Do not cite
The Argentine Collapse:
Hard Money’s Soft Underbelly
Kennedy School of Government
The Argentine crisis has been a humbling experience for policymakers, investors, academics and more than a few op-ed writers. This was not an event that caught people by surprise. Instead, it was a protracted affair that, as it marched inexorably towards a catastrophic demise, garnered the attention of the best minds in Washington, Wall Street and Buenos Aires for months on end.
The main actors of this drama, whether in Argentina or abroad, were well-trained economists, deeply knowledgeable of previous crisis episodes and profoundly aware of the lessons of the past and the policy implications of new theories. During this long agony, many diagnostics were developed, many original and innovative policy initiatives were tabled, much delegation of authority was provided by political parties onto technocrats and significant international intellectual and financial support was offered. And yet the catastrophe proved hard to sidestep.
This was not the first time that Argentina’s Convertibility system had gotten into trouble. During the Tequila crisis of 1995 the system had been tested through a major collapse in capital inflows and deposit demand, and had come out roaring in 1996-1997 without any changes in its currency system. Moreover, it had used the experience to lengthen the maturity of its public debt, improve the liquidity of the Treasury, upgrade its banking regulation and create a path breaking liquidity policy that assured the confidence of domestic agents and kept deposits growing throughout the 1998 recession and until as late as February 20011.
The theories that were present in the minds of the actors spanned the whole scope of the academic literature. For some, the problem had a fiscal origin and required a fiscal response (IMF, Lopez-Murphy, Mussa(2001)). In such a context, a fiscal contraction could actually become expansionary as it would eliminate fears of insolvency and make capital markets more forthcoming. These ideas lead to a sequence of fiscal adjustment efforts that in fact increased the non-social security primary surplus by over 2 percentage points of GDP in spite of the recession2. They involved raising taxes, and eventually cutting nominal public sector wages, pensions and mandated inter-governmental federal transfers. For others, the issue was a multiple equilibria story, in which self-fulfilling pessimism kept interest rates too high and growth too low for the numbers to add up. Some pointed to liquidity concerns and roll-over risks. In response, the government negotiated a major lending package lead by the IMF in November 2000, and negotiated a 30 billion dollar debt exchange in May 2001, but it did not have the expected effects. Some blamed the lack of conviction and forcefulness of policymakers and demanded a powerful leader. This strategy lead to the return of Domingo Cavallo, the architect of the Convertibility Plan of 2001 and arguably a legend in the minds of Argentineans and market participants alike. He demanded and was granted special powers to fix the economy by decree, only to be received by a skyrocketing country risk. For others, the problem was the exchange rate, which had moved in the wrong direction because of the dollars strength and the Real’s weakness. Fearful of the balance sheet and credibility consequences of an exchange rate move, the government in 2001 engineered a fiscal devaluation (i.e. a tariff for imports accompanied by a subsidy for exports, leaving financial transaction and hence balance-sheets untouched) of about 8 percent3. They accompanied this measure with a planned gradual transition away from a pure US dollar basket and into a 50-50 dollar-euro peg. For others yet, the problem was growth and required a supply response. Here again, a massive attempt was made at sectoral competitiveness plans. Markets again remained unimpressed.
It is also hard to blame the Argentine crisis on political gridlock. In spite of an unrelenting recession and with little to show for their efforts, the government repeatedly asked and got from Congress an unprecedented level of delegated power. All major policy requests were granted: a labor market reform and a tax increase in 2000, a special powers act in April 2001, a zero-deficit rule in the summer of 2001 that involved cutting wages and pensions and making their recipients junior to bondholders. And yet, as in a Greek tragedy, destiny proved unavoidable.
This paper tries to make sense of this sad experience. We explore the different interpretations and then propose a framework to analyzed the crisis. Section 2 discusses what happened. Section 3 entitled “what did not happen”, analyzes the failure of the three major paradigms with which actors analyzed the crisis. Section 4 presents and analytical framework that tries to account for the crisis. It puts the accent on the presence of an endogenous external financing constraint. With that framework we ask in section 5 what should have happened, i.e. what were the effective policy options that the authorities could have pursued. We find that fiscal contraction had very little chance of fixing things and that devaluation per se could easily have made matters worse. By contrast, devaluations accompanied by a pesification of financial claims could have resolved the crisis while minimizing the loss to investors. We conclude in section 6 with some additional lessons and policy conclusions.
II. What happened
Argentina collapsed into hyperinflation in the late 1980’s, but was able to right itself by adopting a radical market-oriented reform anchored by a currency board. The reforms delivered rapid growth in the early 90s, with a very rapid recovery of investment. Then came the Tequila crisis in 1995, characterized by a collapse in financial flows and investment that generated a deep recession. Notice, however, that during the crisis exports skyrocketed, growing at real rates in excess of 30 percent in 1995. The subsequent period of 1996 and 1997 saw what appeared to be a very healthy export- and investment-lead growth. Concerns over the competitiveness of the country were laid to rest as the economy was able to extricate itself from the Tequila crisis and rebound back to high growth through exports, without the disruptive devaluation that the Mexicans had undergone.
Then came the Russian crisis in August 1998 and later the Brazilian devaluation of January 1999. Just as under the Tequila, output declined lead by a collapse in investment. Would the economy turn itself around like the last time? Notice that in this opportunity, export prices entered a severe slump, export volumes lost their luster while investment continued to decline. The recovery never came. The earlier magic was not repeated.
For much of the period after the Russian 1998 crisis in which the economy was deteriorating, financial markets were persuaded that the situation was under control. Until the Brazilian devaluation in January 1999, markets perceived Argentina as just another Mexico. After that, while the risks were seen as somewhat larger, Argentina’s country risk was well below that of Brazil, Venezuela or the EMBI+ average4 (Figure 1). It was only in the summer of 2001 that asset prices reflected an ominous future.
Argentina obviously had a streak of bad luck. The terms of trade were negatively impacted after the Asian crisis in the second semester of 1997. Financial markets dried up after the Russian default in August 1998. Brazil abandoned its crawling band and massively depreciated its currency in January 1999. The euro sank by over 20 percent in 2000, further weakening Argentina’s competitiveness vis a vis the important European market. The world entered into recession in 2001, not only weakening commodity prices and export prospects, but creating additional turmoil in financial markets associated with the bursting of the high tech bubble. Throw in underwhelming new authorities in the US Treasury and the IMF and the implications of September 11 and you have the makings of a perfect storm. But Argentina had already gone through a bad experience with the so-called Tequila crisis in 1995 and came out roaring. Was it not just a question of keeping heads cool and policies focused until the economy would turn around? That at least was the dominant view as expressed by the IMF board in May 1999:
"Argentina is to be commended for its continued prudent policies. As with a number of other countries in the region, Argentina has had to bear the adverse consequences of external shocks, which have taken a significant toll on economic performance. Nevertheless, the sound macroeconomic management, the strengthening of the banking system and the other structural reforms carried out in recent years in the context of the currency board arrangement, have had beneficial effects on confidence, and have allowed the country to deal with these challenges." IMF, News Brief No. 99/24, May 26, 1999
III. What did not happen
As the drama unfolded, three major views developed as to the nature of the problem and the appropriate policy response. The dominant view put the accent firmly on self-fulfilling bad expectations. A second view – not completely unrelated – emphasized problems of fiscal sustainability. Another view put the accent on competitiveness and the rigidities of the exchange rate regime. Each of these views was influential in policy circles and lead to important changes in the policy framework.
The self-fulfilling pessimism paradigm probably became dominant as it was seen as the most convincing explanation of the 1995 Tequila crisis, which was associated with a sudden and systemic collapse in the demand for deposits in the banking system, a bank run in fact. Without a lender of last resort, the country was vulnerable to liquidity crises. To avoid future similar crises the authorities developed after the 1995 crisis a highly-praised liquidity policy which involved imposing high liquidity requirements on banks, negotiating contingent credit lines with foreign banks, lengthening the maturity of the public debt and keeping a liquid fiscal position. These policies were handsomely rewarded by the markets through improved confidence and market access. In fact, these policies together with the currency board were seen as providing robust institutions to cope with financial turmoil5. They proved their mettle during much of the subsequent crisis: deposits in the banking system kept growing until February 2001.
With the banking system under control, self-fulfilling negative expectations were seen as potentially causing roll-over problems in the public debt. To avoid such bad equilibria, the authorities negotiated a major expansion of international official support in November 2000 – the so-called “blindaje”. They repeated this strategy in the spring of 2001 with the 30 billion dollar debt exchange designed to lengthen the maturity of shorter-term debt.
Negative expectations were also seen as becoming self-fulfilling not just through liquidity channels but also through fiscal conduits. Pessimism would lead to high interest rates, which would depress growth and weaken the fiscal position, complicating debt service and thus justifying the initial pessimism.
“Despite substantial efforts by the Argentine government to implement the economic program it had announced in December 1999, and which the IMF has supported with a stand-by credit since March 2000, economic performance in 2000 was worse than expected. A major disappointment was the failure to recover from the recession affecting economic activity since mid-1998. After a short-lived pickup in the last quarter of 1999, the economy again stagnated. This reflected in part the impact of the fiscal tightening on domestic demand, but was mainly the result of a drop in business and consumer confidence, and the progressive hardening of financing conditions in international markets, that resulted in rising borrowing costs and reduced market access for Argentine private and official borrowers.” IMF, Press Release No. 01/3, January 12, 2001
In designing a strategy to deal with the crisis the IMF program – revised in May 2001- argued as follows:
"Argentina's program aims at strengthening confidence through fiscal consolidation to achieve the program's targets for 2001 and fiscal balance by 2005, while promoting the recovery of investment and output through fiscal incentives and regulatory changes. Firm implementation of the program is needed to initiate a virtuous circle of stronger public finances, lower interest rates, and a recovery of economic activity. (italics added). IMF News Brief No. 01/44 of May 21, 2001
To check some implications of this story we ran a simple simulation. We assumed that enough “confidence” was reestablished to secure a 3 percent growth rate starting in the fourth quarter of 1998. This simulation intends to illustrate a possible counter-factual path, had the Russian crisis not affected the availability of finance and a move towards a “bad” equilibrium. The simulation intends to use very crude relationships, just to gauge the potential implications of alternative paths. We are not taking account of other real shocks that the economy underwent throughout this period.
To keep things simple, we include a minimum number of behavioral equations but we keep the identities required by the national accounts. First, we incorporate the impact of the higher output on a higher demand for imports. We calculate the marginal propensity by running a regression between imports and output. The estimated coefficient used for these simulations is .261. We also include the higher external debt needed to run the wider current account deficit and we service the additional accumulated debt in future years. We also take account of the impact of the higher output on the primary balance. We calculate the marginal propensity to save the additional fiscal revenues from the data by running a simple regression. The estimated effect is 0.088. We leave all other fiscal variables as they are, except that we count the impact of the lower path for public debt on the interest burden. We do not include the potential impact of the higher domestic demand on a lower level of exports, as this would only make our story even more compelling.
The simulations are presented if Figures a, b, and c. As can be clearly seen, the increased activity would have been enough to maintain the public debt to GDP ratio relatively stable, below 40 percent of GDP instead of rising as it did up to almost 50 percent of GDP by the first quarter of 2001. However, in order to achieve this path, the current account deficit would have had to average in excess of 6 percent of GDP instead of declining to a 4 quarter moving average 3.1 percent of GDP by the first quarter of 20016. This larger deficit implies that external obligations would have had to rise by an additional 20 percent of GDP, even after correcting for the larger denominator, given the higher growth. This implies an even larger increase in the debt to export ratio.
Hence, leaving all other shocks aside, the “good equilibrium”, i.e. a reestablishment of enough confidence to maintain growth at 3 percent would have done away with the fiscal imbalance but would required the funding of sustained 6 percent current account deficits and the accumulation of an additional 20 percent of GDP in external obligations. This assumes that external financial constraints would not bind. As we shall see, if for some reason this amount of financing is not feasible, then the good equilibrium would also not be feasible.
A second view of the crisis put the accent, not so much on pessimism and multiple equilibria, but on the more banal problem of fiscal solvency. After all, the public debt went from 80.3 billion dollars at the end of 1994 to over 140 billion dollars in the summer of 2001. Is this not proof that the fiscal accounts were on an unsustainable path? True, the fiscal problem had been aggravated by the recession, but the debt had increased by 15 billion dollars in the three boom years of 1996-1998. Was this not proof that the country could not enforce a budget constraint? (Mussa, 2002).
The importance of fiscal balance was present in the minds of the authorities and the IMF throughout the slow evolving crisis. In fact, that was the diagnosis with which Minister Jose Luis Machinea defined the economic situation in early 2000 in order to justify his tax measures, the so-called impuestazo. It was also the interpretation of Minister Ricardo Lopez Murphy who took office briefly in March 2001. When Minister Domingo Cavallo took over after him he immediately implemented a financial transactions tax to improve the fiscal situation. He later adopted the zero deficit policy in the summer of 2001.
The view that Argentina was somehow irresponsible in its fiscal management and that this may have been a major cause of the crisis, and not just one of its consequences has become a dominant story ex post. The idea that fiscal policy was somehow inconsistent with the convertibility system has gained currency among many analysts (Reference?) We do not share this view. As we shall see, the fiscal imbalance was not large and was backed up by increased savings of the privatized pensions system. Moreover, as the simulation above illustrates, the fiscal imbalance that emerged was related to the recession and hence is best understood as a consequence rather than a cause of the crisis. Moreover, as we will argue below, it is very hard to make the case that a more forceful fiscal adjustment would have made a very significant difference. In this section we will just present the facts in a way that supports a rather different interpretation (Table 1)
The numbers quickly dispel any argument based on a spending feast. Government spending remained remarkably flat as a share of GDP from 1993 onwards. If we exclude social security payments, other primary spending actually declined by 1.5 percent of GDP (from 13.8 to 12.3 percent) during the pre-crisis period 1993-1998.
It is important to understand the dynamics that were affecting the fiscal accounts in Argentina. First, there was a rising interest burden of the debt. As shown in Table 2 factor payments increased from 1.2 percent of GDP in 1993 to 3.4 percent of GDP in 2000. This was due mainly to three factors:
Some of the Brady Bonds issued during the early 1990s had rising interest rates.
The increase in the official public debt exceeded the accumulated deficit flows between 1994 and 2000 by about 21 billion dollars, half of which was the recognition of pre-existing debts, while the 11.7 billion corresponded to debt issued in order to purchase the collateral of the Brady bonds.
After the Russian crisis the country faced an interest rate on new debt which was higher than the rate paid on average on the existing stock.
A second force affecting the fiscal accounts was the social security reform. This caused revenues to the Social Security system to be diverted towards the new privatized fund administrators. Revenues declined from 5.6 percent of GDP in 1993 to 3.8 percent by 2000 (Table 2). This did not represent a reduction in contributions to the system, only a change in the mechanism of allocation and administration. By December 2000, the private pension fund administrators had assets totaling 20.3 billion dollars7. By contrast, social security payments rose from 5.3 percent to 6.1 percent of GDP by 2000. This caused the social security balance to swing from a surplus of 0.4 percent in 1993 – before the reform – to a deficit of 2.4 percent of GDP by 2000. The cumulative deficit of the social security component of the budget between 1995 and 2000 was US$ 30.9 billion.
In order to confront these pressures on the budget, the authorities pursued a policy of improving the primary surplus of the remaining parts of the budget, i.e. excluding the social security system. This surplus increased from 1.3 percent of GDP in 1995 to 3.3 percent in 2000. In this sense, the primary surplus achieved by Argentina is perfectly comparable to that achieved by Brazil, a country that has not privatized its social security system.
Did the authorities really tighten fiscal policy when they found themselves in trouble in 2000 or was it all just talk? Table 3 explores this issue by running regressions government revenues and primary spending controlling for GDP and including a dummy for the post-impuestazo period, i.e the second quarter of 2000. Several features can be highlighted. First, tax revenues show much more buoyancy than spending. The estimated elasticity of tax revenues to GDP is 1.47, while it is only 0.72 for primary spending excluding social security. This implies that during the booming years of 1996 and 1997government spending was kept subdued relative to revenues. By the same toke, the automatic reaction to the recession would have been a significant increase in the deficit. Second, revenues post the fiscal adjustment in the first quarter of 2000 – the impuestazo – are estimated to have been 11 percent or 1 billion dollars per quarter higher than would have been expected given GDP changes. By contrast, the dummy variable for spending is not statistically significant, meaning the government was essentially just able to cut spending by the expected amount, i.e. with an elasticity of 0.72 relative to the fall in GDP. This explains in part the problems the government faced during the recession: revenues would have fallen more than spending, but significant action was taken to prevent this from happening and securing a continued improvement of the non-social-security primary surplus.
Bringing it all together, it appears that one way to describe the situation emanates quite naturally from the data: the government was able to improve the primary surplus to accommodate the increase in debt service, while the overall deficit was smaller than the deficit of the social security system. However, the latter was backed to a significant extent by the savings in the privatized pension system.
These calculations account for the published deficits. What about the assertion that the growth of debt was out of control? Table 4 below shows the increase in debt during the 1995-2000 period. As can be seen, the total increase in debt of 42.7 billion exceeds the cumulative deficit by 20.9 billion dollars. A bit over half of the difference is explained by the accumulation of assets (11.7 billion), while some 10.4 billion can be explained by the recognition of pre-existing debts. Note that the cumulative overall deficit is 10.1 billion dollars larger than the cumulative social security deficit, and is equal to the accumulation of assets in the pension system.
In conclusion, fiscal policy was adjusted to generate a sufficient primary surplus, (excluding the public social security system) in order to cover the increased debt service. In fact the primary surplus achieved, in spite of the deep recession was of the same order of magnitude as that of Brazil, in spite of the deeper recession. The overall deficit was smaller than the deficit of the social security system, which was backed up to a significant extent by saving flows into the private pension funds. In addition, there was a significant accumulation of assets and documentation of pre-existing debt.
Obviously, the country could have tried to run a tighter fiscal ship, but the numbers here are not those of a profligate country and are hard to square with the catastrophe that followed. What did change significantly from the time when Argentina was perceived as among the safest emerging markets as late as 1999?
The third influential theory was associated with the peculiar exchange rate choice of Argentina. Faced with hyperinflation and in desperate need of a nominal anchor, Argentina chose in 1991 a currency board with the dollar and a bi-monetary financial system, one in which both the US dollar and the Argentine peso were legal tender. The system achieved price stability, but left the country vulnerable to inconvenient movements in the multilateral exchange rate. This possibility became a reality after the Brazilian devaluation of January 1999 and the euro slide of 2000. The story is clearly evident in the data (Figure 2).
It is clear that the nominal appreciation of the multilateral nominal exchange rate (MNER) of Argentina took place at a most inconvenient time. The Brazilian devaluation of 1999 had caused an appreciation of the exchange rate of 14 percent. Between January and July 2001, the exchange rate appreciated a further 13 percent. This was taking place against a backdrop of stagnant export volumes and low and now declining export prices, a development that would have required a movement in the real exchange rate in the opposite direction. Hence, an increasing exchange rate misalignment developed: the worsening external conditions called for a depreciated equilibrium exchange rate, while the actual rate appreciated. Low and falling export prices, a less competitive exchange rate and a rising cost of capital must have wrecked havoc on the profitability of the export sector and thus on its ability to expand supply. Export volume growth, which had averaged over 14 percent per year between 1993 and September 1998, stalled and never again managed to recover its earlier dynamism, in spite of the declining levels of domestic absorption. And yet, it managed to maintain a 3.8 percent growth from the third quarter of 1999 to the third quarter of 2001.
This standard logic can explain the protracted recession and the increasing tension between the achievement of external balance and full employment. But why would it lead to a financial crisis? As we showed in the simulations described in Figures a, b and c, at the prevailing real exchange rate, even modest growth of 3 percent could only be achieved at the expense of large current account deficits and rising debt ratios. Argentina thus found itself in a bind: if it tried to grow it risk accumulating debt to the point of insolvency; if it chose to achieve external balance would have required strongly negative growth rates which would also have imperiled its solvency.
Markets increasingly began to fear this latter risk, as shown in Figure 3. The multilateral exchange rate tracked remarkably well the evolution of the spread between the country risk of Argentina relative to that of Mexico, especially after the Brazilian 1999 devaluation. We take Mexico as a benchmark since both economies had very similar country risk spreads until the 1999. Both countries suffered a common shock when emerging markets floundered after the Russian default, but after the Brazilian devaluation, Argentina started to move in a different direction8.