Chinese Stock and Bond Market Analysis

8/2/2019 Chinese Stock and Bond Market Analysis

1/13

How do policy and information shocks impact co-movementsof Chinas T-bond and stock markets?

Xiao-Ming Li *, Li-Ping Zou

Department of Commerce, Massey University (Albany), Private Bag 102 904, North Shore MSC, Auckland, New Zealand

Received 9 November 2006; accepted 13 April 2007Available online 16 July 2007

Abstract

We investigate the impacts of policy and information shocks on the correlation of Chinas T-bond and stock returns, using originallythe asymmetric dynamic conditional correlation (DCC) model that allows for the coexistence of opposite-signed asymmetries. The co-movements of Chinas capital markets react to large macroeconomic policy shocks as evidenced by structural breaks in the correlationfollowing the drastic 2004 macroeconomic austerity. We show that the T-bond market and the bondstock correlations bear more of thebrunt of the macroeconomic contractions. We also find that the bondstock correlations respond more strongly to joint negative than

joint positive shocks, implying that investors tend to move both the T-bond and stock prices in the same direction when the two assetclasses have been hit concurrently by bad news, but tend to shift funds from one asset class to the other when hit concurrently by goodnews. However, the stockstock correlation is found to increase for joint positive shocks, indicating that investors tend to herd more for

joint bullish than joint bearish stock markets in Shanghai and Shenzhen. 2007 Elsevier B.V. All rights reserved.

JEL classification: G18; G10; C22; O53

Keywords: Macroeconomic austerity; Dynamic correlation; T-Bond market; Stock market; China

1. Introduction

During 2003 and early 2004, China experienced anexcessive investment boom. To cool this economic over-heating, in AprilMay 2004 the government put into prac-tice a series of tight policy measures. Included in thesepolicy measures were the following. The central bank

raised the reserve requirements and tightened credit lines.The China Banking Regulatory Commission required com-mercial banks to nix investment projects deemed to be ill-planned, low quality, and unconformable to the govern-ments industrial policies. The State Development andReform Commission ordered local authorities to controlthe debut of price-hiking projects within their jurisdictions.

According to the news media, following the austerity pro-grams, the Chinese stock and bond markets simultaneouslyunderwent drastic drops, which subsequently had conta-gious effects on financial markets in Hong Kong, the US,Japan, London, Australia and so on (for example, Japansstock price indexes reportedly fell by 400500 points). Ametaphor went: As the Chinese economy is having an

injection for allaying fever, the worlds financial marketssuffer a shivering fit (http://news.xinhuanet.com/fortune/2004-05/14/content_1468420.htm).

These observations and anecdotes seem to suggest that,in China, drastic policy changes have begun to impactdomestic financial markets (as well as international finan-cial markets), which then motivated the present paper toattempt a serious investigation on some related issues intowhich anecdotes do not and cannot provide insights. How-ever, we are not interested in how individual marketreturns, but rather how correlations between them, respond

0378-4266/$ - see front matter 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.jbankfin.2007.04.029

* Corresponding author. Tel.: +64 9 414 0800x9471; fax: +64 9 4418177.

E-mail address: [email protected] (X.-M. Li).

www.elsevier.com/locate/jbf

Available online at www.sciencedirect.com

Journal of Banking & Finance 32 (2008) 347359
http://news.xinhuanet.com/fortune/2004-05/14/content_1468420.htmhttp://news.xinhuanet.com/fortune/2004-05/14/content_1468420.htmmailto:[email protected]:[email protected]://news.xinhuanet.com/fortune/2004-05/14/content_1468420.htmhttp://news.xinhuanet.com/fortune/2004-05/14/content_1468420.htm


2/13

to policy shocks. We have chosen to focus on the correla-tion between T-bond and stock returns for three main rea-sons as follows.

First, to reduce portfolio risk via diversification, a keyinput required by risk managers to hold efficient portfoliosis the correlation between assets included in the portfolio.1

Portfolios that contain stocks and government bonds havebecome popular among investors, as the two asset classesare believed to have different risk-return characteristicsand their correlations to be low or even negative. Becausethe correlation between T-bonds and stocks plays a vitalrole in portfolio risk management and dynamic asset alloca-tion for investors, it has been extensively studied in the lit-erature. For example, an earlier study by Barsky (1989)looks at price co-movements between stocks and bonds,and finds that when investors are scared, they look forsafety. They adjust their portfolios to include more safeassets and fewer risky assets. This kind of movement is usu-ally referred to as a flight to quality. A recent study by

Ilmanen (2003) on the US stockbond correlation reportsthat the correlation between stock market and governmentbond returns was positive through most of the 1900s, butnegative in the early 1930s, the late 1950s, and recently.A negative correlation implies that investors have benefitedfrom the bond market upswing, offsetting some of theirlosses in stock markets. However, this combination mayhave severe implications for pension funding ratios, as bothequities and discount rates decline, sending assets and lia-bilities in opposite directions.

Second, correlations between the stock and bond mar-kets are important to policymakers. Since China entered

the WTO in 2000, the Chinese government has endeavoredto reform its financial system including capital markets, inorder to transform the conduct of macroeconomic policiesfrom being administrative to being market oriented in nat-ure (as required by becoming a WTO member). In the lat-ter context, the central bank cannot set specific pricetargets for stocks and bonds, and so has to utilize the infor-mation contained in the co-movements between the freelyadjusted prices of these assets to gauge, for example, mar-ket participants expectations about growth and inflation.In other words, the stockbond return correlationestimates may provide policymakers with useful comple-mentary information to determine whether market partici-pants are changing their views on inflation or economicactivity prospects. Quantifying contemporaneous relationsbetween the stock and bond markets also helps policymak-ers to estimate and control the unintended consequencesthat policies directed primarily at one market could havefor the other. To our best knowledge, the existing literaturelacks such a study as ours for China, despite the importantpolicy implications of the issues examined in the presentpaper.

Third, correlations between asset returns have beenviewed as an integral aspect of inter-financial market inte-gration, in the literature. Kim et al. (2006) and Berben andJansen (2005) examine the dynamic or time-varying corre-lation between stock and T-bond returns of several Euro-pean countries to infer the state and progress of their

financial integration, taking into account the influence ofthe European Monetary Union as a possible cause of struc-tural change. Kim et al. (2005) also conducted a similarstudy for stock market integration in Europe. In these stud-ies, the authors use return correlations to gauge the degreeof integration between financial markets: a high/low corre-lation implies a high/low level of integration. High, not justlow, stockbond correlations have also been established.For example, Kim et al. (2006) document that the inter-bondstock correlations for each of their sample Euro zonecountries and the weighted average of these for Euro coun-tries and also non-Euro zone countries once reached veryhigh levels, although they have been falling since the mid

1990s. The authors take these results to imply a fallingfinancial integration since the mid 1990s. Apart from corre-lation analysis, the cointegration framework is also a usefultool in studying the degree of market integration, and arecent application of cointegration analysis to the long-run equilibrium relationship among Chinas money, stockand T-bond markets has appeared in Yin (2005). However,a long-run relationship detected by cointegration tests iscredible only if the true relationship is constant or expe-riences few breaks over time. The fact that China has beenreforming its financial systems with frequent debuts of newreform programs allows one to argue that the number of

breaks is too large to permit a meaningful cointegrationanalysis. Such an unstable, time-varying relationshipamong financial markets ought to be modeled more appro-priately within a time-varying correlation framework. Bydoing so, our investigation into the correlation betweenChinas T-bond and stock returns will shed new light oninter-stockbond market integration that may vary acrossdifferent points in time. However, we use the previouslyreported estimates of the European stockbond return cor-relations as a reference point with which to compare theestimates of the Chinese stockbond return correlations,in order to infer the relative degree of stockbond marketintegration in China.

Our work contributes to the existing literature in at leasttwo aspects. One is the link of stockbond correlations toinformation shocks or macroeconomic factors. Recentstudies on these issues include Chordia et al. (2005) andLi (2002). In the former article that uses the US data, theauthors find that innovations to stock and bond marketliquidity and volatility are significantly correlated, andattribute this observation to the possibility that commonfactors such as monetary shocks and money flows driveliquidity and volatility in these markets. The latter papershows that the major trends in stockbond correlation inG7 countries can be explained by their common exposure

to macroeconomic factors, such as expected inflation,

1 An example that the correlation between three assets in a portfolioaffects their optimal weights which maximize its Sharpe ratio is available

from the authors upon request.

348 X.-M. Li, L.-P. Zou / Journal of Banking & Finance 32 (2008) 347359
http://-/?-http://-/?-


3/13

unexpected inflation and the real interest rate. Our studyfocuses on China as the largest emerging economy, andlinks stockbond and stockstock correlations to macro-economic austerity measures of an administrative natureand to the peculiar behavior of Chinese investors inresponse to information shocks. Exploring these unique

characteristics of the Chinese financial markets representsour attempt to fill the void in the literature.The other aspect is related to the technical front for

empirically investigating time-varying return correlations.Scruggs and Glabadanidis (2003) introduces flexibility intotheir specification for the time-varying covariance matrixof stock and bond returns, by assuming that conditionalsecond moments follow an asymmetric dynamic covariance(ADC) process proposed by Kroner and Ng (1998). Theuse of the ADC model enables them to examine how returnshocks and volatility are transmitted between the stock andbond markets. Cappiello et al. (2003) employ an asymmet-ric version of the dynamic conditional correlation (DCC)

model which allows for a structural break to investigateasymmetric dynamics in the correlations of internationalequity and bond returns. They are able to find strong evi-dence of asymmetries in conditional covariance of equityand bond returns, and significant evidence of a structuralbreak in conditional asset correlation upon the creationof the Euro. Connolly et al. (2005, 2007) estimate rollingcorrelations over time to examine whether the time-varia-tion of stockbond and stockstock return comovementscan be linked to stock implied volatility. It appears fromthe above-cited studies that those different approachestaken serve, and depend on, different objectives. Our

research objectives (to be detailed below) determine thatthe asymmetric version of the DCC model with structuralchange as employed by Cappiello et al. (2003) seems tobe a more appropriate econometric tool than thoseemployed by Scruggs and Glabadanidis (2003) and Con-nolly et al. (2005, 2007). However, to adapt to the realityof the Chinese financial markets, we modify the modelemployed by Cappiello et al. (2003) in the following inno-vative manners. First, we propose a new version of themodel which is termed as the mixed asymmetric DCC(MADCC) model, and apply it for the first time in the lit-erature. Our MADCC model is capable of capturing con-current responses, if any, of the conditional covariance ofstandardized return residuals to positive and negativeinformation shocks. Second, we allow for multiple struc-tural breaks in the conditional correlation. This enablesus to test the strength and duration of policy impacts onthe correlation. Third, we fit a Skew-t-GARCH model(Jondeau and Rockinger, 2005) in filtering return volatility,so that the notorious problems of non-normality and skew-ness in the return distribution can be properly addressed.

The objectives of this paper are twofold. One is toexplore the general question of how the correlationresponds to policy shocks. Specifically, we ask: (1) In whatdirection did the 2004 macroeconomic austerities affect the

stockstock and bondstock correlations? (2) How long

did the impacts last for? (3) Were there any differences inthe correlation patterns between the pre- and post-episodeperiods. (4) What are the possible implications of such dif-ferences for financial market integrations, compared to thestockbond market integration in Europe as investigated inprevious studies (e.g., Kim et al., 2006)? Answers to these

questions should carry useful complementary informationfor policymaking.The second objective of this paper is to use the news

impact surfaces of Kroner and Ng (1998) to study asym-metry in the impact of joint information shocks on cor-relation. Previous studies have only examined possibleasymmetry in the reaction of univariate return volatilityto information shocks for China, and evidence is mixedin terms of the sign of asymmetry (see, for example, Li,2003a,b). We extend the literature and see whether correla-tion between two asset returns responds asymmetrically tojoint bad news (represented by their last periods standard-ized return residuals being both negative) and joint good

news (represented by their last periods standardized returnresiduals being both positive), and whether in the same wayas developed economies financial market correlations.Joint bad/good news means that two assets prices fall/risetogether at a particular point in time. If, say, joint badnews (two prices falling together) increases the correlationbetween two assets, the probability that their prices willfurther fall together is high relative to the probability thatthey will further rise together following their previous risestogether. This correlation asymmetry has important finan-cial implications. For example, the standard mean-varianceinvestment theory advises portfolio diversification, but the

value of this advice might be questioned if all the assetstend to fall as they have already fallen. Also, the presenceof asymmetric correlations can potentially cause problemsfor hedging effectiveness, since hedging relies crucially onthe correlation between assets hedged (and the financialinstruments used). For these reasons, asymmetry relatedto joint information shocks is of more concern than asym-metry related to mixed information shocks (i.e., two assetsmoving in the opposite directions).

The paper is organized as follows. Section 2 describesthe data used in this study, and provides their descriptivestatistics. Econometric methods are presented in Section3, followed by empirical results in Sections 4 and 5. Section6 offers our conclusions.

2. Data

Data used in this study include the daily Chinese Gov-ernment Bond (T-bond) Price Index traded on the Shang-hai Stock Exchange, the daily Shanghai Composite PriceIndex, and the daily Shenzhen Composite Price Index,for the period from January 2, 2003 to November 30,2005. The T-bond price index data were obtainedfrom the Security Industry Research Centre, Asia-Pacific(SIRCA), while data on the two equity indices were col-

lected from Datastream. The Shanghai Stock Exchange

X.-M. Li, L.-P. Zou / Journal of Banking & Finance 32 (2008) 347359 349


4/13

(SSE) released for the first time the Chinese T-bond priceindex on January 2, 2003. The T-bond price index, with abase value of 100, is the weighted average of all T-bondstraded on the SSE on the basis of their issuing volumes,and is the only index available for us to empirically inves-tigate a T-bond market in China.2 The Shanghai and

Shenzhen composite price index contain all traded stocks(A-shares and B-shares) on the two exchanges. They areconsidered to reflect the overall price movements in theChinese stock markets.

Figs. 13 plot the (raw) return series of, respectively, theT-bond Price Index, the Shanghai Composite Price Indexand the Shenzhen Composite Price Index, calculated bytaking the first difference of the natural logarithm of thecorresponding price indices. Table 1 presents the descrip-tive statistics for the three return series. The skewness sta-tistic for the T-bond return series is negative and significantat the 1% level, indicating that the return distribution isnegatively skewed or there is a substantial probability of

a big negative return. The skewness statistics for the Shang-hai and Shenzhen stock index returns are both positive andsignificant at the 1% level, implying that there is a substan-tial probability of a big positive return. The excess kurtosisstatistics are all significantly greater than 0 at the 1% levelfor the three return series, suggesting that the T-bond andstock return series are leptokurtic. The above statisticsshow that assuming a normal distribution for the threereturn series is inappropriate, which is confirmed by highlysignificant JarqueBera statistics (at the 1% level) enablingone to reject the null hypothesis of normality.

Regarding the LjungBox statistic, there is little formal

guidance for the choice of the lag length, and so we decidedto use a rule of thumb: the lag length is determined by thesquare root of the sample size as 26.3 With this lag length,the LjungBox Q statistics suggest that the two stockreturn series display no autocorrelation, as the statisticsare not significant at the 5% level. So when demeaningthe two stock return series, no autoregressive terms needto be included. However, the T-bond return series suffersa serious problem of autocorrelation, as the associated Qstatistic is significant at the 1% level. This prompts us toinclude autoregressive terms in demeaning the T-bondreturn series later for the DCC estimation.

The ARCH test statistic is significant at the 1% level forthe T-bond, which suggests that the T-bond return series isheteroskedastic. The insignificant ARCH test statistics atthe 5% level seem to indicate that the two stock return ser-ies are free of the ARCH problem. However, we should becautious about these results, as their underlying normalityassumption does not hold according to the normality testresults just mentioned above. As will become clearer later,

under some non-Gaussian distribution assumptions, heter-oskedasticity in the two stock returns can be detected. Insummary, to capture possible skewness and excess kurtosisshown by the descriptive statistics, we will follow Jondeauand Rockinger (2005) by using the Skew-t-GARCH model

to estimate the volatilities of the three return series.

2 Note that data on inter-bank bond markets are not externallyavailable, and there does not yet exist a price index for all T-bondstraded on the Shenzhen Stock Exchange.

Return rate (%, T-bond)

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07

/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07

/27

2005/09/05

2005/10/20

Fig. 1. Return rate of the T-bond price index.

Return rate (%, Shanghai stock)

-6

-4

-2

0

2

4

6

8

10

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07

/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07

/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07/27

2005/09/05

2005/10/20

Fig. 2. Return rate of the Shanghai stock price index.

Return rate (%, Shenzhen stock)

-6

-4

-2

0

2

4

6

8

10

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07

/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07

/27

2005/09/05

2005/10/20

Fig. 3. Return rate of the Shenzhen stock price index.

3 Slightly changing the lag length around 26 does not alter the test

results.



5/13

3. Econometric methods

This section briefly outlines the econometric methodsused in this paper. Consider a return series ri,t that is gen-erated by

ri;t li;t h1=2i;t ei;t; i 1; 2; . . . ; n; 1hi;t xi die2i;t1 hihi;t1; 2

where li,t, the conditional mean, contains AR(p) terms ofri,t plus a constant,

4 such that the demeaned return seriesri,t li,t will have iid standardized residuals ei,t after theconditional volatility h1

=2i;t is filtered. For the two stock re-

turn series, the order p is chosen to be zero, while for theT-bond return series, p is set to 3, as the correspondingLjungBox Q(26) statistics become insignificant at the 5%level for the respective demeaned return series.

Instead of introducing asymmetry to the GARCHmodel (Eq. (2)) for hi,t while assuming normality on ei,t,we work on the assumption that ei,t is driven by the Sk-t(skewed Student-t) distribution; that is, possible asymmetry

occurs in ei,t rather than in hi,t. This practice follows closelythe existing literature (see, e.g., Jondeau and Rockinger,2005). The Sk-t distribution (pdf) is defined as:

Sk-tei;tjti; ki

bici 1 biei;tai=1ki2

ti2

ti12

if ei;t < ai=bi;

bici 1 biei;tai=1ki2

ti2

ti12

if ei;tP ai=bi;

8>>>>>>>>>>>>>>>:

3

where ai 4kiciti

2

ti1 ; bi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 3k2

i a2

iq

and ci Cti

1

2

=ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipti 2

pC ti

2

. In Eq. (3), the parameter ki introducesasymmetry to the standard Student-t distribution: Ifki = 0, then ai = 0 and bi = 1, giving rise to the standardStudent-t distribution which is symmetric. If 1 < ki < 0(0 < ki < 1), the distribution Sk-t(ei,tjti,ki) is negatively(positively) skewed. The parameter ti (>2) measures thedegree of freedom. A finite ti means positive excess kurtosis

(or fatter distribution tails than the normal distribution).When ti becomes infinite, the normal distribution results.The Sk-t distribution as described by (3) is potentially use-ful for modeling financial time series which are mostly fat-tailed and often skewed.

Using the volatility series hi,t obtained by estimatingEqs. (1)(3) the conditional covariance matrix of assetreturns ri,t (i= 1,2, . . ., n) may be expressed as:

Ht DtRtDt; 4

where

Dt

ffiffiffiffiffiffih1;t

p0 0

0ffiffiffiffiffiffih2;t

p..

....

.

.

...

...

.0

0 0 ffiffiffiffiffiffihn;tp

0BBBBBBB@

1CCCCCCCA

and

Rt

diag

Qt

1

Qt

diag

Qt

1;

5

where Rt is the conditional correlation matrix for standard-ized return residuals {ei}t (i= 1,2, . . ., n) the elements inwhich, qij,t, represent conditional correlation between ei,tand ej,t. Qt may be taken as the conditional covariance ma-trix of the vector et, the elements in which, qij,t, can be usedto compute qij;t qij;t= ffiffiffiffiffiffiffiffiffiffiffiffiffiqii;tqjj;tp .

To model Qt for estimating Rt, we adopt the diagonalversion of the asymmetric DCC model proposed by Shepp-ard (2002). The model allows for different dynamics acrosscorrelations between different assets. With this model, wecan explore the possibility that the correlations dynamic

evolution of, say, the T-bond market vs. the Shanghaistock market is different from that for, say, the Shanghaistock market vs. the Shenzhen stock market. Moreover,this model also allows for asymmetries in the responsesof correlations to information shocks (such as joint goodand/or bad news).

The diagonal version of the asymmetric DCC model isspecified as:

Qt Q A0QA B0QB G0ZG A0et1e0t1A B0Qt1B G0zt1z0t1G; 6

where

Table 1Descriptive statistics

Price index Skewness Kurtosis Q(26) JB ARCH

T-Bond 1.8203** 14.928** 174.66** 6784.9** 156.01**Shanghai stock 0.7965** 2.7095** 27.364 283.44** 0.1025Shenzhen stock 0.5110** 1.9275** 33.587 136.21** 0.0717

Note: The sample size is 700, spanning from January 3 to November 30, 2005. The Skewness column reports the statistics of skewness. The Kurtosiscolumn presents the excess kurtosis statistics. Q(26) is the LjungBox statistic with the lag length equal to 26, to test for autocorrelation. JB denotes theJarqueBera statistics of test for normality. The ARCH column sets out the LM test statistics for ARCH. ** indicates significance at the 1% level.

4 We have also explored the possibility that the 2004 austerity had animpact on the conditional mean by adding a dummy to li,t. The testresults, however, show no significant evidence of structural breaks in thelevel of average returns on T-bonds, Shanghai stocks and Shenzhen

stocks. The results are available from the authors upon request.

http://-/?-http://-/?-


6/13

Q

1 qe12 qe1nqe12 1

..

....

.

.

...

...

.qen1;n

qe1n qen1;n 1

0BBBBBB@

1CCCCCCA

for t 1; 2; . . . ; T;

Z

1 qz12 qz1nqz12 1

..

....

.

.

...

...

.qzn1;n

qz1n qzn1;n 1

0BBBBBB@

1CCCCCCA

for t 1; 2; . . . ; T;

A

a1 0 00 a2

..

....

.

.

...

...

.0

0 0 an

0BBBBB@

1CCCCCA; B

b1 0 00 b2

..

....

.

.

...

...

.0

0 0 bn

0BBBBB@

1CCCCCA

and G

c1 0 00 c2

..

....

.

.

...

...

.0

0 0 cn

0BBBBB@

1CCCCCA:

In addition, et is a column vector of n standardized returnresiduals, while zt is a column vector containing all ele-ments in et that are negative (or positive) and zeros other-wise. qeij denotes the unconditional correlation coefficientbetween ei,t and ej,t, and q

zij denotes the unconditional cor-

relation coefficient between zi,t and zj,t. Notice that, when

cs are set to zero, Eq. (6) collapses to a symmetric DCCmodel; and when ai = a(i= 1,2, . . ., n), bi = b (i=1,2, . . ., n) and ci = c (i= 1,2, . . ., n), Eq. (6) becomes ascalar version of the asymmetric DCC model whichconfines the dynamics to being identical across all the cor-relations between different assets. This restriction is unnec-essary, and will limit the models applicability to differentcircumstances.

To facilitate the exposition of our methodological inno-vations, we rewrite Eq. (6) in an element version as:

qij;t 1 aij bijqij gijqzijh i

aijei;t1ej;t1

bijqij;t1 gijzi;t1zj;t1; 7where aij = aiaj, bij = bibj, gij = cicj, zi,t, zj,t = I(eit < 0)eit,I(ejt < 0)ejt for negative-signed asymmetry (or zi,t, zj,t =I(eit > 0)eit, I(ejt > 0)ejt for positive-signed asymmetry),and i, j= 1,2, . . ., n.

Previous studies using asymmetric DCC models haverestricted asymmetries (represented by gijq

zij and gij, zi,t1,

zj,t1 in (7)) as extra responses to the same type of shocks,such as negative shocks, of all the correlations betweenassets i, j= 1,2, . . ., n. However, it is possible that, in theChinese context, the conditional covariance between differ-ent assets might have, concurrently, extra responses to

news shocks of different types/signs. That is, the condi-

tional covariance between, say, the T-bond and Shanghaistock returns might be more responsive to bad news,whereas at the same time the conditional covariancebetween, say, the Shanghai and Shenzhen stock returnsmight be more responsive to good news. This issue hasnot yet received attention so far. To explore the possibility,

we propose to modify the standard asymmetric DCC toallow for the coexistence of positive and negative leverageeffects, and term such a modified model mixed asymmetricDCC (MADCC), the details of which are given asfollows.

Suppose that for i= 1 only and j= 1, . . ., n, asymmetryis negative-signed (i.e., zi;t zi;t Ieit< 0eit;zj;t zj;t Iejt< 0ejt, and hence qzij qzij EIeit < 0eitIejt 0eit, zj;t zj;t Iejt> 0ejt, and hence qzij qzij EIeit > 0eitIejt> 0ejt). In this case, (7) becomes:

qij;t 1 aij bijqij gijqzijh i aijei;t1ej;t1 bijqij;t1 gijzi;t1zj;t1

for i= 1 and j= 1, . . ., n; and

qij;t 1 aij bijqij gijqzijh i

aijei;t1ej;t1 bijqij;t1 gijzi;t1zj;t1

for i= 2, . . ., n and j= 1, . . ., n. The resulting parameterrestrictions are imposed in the following manner. Whenzi;t

zi;t

I

eit< 0

eit and zj;t

zj;t

I

ejt< 0

ejt;gij

cicj

with i= 1 and j= 1, . . ., n are freely estimated to give theestimates as ^gij ci cj , but gij = cicj with i= 2, . . ., n andj= 1, . . ., n are restricted to be zero. When zi;t zi;t Ieit> 0eit and zj;t zj;t Iejt> 0ejtgij cicj with i= 1and j= 1, . . ., n are restricted to be 0, but gij = cicj withi= 2, . . ., n and j= 1, . . ., n are freely estimated to give theestimates as ^gij ci cj . We will subject these parameterrestrictions, or our search for the particular MADCC, tostatistical tests.

We also consider possible structural breaks in themodel. Take the case where breaks occur in the correlationmean as an example, and thus (6) becomes:

Qt Q1 A0Q1A B0Q1B G0Z1G

Q2 A0Q2A B0Q2B G0Z2G A0et1e0t1A

B0Qt1B G0zt1z0t1G; 8

where Q1 is defined as Q for t < s (break date), and Q2 fortP s. Analogously Z1 is defined as Z for t < s, and Z2 fortP s. The null hypothesis is Q1 Q2 Q over the entiresample period, and it generates a value of the log likelihoodfunction in (9). The alternative hypothesis is Q1 6 Q2, andit generates another value of the log likelihood function. Bycomparing the two values of the log likelihood function

using the likelihood ratio test statistic, we can decide on



7/13

whether or not the null can be rejected in favor of thealternative.

Estimation of the models is carried out by maximizingthe following log likelihood function of the DCC modelwith Sk-t distribution (see Jondeau and Rockinger, 2005):

L XT

t1lnSk-tetjt; k

1

2XTt1

ln jHtgj 9

with respect to the parameter vectors t, k and g (g containsall the parameters in the GARCH and DCC models).

4. Estimation and testing results

Table 2 reports the estimation results for the Skew-t-GARCH model for the three return series. All the d andh parameters for the conditional variances of the threereturn series are positive and statistically significant at the1% level, indicating volatility clustering of the three return

series. Overall, volatility shocks appear to be persistent asd + h is close to unity for all the return series. Recall thatthe parameter ki introduces asymmetry to the standard Stu-dent-tdistribution. For the T-bond index, a negative k sug-gests that its return distribution is negatively skewed. Onthe other hand, a positive k indicates that the return distri-bution is positively skewed for the Shanghai and Shenzhenstock indices. The estimates of the parameter t are all finiteand range between 4 and 11, albeit with a low precision forthe two stock indices, signifying that the three asset returnsare all driven by non-normal, fat-tailed distributions.

Figs. 46 plot three series of return volatility from the

Sk-t-GARCH model. There are clear differences in the vol-atility pattern between the T-bond and the two stock returnseries. Overall, the T-bond volatility is relatively smallerthan the stock volatilities, consistent with the wisdom thatT-bonds have lower risk, or are safer, than stocks. The twostock volatilities are quite similar in terms of their patterns

and magnitudes. There is a notable observation that the T-bond volatility exhibits spikes far above the average ampli-tude around May of 2004 (the peak value, 0.8132, occurredon May 10, 2004). The spikes could be an indication thatthe policy shocks were large enough to have caused breaksin the correlation of the T-bond and stock markets.Whether this is the case, however, needs to be formally

tested, to which we now turn.

Table 2Estimation results of the Sk-t-GARCH model

Parameter T-bondmarket

Shanghai stockmarket

Shenzhen stockmarket

x 0.0003** 0.0605** 0.0341**

(1531) (10.77) (11.63)d 0.2307** 0.0440** 0.0429**

(50.67) (40.31) (26.97)h 0.7693** 0.9195** 0.9384**

(87.40) (254.73) (251.1)t 4.6070** 7.3712 10.040

(11.17) (1.513) (0.651)k 0.0588** 0.1631** 0.1064**

(23.22) (54.13) (35.84)Note: This table gives the estimates of the parameters in Eqs. (2) and (3).When carrying out estimation, li,t in Eq. (1) is a constant for the two stockreturns, but equals a constant plus the AR(1), AR(2) and AR(3) terms ofri,t for the T-bond returns, such that the demeaned three return seriesri,t li,t have iidstandardized errors ei,t after the conditional volatility h1=2i;tis filtered. In parentheses are the t-statistics. ** indicates significance at the

1% level.

Return volatility (T-bond)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07

/27

2005/09/05

2005/10/20

Fig. 4. Return volatility of Chinese T-bonds.

Return volatility (Shanghai stock)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07

/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07

/27

2005/09/05

2005/10/20

Fig. 5. Return volatility of Shanghai stocks.

Return volatility (Shenzhen stock)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

2003/01/0

3

2003/02/2

4

2003/04/0

3

2003/05/2

2

2003/07/0

1

2003/08/0

8

2003/09/1

7

2003/11/0

3

2003/12/1

1

2004/02/0

2

2004/03/1

1

2004/04/2

0

2004/06/0

4

2004/07/1

4

2004/08/2

3

2004/09/3

0

2004/11/1

6

2004/12/2

4

2005/02/0

3

2005/03/2

4

2005/05/1

0

2005/06/1

7

2005/07/2

7

2005/09/0

5

2005/10/2

0

Fig. 6. Return volatility of Shenzhen stocks.

http://-/?-http://-/?-


8/13

We first consider our conjecture that positive and nega-tive leverage effects coexist in the correlation structure ofthe Chinese T-bond and stock markets. To do this, wecompare, via some statistical tests, three versions of theasymmetric DCC: (a) the positive asymmetric DCC(PADCC) where the conditional covariance of standard-

ized return residuals responds more strongly to all theresiduals joint positive than negative shocks; (b) the nega-tive asymmetric DCC (NADCC) where the conditionalcovariance of standardized return residuals responds morestrongly to all the residuals joint negative than positiveshocks; and (c) the mixed asymmetric DCC (MADCC)where the conditional covariance of standardized returnresiduals responds more strongly to some residuals jointpositive shocks than these residuals joint negative shocks,and more strongly to other residuals joint negative shocksthan these residuals joint positive shocks.

The t-statistics associated with the estimates of the ciparameters in Table 3 convey interesting messages regard-

ing the three versions of the asymmetric DCC model. Forthe NADCC, the c estimate associated with T-bond isstatistically significant at the 1% level, while the c esti-mates associated with Shanghai stock and Shenzhenstock are statistically insignificant even at the 10% level.The NADCC thus suggests that c1 = cT-bond > 0, c2 =cSH-stock = 0 and c3 = cSZ-stock = 0; that is, the conditionalcovariance of standardized return residuals has a greaterresponse to negative than positive shocks to the T-bondmarket only.

For the PADCC, the c estimate associated with T-bond is statistically insignificant even at the 10% level,

while the c estimates associated with Shanghai stockand Shenzhen stock are statistically significant at the1% level (see Table 3). The PADCC thus suggests thatc1 = cT-bond = 0, c2 = cSH-stock > 0 and c3 = cSZ-stock > 0;that is, the conditional covariance of standardized returnresiduals has a greater response to all joint positive thannegative shocks to the Shanghai and Shenzhen stock mar-kets, but not to joint positive than negative shocks to the T-bond and Shanghai (or Shenzhen) stock markets.

The fact that the NADCC and PADCC can only cap-ture either one, but not all, of the above-discussed two fac-ets of asymmetry motivates our search for a particularMADCC that is capable of embracing both of these twofacets in one model. Indeed, Table 3 shows that, for theMADCC, all the three c estimates now become statisticallysignificant at the 1% level. Moreover, the likelihood ratiotest statistics significant at a higher than 1% level enableus to reject the NADCC and the PADCC in favor ofthe MADCC. The MADCC suggests that c1 = cT-bond > 0only when shocks to the T-bond market is negative;c2 = cSH-stock > 0 and c3 = cSZ-stock > 0, only when shocksto the Shanghai and/or Shenzhen stock markets are posi-tive. That is, the conditional covariance of standardizedreturn residuals responds more strongly to negative thanpositive shocks to the T-bond market, but more strongly

to joint positive than negative shocks to the Shanghai Table3

E

stimationresultsofdifferentasymmetricDCCmodels

NADCC

P

ADCC

MADCC

ai

bi

ci

a

i

bi

ci

ai

bi

ci

T

-bond

0.0

235

0.7

870

0.0

214**

0.0

000

0.8

710**

0.00

00

0.0

219

0.9

276**

0.0

504**

(0.3

132)

(0.3

104)

(8.5

137)

(0

.0001)

(5.7

618)

(0.00

00)

(1.0

705)

(95

.561)

(3.0

393)

Shanghaistock

0.2

086**

0.7

914**

0.0

000

0.0

000

0.5

617**

0.36

59**

0.2

235**

0.2

058*

0.4

069**

(19

.712)

(81

.680)

(0.0

000)

(0

.0000)

(5.2

382)

(4.40

68)

(16

.154)

(2.1

674)

(42

.424)

Shenzhenstock

0.1

936**

0.6

886**

0.1

178

0.0

790

0.5

623**

0.35

87**

0.1

629**

0.2

146

0.3

217**

(29

.199)

(36

.965)

(1.5

382)

(0

.6594)

(9.7

834)

(16

.257)

(11

.265)

(1.6

357)

(25

.474)

H

0:NADCCistrue;H1:MADCCistrue

LRT=

28

.0172**

[0.0

000]

H

0:PADCCistrue;H1:MADCCistrue

LRT=

18

.7920**

[0.0

003]

H

0:nobreakinmean;H1:twobreaksinme

an

LRT=

26

.788**

[0.0

002]

N

ote:NADCCstandsforthenegativeasymm

etricDCCmodel.

PADCCstandsforthepo

sitiveasymmetricDCCmodel.

MADCCstandsforthemixedasymmetricDCCmodel.LRTstandsforthelog

likelihoodratioteststatistics.Inparentheses

arethet-statistics.Inbracketsarethepva

lues.

**indicatessignificanceatthe1%level.*

indicatessignificanceatthe5%level.



9/13

and/or Shenzhen stock markets. These results provide evi-dence in support of the coexistence of positive and negativeleverage effects in the Chinese capital markets.

Next, consider possible changes in the correlation struc-ture. Since the MADCC model outperforms the NADCCand PADCC models, we report the test results only for

the MADCC model. The tests were performed by search-ing break points over the period between April 2004 andDecember 2004. Table 3 sets out the likelihood ratio statis-tic for testing the null hypothesis of no break in the corre-lation mean against the alternative hypothesis of twobreaks in the correlation mean. The result shows that thenull is rejected at a higher than 1% level in favor of thealternative.

In fact, we also tested the null against the alternative ofone break in the correlation mean over the period fromApril to December, 2004, and the values of the log likeli-hood function associated with one break in the correlationmean were either smaller than or very close to that with no

break. In addition, it is possible that the two structuralbreaks also exist in the dynamics of correlation (i.e., inthe A and B matrixes). In order to see if this is the case,we again used the log likelihood functions to comparebetween breaks in the correlation mean only and breaksin both the correlation mean and dynamics. There wereno significant improvements in the log likelihood functionsthat could allow us to reject the null of two breaks in thecorrelation mean only in favor of the alternative of twobreaks in both the correlation mean and dynamics; and thisapplies to the NADCC, PADCC and MADCC models.

To sum up, we conclude that the MADCC model with

two breaks in the correlation mean is the most appropriatemodel to capture the possibility that there coexist negativeand positive leverage effects, and two structural breaks, inthe correlation of the Chinese T-bond and stock markets.Therefore, the estimation results of the model are used inaid of plotting a number of figures that can provide intui-tive insights into the main issues investigated in this paper.

5. The impacts of policy and news shocks on correlation

Figs. 79 depict the dynamics of correlation, derivedfrom the MADCC model, between, respectively, the T-bond and Shanghai stock returns, the T-bond and Shenz-hen stock returns, and the Shanghai and Shenzhen stockreturns. Recall from the preceding section that two struc-tural breaks were detected by the MADCC model. Thetwo breaks can be clearly seen by an inspection of Figs.79. Several messages emerge from the three figures, andare relevant to the questions posed earlier in Section 1.

First, a series of more austere macroeconomic contrac-tions brought into effect in April 2004 started to be felt con-cretely by the domestic financial markets in May 2004. Thisprovides evidence that the large policy shocks did have cer-tain impacts on financial markets albeit with a lag. Thespikes in the volatility of the T-bond returns as shown in

Fig. 4 are indeed an indication of structural change in the

dynamics of correlation: they were accompanied by anabrupt increase in the correlations between the T-bondand two stock returns as shown in Figs. 7 and 8, and inthe correlation between the two stock returns as shown inFig. 9. In particular, the rises in the two bondstock corre-lations are shown to be significantly greater than that in thestockstock correlation, approximately at a ratio of 0.150.02. This is consistent with the observation that the T-bond volatility underwent relatively large spikes whereasthe stock volatilities did not, around May 2004, implyingthat the T-bond market is more susceptible to the macro-

economic contractions than the stock markets. The rise

Correlation (T-bond vs Shenzhen stock)

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07

/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07

/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07

/27

2005/09/05

2005/10/20

Fig. 8. Plot of the conditional correlation between the T-bond andShenzhen stock returns.

Correlation (Shanghai stock vs Shenzhen stock)

0.82

0.84

0.86

0.88

0.9

0.92

0.94

0.96

0.981

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07/27

2005/09/05

2005/10/20

Fig. 9. Plot of the conditional correlation between the Shanghai andShenzhen stock returns.

Correlation (T-bond vs Shanghai stock)

-0.1

-0.05

0

0.05

0.1

0.15

0.2

2003/01/03

2003/02/24

2003/04/03

2003/05/22

2003/07

/01

2003/08/08

2003/09/17

2003/11/03

2003/12/11

2004/02/02

2004/03/11

2004/04/20

2004/06/04

2004/07/14

2004/08/23

2004/09/30

2004/11/16

2004/12/24

2005/02/03

2005/03/24

2005/05/10

2005/06/17

2005/07

/27

2005/09/05

2005/10/20

Fig. 7. Plot of the conditional correlation between the T-bond andShanghai stock returns.



10/13

in the correlation increased the overall risk of the portfoliosthat include T-bonds and stocks, as the probability thatthese assets prices rise/fall together was increased.

Second, the impact of the April austerity programs onthe correlation lasted for about 5 months, as Figs. 79show that the first break date is May 14, 2004 while the sec-

ond is October 29, 2004, although the second break pointin the stockstock correlation is indiscernible. The newsmedia reported, using data released a couple of monthslater, that the macroeconomic austerity measures turnedout to be effective and successful in three main aspects:the growth of fixed investment had declined rapidly; themoney supply had fallen to the targeted range set by thecentral bank; and the upward trend of price movementshad been stopped or even slightly reversed. Among econo-mists and the economic authorities, consensus was thenreached that macroeconomic stabilization should nowswitch from fighting against overheating/inflation to pre-venting potential overcooling/deflation. This does not

mean a switch from drastic contractions to comprehensiveexpansions. In late 2004, in fact, there was a call for mod-erate macroeconomic policies that would lessen the tight-ness of the existing austerity policies. The above mayexplain 5-month duration of the policy impact on the cor-relation of the domestic financial markets.

Third, before the first structural break, the two bondstock correlation series were positive but took negligiblevalues around 0.025. After the second break, the correla-tion series became negative with a slightly greater averagevalue of 0.05 (in absolute terms). The two observations,low stockbond correlations and change of the sign of

the correlations, are interpreted as follows.We propose that the observed low bondstock correla-

tions should be due mainly to the low level of stockbondmarket integration, for two reasons. First,5 recall from Eq.(6) that correlations investigated are between standardizederrors eit, that is, between returns which are demeaned aswell as volatility-filtered (i.e, mean- as well as risk-adjusted). This suggests that the difference in the risk-return characteristics between stocks and bonds is unableto account for the observed low bondstock correlations.Therefore, that information shocks (represented by eitwhich are iid) to the stock and bond markets are veryweakly correlated could be attributed to the two marketsbeing still highly segmented. Serious underdevelopmentof the bond market relative to that of the stock marketin China, as noted by XIANG Jun the Deputy Governorof the Peoples Bank of China (http://www.china-138.com/news/show.asp?id=46052), is an important rea-son for this segmentation. Second, as reviewed in Section1, previous studies (Kim et al., 2006; Cappiello et al.,2003) have reported quite large estimates of the inter-bondstock correlations for European markets especially

prior to the advent of EMU. Note that Cappiello et al.(2003) also use mean- and risk-adjusted returns for estimat-ing their conditional correlations. Compared with thoseEuropean results, our Chinese results can be taken to implythat the integration level of Chinas T-bond and stock mar-kets is still low, at least relative to that for the European

counterparts at some points in time.The negative bondstock return correlations followingthe second break could be a signal that the flight-to-qualityphenomenon has started to characterize the behavior of theChinese investors, albeit still to a small degree.

In addition to the dynamics of correlation, we alsoexamine asymmetry in the responses of correlation to jointbad news and joint good news. Following Kroner and Ng(1998), we employ the news impact surfaces. If the correla-tion between two asset returns has a greater/smallerresponse to joint bad news than to joint good news, thenews impact surface will delineate intuitively such correla-tion asymmetry. And it is more relevant to examine asym-

metry related to the same-signed information shocks thanthat related to the opposite-signed information shocks.The reason, as stated in Section 1, is because asymmetryfor the same-signed shocks has more important financialimplications than for the opposite-signed shocks.

A news impact surface for correlation is constructedusing all the estimated parameters of a DCC model.6 Thus,different DCC models employed would lead to differentnews impact surfaces. Our news impact surfaces are basedon the parameter estimates of the MADCC model7 withopposite-signed (i.e., both positive-signed and negative-signed) correlation asymmetries, while using the original

KronerNg news impact surfaces would involve theparameter estimates of a DCC model with the same-signed(i.e., either positive-signed or negative-signed) correlationasymmetry. An application of the KronerNg news impactsurfaces for correlation which have greater response tojoint bad news than to joint good news (i.e., negative-signed correlation asymmetry) can be found in Cappielloet al. (2003). These three authors provided the exact formu-lae for calculating the news impact surfaces based on whatwe now call the NADCC model. Compared with their for-mulae, our MADCC-based news impact surface formulaeseem more complex, a price paid for complexity. However,the advantage of our new technology is worth the price. Inthe Chinese context, for example, using the original Kro-nerNg news impact surfaces would be unable to delineatepositive-signed correlation asymmetry if based on theNADCC, or negative-signed correlation asymmetry ifbased on the PADCC. Our MADCC-based news impact

5 We thank a referee for the suggestion to conduct this risk-adjusted

analysis.

6 Kroner and Ng (1998) introduced the notion of the news impactsurfaces for, respectively, covariance, conditional variance and correla-tion. In this paper, we are only interested in the news impact surfaces forcorrelation.7 The formulae for calculating the news impact surfaces using the

parameter estimates of the MADCC model are available from the authors

upon request.

http://www.china-138.com/news/show.asp?id=46052http://www.china-138.com/news/show.asp?id=46052http://www.china-138.com/news/show.asp?id=46052http://www.china-138.com/news/show.asp?id=46052


11/13

surfaces enable one to view all co-existent correlationasymmetries, negative-signed and positive-signed, in theproper perspective.

Figs. 1012 depict three news impact surfaces (each hav-ing two panels (a) and (b) taken from two differentangles) for three correlations, respectively: The T-bond

vs. Shanghai stock correlation, the T-bond vs. Shenzhenstock correlation, and the Shanghai vs. Shenzhen stockcorrelation. One can see that the three correlation newsimpact surfaces are highly asymmetric. Figs. 10 and 11shows a larger response to joint negative shocks (in the, quadrant) than to joint positive shocks (in the+, + quadrant), of the two bondstock correlations. Inother words, the likelihood that T-bonds and Shanghai(or Shenzhen) stocks tend to fall together given that theyhave already fallen together is greater than the likelihoodthat they tend to rise together given that they have alreadyrisen together. The economic intuition of the result is this.After both the T-bond and stock markets have been hit by

bad news (i.e., joint negative shocks), a majority of inves-tors may respond by selling both T-bonds and stocks, lead-ing both the T-bond and stock prices to further falltogether. On the other hand, after both the T-bond andstock markets have been hit by good news (i.e., joint posi-tive shocks), a majority of investors may shift funds fromrisky stocks to relatively safe T-bonds, causing a furtherand large rise in the T-bond price but a small or no riseor even a fall in the stock price. So, bondstock return cor-relations in the case of joint good news are smaller than inthe case of joint bad news.

Fig. 12, however, marks asymmetry for the stockstock

correlation opposite to those for the bondstock correla-tions. It shows a larger response to joint positive shocks(in the +, + quadrant) than to joint negative shocks(in the , quadrant), of the stockstock correlation.Put differently, the probability that Shanghai and Shenzhenstocks tend to rise together given that they have already

risen together is greater than the probability that they tendto fall together given that they have already fallen together.We propose to interpret this finding as follows. After boththe two stock markets have been hit by good news (i.e.,joint positive shocks), more investors respond by purchas-ing both Shanghai and Shenzhen stocks than not doing so,

leading both the two stock prices to further rise together.On the other hand, after both the two stock markets havebeen hit by bad news (i.e., joint negative shocks), moreinvestors do not hurry to sell than those do. These arethe economic intuition for the finding that the probabilityof rising together following rising together is greater thanthe probability of falling together following fallingtogether.

It is worthwhile to devote more space to the storiesbehind the observation that the stockstock correlation isgreater following joint positive shocks than joint negativeshocks. According to Li (2004, p. 326), under informationasymmetry, the Chinese fund managers are risk-lovers.

When a stock market turns bullish, they tend to be pan-icked into purchasing stocks, in the expectation that thestock prices will rise further. When the stock market turnsbearish, however, they are less panicked into selling stocks,in the hope that market downturns will not last for long.These fund managers are aware of the possible existenceand eventual burst of bubbles, but also subjectively believethat they will not encounter the bust of bubbles while theirsubsequent followers will. In addition to this psychologicalbias, there also exist distorted incentives: When an invest-ment is making money, the fund managers share the profits(and the resultant performance-based remuneration is quite

high), but when an investment is losing money, only thefund holders are the bearers of the losses. The psychologi-cal bias and the distorted incentives thus lead to the asym-metric behavior of the Chinese fund managers acrossmarket upturns and market downturns. Li (2004) showsthat such asymmetric behavior is not confined to fund

Fig. 10. The correlation news impact surface for the T-bond and Shanghai stock returns.



12/13

managers: Individual investors are also found to be charac-terized by this behavior. Moreover, researchers have founda significant herding effect in China, even more signifi-cant than in the US [see Li (2004, p. 325), for the reviewof the studies on the herding effect in Chinas stock mar-kets]. It may be this herding effect that helps the asymmet-ric movements of individual markets lead to theasymmetric correlation between stock markets in China.

6. Conclusions

We now summarize the main findings of this study thatare believed to have important implications for policymak-ers, market participants and investors in China.

The co-movements of the Chinese capital markets doreact to large macroeconomic policy shocks as evidencedby structural breaks in the correlation following the drastic

2004 macroeconomic austerity measures whose impact

lasted for about 5 months. The T-bond market and its cor-relations with the Shanghai and Shenzhen stock marketsare found to have borne more of the brunt of the macro-economic contractions than the two stock markets andtheir correlation. Overall, however, the level of Chinasbondstock market integration is still low, at least lowerthan that in Europe at some points in time, although Chi-nas stockstock market integration has reached a quitehigh level. In addition, the relatively small volatility in T-bond returns implies that investors could reap diversifica-tion benefits via flight to quality (i.e., by moving theircapital out of riskier equities and into safer governmentsecurities). Such a portfolio-rebalancing strategy seems tohave started to be considered by Chinese investors afterthe episode of macroeconomic contractions in 2004.

As a methodological innovation, we propose a newversion of the asymmetric DCC model, MADCC,

which seems to have been quite helpful, as it enables us

Fig. 12. The correlation news impact surface for the Shanghai and Shenzhen stock returns.

Fig. 11. The correlation news impact surface for the T-bond and Shenzhen stock returns.



13/13

to successfully capture the coexistence of positive-signedand negative-signed asymmetries in the correlation struc-ture, a unique characteristic of the Chinese capital markets.The news impact surfaces based on the MADCC provideintuitionalized evidence that the bondstock correlationstend to increase only when their returns have both been hit

by bad news, but the stockstock correlations tend toincrease only when their returns have both been hit bygood news.

Acknowledgements

We thank Henk Berkman, Charles Corrado, Ben Jacob-sen, participants at the 14th SFM conference, the 2006AFS Meeting and the Department of Commerce BrownBag Seminar at Massey University, and two anonymousreferees for their helpful comments and suggestions. Theusual disclaimers apply.

References

Barsky, R.B., 1989. Why dont the prices of stocks and bonds movetogether? American Economic Review 79, 11321145.

Berben, R.P., Jansen, W.J., 2005. Bond market and stock marketintegration in Europe. DNB Working Papers 060, Netherlands CentralBank, Research Department.

Cappiello, L., Engle, R., Sheppard, K., 2003. Asymmetric dynamics in thecorrelations of global equity and bond returns. Working paper No.204. European Central Bank.

Chordia, T., Sarkar, A., Subrahmanyam, A., 2005. An empirical analysisof stock and bond market liquidity. Review of Financial Studies 18,85129.

Connolly, R., Stivers, C., Sun, L., 2005. Stock market uncertainty and thestockbond return relation. Journal of Financial and QuantitativeAnalysis 40, 161194.

Connolly, R., Stivers, C., Sun, L., 2007. Commonality in the time-variation of stockbond and stockstock return comovements. Journalof Financial Markets 10, 192218.

Ilmanen, A., 2003. Stockbond correlations. Journal of Fixed Income 12,5566.

Jondeau, E., Rockinger, M., 2005. Conditional asset allocation under non-normality: How costly is the mean-variance criterion? Research paperNo. 132. FAME International Centre for Financial Asset Manage-ment and Engineering, February, 2005.

Kim, S.J., Moshirian, F., Wu, E., 2005. Dynamic stock market integrationdriven by the European Monetary Union: An empirical analysis.Journal of Banking and Finance 29, 24752502.

Kim, S.J., Moshirian, F., Wu, E., 2006. Evolution of international stockand bond market integration: Influence of the European MonetaryUnion. Journal of Banking and Finance 30, 15071534.

Kroner, K., Ng, V., 1998. Modelling asymmetric comovements of assetreturns. Review of Financial Studies 11, 817844.

Li, L., 2002. Macroeconomic factors and the correlation of stock andbond returns. Yale ICF working paper No. 02-46.

Li, X.-M., 2003a. Time-varying informational efficiency in Chinas A-share and B-share markets. Journal of Chinese Economic and BusinessStudies 1, 3356.

Li, X.-M., 2003b. China: Further evidence on the evolution of stockmarkets in transition economies. Scottish Journal of Political Econ-omy 50, 341358.

Li, X.-D., 2004. Behavioral Finance: Theory and Evidence from ChineseStock Market. Shanghai Sanlian Shudian Publishing House, Shanghai.

Scruggs, J.T., Glabadanidis, P., 2003. Risk premia and the dynamiccovariance between stock and bond returns. Journal of Financial andQuantitative Analysis 38, 295316.

Sheppard, K., 2002. Understanding the dynamics of equity covariance.Manuscript, UCSD.

Yin, J., 2005. The linkage between financial markets in China: 20002004.Forum for Economic Development, Working Paper No. FC20050037.


Chinese Stock and Bond Market Analysis

Documents

Transcript of Chinese Stock and Bond Market Analysis