Annex I The Prague Framework for Private Sector Involvement32
65. The Committee endorses the report by the Managing Director on the involvement of the private sector in crisis prevention and management. It welcomes the progress on developing a framework for involving private creditors in the resolution of crises. The Committee notes that this approach strikes a balance between the clarity needed to guide market expectations and the operational flexibility, anchored in clear principles, needed to allow the most effective response in each case. The Committee notes that Fund resources are limited and that extraordinary access should be exceptional; further, neither creditors nor debtors should expect to be protected from adverse outcomes by official action.
66. The Committee agrees that the operational framework for private sector involvement must rely as much as possible on market-oriented solutions and voluntary approaches. The approach adopted by the international community should be based on the IMF’s assessment of a country’s underlying payment capacity and prospects of regaining market access. In some cases, the combination of catalytic official financing and policy adjustment should allow the country to regain full market access quickly. The Committee agrees that reliance on the catalytic approach at high levels of access presumes substantial justification, both in terms of its likely effectiveness and of the risks of alternative approaches. In other cases, emphasis should be placed on encouraging voluntary approaches, as needed, to overcome creditor coordination problems. In yet other cases, the early restoration of full market access on terms consistent with medium-term external sustainability may be judged to be unrealistic, and a broader spectrum of actions by private creditors, including comprehensive debt restructuring, may be warranted to provide for an adequately financed program and a viable medium-term payments profile. This includes the possibility that, in certain extreme cases, a temporary payments suspension or standstill may be unavoidable. The Fund should continue to be prepared to provide financial support to a member’s adjustment program despite arrears to private creditors, provided the country is seeking to work cooperatively and in good faith with its private creditors and is meeting other program requirements. The Committee urges progress in the application of the framework agreed in April 2000, and in further work to refine the analytical basis for the required judgments, and it looks forward to a progress report by its next meeting.
Annex II An Overview of Past Exceptional Access Programs
67. This annex considers experience with past exceptional access cases, to provide a factual basis for consideration of possible changes in access policy. The cases examined are Mexico (1995), Thailand (1997), Indonesia (1997–2000), Korea (1997), Russia (1998), Brazil (1998), Turkey (2000–2001) and Argentina (2001). These eight cases constitute most of the Fund arrangements during the period 1995-2001 where the Fund invoked exceptional circumstances and/or total access substantially exceeded normal limits.33
Annex III Sustainability Analysis of Past Exceptional Access Programs
88. Only two initial program documents of the eight exceptional access cases contained sensitivity analyses of the medium-term projections. The main risks to the programs were viewed in terms of dichotomous events, such as rollover or no rollover of short-term debt. The consequences of such events on key sustainability indicators, however, were not examined in detail. Applying a systematic sensitivity analysis at the time of program approval could have illustrated the risks in a more comprehensive way and might have led to a reassessment of vulnerabilities.
89. A modified sustainability framework, including a standardized sensitivity analysis, has been developed recently.43 The sensitivity analysis comprises six basic tests and two composite tests. The basic tests assess the impact on the external debt-to-GDP ratio of (i) interest rates, GDP growth, U.S. dollar GDP deflator, and non-interest current account balance at the average of the ten years preceding the crisis rather than the forecasted levels, (ii) a one-off depreciation of the exchange rate by 30 percent, and changes by two standard deviations from the average of the ten years preceding the crisis of (iii) interest rates, (iv) GDP growth, (v) U.S. dollar GDP deflator, and (vi) the non-interest current account balance. The composite tests apply a combination of one-standard deviation shocks, and country-specific standard deviation shocks.
90. A retrospective sensitivity analysis undertaken for six out of the eight cases is illustrated in Figure 9.44 45The charts show the baseline program projections for the external debt-to-GDP ratio, actual outcomes, and the range of possible outcomes indicated by a sensitivity analysis. The most significant risks stemmed from sharp depreciations of currencies (only in the case of Mexico was the main risk a negative shock to the current account). In most cases, these were not anticipated to the extent they actually occurred, and impaired the containment of external debt-to-GDP ratios.
In Mexico, the baseline scenario at the time of program approval indicated that the debt-to-GDP ratio would increase by about 5 percentage points of GDP in three years (after a jump of 7½ percentage points in the crisis year), largely due to the depreciation of the currency. While the actual outcome three years after the crisis was close to forecasts, the standard sensitivity analysis indicates that the debt-to-GDP ratio could have risen by up to 20 percentage points of GDP (but could also have fallen by up to 10 percentage points).
The contrasting case is that of Indonesia. Projections in 1997 envisaged an increase by almost 9 percent in the debt-to-GDP ratio until 2000. The outcome, an increase by about 59 percent, could not have been captured by any of the standard tests. However, the test assuming a 30 percent depreciation could have highlighted the risk that materialized when the currency subsequently lost value rapidly, by forecasting an increase in the debt-to-GDP ratio of about 29 percentage points of GDP.
Not as extreme, but nevertheless disquieting is the case of Russia, where the debt-to-GDP ratio increased by 23 percentage points of GDP three years after the crisis, while projections had envisaged an increase of about 5 percent. Here, however, the actual outcome after three years lay within the range indicated by the sensitivity tests.
In Brazil, a substantial increase in the debt-to-GDP ratio was programmed; by end-2001, the ratio was expected to increase by almost 9 percentage points of GDP. The sensitivity analysis assuming a 30 percent depreciation indicated, however, that the ratio could increase by up to 19 percentage points. The actual outcome, of 11½ percentage points, was lower than this result, largely because the slower-than-anticipated build-up of external debt mitigated the impact of a strong depreciation.
In Argentina, projections in 1999 indicated that the external debt-to-GDP ratio would rise by 2 percentage points of GDP, with a sensitivity analysis indicating that the rise could be as much as 27 percentage points of GDP in the event of a devaluation by 30 percent. A scenario that would have been considered more likely at the time, which assumed lower-than-projected real GDP growth, pointed to an increase in the debt-to-GDP ratio of about 10 percentage points, close to the actual outcome of 8-percentage points.
Projections in 1999 for Turkey indicated that the debt-to-GDP ratio would rise by 11 percentage points by 2001; a sensitivity analyses indicates that the increase could have been as high as 42 percentage points if a combination of shocks to the exchange rate, interest rates, and real GDP growth occurred. The actual outcome, and increase of the debt-to-GDP ratio by 30 percentage points would thus have been captured in the sensitivity scenarios.
See Sovereign Debt Restructuring Mechanism -- Further Considerations (forthcoming); Collective Action Clauses in Sovereign Bond Contracts—Encouraging Greater Use, June 7, 2002 (SM/02/175); The Design and Effectiveness of Collective Action Clauses, June 7, 2002; (SM/02/173) and Fund Policy on Lending into Arrears to Private Creditors—Further Considerations of Good Faith Criterion (forthcoming).
Review of Access Policy in the Credit Tranches and Under the Extended Fund Facility August 8, 2001 (EBS/01/133).
“Exceptional” access is defined in this paper as access above normal limits that apply under the credit tranches and EFF.
See Ghosh, Lane et. al., IMF-Supported Programs in Capital Account Crises: Design and Experience, Occasional Paper No. 210, 2002.
This paper does not emphasize the use of such intermediate forms of PSI, often involving moral suasion, as their effectiveness has often been shown to be limited in recent cases.
GFN is defined as the current account balance excluding official transfer receipts, plus Fund repurchases, plus other amortizations of maturities in excess of one year, plus targeted arrears reduction, plus buildup of gross reserves.
In most cases, approved access as a share of ex post gross financing was higher since the capital account was poorer and current account adjustment greater than programmed. Indeed, the current account adjusted far more than was needed to achieve any reasonable measure of sustainability, and this excessive adjustment was brought about mainly through the effects of severe recessions on imports.
Some comparative analysis has been conducted of the case of Malaysia, which adopted similar policies to those in Thailand, Indonesia and Korea, but did not have Fund financial support for its adjustment efforts. However, the analysis has focused mostly on the effects of controls on capital outflows, which Malaysia adopted and the other countries did not. There were also differences in the initial position. In particular, Malaysia’s external debt at the outset of the crisis was only 17 percent of GDP, its initial fiscal position was strong, and the regulatory framework for the financial sector was relatively well developed. See IMF-Supported Programs in Capital Account Crises, op. cit., and Capital Controls: Country Experiences with Their Use and Liberalization, IMF Occasional Paper 190, 2000.
For a more comprehensive discussion of this question, see Lane and Phillips Does IMF Financing Result in Moral Hazard, IMF Working Paper No. 00/168, October 2000, and Kamin Identifying the Role of Moral Hazard in International Financial Markets, draft Federal Reserve Board, December 2001; Dell’ Ariccia, Gödde and Zettelmeyer Moral Hazard and International Crisis Lending: A Test, draft, November 2000.
The SRF can be used by “a member that is experiencing exceptional balance of payments difficulties due to a large short-term financing need resulting from a sudden and disruptive loss of market confidence reflected in pressure on the capital account and the member’s reserves, if there is a reasonable expectation that the implementation of strong adjustment policies and adequate financing will result, within a short period of time, in an early correction of such difficulties.” See Decision No. 11627-(97/123) SRF, December 17, 1997.
A member that is experiencing such pressures may decide not to draw available resources immediately (Brazil, 2001). While the current system allows the Fund to approve precautionary arrangements that provide for exceptional access, it may be worthwhile to consider, at a later stage, such arrangements explicitly (including precautionary SRF arrangements), and the implications for the CCL.
Assessing Sustainability, SM/02/166 (May 28, 2002).
The latter is a standard criterion for determining access.
Uniformity is the basic legal principle that governs all the activities of the Fund. This principle is based on the Articles of Agreement, which, with very limited exceptions (Article V, Section 8(c); Article V, Section 12(f) (ii) and (iii)), establish the same rights and obligations for all members. Moreover, some provisions in the Articles are specific in declaring that uniformity must be observed (Article II, Section 2, second sentence, on membership terms; Article V, Section 8(d), on charges; Article V, Section 9(a), on remuneration).
See Need as a Condition for Use of Fund Resources, SM/94/299, December 16, 1994.
Narrow GFN comprises current account deficits, excluding official transfers and amortization of medium and long-term loans (including Fund repurchases). Broad GFN includes, in addition to narrow GFN, reserve accumulation and arrears clearance. Augmented GFN is narrow GFN plus short-term debt outstanding. However, projections of short-term debt flows are highly uncertain.
The Articles of Agreement prohibit use of Fund’s resources to meet a large and sustained outflow of capital and allow the Fund to request a member to impose capital controls in order to prevent such use (Article VI, Section 1). Consistent with this, the programs with exceptional access are aimed at restoring confidence with the aim to stop capital outflows, rather than to finance capital outflows themselves. For further discussion of these issues, see Supplemental Reserve Facility, December 5, 1997, EBS/97/225.
The Emergency Financing Mechanism (EFM) provides for exceptional procedures that, in the event a member faced a crisis, would facilitate rapid approval of Fund support while ensuring the conditionality necessary to warrant such support. The EFM, however, does not necessarily cover all cases of exceptional access, as the trigger is the use of emergency procedures, not the level of access. The EFM has been used less frequently in recent crises as ongoing consultations have been the norm. See Summing up by the Chairman: Emergency Financing Mechanism Executive Board Meeting 95/85, September 12, 1995.
The Fund has conducted an unpublished review of the Mexico crisis, a study of the Asian crisis IMF-Supported Programs in Indonesia, Korea, and Thailand: A Preliminary Assessment, IMF Occasional Paper No 178, and a study of the eight capital account crises in the 1990s IMF-Supported Programs in Capital Account Crises, Occasional Paper 210.
J. Gold, Voting and Decisions in the International Monetary Fund (1972).
On the other hand, some constructive ambiguity in the Fund’s lending may help to reduce moral hazard, as with the lender of last resort function of central banks.
A presumptive limit might be derived from prudential considerations, as considered in the next section.
An example of such reporting requirements can be found in the recent Uruguay staff report EBS/02/108.
The External Audit Committee has also called attention to these risks (see: “Briefing of the Executive Board by Mr. Loli, Chairman of the External Audit Committee for FY 2002; FO/DIS/02/196; June 18, 2002).
This derivation of a maximum exposure limit does not take account of possible arrears on principal repayments. The Fund’s policies on precautionary balances are the subject of a staff paper scheduled for Board discussion in November 2002.
A limit of SDR 12 billion would be smaller than the exceptional access granted to 8 out of the 11 cases listed in Table 1.
Use of credit above 200 percent of quota will carry a surcharge of 100 basis points, and the surcharge will rise to 200 basis points for use of credit above 300 percent of quota. It was agreed at the same time that the surcharges adopted during the review of facilities should not be changed for a period of at least four years. Summing Up by the Acting Chairman, Review of Fund Facilities—Proposed Decisions and Implementation Guidelines, Executive Board Meeting 00/113, November 17, 2000 (Buff/00/175).
Initially 300 basis points above the normal rate of charge, the spread increases to a maximum of 500 basis points depending on how long the purchases are outstanding.
This maturity structure was first proposed in Supplemental Reserve Facility, EBS/97/225, December 5, 1997.
The SRF currently is available to members where expected access would exceed the limits under the credit tranches and EFF.
Paragraphs excerpted from the Communiqué of the International Monetary and Financial Committee of the Board of Governors of the International Monetary Fund, September 24, 2000.
In Turkey (1999), exceptional circumstances were invoked but access beyond the limits was not substantial. The case of Brazil (2001) is excluded; the authorities did not intend to draw at the outset.
An earlier case was Turkey (1980) with cumulative access of 625 percent of quota (at a time when the cumulative access limit was 465 percent of quota). Apart from this, and before the exceptional access cases, the highest cumulative access granted had been in the early 1980s, following the oil crisis and immediately before a large quota increase. In 1983 there were four cases in excess of 500 percent of quota, the highest being Jamaica (553 percent of quota), and the others being Turkey, Côte d’Ivoire and Korea. These occurred after approval of the 8th quota increase in March 1983, but before its effectiveness in November 1983. Since the mid-1980s the highest cumulative access approved had been 411 percent of quota for Mexico in 1990 in the context of a debt reduction operation under the Brady Plan.
This is true in every case except Thailand and Indonesia, where access was lower. In Turkey the access approved in the 2001 augmentation was 15 billion SDRs. Total access outstanding under the revised SBA approved in 2002 would peak at 18 billion SDRs in 2004 if all purchases were made on schedule.
These were Gabon 1995 (5.4 percent), Ukraine 1995 (5.0 percent) Moldova 1996 (8.2 percent), Bulgaria 1998 (7.1 percent) and Ukraine 1998 (5.5 percent).
Not all bilateral commitments were actually disbursed. For example, in Indonesia and Korea some funds pledged by bilateral creditors formed a “second line of defense” but were not subject to well-defined terms and conditions and were never disbursed. For details see IMF-Supported Program in Indonesia, Korea and Thailand: A Preliminary Assessment, IMF Occasional Paper No. 178, June 1999.
In Korea, access was rephased to allow a larger early drawing, but was not augmented.
In most countries, there has also been a rise in public debt as a share of GDP. However, in some cases, the rise in the ratios primarily reflects the public sector taking on the responsibilities of failed banks (and by extension corporations), and, since their financial position was precarious before the crisis—and indeed led to the crisis in some countries—the rise in the public debt constitutes recognition of already existing problems rather than an intensification of problems.
Net private capital flows to developing countries in 1996 amounted to US$192 billion, declining to US$33 billion in 1999 and US$29 billion in 2001 (WEO, Winter 2002 Vintage).
Lane and Phillips argue that “in the vast majority of instances, events that would plausibly have sent signals to markets regarding the availability and magnitude of IMF financing did not have the effects predicted under the hypothesis that moral hazard is important” and that “the role of moral hazard in recent crises may have been seriously overstated by some observers.” Does IMF Financing Result in Moral Hazard, IMF Working Paper No. 00/168, October 2000. In contrast, dell’Ariccia, Gödde and Zettelmeyer find that prior to mid-1998, there was a compression of spreads that is consistent with moral hazard during the period between the Mexican and the Russian crises; however, there is no evidence of moral hazard after the Russian crisis, Moral Hazard and International Crisis Lending: A Test, draft, November 2000.
Indeed, if debtor moral hazard was a serious problem, there is more evidence of it before the capital account crises than after them. All eight exceptional access cases had used Fund resources previously, though in the Asian countries the last Fund arrangement had been more than a decade before. Some countries had arrangements in place already when the crisis hit. Brazil and Mexico last had Fund arrangements in the early 1990s. Argentina, Russia and Turkey all had arrangements at the time of the crisis. Arguably, the severity of the crises has made the members concerned both more conscious of specific risks and more risk averse than they were before the crisis, hence the build up of reserves.
See Assessing Sustainability, SM/02/166, May 28, 2002.
In the cases of Argentina and Turkey, the analysis refers to 1999 instead of the actual crisis year, since these countries already had programs at that time and to allow for a three-year horizon of projections.
Not all tests were conducted, since the country-type specific standard deviations have yet to be developed. Also, to determine the pre-crisis averages, instead of a ten-year period preceding the crisis, a five-year period was selected.