Herd behavior and the 1997 Asian crisis: further evidence.

Khan, Saleheen ; Islam, Faridul ; Park, Kwang Woo "Ken" 等

Abstract

This paper presents empirical evidence in support of the claim that herd behavior can explain the severity of the 1997 Asian crisis above and beyond macroeconomic fundamentals. We compare cross-country correlations between crisis and tranquil periods among the nations of Thailand, Malaysia, Indonesia, Korea, and the Philippines, with regard to such macro variables as, stocks returns, interest rates, exchange rates, and foreign reserves. Macro models are constructed and implemented to capture the pure contagion effects on markets. After controlling for the economic fundamentals for these economies, we find strong evidence of herding contagion.

JEL classification: F30, G15

Keywords: Financial Markets, Currency crisis, Herd behavior, Contagion, Correlation

I. INTRODUCTION

The 1997 Asian crisis has rekindled the debate over currency contagion. When the Thai Government abandoned the dollar exchange rate peg on July 2, 1997, the looming crisis quickly spread through much of East Asia. How this crisis spread so rapidly from the first victim to the rest has baffled the policy makers and has attracted the attention of a growing number of economists and professionals, all searching for answers.

According to Calvo and Reinhart (1996), herding contagion is rooted in factors that are independent of economic fundamentals. Such contagion is more likely to occur when common shocks or all channels of interdependence are not present or controlled for. In other words, a herding framework assumes that individual investor simply follow other investors where market sentiment provides the main dynamic force. As Radelet and Sachs (1998) argue, to a foreign creditor the basic economic characteristics of the Southeast Asian countries prior to the crisis were indistinguishable from one another. To them, if Thailand was in trouble, any other country in the region could be the next. Therefore, creditors' mood was to get out of the region as soon as possible, something that potentially can lead to herding behavior in the financial market.

The objective of the paper is to examine the 1997 Asian crisis in an effort to provide further empirical evidence in favor of the existence of herding contagion; and also to offer a set of general policy insights based on the observed facts. We define contagion as shocks that are in excess of what can be normally explained by economic fundamentals alone. (1) The paper tests for contagion by comparing the cross-country correlations between crisis and tranquil period, among the macro variables for Thailand, Malaysia, Indonesia, Korea, and the Philippines. The variables selected relate to stocks index, interest rates, exchange rates, and foreign reserves. The approach used here is relatively common in the prevailing literature on contagion (2), but differs in one major aspect. Most of the prevailing research uses high frequency data and thus, cannot control for macroeconomic fundamentals and global shocks. By contrast, this paper uses monthly data which allows us to address the macro issues. Given that our research focus is about herding contagion, we consider it is of significant importance to control for macroeconomic fundamentals and global shocks to better understand the forces at play.

The paper closely relates to Baig and Goldfajn (1999), but differs from them in approach and methodology. While these authors examine whether cross-country correlations among currencies, stock returns, interest rates, and sovereign spreads in emerging markets increased during the Asian crisis, we examine the cross-country residual correlations among variables of currencies, stock returns, interest rates, and foreign reserves after controlling for macroeconomic fundamentals and global shocks. The prevailing models try to capture the contagion using the correlations of a set of macro variables across nations. The correlations obtained by using this procedure can however, be biased due to its link with macro variables and thus may fail to capture pure contagion. Masson (1998), and Pindyck and Rotemberg (1990) also use the concept of pure contagion and argue that co-movements of macro variables between nations cannot be explained by economic fundamentals alone, because of the role of market sentiments. Residuals filter out the effects of other included relevant macro variables. So, correlations obtained from the residuals are expected to better represent market sentiments or herding contagion. The paper thus is a significant contribution to the literature.

The paper is organized as follows. Section II briefly reviews the literature. Section III describes methodology, and data. Empirical results are reported in Section IV. Section V offers some policy perspective. Section VI concludes the paper.

II. REVIEW OF LITERATURE

Following the US stock market crash of 1987 changes in the cross-country correlation have been the standard to measure contagion. King and Wadhwani (1990) applied this approach to the US, British, and the Japanese data and found evidence of significant rise in correlations after the crash. Lee and Kim (1993) also find similar results. Calvo and Reinhart (1996) find evidence of a rise in correlation between weekly returns on equities and Brady bonds for Asian and Latin American emerging markets, after the Mexican crisis. Baig et al. (1999) and Khan et al. (2005) also find evidence of increased cross-market correlations. Forbes and Rigobon (2002) counsel caution in interpreting the increased correlations as evidence of contagion because the returns correlations can be the product of statistical artifact, and more so in a volatile stock market. Using a corrected conditional heteroskedasticity method they showed that once the bias is adjusted, the evidence of contagion disappears in all three major crises (Asia 1997, Mexico 1994 and the US 1987). Bartram and Wang (2005) and Corsetti, Pericoli, and Sbracia (2005) question Forbes and Rigobon (2002) results because of its reliance on particular assumptions about the underlying stochastic process of the stock returns. The authors show that the adjustment produces serious biases in test in favor of the null hypothesis of "no contagion." Masson (1998), and Pindyck and Rotemberg (1990) also use the concept of pure contagion and argue that co-movements of macro variables between countries cannot be explained by economic fundamentals alone, because of the role of market sentiments.

III. EMPIRICAL FRAMEWORK

(a) Data

The paper uses monthly data on nominal and real exchange rate, interest rate, consumer price index, money supply (M2), trade volume, domestic credit, and stock return. The sample period covers 1994.1 through 1999.12. For interest rate, we use the money market rates for Thailand, Indonesia, and Korea; and Treasury bill rates for Malaysia and the Philippine. All data are taken from the IFS CD ROM, except for stock return and real exchange rate. Stock return series are from the Bloomberg, and real exchange from JP Morgan. The 3-month US Treasury bill rate represents foreign interest rate.

(b) Methodology

In line with Baig et al. (1999) and Forbes and Rigobon (2002), we define contagion as a significant increase in cross-market linkages, after an initial shock to one country or a group of countries. Within this framework, test for contagion boils down to verifying if the cross-market co-movements increase significantly after a shock. The argument is: If correlations increase significantly in the crisis compared to tranquil period, one may conclude in favor of herding contagion because international financial markets tend to move more closely together during a period turbulence. Due its simplicity, the approach is standard in the literature on contagion.

Whether or not the Asian currency crisis was contagious is an empirical question. As defined, contagion refers to a substantial increase in the cross-market linkages following a shock to an single or a group of countries (3). Our task is to tests if such a rise in some of the channels of economic linkages did take place after the shock. Possible channels for shock transmission include; stock market, interest rate, currency, and foreign reserves for the economies of Indonesia, Korea, Malaysia, the Philippines, and Thailand (4).

We obtain two different measures of correlation for the above noted series for the five Asian nations, most hit by the crisis. First, we compute and compare the cross-market correlations for these variables both for the crisis and for the tranquil periods. Failure to control for global shocks or macro fundamentals can produce an upward bias in the estimated correlation coefficients.

A high correlation among the variables could simply be the result of the similarities in economic fundamentals and common global shocks. To correct for this, a second measure of correlation was computed from the residuals in lieu of the actual series. These residuals were obtained by running a several regressions (levels data). A total of five regressions, one each for the variables of stock return, interest rates, exchange rates, and foreign reserves produced five sets of residuals series for each of the five sample countries. The regressions were run on a set of macroeconomic variables (levels), which were chosen for their theoretical and empirical relevance. The first residual series for the stock return was obtained by regressing it on nominal exchange rate, interest rate, price level, and the US interest rate. Likewise, we specify the other equations for the variable of interest rates, exchange rates, and foreign reserves, and ran regression for each of the five countries (5,6) The U.S. interest rate serves as a proxy for global shock and tacitly recognizes its influence on the emerging markets. These residuals, and not the actual series, were used to compute the correlations which produced ten pairs of such coefficients for each of the five variables. Interest rates were regressed on money supply, foreign reserve, price level, and U.S. Treasury bond rates. Nominal exchange rates were regressed on money supply, real exchange rates, bank credit, and trade volume. Foreign reserves were regressed on money supply, nominal exchange rate, pride level, and interest rates.

We employ the methodology developed by Forbes and Rigobon (1999) and Baig and Goldfajn (1999). We apply a two-sample t-test to check whether correlations are significantly different in two periods. The test hypotheses are the following:

[H.sub.0] : [[rho].sup.0.sub.i,j] [greater than or equal to] [[rho].sup.1.sub.i,j] [H.sup.1] : [[rho].sup.0.sub.i,j] < [[rho].sup.1.sub.i,j]

Where [[rho].sup.i.sub.i,j] is the correlation coefficient between country i and country j over period t. The tranquil and crisis period is denoted by "0" and "1" respectively. Baig and Goldfajan (1999) derived following test statistic:

T = [[bar.x].sub.0] - [[bar.x].sub.1]/[([s.sub.0.sup.2]/[n.sub.0] + [s.sub.1.sup.2]/[n.sub.1]).sup.1/2] (1)

'The test statistics follows the t-distribution, and degrees of freedom are calculated as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (2)

[[bar.x].sub.t], and [s.sub.t.sup.2] refer to the estimated sample mean and variance, and [n.sub.t] refers to the sample size. The correlation coefficients are transformed through a Fisher procedure. The estimated correlation coefficients are approximately normally distributed, with mean and variance given by:

[[micro].sub.t] = 1/2ln (1 + [[rho].sup.t.sub.i,j]/1 - [[rho].sub.i,j.sup.t])

[[rho].sub.t.sup.2] = 1/[n.sub.t] - 3 (3)

We also apply the likelihood ratio test to investigate the significance of the groupwise correlations. Following Valdes (1997) and Pindyck and Rotemberg (1990) the hypotheses under test can be stated as follows:

[H.sub.0] : No groupwise correlations

[H.sub.1]: The null is not true

The test statistic: LR = -N log[R] is [chi square] distributed with 1/2q(q - 1) degrees of freedom. Where, [absolute value of R] is the determinant of the correlation matrix, N is the number of observation in the pooled sample, and q is the number of series being tested.

IV. RESULTS

Tables 1 and 2 present residuals correlations of stock returns controlling for macroeconomic fundamentals and global shock. The two-sample t-test under this scenario reveals that eight of the ten pairs of residual correlations are significantly greater in the crisis period. Further, the signs of correlations change from a strong negative to strong positive; for the pairs of Malaysia-Thailand and the Philippine-Thailand. However, Indonesian markets do not show significant increase in the correlations in the crisis period compared to Malaysian and the Philippine markets, after controlling for fundamentals. Therefore, it appears that the contagion is present in the stock markets. This is supported by rise in correlation in the crisis period, even after controlling for fundamentals and a global shock in the other markets. Furthermore, the LR test reveals statistically significant group-wise correlations of residuals of stock returns for both tranquil and crisis period.

Tables 3-4 present residual correlations for interest rates after controlling for macro fundamentals and global shock. The two-sample t-test shows that nine of the ten pairs of residual correlations are significantly greater in the crisis period. The correlations for the pairs, Indonesia-Korea, Indonesia-Thailand, Korea-Malaysia, and Korea-Philippine change from mild negative to strong positive. This implies that even after controlling for fundamentals and a global shock, contagion seem to permeate in to the interest rate markets; and that it increases in the crisis, compared to the tranquil period. Even though the LR test fails to establish statistically significant group-wise correlations for the residuals for the tranquil period; it shows significant group-wise correlations for the crisis period.

For exchange rates, the cross-market correlations increase significantly for seven pairs, from the tranquil to the crisis period after controlling for macro fundamentals and global shock. These results are presented in Tables 5 and 6. The LR test reveals statistically significant group-wise correlations for the crisis period. In particular, after controlling for the economic fundamentals, there is no evidence of significant groupwise correlations for the tranquil period; whereas it turns out to be very significant for the crisis period.

Tables 7 and 8 report the cross-market correlations for foreign reserve, for both the tranquil and the crisis periods. The two-sample t-test shows that cross-market correlations are significantly higher in the crisis period. When economic fundamentals are not controlled, six of the ten pairs of correlations increase significantly during the crisis period. Nine of the ten pairs of residual correlations of foreign reserves increased significantly in the crisis period. The LR test reveals significant group-wise correlations for both the tranquil and the crisis periods.

Sensitivity Analysis

Our finding that correlation coefficients increase significantly in the crisis period compared to the tranquil period is notably obvious. Various sensitivity tests were conducted whereby outliers were excluded or sample periods in the data were changed. The results appear to be relatively robust with respect to this test.

V. POLICY ISSUES

Crises can be transmitted either through temporary or permanent channels. In the case of former channels, short run isolation strategies, e.g., capital control can be effective in controlling contagion. If crises are transmitted mainly through latter channels that existed even before the crisis, then short run isolation strategies will only delay countries adjustment to a shock. Since we find evidence of contagion even after we control for macro control and global shocks, we argue capital controls as an effective method in controlling contagion. At the same time, policies should also aim at reducing countries' vulnerability to long- run linkages through trade, macro economic interdependence, and financial linkages. Even though we find evidence of herding contagion in the Asian crisis, it is possible for the markets to have long run linkages.

The 1990's Asian financial crisis has invoked strong interest in the search for its causes and for a prescription for future guidance. Most agree that lack of transparency, poor market structure and excessive capital flows were at the center of the crisis. The similarities across the victims were sufficient for developing common guidelines. Policy however, must be tailored to the specific needs and structure of a particular nation. Knowledge of the early warnings signs can help policy makers initiate measures even prior to the onset of a crisis, or to minimize its effects when it takes place.

Both the Mexican and the Asian crises point to the risk of currency peg. Usually maintained to the US dollar, peg masks real fall in the value of the domestic currency and thus makes it difficult to respond to sudden shock like large capital outflows. Even so, peg is adopted for non-market or political reasons. Strong external link facilitates transmission of contagion via co-movement of exchange rates. While foreign direct investment aids economic growth, speculative capital inflow remains a major threat. Sound macro policy along with strict accountability act as deterrent against speculation. As the recent crises demonstrate, peg policy contains seeds of the crisis. In small economies, peg increases financial instability, reduces export competitiveness, widens trade deficit, and encourages speculative attack. Large foreign debt hastens a crisis. In flexible rates regime information availability makes the agents aware of the risk and allows policymakers to respond, as needed.

Manipulative exchange rate to gain competitive advantage in the short run is not uncommon, although it can be very risky in the long run. In a non market exchange rate regime, trade tends to favor some nations at the expense of others, threatening international monetary stability and promoting "beggar thy neighbor policy" (Corsetti et al., 1999). It happens by creating a parallel flow of capital to balance the trade sector.

Sudden capital flows can trigger crisis, destabilizing the exchange rate market. Capital outflows can be associated with dumping local currency and lead to devaluation, as in Thailand. Capital outflow due to exchange rate crisis is a symptom of fundamental problems, not a cause of the crisis. Unlike Korea and Thailand, Malaysian achieved limited success by capital control. Yet such control is not recommended because the costs often exceed the benefits.

Strong banking sector helps maintain sound economic fundamentals. When banks receive implicit government backing, a fragile financial structure can result (Dekle and Kletzer 2001). Repressive financial system also leads to total breakdown (Edward 2001). Anay distortions cause market overreaction, promote herding behavior, and lower economic activities far more than initially anticipated (Chang and Velasco 1999). A strong bond market creates a sobering effect on exchange rates volatility. Governments should promote macro and financial policies for proper bond market development (Eichengreen et al. 2004). As part of overall financial architecture, a sound banking system insulates an economy from crisis. Independent regulatory bodies insure oversight and implement rule of law. Strong viable financial institutions, with limited regulation on capital flows and transparency can lead to the desired outcomes and can help locate sources of financial fragility (Cowan and Gregorio 2005; Kaplan and Rodrik 2001). Over-regulation creates financial vulnerabilities and even causes sharp capital account reversals. Cronyism only makes things worse. Sound 'financial architecture,' can establish balance between exchange rate regimes, capital flows, and currency crises. The Mexican, Asian, Russian, and Brazilian crises point to the need for reform of financial architecture. (7).

Trade diversification, by commodity and by regions, may shield from the severit of contagious crisis by de-linking overly dependent trade relation. A well planned trade expansion policy can also minimize the risk of exposure to major external economic shock. Free trade zone or common currency, like Euro from Australia to East Asia may be beneficial.

Globalization and openness is desirable for benefits of the global community but rush in to it will have undesirable side effects. Pressure for economic liberalization will be ever greater in the future. Increased openness adds linkages and more vulnerability in the short run. Failure to address the changing global dynamics can produce the exact opposite effect of free trade. Some nations may have taken decisions when they were not ready.

VI. CONCLUSION

The empirical framework used in this paper supports herding in the Asian crisis. When the crisis erupted in Thailand, investors quickly pulled out of Asian economies without drawing any distinction between them. The general consensus among the investors was, if Thailand was in trouble, so would be other Asian "tigers".

The paper finds evidence of herding contagion among the East Asian financial markets during the crisis period of 97. We investigated the cross-country residual correlations among variables of currencies, stock returns, interest rates, and foreign reserves after controlling for macro fundamentals and global shocks. We find that correlations among the financial markets increased significantly during the crisis, compared to the tranquil period. Historically, some of the markets were strongly correlated in the tranquil period and the correlations increased significantly in the crisis periods. The evidence of herding contagion leading to the 1997 Asian crisis seems strong.

Knowledge of the transmission of crisis, temporary or permanent is helpful. Policy response would vary by the type of channel. The evidence of contagion even after controlling for macro variables and global shocks suggest that capital controls is likely to be more effective. We find evidence of herding contagion in the Asian crisis but that does not preclude the prospect of cross market long run linkages. The policy should aim at reducing the vulnerability to a shock using both short run adjustment of capital markets and long run adjustment of linkage parameters. Monetary authorities need to monitor exchange rates vis-a-vis balance of payments situation. Responsible fiscal policy can help maintain appropriate interest rate which is a major force of economic instability. These two policies must be aligned with a sensible and internally consistent commercial policy to guide import and export in response to emerging needs to insure both internal and external balance.

References

Baig, T. and I. Goldfajn (1999), "Financial Market Contagion in the Asian Crisis," IMF Working Paper, 46(2): 167-95.

Bartram, T. and Y. H. Wang (2005), "Another Look at the Relationship Between Cross-Market Correlation and Volatility," Finance Research Letters, Vol. 2, 75-88.

Calvo S. and C. Reinhart (1996), "Capital Flows to Latin America: Is there Evidence of Contagion Effect?" in Private Capital Flows to Emerging Markets After the Mexican Crisis (G. A. Calvo, M. Goldstein, and Eduard Hochreiter, eds.) Institute for International Economics, Washington, D. C.

Chang R. and A. Velasco (1999), "Liquidity Crises in Emerging Markets: Theory and Policy," NBER Working Papers 7272.

Corsetti, G, P. Pesenti, and N. Roubini (1999), "What Caused the Asian Currency and Financial Crisis," Japan and the World Economy, Vol. 11, 305-373.

Corsetti, G., M. Pericoli, and M. Sbracia, (2005), "Some Contagion, Some Interdependence: More Pitfalls in Tests of Financial Contagion," Journal of International Money and Finance, Vol. 24, 1177-1199.

Cowan K. and J. D. Gregorio (2005)," International Borrowing, Capital Controls, and the Exchange Rate: Lessons from Chile," NBER Working Papers 11382.

Dekle R. and Kletzer K. (2001), "Domestic Bank Regulation and Financial Crises: Theory and Empirical Evidence from East Asia," NBER Working Papers 8322.

Dornbusch R., Y. C. Park, and C. Stijn (2000), "Contagion: Understanding How it Spreads," World Bank Research Observer, Vol. 15, 177-97.

Eichengreen B. and P. Luengnaruemitchai (2004), "Why Doesn't Asia Have Bigger Bond Markets," NBER Working Papers 10576.

Fratzscher M. (1998), "Why are Currency Crisis Contagious? A Comparison of the Latin American Crisis of 1994-1995 and the Asian Crisis of 1997-1998," Weltwirtschaftliches Archiv, Vol. 134.

Forbes, K and R. Rigobon (2002), "No Contagion, only Interdependence: Measuring Stock Market Comovement," Journal of Finance 57 (5), 2223-2261.

Forbes K. and R. Rigobon (1999), "Measuring Contagion: Conceptual and Empirical Issues," mimeo, MIT.

Kaplan E. and D. Rodrik (2001), "Did Malaysian Capital Controls Work?" NBER Working Papers No 8142.

Khan, S., F. Islam, and S. Ahmed (2005), "The Asian Crisis: An Economic Analysis of the Causes," Journal of Developing Areas, Vol. 39, 1, 169-90.

King, M. and S. Wadhwani (1990), "Transmission of Volatility between Stock Markets," Review of Financial Studies, Vol. 3, 2, 5-33.

Lee, S. B. and K. J. Kim (1993), "Does the October 1987 Crash Strengthen The Co-Movements Among National Sock Markets?" Review of Financial Economies, Vol. 3, 89-102.

Masson P. (1998), "Contagion and Monsoonal Effects, Spillovers, and Jump between Multiple Equilibria" IMF Working Paper No. 98/142.

Pindyck, R. and J. Rotemberg (1990), "The Excess Co-Movement of Commodity Prices," Economic Journal, Vol. 100, 4.

Radelet, S. and J. Sachs (1998), "The Onset of East Asian Financial Crisis," NBER Working Papers 6680.

Sebastian E. (2001), "Exchange Rate Regimes, Capital Flows, and Crisis Prevention" NBER Working Papers 8529.

Valdes, R. (1997), "Emerging Markets Contagion: Evidence and Theory," Mimeo, MIT.

SALEHEEN KHAN

Minnesota State University, Mankato

FARIDUL ISLAM

Utah Valley University, Orem

KWANG WOO (KEN) PARK

Minnesota State University, Mankato

Notes

(1.) Dornbusch, Park, and Claessens (2000) define pure contagion as comovement that cannot be explained on the basis of fundamentals or global shocks.

(2.) See King & Wadhwani (1990), Lee and Kim (1993), Calvo and Reinhart (1996), and Baig and Goldfajn (1999).

(3.) Rigobon and Forbes (1999) and Dornbusch, et al. (2000) also defined contagion in similar manner.

(4.) For Tranquil period, the sample period refers to 1/1994-12/96 and for the crisis period it is 1/1997-1298.

(5.) All variables are in logs.

(7.) For more description on this issue, see Sebastian Edwards (2001).

Table 1
Residuals Correlation (Stock Returns) Tranquil Period: (1/94-12/96)

 Indonesia Korea Malaysia Philippines

Korea .225
Malaysia .677 .256
Philippines .473 .190 .493
Thailand .158 .286 -.098 -.093

(**) indicates statistical significance at the 1% level
Macroeconomic fundamentals and global shock is controlled for
LR Test statistic = 39.321 **

Table 2
Residuals Correlation (Stock Returns) Crisis period: (1/97-12/99)

 Indonesia Korea Malaysia Philippines

Korea .494 **
Malaysia .478 .691 **
Philippines .477 .711 ** .903 **
Thailand .618 ** .703 ** .634 ** .705 **

(**) indicates statistical significance at the 1% level
Macroeconomic fundamentals and global shock is controlled for
LR Test statistic = 119.059 **

Table 3
Residuals Correlation (Interest Rate) Tranquil Period: (1/94-12/96)

 Indonesia Korea Malaysia Philippines

Korea -.124
Malaysia .195 -.079
Philippines .207 -.018 .051
Thailand -.375 .268 .117 .208

(**) indicates statistical significance at the 1% level
LR Test statistic = 15.049

Table 4
Residuals Correlation (Interest Rate) Crisis period: 1/97-12/99)

 Indonesia Korea Malaysia Philippines

Korea .467 **
Malaysia .099 .489 **
Philippines .457 ** .793 ** .671 **
Thailand .458 ** .723 ** .614 ** .782 **

(**) indicates statistical significance at the 1% level LR Test
statistic = 96.763 **

Table 5
Residuals Correlation (Exchange Rate) Tranquil Period: (1/94-12/96)

 Indonesia Korea Malaysia Philippines

Korea -.043
Malaysia .201 .409
Philippines .263 .098 .253
Thailand .057 .207 .192 -.003

(**) indicates statistical significance at the 1% level LR Test
statistic = 13.38

Table 6
Residuals Correlation (Exchange Rate) Crisis period: (1/97-12/99)

 Indonesia Korea Malaysia Philippines

Korea .293 **
Malaysia .275 .669 **
Philippines -.143 .116 .444 **
Thailand .313 .698 ** .798 ** .560 **

(**) indicates statistical significance at the 1% level;
LR Test statistic = 84.036 **

Table 7
Residuals Correlation (Foreign Reserve) Tranquil Period: (1/94-12/96)

 Indonesia Korea Malaysia Philippines

Korea -.367
Malaysia -.167 -.073
Philippines .640 -.329 .130
Thailand -.224 .354 -.313 -.274

(**) indicates statistical significance at the 1% level;
LR Test statistic = 42.083 **

Table 8
Residuals Correlation (Exchange Rate) Crisis period: (1/97-12/99)

 Indonesia Korea Malaysia Philippines

Korea .398
Malaysia .478 ** .627 **
Philippines .670 .521 ** .532 **
Thailand .345 ** .267 .105 ** .355 **

(**) indicates statistical significance at the 1% level;
LR Test statistic = 55.1761 **