An Empirical Analysis Of The U.s. Dollar, Yen And ...



An Empirical Analysis of the U.S. Dollar, Yen and Eurodollar Exchange Shock Mean and Volatility Spillover to domestic and China Stock Markets

by

Ching-Chun Wei

Department of Finance, Providence University,

200 Chung-Chi Rd. Shalu, Taichung 43301, Taiwan

Phone: 886-4-26328001-Ext.13063

E-mail:ccw@pu.edu.tw

An Empirical Analysis of the U.S. Dollar, Yen and Eurodollar Exchange Shock Mean and Volatility Spillover to domestic and China Stock Markets

by

Ching-Chun Wei[1]

Department of Finance, Providence University, 200 Chung-Chi Rd. Shalu, Taichung 43301, Taiwan

Abstract

This paper used the Constant Conditional Correlation (CCC) and Dynamic Conditional Correlation (DCC) Multivariate EGARCH-M model to analyze the U.S. Dollar (USD), Japanese Yen, and Eurodollar to Renminbi (RMB) unexpected exchange rate shock mean and volatility spillover to the domestic and China stock markets. Under the CCC-MEGARCH model, we found that exchange rate markets have significant mean spillover to the China Shenzhen Composite index (SJC) stock market, and the cross-section volatility spillover effects and leverage effects are significant in the three exchange-rate models. Next, based on the DCC-MEGARCH-M model, the mean spillover effect has been found in the USD to RMB exchange rate model, the cross-section volatility spillover effects are significant in the USD and Yen to RMB exchange rate market, and the asymmetric effects and volatility persistence are significant in the USD and Eurodollar to RMB exchange rate market. We found that the USD to RMB exchange rate has more significant mean spillover effect, volatility cross-section effect, and asymmetric and volatility persistence effects than other markets.

Keywords: CCC and DCC MEGARCH-M; Volatility spillover; Unexpected exchange rate shock

JEL Classification: C12; F31; G15.

1. INTRODUCTION

Over half of China’s exports go to the United States, the European Union, and Japan (with over 30% going to the U.S. alone). If China’s rapidly rising current account surplus, huge accumulation of reserves, and limited appreciation of the renminbi (RMB) persuade legislators and policymakers in the major industrial countries that China is blocking effective balance-of-payments adjustment, China may well find its access to these markets constrained by new protectionist barriers. In response to international pressure over its policy for pegging RMB to the USD, the Chinese government on July 21, 2005 announced it would immediately appreciate the RMB to the dollar by 2.1% and adopted a currency policy based on a basket of currencies (including the USD, Yen, and Eurodollar). However, the 2.1% revaluation of the RMB has symbolic value, notably in the United States. It is indicative of the recognition that China now where responsibility for the stability of the global economy. A more flexible exchange rate should enable the People’s Bank of China to more effectively tailor monetary condition to local needs as it moves toward a more market-based financial system. There are two approaches that suggest the interrelationship between the exchange rates and stock prices of two countries. Based on the “asset approach” of Frankel (1983), the exchange rate adjust equate supply and demand for financial assets, and the expectation of financial asset price movements affect exchange rate dynamics. Suppose the stock market affects are large, change the impact of expansionary monetary policy on the exchange rate lead to an appreciation rather than depreciation. On the “goods market” model of Dornbusch and Fisher (1980), currency movements affect the international competitiveness of firms and the balance of trade, thereby influencing real income and output, and eventually affect current and future cash flows of companies and their stock prices.

Bodar and Reding (1999) examined the impact of Germany exchange rate fluctuations on the stock market volatility and the correlation between the Germany stock market and European markets. They found that there is no strong evidence that a higher exchange rate variability increase stock market volatility. Colavecchio and Funke (2006) used the multivariate GARCH model to study volatility spillover between the Chinese non-deliverable forward market and seven of its Asia-Pacific counterparts and examine co-movements between China and other Asian forward exchange rates and the volatility of Asian currencies is affected by RMB exchange rate developments. Mum (2007) examined the exchange rate fluctuations in international stock markets and to what extent volatility and correlations in equity markets are influenced by exchange rate fluctuations. Empirical evidence indicated that a higher foreign exchange rate variability increase local stock market volatility but decrease volatility for the U.S. stock market. Tai (2007) used an asymmetric MGARCH(1,1)-M model examined is whether there are pure contagion effects between stock and foreign exchange markets for Asian country during the 1997 Asian crisis. The empirical results show that strong positive impact of return shocks originating from the domestic stock market to its foreign exchange market during the crisis is found.

Some recent authors have used the GARCH and EGARCH model to investigate the dynamic relationship between stock return volatility and trading volume for individual stocks listed on the Chinese stock market (Wang, Wang, Liu (2005); Copeland and Zhang (2003); Lee and Rui (2000)). Extensive studies on conditional volatility of financial markets in the economics of Greater China have been conducted. These include the works of Friedmann and Sanddorf (2000) for stock index, as well as those of Yeh and Lee (2000), Ho and Tsui (2004), and Colavecchio and Funke (2006) for non-deliverable forward exchange rate. Friedmann and Sanddorf-Kohle (2000) apply both the EGARCH model and the GJR model to daily Chinese stock index returns. They found that the EGARCH can well be at least as robust in periods of high volatility than the GJR GARCH model. Kin Yip Ho and Tsui (2004) searched for evidence of conditional volatility in the quarterly real GDP of Greater China, which comprises the economies of Mainland China, Hong Kong, and Taiwan. They used the EGARCH model to capture the existence of asymmetric

conditional volatility in real GDP. Yeh and Lee (2000) investigated the response of investors to unexpected returns and the information transmission in the stock markets China. They analyzed the asymmetric reaction of return volatility to good and bad news by utilizing the GARCH model. They found that the impact of bad news of volatility is greater than the impact of good news of the same effect in Taiwan and Hong Kong, but not for China. Colavecchio and Funke (2006) used the multivariate GARCH model to study volatility spillover between the Chinese non-deliverable forward market and seven of its Asia-Pacific counterparts over the period January 1998 to May 2005. The objective of this study is to examine the nature of co-movements between China and other Asian forward exchange rates and the volatility of Asian currencies is affected by renminbi exchange rate developments.

The objectives of this paper by using the CCC and DCC-MEGARCH-M models that include the unexpected exchange rate shock from the exchange rate market into the mean equation are to capture the mean spillover effect from one market to another, and to capture the volatility spillover effects in the variance equation, while also estimating the asymmetric effect of exchange rate volatility to the stock market. The remainder of this paper proceeds as follows: Section 2 presents the Constant Conditional Correlation (CCC) and the Dynamic Conditional Correlation (DCC)-MEGARCH-M models. Section 3 starts the description of the data employed in this study and presents the empirical results. Section 4 presents the conclusion of the paper findings.

2. METHODOLOGY

In the financial literature, it has been suggested that the ARCH models are well suited to capturing exchange rate and stock return movements. However, there are different approaches to proxy exchange rate uncertainty. One is the use of variance of the exchange rate return (Goldberg and KolstadK, 1995). Another is the estimated standard GARCH model to obtain a conditional measure of volatility. GARCH methods have been used to derive measures of uncertainty (Engle, 1982). Here, we use the GARCH method error term to be the proxy for measuring exchange rate uncertainty or shocks. Kutmos (1995) defines “price spillover” as the impact of an innovation from market i on the conditional mean of market j, whereas “volatility spillover” is the impact of an innovation from market i on the conditional variance of market j. Here, we put the exchange rate variable into the mean equation to examine the impact of unexpected exchange rate shock to the stock markets. The advantage of the MEGARCH specifications is that they permit time-varying conditional covariance as well as variances; thus, they allow for possible interactions within the conditional mean and variance of two or more financial series.

Our study use a Constant Conditional Correlation (CCC) and Dynamic Conditional Correlation(DCC) form of the multivariate EGARCH model to investigate market interdependence and volatility transmission between unexpected exchange rate market to stock markets in US, Japan and European to China. We estimate the MEGARCH model suggested by Bollerslev (1990) is used as a framework to take account of asymmetric volatility spillovers and the standard multivariate EGARCH model that assumes that the underlying correlations between shocks are constant over time. This constant correlation specification has generally a well-behaved likelihood function as well as limiting the number of estimation of coefficients to a workable level. However, the dynamic conditional correlation model which allow these correlations to be a time-varying.

Following Koutmos and Booth(1995) and Antoniou et al. (2003), we specifically the multivariate EGARCH-M model as follows:

[pic] (1)

[pic] (2)

[pic] (3)

[pic] (4)

From the mean equation, the dynamic relationships in returns are captured by using a Vector Autoregressive (VAR) model, where the conditional mean in each market (Ri,t) is a function of own past returns and cross-market past returns(Ri,t). However, βi,j, captures the lead-lag relationship between returns in different markets, for [pic].The coefficients of [pic] measures the unexpected exchange rate mean spillover on form three unexpected exchange rate markets to the local and china stock markets. Gj is the unexpected exchange rate shock from market j. The coefficients of [pic] measures the three unexpected exchange rate price spillover to the local and china stock markets. The significantβi,j value, implying market j leads market i. Here, the first-order VAR is adopted because we think the stock market will quickly respond to information from other markets.

The conditional variance equation describes the conditional variance in each market as an exponential function of past standardized innovation,([pic]), that is from its own market and other markets. The estimated value of [pic] that measure the persistence of volatility. If [pic] =1, then the unconditional variance doesn’t exist and the conditional variance follows an integrated process of order one. Spillovers are captured by the coefficients[pic], while asymmetry implies negative [pic] The asymmetric influence of innovation on the conditional variance is captured by the term ([pic]). Here, a statistically significant positive rj together with a negative(positive) [pic] show that negative innovations in market j have a greater impact on the volatility of market i than positive(negative) innovations. The relative importance of the asymmetry(or leverage effect) can be measure by the ratio [pic].

The term [pic] measures the size effect. Assuming positive [pic] the impact of [pic]on [pic] will be positive(negative) if the magnitude of [pic] is greater(smaller) than its expected value [pic]. Assuming the conditional joint distribution of the returns of the three markets is normal, the log likelihood for the multivariate EGARCH model can be written as:

[pic] (5)

where N is the number of equation, T is the number of observations, θis the [pic] parameter vector to be estimated [pic] is the [pic] vector of innovations at time t, St is the [pic] time varying conditional variance-covariance matrix with diagonal elements given and cross diagonal elements are given. The log-likelihood function is highly non-linear inθ,and therefore, numerical maximization techniques have to be used. The disturbance error term of mean equation are assumed to be conditional multivariate normal with mean zero and conditional covariance matrix[pic]:

[pic] (6)

and

[pic] or [pic] (7)

Under the conditional covariance matrix, Dt is a n time n diagonal matrix with the time-varying standard deviations of (3) on the diagonal and Di is a time-varying symmetric correlation matrix:

[pic] , [pic] (8)

The dynamic correlations are captured through the asymmetric general diagonal DCC equation:

[pic] (9)

The [pic] and [pic] are the conditional correlation matrices of Zt and[pic] .Above model is a generalization of the DCC model of Engle(2002) to capture asymmetric correlations and first used by Capiello et al. (2003). The matrix of A, B and C are restricted to being diagonal for practicality in the estimation of model.[pic]t is positive definite with probability one if ([pic] is positive definite. Through the final term of (7), the time-varying correlations will respond asymmetrically to positive or negative shocks in each market. Next, we scale [pic] to get the correlation matrix qt:

[pic] (10)

The multivariate EGARCH allow us to examine both volatility spillover between markets and asymmetry. However, it is not useful to apply such an EGARCH specification to the conditional correlations, both because it would unduly restrict the conditional correlations to be always positive and because it has too many parameters. The DCC specification of (6) does not have these problems, but nevertheless allows the possibility of asymmetric effects.

The model is estimated by maximum likelihood. As [pic] is normally distributed, the log-likelihood can be express as:

[pic] (11)

where n is the number of equations, T is the number of observations, [pic] is the parameter vector to be estimated, [pic], is the vector of innovations at time t, Ht, is the time varying conditional variance-covariance matrix with diagonal elements give by equation (2) and cross diagonal elements are give by equation (4).

3. DATA SUMMARY AND EMPIRICAL RESULTS

3.1 Data Summary

In order to investigate the price and volatility spillover of unexpected exchange rate shock in the China stock markets, the data consist of the daily data of five stock market indexes and three countries’ exchange rates from the Taiwan Economic Journal (TEJ) data bank. These are the New York Dow-Jones index (NYD), Japan Nikkei 225 index (J225), European Amsterdam AEX index (AEX), China Shanghai Composite index (SHC), Shenzhen Composite index (SJC), and the exchange rates for the USD(CAR), Japanese Yen (CJR) and Eurodollar (CUR) to Chinese RMB from July 21, 2005 to January 4, 2007. Here, the variables of returns in each market are calculated as the first difference in the natural logarithms of the stock market indexes.

Table 1 presents the descriptive statistics for each variable series. The skewness statistics suggest lack of normality in the distribution of the return series. The CAR, NYD, AEX, J225, SHC and SJC have return distributions that are negatively skewed, while those of CJR and CUR are positively skewed. The value of kurtosis indicate that each of the return series is leptokurtosis. The Jarque-Bera (JB) normality test rejects the null hypothesis of normality. The significant value of the Ljung-Box Q statistics for return series rejects the null hypothesis of white noise and indicates the presence of autocorrelation. The ARCH-LM for all variables is significant at 1% level, indicating the existence of ARCH phenomenon for all variable series. Here, the stationarity of the series was investigated by employing the unit root tests developed by augmented Dickey Fuller (ADF) (Dickey and Fuller, 1981), and Phillips and Perron (1988) tests. The calculated ADF statistics are less than the critical value only for the first difference variables; therefore, it could be concluded that all variables have difference stationarity.

3.2.1 Constant Conditional Correlation (CCC)-MEGARCH (1, 1)-M Model

3.1.1 Mean and volatility spillover- from U.S. to China

The maximum likelihood estimates of the multivariate model with no parameter restrictions are reported in Table 2. The full model considers both price and volatility spillovers from the last two markets to close to the next market to trade. Table 2 shows that there are significant unexpected exchange rate shock price spillovers to the China SJC stock market (-0.3956). The value of feedback effects (0.1321 and -0.3771) have been found significant between two China stock markets. The effect from the New York stock market to the China stock markets is a significant spillover. However, we find no significant effects from the China stock markets to the New York stock market (-0.3288 and -0.3689). Table 2 also presents second moment interactions or volatility spillover between the domestic stock market and the two China stock markets. The impacts of the past own and cross-market innovations on current market volatility are investigated through parameter estimates. Table 2 reports these parameter estimates, as well as the parameters measuring the asymmetric volatility spillover effect,[pic]. The estimate parameters of the impact of past own innovation on current volatility ([pic] ) are all positive and significantly different from zero for all three markets. These results indicate the presence of significant own volatility spillover in these markets. The estimate parameters of the cross-market innovation spillover parameters,[pic], indicate significant impact of past volatility shocks from the U.S. stock market to the two China stock markets. Here, we also find that these are feedback impacts of volatility innovation from those markets.

Within China, we find that feedback volatility spillover occurs in two China stock markets with the SHC more significant than the SJC markets. Later, we compare these volatility spillover effects; all markets have significant feedback spillover, and the impact spillover of the U.S. to the China stock markets has more significance than that of China to the U.S. Here, the leadership roles within the mean and volatility equation should be noted, with U.S. as the key market for those markets. More importantly, Table 2 also shows the volatility transmission mechanism to be asymmetric for all markets. That is, the coefficients measuring asymmetric, namely, rj, are significant for the three markets. This result reinforces our assertion that bad news increases volatility more than good news does. Except for the SHC market, the estimates of the asymmetry parameter are negative and significantly different from zero, suggesting that negative past innovations in these stock markets increase volatility more than positive innovations do. The persistence of volatility is measured by ri, Table 2 reports that those values are less than one for all markets. However, these estimates are highly significant, indicating high volatility persistence in U.S. and two China stock markets.

The residual-based diagnostics are reported in Table 2. The LB and LB2 statistics show no serious linear or nonlinear dependencies for the standardized residuals for the three markets. Specifically, the standardized residual results show no evidence of autocorrelation, which means that such effect was successfully captured by our model. From the ARCH-LM test, we can conclude that both linear and nonlinear dependencies in the return series have been effectively filtered.

However, skewness and kurtosis coefficients, as well as J-B statistics, indicate various departures from normality. Thus the conditional tri-variate normality assumption may be violated. The validity of the assumption of constant conditional correlation can be assessed by testing for serial correlation in the cross-product of the standardized residuals. The Ljung-Box statistics for eight lags are reported in Table 2, which show no evidence of serial correlation, so that constant correlation specification appears to be a reasonable parameterization of the variance–covariance structure of the three markets.

3.1.2. Mean and Volatility Spillover- from Japan to China

At the Yen-RMB exchange rate in the Japan and China stock markets, there are significant unexpected exchange rate price spillover effects to the China stock market (-0.1158 and -0.1586) as shown in Table 3. At the Chinese stock markets, we have found negative and significant feedback price spillover effect (-0.1477 and -0.1259). However, the unexpected exchange rate shock is negative, and the significant spillover to the two China stock markets indicates that negative exchange rate shock reduces the stock market returns. The variance equation of this table indicates the volatility spillover between stock markets. The estimate parameters of the[pic],[pic] and [pic] (0.2658, 0.1206 and 0.2050) are all positive and significantly different from zero for all three markets, indicating the presence of significant own volatility spillover in those markets. The cross-market innovation spillover parameter[pic] indicates significant impact of volatility from the Japan stock market to the SJC stock market (-0.0291), from the SHC stock market to the Japan and SJC stock markets (-0.0516 and 0.1607), and from the SJC market to the Japan and SHC stock markets (-0.0675 and 0.0404). Also, there are significant feedback volatility spillover effects in the two China stock markets. Table 4 reports the volatility transmission mechanism to be asymmetric for all markets, and the coefficients (-0.3652, 0.0941, and -0.0687) are all significant for all markets. The Ljung-Box statistics show no serious evidence of linear and non-linear dependence in the standardized residual; the ARCH-LM test shows that no ARCH effect exists; skewness, kurtosis, and J-B statistics show that normality does not exist. The Ljung-Box Q statistics tests for the cross-product of standardized residuals show no evidence of serial correlation.

3.1.3 Mean and Volatility Spillover-from Europe to China

The maximum likelihood estimates of the Eurodollar–RMB exchange rate to the China stock market of the MEGARCH(1,1)-M model are reported in Table 4. Focusing on the parameters in the conditional mean equation, we can see that current returns in the European markets are not influenced by past returns in China’s stock markets. Returns in China are not correlated with past returns in the European markets. However, we did not find a significant feedback price spillover effect between the European and the China stock markets. However, the Eurodollar–RMB unexpected exchange rate has a negative price spillover effects on the two China stock markets. Table 4 also presents the estimation result of the measurement coefficients of the volatility interactions. The estimate parameters of [pic], and [pic](0.1791,-8.436e-03 and0.0240) are all significantly different from zero at the European and the China stock markets. It shows the presence of a significant own volatility spillover in those markets. The cross-market innovation spillover effect shows the significant impact of volatility from the SJC stock market to the European market (0.0577), and the feedback effect between the two China stock markets. The reporting parameters of the volatility transmission mechanism to be asymmetric are significant for all markets (-0.2273, 0.0469, and -0.0281). Finally, the parameters of persistence volatility (0.8551, 0.9321, and 0.5916) show high volatility persistence in the three markets. Table 4 shows that Ljung-Box Q statistics has no serious evidence of serial autocorrelation, except for the SJC stock market. The estimated coefficients of skewness, kurtosis, and J-B statistics imply a violation of the normality assumption. The Ljung-Box Q statistics test for the validity of the assumption of constant conditional correlation displays no evidence of serial correlation, so that constant correlation specification will be suitable for those markets.

3.2. The Dynamic Conditional Correlation (DCC)-MEGARCH (1, 1)-M Model

3.2.1 Mean and Volatility Spillover- from U.S. to China

Table 5 presents the estimation results for asymmetric DCC-MEGARCH(1,1)-M model. In terms of the first moment interdependences, the coefficient of [pic] indicate that there are negative significantly unexpected price spillover effects of unexpected exchange rate shock to the China stock markets. Turning to the volatility spillover equation, it is observed that the estimated parameters[pic] are positive and significant at 1% level. The coefficients value of ri measures the leverage effect to be less than one and negatively significant at SHC stock markets. The residual-based diagnostic tests for the model of Ljung-Box Q statistics show no serious evidence of linear and non-linear dependence in the standardized residuals. The ARCH-LM test represents that no ARCH effect exist.

3.2.2 Mean and Volatility Spillover-from Japan to China

The estimated results reported in Table 6 indicate that the price spillover effects between those stock markets are all insignificant. This table also reports the estimation result of the volatility spillover between stock markets. The estimated parameters of [pic] are all positive and significant at least at the 10% level, indicating the presence of significant volatility spillover in those markets. The estimated coefficients of persistence volatility ri are only significant at the Japan stock markets.

3.2.3 Mean and Volatility Spillover-from Europe to China

Finally, Table 7 displays the estimation result of the European to the China stock market. The empirical results show a significant price spillover effect only at the European stock market but not for the other two China stock markets. The estimated parameters of [pic] are also significant only at the European stock market. The parameters of volatility persistence show a significant volatility persistence at the European and SHC stock markets. In this model, we did not find any significant price and volatility spillover effect among those markets.

4. CONCLUSION

Through the CCC and DCC-MEGARCH(1,1)-M models, this paper includes the unexpected exchange rate shock of the exchange market in the mean equation to capture the price spillover effect from one market to another. In the CCC-MEGARCH-M model, we found that the three exchange rate markets have significant price spillover to the China SJC stock market. The effects of the cross-section volatility spillover effect have been found to be significant at the USD, Yen, and Eurodollar to RMB exchange rates. The measurement of leverage effects is also significant in all three exchange-rate volatility spillover models. Furthermore, the persistence of volatility measurement is significant which show the volatility persistent.

Next, based on the DCC-MEGARCH(1,1)-M model, the price spillover effect has been found in the USD–RMB exchange rate model, but not in the other two exchange rate models. The cross-section volatility spillover effects are also significant at the USD and Yen to RMB exchange rate markets except for the Eurodollar exchange rate market. The asymmetric effects and volatility persistence are significant at the USD and Eurodollar to RMB exchange rate market. In summary, it is obvious that the USD to RMB exchange rate has more significant price spillover effect, volatility cross-section effect, asymmetric, and volatility persistence effect than others. In this study, we found that USD and RMB exchange rates play a significant role in the China stock market. In particular, U.S. and China are more integrated with each other. Compared with the U.S stock market, however, the yen and Eurodollar to RMB exchange rates have weakly significant effect on the China stock prices

References

Antoniou, A., Pescetto, G., and Violaris, A. (2003). Modeling International Price Relationships and Interdependencies Between the Stock Index Future Markets of Three EU Countries: A Multivariate Analysis, Journal of Business & Accounting, 30, 645-667.

Bollerslev, T. (1990). Modeling the Coherence in the Short-Run Nominal Exchange Rates, Review of Economics and Statistics, 72, 498-505.

Cappiello,L., Robert, E., and Kevin,S. (2003). Asymmetric Dynamics in the Correlation of Global Equity and Bond Returns, European Central Bank, Working Paper Series, 245, 1-65.

Colavecchio,C., and Michael F. (2006). Volatility Transmission between Renminbi and Asia-Pacific on shore and off-shore U.S. Dollar Futures, Bank of Finland, BoFIT Discussion Paper.

Copeland, L., and Zhang B. (2003). Volatility and Volume in Chinese Stock Markets, Journal of Chinese Economic and Business Studies, 1, 287-300.

Dornbusch., R. and Fisher., S.(1980).Exchange Rates and the Current Account, American Economic Review, 70, 960-71.

Frankerl, J.A. (1983). Monetary and Portfolio-Balance Models of Exchange Rate Determination, in J.S. Bhandari and B.H. Putman (eds.), Economic Interdependence and Flexible Exchange Rates. Cambridge, MA.: MIT press.

Friedmann,R., and Sanddorf-Kohle. (2002). Volatility Clustering and Nontrading Days in Chinese Stock Markets, Journal of Economics and Business, 54,193-217.

Goldberg,L., and Kolstand, C.(1995).Foreign Direct Investment, Exchange Rate Volatility and Demand Uncertainty, International Economic Review, 36, 855-873.

Ho, K-Y., and Tsui, K.C. (2004). Volatility Dynamic of the Tokyo Stock Exchange: A Sectoral Analysis based on the Multivariate GARCH Approach, Money Macro and Finance (MMF) Research Group Conference.

Koutmos, G., and Booth,G..(1995). Asymmetric Volatility Transmissions in International Stock Markets, Journal of International Money and Finance, 14, 747-762.

Lee,C.F., and Rui, O.M. (2000). Does Trading Volume Contain Information to predict Stock Returns? Evidence from China’s Stock Markets, Review of Quantitative Finance and Accounting, 14, 341-60.

Yeh, Y-H and Lee, T-S. (2000). The Interaction and Volatility Asymmetry of Unexpected Returns in the Greater China Stock Markets, Global Finance Journal, 11, 129-149.

Wang, P., Wang P., and Liu, A. (2005).Stock Return Volatility and Trading Volume: Evidence from the Chinese Stock Market, Journal of Chinese Economic and Business Studies, 3, 39-54.

|Table 1. Descriptive Statistics |

| |CAR |CJR |CUR |NYD |J225 |AEX |SHC |SJC |

|Mean |0.1091 |14.7528 |0.1002 |11297.71 |15614.20 |5811.692 |1664.964 |407.9452 |

|Mediam |0.1087 |14.7486 |0.0992 |11124.37 |15960.62 |5850.800 |1589.540 |396.7700 |

|Maximum |0.1123 |16.2095 |0.1069 |12786.64 |18215.35 |6444.400 |3197.540 |841.4800 |

|Minmum |0.1068 |13.5931 |0.0949 |10215.22 |11695.05 |5142.100 |1020.630 |240.3700 |

|Std.Dev. |0.0015 |0.5788 |0.0030 |686.7813 |1548.216 |325.1345 |580.9623 |144.1937 |

|Skewness |0.5959 |0.3340 |0.3071 |0.6051 |-0.8980 |-0.1500 |1.12492 |1.2414 |

|Kurtosis |2.2923 |2.9739 |1.8562 |2.2302 |3.0088 |2.0123 |3.1669 |3.8527 |

|Jarque-Bera |32.7460** |9.2088** |28.723** |35.0626** |54.9815** |18.1556** |86.7373** |117.4557** |

|ARCH-LM(8) |94.286*** |21.544*** |11.844*** |2.366*** |3.969*** |8.1103*** |7.368*** |12.016*** |

|LB (8) |113.13*** |22.22*** |35.252*** |25.74** |116.12* |28.16** |24.85** |40.766** |

|[pic] |191.29*** |51.57*** |43.273*** |27.98** |107.63* |197.96* |96.396* |158.66* |

Note：1.The significant value of the LB-Q statistics for the squared returns suggests the presence of autocorrelation in the square of stock returns. ARCH-LM statistics proposed by Engle (1982) aimed to detect ARCH. In fact the value of ARCH-LM(8) are all significant at 1% level, indicating the existence of ARCH phenomena for all variable series.

2. *** indicated at least significant at 1% level. ** indicated at least significant at 5% level. *indicated at least significant at 10% level.

|Table 2. The CCC-MEGARCH(1.1)-M Model for U.S.D-RMB unexpected exchange rate mean and volatility spillovers to U.S. and china stock |

|markets |

|Mean Equation | | | | | |

| |DLNYD | |DLSHC | |DLSJC |

|[pic] |4.2284e-04*** |[pic] |1.8319e-03*** |[pic] |1.8087e-03* |

| |(2.2088e-04) | |(4.3529e-04) | |(4.8847e-04) |

|[pic] |0.0488 |[pic] |-0.3288*** |[pic] |-0.3689*** |

| |(0.05) | |(0.05) | |(0.06) |

|[pic] |-0.0194 |[pic] |-0.1148*** |[pic] |-0.3771*** |

| |(0.02) | |(0.03) | |(0.04) |

|[pic] |8.7562e-03 |[pic] |0.1321 |[pic] |0.3916*** |

| |(0.02) | |(0.02) | |(0.03) |

|[pic] |0.0704 |[pic] |-0.3463*** |[pic] |-0.3956*** |

| |(0.09) | |(0.17) | |(0.17) |

|Variance Equation | | | | |

|[pic] |9.5419e-07*** |[pic] |1.3232e-04*** |[pic] |2.3393e-04*** |

| |(1.6861e-07) | |(7.1824e-06) | |(2.1378e-06) |

|[pic] |0.0663*** |[pic] |0.0896*** |[pic] |-0.0196*** |

| |(0.01) | |(0.01) | |(7.3792e-03) |

|[pic] |-0.0232** |[pic] |0.0959*** |[pic] |0.0894*** |

| |(0.01) | |(3.1042e-03) | |(3.9394e-03) |

|[pic] |-0.0703*** |[pic] |-0.0357*** |[pic] |0.0789* |

| |(9.334e-03) | |(0.02) | |(4.8602e-03) |

|[pic] |-0.1590*** |[pic] |0.1624*** |[pic] |-6.5727e-03*** |

| |(0.02) | |(0.01) | |(3.5835e-03) |

|[pic] |0.9878*** |[pic] |-0.9950*** |[pic] |-0.8074*** |

| |(6.3258e-03) | |(5.9603e-03) | |(0.05) |

|Diagnostic test checking | | |

| |DLNYD |DLSHC |DLSJC |

|L-BQ(8) |18.2433 |25.8624 |12.2540 |

|[pic] |28.2762 |50.2753* |33.4546* |

|ARCH-LM(8) |4.1007 |8.6746 |8.1212 |

|Skewness |-0.4305* |-1.2746* |-1.2718* |

|Kurtosis |2.3295* |5.5509* |4.6204* |

|Jarque-Bera |124.9087* |721.3709* |537.8280* |

Note: 1. DLNYD, DLSHC and DLSJC are the difference log of the New York Dow-Jones Stock index, China Shanghai Composite Stock index and Shenzhen Composite Stock index, respectively. 2. *** indicated at least significant at 1% level, ** indicated at least significant at 5% level, * indicated at least significant at 10% level.

|Table 3. The CCC-MEGARCH(1.1)-M Model for Yen-RMB unexpected exchange rate mean and volatility spillovers to Japan and China stock |

|markets |

|Mean Equation | | | | |

|Coefficient |DLJ225 | |DLSHC | |DLSJC |

|[pic] |1.144e-03** |[pic] |1.4519e-03*** |[pic] |1.5948e-03*** |

| |(4.6558e-04) | |(2.8262e-04) | |(3.4168e-04) |

|[pic] |0.0356 |[pic] |0.038*** |[pic] |-4.6720e-03 |

| |(0.0366) | |(0.02) | |(0.02) |

|[pic] |0.0547** |[pic] |-1.4658e-03 |[pic] |-0.1259*** |

| |(0.03) | |(0.027) | |(0.02) |

|[pic] |-0.1154*** |[pic] |-0.1477*** |[pic] |0.0428** |

| |(0.03) | |(0.01) | |(0.02) |

|[pic] |-0.0723 |[pic] |-0.1158*** |[pic] |-0.1586*** |

| |(0.07) | |(0.03) | |(0.04). |

| | | | | |

|Variance equation | | | | |

|[pic] |2.1707e-05 |[pic] |1.1939e-04*** |[pic] |2.3817e-04*** |

| |(3.3601e-06) | |(1.7323e-06) | |(1.8234e-07) |

|[pic] |0.2658*** |[pic] |-1.7070e-03 |[pic] |-0.0291*** |

| |(0.07) | |(6.4842e-03) | |(1.7292e-03) |

|[pic] |-0.0516*** |[pic] |0.1206*** |[pic] |0.1607*** |

| |(0.01) | |(4.86e-04) | |(2.5516e-05) |

|[pic] |-0.0675*** |[pic] |0.0404*** |[pic] |0.2050*** |

| |(0.01) | |(2.2939e-03) | |(5.6536e-04) |

|[pic] |-0.3652*** |[pic] |0.0941*** |[pic] |-0.0687*** |

| |(0.07) | |(0.02) | |(0.02) |

|[pic] |0.7449*** |[pic] |0.4649*** |[pic] |-0.8023*** |

| |(0.04) | |(5.8158e-03) | |(0.03) |

|[pic] |5.6995 | |13.0672 | |11.7211 |

| [pic] |14.1878 | |15.5516* | |13.5835 |

|ARCH-LM |0.4168 | |2.31609 | |3.3284 |

|Skewness |-0.29742*** | |-1.27467* | |-1.2718* |

|Kurtosis |0.70088** | |5.55096* | |4.62048* |

|Jarque-Bera |16.37364* | |721.37099* | |537.828* |

Note: 1. DLJ225, DLSHC and DLSJC are the difference log of the Japan Nikkei Stock index, China Shanghai Composite Stock index and Shenzhen Composite Stock index, respectively.

2. *** indicated at least significant at 1% level, ** indicated at least significant at 5% level, * indicated at least significant at 10% level.

3.[pic] and [pic]is the Ljung-Box Q statistic value for return and squared return

autocorrelation at lag period eight.

|Table 4. The CCC-MEGARCH(1.1)-M Model for Eurodollar-RMB unexpected exchange rate mean and volatility spillovers to Europe and China |

|stock markets |

|Mean Equation： | | | | |

| |DLAEX | |SHC | |SJC |

|[pic] |4.0483e-04 |[pic] |3.2085e-03*** |[pic] |3.1469e-03*** |

| |(3.0687e-04) | |(6.2143e-04) | | |

|[pic] |-0.0833** |[pic] |-0.0412 |[pic] |-6.3177e-03 |

| |(0.05) | |(0.09) | |(0.10) |

|[pic] |-0.0296 |[pic] |0.0558 |[pic] |-0.0715 |

| |(0.04) | |(0.09) | |(0.10) |

|[pic] |0.0233 |[pic] |-0.0665 |[pic] |0.0964 |

| |(0.04) | |(0.09) | |(0.10) |

|[pic] |0.0274 |[pic] |-0.2177* |[pic] |-0.2557*** |

| |(0.05) | |(0.08) | |(0.09) |

| | | | | |

|Variance Equation | | | | |

|[pic] |3.7517e-06*** |[pic] |1.4813e-05*** |[pic] |1.2004e-04** |

| |(1.2474e-06) | |(7.5191e-07) | |(2.4851e-05) |

|[pic] |0.1791*** |[pic] |0.0343 |[pic] |0.0583** |

| |(0.05) | |(0.03) | |(0.03) |

|[pic] |0.0341 |[pic] |-8.436e-03*** |[pic] |7.1375e-03** |

| |(0.03) | |(1.566e-02) | |(3.9240e-03) |

|[pic] |0.0577** |[pic] |6.383e-03*** |[pic] |0.0240*** |

| |(0.03) | |(3.7196e-03) | |(7.7002e-03) |

|[pic] |-0.2273*** |[pic] |0.0469*** |[pic] |-0.0281*** |

| |(0.05) | |(6.0305e-03) | |(0.01) |

|[pic] |0.8551*** |[pic] |0.9321*** |[pic] |0.5916*** |

| |(0.05) | |(6.0954e-03) | |(0.07) |

|L-BQ(8) |16.2452 | |19.1684 | |10.7551 |

|[pic] |37.8126 | |19.2844 | |15.1049*** |

|ARCH-LM(8) |24.975* | |3.1455 | |4.1420 |

|Skewness |-0.4090* | |-1.2747* | |-1.2718* |

|Kurtosis |1.6942 | |5.5510* | |4.6205* |

|Jarque-Bera |71.5266* | |721.3710* | |537.8289* |

Note: 1. DLAEX, DLSHC and DLSJC are the difference log of the Europe Amsterdam Aex Stock index, China Shanghai Composite Stock index and Shenzhen Composite Stock index, respectively. 2. *** indicated at least significant at 1% level, ** indicated at least significant at 5% level, * indicated at least significant at 10% level. 3. [pic] and [pic]is the Ljung-Box Q statistic value for return and squared return autocorrelation at lag period eight.

|Table 5. The DDC-MEGARCH(1.1)-M Model for U.S.D-RMB unexpected exchange rate mean and volatility spillovers to U.S. and China stock |

|markets |

|Mean Equation： |DLNYD |DLSHC |DLSJC |

|Coefficient | | | |

|[pic] |4.4154e-04* |4.2247e-03*** |4.8600e-03*** |

| |(2.5308e-04) |(4.2462e-04) |(4.3388e-04) |

|[pic] |-0.0360 |0.0549*** |0.1082*** |

| |(0.04) |(0.01) |(0.02) |

|[pic] |-0.0333*** |-0.2370*** |-0.2174*** |

| |(0.04) |(0.06) |(0.06) |

|Variance equation | | | |

|[pic] |3.1890e-05*** |1.7000e-04*** |1.7648e-04*** |

| |(1.4861e-06) |(1.2561e-05) |(1.7555e-05) |

|[pic] |0.0483*** |0.1788*** |0.1009*** |

| |(1.84e-03) |(0.02) |(0.01) |

|[pic] |0.1602*** |0.1662*** |0.2831*** |

| |(0.04) |(0.05) |(0.06) |

|[pic] |-0.1759*** |-0.0664* |9.9148e-03 |

| |(3.3917e-03) |(0.04) |(0.04) |

|Diagnostic test checking |

|L-BQ(8) |7.7199 | | |

|[pic] |11.9951 | | |

| ARCH-LM(8) |0.7783 | | |

| Log-likelihood |4683.25 | | |

Note: 1. DLNYD, DLSHC and DLSJC are the difference log of the New York Dow-Jones Stock index, China Shanghai Composite Stock index and Shenzhen Composite Stock index, respectively.

2. *** indicated at least significant at 1% level, ** indicated at least significant at 5% level, * indicated at least significant at 10% level.

3.[pic] and [pic]is the Ljung-Box Q statistic value for return and squared return

autocorrelation at lag period eight.

|Table 6. The DDC-MEGARCH(1.1)-M Model for Yen-RMB unexpected exchange rate mean and volatility spillovers to Japan and China stock |

|markets |

|Mean Equation： |DLJ225 |DLSHC |DLSJC |

|Coefficient | | | |

|[pic] |1.0627e-03 |2.1424e-03 |2.9674e-03*** |

| |(5.8503e-04) |(1.5454e-03) |(6.6629e-04) |

|[pic] |-0.0103 |5.9904e-03 |0.0620 |

| |(0.10) |(0.02) |(0.04) |

|[pic] |-0.0143 |-0.0111 |-0.0433 |

| |(0.11) |(0.03) |(0.04) |

|Variance equation | | | |

|[pic] |1.0617e-04*** |2.1187e-04*** |2.2556e-04*** |

| |(8.5108e-06) |(2.5924e-05) |(1.1984e-05) |

|[pic] |0.1885** |0.1246*** |0.0966*** |

| |(0.10) |(0.04) |(0.02) |

|[pic] |0.1242 |-0.0120 |-0.2443*** |

| |(0.14) |(0.04) |(0.13) |

|[pic] |-0.2443** |0.550 |0.0343 |

| |(0.13) |(0.07) |(0.03) |

|Diagnostic test checking |

|L-BQ(8) |2.006 | | |

|[pic] |4.520 | | |

|ARCH-LM(8) |0.073639 | | |

|Log-likelihood |4410.0327 | | |

Note: 1. DLJ225, DLSHC and DLSJC are the difference log of the Japan Nikkei Stock index, China Shanghai Composite Stock index and Shenzhen Composite Stock index, respectively.

2. *** indicated at least significant at 1% level, ** indicated at least significant at 5% level, * indicated at least significant at 10% level.

3.[pic] and [pic]is the Ljung-Box Q statistic value for return and squared return

autocorrelation at lag period eight.

|Table 7. The DDC-MEGARCH(1.1)-M Model for Eurodollar-RMB unexpected exchange rate mean and volatility spillovers to Europe and China |

|stock markets |

|Mean Equation： |DLAEX |DLSHC |DLSJC |

|Coefficient | | | |

|[pic] |3.0794e-04*** |2.6471e-03*** |2.6228e-03*** |

| |(1.1558e-04) |(3.8402e-04) |(4.7437e-04) |

|[pic] |0.4632*** |0.0623*** |0.1567*** |

| |(0.06) |(0.02) |(0.03) |

|[pic] |-0.5550*** |8.7791e-03 |0.0803 |

| |(0.08) |(0.06) |(0.05) |

|Variance equation | | | |

|[pic] |6.1826e-05 |1.9205e-05 |4.8109e-05 |

| |(2.549e-06) |(1.4964e-05) |(3.0801e-05) |

|[pic] |0.2327*** |0.0796 |0.1159 |

| |(0.04) |(0.05) |(0.08) |

|[pic] |0.7558*** |0.7862*** |0.6423*** |

| |(0.07) |(0.09) |(0.17) |

|[pic] |-0.2724*** |0.1262* |0.1348 |

| |(0.05) |(0.07) |(0.08) |

|Diagnostic test checking |

|L-BQ(8) |3.137 | | |

|[pic] |1.237 | | |

|ARCH-LM |3.089 | | |

|Log-like |4692.1838 | | |

Note: 1. DLAEX, DLSHC and DLSJC are the difference log of the Europe Amsterdam Aex Stock index, China Shanghai Composite Stock index and Shenzhen Composite Stock index, respectively.

2. *** indicated at least significant at 1% level, ** indicated at least significant at 5% level, * indicated at least

significant at 10% level.

3.[pic] and [pic]is the Ljung-Box Q statistic value for return and squared return autocorrelation at lag period eight.

-----------------------

[1] È[2]È[3]È[4] È[5]'È[6].È[7]dXXXX

$[pic]$[pic]If[pic]a$[pic]gd  ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download