Chapter 9: Pakistan

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

1. Pakistan is one of the most prolonged users of IMF resources and has been under IMF-supported programs almost continuously since the late 1980s.

2. This report aims to cast light on what successive IMF-supported programs achieved and failed to deliver and on the factors underlying their limited success.1 In particular, two such factors, which appear to have been critical in the Pakistan case, are analyzed in depth, namely (i) program design and implementation problems; and (ii) internal IMF governance issues affecting the rationale for IMF involvement, the effectiveness of surveillance, and the program design itself. The report concludes by highlighting a few key lessons from this experience and outlining suggested remedies.2

3. This evaluation was conducted based on (i) reviews of IMF staff reports and internal documents (including mission briefs, internal review comments, and selected technical assistance reports); (ii) inter-views with IMF staff, a broad range of Pakistani stakeholders, and staff of the World Bank;3 (iii) a survey of relevant academic literature; and (iv) independent work commissioned by the IEO from the Center for Development Research at the University of Bonn (Appendix 1).

Pakistan’s Prolonged UFR Experience Points to a Limited Effectiveness of Its IMF-Supported Programs

4. Pakistan’s economic history over the last 30 years can be subdivided in two periods. From 1970 to the late 1980s, Pakistan enjoyed an impressive growth performance (6–7 percent a year on average). Fiscal and external imbalances were large during most of that period, but unlike in many other developing countries, they did not lead to hyperinflation or to a debt crisis, which led Pakistan to be sometimes referred to as a “development puzzle.”4 However, the picture deteriorated markedly from the late 1980s onward, as growth faltered and the continued failure to rein in the fiscal and current account deficits led the debt—which had been accumulating for over two decades—to become unsustainable. Pakistan made an intensive use of IMF resources during both periods, but became continuously dependent upon IMF-supported programs only in the second one.

Overview of Pakistan’s history of use of IMF resources since 19705

1970/71–1987/88: repeated but discontinuous use of IMF resources

5. Pakistan had four one-year Stand-By Arrangements6 (SBAs) with the IMF between 1972 and 1977. They were followed by a three-year extended arrangement in 1980, which was to provide close to SDR 1.3 billion (445 percent of quota) over three years. The programs supported by these arrangements were classic macroeconomic stabilization programs, which put little emphasis on structural reforms (although the program supported by the 1980 EFF did consider reforms in the areas of taxation, tariff reform, and price liberalization).

6. All SBAs except the first one were entirely disbursed. However, they did not succeed in correcting durably the underlying imbalances. As the dollar (to which the Pakistani rupee was pegged) began appreciating in 1979, pressures on competitiveness increased, and a new recourse to IMF resources proved necessary. In spite of strong policy implementation in the first year of the EFF-supported program, in 1981 devaluation could not be avoided. Thereafter, slippages in the fiscal and monetary area grew larger and the pace of reform slowed, so that after several delays in the completion of reviews, the program was eventually declared off-track.

1988–2000: almost continuous IMF arrangements

7. Pakistan had seven different arrangements with the IMF over the 1988–2001 period,7 of which four were short term and three were multiyear arrangements. All put a strong emphasis on restrictive demand management policies and a variety of structural reforms to correct financial imbalances. The total amount of funds committed under these programs amounted to over SDR 4 billion.8 All but the last arrangement (the 2000 SBA) suffered from substantial policy slippages and soon went off-track, usually after the first or second review. As a result, a large share of the committed financing was not disbursed: undrawn balances averaged a little over half the committed amounts, compared to one-third for all users of IMF resources and a quarter for prolonged users.9 This suggests a rather poor implementation record overall.

Economic performance over 1988–2000 was unimpressive compared with previous decades

Macroeconomic performance deteriorated and financial imbalances largely persisted

8. GDP growth fell to a little under 4 percent a year over 1988–2000, compared to almost 6 percent in the two previous decades, with a sharp slowdown in capital formation. Export growth slowed to under 3 percent a year, against over 10 percent in the 1970s and 1980s. Poverty rose steadily, according to most measures, after two decades of decline, and by the end of the 1990s close to 30 percent of the population lived below the poverty line, against less than 20 percent a decade earlier.

9. Only a modest correction of financial imbalances was achieved over the period. Inflation was halved to 4 percent by 2000, but the fiscal deficit declined only from 7.7 percent of GDP in 1989 to 5.2 percent in 2000. The current account deficit fluctuated around 4 percent of GDP. Gross international reserves fell as a share of imports over the period to under one month in June 2000, the symbolic threshold of three-month import coverage was reached only once, and the coverage of effective short-term debt by official reserves averaged just over 15 percent over 1991–2000.

Structural reforms progressed in some areas but significant challenges remain

10. Significant progress was achieved relatively early on in the areas of interest rate liberalization and public debt management; liberalization of external transactions, both on the current and the capital account; and trade liberalization: tariffs were brought down sharply,10 their structure was simplified, and quantitative restrictions on both imports and exports were substantially reduced. Nevertheless, Pakistan’s trade regime remains relatively restrictive in comparison with most Asian economies.

11. In other areas, such as the implementation of a broad-based general sales tax (GST), taxation of the agricultural sector, liberalization of administered prices, and the setting of utilities tariffs, the reform process was very protracted. Progress was achieved many years later than initially intended, and most of these reforms have yet to be fully effective. By contrast, very little was achieved in the areas of tax administration or income tax reform, and public enterprises were still a considerable drain on the government’s finances in 2000.

12. One indicator of limited progress in bringing about core structural changes needed for longer-term sustainability is the evolution of the tax revenue to GDP ratio: in spite of all attempts to increase it under repeated IMF-supported programs, the ratio was lower in 2000 than in 1988 (12.1 percent against 13.5 percent). This decrease in the overall tax ratio reflects a shift in the structure of taxation away from international trade taxes, which ceteris paribus should have increased tax efficiency. But these possible gains must be weighed against the efficiency losses induced by the de facto very narrow base of domestic taxes, which fall on a very small number of taxpayers11 despite significant steps taken to broaden the legal definition of the tax base.

Economic institutions do not seem to have benefited from prolonged UFR

13. In Pakistan, the 1990s has been characterized as an era of “institutional decay.”12 This was manifested through increased political interferences in the management of public enterprises and in the operations of the predominantly public banking sector, more widespread corruption in tax administration, and increased clout of vested interests in all areas of public policymaking. Meanwhile, statistics and public accounts remained of very poor quality, as did technical skills at all but the highest level of public administration. The greater independence gradually granted to the central bank (State Bank of Pakistan) is one exception to that general trend.13

14. These problems obviously had other, deeper causes that were not directly associated with IMF-supported programs, but they proliferated in spite of these programs, which has prompted many in Pakistan to blame the IMF for failing to tackle governance issues directly. Addressing governance issues did not explicitly become part of the agenda of IMF-supported programs until 1997, and those negotiated with Pakistan were no exception in that respect. Within that context, however, the issue of institutional effectiveness could have received more emphasis than it did, both in the design of programs and in the policy dialogue with the authorities in the context of surveillance and technical assistance.

15. Moreover, many officials in Pakistan were of the view that protracted UFR has weakened the independent policy formulation capacity, in part because the policy closure process revolved around negotiations with the IMF, with little room for domestic initiative and open discussion of policy alternatives.14 Most officials also viewed as limited the capacity building effect on economic management skills sometimes assumed to be associated with IMF-supported programs, because this “skills transfer” was largely concentrated in narrow areas geared to the technical implementation and monitoring of IMF-supported programs. Likewise, the unusually large amount of resources invested by the IMF in technical assistance to Pakistan produced a considerable amount of blueprints for reform.15 But, according to the authorities, they failed to have a significant impact because the knowledge transfer involved in their conception was limited, and they were not sufficiently focused on implementation.16

The Lack of Effectiveness of Pakistan’s IMF-Supported Programs Stems from Both Design and Implementation Issues

16. At first glance, it is tempting to blame the lack of effectiveness of IMF-supported programs in Pakistan on their poor implementation by the authorities.17 The prolonged political instability and regional disruptions undoubtedly weakened policies and the ability of the economy to respond to adjustment initiatives. To the extent that policy slippages prevented the full disbursement of all but one of Pakistan’s arrangements since 1980, poor implementation is also undeniable. However, there is a strong perception in Pakistan that the programs had a low probability of success from the outset, owing to various design flaws and weak ownership at the highest political level. These problems were compounded by shortcomings in conditionality and program “enforcement” by the IMF that resulted in some core underlying problems remaining unresolved. Many of these difficulties reflected “systemic” issues associated with the IMF’s approach in Pakistan rather than just technical issues associated with the IMF staff’s analysis (see the section “Some Program Design Problems Were Rooted in Deeper IMF Governance Issues” below).

Most IMF-supported programs had several design flaws

Overoptimistic assumptions and unrealistic objectives

17. Most programs, particularly from 1993 onward, were based on overly optimistic projections as regards such key elements as GDP and export growth, as well as regarding the growth of domestic savings and investment (Figure 9.2).18 Any such comparison needs to take account of the fact that macroeconomic projections made in IMF-supported programs are not unconditional forecasts: they implicitly assume that the program will be implemented as planned, which was clearly not the case in Pakistan. However, the discrepancies between projections and outturns are so large as to make it difficult to assess whether poor implementation was the cause or, at least in part, the consequence of these discrepancies.

Figure 9.1.Pakistan: History of Lending Arrangements

Sources: IMF Treasurer’s Department and IEO calculations.

Figure 9.2.Pakistan: Growth of GDP, Exports, Gross Domestic Saving, and Gross Domestic Investment

Source: IMF staff reports.

18. The growth rate was overestimated on average by 1½ percentage points (over 2 percentage points from 1993 on). Whether this gap resulted from an excess of optimism ex ante or from unforeseen exogenous shocks, the IMF generally proved reluctant to adjust the program framework and key objectives, especially the fiscal deficit target. Reviews of internal documents suggest that this reluctance reflected concerns to avoid accommodating policy slippages that would undermine the incentives to persevere with the core fiscal reforms needed for long-term sustainability.19 However, since most programs did not incorporate specific contingency plans to deal with the effects of lower-than-projected growth, and in the absence of adequate methodology to distinguish the impact of policy slippages from that of exogenous shocks, the result was often lengthy negotiations on the scope and nature of the corrective actions needed. In most cases, this process resulted in fiscal targets generally being met at least up to the first review, at the cost of ad hoc efforts that were neither sustainable nor economically efficient, while negotiations on adaptations to the program framework and necessary policy changes dragged on thereafter, with lack of agreement on a suitable course of action de facto sending the program off-track.

19. Export growth projections also proved far too optimistic: instead of rising by roughly 15 percent a year, which was the average program projection and would have represented an acceleration from the 11 percent growth rate observed in the 1970s/80s,20 exports rose by barely 5 percent annually during the 1990s. As a result, current account projections typically projected a too rapid improvement, while capital account projections were, on the whole, too conservative. The outcome was a substantial underestimation of the buildup of external debt servicing charges from the early 1990s onward (Figure 9.3).

Figure 9.3.Pakistan: Current and Capital Account Projections and External Debt

Source: IMF staff reports.

20. Similarly, some program targets proved unrealistically ambitious given the time frame and the implementation capacity available. This was especially the case for tax revenues (Figure 9.4). These targets were also very ambitious compared to IMF- wide averages.21 Indeed, these targets were never met over the 1988–2000 period in spite of frequent in-program downward revisions.

Figure 9.4.Pakistan: General Government Balance and Tax Revenues

(In percent of GDP)

Source: IMF staff reports.

21. While stakeholders, including IMF staff, now generally agree that tax revenue targets were generally unrealistic ex ante, the reasons for it are not entirely clear. The authorities argue that the overoptimism of revenue projections reflected an overestimation of their implementation capacity. IMF staff, on the other hand, contend that the lack of realism of revenue projections reflected the authorities’ pressures and was fueled by the political economy of Pakistan’s budgetary process.22 Staff memoranda exchanged during the internal review process indicate that they resisted what they viewed as overoptimistic projections as much as possible, but in the end had to defer to the authorities’ knowledge of their own abilities on that issue. In any event, the overoptimism of revenue projections undoubtedly helped paper over difficult policy choices in a context of strong pressures to agree on a program.

22. As regards structural reforms, the overoptimism was reflected in the length and diversity of the reform agenda embedded in IMF-supported programs from 1988 onward and from the overambitious timetable envisaged for reforms that take time to implement even in countries with far more administrative resources than Pakistan.23 The number of areas of economic policy covered by explicit conditions increased from four in the 1988 SBA/SAF to eleven in the 1997 EFF/ESAF. The result was insufficient prioritization and an overburdening of the policy formulation and implementation capacity of the authorities—a capacity that was limited in part by technical constraints, but mostly by the lack of political will to take measures with significant short-term costs (for instance, as concerning the removal of tax exemptions).

Some of the policy prescriptions had adverse side effects

23. Some of the policy prescriptions embedded in IMF-supported programs turned out, owing to implementation difficulties, to have unintended side effects.

24. In the fiscal area, two mutually reinforcing problems occurred. First, the strategic orientation of shifting taxation from international trade to domestic activities, an important feature of all IMF-supported programs since 1980, proceeded at an uneven pace that caused de facto sequencing problems. While the reduction of tariffs and other taxes on international trade was relatively fast, it took much longer for the general sales tax (GST) instituted in 1990 to yield a comparable revenue, owing to numerous exemptions that took no less than ten years to eliminate, and to the shortcomings of tax administration.24 Likewise, income tax was prevented by poor tax administration and too narrow a base from making a sufficient contribution to the revenue collection effort. These factors explain the bulk of the revenue shortfalls that were a common characteristic of all IMF-supported programs since 1988 (see Figure 9.5).

Figure 9.5.Pakistan: Evolution of Tax Revenue Structure

Revenue to GDP ratio (in percent)

Sources: IMF staff reports and IEO calculations.

25. The second problem was that attempts to meet the fiscal deficit targets led to the frequent adoption of ad hoc tax increases and expenditure cuts in the course of the program. These measures were frequently inconsistent with the medium-term strategy pursued under the program. In particular, hikes in tax rates and surcharges, which in the short term appeared to be the most effective way to fill the revenue gap, might have contributed to reducing the tax base further by encouraging taxed activities to shift to the informal sector. They also led to an increased complexity of the tax system.25 Many observers in Pakistan commented that such ad hoc measures have had a detrimental effect on business sentiment by making tax legislation unpredictable (owing to frequent legal changes but also to the increased scope for taxpayer harassment by the employees of the Central Board of Revenue, who traditionally have large discretionary powers in the implementation of tax policy). As regards expenditure cuts, given the inflexible structure of expenditure (due to the large share of military spending and the weight of interest expenditure) cuts inevitably fell on development and social spending, which resulted in a marked decline in public investment (from about 10 percent of GDP in 1992 to 4½ percent in 2000).

26. As regards financial sector reforms, the same process of uneven implementation of a package of reforms resulted in ex post sequencing problems. The 1988 SAF envisaged the simultaneous implementation of a liberalization of interest rates and lending practices, a major regulatory reform of the banking sector, and a significant reduction in the borrowing requirement of the public sector. In the event, only the first leg of this tripod was delivered. As a result, the financing costs of fiscal deficits ballooned, making it ever more difficult to balance the budget, and the soundness of the banking sector deteriorated as governance weakened. This led to the accumulation of large volumes of nonperforming loans. The very deteriorated state of public banks’ loan portfolios also accounts for the high level of interest rate spreads that has had a depressing impact on economic activity in recent years.

27. However, it is worth emphasizing that the two types of side effects discussed above do not imply that the fundamental policy prescriptions were wrong. In particular, it is unlikely that economic performance would have been better had tariffs not been lowered until sufficient revenue could be obtained from domestic taxes (i.e., in the late 1990s), or had interest rates not been liberalized until the fiscal deficit had been sharply reduced (i.e., not yet). Rather, these side effects demonstrate that a critical mass of reforms is needed for adjustment to take hold, and that the adjustment path chosen cannot ignore the longer time frame needed to implement the most complex of these reforms. Nonetheless, the trade-offs and risks involved (of which internal documents suggest that IMF staff was keenly aware) would certainly have warranted a more open debate in policy discussions with the authorities and in reports to the Executive Board.26

Insufficient emphasis was placed on institutional reforms

28. In retrospect, there is ample recognition—both in the IMF (more in the views expressed in staff interviews than in reports) and in Pakistan—that IMF-supported programs should have put much stronger emphasis on institutional reforms, particularly in tax administration, the banking sector, and public enterprises, all of which have only begun to be addressed in recent years.27 In addition to directing attention to implementation capacity issues, thereby making program objectives less unrealistic, an explicit emphasis on institutional reforms would have mitigated the program design problems discussed above.

29. As far as tax administration is concerned, interviews with Pakistani officials and a review of internal documents suggest that there was an implicit understanding on both sides that the revenue targets could only be met if far-reaching tax administration reforms were undertaken in addition to the changes in tax policy explicitly monitored under IMF-supported programs. However, none of the programs had an explicit focus on tax administration reform, and even though Pakistan received several technical assistance missions from the IMF in relation to tax administration problems, the actual implementation of reforms was never pursued through specific conditionality until the 1997/98 EFF/ESAF.28 As a result, no significant progress in revenue collection occurred during that period, in spite of protracted but eventually significant improvements in tax policy.

30. Likewise, stakeholders on both sides now generally agree that much of the deterioration in the quality of banks’ portfolios that occurred in the 1990s might have been avoided had the financial sector reforms of the beginning of the decade been designed differently. In particular, giving the same weight to regulatory improvements as to interest rate liberalization and paying due attention to banks’ management flaws (in particular, political interferences in lending decisions) would probably have led to a better outcome.

31. Finally, the effectiveness of IMF-supported programs would have been enhanced had public enterprises been handled from an institutional reform perspective from the start: IMF-supported programs’ focus on preventing too large misalignments between public enterprises tariffs and world market prices was warranted, but it ignored the broader restructuring needs of these enterprises. The drawbacks of this approach are particularly visible in the case of the largest public enterprises (discussed further below), whose impact on the economy expand way beyond that of potentially distortionary prices and eventually took the form of a drain on fiscal resources, expensive and unreliable supply of basic utilities, accumulation of nonperforming loans in the banking system, and massive cross-arrears within the broader government sector.

Lack of ownership and inconsistent monitoring resulted in poor implementation

Ownership was generally weak

32. Pakistan suffered from a very unstable political environment throughout most of 1988–2000. No single government managed to stay more than three years in power, with an average tenure of 18 months.29 The effects of this instability on successive governments’ ability (and willingness) to implement comprehensive reforms and carry through an ambitious adjustment effort seem to have been largely underestimated—at least in reports submitted to the Executive Board. Most UFR staff reports during that period acknowledged the fluidity of the political situation, but usually stated that governments had strong ownership of the programs and were strongly committed to their implementation, even when programs were negotiated with caretaker governments and endorsed at the last minute by the incoming cabinet, as occurred for the 1988 SAF and the 1994 EFF/ESAF. While the claim made by successive IMF reports that there was a broad consensus across major political parties on economic policy matters was correct, this consensus gave no assurances as to the political ability of elected governments to carry through unpopular adjustment policies (Box 9.1).

33. In the event, as political difficulties arose, it often appeared that the reform agenda of the economic team had little backing at the highest political level,30 at least in the sense that top decision makers were not sufficiently convinced that the reforms were necessary and that the economic price of postponement outweighed the political costs of early implementation. As a result, implementation efforts often limited themselves to the minimum needed to ensure the continued disbursement of IMF resources. In many instances, what was observed was more the letter of the conditions than their spirit, and even though the tightening of conditionality over time gradually reduced the room for such practices, it was insufficient to compensate for the lack of ownership. For instance, in the fiscal area, performance criteria initially targeted bank financing of the deficit, while the overall deficit was only a benchmark. Since bank financing was not a predominant source of financing of the deficit, Pakistan was able to comply with the fiscal performance criteria until 1994 in spite of substantial fiscal overruns.

34. As regards structural reforms, conditionality was initially set in the form of general commitments in the letter of intent or structural benchmarks, which were at best partially observed (e.g., the removal of exemptions from excises, custom duties, and other taxes targeted by the 1980 EFF and the 1988 SAF). When conditionality was “hardened,” in the mid-1990s, conditions were often met in ways that minimized their impact. A few examples can illustrate the general problem:

  • by enacting a law but not implementing it (e.g., extension of GST to the services sector and taxation of agricultural income);
  • by adopting a new tax but with so many exemptions as to make its additional yield negligible (e.g., the GST act in 1990);
  • by abolishing existing tax exemptions while simultaneously creating new ones;
  • in other cases, measures adopted were subsequently reversed or suspended (e.g., the petroleum price adjustment mechanism from 1995 to 1998).

Box 9.1.Ownership Assessment in the Context of Pakistan Arrangements with the IMF

A brief analysis of the politico-administrative context of Pakistan at the onset of the multiyear arrangements approved in 1993/94 and 1996/97, undertaken by a political scientist at the IEO’s request, shows that there were conflicting factors at play, making the assessment of strength of ownership and political feasibility particularly difficult:

  • On the positive side, governments in power at the time of program approval enjoyed a comfortable majority in Parliament and in key provincial governments and the reforms they put in motion before the program’s approval were consistent with its spirit, they could rely on well-structured and, at the top echelons, skillful and stable bureaucracy. Moreover, there appeared to be a broad consensus across major political parties about the main thrust of economic reforms.
  • On the negative side, vested interests weighed heavily in the political power base of every government and were in a position to block the approval of reforms (i.e., they had a “veto power”), many of which directly or indirectly threatened their privileges, implementation capacity was weak due to serve deficiencies in the rule of law and law enforcement; decision-making style (restricted to a narrow circle), and a policy dialogue that presented reforms as IMF-imposed, were not conducive to a broad ownership of reforms. Taking all these factors into consideration, serious doubts should have been raised about the prospects for consistent implementation of the programs.
See Appendix 1 for a more comprehensive presentation of the analysis conducted.

35. Furthermore, during much of that period, the government’s practice was to use the IMF as a scapegoat for unpopular decisions, a strong indication of the limited degree of actual ownership. Over time, the result was that the surest way to undermine popular support for any measure was to give it an “IMF” label.31

The mitigating devices available to the IMF were not used fully

36. The extent of political instability and ensuing low ownership during the period of prolonged UFR was such that even using the whole arsenal of safeguards at the disposal of the IMF would have been unlikely to produce the results promised in successive programs. However, many of the safeguards—which the Board had established as part of a strategy for minimizing prolonged use (see Chapter 3 of Part I)—were not used fully until late 2000. At that stage, practice shifted to the strictest standards of track record probing.32 However, ownership too was higher, which was probably more determining in the success of the program than any parameter under the IMF’s control.

Track record requirements

37. When there are significant doubts about the authorities’ commitment or ability to pursue adjustment policies over a sustained period of time—perhaps as a result of previous policy slippages—it is customary practice for the IMF to ask a government to build a track record before engaging in a multi-year arrangement. In Pakistan, however, track record requirements were often squeezed out by protracted negotiations. For example, the 1994 EFF/ESAF, for which negotiations lasted 18 months, was preceded by a Stand-By Arrangement, but the SBA was not allowed to run its course: it was cancelled after six months to be replaced by the multiyear arrangement, even though the first review had not been completed due to slippages.33 Negotiations for the 1997 EFF/ESAF started in mid-1996, at a time when Pakistan was under a Stand-By Arrangement following the collapse of the previous EFF/ESAF. The EFF/ESAF was eventually presented to the Board after a six-month track record—not under the SBA, which had been irremediably interrupted owing to very large policy slippages, but under a staff-monitored program (SMP) whose targets had been revised several times to accommodate various slippages and shocks.

38. Other tools available to the IMF are prior actions and the phasing of disbursements, with a back-loaded schedule providing a greater incentive to sustain the policy effort over the medium term.

  • Prior actions were used in most UFR requests since 1988. However, these prior actions did not always include the key conditions whose nonobservance had sent the previous program off-track (e.g., GST base extensions in the 1994 program, or the petroleum price adjustment mechanism in 1997). In addition, prior actions, like any other conditionality, are subject to superficial or temporary observance only. Two examples drawn from agricultural taxation illustrate this point: the 1993 SBA included a prior action related to the extension of taxation to the agricultural sector. The prior action was considered to be met, but legal impediments that subsequently surfaced made it necessary to impose a new prior action in the 1994 ESAF, this time related to the parliamentary approval of the ordinance on federal agricultural taxation. Neither prior action resulted in meaningful taxation of agricultural incomes (Box 9.2).34 Thus, the issue in Pakistan’s programs appears to have been the prioritization of prior actions and their integration into program design, rather than their quantity.
  • The tranching of disbursements was generally mildly front-loaded,35 even in cases where facility specific rules allowed flexibility. This favorable tranching contrasts with the generally back-loaded design of the policy agenda: even in the 1997 ESAF/EFF, which came after a series of interrupted programs, half of the structural conditions specified from the outset were related to the second program year, and several pivotal measures, such as income tax reform, civil service reform, extension of the GST to the retail level, and tariff cuts were only planned for the third program year.


39. The conditionality response to the practices induced by the low ownership of programs by the authorities was to gradually close as many loopholes as possible, through an increase in the overall number of both macroeconomic and structural conditions, through a larger recourse to performance criteria and prior actions as opposed to general commitments in the letter of intent, and through the use of continuous performance criteria and other no-reversal clauses (Figure 9.6).

Figure 9.6.Pakistan: Evolution of Structural Conditionality

(Average number of conditions per program year)

Sources: IMF staff reports and IEO calculations.

40. It is doubtful that any form or volume of conditionality would have been sufficient to compensate for the authorities’ fundamental unwillingness or inability to implement many of the policies promoted by successive programs. However, two lessons can be derived from this experience. First, strict implementation and enforcement would have been easier had conditionality been more focused on truly critical areas. Second, the approach—eventually adopted with success in the areas of tax exemptions and utilities price adjustment—consisting in setting conditionality on policy rules rather than discretionary actions by the authorities might have speeded up the reform process if adopted and generalized earlier.

41. More minor problems that hindered the effectiveness of conditionality in Pakistan are the following. First, the tightening of conditionality always occurred with a lag following repeated failures to deliver on policy undertakings, a lag that in some cases was very long (e.g., conditionality on the financing of the fiscal deficit or utilities price adjustments). Second, in some areas, such as tax administration, utilities pricing, or public enterprise reform, the effectiveness of conditionality was reduced by the lack of a sustained, consistent approach. These areas, which were presented as key to the success of the programs of the 1980s, subsequently ceased to be covered by conditionality for several years, despite very limited progress and occasional backsliding. In other cases (e.g., agricultural taxation), the inconsistency stemmed from the lack of follow-up on undelivered commitments from one review or program to the next.

Box 9.2.A Chronology of IMF Conditionality on Agricultural Taxation in Pakistan1

Agricultural taxation was pursued through IMF-supported programs since 1981 through the following conditionality (implementation status as reported in staff reports noted in parentheses). None of this conditionality—much of which was actually observed, in a narrow sense—had substantially increased the contribution of the agricultural sector to tax revenues by the end of the decade.

  • 1981 (EFF): commitment in the authorities’ Memorandum of Economic Policies (MEP) to tax agricultural output within one year (formally implemented, but with a very low yield).
  • 1991 (SAF): commitment in the MEP to extend the scope of agricultural taxation in the 1992/93 budget (not implemented).
  • 1993(SBA): prior action on extension of taxation to the agricultural sector (met).
  • 1994 (EFF/ESAF): prior action on parliamentary approval of ordinance relating to federal agriculture taxation (met) and performance criteria on expansion in agricultural tax base including through Produce Index Units (PIU) adjustment, within six months (met).
  • 1995 (SBA): performance criteria on inclusion of provisions in 1996/97 budget to broaden agricultural taxation through an increase in PIU to PRs 400 and extension of its base (met).
  • 1996 (SBA, revised): performance criteria on adoption by all provinces (by end-December 1996) of ordinances on agricultural taxation (not met); establishment of a task force on the potential revenue from provincial agricultural taxes (met); and task force to present recommendations for 1997/98 budget and the medium term.
  • 1997 (EFF/ESAF): performance criteria on harmonizing provincial taxation of agricultural income with that prevailing in Punjab, by end-June 1998 (not met; waived and rescheduled).
  • 1998 (EFF/ESAF): performance criteria on harmonizing provincial taxation of agricultural income with that prevailing in Punjab, by end-June 1999 and structural benchmark on finalization by the same date of the rate and threshold structure for the provincial tax on agricultural income capable of yielding PRs 3.5 to PRs 4 billion in 1999/00 (both conditions partially implemented, with delay).
  • 2000 (SBA): performance criteria on GST extension to urea fertilizers and pesticides by March 2001, and to all other agricultural inputs by September 2001 (both performance criteria met); structural benchmarks on full implementation of agricultural income taxes on the basis of provincial implementing regulations to become effective in each province by end-June 2001 (met with delay in two provinces, unmet in other two).
1 Agriculture accounts for roughly 25 percent of GDP, employs half of the labor force, and is one of the largest earners of foreign exchange. Under the Constitution of Pakistan, farmers have been exempt from taxes on their incomes from agriculture, and only the provincial governments are permitted to levy a land tax. The political power of large landowners has long prevented the federal government from seeking the legal changes needed to give it authority to tax agricultural income or land and Provincial governments from using the authority they have to pursue the same goal. This situation has led the agricultural sector to become a legal, and sometimes illegal, shelter for other forms of income.

42. In the few cases where conditionality responded proactively and consistently, better results were eventually obtained, e.g., in the area of tax exemptions: conditionality evolved from a simple LOI commitment to phase out existing tax exemptions (in 1988), to a continuous performance criterion on the nonintroduction of new exemptions (in 1995), to a prior action on the passage of an act prohibiting the government from creating new exemptions (in 1998).

43. Finally, the relative generosity of the IMF in granting waivers to the authorities, or in consenting to renew its support soon after a program interruption typically caused by large fiscal slippages, led—according to many Pakistan officials—to a form of moral hazard: the expectation that the IMF would eventually provide financing weakened incentives to tackle the fiscal deficit forcefully. Indeed, of the seven program reviews completed in the 1990s, five involved at least one and generally several waivers, mostly concerning fiscal performance criteria.36 Furthermore, in spite of the many interruptions suffered by IMF-supported programs, the interval between two disbursements of IMF resources was generally short—never exceeding 12 months over 1991–99.37

Some Program Design Problems Were Rooted in Deeper IMF Governance Issues

Institutional considerations weighed heavily in decisions regarding IMF involvement in Pakistan

Geopolitical considerations

44. “Most IMF-supported programs primarily served political purposes. Thus, it should come as no surprise that they did not achieve much in terms of economic results.” That view, expressed by a senior Pakistan official interviewed by the IEO, appears to be very widely shared in Pakistan, both within and outside official circles. A large proportion of IMF staff involved in Pakistan programs were also of the opinion that political considerations had, at times, prevailed over technical judgments, not necessarily on the details of program design but in terms of the over-all threshold required for a program to be supported.

45. While there is no hard evidence or “smoking gun,” either in internal memoranda or in the record of Board discussions, that noneconomic considerations played a predominant role in IMF decisions to support a request for use of Fund resources or to complete a program review, there is no shortage of anecdotal evidence—coming from both former authorities and the IMF staff—that such considerations did matter importantly on some occasions. Moreover, a few significant “program events” closely followed major geopolitical developments, for instance the 1980 EFF following the Soviet invasion of Afghanistan; and the mid-1998 program interruption following Pakistan’s nuclear tests.

46. More generally, until 1998, the prevailing perception within IMF staff was that the principal IMF shareholders, no matter how demanding they claimed to be on the substance of programs, were not willing to take the risk of major turmoil in Pakistan that an interruption of IMF support might have caused. This resulted in a general sense, shared by staff and the authorities, that the IMF would remain involved irrespective of performance under a specific program. The succession of adverse events of 1998–99 (nuclear tests, a military coup, and the unveiling of past misreporting of fiscal data to the IMF) caused a dramatic reversal in shareholders’ views, after which IMF support was perceived by IMF staff and the authorities as unlikely—short of a very strong performance under an extremely ambitious program.

47. Assessing with accuracy to what extent geopolitical considerations did supersede economic ones in program or program-design decisions is virtually impossible, given the element of judgment appropriately present in any UFR decision. But the simple fact that IMF-supported programs are widely perceived as heavily influenced by political factors—both by the authorities who sign onto them and by the economic agents whose behavior they are meant to influence—probably weakened the efficacy of these programs.38

“Systemic” considerations related to the IMF mandate

48. There are four dimensions of the IMF mandate that might account for some of the design problems discussed above. First, as a monetary institution, the IMF is traditionally viewed as primarily mandated to provide temporary balance of payment support to its members, not long-term financing. This means that its interventions are expected to allow a restoration of balance of payments viability over a relatively short period. While the creation of the EFF and, later on of the SAF, ESAF, and PRGF, has lengthened the time frame for the restoration of viability, until recently it would have been difficult to present for the consideration of the Executive Board a program that did not lead to a restoration of balance of payments viability within a short- to medium-term time frame. This constraint is likely to have contributed to two related problems: (i) a tendency toward overoptimism both in the speed with which the real economy would respond to the policy measures and in the pace at which difficult structural reforms could be implemented; and (ii) a focus on types of structural “conditions” that can be clearly delivered within programs’ time frame—or even in the few weeks separating the finalization of the program from its presentation to the Executive Board, in the case of prior actions. Complex institutional reforms, which are key to longer-term sustainability, are much harder to fit within such a framework.39 As far as Pakistan is concerned, program assumptions did not become markedly overoptimistic until the early 1990s (see the figures in the section “The Lack of Effectiveness of Pakistan’s IMF-Supported Programs Stems from Both Design and Implementation Issues”). From then on, internal memoranda do suggest that several review departments had serious reservations about the optimism of program assumptions, particularly regarding GDP and export growth, as well as revenue targets. The “systemic constraint” discussed above is a plausible explanation for why programs went ahead despite these reservations.

49. Second, until the late 1980s, the IMF interpreted its mandate as narrowly focused on macroeconomic policies and a few structural areas. That conception changed dramatically when concessional facilities were created, which led the IMF to embrace a broad agenda of structural reforms, without necessarily having the expertise needed for an optimal selection, sequencing, and design of these reforms, at least initially. While in theory these shortcomings could have been mitigated by a close collaboration with the World Bank (and the AsDB), in practice collaboration was imperfect. As a result, key areas, such as governance and institutional reforms in tax administration, civil service, and public enterprises, ended up being handled inadequately by both IFIs for lack of emphasis and coordination of their efforts. Both institutions implicitly or explicitly recognized that these issues were central to long-term macroeconomic sustainability. But they do not appear to have reached an effective agreement, among themselves and with the government, on what the priorities should be and on coordinated action plans to address them, until the EFF/ESAF approved in late 1997.

50. The problems of the power sector provide a good illustration. Initially (i.e., in the 1988 SAF), the IMF focused on power tariff adjustments, on an “as required” basis, with the primary purpose of containing the fiscal impact of the operational losses and investment needs of the power sector. However, that condition was omitted in subsequent arrangements,40 while at the same time, the World Bank was shifting its focus toward human development, limiting its interventions in the energy sector to technical aspects. Meanwhile, severe institutional problems intensified.41 As a result, by 1996–97, the overall borrowing requirement of the seven major public enterprises had reached 2½ percent of GDP (the bulk of which came from WAPDA and KESC, the two main public power producers). Their nonperforming loans had piled up in public banks to the point of threatening their stability and substantial cross-arrears had accumulated among public enterprises and vis-à-vis the government. At that point, the seriousness of the problem brought a greater focus on it by both IFIs. The World Bank helped the authorities draw up operational and financial restructuring plans, whose implementation was subsequently monitored through IMF-supported programs’ conditionality.42 By end 2000, both the financial and the operational restructuring were under way but are likely to take considerable time to yield substantial efficiency gains and cost reductions.

51. Third, the “seal of approval” function effectively given to IMF-supported programs by donors and creditors might account for the tendency to over-promise in programs and subsequently undersanction when the authorities did not deliver. The fact that the IMF acted as gatekeeper for access to many other sources of financing in one sense gave it great leverage, but also meant that the consequences of a prolonged interruption in programs would be very severe. Ultimately the IMF—presumably reflecting the views of its shareholders—proved extremely reluctant to risk the costs of such major disruption, which meant that there was an inbuilt tendency not to insist too hard on the core issues.

52. Fourth, the IMF’s mandate gives it an obligation to support member countries making the necessary efforts to address their economic difficulties. Thus, any government committing itself to making such efforts has to be given the benefit of the doubt, at least initially. In Pakistan, since 1988, political instability was such that each new IMF arrangement practically coincided with a new government and it would have been extremely difficult not to give such a government the initial benefit of the doubt on its declared policy intentions.

Defensive lending considerations

53. It has been argued43 that a factor contributing to the prolonged use of IMF resources when adjustment fails to take place is the need to ensure that obligations falling due are met. The IMF did have a significant exposure to Pakistan throughout the period,44 although Pakistan’s record of repayment of its obligations to the IMF was impeccable even at times of severe stress on its international reserves position. The only times when there is evidence that “defensive lending” considerations could have been a significant factor, as reflected in explicit concerns about a possible default expressed in internal memoranda, were when Pakistan came closest to the brink of a foreign exchange crisis: (i) in late 1996, just before the revival and augmentation of the SBA; (ii) in the months preceding the completion, in January 1999, of the second review of the 1997 EFF/ESAF following its de facto interruption in May 1998; and (iii) in mid/late 2000, prior to the approval of the SBA.

Crowded out by UFR, surveillance played only a limited independent role

54. In Pakistan, all but two of the Article IV consultations of the 1988–2000 period were conducted jointly with UFR activities. A comparison of the depth and quality of surveillance in stand-alone Article IV reports with joint Article IV/UFR reports and with surveillance guidelines over 1982–2000 indicates that joint consultations tended to make a more cursory treatment of such key issues as medium-term perspectives, sensitivity to shocks, and vulnerabilities.45 They also tended to be less concerned about exploring trade-offs between policy options and less candid about divergences of views between the staff and the authorities.

55. Once again, this “crowding out” of aspects of surveillance appears to reflect systemic factors, rather than a particular shortcoming of the IMF staff’s work on Pakistan. For example, since there is an inbuilt incentive not to undermine the catalytic role of an IMF-supported program in mobilizing other financing, surveillance activities that are closely associated with programs tend not to raise too many questions about alternative policy design or downside risks.

Implications of short-term stabilization measures for longer-term growth and sustainability of adjustment were not analyzed sufficiently

56. There are three aspects where, with the benefit of hindsight, it appears that surveillance could have played a larger role in analyzing some of the longer-term implications for sustainability. First, the consequences for long-term growth and hence sustainability of the structure of fiscal adjustment could have been given greater attention—although more analysis was undertaken in internal reports than was reflected in Board papers.46 Specifically, the adverse long-term effects of ad hoc revenue measures taken by the authorities to compensate for revenue shortfalls, or ad hoc expenditure cuts instead of fundamental improvements in the structure and quality of expenditure were not emphasized in surveillance reports, even when strong reservations were expressed about them in the internal review process.

57. Second, as most programs failed to deliver the required adjustment but nevertheless unlocked substantial financing flows, over time financing largely substituted itself for adjustment. To a large extent, this merely reproduced the pattern of the 1970s and 1980s, but this time in a low-growth environment with a much higher cost of financing (workers’ remittances and ODA flows, relative to GNP, both fell by over 40 percent in the 1990s compared with the average of the two previous decades, and the average grant element also fell from 46 percent to 32 percent). As a result, Pakistan got trapped into a debt sustainability problem, which was not fully analyzed in IMF reports until 1997, apparently in part because of concerns about undermining the credibilityenhancing effects of programs.47

58. Third, surveillance failed to sound the alarm loudly enough at a sufficiently early stage about the gradual buildup of an eventually very large uncovered exposure of the central bank to foreign currency deposits (FCDs)—a central avenue through which the debt buildup occurred.48 By the time the 1997 ESAF/EFF was approved, FCDs had reached over $11 billion, that is, the equivalent of 17 percent of GDP and 9½ times the level of gross international reserves. This vulnerability turned into a foreign exchange liquidity crisis when capital outflows intensified in mid-1998 after Pakistan’s nuclear tests and ensuing international sanctions, leading the authorities to impose a deposit freeze.

59. Surveillance reports, while factually accurate, never gave much prominence to the issue, even when the downsides of the heavy reliance on FCDs were mentioned, and mission briefs suggest that the buildup of large uncovered foreign exchange exposure through FCDs was never a pivotal issue of either policy or program discussions with the authorities until late 1996 (Box 9.3). However, we understand from interviews with staff and others that the potential vulnerabilities involved were raised with the authorities beginning in the early 1990s, even though the issue was not given prominence in IMF reports until much later.

Factors that might have undermined confidence in the programs were downplayed

60. Discussions of risks to the program outlook and sensitivity analyses of medium-term projections were generally limited, in the sense that sensitivity analyses never involved sufficiently large adverse shocks that they would push the medium-term outlook off its “sustainable” path, even though the Pakistan economy was subject to a variety of shocks of much broader magnitude.49 Likewise, only exceptionally were substantive ex ante analyses of the implications of policy slippages offered (the 1993 SBA request report and the 1991 and 1995 Article IV consultations are the only, but welcome, counter-examples). When such analyses were provided, they did not spell out the side effects that might result from an uneven implementation of reforms intended as a package.

Box 9.3.The Coverage of the Issue of Foreign Currency Deposits in IMF Staff Reports

From 1992 to late 1995, surveillance reports gave factual accounts of the evolution of FCDs,1 and a few noted that they compounded the weakness of gross international reserves, but they did not flag them as a serious vulnerability,2 nor did they offer any specific policy advice to address the problem. The IMF’s approach changed dramatically in 1996: there was no Article IV consultation, but the staff report on the second review of the SBA, which was completed as Pakistan was on the verge of a foreign exchange crisis partly owing to withdrawals from FCDs, clearly emphasized the risks of Pakistan’s heavy reliance on short-term liabilities to finance its current account deficits and, for the first time, set specific conditionality on the preparation, at a future date, of an exit strategy from the forward exchange cover provided to banks by the SBP. In the event, the definition of a strategy took somewhat longer than initially expected, by which time the immediate crisis had passed, and the 1997 Article IV returned to a relatively relaxed stance: while the presence of a large uncovered foreign exchange exposure was still identified as a “major concern,” the emphasis was more on the constraint this exposure represented for the conduct of exchange rate policy and the distortions to incentives induced by the too low price of the SBP forward cover fee, than on the looming risks of foreign exchange crisis.3 The EFF/ESAF arrangement approved at the same time did outline a detailed exit strategy, based on the recommendations of an IMF TA mission, which was partially supported by explicit conditionality. However, to the extent that financing needs had not diminished, it is doubtful that the authorities would have pressed ahead with its implementation even if the mid-1998 crisis had not struck.4

1 However, from 1993 onward, those factual accounts were only partial, since—at the authorities’ request—FCDs owned by residents were reported in the balance of payments “above the line,” as part of private transfers, and even FCDs held by nonresidents were not included in the stock of external debt in tables presenting key indicators. Furthermore, FCDs held by residents, even though they represented a liquid claim on the central bank’s reserves, were not netted out from international reserves for the purpose of IMF monitoring of net international reserves.2 The first staff report that put some emphasis on the vulnerability aspect was the background paper to the 1995 Article IV report. Internal documents attest that the vulnerability aspect had been fully understood as early as 1993: a memorandum from the Research Department noted that “attempts by domestic residents to reduce their holdings of foreign currency deposits in the domestic banking system could precipitate not only balance of payment difficulties, but also a major banking crisis. Indeed, given the degree of mismatch between Pakistan’s low gross international reserves and the uncovered foreign currency accounts… , even modest capital outflows could trigger a loss of confidence in the authorities’ ability to sustain the convertibility of the foreign currency deposits. Avoidance of a foreign exchange crisis and reversal of the liberalization measures would be a major success at the current juncture.”3 However, the paper did note, in a separate section, that “Pakistan’s fragile external reserve position is compounded by the accumulation of short-term foreign currency liabilities by the banking system, mostly… FCDs.”4 Indeed, the staff report on the first review of the ESAF/EFF, completed in March 1998, noted that the authorities viewed with concern the decline in FCDs registered in previous months.

61. For the same reasons, most joint surveillance-UFR reports provided little analysis of trade-offs between alternative policy strategies and the divergences of views between IMF staff and the authorities as to the best option. For instance, when programs went along with the authorities’ preference for imposing a turnover tax on traders instead of putting them under the GST net, as favored by staff, or for a tax registration drive instead of focused improvements in tax collection, only the retained options were presented, with no detailed discussion of their merits and downsides relative to the alternative. Likewise, disagreements between staff and the authorities (as well as within staff) on the appropriate exit strategy from the FCDs freeze were not discussed in the reports.

Only limited ex post evaluation of UFR was undertaken

62. Another function of surveillance that was regrettably toned down in most Article IV reports—and in other Board documents—since 1988 is the ex post evaluation of UFR experiences. Among earlier reports, only the report on the 1991 Article IV consultation provides a detailed review of performance under previous programs. Subsequent surveillance reports gave only limited factual accounts of the most recent UFR performance, and were generally silent about the recurrent implementation difficulties that were encountered, in contrast with the more critical tone of internal documents.50 The 2000 Article IV report represented a marked improvement in that respect, by taking a long-term look at past economic performance, but it did not examine the reasons for the failures of the previous, or earlier, programs.

Internal governance issues occasionally contributed to program design imperfections

Excessive attention is paid to fine-tuning the financial programming framework

63. While a sound and internally consistent macroeconomic framework is essential to the success of any adjustment program, the IMF’s “financial programming” framework, as implemented in practice in Pakistan, suffered from several imperfections that took extra significance owing to one major weakness: the soundness of the framework itself is heavily dependent upon the realism of growth targets. In Pakistan, for reasons discussed above, strong incentives to be overoptimistic were at play.

64. First, the precision with which quantitative targets are set (and monitored) is at odds with the severe imperfections of the data they are based on. Second, the weight attached to these targets in setting appropriate macroeconomic policies (and later appraising them) often does not recognize sufficiently the considerable uncertainties attached to the underlying behavioral relationships. In light of these considerations, the authorities and most staff were of the view that too much time had been spent on “finetuning” the financial programming variables, at both the negotiation and subsequent monitoring stages, when the time could have been better spent on more fundamental issues.

65. Issues that, according to IMF staff members themselves, did not receive sufficient attention as a result of this process in Pakistan are real economy dynamics (in particular the links between the policy variables monitored under programs and economic outcomes), analysis of the sources of growth, and impact on economic performance of a variety of exogenous shocks.

Some of the incentives to which IMF staff is subject may have perverse effects

66. There is a rather widespread perception, shared by Pakistan officials and many IMF staff, that the decision-making process of the IMF was biased toward programs that look “tough” on paper, even if there were substantial doubts about their realism. A number of IMF staff who have worked on one or more programs with Pakistan noted that the internal incentive system rewarded toughness more than realism, and that negotiating a program with ambitious objectives and few departures from the mission brief smoothed the internal review process considerably, whereas attempts to be realistic and accommodative of the authorities’ concerns—legitimate or not—did not.

67. Furthermore, a majority of staff interviewed expressed the view that the excessive attention paid to the fine-tuning of the financial programming reflected the focus of the internal review process on that aspect of programs,51 at the expense of much greater attention to judgments on ownership and the capacity of the authorities to implement the policy undertakings couched in the program. While the move from the ESAF to the PRGF/PRSP approach was intended to bring about improvements in that respect, it is too early too judge whether this shift in emphasis will succeed.52

68. While there is no such intentional bias in internal guidelines, the fact remains that most of the concerns expressed on Pakistan’s UFR requests by internal reviewers and in the record of Executive Board meetings were related to the lack of ambitiousness of the programs, in relation to both macro-economic and structural aspects, not to their lack of realism or difficulty to implement.

69. A related problem is staff turnover, which accelerated dramatically in the 1990s.53 While Pakistan’s experience is broadly in line with recent Fund-wide practice (see Chapter 4 of Part I), it has led the authorities to feel that significant amounts of time are wasted, in the conduct of negotiations, as new mission members gain familiarity with the issues at stake.

70. Finally, Pakistan officials noted that IMF Resident Representatives played a very useful role, and they were generally very complimentary about their inputs. However, they felt that the Resident Representatives’ better understanding of “ground realities” was not sufficiently taken into account in policy formulation at IMF headquarters.

Several internal guidelines were not fully implemented

71. Over time, the IMF has developed a series of internal guidelines aimed at ensuring an efficient use of its resources, either specific to prolonged users or applicable to all UFR cases (see Chapter 3 of Part I for a detailed discussion). While it is unlikely that a full implementation of these guidelines would have avoided Pakistan’s prolonged UFR, or would have had a dramatic impact on program effectiveness, two procedural slippages are nevertheless worth mentioning.

72. First, general guidelines on the justification of the level of access and the capacity to repay the IMF call for particularly strong motivation in cases involving prolonged use and/or a poor track record. In practice, such discussions have been largely superficial in successive UFR reports on Pakistan, even though approved access was always on the generous side of the range until the end of the period and increased moderately over time instead of decreasing (see Figure 9.7).

Figure 9.7.Pakistan: Evolution of Access in IMF Arrangements

(Average annualized access in percent of quota)

Source: IMF Policy Development and Review Department.

73. Second, the guidelines specific to prolonged users, as endorsed by the Executive Board in 1984 and 1991, do not appear to have been fully adhered to. These guidelines require a proactive use of program design to ensure strong implementation (track record requirements, prior actions, back-loading of disbursements, diminishing access, etc.), along with a critical appraisal in staff reports of performance under previous programs and an analysis of the reasons why their objectives were not achieved. As noted earlier, candid ex post assessments were undertaken only rarely.

Conclusions and Suggested Remedies

74. The various factors discussed in this report are present to some extent in many IMF-supported programs, and it is clearly their combination, more than the isolated influence of any single one of them, that resulted in a limited effectiveness of programs and a consequent prolonged use of IMF resources. Moreover, the IMF was clearly faced with an extremely difficult situation in Pakistan, and one where the ability of any external agency to achieve a better out-come would be limited. In the absence of any way of testing a counterfactual, it is impossible to state with certainty that any alternative course of action by the IMF would have led to a better outcome. However, the following lessons are worth pointing out. These issues are discussed in greater depth in Part I.

Program design was affected by pressures to overpromise and downplay risks

75. A number of factors led IMF staff and the Pakistan authorities to agree to programs that were overoptimistic in their assumptions and that favored quick fixes over more difficult but essential reforms. At the stage of program implementation, these pressures led to actions that sometimes met the letter but not spirit of the agreements. In other cases, the agreed policies were only partially implemented, in a way that turned out to have adverse side effects (e.g., reducing tariffs with no replacement tax in place), while other critical steps (e.g., the strengthening of domestic tax collection) were not insisted upon. A more consistent and proactive use of conditionality, track record requirements, and other monitoring devices might have improved the outcome of successive programs. But beyond that, to address these problems in future IMF programs, each of the following factors deserves some attention:

  • The pressures for an IMF-supported program to produce “success” in a very short time frame: it is counterproductive to try to cram long-term reforms into an unrealistically short time frame, especially with regard to the implementation of complex reforms that may be central to longer-term sustainability. In some cases, this might call for more realistic assessments of the possible speed of adjustment. In this context, more thought needs to be given to the question of how to deal with the implementation of long-term institutional reforms, such as reform of the tax system/administration, that stretch beyond the time period of programs and where, by their nature, it is often not possible to condense the needed action into only a few concrete measures. This is true whether these reforms are critical to macro-economic stability (i.e., clearly related to a core IMF responsibility) or critical to future growth/poverty reduction in some other way that is less clearly an IMF core responsibility (e.g., public enterprise reform and other forms of institutional reform). The Pakistan case illustrates that just requiring a law to be passed, for example, is not sufficient. On the latter aspects, an important lesson is the importance of the IMF engaging with partners such as the World Bank and others to ensure that all such reforms are given appropriate external support (as a better option than trying to expand the IMF’s own mandate).
  • Pressures to agree a “seal of approval” for other financing. This raises questions about whether the seal of approval could be given in another way, and the risks of devaluing the seal of approval if these pressures lead to poor quality IMF-supported programs.
  • The perception, at least, that political influences on the IMF determined the outcome: eliminating political considerations altogether would not be realistic, since the IMF as an institution should respond to its shareholders, whose views should be taken into account in difficult cases where judgments as whether or not to proceed are finely balanced. But such political judgments should be made in as transparent a manner as possible, that is, at the level of the Executive Board, which is accountable for them, and should be distinct from technical assessments by the staff. Candid staff reports, with a clear discussion of implementation risks and their consequences, is one way of ensuring such transparency.

Programs did not always focus on the right issues

  • Tailoring programs as closely as possible to country-specific circumstances would be instrumental in “getting the program design right,” that is, (i) building as much as possible on domestic policy formulation, based on local expertise, to determine what reforms need to be made, in what sequence, what is feasible in a given time frame, and what milestones would constitute meaningful benchmarks to measure progress (in the spirit of the strategy followed for tax administration re-form in the recent PRGF); (ii) showing flexibility in the face of unexpected developments: either exogenous shocks, or when the policies agreed upon do not produce the intended/expected results. That implies being prepared to question the validity of initial assumptions and to communicate about revisions made necessary by initial design flaws in a way that clearly distinguishes them from policy slippages. Meaningful sensitivity analyses ex ante could help prepare adequate contingency plans.
  • More attention should be devoted to debt sustainability issues. The debt crisis faced by Pakistan in late 1999/2000 rightly, but belatedly, led IMF staff to focus on that issue. Recent initiatives to bring more discipline and consistency to sustainability assessments are a further step in the right direction.

Ownership matters enormously, and should be linked to greater selectivity

76. This issue goes beyond ownership of the program by the immediate economic team. Had the IMF insisted more on the presence of strong ownership and devoted greater efforts to promoting it, ex ante design flaws might have been reduced by a more substantive negotiation process, as the authorities would have refused to commit themselves to under-takings viewed as unrealistic or ill-suited to the needs of the country. Similarly, ex post side effects could have been mitigated by stronger and more consistent implementation. Political and institutional factors would have still existed, but they would have become less critical.

77. This approach also implies greater selectivity: the IMF should refrain from providing resources in support of a program that is not genuinely owned by the authorities and when there is not a strong commitment to a core set of necessary adjustment and reform measures. This can imply either fewer IMF-supported programs or more focused programs. It also implies dedicating more resources and attention to assessing implementation capacity (of which ownership is a strong component) and political feasibility, and being more candid about uncertainties and downside risks in documents submitted to IMF management and to the Executive Board. In retrospect, some of the programs with Pakistan would probably not have been entered into had these precautions been taken (especially considering that, in the absence of candid assessments of the implementation risks, the Board generally regretted that those programs were not more demanding). Clearly, greater selectivity in those circumstances might have implied a worsening of economic conditions in the short run, perhaps to the point of a full-blown crisis, and such implications are not to be taken lightly. However, such an evolution might have been more conducive to the strengthening of ownership needed for adjustment to take root than the provision of new financing in a context where it was de facto little more than a device to postpone hard decisions. Clearly, such a judgment is easier to make in hindsight, but this does not mean that the tradeoff should not have been faced in those terms at the time.

IMF surveillance did not act as a second opinion on program design and performance

78. This raises the issue of whether the surveillance function can or should be separated from program design and monitoring. At the very least, care must be taken in the relationship with the member country to avoid being trapped in a narrow program perspective. To that end, surveillance should be used as a tool for stepping back from time to time, to take a broader strategic look and to reexamine vulnerabilities, assess alternative strategies, and foster a debate on key reforms that can help build the necessary consensus.

Appendix 1 An Illustration of Ownership and Political Feasibility Assessment in the Context of the 1993/94 Programs Through the Use of Basic Political Science Tools

The developments below are drawn from a paper prepared at the IEO’s request on “Political Science Tools for Assessing Feasibility and Sustainability of Reforms.”54 They are meant to illustrate the type of analyses that might have been undertaken by IMF missions to make a methodical assessment of the authorities’ ownership of the programs supported by the IMF and of their ability to implement them. There are essentially three different tools: (i) stakeholder analysis, which focuses on the balance of power, policy preferences, and modes of interaction of key individuals and interest groups having a stake in the decision-making process; (ii) institutional analysis, which looks at the institutional dynamics of this process, including the identification of institutions in a position to veto certain reforms, along with an analysis of capacity constraints; and (iii) “Delphi” study, which involves seeking the views of an expert panel in a systematic manner on each of the relevant dimensions of the political economy setting. Each of these tools can be used in three different modes, namely (i) trend extrapolation; (ii) impact analysis; and (iii) scenario building. Scenario building and “Delphi studies” are not meaningful when applied retrospectively. Therefore, they are not discussed in the following presentation.55 The overall analysis summarized below and the characterizations of the Pakistan politico-administrative system contained therein are those of Wimmer and others (2002) and do not necessarily reflect the views of the IEO.

Stakeholder Analysis

Trend extrapolation

  • Reforms under way. Since coming to power in 1990, the government of Nawaz Sharif favored economic liberalization and had already launched a deregulation program strengthening the private sector. The budget deficit had been brought down (by cutting expenditure in health and education and by reducing public works programs). However, there were no plans to introduce an agricultural tax or increase tax revenues in general. Looking at the measures already undertaken, the government thus seemed to have the political will to continue the process of reforms, and it seemed to enjoy a large enough political basis both at the national and provincial levels to do so. But the existing reform trend was clearly selective and avoided important areas that would have touched the entrenched interests of groups on whose political support the government depended (such as the Jamoori Ittahad coalition—IJI) in Punjab, with its important landowners.
  • Decision-making style. Political decision making was mostly concentrated in the higher echelons of the government, with major reform steps being decided upon by the inner circle of a relatively isolated group of decision makers and then left to the respective ministers to implement. Despite political rivalries, there was agreement between the main political parties about the necessity to continue reforms. Indeed, Benazir Bhutto, the leader of the opposition in 1993, had declared she would not reverse the process of privatization if she came back to power, and she did not, but under her government decision making followed the same pattern and was likewise restricted to a very small group of advisers. Given this structure of decision making, a broad ownership of the reforms had certainly not yet developed.
  • Attribution of agency. IMF-sponsored reforms were generally presented as a bitter pill that the country was forced to swallow by a powerful outsider. Implicitly and sometimes explicitly, however, the message was that the pill would not be consumed as bitter as it looked at the time of the negotiations, given the apparently widespread assumption in the informed public that lending was politically motivated (i.e., the reward of the political alliance with the United States). The style of communication on the reforms thus does not show signs of genuine ownership by the major political forces.

Impact analysis

The implementation of the program would not have affected the power balance between the army, the prime minister, and the president, given the generous treatment of military expenditures in the proposed agreement. However, the other parts of the political equation would have changed quite a bit. Taxation of agricultural incomes, one of the cornerstones of the agreement, would have seriously reduced the support of the government by the IJI coalition. The increase in indirect taxes may have heightened public discontent and may have strengthened opposition parties. Thanks to the broad agreement between the main parties on the general direction of reforms, this would perhaps not have stopped them, but it would certainly have brought about increased pressures for softening the consequences for the public and taking tax reforms back. The reform of credit-awarding mechanisms would have seriously limited the capacity of government employees to distribute credit along the lines of political patronage, which may have weakened the support of the bureaucracy, traditionally regarded as another important power center in Pakistan. In sum, it seems that effective implementation of the reform program would have shaken at least part of the political basis of the regime and it is doubtful whether it would have survived a comprehensive enforcement of reforms in the tax and financial sectors.

Institutional Analysis

  • Veto point analysis. The institutional position of the government was restricted by two factors inherited from military rule: (i) the constitution allowed the indirectly elected president to dismiss the directly elected prime minister and his government; (ii) the army was still the most important veto actor that could influence all government decisions (as illustrated by the exclusion of the defense budget from any expenditure cuts undertaken under most IMF-supported programs). Besides these two institutional actors, the bureaucracy and the national assembly (MNA) would have been two other important veto players. Given the strong representation of landed interests within these groups, it was highly unlikely that any law would have been enforced that would seriously introduce taxation on agricultural income. Large parts of the bureaucracy would also not have been in favor of privatization programs that would have implied serious cuts in their domains of political influence.

Figure 9.8.Political Economy Analysis Toolbox

Note: Tools in italics are those whose conclusions are reported above.

  • Implementation capacity. Despite the long tradition of military rule, Pakistan showed a serious weakness of its law and policy enforcing authorities. Widespread corruption, clientelistic practices, and recruitment procedures based on patronage explain at least in part the difficulties in enforcing basic rights and duties in different areas. Most ministries use the large discretionary powers that these deficits imply in order to build up their own network of patronage relationships and therefore enhance their standing in the all-embracing web of political alliances. A good example of this is the way that the Central Board of Revenues interprets the myriads of exemptions in tax law on a case-by-case basis. The situation in public enterprises and in the banking sector was comparable. Every government would face the problem of such deficient implementation capacities, a weakness that would also have an impact on the design of the reform programs. The rise of indirect taxes, for instance, would be favored in order to circumvent the enforcement problems of direct taxation.

Overall Assessment

Serious doubts about the prospects of future implementation of the program would have had to be raised. The decision-making coalition endorsing the reform was not broadly built. It did not include large sections of the public in order to counterbalance the possible loss of support that effective implementation would have brought about from the power base of the current regime and from within the administration.

Appendix 2 Pakistan: List of People Interviewed in Connection with the Evaluation of Prolonged Use of IMF Resources

Senior Officials

Mr. Mueen Afzal, Secretary General Finance

Mr. Shaukat Aziz, Ministry of Finance

Mr. A.R. Chugtai, Deputy Governor, State Bank of Pakistan

Mr. Ishrat Husain, Governor, State Bank of Pakistan

Mr. Ashfaque H. Khan, Economic Adviser, Ministry of Finance

Mr. Yunis Khan, Finance Secretary

Mr. Mushtaq Malik, Joint Secretary (External Finance), Ministry of Finance

Mr. Riaz Ahmad Malik, Chairman, Central Board of Revenue

Mr. Altaf M. Saleem, Minister for Privatisation

Mr. Murtaza Ahmad Shaikh, Special Assistant to Deputy Chairman, Ministry of Planning & Development

Dr. Abdul Naseer, Economic Adviser, State Bank of Pakistan

Former Senior Officials

Mr. Aitzaz Ahsan, former Minister of Law and Interior

Mr. Qazi M. Alimullah, former Finance Secretary

Mr. Sartaj Aziz, former Minister of Finance

Mr. H.U. Beg, Chairman, Ad-hoc Public Accounts Committee, and former Finance Secretary

Mr. Mushahid Hussain, former Minister for Information

Mr. Fakhar Imam, former Minister

Mr. Vaseem A. Jafarey, former Governor of State Bank of Pakistan and Adviser to the Prime Minister

Mr. A.G.N. Kazi, former Governor, State Bank of Pakistan

Mr. M. Farooq Leghari, former President

Mr. Saeed Qureshi, former Secretary General Finance


Mr. Akhtar A. Hai, Senior Research Economist/Associate Professor, Applied Economics Research Centre, University of Karachi

Dr. Akmal Hussain, Economist, Syeed Engineering

Dr. A.R. Kamal, Director, Pakistan Institute of Development Economics

Banking Sector

Mr. S. Ali Raza, President & Chief Executive Officer, National Bank of Pakistan

Mr. Masood Karim Shaikh, Chief Financial Officer, National Bank of Pakistan

Mr. Zubyr Soomro, Chief Executive & Country Corporate Officer, Citibank

Business Community

Dr. Anwarul Haque, Secretary General, The Federation of Pakistan Chambers of Commerce & Industry

Mr. Sheikh Javaid, Chairman, The Federation of Pakistan Chambers of Commerce & Industry

Mr. Tahir Khaliq, Chief Executive, Chamber of Commerce & Industry, Karachi-Pakistan

Dr. Mohammad Zubair Khan, Managing Director, Financial Techniques Internationals (and former Minister of Commerce)

Mr. Haroon Rashid, Vice President, The Federation of Pakistan Chambers of Commerce & Industry


Mr. Farhan Bokhari, Pakistan Correspondent, Financial Times

Mr. Nadeem Malik, The News

Mr. M. Ziauddin, Resident Editor, Dawn

Mr. Arshad A. Zuberi, Deputy Chief Executive, Business Recorder

Mr. M.A. Zuberi, Editor, Business Recorder

Nongovernmental Organizations

Mr. Khadim Hussain, Senior Programme Officer, Action Aid Pakistan

Dr. Asad Sayeed, Social Policy and Development Centre

Trade Union Representatives

Mr. M. Zahoor Awan, Secretary General, All Pakistan Federation of Labour

Mr. Raja Khalique A. Khan, Vice President, Pakistan National Federation of Trade Unions

The mission also met with a large number of current and former IMF and World Bank staff involved with these institutions’work on Pakistan.

Appendix 3 Pakistan: History of Lending Arrangements
Amount agreedAmount drawn
FacilityDate of arrangementInitial date of expirationActual date of expiration1(In thousands of SDRs)Percent undrawn
1SBADec. 8, 1958Dec. 7, 1959Sep. 22, 195925,0000100
2SBAMar. 16, 1965Mar. 15, 196637,50037,5000
3SBAMay 18, 1972May 17, 1973100,00084,00016
4SBAAug. 11, 1973Aug. 10, 197475,00075,0000
5SBANov. 11, 1974Nov. 10, 197575,00075,0000
6SBAMar. 9, 1977Mar. 8, 197880,00080,0000
7EFFNov. 24, 1980Nov. 23, 19831,268,0001,079,00015
8SAFDec. 28, 1988Dec. 27, 1991Dec. 15, 1992382,410382,4100
8SBADec. 28, 1988Mar. 7, 1990Nov. 30, 1990273,150194,48029
9SBASep. 16, 1993Sep. 15, 1994Feb. 22, 1994265,40088,00067
10ESAFFeb. 22, 1994Feb. 21, 1997Dec. 13, 1995606,600172,20072
10EFFFeb. 22, 1994Feb. 21, 1997Dec. 13, 1995379,100123,20068
11SBADec. 13, 1995Mar. 31, 1997Sep. 30, 1997562,590294,69048
12PRGFOct. 20, 1997Oct. 19, 2000682,380265,37061
12EFFOct. 20, 1997Oct. 19, 2000454,920113,74075
13SBANov. 29, 2000Sep. 30, 2001465,000465,0000

If different from initial date of expiration.

If different from initial date of expiration.


In keeping with the IEO’s terms of reference, which prevent it from commenting on ongoing operations, the PRGF arrangement approved in December 2001 is not within the scope of this review.


Lessons and recommendations are elaborated in greater detail in Part I.


A full list of people outside the IMF interviewed by the IEO is provided in Appendix 2.


In fact, as the subsequent discussions will show, the achievements of that period hinged upon the buildup of partially disguised debt vulnerabilities.


Figure 9.1 provides an overview of this history.


These arrangements were supplemented by rather large drawings under special facilities (the Oil Facility and the Compensatory Financing Facility for export shortfalls).


This count does not include the PRGF arrangement approved in late 2001, which, as an “on-going operation,” must be considered outside the scope of this evaluation.


In addition, over that period, Pakistan had access to sizable resources under special facilities.


Not taking into account precautionary arrangements, that is, those where the authorities indicate at the outset that they do not intend to avail themselves of the resources committed under the program.


The weighted average rate went down from 65 percent in 1988 to 19 percent in 2000. Pakistan is rated 7 on the IMF’s 10-point index of trade restrictiveness (where 10 is the most restrictive), that is, below India (10), but well above China and Sri Lanka (5) or Indonesia, Malaysia, and the Philippines (4).


In theory, broad consumption- or income-based taxes should be more efficient than trade taxes, but the problem in Pakistan is that these taxes have relatively high rates to compensate for the lack of broad base: in 2000, there were less than 100,000 registered GST taxpayers and under 2 million income taxpayers (compared to 1 million in 1990), in a population of roughly 140 million.


See, for instance, “Economic Reforms and Macroeconomic Management in Pakistan (1999–2001)” by Ishrat Husain, Governor of the State Bank of Pakistan.


Admittedly, changes in the legal framework of the central bank’s operations do not necessarily imply that its de facto independence increased. Indeed, one former Governor of the SBP told the IEO that he had been able to perform his functions in great independence at a time when no institutional safeguards guaranteed it. However, there was broad agreement among both IMF staff and Pakistan stakeholders that the SBP was significantly strengthened during the 1990s.


However, former IMF mission chiefs for Pakistan pointed out that there were few cases where the authorities had approached them with viable policy alternatives. When they did, particularly from late 1998 onward, their proposals were often incorporated into the program framework (e.g., as concerns the exit strategy from the foreign currency deposit freeze or as regards the exchange rate regime).


From 1990 to 2000, Pakistan received 40 IMF technical assistance missions, with over half of them concerning fiscal issues and about a quarter on monetary and banking issues. During that period, it also benefited from the presence of resident advisors for a cumulative total of five years. Furthermore, since FY1996, IMF TA to Pakistan has represented the equivalent of 13 staff years.


Examples of such problems include the 1992 and 1997 TA reports on tax administration reform or the 1999 TA report on modernizing the income tax system. Numerous reports were also prepared on the reform of indirect taxation, with limited concrete results until the last couple of years.


This dimension was frequently emphasized in performance assessments done by the IMF. For instance, the 2000 Article IV staff report notes that “Pakistan has had a series of adjustment and reform programs supported by Fund arrangements during the past decade. Policy implementation and economic performance have been disappointing…. This weak performance stems in large part from the failure of successive governments to carry through sustained reforms…”.


The starting point of projections is occasionally off the “actuals” line because of subsequent revisions to the data on which the projections were based.


A justification for this reluctance is provided by the debt sustainability concerns that surfaced at the end of the 1990s. But, more than the level of the target itself, what appears to have hindered fiscal adjustment most is the stop-go process stemming from the inadequate macroeconomic framework, with the slippages that occurred during off-track periods more than undoing any progress previously made.


These projections were also more optimistic than contemporaneous WEO forecasts for developing countries as a whole (9 percent) and for Pakistan’s WEO country group (i.e., Asia: 10 percent).


In multiyear arrangements approved since 1993, the average targeted revenue increase over the three-year program period was 2.2 percentage points of GDP in Pakistan, compared to 0.7 points for prolonged users as a whole and 1.3 points for “temporary” users. Overall and primary fiscal balance targets were also more ambitious in Pakistan than in both “temporary” and prolonged users’ programs on average. (Source: MONA.)


The logic of the argument is the following: building the budget on the basis of overly optimistic revenue projections avoids making tough decisions on expenditure at the time of the budget debate, while the deficit that inevitably arises when revenue expectations fail to materialize makes it easier for the Finance Ministry to impose a degree of expenditure restraint that would have been unacceptable before. (Nonetheless, the cuts adopted in that context may not be—and in fact often were not—optimal in terms of economic efficiency.)


For instance, effective taxation of the agricultural sector was expected to be put in place within one year and the implementation of a broad-based GST within two years. In both cases, after ten years full effectiveness had yet to be achieved.


In theory, a shift from trade taxes to broad-based domestic taxes, such as the VAT, should be revenue enhancing since the latter taxes would typically also include traded goods in their base. However, in Pakistan, this effect largely failed to materialize because of severe weaknesses in tax administration and of generous and widespread tax exemptions, which persisted in spite of specific conditionality to that effect in each of the programs.


In particular, in the form of multiple tax rates, cascading sales taxes, multiple withholding taxes, and complex excises.


For instance, an internal country strategy paper prepared in early 1993 noted, in drawing the lessons from past experience of IMF-supported programs in Pakistan, that “in sequencing the program measures, primary emphasis needs to be placed on fiscal consolidation and reform, taking full account of the budgetary costs associated with trade and financial sector reforms” (emphasis added).


To some extent, this criticism reflects the application of current standards to past programs. Program documents typically did say that these aspects were important, but they put little emphasis on the detailed monitoring and implementation of deeper institutional reforms, in keeping with the IMF’s own institutional culture at the time. Moreover, a number of these issues were rightly identified as being the primary responsibility of the World Bank but, as will be discussed in the section “Some Program Design Problems Were Rooted in Deeper IMF Governance Issues” below, this identification of the division of labor between the two institutions did not lead to an effective operational approach to ensure that the issues were addressed.


The fact that the achievement of revenue projections hinged critically on improvements in tax administration was repeatedly underscored in internal memoranda by the Fiscal Affairs Department, which also supplied abundant technical assistance in that area and frequently expressed concern that their recommendations did not appear to receive consistent follow-up. At the same time, they emphasized in their comments that improvements in tax administration should not be strongly relied upon to deliver large and quick increases in tax revenue. This might explain why improvements in tax administration were not aggressively pursued through conditionality until the most recent programs.


By contrast, at the administrative level, there was a considerable degree of stability at the top of the two main institutions involved, namely the central bank and the Ministry of Finance.


This problem was recognized in the 1993 country strategy paper, which highlighted as the first lesson to be learned from past experience that “it is important that the policy dialogue involve, at an early stage, full commitment at the highest political levels in Pakistan.”


One telling anecdote heard during the evaluation team’s mission to Pakistan illustrates how a long series of weakly owned programs can lead to perverse results: when the IMF suggested including in the program supported by the current PRGF arrangement key elements of the tax administration reform prepared by the authorities, some government officials initially resisted on the grounds that these reforms were ones they truly wished to implement, whereas including them in the LOI would give the impression they were IMF-imposed and hence would reduce their chances of implementation.


The 2000 SBA was adopted only upon completion of an unusually long list of prior actions, going beyond the unfulfilled commitments of the preceding interrupted program, and its disbursements were slightly back-loaded. The current PRGF arrangement was approved only after one year of satisfactory performance under a Stand-By Arrangement.


End-1993 targets, however, were reportedly met, which allowed the request for an EFF/ESAF to be presented to the Board and approved in February 1994.


For a more comprehensive review of attempts to tax agriculture in Pakistan and impediments thereto, see Khan and Khan (1998).


In Stand-By Arrangements during the 1990s, the amount disbursed upon approval by the Board was on average 12 percent higher than what would have resulted from a division of the total amount committed under the arrangement into tranches of equal size.


Over that period, Pakistan was granted 18 waivers of compliance, of which 8 concerned quantitative PC, another 8 concerned structural PC, and 2 continuous PC. In all three categories, half of the waivers concerned fiscal PC (two-thirds if PC on utilities prices are treated as “fiscal” conditionality). All the waivers on quantitative PC were requested for reasons other than “minor technical deviation” or “exogenous shocks.”


Somewhat longer intervals occurred at the beginning and end of the decade, and coincided with episodes of major political dis-order: from December 1989 to December 1991 and from June 1999 to November 2000.


In this connection, it has been suggested to the IEO by some IMF staff members that the program design weaknesses discussed in the section “The Lack of Effectiveness of Pakistan’s IMF-Supported Programs Stems from Both Design and Implementation Issues,” such as unrealistic macroeconomic projections, the pretence of toughness, and so on were symptomatic of attempts to find a face-saving way to justify continued lending to Pakistan.


Examples of such difficulties include tax reform, where setting conditions on the enactment or parliamentary approval of given measures failed to ensure their actual implementation (e.g., for agricultural taxation, where the first legal changes were approved in 1994, but full implementation was an unmet structural benchmark (SB) as of mid-2001, or for the extension of the GST to traders/retailers, where the timetable was similar); improvements in tax administration, where conditionality on the creation of a given administrative structure failed to ensure that it performed effectively the tasks it had been set up for (e.g., tax audit unit, an SB for 1997); or yet again privatization, where conditionality on “bringing to the point of sale” specified public enterprises, does not guarantee that they will ever be privatized (as was done in 1996/97 for companies in the oil and gas sector, which for the largest part had yet to be sold by end-2000). In such a context, an alternative is to specify the condition in terms broad enough to give staff room for judgment in appraising the extent of progress made at the time of the program review. However, evidence from other case studies (especially the Philippines) suggests that so-called “review” conditionality is no panacea.


Presumably because of a change of strategy, which, from 1993 onward, focused on privatizing these enterprises.


The World Bank diagnosed them in 1998 in the following terms: “Public enterprises suffer from operational inefficiencies, overstaffing, inappropriate incentives, misdirected investments, inadequate O&M [operation and maintenance], political intervention in their decision making, and other problems which adversely affect their performance… [they] have become a vehicle for employment creation, political patronage and corruption…. Their inefficiency has resulted in high costs to the private sector as a result of high prices of utilities… [and] in many cases poor quality of supply. . . . Problems are most acute in the power sector.” (“Pakistan Public Expenditure Review,” 1998, pp. 15–16.)


Specific conditions included: introducing performance improvement arrangements between the government and WAPDA (by 12/97); developing action plans for restructuring seven major public enterprises (by 6/98); strengthening the National Electric Power Regulatory Authority (by 6/98); reconciling and settling electricity bills of federal and provincial governments (prior action, 1999); and implementing power sector restructuring program (by 3/99).


See, for instance, Birdsall and others (2001).


Pakistan’s average outstanding obligations over 1988–2000 were SDR 904 million; annual repayments and repurchases averaged SDR 148 million over the same period.


The quality of surveillance over that period was assessed by systematically rating the performance of each surveillance report for nine functions viewed by the IEO as “key elements” of surveillance in a program context (see Chapter 6 of Part I for a discussion of how these elements were identified). These nine functions are (i) provision of realistic medium-term and alternative scenarios; (ii) provision of meaningful sensitivity analyses; (iii) discussion of risks to the assumptions and projections; (iv) discussion of the risks and impact of policy slippages and of vulnerabilities; (v) balanced reporting of the authorities’ views, including any significant differences with staff; (vi) cogent presentation of proposed policy course; (vii) discussion of policy alternatives and trade-offs; (viii) critical and frank review of previous UFR performance; and (ix) presentation of collaboration/interaction with the World Bank. The diagnosis is consistent with the findings of the 2002 “Biennial Review of Surveillance.” In other words, the Pakistan exception lies not so much in this absence of strong, independent role for surveillance as in the length of time it lasted.


In particular, several research papers of the early 1990s (see, for instance, Haque and Montiel, 1992) emphasized that in the Pakistan context, a deficit reduction strategy emphasizing cuts in public investment would have large output costs, because of lower public and private capital stocks, “since the smaller public capital stock would have depressed private investment as well. Crowding-in through lower interest rates does not materialize in this case because the lower public capital stock actually represents a substantial negative supply shock.” (pp. 9–10.) Interestingly, the 1993 country strategy paper for Pakistan explicitly acknowledged the findings of that paper. To be sure, it could also be argued that in a corrupt environment, the supply impact of public capital expenditure is limited and, therefore, reducing capital expenditure makes sense in the short term, provided public expenditure management reforms are undertaken at the same time.


As noted earlier, more analysis was undertaken internally—for example, in the late 1980s by economists from the IMF Re-search Department. However, their work was not approved for publication as an IMF Working Paper until late 1992, apparently because of concerns about country sensitivity. This paper (Haque and Montiel, 1992) made it clear that “although the average cost of servicing this debt may have been low in the past, the marginal cost of debt service can be expected to rise in the coming years, since, as a result of the opening up of the economy, international interest parity is likely to prevail. . . . The high level of government indebtedness implies that debt servicing has the potential to frustrate future deficit reduction plans.” Under assumptions of constant revenue to GNP and primary expenditure to GNP ratios and with a growth rate constant at 5.8 percent (i.e., much higher than it turned out to be), they found that the deficit to GNP ratio would rise to 8 percent over five years just through increased debt-servicing costs. Under similar GNP growth assumptions, they also found that an average deficit of 4 percent of GDP over the next five years would be consistent with lower macroeconomic imbalances (in fact, the fiscal deficit averaged 7 percent of GDP during the 1990s).


The deposits, after being surrendered to the SBP, were effectively on-lent to the government and therefore did not result in a corresponding buildup in reserves. Hence the uncovered foreign exchange exposure. FCDs increased at a rapid pace starting in 1991, when they were first allowed for residents, partly reflecting various price incentives for both banks and depositors, partly owing to the advantages of their “no questions asked” status.


For instance, the only shock analyzed in the sensitivity analysis attached to the 1997 ESAF/EFF was a 1 percent lower than projected cotton exports.


Even internal documents, however, did not go beyond the stage of taking note of past implementation difficulties. The only effort to take stock of past experience and draw strategic orientations for future IMF involvement was done in a 1993 country strategy paper (CSP), but its close links with the briefing paper for the negotiation of what became the 1994 EFF/ESAF limited its “strategic” value, and the lessons it drew from past experience did not appear to be consistently reflected in the design of subsequent programs. No other CSP was prepared since then, even after CSPs were given new impetus in 1997 through guidelines aimed at making them a prime vehicle for staff to step back from the contingencies of program negotiations, learn lessons from past country experience, and design an optimal strategy for the IMF’s involvement in the country.


This focus itself is in large part a reflection of the legal framework of programs, which involves cumbersome and—from the authorities’ standpoint—embarrassing procedures in case of deviations from targets, no matter how small.


The first PRGF for Pakistan was approved in late 2001, and the interim PRSP was finalized not long before the IEO mission to that country. As such, both are beyond the scope of this evaluation.


Pakistan had six different mission chiefs and nine different desk economists over 1990–2000. Only two of the desk economists subsequently became mission chief. However, some greater continuity was provided by the fact that the Department Director, or another senior staff member, typically participated in key aspects of the discussions.


“Political Science Tools for Assessing Feasibility and Sustainability of Reforms” by Professor Andreas Wimmer, Director of the Department of Political and Cultural Change at the Center for Development Research, University of Bonn, with Indra de Soysa and Christian Wagner. This paper is available on the IEO’s website at


A schematic presentation of these tools and modes is provided in Figure 9.8.

    Other Resources Citing This Publication