Spillovers of the U.S. Subprime Financial Turmoil to Mainland China and Hong Kong SAR: Evidence from Stock Markets 1
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
  • | 2 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: tsun@imf.org; zhang_xj@cass.org.cn

This paper focuses on evidence from stock markets as it investigates the spillovers from the United States to mainland China and Hong Kong SAR during the subprime crisis. Using both univariate and multivariate GARCH models, this paper finds that China's stock market is not immune to the financial crisis, as evidenced by the price and volatility spillovers from the United States. In addition, HK's equity returns have exhibited more significant price and volatility spillovers from the United States than China's returns, and past volatility shocks in the United States have a more persistent effect on future volatility in HK than in China, reflecting HK's role as an international financial center. Moreover, the impact of the volatility from the United States on China's stock markets has been more persistent than that from HK, due mainly to the United States as the origin of the subprime crisis. Finally, as expected, the conditional correlation between China and HK has outweighed their conditional correlations with the United States, echoing increasing financial integration between China and HK.

Abstract

This paper focuses on evidence from stock markets as it investigates the spillovers from the United States to mainland China and Hong Kong SAR during the subprime crisis. Using both univariate and multivariate GARCH models, this paper finds that China's stock market is not immune to the financial crisis, as evidenced by the price and volatility spillovers from the United States. In addition, HK's equity returns have exhibited more significant price and volatility spillovers from the United States than China's returns, and past volatility shocks in the United States have a more persistent effect on future volatility in HK than in China, reflecting HK's role as an international financial center. Moreover, the impact of the volatility from the United States on China's stock markets has been more persistent than that from HK, due mainly to the United States as the origin of the subprime crisis. Finally, as expected, the conditional correlation between China and HK has outweighed their conditional correlations with the United States, echoing increasing financial integration between China and HK.

I. Introduction

The subprime turmoil that began in the summer of 2007 initially manifested itself as a problem for U.S. financial institutions. At first, the turmoil was simmering within the United States, but subsequently it boiled over and set off vortices and currents not only in the U.S. financial markets, but also in the global markets. As a result of this tremendous financial turmoil, dramatic changes have taken place in the financial landscape and the global financial markets have been seriously affected.

The linkages between emerging market (EM) economies and advanced economies have become a major topic of debate during this episode of financial turbulence. Some argue that the ripple-effect of the mature market turbulence on EM countries has so far been muted and uneven, and that EM economies are still relatively unscathed, in part because the use of structured products was much less prevalent. On the other hand, others claim that the increasing global financial integration has potentially raised EM economies’ vulnerability to abrupt reversals in market confidence related to those subprime external shocks and spillover effects. That is, distant events in the United States can have sharp impacts on EM economies.

The question therefore naturally arises whether such financial turmoil has actually had any tangible effects on the daily movements of the stock prices in mainland China (China) and Hong Kong SAR (HK). This question remains relevant because China and HK experienced a long run-up in asset prices, including equities, despite the partial reversal since late 2008. Moreover, China and HK’s resilience will likely continue to be tested if financial uncertainty remains protracted and the global economic slowdown worsens.

This paper examines whether the U.S. subprime financial turmoil has had any statistically significant effect on both the daily returns and the conditional volatility of stock prices in China and HK. To capture the possible spillover effects, we employ a two-stage procedure; in the first stage we estimate a group of univariate GARCH models (referred to as UGARCH models below), and in the second stage we set up a group of multivariate GARCH models (referred to as MGARCH models below) to further identify the sources and magnitudes of the spillovers.

Using both univariate and multivariate GARCH models, this paper finds that (i) China’s stock market has not been immune to the financial crisis and there is no decoupling story, as clearly evidenced by the price and volatility spillovers from the United States to China in MGARCH models; (ii) HK’s equity returns exhibit more significant price and volatility spillovers from the United States, and past volatility shocks in the United States have a more persistent effect on future volatility in HK than in China, indicating that HK has been more affected due to its role as an international financial center; (iii) the lower cross-volatility between HK and China than between the United States and China shows that the impact of the United States on China is greater than that of HK in the context of volatility persistence, due mainly to the United States as the origin of the subprime crisis; and (iv) as expected, the conditional correlation between China and HK has outweighed their correlations with the United States, echoing increasing financial integration between China and HK.

This paper contributes to the existing literature in three important ways:

  • First, as far as we know, this paper is the first attempt to gauge the impact of the U.S. subprime turmoil on China and HK’s stock markets using a group of univariate and multivariate GARCH models.

  • Second, this paper compares the different responses of China and HK’s stock markets to the subprime turmoil, reflecting the different degree of China and HK’s financial openness and integration with the rest of the world.

  • Third, this paper can be helpful to the authorities in designing policy responses to the external shocks, particularly given the background of increasing capital account liberalization in mainland China.

The remainder of the paper is organized as follows: Section II provides some stylized facts regarding stock prices in the United States, China and HK, followed by a preliminary discussion of some spillover channels and basic characteristics of China and HK’s stock markets. Section III briefly reviews the related literature, while Section IV describes the data properties and methodology. Section V discusses the results of the estimated UGARCH and MGARCH models. Section VI concludes and provides policy implications.

II. Performance of China and HK’s Stock Markets—Stylized Facts and Some Preliminary Observations

This section first traces developments in the stock prices of several advanced and EM stock markets during the recent run-up and correction, and demonstrates this cycle by showing the previous peak and trough as well as market changes during the period September 15 – October 31, 2008, when the financial distress worsened dramatically following Lehman’s collapse. The section then discusses three characteristics of China and HK’s stock markets.

China and HK’s financial systems have coped relatively well with shocks from international capital markets, owing to the good fortune of having little exposure to the securitized products. However, China and HK do not seem to have been immune to the events in international markets. As Table 1 shows, by end-October 2008 stock prices in major economies had declined at least 30 percent since their peak in October 2007. In particular, those in China and HK experienced the largest declines (71 percent and 56 percent respectively, Figure 1). Coupled with the stock price declines, volatility in the U.S. stock market had risen since late 2006, with a noticeable spike in mid-2007 and September-October 2008 in the wake of the subprime crisis and the increased financial stress after Lehman’s collapse (Figure 2).

Figure 1.
Figure 1.

Stock Price Indices

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Source: Bloomberg L.P.
Figure 2.
Figure 2.

U.S. Market Volatility

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Source: Bloomberg L.P.Note:1The implied volatility of interest rate swaptions with maturities between one and six month. 2VIX represents the implied volatility of the S&P 500 index.
Table 1.

Emerging Stock Market Peaks and Troughs: Current Episode

(In percent)

article image
Sources: Bloomberg L.P. and authors’ calculations.

Such co-movement between advanced and EM stock markets is unlikely to be a coincidence. Some transmission channels translate the Wall Street hurricane into the fluctuations in EM stock markets. The possible spillover channels from the United States to China and HK are as follows:

  • Loss exposures. Financial institutions in China and HK had some direct loss exposures to the U.S. subprime crisis. For instance, the direct exposure of the Chinese financial system to U.S. structured credit products is reported to be about $10 billion.2 The direct exposure to subprime-related assets accounted for 3.7 percent of total banking assets (CBRC, 2008). Moreover, there are several channels through which Lehman’s collapse had a further impact on China and HK’s financial institutions, such as loan exposures, credit derivatives exposures and bond exposures. It was reported that seven Chinese listed banks announced total bond holdings of $721 million in the bankrupt U.S. investment bank Lehman Brothers as of September 22, 2008. HK banks’ total exposure to U.S. subprime securities and structured assets remained well below ½ percent of banking system assets (IMF, 2008a).

  • Loss of confidence. China and HK’s stock markets were increasingly affected by the indirect influence of confidence shocks. For example, at the beginning of the subprime crisis in August 2007, China and HK’s stock markets were still moving up. However, with the increasing number of news releases about the losses at China’s financial institutions and the pessimistic projections for the US economy, market sentiment in China and HK’s stock markets shifted to the downside. Both stock markets substantially echoed the incidents of the bankruptcy of Lehman, the sale of Merrill Lynch and the rescue plan of the AIG.

  • Uncertain direction of capital flows. Capital flows to China and HK could be affected by the subprime crisis, which might have two opposite effects. On the one hand, the credit crunch effect in advanced economies drove foreign capital out of China and HK to meet foreign investors’ liquidity needs. On the other hand, some foreign investors saw China and HK as still-resillence markets in which to invest.3 For example, in the first two quarters of 2008, China experienced a hot money inflow of about USD 130 billion, even larger than that for the whole of 2007 (US$ 124.9 billion), while in the third and fourth quarter it experienced a hot money outflow of about USD 7.2 billion and 90 billion (Figure 3) 4. The U.S. Resident’s Net Foreign Transactions in Foreign Corporate Stocks, as a proxy of capital flows to China and HK, has also showed more volatility in capital flows to HK since the beginning of the subprime crisis (Figure 4). Therefore, the volatility of large capital flows could potentially increase stock price volatility in China and HK.

  • Slower export growth. China and HK’s stock markets might also be negatively affected by the global slowdown and the consequent reduction in external demand. Slower export growth from China and HK to the United States would be a material drag on both economies’ growth. The impact of market expectations for slower export growth could be detrimental to stock market sentiment.

Figure 3.
Figure 3.

Hot Money Flows to China

(In billions of U.S. dollars)

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Source: State Administration of Foreign Exchange in China.Note: Newly added foreign exchange reserve is equal to the sum of foreign direct investments, trade surplus and hot money (other)
Figure 4.
Figure 4.

U.S. Resident’s Net Foreign Transactions in Foreign Corporate Stocks

(In millions of U.S. dollars)

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Sources: Haver Database; and U.S. Treasury.Note: The U.S. Resident’s Net Foreign Transactions in Foreign Corporate Stocks are used as a proxy of capital flows to China and Hong Kong.

Moreover, three characteristics of China and HK’s stock markets need to be briefly discussed for the purpose of this analysis below.

  • The development of China’s stock market has been greatly supported by the tremendous structural changes in the past four years. Before 2005, the two-tier share structure, under which about two-thirds of the shares of China’s listed companies were nontradable, had a negative effect on the development of the stock market.5 The uncertainties related to this split structure prevented the market from recovering from the previous decline, and affected the overall credibility of Chinese stock market. To cope with this problem, the Chinese authorities initiated the so called “reform of nontradable shares” in the capital market in 2005. With the gradual floating of nontradable shares in the stock market, the real market value of the shares began to be realized and contributed to the rising stock prices since 2005. The institutional reform has been a key driving factor in China’s stock market development.

  • China and HK’s financial markets are not homogeneous in their openness. China is still in the process of opening up its capital account and there are various sorts of restrictions on inward and outward capital flows, while HK is an open market to all foreign investors and issuers, and does not impose any restrictions on its residents investing or obtaining funding in foreign financial markets. This difference manifests itself in the very openness of the HK stock market and banking sector to Chinese issuers and institutions as opposed to the relatively limited openness of China’s financial market to HK’s institutions.

  • Policy choices and financial market development in China have important implications for HK to maintain its status as an international financial center with the increasing financial integration between the two sides. On the one hand, HK has been playing an important role in the opening up of China’s capital account and the development of China’s financial markets. On the other hand, China’s rapid economic growth and financial market development provided tremendous scope for HK to fortify its competitiveness as a leading international and regional financial center. This bilateral promotion has laid a foundation for the increasing comovements of stock prices on both sides.

III. Related Literature

A substantial amount of theoretical and empirical work has documented how stock returns and volatility are transmitted across economies.

On the theoretical side, a number of theoretical analyses based on the “revision of expectations” have been promoted. For instance, Kodres and Pritsker (1998) suggest that the existence of feedback traders and asymmetric information could lead to the propagation of shocks through portfolio rebalancing effects without changes in fundamentals. Calvo (1999), and Calvo and Mendoza (2000) argue that comovements in stock markets are caused by herd behavior among portfolio managers.

On the empirical side, correlation and contagion have been used to study spillovers. The correlation of stock returns across different markets has been widely applied to evaluate the spillover effects across stock markets. For example, after controlling for own-country news and a few other fundamental factors, Baig and Goldfajn’s research (1999) shows that the cross-country correlations in the stock markets remain large and significant. By identifying potential channels for financial market spillovers in twelve transition economies, Gelos and Sahay (2000) demonstrated that a visible increase in stock market correlations during the 1994-99 period, points to increased financial market integration. Chan-Lau, Mathieson, and Yao, using extreme value theory to uncover nonlinear relationships and analyze contagion in financial markets, found that contagion is higher for negative returns than for positive returns (2004).

Beyond correlation analysis, Vector Autoregression (VAR) models have also been used to gauge spillover effects. Using a variety of statistical methods including VAR models, Guimaraes-Filho et al. (IMF, 2008b) found that spillovers from the United States to Asian economies have grown stronger over time. By examining both the sources and size of spillovers across major industrial country regions, Bayoumi and Swiston (2007) concluded that global financial conditions were the main source of spillovers, and that the reduction in global financial volatility as a result of a more stable U.S. environment was crucial for lower global output volatility and greater financial certainty.

GARCH models are another set of techniques to weigh the spillover effects. For example, Chan-Lau and Ivaschenko (2003) found that the price changes and volatility spillovers generated from the United States—the Wall Street virus hypothesis—are the most important carrier of the price change spillovers between the United States and the Asian region, while volatility spillovers within Asia do not appear to play an important role. Booth, Martikainen, and Tse (1997) analyzed the price and volatility spillovers across Scandinavian stock markets using a multivariate EGARCH model and found that volatility transmission is asymmetric, spillovers being more pronounced for bad than good news. Worthington and Higgs (2001) examined the transmission of equity returns and volatility among Asian stock markets and investigated the differences that exist in this regard between the developed and emerging markets by using MGARCH to identify the source and magnitude of spillovers. Their results indicate the presence of large and predominantly positive mean and volatility spillovers. Nevertheless, mean spillovers from the developed to the emerging markets are not homogenous across the emerging markets, and own-volatility spillovers are generally higher than cross-volatility spillovers for all markets, but especially for the emerging markets.

Some other methodologies have also been used to discuss the spillovers. For example, Jobst and Kamil (IMF, 2008c), using multivariate extreme value theory (EVT) to quantify the joint behavior of extreme realizations of returns across different markets, find that the degree of financial contagion between stock markets in the United States and Latin America appears to be greater during periods of financial turmoil.

There are some very recent studies on the spillovers from advanced to emerging markets through stock market channels. The Global Financial Stability Report (IMF, 2008d), by developing an empirical (panel and VAR) framework to assess what drives EM stock prices, found that global factors can act as a channel for spillovers when the international economic environment changes.

IV. Data and Methodology

The analysis in this paper uses daily close-of-day stock market price indices based on national currencies in the United States, China and HK. Price returns are calculated as changes in log stock market prices, Rt=LnPt- LnPt-1 (Figure 5).6

Figure 5.
Figure 5.

Daily Equity Returns (January 1, 2007-October 31, 2008)

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Sources: Bloomberg L.P. and authors’ calculations.Note: LOGSHCOMP, LOGSHFSUB, LOGFXI, LOGHSI, and LOGHSF represent the log difference in Shanghai Composite Index, Shanghai Financial Index, iShares FTSE/Xinhua China 25 index fund, Hang Seng Index, and Hang Seng Financial Index, respectively.

Our framework tests whether the subprime financial turmoil has had any significant effect on daily changes in stock prices. This requires us to control for the effect of various domestic underlying factors (such as the state of monetary conditions and economic activity) and global financial market volatility variables. Therefore, two groups of independent variables are used: (i) domestic control variables in China and HK, and (ii) global financial market volatility variables.

Following Liu, Pauwels, and Chan (2008), we took the unanticipated effects or the “surprise” from these domestic factors to control for the effect owing to domestic macroeconomic conditions. We measure the surprises from these underlying macroeconomic factors by using the unanticipated effects of these monthly macroeconomic indicators or macroeconomic surprises on stock prices, which are measured as the differences between the official data on their release dates (real-time data) and their corresponding market forecasts that reflect market expectations (see Appendix C for a description of these market forecasts)7. Specifically, the interest rate differential (which attempts to reflect the interest rate parity condition), surprises in M2 growth, CPI inflation, and industrial production are taken as domestic control variables. The effect of external balances on stock prices is accounted for by surprises in the monthly changes of China’s trade balance and the difference between HK’s export and import growth.8

The market forecasts are obtained from Bloomberg and the Consensus Forecast, which conduct regular surveys of financial institutions both in China and abroad every month before the official data releases of these monthly indicators (Appendix C). The sample used for the analysis of this paper spans the period from January 1, 2005 to October 31, 2008.

On the global market variables, we take the Chicago Board Options Exchange’s Volatility Index (VIX) and Lehman Brothers Swaption Volatility Index (LBSPX) as the variables to capture market volatility and interest rate volatility.9 A detailed description of all domestic and global variables and their definitions are presented in Table 2.

Table 2.

Data Description and Transformation

article image
Sources: Bloomberg L.P.; and Consensus Forecast.
Table 3.

Daily Equity Price Returns: Summary Statistics

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: All equity indices are calculated in log difference. The Jarque-Bera test is a goodness-of-fit measure of departure from normality, based on the sum of sample kurtosis and skewness.

In order to test whether the subprime financial turmoil has had any significant effect on daily changes in stock prices, we use GARCH models to analyze stock return and conditional volatilities across the different stock markets described in the previous section. For asset returns, the GARCH class of models involves the estimation of an equation for asset returns and a conditional variance (σt2) specification.

For this analysis, we establish two groups of GARCH models: (i) a group of UGARCH models with subprime events; and (ii) a group of MGARCH models. These two groups of models are specified as follows.

UGARCH models

An important feature in the UGARCH models is that we construct a variable for subprime events by collecting daily financial news pertaining to the subprime turmoil (Appendix A lists the events) following the rule below:

event =1, if there is subprime negative news

event =0, otherwise

The event models are designed to capture the responses of China and HK’s stock markets to subprime events. In the case where asset returns follow an autoregressive process and are dependent on other variables, the model specification takes the following form:

Rt=constant+ζtRt1+λXt1+φVtf+ðeventt+t(1)

where the price return, Rt, is a linear function of its 1-day lagged-return. The vector X t-1 represents control variables, the vector Vtf represents measures of global financial market volatility, “eventt” represents subprime events, and єt is the error term in the return equation.

The conditional variance of the error term, σt2 is given by :

σt2=α0+α1  t12+β1σ t1  2+μevent(2)

where the regressors є2 t-1, and σt-12 are commonly denoted as the ARCH and GARCH components, respectively.

MGARCH models

While modeling volatility of the equity returns with the subprime events variable is a main focus of attention in this paper, understanding the co-movements of equity returns is also of great interest. It is therefore important to extend the considerations to multivariate GARCH models, which have been widely used to investigate volatility and correlation transmission and spillover effects (Tse and Tsui, 2002; and Bae, Karolyi, and Stulz, 2003). Since MGARCH models explicitly parameterize the conditional cross-moments, they can take account of the correlation in various stock returns and their volatility better than univariate GARCH models. Therefore, to further identify the sources and magnitudes of the spillover impact, we establish a group of MGARCH models.

Theoretically, the specification of an MGARCH model should be reasonably parsimonious while maintaining flexibility. For the purpose of the following analysis, a common form, Diagonal VECH model, is employed that restricts A and B to be diagonals.10 This Diagonal VECH model, where the coefficient matrices are rank one matrices, is identical to the Diagonal BEKK model (Engle, 1982; Bollerslev 1986; Nelson, 1991). One important feature of this specification is that it reduces the number of parameters estimated and guarantees that the conditional covariance matrix is positive semi-definite. Also, this specification allows us to identify the own-volatility spillover effects as well as cross-volatility spillover effects.

Since the standardized residual showed evidence of excess kurtosis, we assume that the errors follow a Student’s t-distribution to model the thick tail in the residuals.

The following MGARCH models are developed to examine the joint processes relating to the daily rates of return for the three markets, namely the United States, China, and HK, from January 1, 2005 to October 31, 2008. The sample period is also chosen to include only the subprime turmoil period (January 1, 2007–October 31, 2008) for the purpose of emphasizing the subprime crisis period. The following conditional expected return equation accommodates each market’s own returns and the returns of other markets lagged one period. Moreover, for China and HK stock markets, we also include the domestic macroeconomic variables and external shock variables that have been used in the previous UGARCH models. That is, the surprise data (differences between the official data on their release dates and their corresponding market forecasts that reflect market expectations) are taken as the value for this release date, while 0 is taken as the value for other days in the corresponding month.

Rt=constant+ζtRt1+ηRt1f+λXt1+φVtf+t(3)
σt2=α0+α1 t12+β1αt1  2(4)

The difference between equations (1) and (3) is that the latter has the lagged return on the United States, China and HK stock markets, Rft-1, where the superscript “f” refers to the foreign country relative to the independent variable for that equation, which is assumed to be the source of spillovers, replacing the subprime events variable in equations (1) and (2).

V. Empirical results

Following the above-mentioned steps, we obtain two groups of GARCH models: one is a group of UGARCH models, the other is a group of MGARCH models. Moreover, in each group, we run regressions respectively on composite stock prices (i.e., S&P 500 Index in the United States) and the financial equity index (i.e., S&P 500 Financials Index in the United States).

Autocorrelation functions for stock price returns implied persistence in the series and suggest an AR formulation for the returns equation. ADF tests (Table 4) suggest that the log change in the stock price series was I(0); the ADF tests also suggest that the VIX and LBSPX index were I (1). The Akaike Information and Schwarz Information Criteria suggest one lag of the dependent variable and regressors in the estimated equations (Table 5).

Table 4.

Equity Prices and Volatility Indices: Augmentted Dickey-Fuller Tests Statistics

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: All indicators are tested in the form of those as denoted in Table 2. *, ** and *** represent the Judgment parameters c, t and k represent intercept, trend and lags used in the ADF test significance at 10%, 5% and 1% level.
Table 5.

VAR Lag Order Selection Criteria

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: Two lags are tested and one lag is selected based on the minimum Akaike information criterion and Schwarz criterion value

The squared returns also exhibit patterns of persistence and clustering within China and HK over time (Figure 6). ARCH tests confirm the appropriateness of a GARCH formulation. The distribution of squared returns was also markedly skewed and leptokurtic, suggesting that the error term was nonnormally distributed (Table 6).11

Figure 6.
Figure 6.

Squared Returns (January 1, 2007-October 31, 2008)

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Sources: Bloomberg L.P. and authors’ calculations.Note: LOGSHCOMP, LOGSHFSUB, LOGFXI, LOGHSI, and LOGHSF represent the log difference in Shanghai Composite Index, Shanghai Financial Index, iShares FTSE/Xinhua China 25 index fund, Hang Seng Index, and Hang Seng Financial Index, respectively.
Table 6.

The Distribution of Squared Returns

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: The Jarque-Bera test is a goodness-of-fit measure of departure from normality, based on the sum of sample kurtosis and skewness.

A. UGARCH Models

In each UGARCH model, we take different dependent variables for China and HK. For China, we use three stock price indices: the Shanghai Composite Index, the Shanghai financial index and FXI to capture the spillover to the general equity index and the financial index 12. For HK, in the same vein, we use two stock price indices: the Hang Seng index and the Hang Seng financial index.

We now examine the empirical findings regarding the effect of the subprime events on both the mean and the conditional variance of China and HK’s stock price returns by taking January 2007 to October 2008 as the sample period to interpret the econometric results.

China

Table 7, which presents the empirical findings for China, shows that in all cases, the AR(1)-GARCH(1,1) specifications yield acceptable models of returns and conditional variance for the sample periods.

Table 7.

Regression Results of the Event Models: China

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: The dependent variable is the daily return of the equity prices and all explanatory variables are lagged one period. Berndt-Hall-Hall-Hausman (BHHH) is used as the optimization algorithm. All regressions follow the GARCH (1,1) model and are estimated by maximum likelihood. Standard errors are presented in parentheses.***significant at 1 percent level; **significant at 5 percent level; *significant at 10 percent level.

We first look at the intercept terms and the lagged daily return variable of the stock prices. The intercept terms for the Shanghai Composite Index and financial index are insignificant, while that for the FXI is significant at the 10 percent level. The sign of the FXI is negative, indicating that the FXI declines at a pace of 0.02 percent per day on average, holding other things equal. Moreover, the lagged daily return variable is significant in the FXI model. This distinctive performance of FXI might be due to the fact that its companies are listed in HK stock market, which is more responsive to external shocks.

We then look at the set of control variables pertaining to monetary conditions, economic activity, and external imbalances. The interest rate differential variable is statistically significant and with a negative sign for the Shanghai Composite Index and FXI, indicating that monetary tightening has a negative effect on China’s stock prices. Most other domestic macroeconomic variables are statistically insignificant. These results show that the “surprises” in domestic macroeconomic information have had no major impact on stock price returns in China.13

We now discuss the effect of external shocks on daily stock prices. The global market volatility variables and subprime events are generally statistically insignificant in all three models. These findings suggest that external shocks have not had any significant influence on China’s daily stock prices.14

The results from the variance equations of the GARCH (1,1) model are presented in the bottom panel of Table 7. We find that the signs of the coefficients of the ARCH, GARCH and subprime events for all three models are both positive and statistically significant, indicating that the subprime turmoil has increased the conditional volatility of the financial sector and HK-listed companies’ stock prices. Moreover, the coefficients of these variance equations were also nonnegative in all cases, as required, to ensure that the conditional variances are well defined. In addition, in all cases, α11 was less than 1, producing (positive) finite estimates of unconditional variances. More interestingly, the subprime events variables are significant in the models of Shanghai Composite Index and FXI, indicating that the volatility of the financial sector and FXI equity indices have responded to the external subprime shocks, which were basically driven by the financial sectors in advanced economies.

Hong Kong

Table 8, which presents the empirical findings for HK, shows that while there are similarities to the models on China, such as similar insignificance of most domestic macroeconomic variables, several distinguishing features stand out.

Table 8.

Regression Results of the Event Models: Hong Kong SAR

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: The dependent variable is the daily return of the equity prices and all explanatory variables are lagged one period. Berndt-Hall-Hall-Hausman (BHHH) is used as the optimization algorithm. All regressions follow the GARCH (1,1) model and are estimated by maximum likelihood. Standard errors are presented in parentheses.***significant at 1 percent level; **significant at 5 percent level; *significant at 10 percent level.

First, in contrast to that of China, the intercept term for all HK models is positive, reflecting that HK’s stock prices increase at a pace of 0.01 percent per day on average, holding other things equal.

Second, the interest rate differentials are also statistically significant, but with positive signs. This might reflect HK’s role as an international financial center which attracts capital inflows to the HK stock market when HK’s interest rate is higher than that of the United States.

Third, the VIX, one of the global market volatility variables, is significant in both models with negative signs, indicating the negative impact of market volatility on HK stock price returns.

Lastly and most importantly, the subprime event variable is not only statistically significant in the variance equation (with positive signs), but also in the mean equation (with negative signs), indicating that the subprime events have had material negative effects on the conditional volatility and levels of HK stock price returns.

B. MGARCH models

We run two groups of MGARCH models based on composite and financial stock prices in the United States, China and HK.

The estimated coefficients and standard errors for the conditional mean return equations using composite indices are presented in Table 9. The price spillovers from the United States to HK have been more significant than to China with a coefficient of 0.27 to China and 0.76 to HK. This may reflect China’s less financial openness than HK.

Table 9.

Estimated Coefficients for Conditional Mean Return Equations

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: A common form, Diagonal VECH model, is employed that restricts A and B to be diagonals. All regressions follow the GARCH (1,1) model and are estimated by maximum likelihood using the Berndt-Hall-Hall-Hausman (BHHH) maximization algorithm. Standard errors are presented in parentheses. ***significant at 1 percent level; **significant at 5 percent level; *significant at 10 percent level.

The conditional variance covariance equations incorporated in the current paper’s multivariate GARCH methodology effectively capture the volatility and cross volatility spillovers between the three markets. Table 10 presents the estimated coefficients for the variance covariance matrix of equations using composite indices. These quantify the effects of the lagged own and cross innovations and lagged own and cross volatility persistence on the present own and cross volatility of the three markets.

Table 10.

Estimated Coefficients for Variance Covariance Equations

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: A common form, Diagonal VECH model, is employed that restricts A and B to be diagonals. All regressions follow the GARCH (1,1) model and are estimated by maximum likelihood using the Berndt-Hall-Hall-Hausman (BHHH) maximization algorithm. Standard errors are presented in parentheses. ***significant at 1 percent level; **significant at 5 percent level; *significant at 10 percent level.

Own-volatility spillovers in all three markets are large and significant indicating the existing strong ARCH effect. Specifically, the own-volatility spillovers in the United States are 0.04, while those for China and HK are 0.13 and 0.16, respectively. These are in line with the previous study that own-volatility spillover effects are generally higher for emerging markets than for developed markets (Worthington and Higgs, 2001). In terms of cross-volatility effects, the cross-volatility effects between the United States and China and HK were 0.07 and 0.08 respectively, and the cross-volatility effects between HK and China were 0.15, indicating higher cross-volatility between China and HK. This could be taken as further evidence that China and HK have become more financially integrated with each other than with the United States

In the GARCH set of parameters, most of the estimated coefficients are significant (Table 10). The lagged volatility persistence was 0.97 and 0.83 for the United States and HK respectively, while that for China is insignificant. This indicates a higher own volatility persistence in more developed and open stock markets. In addition, the cross-volatility persistence between the United States and China and HK was 0.40 and 0.90, respectively, while the cross-volatility between HK and China was 0.37. That is, past volatility shocks in the United States have a more persistent effect on future volatility in HK than in China, which is further evidence of HK’s position as an international financial center. In addition, this lower cross-volatility between HK and China also shows that the United States has a slightly larger impact on China than HK in the context of volatility persistence, due mainly to the United States as the origin of the subprime crisis

Conditional correlation analysis shows that the correlation between the U.S. S&P 500 index and the Shanghai Composite index has been much lower than that between the U.S. S&P 500 index and the Hang Seng index. In addition, the correlation between the Shanghai Composite index and the Hang Seng index has been the highest among the three stock markets. These results indicate limited openness in China, but the correlation between China and HK has outweighed correlation with the United States, echoing increasing financial integration between China and HK (Figure 7).

Figure 7.
Figure 7.

Conditional Correlation between the Composite Indices (January 1, 2007-October 31, 2008)

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Sources: Bloomberg L.P. and authors’ calculations.Note: LOGSPX, LOGSHCOMP, and LOGHSI represent the log difference in Standard & Poor’s Index, Shanghai Composite Index, and Hang Seng Index, respectively.

To assess the reliability of the results, we employ a battery of robustness checks by testing different dependent variables and sample period using UGARCH and MGARCH models. First, the results of the mean equation and variance equations generally hold for UGARCH and MGARCH models using financial sector stock prices (Table 7, 8, 11 and 12, Figure 8).

Figure 8.
Figure 8.

Conditional Correlation between the Financial Indices (January 1, 2007-October 31, 2008)

Citation: IMF Working Papers 2009, 166; 10.5089/9781451873139.001.A001

Sources: Bloomberg L.P. and authors’ calculations.Note: LOGSPXFINL, LOGSHFSUB and LOGHSF represent the log difference in Standard & Poor’s Financial Index, Shanghai Financial Index, and Hang Seng Financial Index, respectively.
Table 11.

Estimated Coefficients for Conditional Mean Return Equations Using Financial Sector Indices

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: A common form, Diagonal VECH model, is employed that restricts A and B to be diagonals. All regressions follow the GARCH (1,1) model and are estimated by maximum likelihood using the Berndt-Hall-Hall-Hausman (BHHH) maximization algorithm. Standard errors are presented in parentheses. ***significant at 1 percent level; **significant at 5 percent level; *significant at 10 percent level.
Table 12.

Estimated Coefficients for Variance Covariance Equations Using Financial Sector Indices

article image
Sources: Bloomberg L.P. and authors’ calculations.Note: A common form, Diagonal VECH model, is employed that restricts A and B to be diagonals. All regressions follow the GARCH (1,1) model and are estimated by maximum likelihood using the Berndt-Hall-Hall-Hausman (BHHH) maximization algorithm. Standard errors are presented in parentheses. ***significant at 1 percent level; **significant at 5 percent level; *significant at 10 percent level.

The minor differences are that the coefficients for mean spillover effects and ARCH effects are slightly lower, while the GARCH effects are higher than those in the composite index models with China’s own-volatility coefficient is now significant. Second, besides incorporating various additional dependent and explanatory variables and making various changes in the estimation methodology used in the previous part of this section, we conduct robustness checks for all UGARCH and MGARCH models, with respect to the extended starting sample date from January 1, 2007 to January 1, 2005 (Table 7 and 8). Overall, we find that the results are rather robust with respect to the sample period, additional dependent and explanatory variables, and various changes in the estimation methodology (Table 9 to 12).

VI. Conclusions and Policy Implications

The paper finds that China’s stock market is not immune to the financial crisis, as evidenced by the price and volatility spillovers from the United States. In addition, HK’s equity returns have exhibited more significant price and volatility spillovers from the United States than China’s returns, and past volatility shocks in the United States have a more persistent effect on future volatility in HK than in China, reflecting HK’s role as an international financial center. Moreover, the impact of the volatility from the United States on China’s stock markets has been more somewhat persistent than that from HK, likely due to the United States as the origin of the subprime crisis. Finally, as expected, the conditional correlation between China and HK has outweighed their correlations with the United States, echoing increasing financial integration between China and HK.

Based on this empirical evidence, a number of policy implications arise from the interlinkage between national and global developments and between economic and financial market developments.

  • There is no “decoupling” story for China and HK. No economy can be totally immune to the subprime financial turmoil. Our regression results strongly support this. In our case, even though the Chinese and HK economies may be diversifying their exports and becoming less dependent on the United States, their financial markets are still very much, if not more, influenced by U.S. monetary and financial conditions. China and HK have become increasingly integrated into the global financial system, and the authorities should be alert to the drift toward negative spillover effects.

  • Greater attention should be paid to the risk of a virulent feedback loop between the financial markets and the economy. The deteriorating subprime crisis has increased the downside risks to the United States and global economies, potentially taking their toll on the economies of China and HK.

  • International policy coordination has become more important. The tendency for the markets to transmit volatility rapidly leaves little time for policy-makers to intervene. Recent unprecedented circumstances have called for commensurate action to be taken by central banks, given the rising interdependence among economies. The fact that recent policy actions have been taken by both advanced and emerging market central banks (including China) underlines the need for coordination. Moreover, as evidenced by the increasing volatility spillover between China and HK’s stock markets in our empirical work, the interdependence has been increasing with the strengthening financial integration between the two markets, necessitating further cooperation between the mainland and HK authorities.

  • The Chinese authorities should be more cautious in their approach to capital account liberalization. Given increasing trade openness and financial spillovers from the international market, the capital account is de facto becoming more open over time irrespective of government attempts to control it (Prasad and Rajan, 2008). In this natural opening-up process, it become more urgent for the authorities to strengthen financial markets and relevant infrastructure (governance, the exchange rate system, the supervisory system, etc.) alongside the de facto capital account liberalization. It is argued that in the Asian financial crisis, China suffered less than other Asian countries just because of its lower level of financial openness. And even now, some consider that China’s limited exposure to the subprime crisis is due mainly to its lower level of financial openness. However, in the era of financial globalization, this limited financial openness, which has insulated China well in the past, may be much harder to maintain in the future. The key issue will be how to choose a pragmatic approach to capital account liberalization.

Appendix A.

Lists of Subprime Events

article image
article image
article image
article image
article image
article image
Sources: Bloomberg L.P. and Reuters.Note: These ongoing news events about the subprime crisis are used to capture their possible impact on price and volatility of equity markets.

Appendix B.

Members of FXI US Equity

article image
Source: Bloomberg L.P.

Annex C

Table 1.

Market Forecasts of Monthly Economic Indicators: China

article image
Sources: Bloomberg L.P.; and Consensus Forecast.Note: We take the difference in the growth rate between export and import as the indicator reflecting the trade balance in China. In addition, we take M2 as the indicator of money supply in China.
Table 2.

Market Forecasts of Monthly Economic Indicators: HK

article image
Sources: Bloomberg L.P.; and Consensus Forecast.Note: For the data availability reason, we take the difference between the growth in export and import as the indicator reflecting the trade balance in HK. Similarly, we take M1 as the indicator of money supply in HK.