was planned).

    Figure 3.11.
    Figure 3.11.

    Classification of Financial Sector Conditionality

    Citation: Policy Papers 2012, 043; 10.5089/9781498340403.007.A001

    Source: WEO, MONA, and Staff calculations.

  • In PRGT countries, the emphasis shifted more to assessing financial stability and building capacity for crisis contingency and less to bank restructuring, reflecting the weaker financial sector transmission channel during the crisis.

  • Programs adapted to country-specific vulnerabilities, including pre-existing insolvency of key banks (e.g., Iceland, Latvia, and Ukraine), corporate sector distress (e.g., Georgia, Latvia, and Pakistan), and private debt restructuring (e.g., Iceland, Latvia, and Serbia).

35. Efforts to address concerns over maintaining cross-border inter- and intra-bank financing met with mixed success. The high incidence of foreign ownership of domestic banks in some program cases was initially a cause for concern, given that parent banks were under pressure. For certain programs in Europe, this issue was addressed by way of the European Bank Coordination Initiative (or “Vienna Initiative”) (Box 7). In other programs, pre-conditions for similar initiatives were not as propitious and financing pressures often did not appear as acute. Nevertheless, it would be useful to explore the possibility, during surveillance, of contingency planning and preparatory work for similar initiatives in other regions—much as the Vienna Initiative has continued as a forum for discussions in Europe.

G. Macroeconomic Policy Mix and Financing/Adjustment Balance

36. Program design used policy mixes and financing/adjustment balances tailored to particular country challenges, although some areas for improvement stand out. Based on the preceding analysis, these policy mixes and balances generally made economic sense, particularly from the point of view of achieving stabilization and sustainability while minimizing the cost in terms of output losses.31 This preoccupation was especially appropriate during the immediate aftermath of the global financial crisis. For example, program design has proven flexible in providing direct budget support where needed during this period, while maintaining Fund lending principles (Box 8). The following assessment focuses on the policy mixes and financing/adjustment balances for programs with imbalances in the top quartile of all programs. The analysis suggests that recent program design has to a large extent drawn on lessons from past mistakes (e.g., capital account crises). But certain challenges and risks in the specific cases of capital account crises and perhaps fragile countries (examined further below) may not yet be managed as well as possible (Figure 3.8).

  • Adjustment for a high fiscal deficit. The typically gradual fiscal adjustment and accompanying accommodative monetary policy appropriately sought to mitigate output losses. Fiscal adjustment in high fiscal deficit programs was larger than in the entire program sample and on average sufficient to reach a stable path of public debt. Programmed adjustment appeared realistic and attainable on average, which contributed to strong ownership and implementation. At the same time, monetary policy was supportive of output, with money growth slowing only marginally in PRGT cases and with expansionary policy in GRA programs. GRA cases with high initial fiscal deficits or debt burdens also received higher access, which was appropriate given higher financing needs. However, more recent GRA programs with very high public debt loads have highlighted the challenges for this type of program, including the vulnerabilities stemming from high debt and the nexus of fiscal adjustment and growth (see BP3). One lesson for program design is that higher risk cases require additional, more in-depth analysis.

  • Adjustment for high inflation. Programs with higher initial inflation successfully achieved disinflation while mitigating output costs. PRGT programs targeted both slower disinflation and sharper monetary and fiscal adjustment, compared to GRA cases—likely because inflation in the former cases responded less rapidly to adjustment. GRA programs relied on a more gradual monetary tightening; fiscal policy was not supportive initially, perhaps reflecting a missed opportunity to coordinate policies. Access was somewhat below average in GRA cases and near average in PRGT cases, while monetary adjustment was well above average in both cases, as one would expect. Outturns for these high inflation cases in both program types missed targets by about 3-5 percentage points on average in the two to three years after the program initiation—which nevertheless represented significant disinflation.

  • Adjustment for a high current account deficit. Adjustment to initially high current account deficits achieved its goal of restoring sustainability with a generally appropriate gradual approach. In PRGT programs, adjustment relied on a mix of initially tighter fiscal and monetary policy, while adjustment was limited and gradual in GRA cases. In PRGT programs with high current account deficits, actual adjustment exceeded programmed targets; it may be worthwhile in retrospect to try to draw lessons from why projections were too pessimistic. Access was average for both GRA and PRGT cases with high current account deficits; while this finding might run counter to expectations, the feasibility and success of adjustment was reassuring. Decisions on whether to increase exchange rate flexibility supported adjustment in appropriate ways, reflecting initial conditions (especially reserve levels) and country policy preferences. This approach supported ownership and implementation.

Direct Budget Support in Recent Fund-Supported Programs

While a long-standing practice, the incidence of direct budget support has increased in recent years. Direct budget support occurs when a disbursement goes directly, at least in part, to the government’s fiscal authority. Direct budget support can be appropriate and consistent with the Fund’s mandate and legal framework provided a balance-of-payments (BoP) need exists (Staff Guidance Note on the Use of Fund Resources for Budget Support). Analytically, fiscal needs are seldom disjointed from BoP needs, and Fund financial support has often addressed both.

From 2006 onwards, 36 percent of Fund-supported programs had some direct budget support, driven by needs related to the global crisis and institutional changes in member countries. The increase was particularly notable for GRA programs. This increase reflected both fiscal financing needs to respond to the impact of the global crisis, including more recently sovereign debt crises (and an accompanying loss of access to private external financing) and institutional changes in member countries, especially the move to greater central bank independence. All of these programs have been consistent with the requirement of a balance of payments need and circumstances identified as warranting direct budget support—fiscal financing needs and institutional constraints. Based on econometric analysis, these programs have not differed greatly from others in economic design, controlling for other factors (Appendix Table I.8). The main difference observed was that these programs tended to have somewhat larger fiscal adjustment.

Appendix Table I.8.

Regression Evaluating Impact of Budget Support in GRA Programs

article image
Source: IMF staff estimations.Notes: ***, **, * denote 1, 5, and 10 percent significance levels, respectively.

Given the large uncertainty surrounding the exact model specification, the iterative Bayesian Model Averaging (BMA) methodology was applied to a larger 2002-11 program sample to obtain the baseline specifications (see Review of Conditionality Background Paper 2 on Program Design). The BMA methodology requires a set of candidate regressors, which here is constituted by all the variables listed. BMA then selects the best model by using a selection criterion that trades off parsimony and goodness of fit. The estimates presented here then result from an OLS estimation of this best model.


The regressor is excluded from the set of candidate regressors used by BMA to determine the best model.