Unlike in the cases of most IMF-supported programs, pre-crisis vulnerabilities in capital account crisis countries reflected various stock disequilibria and structural weaknesses more than purely macroeconomic imbalances. The crises were manifested in sharp exchange rate movements and massive capital outflows, far in excess of initial imbalances. The focus of program design, and of macroeconomic and structural policies, therefore, was on restoring confidence, to stem and reverse capital outflows.
Pre-Crisis Conditions and Emergence of the Crisis
Pre-program conditions in capital account crisis countries differed in important respects from those prevailing in most other countries that adopted IMF-supported standby and Extended Fund Facility (EFF) arrangements in the past decade. For one thing, with the exception of Turkey (1994), which suffered from erratic growth and high inflation, most traditional macroeconomic indicators were, on average, more favorable: fiscal imbalances were smaller, inflation was lower, and growth was stronger (see Table 2.1).
Selected Macroeconomic Indicators for Capital Account Crisis Countries
See Appendix I for sample definitions.
Sample median.
Selected Macroeconomic Indicators for Capital Account Crisis Countries
t−3 | t−2 | t−1 | t | t+1 | ||
---|---|---|---|---|---|---|
Capital account crisis countries1 | ||||||
Real GDP (growth, in percent per year) | 6.0 | 5.4 | 3.8 | −5.5 | 5.1 | |
Consumer price index (growth, in percent per year)2 | 12.5 | 8.2 | 5.3 | 8.9 | 6.8 | |
Fiscal balance (as percent of GDP) | −1.2 | −1.8 | −3.5 | −4.6 | −3.5 | |
Current account (as percent of GDP) | −3.7 | −4.2 | −3.7 | −3.5 | 2.6 | |
Official reserves (in months of imports) | 3.2 | 3.3 | 2.4 | 3.6 | 4.0 | |
External debt (as percent of GDP) | 37.4 | 38.5 | 43.7 | 66.6 | 53.4 | |
Other program countries1 | ||||||
Real GDP (growth, in percent per year) | −2.2 | 0.0 | −1.6 | 2.1 | 3.1 | |
Consumer price index (growth, in percent per year)2 | 38.6 | 32.0 | 21.7 | 32.9 | 21.0 | |
Fiscal balance (as percent of GDP) | −7.0 | −5.5 | −5.1 | −3.6 | −3.3 | |
Current account (as percent of GDP) | −2.5 | −3.5 | −4.2 | −3.2 | −3.7 | |
Official reserves (in months of imports) | 1.4 | 1.7 | 1.7 | 2.3 | 2.4 | |
External debt (as percent of GDP) | 63.2 | 61.3 | 63.0 | 65.9 | 59.8 |
See Appendix I for sample definitions.
Sample median.
Selected Macroeconomic Indicators for Capital Account Crisis Countries
t−3 | t−2 | t−1 | t | t+1 | ||
---|---|---|---|---|---|---|
Capital account crisis countries1 | ||||||
Real GDP (growth, in percent per year) | 6.0 | 5.4 | 3.8 | −5.5 | 5.1 | |
Consumer price index (growth, in percent per year)2 | 12.5 | 8.2 | 5.3 | 8.9 | 6.8 | |
Fiscal balance (as percent of GDP) | −1.2 | −1.8 | −3.5 | −4.6 | −3.5 | |
Current account (as percent of GDP) | −3.7 | −4.2 | −3.7 | −3.5 | 2.6 | |
Official reserves (in months of imports) | 3.2 | 3.3 | 2.4 | 3.6 | 4.0 | |
External debt (as percent of GDP) | 37.4 | 38.5 | 43.7 | 66.6 | 53.4 | |
Other program countries1 | ||||||
Real GDP (growth, in percent per year) | −2.2 | 0.0 | −1.6 | 2.1 | 3.1 | |
Consumer price index (growth, in percent per year)2 | 38.6 | 32.0 | 21.7 | 32.9 | 21.0 | |
Fiscal balance (as percent of GDP) | −7.0 | −5.5 | −5.1 | −3.6 | −3.3 | |
Current account (as percent of GDP) | −2.5 | −3.5 | −4.2 | −3.2 | −3.7 | |
Official reserves (in months of imports) | 1.4 | 1.7 | 1.7 | 2.3 | 2.4 | |
External debt (as percent of GDP) | 63.2 | 61.3 | 63.0 | 65.9 | 59.8 |
See Appendix I for sample definitions.
Sample median.
Current account deficits prior to the crises were substantial but they were the counterpart of large private capital inflows that in most cases had been sustained for years. Although current account deficits were, to varying degrees, an ingredient of the crisis in all countries, these deficits in and of themselves do not explain the abruptness and magnitude of the reversals of capital flows. In some countries, notably Mexico and Thailand, current account deficits were very large, but overall external debt was manageable and solvency was not a concern.3 Although the capital inflows registered in the precrisis years could not necessarily be expected to persist, the magnitude of these flows gave little warning that the adjustment in the capital account would be as sudden and severe as that which eventually took place. Nor did the appreciation of the real effective exchange rate, experienced in varying degrees in most countries in the years prior to the crises, foreshadow the sudden sharp depreciations, with significant overshooting, that occurred (Figure 2.1). In fact, in the majority of the countries considered, the crisis intensified after the initial exchange rate adjustment. In the absence of other vulnerabilities, current account adjustment and a correction of the real exchange rate could have been achieved in a more orderly fashion, although perhaps not without some temporary turbulence.4
Exchange Rate Movements in Capital Account Crisis Countries
Sources: IMF, Information Notice System; and IMF staff estimates.The large capital outflows were thus less the underlying cause of the crises than their manifestation. The underlying vulnerabilities primarily reflected stock imbalances—high levels of public debt, maturity or currency mismatches in the structure of private sector liabilities, or highly leveraged positions 5—and the correction of these imbalances would likely have entailed at least some macroeconomic disruption, even in the absence of the very large capital outflows that eventually transpired. The precise propagation mechanism between these stock imbalances and the currency crises varied across countries. As many “second-generation” crisis models would suggest, however, the market’s perception that the authorities might be reluctant to raise interest rates—in some cases because of vulnerabilities in the corporate and financial sectors, in others because of the adverse impact on public debt dynamics—may have been an important factor that helped trigger self-fulfilling runs on the currency.
The fact that the vulnerabilities were mainly related to stock imbalances, rather than traditional flow disequilibria, complicated the macroeconomic policy response and created an environment in which it became rational for investors to run for the exit in response to relatively small changes in information about the fundamentals.6 The sources of these vulnerabilities differed across crises: in the Latin American countries and Turkey, they were primarily rooted in the public sector; in Asia, in the private sector.
The Mexican crisis of 1994–95 demonstrated that such vulnerabilities do not necessarily reflect longstanding imbalances but can build up relatively quickly, particularly when the policy response fails to come to grips with emerging difficulties. Prior to the crisis, in 1992–93, public finances were on a sound footing, with a surplus of the non financial public sector, declining public and external debt ratios, and short-term external debt more than covered by official reserves. In 1994, however, as a combination of domestic political shocks (uncertainty about the election outcome and internal armed conflict), rising international interest rates, and a hesitation to tighten monetary policy had resulted in substantial reserve losses, the government swapped large amounts of peso-denominated treasury bills for U.S. dollar indexed bills both to reduce its funding costs and to underscore its commitment to the exchange rate peg. Although the move reduced interest payments in the short run, it also increased the government’s exposure to exchange rate risk significantly. By early 1995, Mexico was facing an external funding crisis in the wake of the floating of the peso as doubts about the government’s ability to service its maturing obligations led to massive outflows.
In contrast, the crises in Turkey (1994) and Brazil (1998) primarily reflected considerations that were medium-term in nature, related to adverse public sector debt dynamics.7 Although public sector debt ratios were comparable to those in other countries, the legacy of high inflation—ongoing in Turkey and only recently conquered in Brazil—and low policy credibility was reflected in high real interest rates and a correspondingly large burden of this debt on the fiscal accounts; each country ran large and increasing deficits in the run-up to its financial crisis. Moreover, this situation gave rise to an inherent policy dilemma: any monetary accommodation for exchange rate depreciation threatened to accelerate or reignite inflation, but raising interest rates in response to a shift in the capital account could also fuel inflation through “unpleasant monetarist arithmetic,” 8—that is, by aggravating the already heavy debt servicing burden of the public sector.
This dilemma was at the core of Brazil’s vulnerability to contagion in the aftermath of the Asian and Russian financial crises, undermining market confidence in the exchange rate peg despite a comfortable cushion of official reserves (in relation to imports or short-term debt). Temporary interest rate hikes failed to stem capital outflows amid doubts about the government’s ability to achieve a sustained fiscal consolidation. The dilemma continued as the IMF-supported program sought to maintain the exchange rate peg, and was only resolved once structural fiscal reforms gained credence and the peg was replaced by a more credible monetary policy framework.
Argentina, like Brazil, had a history of high inflation, which had been stabilized a few years before the Mexican crisis of 1994–95. In contrast to Brazil, however, its policy credibility was buttressed by a currency board-type arrangement and, with the fiscal position close to balance in 1992–93, public sector debt dynamics were not a concern. Moreover, with external debt accounting for over 70 percent of total public sector debt, the fiscal position was partially insulated from changes in domestic financing conditions. But two weaknesses of these arrangements nevertheless made Argentina especially vulnerable to the effects of the Mexican crisis: reliance on external financing, which made the government dependent on sentiments in international capital markets, and the severely limited ability to absorb bank balance sheet weaknesses that the currency board arrangement imposed on the central bank. These concerns triggered an external funding crisis as the government temporarily lost access to international capital markets amid massive withdrawals of bank deposits and large capital outflows.
Pre-crisis conditions in the Asian program countries differed significantly from those in Latin America and Turkey. These countries had a strong track record of low inflation and fiscal positions that had, on average, been close to balance (Indonesia, Korea, and Philippines) or in surplus (Thailand) and public sector debt ratios that were either very low (Korea and Thailand) or declining steadily (Indonesia and Philippines). The vulnerabilities that exposed these countries to a shift in market sentiment were rooted in the private sector, reflecting both weak financial systems and fragilities in heavily indebted corporate sectors. As in other countries, these vulnerabilities created a serious dilemma for monetary and exchange rate policy.
The weaknesses in the financial systems in Asia are by now well known: financial institutions ill equipped to handle risk; inadequate regulation and supervision; highly leveraged corporate sectors; and substantial unhedged and short-term foreign borrowing—encouraged, in part, by remaining controls on long-term flows (Korea) or special facilities (Thailand) and by the implicit guarantee of a pegged exchange rate.
The proximate trigger of the Asian crisis was a decline in export growth, against the background of weakening demand in partner countries, some real exchange rate appreciation especially as the U.S. dollar (to which these countries’ currencies were explicitly or implicitly pegged) strengthened against the yen, and a sharp decline in prices for key exports, notably semiconductors, which in varying degrees affected many countries in the region. In Thailand, the resulting drop in exports prompted a reassessment of the sustainability of the country’s large current account deficit and of inflated domestic asset prices. Asset prices began to decline, and the downturn of the economy increasingly exposed the vulnerabilities in the financial sector and led to capital outflows and increasing pressures on the exchange rate.
Contagion from the crisis in Thailand quickly spread throughout the region, affecting, among others, the Philippines, Indonesia, and, eventually, Korea. Like Thailand, these countries hesitated to raise interest rates and initially tried to support the exchange rate through intervention. (Indonesia, however, did attempt an initial hike in interest rates that was later reversed.) As in other countries, this policy failed to solve the flow disequilibrium in the foreign exchange market, while creating (in Indonesia and the Philippines) or severely aggravating (in Korea) imbalances between the stocks of short-term debt and usable reserves. These imbalances provided additional incentives for investors to run for the exit, particularly in Korea. Faced with significant reserve losses, 9 these countries were eventually forced to abandon their implicit or explicit pegs.
Despite differences in the underlying causes of the crises, the broad pattern of balance of payments developments before and during the onset of the crises is strikingly similar in the capital account crisis countries (Figure 2.2). Prior to the crises, private capital inflows averaged 4–6 percent of GDP, matched by current account deficits of similar orders of magnitude (official capital and reserve flows being relatively small). In each of these countries, the initial response to the shift in private capital inflows was to try to maintain the implicit or explicit currency pegs through large-scale intervention and reserve losses. With the exception of Argentina, the pegs were abandoned and the counterpart to private capital outflows was official financing together with current account adjustment. One notable difference between Latin America and Turkey, on the one hand, and the Asian crisis countries, on the other, was the duration of the capital outflows when the crises broke. In Latin America and Turkey, capital outflows tended to be sharper—substantial, but of (relatively) short duration. In East Asia, by contrast, capital outflows lasted for several quarters (and, in some cases, inflows have yet to resume on a sustained basis). Eventually, though usually only after a wrenching macroeconomic adjustment, private capital outflows subsided or were even reversed, allowing for a buildup of reserves. In turn, this pattern of capital flows had profound implications for program design, the need for financing, and the appropriate role of macroeconomic policies.
Balance of Payment Developments
(In percent of quarterly GDP)
Source: IMF, International Financial Statistics.Implications for Program Design
When the countries turned to the IMF for support, 10 they had already encountered one or several waves of capital outflows. Faced with the dilemma that both sharp increases in interest rates and a depreciation of the exchange rate would likely have serious negative effects, most of these countries had tried to counter the outflows through some form of exchange market intervention, which was largely sterilized. In the process, they had incurred significant reserve losses, which in some cases (Korea, Mexico, and Thailand) had led to a virtual depletion of net reserves, significantly increasing the risk of a full-fledged funding crisis. With the exception of Turkey, all of the countries had some form of exchange rate peg prior to the crisis, 11 but by the time the programs were negotiated only Argentina and Brazil were still maintaining their pegs. (The initial program in Brazil explicitly sought to maintain the peg; however, the rate had to be floated soon after the program was approved.) The other countries had already experienced substantial depreciations of their exchange rates, which generally exceeded pre-crisis estimates of possible overvaluations. These capital outflows, if extrapolated, would require massive shifts in current account balances well beyond what seemed consistent with medium-term sustainability and what would have been considered adequate prior to the crisis.
These circumstances held three key implications for program design. First, in the immediate run, restoring confidence required convincing financing strategies that alleviated concerns about the countries’ ability to meet their maturing obligations. Second, the fundamental thrust of macroeconomic policy had to be directed more at stemming capital outflows than at fostering greater external adjustment (as in traditional programs). Third, restoring confidence required addressing the various vulnerabilities that had prompted the shift in market sentiment in the first place.
As discussed below, none of these desired criteria for program design would be easy to achieve. If programs were successful in restoring confidence, the large official financing would be superfluous. But given uncertainty about market reactions, programs ran the risk that the projections of market turnaround on which they were predicated would turn out to have been overly optimistic. At the same time, assuming the worst case scenario—or heavy-handed attempts at private sector involvement—could undermine confidence further and exacerbate outflows.
On macroeconomic policies, monetary policy faced a dilemma: should policies be tightened to stem outflows, or eased to forestall an excessive weakening of economic activity (or a worsening of the public debt dynamics)? Fiscal policy could play a key part where there were underlying weaknesses in public sector finances. Beyond this, however, fiscal policy’s role was difficult to define because the contribution it could make to external adjustment depended critically on the response of the private sector to the crisis, which was difficult to predict.
Finally, addressing structural vulnerabilities was relatively more straightforward in the countries where these vulnerabilities were primarily rooted in the public sector and where the required policy adjustments fell primarily in the domain of the IMF’s traditional expertise. In contrast, in the countries where the structural vulnerabilities were rooted in the private sector and required extensive structural reforms in the financial system and the corporate sector, the vulnerabilities presented a new set of challenges.