The Brazilian exchange rate crisis of January 1999: Errors in economic policy or contagion from Asia and Russia?

From the monetary reform of July 1994 to January 1999 Brazil followed the policy of pegging the new currency (the real) to the US dollar. The central rate was initially fixed at 1:1 to the US dollar but no fluctuation band was set and the market rate was allowed to fluctuate substantially. After a sharp appreciation of up to 15% the real remained at a premium to the dollar for two years (until June 1996). In March 1995, following the Mexican crisis, the Banco Central do Brasil adopted a crawling band without preannounced depreciations. This change in policy was meant to increase somewhat the flexibility of the exchange rate regime while still maintaining an anchor for inflationary expectations. The market rate depreciated by 13.9% in the course of 1995 (December 1995 on December 1994), 7.1% in 1996, 7.3% in 1997 and 8.3% in 1998. By December 1998 it had reached 1.2054 to the US dollar, a depreciation of only 20% with respect to the central rate fixed at the end of the hyperinflation but about 40% with respect to the rate prevailing in July 1994.

During the four and a half years from July 1994 to end 1998 the price index for non-traded goods increased by 120% and the price index for traded goods increased by about 26.7%, resulting in an enormous loss of competiveness of Brazilian exports on world markets, a substantial worsening of the current account which moved from a surplus of 1% of GDP in 1992 to a deficit of 4.5% in 1998 (both were recession years) and a significant increase in the Brazilian foreign debt both private and public. Worst of all, Brazil’s growth rate slowed down sharply since the second half of 1997 and did not recover until the second half of 1999, a dismal performance for a country whose GDP per capita had not been growing for almost two decades.
By the second half of 1998 the costs of the exchange rate based stabilization and of the excessive rigidity of the nominal exchange rate had become clear even to the government. In January 1999 the exchange rate was devalued by 8% with respect to the US dollar with the upper limit of the band being fixed at 1.32 from 1.22: too little too late. The government was unable to defend the new peg and was forced to move within days to a flexible exchange rate system. By February 1999 the real reached the maximum of 2.2 to the US dollar. It soon reversed the excessive depreciation and with ups and downs (from a minimum of about 1.65 to a new maximum of 1.99 in October) it stood at about 1.75 at the time of writing (March 2000).
The exchange rate based stabilization pursued by Brazil after the hyperinflation was the most reasonable policy to follow and can be considered as successful. However, it was pursued for too long at the cost of a large loss in competiveness first and of economic growth later. The time was ripe already before the Thai crisis of July 1997 to introduce greater exchange rate flexibility and full Central Bank independence coupled with inflation targeting. The Asian and Russian crises caught Brazil with an excessively overvalued exchange rate and a worsening current account. In addition to the mistake of continuing with the exchange rate policy of the previous years the government committed an additional error which made the policy mix even more incompatible – fiscal policy became much too expansionary in the election year 1998.
This paper analyses the causes of the Brazilian exchange rate crisis and tries to address the question of how much it was the result of errors in domestic policy and how much simply the result of contagion from Asia and Russia and excessive reliance on foreign capital subject to herd behaviour. The paper is divided in six sections. Section 1 focuses on measures of overvaluation of the exchange rate. Section 2 on the development of the current account and of foreign debt. Section 3 on government finances. Section 4 on monetary policy and measures of illiquidity ratios as leading indicators of financial and/or exchange rate crisis. The period taken into consideration is usually from the end of the hyperinflation (July 1994) to the end of 1998, just before the government lost control of the exchange rate. In section 5 we analyse structural changes in the Brazilian banking system since 1994 and adddress the question of why the 1999 exchange rate crisis was not accompanied by a banking crisis, unlike what happened in most Asian countries in 1997. Section 6 concludes by summarizing the main results of the paper and by drawing some general conclusions on the Brazilian exchange rate crisis and on financial crises in general. We shall also deal with issues such as the dangers of contagion and of excessive freedom of international capital movements.

1 – Was the real overvalued?
Measuring meaningfully the real exchange rate and the fundamental equilibrium exchange rate is always a difficult task. Nevertheless, already towards the end of 1997 our calculations showed that the real was dangerously overvalued with respect to the US dollar and in effective terms. At a conference in Buenos Aires held in December 1997 we warned that the policy of pegging the real to the US dollar could not be continued much longer (Tullio, 1998). This view was shared by several Brazilian economists.
By December 1997 the nominal exchange rate of the real had fallen in relation to the US dollar by 11.3% with respect to the central rate fixed in July 1994 and by 19.5% with respect to the commercial rate prevailing in the market in the same month, while the price index for traded goods had increased from August 1994 to December 1997 by 27.5% and the one for non traded goods by 112.2% (Table 1). Even if the figures for July and August 1994 might be distorted by the end of the hyperinflation the data presented in Table 1 suggest that the nominal exchange rate may have been unsustainable already in 1997. This conclusion is reinforced by the behaviour of the trade and current account balance. The trade balance had moved from a surplus of 15.2 billion dollars in 1992 to a deficit of 8.4 billion in 1997 and the current account from a surplus of 6.1 billion to a deficit of 33.3 billion. Table 1 also shows the cumulated changes of traded and non traded goods prices up to the end of 1998. The disequilibrium in domestic relative prices continued to increase in 1998, fed by excessive domestic demand growth.
Despite these developments the exchange rate policy of Brazilian monetary authorities remained largely unaffected by the outbreak of the Asian crisis (July 1997). The changes in the nominal exchange rate of the real are summarized in Table 2. Only after the Russian crisis of August 1998 did the Brazilian monetary authorities accelerate the rate of devaluation of the currency, but only marginally so. This is shown in the bottom half of Table 2 where the rates of changes of the currency are presented by semester. Until June 1998 the rate of change of the currency by semester was in the range of 3.6-3.7%. In the second semester of 1998 the rate of depreciation was increased to 4.4%, too little too late given the very large disequilibria accumulated in 1995-97 and the nervousness of international financial markets since July 1997.
Table 3 shows the growth rates of real GDP in Brazil from 1995 to 1998. Considering that during the last 10 years or so of the hyperinflation period there had been virtually no growth in Brazil and that the ends of hyperinflations always entail large welfare gains and increases in economic efficiency arising from the removal of distortions, the performance of Brazil has been far from satisfactory since the end of the hyperinflation. In addition since the second half of 1997 the rigid exchange rate policy combined with the effects of the Asian and Russian crises led to a sharp slowdown in Brazilian economic growth.
At the same time yearly unemployment went up steadily from 4.6% in 1995 to 5.4% in 1996, 5.7% in 1997 and 7.6% in 1998. In a separate more technical paper we test a model for Brazil in which we explain devaluation expectations as a function of unemployment, unemployment as a function of the real interest rate and show in addition that the nominal interest rate was positively influenced by devaluation expectations (Tullio and Ferreira, 2000). These tests suggest that the defence of an unrealistic exchange rate by means of high interest rates may eventually lead to a politically unsustainable increase in unemployment which undermines the credibilty of the pegged exchange rate (Obstfeld, 1994 and Eichengreen and Jeanne, 1998).
The real effective exchange rate of the Brazilian real is shown in Figure 1. Two measures are shown, one calculated by using consumer prices at home and abroad and the other using wholesale prices .

The effective exchange rate deflated by wholesale prices (EXRATEWPI) shows less variability than the one deflated by consumer prices (EXRATECPI) because the former contains a much higher proportion of traded goods the prices of which are more heavily influenced by the behaviour of the exchange rate itself. For this reason we consider the effective exchange rate deflated by consumer prices as a better proxy of the competitiveness of Brazilian goods. Considering the years 1992-93 as years during which the exchange rate was roughly in equilibrium, the Brazilian currency was overvalued by about 20% in mid 1996. In 1997 and 1998 Brazil gained back some of the lost competiveness (10% by end 1998 with respect to the worst moment in mid 1996) thanks to the depreciation of the nominal exchange rate and to repressed inflation at home which acccompanied the slowdown in economic growth and the increase in unemployment. With the nominal depreciation of about 50% in January 1999 and in the succeeding months (from 1.2 to 1.8 to the dollar) and thanks to the fact that consumer price inflation at home remained very subdued contrary to the initial expectations of most analysts (consumer price inflation was only 9% in the period December 1998 to December 1999) Brazil recorded in 1999 an improvement in competitiveness of about 40% setting the stage for a recovery in economic activity.
In consideration of what has been said so far it is surprising that Stanley Fischer, Deputy Managing Director of the IMF, was stating publicly in road-shows in November 1998 “that the real is certainly not as overvalued as one reads in the press” (Il Sole-24 Ore, November 18th 1998). There must have been political reasons for the IMF to support the exchange rate policy of Brazil in 1998: the reelection of President Cardoso in October 1998 and the fear that a devaluation of the real would upset trade links in Mercosur and put Argentina’s currency board and the legally fixed peg of the peso to the US dollar in a very difficult situation. There was also the fear that China would devalue if Brazil’s currency would fall. There were talks of a systemic crisis, of deflation and avoiding a devaluation in Brazil was seen by some as a crucial step in preventing a worsening of the world crisis. Whatever the political and economic reasons behind the huge aid package put together by the IMF to help Brazil in the second half of 1998 (41 billion dollars!), markets may have been led to believe for a short time by statements like the one cited above and by the aid package itself that for the effective exchange rate of the Brazilian currency the normal laws of gravity (by which we mean the return towards some sort of equilibrium) could be suspended by decree. But markets did not seem to buy this story for long: after August 1998 capital flight forced Brazil to raise again interest rates to unsustainable levels. However, in a sense the aid package and the interest rate policy were successful: Cardoso was reelected in October 1998 but Brazil paid dearly the policies pursued with two years of virtual standstill in growth (mid 1997 to mid 1999).

2 – The balance of payments and foreign debt.
Table 4 shows how the trade balance of Brazil worsened significantly from 1992 to 1997, moving from a surplus of 15.3 billion US dollars in 1992 to a deficit of 8.3 billion in 1997, a turnaround of roughly 3 percentage points of GDP. The slight improvement in 1998 was due to the sharp slowdown in imports which outweighted the effects of the Asian and Russian crisis on Brazilian export volumes and prices. Table 5 shows that the prices of Brazilian exports of raw materials fell by 16% in 1998. This fall was only partly compensated by an increase in volumes of 7%. A similar thing happened with prices and volumes of exports of semi-manufactures (-8% for prices and plus 4% for volumes). However, raw materials accounted in 1998 for only 26% of Brazilian exports and semi-manufactures for only 16% (Table 6). For manufactured goods, which accounted for 58% of exports, prices fell by only 1.5% while volumes rose by 2%. Thus the value of Brazilian exports was not dramatically affected in 1998 by the Asian and Russian crises and by the fall in raw material prices which ensued. Contagion via trade links is not a major issue in the explanation of the Brazilian crisis of January 1999. Brazil is not comparable to Malaysia and Indonesia after the Thai crisis, nor is it comparable to Russia which is largely a raw material exporting country. This is one of the reasons which explain why the Asian crisis did not initially affect Brazil.
With the devaluation of January 1999 the trade balance improved by somewhat over 5 billion dollars in the course of the year, not very much (Table 4). However, world demand and prices were still weak in 1999 and price elasticies of trade flows are known to be rather small in the short run. There was a further improvement in the first quarter of the year 2000. The volume of exports of manufactures picked up sharply already in the second half of 1999 as did the volume of exports of semi-manufactures. The improvement in the volume of raw material exports was even more pronounced (Table 5).
Table 7 shows the current account from 1990 to 1999. From its peak in 1992 when it recorded a surplus of 6.1 billion US dollars, it moved into large deficits in the course of the decade. The deterioration was particularly sharp in 1995 and continued in 1998 mainly because the increase in interest payments on foreign debt was larger than the improvement in the trade balance (Table 8). In 1998 the current account recorded a deficit of 35 billion dollars or about 4.5% of GDP, a turnaround of about 5.5 percentage points of GDP with respect to 1992.
The Brazilian current account deficit of 1998 was not among the highest in the list of countries which experienced major foreign exchange crises. Mexico which experienced a major crisis in December 1994 had a deficit of 7% of GDP in that year, Thailand where the crisis broke out in July 1997, had a deficit of 8.1% in 1996 and Malaysia of 8.4% (Table 9).
The large current account deficits recorded since 1995 led to a roughly equivalent increase in the foreign debt of Brazil. From end 1995 to end 1998 total foreign debt increased by 75 billion US dollars (Table 10). At the end of 1998 total foreign debt stood at about 30% of GDP. In the course of 1998 the short term foreign exposure of both the public and the private sector fell significantly as a result of capital flight. At the same time the medium and long term exposure of the public and private sector increased by 47 billion. Table 11 shows the composition of Brazilian foreign debt. At the end of 1998 short term debt was 10% and medium and long term debt 90% of total. The share of public debt fell significantly from 1995 to 1998 for all maturities.
The analysis of Brazilian economic developments since the end of the hyperinflation suggests the following interpretation of the 1994-98 period. A sustained growth of domestic demand was accompanied by a less significant rise in domestic production and spilled over into prices of non-traded goods (Table 1) and into foreign goods and services (Table 7). The resulting current account deficits were financed by accumulating foreign debt on a large scale (Tables 10 and 11). As we shall see in Section 3 fiscal policy contributed significantly to the accumulation of these economic disequilibria and in this sense the Krugman first generation model of exchange rate crisis is a valid simplification of Brazilian developments (Krugman, 1979), especially from 1997 onwards.
Table 12 shows that the rise in domestic demand was mainly led by consumption. The share of consumption in GDP grew from 79.3 % in 1992-94 to 82.8% in 1997 (a rise of 3.5 percentage points) while the share of investment grew in the same period by 0.6 percentage points only. In the next section we turn to errors in fiscal policy.

3 – Fiscal policy
From 1995 to 1998 total government expenditures grew by 31% in real terms (Table 13). Interest payments on public debt increased by 108% and transfers to States and Municipalities by 17%, while real GDP grew in the same period by a mere 10.7%. Particularly worrisome were the developments of total real government expenditures in the election year 1998: they grew by 22% after a rise of 8% in 1997. The large increase in interest rates to defend the currency after the outbreak of the Asian and the Russian crises was the main culprit. However, other items of the budget also contributed to this outcome: expenditures for personnel, amounting to over 30% of the total, grew by 11% in real terms and transfers to States and Municipalities by 15%.
It is clear that the Brazilian stabilization was also marred by fiscal indiscipline, besides a wrong exchange rate policy. This is especially true for 1997 and 1998, when fiscal authorities reacted wrongly to whatever Asian contagion there might have been and to the loss of confidence in Brazilian economic policies. Despite the so called “pacote 51” announced on 10 November 1997 and successive measures taken to appease financial markets the actual behaviour followed by the authorities after the outbreak of the Asian crisis was totally different. They should not have increased total government expenditures, but cut other items of the budget to compensate for the increase in interest payments. It could be doubted whether Cardoso’s electoral campain really benefited from the expansionary fiscal policy followed. An orderly readjustment of the real exchange rate in the second half of 1997 coupled with fiscal restraint in 97/98 and legislation granting more independence to the Banco Central do Brasil would have very likely caused more economic growth in 1998.

The fact that the only year during which fiscal authorities did not “misbehave” was 1996 is confirmed by the development of the fiscal deficit, shown in Table 14. In that year total government expenditures fell in real terms and the nominal deficit fell by 1.3 percentage points of GDP. In 1997 the deficit fell further by 0.8 percentage points of GDP but real expenditures were rising.
Table 15 shows total public sector net debt subdivided into internal and external. Total debt increased from 28.5% of GDP at the end of 1994 to 49.4% in September 1999. The fact that net debt fell in 1998 despite a large public sector borrowing requirement of 8% (cfr. Tables 15 and 14) requires an explanation. In that year the dynamics of net debt was influenced by privatizations, the proceeds of which are not fully included in the deficit figures of Table 14. There were large privatizations in the telecommunication sector (Telebras) and auctions for concessions of mobile phones. In addition in 1998 non-active bank deposits of non-identified owners were transferred to the government. In the late 1980s the government had authorized banks to issue deposits without the owners having to identify themselves. This authorization was later withdrawn and some deposit owners, for reasons that one can easily guess, never identified themselves and never claimed their deposits back. The most dynamic component of total net debt was internal debt. Although by international standards Brazilian public sector debt was not very high in 1999, its growth since 1994 has been astonishing .

4 – Monetary policy
An interesting question is what allowed Brazil to follow the exchange rate policy it did until the crisis of January 1999 and pursue at the same time a very expansionary fiscal policy. The main answer is to be found in the restrictive monetary policy pursued when international reserves were falling. However, there were other more fundamental factors at work like the policy of trade integration within Mercosur, the large reductions in import duties, the successful privatization policy of the government, the resumption of economic growth and the initial success of the exchange rate based stabilization. These factors boosted foreign confidence in the country and led to an increasing flow of foreign direct investments which remained unscathed by the Asian, the Russian and even the Brazilian crisis. In addition they led to several waves of massive inflows of capital which not only financed increasing current account deficits but allowed also an increase in the stock of international reserves by 42.3 billion US dollars. The reserves rose from 31.5 billion in June 1995 to a peak of 73.8 billion in April 1998. However, foreign capital movements were very unstable and their unstable nature put at times Brazilian monetary policy under severe strain.
Table 16 reports foreign direct investments in Brazil. They rose steadily from 1.7 billion US dollars in 1994 to 26.1 billion in 1998 (3.4% of GDP). In the first nine months of 1999 they were 28% higher than in the same period of 1998, suggesting that the exchange rate depreciation did not affect the positive expectations about the future of the Brazilian economy. On the contrary the devaluation may have helped the flow of foreign direct investment.
The analysis of the behaviour of international reserves is a good starting point to study the monetary consequences of the sharp turnarounds in international capital flows. In addition only after discussing at some lenght the irregular nature of capital flows to and from Brazil we may attempt to draw some conclusions about the need to restrict them, a rather hotly debated issue at present. Table 17 shows the stock of Brazilian international reserves and suggests that the period since the end of the hyperinflation can be roughly divided into 7 phases.

1) During the first which goes from July 1994 to June 1995 reserves were falling moderately but steadily. This period was characterized by a “wait and see” attitude with respect to the outcome of stabilization with a good dose of scepticism. Also the Mexican crisis of 1994/95 played some role in this period.

2) During the second period which goes from June 1995 to June 1996 reserves almost doubled from 31.5 to 58.6 billion. In February 1995 a restrictive monetary policy was put in place to check a consumption boom which was threatening the success of the stabilization. As a result confidence surged soon thereafter bringing with it a substantial wave of capital inflows.

3) During the third period which goes from June 1996 to June 1997 (the 12 months preceding the outbreak of the Thai crisis) reserves remained stable at the high levels reached at the end of the previous period.

4) The second semester of 1997 was marked by the effects of the Asian crisis. These effects were felt relatively late in Brazil. On 27 October the Dow Jones Industrial average fell by 7.2% and the stock market of Sao Paulo lost about a third of its value in six sessions. In the foreign exchange market the Brazilian Central bank lost some 8 billion US dollars. By doubling interest rates to 43% at the end of October and by annoncing the “pacote 51” on 10 November calm returned to the markets. Overall reserves had fallen by year-end by 11 billion US dollars with respect to their peak in August. Thus the effect came late and was not extremely large. Brazil has very few trade links with Asia and its banking system was solid, at least by Asian standards. In addition, as stated above, monetary policy was very restrictive in November and December. The key Central Bank interest rate (Selic) almost doubled from an average of 22% in October to an average 42.6 in the November-December period (Figure 2).

5) From January to August 1998 confidence in Brazil returned on a massive scale. Reserves went up by 22.5 billion US dollars in the four months from end 1997 to April 1998, then came slightly off their peak until August. During the same period the key interest rate was more than halved to 19% (Fig.2) to a level which was lower than the one prevailing before the Asian crisis. The return of confidence on such a scale is difficult to explain in view of the developments of the real exchange rate, of slowing economic growth and especially of the fiscal policy pursued by the authorities. The delays with which the actual behaviour of fiscal variables becomes known, coupled with the fact that announcements of fiscal policy were in the direction of austerity may have played some role. The behaviour of fiscal and monetary policy in the first 8 months of 1998 was certainly influenced by the desire to stimulate activity in view of the presidential elections of October. This was a major economic error (although possibly not a political one). Brazilian economic policy had become utterly inconsistent by mid 1998 and it was only a matter of time when markets would dicover it. Brazilian authorities must have been misled into greater misbehaviour than would have otherwise been possible by the large return of confidence in the country, which seems to us difficult to explain in the light of wisdom after the event.

6) The sixth period goes from August to December 1998. The Russian crisis and the fall in all stock markets around the world hit Brazil hard. Reserves fell from 69.4 billion US dollars in July to 45 billion in September, a 24.9 billion drop in only two months, with most of the change occurring in September (21.5 billion) (Table 17). In the remaining part of the year the fall in reserves slowed down. At the same time in order to stem the capital outflow the Selic rate had to be increased again from 19.3% in August to 41.6% in October. Were these changes in Brazilian international reserves and interest rates solely due to the Russian crisis and contagion? Certainly not! When the Asian crisis broke out and only the Brazilian exchange rate policy was clearly wrong, contagion, if any, was very small and very delayed as we have seen above. In 1998 all policies were wrong in Brazil (monetary, fiscal and the exchange rate policy) and what many well known economists like to call contagion may more simply be the lagged realization by international investors that economic policies were utterly inconsistent. The outbreak of a crisis like the Russian one may simply represent the occasion for research departments of major international banks to give a closer look at the inconsistency of a country’s economic policies. No econometric model will ever be able to tell us exactly how much of the fall in reserves was purely due to contagion and how much it was the result of previously inconsistent policies. However, the analysis presented in this paper suggests that in the case of Brazil one should not exaggerate the role played by contagion. The data presented in Tables 5 and 6 showing the development of prices and quantities of Brazilian exports by major commodity groups reinforce this conclusion (see Section 2).

7) The last phase goes from January 1999 to May 1999. Reserves fell sharply from 43.6 billion in December 1998 to 32.8 billion in March 1999, then they recovered in April as the worst of the exchange rate crisis seemed to be over. In order to understand the fall in reserves which occurred in this period and especially in the run up to the currency crisis of late January 1999 is worth noting that the Central Bank reduced the Selic rate from 41.6% in November to 32.9% in December, a reduction of almost 10 percentage points.

The fact that capital flows to Brazil were so irregular led, as we have seen in analysing the behaviour of interest rates, to difficulties in the conduct of monetary policy. The growth rates of M2, reported in Table 18, show a very irregular behavior.
Table 19 presents two measures of international illiquidity: the ratio of short term foreign debt to international reserves and of the money stock (M2) to reserves. These ratios have become common statistics used in the recent literature on financial crises. They measure the pressure on international reserves which could arise from market expectations about impending currency changes. Especially the last measure suggests a steady worsening of the liquidity position of Brazil from the end of 1995 to the end of 1997. During this interval all numerators of the measures increased significantly while reserves did not change, suggesting that well before the outbreak of the Asian crisis Brazil’s short term foreign debt policy was on a dangerous path. Between December 1997 and June 1998 the growth of international reserves outpaced the growth of money, leading to a sharp reduction in the ratios and the illusion that the worst was over. However, short term foreign debt of Brazilian monetary authorities and of the private sector fell by 12.5 billion US dollars in 1998 (see Table 10). As a result in the second semester of 1998, when things turned sour, the first illiquidity ratio worsened very little, while the other reached unprecedented levels.

5 – The Brazilian Banking System.
Most Asian countries which experienced exchange rate crises in 1997 had also weak financial and banking systems. Their crisis was also a banking and financial crisis. The Brazilian crisis of 1999 was instead only a foreign exchange crisis as the banking system was rather solid and its foreign indebtedness was minor compared to Asian countries. Hence no major government intervention was needed in the finacial sector.
The Brazilian banking system had been badly hurt by the end of the hyperinflation as a substantial part of the inflation tax accruing to it suddenly disappeared (Tullio and Ronci, 1997) and the spreads between rates on lending and deposits which were huge had to come down to more acceptable levels, although they remained still unreasonably high well into the year 2000. As a result when the hyperinflation ended banks had to start a painful restructuring process although some banks had enough foresight to start adjusting before. The share of the financial system in GDP which was 15.6% in 1993 fell to 6.9% in 1995. The Brazilian government and the Central Bank played a very active role in helping this restructuring process with budgetary funds (4% of GDP between 1994 and 1998), by easing the access of foreign banks to Brazil, by encouraging mergers, acquisitions and closures of weak banks and increasing capital requirement ratios towards the standard of industrial countries. All in all the gradual restructuring and strengthening of the Brazilian banking system from 1994 to today was a success story and the government and the Central Bank have to be praised for this.
From 1994 to 1998 there were within the so-called Proer Program (Programa de Estimulo à Reestructuracao e ao Fortalecimento do Sistema Financero National) 62 changes in the controlling rights, 33 acquisitions and 44 liquidations of banks. A Deposit Insurance Scheme was created protecting individual deposits up to twenty thousand reais (US$ 11 500 at todays exchange rates).
Very significant was also the increase in the number of foreign banks (from 37 in 1994 to 58 in 1998) and of their branches (from 446 to 2142). The assets of foreign banks increased from 9.5% of total assets in June 1994 to 18.7% in June 1999. The biggest acquisition by foreign banks was the one of Banco Real by the Dutch bank ABN-AMRO. In mid 1999 the most important foreign banks were ABN-AMRO, Bank Boston, HSBC, Citibank, Santander and Sudameris. Their presence increased competition and brought foreign know-how into Brazil.
Finally the number of bank closures accelerated through time reaching a peak in 1999 as capital requirement ratios were progressively increased. As a result, the degree of concentration in the Brazilian banking system increased.
The Brazilian experience suggests that a weak banking system is neither a necessary nor a sufficient condition for an exchange rate crisis to develop. It is not a sufficient condition because Brazil resisted quite well to the Tequila effect of 1994/95 when its banking system was very weak. At that time, its exchange rate was not so overvalued and fiscal policy was far more prudent than in 1997 and 1998. Hence the Brazilian experience of 1994/95 indicates that a weak banking system is not a sufficient condition for a foreign exchange crisis to develop. It is not a necessary condition because in 1999 the Brazilian banking system was quite solid and the crisis broke out despite its strength.

6 – Conclusions: to what extent are foreign exchange crises to be blamed on excessive freedom of international capital flows.
The main conclusions which can be drawn from this paper are first that the exchange rate of the real was already excessively overvalued when the Asian crisis hit Brazil in October 1997. The stabilization of the currency following decades of very high inflation had been successfully based on the nominal exchange rate anchor. However, by mid 1997 the nominal exchange rate rigidity had become excessive. From mid 1994 to mid 1997 the elimination of the distortions caused by the hyperinflation (and the consumption boom of 1994-95 led by consumer credit) allowed the economy to grow moderately despite the overvaluation. Afterwards there were not enough sources of domestic productivity gains and efficiency improvements to move the already overvalued real exchange rate sufficiently towards its equilibrium level. The huge expansion of the weight of the government sector in the economy recorded from 1994 to 1998 contributed to this negative outcome (both via the supply and via the demand side).
Second after the Asian and Russian crises broke out Brazil did not increase the rate of devaluation of the nominal exchange rate with the dollar. This implied a continuation of the previous policy to marginally devalue the currency in real terms.
Third the huge expansion of the government sector and increasing fiscal deficits and debt, especially in 1997 and 1998, were the main reason for the accumulation of large disequilibria in the economy (relative prices between traded and nontraded goods, current account imbalances, public debt and foreign debt). Thus first generation models of currency crisis à la Krugman describe a very relevant feature of the Brazilian currency crisis.
Fourth defending an excessively overvalued currency with increases in interest rates leads sooner or later to a halt in economic growth, an increase in unemployment and a high political cost of the defence which eventually undermines the credibility of the nominal exchange rate anchor itself. This vicious circle was present in the case of France in the 1980s and early 1990s when its defence of the parity with the mark was not yet credible enough, in Italy in the early 1990s in the run-up to the September 1992 foreign exchange rate crisis and in Great Britain in the period 1925-1931 (Eichengreen and Jeanne, 1998). This vicious circle, which is central in second generation models of currency crisis (Obstfeld, 1994), was present also in Brazil in 1995-99 (Tullio and Ferreira, 2000).
Fifth the Brazilian experience of 1994/95 and of 1999 shows that a weak banking system is neither a necessary nor a sufficient condition for a currency crisis to develop when the virus of contagion sweeps around the world economy. Brazil resisted the Tequila effect in 1994/95 when it had a weak banking system and fell seemingly pray to the Russian contagion when it had a strong banking system. Our general conclusion is that necessary and sufficient conditions for a currency crisis to develop are an excessively overvalued currency coupled with wrong fiscal and/or wrong monetary policy, independently of the stregth of the banking system. For instance Italy had a strong banking system and experienced three major currency crises, in 1963, in 1976 and in 1992.
Sixth our analysis suggests that contagion from Asia and Russia should not be exaggerated in explaining the Brazilian currency crisis. Brazilian financial markets started feeling the effects of the Asian crisis only 4 months after its outbreak and the devaluation of the real occurred 19 months after. As to the effects of the Russian crisis they were felt immediately and the devaluation occurred 5 months later. However, as we hope to have shown in this paper, by the time the Russian crisis occurred, Brazil had been “walking on a razor’s edge” for quite some time and it seems to us superficial to attribute the aggravation of the crisis in September-October 1998 just to Russian contagion. In the case of Brazil the virus of contagion travelled slowly enough to give the government a sufficient breathing space to put its house in order and devalue the currency in an orderly manner. Instead of doing that Brazil did exactly the opposite: it made the exchange rate marginally more rigid and it embarked on a major fiscal spending spree. The Brazilian crisis would have occurred anyway, given the rigidity of the exchange rate and the fiscal policy pursued. In this sense what we have described in this paper is a chronicle of a death foretold. This is not to deny that contagion exists and that international financial markets may be at times subject to “herd behaviour” and irrational reactions. However, in the case of Brazil we doubt they were very irrational. On the contrary it seems to us that they sensed better than the Brazilian government and the IMF that the policies pursued by Brazil were unsustainable. If we find some traces of irrationality it is in the overoptimism with which they assessed the situation of Brazil in the first semester of 1998.
Seventh what does the Brazilian experience teach us with regard to the dangers of excessive freedom of private international capital flows? A few observations are in order here.

a) Capital inflows are beneficial to a country. They allow a country to finance a higher investment rate and hence a higher growth rate. In addition they allow a country to run larger current account disequilibria than would otherwise be possible and for longer periods of time. Problematic sudden withdrawals of private capital usually occur if a country forgets for too long that there is an intertemporal international budget constraint which has to be respected. The higher growth rate allowed by private capital mobility has to be weighted against the dangers on the part of policy makers not to be able to realize when they have gone too far with the disregard of the long run solvency of the country. Since judging when a country has gone too far is a very complicated exercise it would be better if countries erred on the side of prudence. They often do not.

b) What may in principle have to be controlled are short term capital flows not long term ones. However, the distinction between short and long term capital flows is easy in theory but difficult in practice. Controlling the first without controlling also the second is rather difficult. Once controls are put in place for all capital flows the country becomes a less attractive place for foreign investors, because market oriented governments are more trustworthy to international investors. The country ends up paying more for the capital it attracts and the amount it receives is smaller. Take the case of Italy which had very tight capital controls until 1990. It was a very unattractive place for foreign portfolio investment until well into the second half of the 1980s. Most of the foreign capital which was flowing into the country was in the form of borrowing from international banks and bond issues in international capital markets by the government and government agencies. In addition the control of capital flows entails non negligible adminstrative costs and they may interfere with current account transactions by increasing the costs of foreign trade.

c) One could make an analogy between capital mobility and the Internet. Both are good but could be abused. Consumer protection is still limited for sales via Internet and Internet can be used to transmit indecent material. Are these sufficient reasons to forbid Internet? Clearly not. The abundance of foreign capital can also be abused by recepient countries. However, this is not a reason to forbid capital mobility. The experience of Brazil after 1997 can be probably considered a case of such an abuse of free access to international capital markets.

We are therefore unsympathetic to the idea of generalized controls on capital flows and agree with the position of the IMF and of the US government on this matter.

References

Calvo G. and Mendoza E.G., “Mexico’s Balance-of-Payments Crisis: a Chronicle of a Death Foretold”, Journal of International Economics, vol. 41, 1996, pp. 235-264.

Cardim De Carvalho Filho J. “The Real Stabilization Plan and the Banking Sector in Brazil”, Banca Nazionale del Lavoro Quarterly Review, vol. LI, no. 206, September 1998, pp. 291-326.

Cardoso E. and Goldfajn I., “Capital Flows to Brazil: the Endogeneity of Capital Controls”, IMF Staff Papers, vol. 45, March 1998, pp. 161-202.

Corsetti G., Pesenti P. and Roubini N., “What caused the Asian Currency and Financial Crisis?”, Temi di Discussione, Servizio Studi Banca d’Italia, no. 343, December 1998.

Dornbusch R., Goldfajn I. and Valdes R., “Currency Crises and Collapses”, Brooking Papers on Economic Activity, no. 2, 1995, pp. 219-293.

Dos Reis Carvalho J. C. and Gianbiagi F. “Cambio real: onde ficamos e para onde vamos”, Conjuntura Economica, June 1999, pp.16-20.

Eichengreen B. and Jeanne O., “Currency Crisis and Unemployment: Sterling in 1931” NBER Working Paper no. 6535, (http://www.nber.org/papers/w6563), Cambridge USA, May 1998.

Eichengreen B. and Masson P. “Exit Strategies: Policy Options for Countries Seeking Greater Exchange Rate Flexibility”, IMF Occasional Paper no. 168, Washington D.C., September 1998.

Goldfajn I. and Gupta P. “Does Monetary Policy Stabilize the Exchange Rate Following a Currency Crisis?”, Texto para Discussao no. 396, Departamento de Economia PUC -Rio, Rio de Janeiro, February 1999.

Goldfajn I. and Valdes R.O., “Are Currency Crisis Predictable?”, European Economic Review, vol. 42, 1998, pp. 873-885.

Krugman P. “A Model of Balance of Payments Crises”, Journal of Money, Credit and Banking, vol. 11, 1979, pp. 311-25.

Masson P., “Gaining and Losing Credibility: The Case of the United Kingdom”, The Economic Journal, vol. 105, May 1995, pp.571-582.

Micossi S. and Padoan P.C., “Italy in the EMS: after Crisis, Salvation?” in “The Monetary Economics of Europe: Causes of the EMS Crisis”, Christofer Johnson and Stefan Collingnon (eds.), Rutherford, N.J., Fairly Dickinson University Press, 1995, pp. 61-83.

Nogueira Batista Jr. P. “O Brasil depois do Plano Real”, Economia Aplicada, vol. 3, March 1999, pp. 95-107.

Obstfeld M. “Rational and Self-Fulfilling Balance-of-Payments Crises”, American Economic Review, vol. 76, 1986, pp. 72-81.

Obstfeld M. “The Logic of Currency Crises” NBER Working Paper no. 4640, 1994

Obstfeld M. “Models of Currency Crises with Self-Fulfilling Features”, European Economic Review, vol. 40, April 1996, pp. 1037-47.

Radlet S. and Sachs J. “The East Asian Financial Crisis: Diagnosis, Remedies, Prospects”, Brooking Papers on Economic Activity no. 1, 1998, pp. 1-90.

Sachs J., Tornell A. and Velasco A. “Financial Crises in Emerging Markets: the Lessons of 1995”, Brooking Papers on Economic Activity, no. 1, 1996a, pp.147-217.

Sachs J., Tornell A. and Velasco A. “The Collapse of the Mexican Peso: what have we learned?”, Economic Policy, 1996b, pp.13-63.

Schwartsman A. “A crise cambial e o ajuste fiscal”, Revista de Economia Politica, vol.19, no.1, January-February 1999, pp.5-29.

Tullio G. “An Empirical Note on Monetary Expansion and Depreciation: the Case of the Lira-Swiss Franc Exchange Rate, February 1973-December 1976”, European Economic Review, vol. 12, 1979, pp. 91-100.

Tullio G., “Exchange Rate Policy and Monetary Integration in Europe and Mercosur”, Nova Economia, vol 8, no. 1, July 1998, pp.9-34.

Tullio G. and Ferreira A., “Unemployment and the Credibilty of Exchange Rate Pegs: Evidence from the Brazilian Currency Crisis of January 1999”, unpublished paper, Federal University of Minas Gerais, Belo Horizonte, April 2000.

Tullio G. and Ronci M., “Brazilian Inflation from 1880 to 1993: Causes, Consequences and Dynamics”, Journal of Latin American Studies, vol. 28, no. 3, October 1996.

—————————————————————————————————————————
TABLE 1 – BRAZIL – KEY ECONOMIC INDICATORS – CUMULATIVE PERCENTAGE CHANGES
—————————————————————————————————————————
Exchange Exchange Consumer Price of Price of
PERIOD Rate Rate Price traded nontraded
R$/US$(1) R$/US$(2) Inflation goods goods
—————————————————————————————————————————
JULY 1994/
DECEMBER 1997 11.3 19.5 54.4 27.5(3) 112.2(3)

JULY 1994/
DECEMBER 1998 20,5 29.5 58.3 26.5(4) 120.1(4)

DECEMBER 1997/
DECEMBER 1998 8.3 8.3 2.4 -0.9 3.7
—————————————————————————————————————————
SOURCE: FGV – CONJUNTURA ECONÔMICA, various issues.

(1) with respect to the official rate prevailing in July 1994 (R$ 1 = US$ 1); (2) with respect to the commmercial rate prevailing in July 1994 (R$ 0.931 = US$ 1); (3) August 1994/December 1997; (4) August 1994/December 1998.

—————————————————————————————-
TABLE 2 – BRAZIL – CHANGES IN THE COMMERCIAL
EXCHANGE RATE WITH RESPECT TO THE US DOLLAR (%)
—————————————————————————————-
December 94 – December 95 13.9
December 95 – December 96 7.1
December 96 – December 97 7.3
December 97 – December 98 8.3

December 96 – June 97 3.6
June 97 – December 97 3.6
December 97 – June 98 3.7
June 98 – December 98 4.4
—————————————————————————————-
SOURCE: Boletim do Banco Central do Brasil, various issues.

—————————————————————————————————————————TABLE 3 – BRAZIL – REAL GDP GROWTH AND RATE OF UNEMPLOYMENT (%)
—————————————————————————————————————————
QUARTER GDP GROWTH (1) UNEMPLOYMENT(2)
—————————————————————————————————————————
1995.2 5.97 4.20

1995.3 0.36 4.86

1995.4 -1.46 5.41

1996.1 -1.70 5.52

1996.2 1.73 5.58

1996.3 6.30 5.33

1996.4 4.96 5.11

1997.1 4.66 5.30

1997.2 4.60 5.58

1997.3 2.59 5.71

1997.4 1.99 6.03

1998.1 1.24 7.28

1998.2 1.47 7.52

1998.3 -0.16 7.64

1998.4 -1.95 7.90
—————————————————————————————————————————
SOURCE: FGV – CONJUNTURA ECONÔMICA, various issues.

(1) with respect to the same quarter of the previous year; (2) seasonally adjusted.

—————————————————————————————————–
TABLE 4 – BRAZIL – TRADE BALANCE – 1990/1999 – US$ BILLION
—————————————————————————————————–
PERIOD PERIOD
—————————————————————————————————–
1990 10.8 1995 -3.3
1o. semester 5.7 1o. semester -4.3
2o. semester 5.1 2o. semester 1.0

1991 10.6 1996 -5.5
1o. semester 7.2 1o. semester -0.3
2o. semester 3.4 2o. semester -5.2

1992 15.3 1997 -8.3
1o. semester 7.0 1o. semester -3.7
2o. semester 8.3 2o. semester -4.6

1993 13.3 1998 -6.4
1o. semester 7.1 1o. semester -2.0
2o. semester 6.2 2o. semester -4.4

1994 10.5 1999 -1.2
1o. semester 6.8 1o. semester -0.6
2o. semester 3.7 2º. Semester -0.6
—————————————————————————————————–
SOURCE: FGV – CONJUNTURA ECONÔMICA, various issues.

—————————————————————————————————————————
TABLE 5 – BRAZIL – EXPORTS – PRICE AND QUANTITY INDICES
—————————————————————————————————–
PERIOD RAW MATERIALS SEMI-MANUFACTURES MANUFACTURES

Price Quantity Price Quantity Price Quantity
—————————————————————————————————–
1995 92.3 97.6 114.9 95.3 99.4 97.5

1996 100.0 100.0 100.0 100.0 100.0 100.0

1997 108.0 112.6 98.5 99.9 98.1 112.7

1998 90.7 120.2 91.0 103.5 96.8 115.0

10/98 83.4 113.9 85.0 109.0 94.5 112.6

11/98 82.8 95.3 81.6 101.0 95.8 107.7

12/98 85.0 98.5 81.3 113.9 95.3 114.5

01/99 86.8 77.5 80.0 100.0 94.3 79.4

02/99 82.4 90.7 77.3 103.2 92.5 93.6

03/99 79.2 109.6 75.9 116.0 90.2 112.1

04/99 76.3 130.8 76.1 102.1 87.9 107.6

05/99 76.0 171.9 73.5 129.8 87.6 119.7

06/99 75.4 153.7 72.6 134.8 87.8 123.1

07/99 75.1 152.7 75.3 119.0 86.4 117.5

08/99 74.7 149.3 76.6 118.1 85.9 129.0

09/99 73.0 145.6 77.4 122.9 85.4 126.4

10/99 73.4 135.5 80.5 128.9 86.5 130.9
—————————————————————————————————–
SOURCE: Fundação Centro de Comércio Exterior – FUNCEX.

—————————————————————————————————————————
TABLE 6 – BRAZIL – COMPOSITION OF EXPORTS – 1995/1998 – % OF TOTAL
—————————————————————————————————–
1995 1996 1997 1998
—————————————————————————————————–
RAW MATERIALS 24.0 25.4 27.8 25.7

SEMI-MANUFACTURES 20.0 18.3 16.3 16.1

MANUFACTURES 56.0 56.3 56.0 58.2

TOTAL (%) 100.0 100.0 100.0 100.0
TOTAL (US$ Billion) 46.51 47.75 52.99 51.12
—————————————————————————————————–
SOURCE: BOLETIM DA MACROMETRICA, various issues.

—————————————————————————————
TABLE 7 – BRAZIL – CURRENT ACCOUNT BALANCE – 1990/1999 – US$ BILLION
—————————————————————————————
PERIOD PERIOD
—————————————————————————————
1990 -3.78 1995 -17.97

1o. semester -0.86 1o. semester -11.88
2o. semester -2.92 2o. semester – 6.09

1991 -1.41 1996 -24.35

1o. semester 1.02 1o. semester – 7.68
2o. semester -2.43 2o. semester -16.67

1992 6.14 1997 -33.27

1o. semester 3.61 1o. semester -14.68
2o. semester 2.53 2o. semester -18.59

1993 -0.59 1998 -35.10

1o. semester -0.03 1o. semester -14.21
2o. semester -0.56 2o. semester -20.89

1994 -1.69 1999 -24.38

1o. semester 1.39 1o. semester -12.29
2o. semester -3.08 2º. Semester -12.09
—————————————————————————————
SOURCE: FGV – CONJUNTURA ECONÔMICA, various issues.

—————————————————————————————————————————-
TABLE 8 – BRAZIL – DECOMPOSITION OF THE CURRENT ACCOUNT – 1990/1999 – US$ BILLION
—————————————————————————————————————————-
YEAR TRADE SERVICES FACTOR TOURISM UNILATERAL CURRENT
BALANCE SERVICES TRANSFERS ACCOUNT
—————————————————————————————————————————-
1990 10.75 -15.37 -11.34 -0.12 0.83 -3.78

1991 10.58 -13.54 – 9.29 -0.21 1.56 -1.41

1992 15.24 -11.34 – 7.83 -0.32 2.24 6.14

1993 13.31 -15.58 -10.11 -0.80 1.69 -0.59

1994 10.47 -14.74 – 8.82 -1.18 2.59 -1.69

1995 -3.35 -18.59 -10.75 -2.42 3.97 17. 97

1996 -5.55 -21.71 -12.21 -3.59 2.90 -24.35

1997 -8.36 -27.11 -15.99 -4.38 2.22 -33.27

1998 -6.42 -30.67 -19.26 -4.27 1.91 -35.18

1999 -1.20 -25.21 na -1.44 2.03 -24.37
—————————————————————————————————————————-
SOURCE: FGV – CONJUNTURA ECONÔMICA – various issues

——————————————————————————————————
TABLE 9 – SEVERAL COUNTRIES – CURRENT ACCOUNT DEFICIT IN
THE YEAR BEFORE THE CURRENCY CRISIS
——————————————————————————————————
COUNTRIES YEAR CURRENT ACCOUNT
DEFICIT – % OF GDP
——————————————————————————————————
MEXICO 1994 -7.0

THAILAND 1996 -8.1

SOUTH KOREA 1996 -4.7

INDONESIA 1996 -3.4

MALAYSIA 1996 -8.4

BRAZIL 1998 -4.5
——————————————————————————————————
SOURCE: IFS-IMF

————————————————————————————————————————-
TABLE 10 – BRAZIL – COMPOSITION OF THE EXTERNAL DEBT – 1995/98 – US$ BILLION
————————————————————————————————————————-
Dec. 1995 Dec. 1996 Dec. 1997 Dec. 1998 Apr. 1999
————————————————————————————————————————-
A. Short Term 30.52 37.79 36.72 24.03 21.05

B.1 Public Sector 3.71 5.23 5.74 4.30 3.61
B.2 Private Sector 26.82 32.55 30.98 19.37 17.44

B. Medium and
Long Term 128.73 142.15 163.28 210.65 210.55

A.1 Public Sector 91.42 88.43 79.97 90.59 100.50
A.2 Private Sector 37.31 53.72 83.32 120.06 110.05

C. Total 159.26 179.93 200.00 234.69 231.60

B / C 0.19 0.21 0.18 0.10 0.10
(A.1 + B.1) / C 0.60 0.52 0.43 0.40 0.45
B / Reserves 0.60 0.64 0.71 0.56 0.48
————————————————————————————————————————-
SOURCE: BOLETIM DO BANCO CENTRAL, various issues.

—————————————————————————————————————————
TABLE 11 – BRAZIL – COMPOSITION OF THE EXTERNAL DEBT – 1995/1998 – % of total
—————————————————————————————————————————
Dec.95 Dec. 96 Dec. 97 Jun. 98 Dec.98 Apr. 99
————————————————————————————————–