Financial contagion in Asia and Latin America.

Chakraborty, Debasish ; Cebi, Merve

I. INTRODUCTION

The Asian financial crisis that originated in Thailand in the summer of 1997, spread globally hitting Russia in the summer of 1998 and Brazil in November of 1999. The crisis was not confined to Asia and did spread into countries that had little interaction with each other. The currency crisis became global. There seems to be an emerging consensus that the crisis originated from speculative attacks on currencies and that these attacks are temporarily correlated at least in the short run. However, what triggered these speculative attacks gave rise to considerable debates among economists and policy planners. This paper considers and explains some of these debates. The paper is organized in the following manner. Section 2 provides a comprehensive survey of literature on currency crisis. Section 3 uses data on three countries in Latin America (Argentina, Brazil and Mexico) to show evidence of contagion; section 4 highlights the difference in currency crisis that erupted in Asia and Latin America, and section 5 summarizes the findings.

II. SURVEY OF LITERATURE

The theoretical literature, for this paper, can be divided into three different types of models for currency crisis. Although they are far from being mutually exclusive, they have very different emphasis on what precipitates currency crisis. While the first generation models sees currency crisis arising as a consequence of unsustainable macroeconomic fundamentals and policy initiatives, the second generation models views currency crisis as a result of self-fulfilling exchange rate panics. The third generation model generally link the currency crisis with banking crisis, although there is no consensus on the issue of causality.

1. First Generation Model

The origin of the first generation model can be traced back to Krugman (1979) where he blames inconsistent government policies as a source of the currency crisis. Krugman argues that any government that runs huge budget deficit and finance that deficit with credit creation and is simultaneously committed to defend the exchange rate runs the risk of precipitating a currency crisis. Specifically if the domestic credit creation is in excess of domestic growth in money demand, then economic agents will shift their portfolio allocation from domestic to foreign currency, which puts a pressure on domestic currency to depreciate. However, if the domestic central bank is committed to defend its currency, it will have no option but to exchange foreign currency for domestic money at the fixed exchange rate. This leads to the depletion of foreign exchange reserves. The initial gradual depletion of reserves will gain momentum once market participants are convinced that the central bank cannot defend the currency any more and devaluation is simply unavoidable. Ultimately, the speculative attack will eliminate the entire stock of international reserves and the fixed exchange rate regime will collapse.

The Model

The first generation model as outlined by Krugman makes the following assumptions:

1. Central bank fix the exchange rate at e and remains committed to maintain that rate.

2. Government runs a budget deficit and the deficit is finance by borrowing from the central bank.

3. Perfect capital mobility, which implies r = [r.sup.*]

4. Private economic agents are rational

The money market equilibrium condition implies that the demand for money ([M.sup.d]) must equal the supply of money ([M.sup.s]). Thus

[M.sup.d]= [M.sup.s] (1)

[M.sup.s] = [B.sub.h] + e [B.sub.f] (2)

Where [B.sub.h] is the government securities held by the central bank and [B.sub.f] is the domestic currency value of international reserves, e is the exchange rate.

[M.sup.d] = P*L (y [r.sup.*]) (3)

Therefore [B.sub.h] + e[B.sub.f] = P.L (Y [r.sup.*])

Assuming fLxed exchange rate we have

* e[B.sub.f] = PL(Yr*) - [B.sub.h] (4)

If [M.sup.d] > [M.sub.s] there is an increase in the interest rate and r > [r.sup.*]. This leads to capital inflow and an increase in international reserves. The process continues as long as [M.sup.d] > [M.sup.s] and it stops when the money market is in equilibrium. If [M.sup.s] > [M.sup.d] that leads to a decline in the interest rate and r < [r.sup.*]. This leads to capital outflow and a decline in international reserves.

In the context of the model [B.sub.f] is treated as a residual between the demand for money balance and [B.sub.h]. If there is a persistent increase in [b.sub.h] with a constant demand for money that implies a fall in [B.sub.f] (assuming that the exchange rate e is constant).

Then if [B.sub.h] approaches a point such that [B.sub.f] approaches 0 then the government is forced to abandon its peg. An important insight that emerges from the first generation model is regarding the timing of the crisis. Even before [B.sub.f] becomes 0, rational economic agents can anticipate this and buys up the entire foreign exchange reserves of the central bank and precipitates the crisis.

In the context of the spread of this crisis, from one country to another, suppose we have two countries A and B. Assume that both the countries indulge in excessive credit creation. With excessive credit creation in country 1, its currency depreciates and by definition leads to currency appreciation in country 2. This leads to a decline in output and price in country 2 and a corresponding decline in the demand for money. This leads to a decline in international reserves in country 2. The lower foreign exchange reserve reduces the ability to withstand speculative attack on currency 2 and accelerate the collapse of the second currency. Thus the currency crisis can spread to the second country where the economic fundamentals are unchanged.

2. Second Generation Models

In the second generation models, instead of focusing on the failure of government policies as the source of currency crisis, the focus is on the market itself and self fulfilling panics. The representative second generation model as formulated by Obstfeld (1986) hinges on the fact that the exchange rate policy followed by most central bank is a result of optimizing a loss function subject to the government budget constraint. The loss function is given by

L = 1/2 ([alpha][[PI].sub.t.sup.2] + [X.sub.t.sup.2]), where (1) (5)

[PI] is the actual exchange rate depreciation and X is the flow of net tax revenue. With the assumption of a small open economy and purchasing power parity, the rate of inflation is equal to the exchange rate depreciation. With fixed exchange rate and no inflation abroad [PI] is 0.

The loss function is minimized subject to the government budget constraint of the form

[rb.sub.t] = [X.sub.t] + [theta]([[PI].sub.t] - [[PI].sub.t.sup.e] (6)

Where [rb.sub.t] is total government obligation with r being the interest rate and [b.sub.t] is the value of all outstanding government debt. [[PI].sub.t.sup.e] is the expected rate of inflation and so ([[PI].sub.t] - [[PI].sub.t.sup.e]) can be interpreted as inflation tax revenue.

Optimal loss function when devaluation is an option is given by

[L.sup.d] ([b.sub.t], [[PI].sup.e.sub.t]) = 1/2 [lambda] [([rb.sub.t] + [theta] [[PI].sup.e.sub.t]).sup.2] (7)

Where [lambda] is <1.

Optimal loss function without devaluation as an option is given by

[L.sup.f] ([b.sub.t][[PI].sup.e.sub.t]) = 1/2 [([rb.sub.t] + [theta] [[PI].sup.e.sub.t]).sup.2] (8)

The optimal value of the loss function with devaluation as an option is lower than the optimal value of the loss function when the government does not have devaluation as an option. Thus it seems that a government can reach socially optimal result by surprise devaluation. However, devaluation also entails cost, mainly in the form of lost credibility of the government. Let such cost be represented by c.

Therefore the government will have incentive to devalue if

[L.sub.d] + c [L.sub.f] (9)

This condition will imply that devaluation will occur if the expectation of devaluation (as measured by the expected inflation rate) is too high or the inherited government debt is too low. Private economic agents are aware of this possibility and they act accordingly. This can turn into a currency crisis if the number of rational economic agents who expects devaluation is greater then the number of participants who don't. The majority action in this case will precipitate a currency crisis. Also since small investors do not have adequate resources to gather information and thus follow big investors who does have the resources to gather and process the information. Hence if the big investor is convinced of an impending devaluation and act on that perception, small investors will follow suit precipitating a currency crisis.

3. Third Generation Model

The third generation models of currency crisis evolved after the Asian financial crisis in 1997. The Asian financial crisis could not be easily explained as resulting from huge budget deficits and subsequent credit creation to finance (first generation model) or from self fulfilling temptations to devalue (second generation model). The crisis was explained by problems arising in the financial and banking sector (Krugman 1998).

The Model

Suppose the country produces one good using capital and labor in a small open economy. Suppose the production is characterized by Cobb--Douglas production function of the type

[Y.sub.t] = [K.sub.t.sup.[alpha]] + [L.sup.(1-[alpha]).sub.t] (10)

Capital in period t is a result of investment in period t-1. Spending is divided between domestic and foreign goods. Suppose that the fraction of consumer and investment spending on foreign goods is given by i and the fraction of spending on domestic goods given by (1-[mu]).

Equilibrium in the goods market is given by

[Y.sub.t] = (1-[mu]) [C.sub.t] + (1-[mu]) [I.sub.t] + [rho][X.sub.t], (11)

where p is the real exchange rate and [rho]Xt is the domestic currency value of export.

With regard to consumption, it must be noted that consumer spends all their labor income (1-[mu]) [Y.sub.t]. Thus we have

[Y.sub.t] = (1-[mu]) (1-[alpha]) [Y.sub.t] + (1-[mu]) [I.sub.t] + [[rho].sub.t] [X.sub.t], (12)

and

[[rho].sub.t] = {1 - (1-[mu])(1-[alpha])[Y.sub.t]] - (1-[mu]) [I.sub.t]}/[Y.sub.t] (13)

With regard to investment, it is assumed that the investor's ability to invest may be limited by its ability to borrow and its ability to borrow may be limited by its wealth. Thus investment is given by

[I.sub.t] = (1 + [theta]) [W.sub.t] (14)

Equation 14 implies that investors can borrow up to e times the wealth and wealth is given by the share of output [alpha][Y.sub.t] minus the debt owned in domestic currency and in foreign currency.

Thus Wealth is given by

[W.sub.t] = [alpha] [Y.sub.t] - [D.sub.t] - [[rho].sub.t][F.sub.t] (15)

Where total wealth W is defines as the total share of capital [alpha][Y.sub.t] minus the debt owed in terms of domestic currency [D.sub.t] and foreign currency [[rho].sub.t][F.sub.t].

Krugman argues that investment decisions are taken by comparing real return on domestic and foreign assets. Thus domestic investment will be attractive as long as

(1 + i) ([P.sub.t] / [P.sub.t+1]) > 1 + [i.sup.*] (16)

Third generation models see the process unfolding in the following way: A decline in capital inflow causes the currency to depreciate, which impacts wealth and consequently the ability to borrow and invest. This further reduces capital inflow and the process continues triggering a crisis.

To see the process more clearly suppose the offer of credit is contingent upon the value of the collateral, which in this model is the wealth. In this model wealth in turn is a function of the real exchange rate as indicated in equation 15.

The model goes further to illustrate that the actual level of investment that can be financed is indeed a function of the expected level of investment. The expected level of investment impacts the real exchange rate (equation 13), which impacts wealth, which in turn impacts the value of the collateral and hence the actual value of finance available and the actual level of investment. Thus

dW/dI = [dW/d[rho]]. [D[rho]/dI] = [(1-[mu])F]/X 17

If wealth is a binding factor for borrowing and subsequent investment, then the level of investment that could actually be financed ([I.sub.f]) is given by

[I.sub.f] = (1 + [theta]) W (18)

Thus the relationship between the actual investment that can be financed (If) and expected level of investment I is given by

dIf/dI = [dIf/dW].[dW/dI] = [(1 + [theta])(1-[mu])F] IX (19)

At very low levels of expected investment, the real exchange rate depreciates and the lenders believe that the borrower has no collateral and hence refuse to loan. So the actual level of investment that can be financed falls dramatically. The model imposes an additional constraint that

[I.sub.t] [greater than or equal to] 0 (20)

Equation 20 implies that the level of investment cannot be negative. This is one extreme. At very high level of expected investment, the wealth constraint becomes non binding and actual investment depends on relative returns on assets. This is the other extreme.

In between these two extreme points, the size of the actual investment is a function of the expected level of investment. This is the zone where d[I.sub.f]/DI > 1.

If for some reason investors are pessimistic, and the expected level of investment falls, that scenario can easily work itself to impact the real exchange rate and make the lender loose confidence on the borrower and refuse to lend. This can explain the crisis. Notice that this crisis may not be due to unsound investment in the past.

It is clear from equation 19 that the factors that contribute to this crisis may be any or all of the following factors: high value of [theta] (high leverage), high value of F/X (high foreign currency debt to export), and low value of [mu] (marginal propensity to import).

III. EMPIRICAL EVIDENCE FROM ARGENTINA, BRAZIL, AND MEXICO

The aim of this section is to investigate the existence of financial spillover or contagion that creates common outcomes in countries with heterogeneous macroeconomic fundamentals. Evidence of contagion in highly integrated financial markets stems from the existence of excess co movement across stock and commodity prices, bond yields, exchange rate changes, and capital flows.

Analyzing the behavior of stock returns, exchange rate changes, and interest rates for the developing countries of Latin America; Brazil, Chile, and Mexico, for the period June 1998--September 2001, we search for "excess co movement" across countries.

Table 1 presents the cross-country correlations on monthly stock market returns of three countries for the period under consideration. As it is apparent in the table, stock returns of Brazil and Mexico are highly correlated; therefore, deserves investigation. Similar results appear with the correlations of exchange rate changes and interest rates as it is shown in Table 2 and 3 respectively. High correlation in interest rates relative to other variables is noteworthy in Table 3.

To analyze the behavior of these correlations over time, we apply a rolling window technique where each window consists of twelve months.

To examine the existence of the so-called "true contagion" we follow a procedure similar to the methodology employed by Valdes (1995), Calvo and Reinhart (1996) to test contagion.

The model is formally,

[y.sub.it] = [[beta].sub.0i] + [[beta].sub.2i][x.sub.1it] + [[beta].sub.2i][x.sub.2it] + [[beta].sub.3x3t] + [[epsilon].sub.it] where,

[x.sub.1] = Country-specific fundamentals

[x.sub.2] = Country-specific international transmission mechanisms fundamentals

[x.sub.3] = Common international fundamentals

[epsilon] = Idiosyncratic country shocks

i refers to country, while t denotes time

In the absence of contagion, [[epsilon].sub.i] is orthogonal across countries. Therefore, the tests involve searching for co movement across countries in these residuals.

We use the ratio of Central Bank foreign assets to reserve money, production growth, ratio of banking sector foreign liabilities to foreign assets and inflation as countryspecific fundamental variables and U.S. Treasury bill rate as common international fundamental variable external to the developing countries of Latin American.

Higher interest rates in the U. S. attract investors to the high-interest yields and deteriorate economic prospects of developing economies. Considering the high external debt burden of these countries, high world interest rates worsen the creditworthiness of debtor countries that borrow at these rates (Fernandez-Arias, 1993). The deteriorated creditworthiness is in turn reflected in the rise in secondary market prices of bank claims on most of the heavily indebted countries.

Ratio of banking sector foreign liabilities to foreign assets come into the picture when deposit banks are required to balance their international accounts every quarter. Their demand for both national currency and foreign currency derive the variations in market for national currency and foreign currency leading an upward trend in interest rates and depreciation given that exchange rate is flexible.

We apply fixed effects panel estimation for each dependent variable and obtain corresponding residuals for each country. Cross-section weighted estimation is preferred to account for heteroskedastic behavior of high frequency data. The objective is to separate the variation in the dependent variable led by fundamental fluctuations. The results are presented in Table 4. Fundamental variables do not significantly explain the variation in stock market returns and depreciation; nevertheless, stock market returns seem to respond Central Bank and commercial banking sector foreign asset position and U.S. Treasury Bill interest rate. Since the countries under investigation tend to use exchange rate as an anchor to monetary stabilization, estimations yield poor results in explaining depreciation via fundamentals. The same reason brings us to better results in interest rate since it is the one to adjust through market mechanism. U.S. T-Bill rates, inflation rate and banking sector fragility tend to explain 68% of the variation in interest rate of 1% significance.

Once we have eliminated fundamental effects on these variables, we next observe the interaction between the residuals obtained from panel estimation to examine the existence of contagion. Formally,

[[epsilon].sub.it] = [[beta].sub.1i] [[epsilon].sub.jt] + [[beta].sub.2i] [[epsilon].sub.kt] + [v.sub.it]

where [v.sub.it] is the lid error term and i, j, k correspond to three different countries.

A constant term is deliberately excluded from the equation since fixed effects are already employed in panel estimations. Above model is applied for every permutation of countries and for each financial variable under analysis. Tables 5-7 display the results of least squares estimations.

High degree of spillovers between Brazil and Mexican stock markets (exchange rate changes, interest rate) are easily observed. 40% of variation in Brazilian stock returns that is not explained by fundamentals is explained by disturbances in Argentinean and Mexican stock markets, which is nothing but an indicator of a true contagion. Decreased mobility of short-term capital due to active government policies in Argentina provides a relatively impermeable financial structure and hence dccreased significance in our estimations. Mexican and Brazilian markets seem to be more volatile in this regard (2).

As mentioned earlier contagion in exchange rate markets barely exist since depreciations in especially Brazil and Mexico are outcomes of accumulated financial distress and usually require government involvement. As expected, disturbances to interest rates in the three countries are highly correlated. Each country owes more than 40% of its variation in interest rates to variation in the other two.

4. NARRATIVE OF THE CURRENCY CRISIS IN ASIA AND LATIN AMERICA

1. Mexican Currency Crisis

The Mexican currency crisis can be explained by first generation currency crisis model. The government of Mexico was firmly committed to maintaining a fixed exchange rate. In 1994, there was a political crisis in Mexico with the farmer rebellion in Chiapas, and the assassination of the Mexican ruling party presidential candidate. Following these two political crisis, the Mexican government decided to follow expansionary monetary and fiscal policy to enhance output and employment. This easy monetary policy to finance budget deficit reduced the interest rate and international reserve. This reduction of the international reserves made it very difficult for the central bank to defend its currency. The central bank then decided to devalue its currency just after the presidential election by 15%. That amount of devaluation proved to be inadequate and the investors interpreted this to be a continuation of the government policy to keep its currency pegged. This had an impact on the credibility of the government. The peso fell sharply to about 50% of its pre crisis value. This led to an increase in import prices and inflation. To counter this move the government had to raise interest rate, which reduced aggregate demand and output. Thus the Mexican currency crisis was a result of unsustainable budget deficit, financed by credit creation and a parallel commitment of the government to maintain a fixed exchange rate.

2. Argentina Currency Crisis

Argentina currency crisis can be adequately explained by the second generation models. Argentina had sound financial fundamentals. Argentina had a currency board which tied the peso to the dollar and every peso in the monetary base was backed by a dollar in the international reserve. However, there was a speculative attack on Peso. With the devaluation of Mexican Peso, there was an upward pressure on the Argentine peso. Speculators were convinced that the Argentine government will decide to abandon its commitment to fixed exchange rate to reduce unemployment. This speculative attack caused a massive capital outflow, and reduced the monetary base which precipitated banking and currency crisis.

3. Brazilian Currency Crisis

Brazilian currency crisis was a combination of both the first generation and second generation model. The Brazilian government could not keep the budget deficit under control. The government realized that this deficit could have an impact on the value of it currency as explained by the first generation models. The government negotiated with the IMF for a three year thirty billion dollar stand by loan. The IMF approved that loan with the understanding that reform package, mainly dealing with the fiscal deficit, would be approved by the Brazilian legislatures. That did not happen and the central bank had to finance the deficit with credit creation. This reduced the interest rate and led to capital outflow and precipitated a currency crisis. The government abandoned its commitment to fixed exchange rate and the real was allowed to fluctuate in the market. Although the real was allowed to fluctuate freely, there was a panic by investors which interpreted the situation as a deep flaw in the economic fundamentals. This panic resulted in further fall of the real, and the real fell 58% against the US dollar in two months.

5. Asian Currency Crisis

The Asian currency crisis however, could not be explained either by the first generation model, since before the crisis their money supply did not grow faster then their GDP, inflation was under check and government run a budget surplus. This combination ruled out the first generation models as a basis for explaining the crisis. Although their economy was slowing down it was not sever enough to create a panic run on the currency. This ruled out the use of the second generation models to explain the basis of the currency crisis that engulfed the East Asian economies in the late 1990s. To look into the real cause of the currency crisis in East Asia, one has no choice but to look into the third generation models which explained the crisis in terms of banking balance sheet crisis.

The Plaza accord (the meeting of the G7 countries) in 1985 resulted in the appreciation of the Japanese currency against the America dollar. This led the Japanese to invest heavily in East Asia for two reasons. The first reason was that these currencies were tied to the US dollar and secondly these countries had cheap labor. In addition, the Japanese economic slow down in the early 90s, prompted the Japanese Central bank to ease money supply to boost aggregate demand. However, this increased liquidity did not boost Japanese aggregate demand but found way to East Asia where the interest rate was much higher. The countries of East Asia namely Korea, Thailand, Malaysia, Indonesia did experience high capital inflow. At the same time that these countries were experiencing high capital inflows they were deregulating their capital markets and lifting all restrictions on capital inflows and outflows. The average savings rates in these countries were around 30%, and since other financial instruments like bonds and stocks were not in fashion, the banks were the main intermediaries. The bank channeled the household savings into corporate debt and with no proper regulation the debt equity ratio of the corporations became very high. With bad loans and the ratio of non performing assets increasing, the banking sector faced a crisis. This crisis led to a currency crisis in two ways. First with the banks being insolvent was unable to advance any further credits. This impacted investment adversely and affected the real sector. Second the Clinton administration reduced the American budget deficit and that appreciated the US dollar against the Japanese yen. Since the currency of the East Asian economy was tied to the dollar, they found that there was appreciation of their currencies which led to a serious current account deficit. Current account deficit increased from around 2% in 1993 to about 5% in 1996. These huge current account deficits were previously financed by capital inflow. With falling currency values, the central bank needed to defend the currency. But with weak balance sheets of commercial banks, the central bank was unable to defend their currencies. With falling currencies, many debtors were unable to pay of their foreign short term loans and foreign and domestic investors started withdrawing their deposits. The capital outflow weakened the currencies further and eventually precipitated a currency crisis.

5. CONCLUSIONS

Currency crisis as witnessed in Asia and Latin America had its origin in different macroeconomic conditions. Whereas the Crisis in Mexico can be traced to unsustainable and inconsistent macroeconomic fundamentals, the crisis in Argentina was a result of pure speculative attacks although the macro-economic conditions were not that unusual. The crisis in Brazil was a combination of weak fundamentals and brutal speculative attacks by investors. The crisis in Asia was not that diverse in its origin. It seemed that all the countries in East Asia faced a currency crisis that resulted from a crisis within the banking sector. However there seems to one commonality in all the crises. The large scale capital market liberalization with no effective monitoring and the commitment by the government to peg the exchange rate seems to the driving force behind the crises.

APPENDIX A--THE DATA

Data used in estimations is obtained from country pages of monthly publications of International Financial Statistics (IFS).

Depreciation: End of period values of official exchange rate expressed in national currency units for SDR (the unit of account for IMF) is used to calculate monthly depreciation.

Foreign Assets/Reserve Money: Foreign assets of monetary authorities divided by stock of reserve money comprising currency in circulation, deposits of the deposit money banks, and deposits of other residents, apart from the central government, with the monetary authorities. Monetary authorities data in IFS generally consolidates the accounts of the central bank with the account s arising from monetary functions undertaken by other institutions.

Foreign Liabilities/Foreign Assets: Foreign Liabilities of deposit money banks divided by Foreign Assets of those banks. Deposit money banks comprise commercial banks and other financial institutions that accept transferable deposits, such as demand deposits.

Interest Rates: Money market rate is taken for Brazil and Mexico. Lending rate is taken in case of Argentine as a proxy since the former is not available.

Inflation: Consumer Price Index is used to calculate monthly inflation figures.

Output Growth: Industrial production is used to calculate monthly growth rate of output in case of Mexico and Brazil As for Argentina, manufacturing output is used as a proxy to production.

Stock Returns: Share price indices of IFS consisting of common shares of companies traded on national or foreign stock exchanges are used to calculate monthly logged returns.

USTB: 3-month U.S. Treasury Bill interest rates in the secondary market are used.

References

Burnside, Craig, Martin Eichenbaum, and Sergio Rebelo (1998), "Prospective Deficits and the Asian Currency Crisis," NBER Working Paper No. 6758.

Calvo, Guillermo A., Leonardo Leiderman, and Carmen M. Reinhart (1996), "Inflows of Capital to Developing Countries in the 1990s," The Journal of Economic Perspectives, Vol. 10, No. 2 (Spring), pgs. 123-139.

Calvo, S., Carmen M. Reinhart, "Capital Flows to Latin America: Is There Evidence of Contagion Effects?" Private Capital Flows to Emerging Markets After the Mexican Crisis (Guillermo Calvo, Morris Goldstein, and Eduard Hochreiter, eds.), Washington DC: Institute for International Economics.

Calvo, Guillermo A. and Enrique G. Mendoza (1997), "Rational Herd Behavior and the Globalization of Securities Markets,", Duke Economics Working Paper No. 97-26.

Caramazza, Francesco, Luca Ricci, and Ranil Salgado (2000), "Trade and Financial Contagion in Currency Crises,", IMF Working Paper 00/55 (March).

Corsetti, Giancarlo, Paolo Pesenti, and Nouriel Roubini (1999), "Paper Tigers? A Model of the Asian Crisis," 1999, European Economic Review 43, No. 7 (June), pgs 1211-36.

Dooley, Michael (1997), "A Model of Crises in Emerging Markets,", NBER Working Paper No. 6300.

Eichengreen, Barry, Andrew K. Rose, and Charles Wyplosz (1995), "Exchange Market Mayhem: The Antecedents and Aftermath of Speculative Attacks,", Paper Prepared for the Economic Policy Panel, April 21-22.

Eichengreen, Barry, Andrew K. Rose, and Charles Wyplosz (1997), "Contagious Currency Crises,", NBER Working Paper, No. W5681.

Glick, Reuven, Andrew K. Rose (1998), "Contagion and Trade: Why are Currency Crises Regional?" CEPR Discussion Papers, No. 1947.

Fernandez-Arias, Eduardo (1993), "The New Wave of Private Capital Inflows: Push or Pull?" Mimeo, World Bank.

Gandolfo Giancarlo (2002), "International Finance and Open-Economy Macro-Economics" Springer Publications.

Goldfain, Ilan, Rodrigo Valdes (1997), "Capital Flows and the Twin Crises: The Role of Liquidity", IMP Working paper 97/87.

Kaminsky, Graciela, Saul Lizondo, and Carmen M. Reinhart (1998), "Leading Indicators of Currency Crises", IMF Staff Papers, Vol. 5, No. 1 (March): 1-48.

Kaminsky, Graciela, and Carmen Reinhart (1999), "The Twin Crises: The Causes of Banking and Balance-of-Payments Problems", American Economic Review, 89, No. 3, 473-500.

Kaminsky, G. L., Carmen M. Reinhart (2000), "On Crises, Contagion, and Confusion", Journal of International Economics, 51: 145-68.

Krugman, Paul (1979), "A Model of Balance-of-Payments Crises" Journal of Money, Credit, and Banking, 11, 311-25.

Krugman, P. (1998a), "Currency Crisis", http://web.mit.edu/krugman/www/crisis.html

Krugman, P. (1998b), "What Happen to Asia"?. January 1998, http://web.mit.edu/krugman/www/ DISINTER.html

Krugman, P. (1999), balance Sheets, the Transfer Problem, and Financial Crisis in P. Isard, A. Razin and A. K. Rose (eds), International Finance and Financial Crisis: essays in honor of Robert. P. Flood, Jr., Norwell, Mass: Kluwer.

MdKinnon, Ronald, and Huw Pill (1997), "Credible Economic Liberalizations and Overborrowing", American Economics Review, 87, No. 2, (May): 189-93. Papers and proceedings of the 109th annual meeting of the American Economic Association, 189-193.

Mishkin, Frederic S. (1996), "Understanding Financial Crises: A Developing Country Perspective", In Annual World Bank Conference on Development Economics, Washington, DC: World Bank, 29-62.

Obstfeld, Maurice (1986), "Rational and Self-Fulfilling Balance-of-Payments Crises", American Economic Review, 76, 72-91.

Obstfeld, Maurice (1994), "The Logic of Currency Crises", Cahiers Economiques et Monetaries, 43, 189-213.

Pesenti, Paolo and Cedriv Tille (2000), "The Economics of Currency Crises and Contagion: An Introduction", Economic Policy Review, Vol. 6, No. 3.

Salant, Stephen, and Dale Henderson (1978), "Market Anticipations of Government Policies and the Price of Gold", Journal of Political Economy, 86, 627-48.

Valdes, Rodrigo (1997), "Emerging Markets Contagion: Evidence and Theory", Publications of Banco Central de Chile, 03/97, Also available at http://www.bcentral.cl/Estudios?DTBC/07/07.htm.

Velasco, Andres (1987), "Financial and Balance of Payments Crises", Journal of Development Economics, 27: 263-83.

Notes

(1.) For a detailed discussions of the models refer to "International Finance and Open-Economy MacroEconomics by Giancarlo Gandolfo.

(2.) Though not reported in this paper, spillovers in stock markets are observed to be of an asymmetric nature where negative shocks tend to affect other countries more than positive shocks.

DEBASISH CHAKRABORTY

Central Michigan University, Michigan

MERVE CEBI

Michigan State University, East Lansing, Michigan

Table 1
Correlations of Monthly Returns on Share Prices

Country Brazil Argentina Mexico

Brazil 1.00
Argentina 0.38 1.00
Mexico 0.71 0.19 1.00

Table 2
Correlations of Monthly Depreciation Rates with Respect to SDR

Country Brazil Argentin Mexico

Brazil 1.00
Argentina 0.13 1.00
Mexico 0.2 0.38 1.00

Table 3
Correlations of Monthly Interest Rates in the Money Market

Country Brazil Argentina Mexico

Brazil 1.00
Argentina 0.60 1.00
Mexico 0.41 0.38 1.00

Table 4
Panel Estimation Results

Dependent
Variable

Explanatory Stock Returns Exchange Rate Interest Rates
Variables Depreciation

CB Asset/R.eserves -0.05 5.35 4.76
 (0.03)([dagger]) (2.54) * -3.37
Deposit Banks -0.05 -1.01 -11.92
Liabilities/Assets (0.03)([dagger]) -3.19 (4.15) **

Output Growth -0.0001 0.03 -0.01
 -0.0002 -0.02 -0.05

Inflation Rate -0.006 -0.66 5.64
 -0.008 -0.96 (1.28) **

US treasury Bill -0.013 -0.42 -1.96
Interest Rates (0.004) ** -0.53 (0.72) **

Constant-Brazil 0.15 -2.28 24.69
Constant-Argentina 0.11 0.29 21.89
Constant-Mexico 0.24 -4.10 39.32
(Weighted) Adjusted 0.05 0.05 0.68
[R.sup.2]
(Weighted) F 2.03 1.09 57.9
Statistic

** Notes: Figures in Parentheses are the standard deviations.
([dagger]), *, ** show significance at % 10, % 5, %1 levels
respectively.

Table 5
Least Squares Estimation Results for disturbances in Stock Returns

Dependent
Variable

Explanatory [epsilon] Brazil [epsilon] Argentina Mexico
Variables

[epsilon] Brazil -- 0.18 0.39
 (0.07) * (0.09) **
[epsilon] Argentina 0.94 -- -0.31
 (0.38) * -0.27
[epsilon] Mexico 0.84 -0.13 --
 (0.21) ** -0.11 -0.05
Adjusted [R.sup.2] 0.4 0.14 0.3

F Statistic 22.4 ** 6.24 * 15.31 **

Notes: Figures in Parentheses are the standard deviations.
([dagger]), *,** show significance at %10,%5,%1 levels respectively.

Table 6
Least Squares Estimation Results for Disturbances in Depreciation

Dependent
Variable

Explanatory [epsilon] Brazil [epsilon] Argentina Mexico
Variables

[epsilon] Brazil -- 0.01 0.03
 (0.04) (0.05)
[epsilon] Argentina 0.19 -- 0.51
 (0.73) (0.20) *
[epsilon] Mexico 0.31 0.33 --
 (0.58) (0.13) *
Adjusted [R.sup.2] -0.013 0.15 0.16
F Statistic 0.56 6.78 * 7.05 *

Notes: Figures in Parentheses are the standard deviations.

([dagger]), *,** show significance at %10, %5, %1 levels
respectively.

Table 7
Least Squares Estimation Results for Disturbances in Interest Rate

Dependent
Variable

Explanatory [epsilon] [epsilon] [epsilon]
Variables Brazil Argentina Mexico

[epsilon] Brazil -- 0.38 0.29
 (0.11) (0.07) **
[epsilon] 0.75 -- 0.02
Argentina (0.21) ** (0.13)

[epsilon] Mexico 1.05 0.03 --
 (0.28) (0.24) *
Adjusted [R.sup.2] 0.60 0.42 0.44
F Statistic 49.2 ** 24.15 ** 26.25 **

** Notes: Figures in Parentheses are the standard deviations.
([dagger]), *,** show significance at %10,%5,%1 levels respectively.