Testing for market segmentation in the A and B share markets of China

  • Published on
    14-Feb-2017

  • View
    214

  • Download
    2

Transcript

  • This article was downloaded by: [University of Nebraska, Lincoln]On: 09 October 2014, At: 13:51Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

    Applied Financial EconomicsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/rafe20

    Testing for market segmentation in the A and B sharemarkets of ChinaPatrcia Chelley-Steeley a & Weihua Qian ba Finance, Accounting and Law Group , Aston Business School, University of Aston , AstonTriangle, Birmingham, B4 1AL, UKb Everbright Pramerica Fund Management Co Ltd , Shanghai, Chinac Finance, Accounting and Law Group , Aston Business School, University of Aston , AstonTriangle, Birmingham, B4 1AL, UK E-mail:Published online: 19 Aug 2006.

    To cite this article: Patrcia Chelley-Steeley & Weihua Qian (2005) Testing for market segmentation in the A and B sharemarkets of China, Applied Financial Economics, 15:11, 791-802, DOI: 10.1080/09603100500118930

    To link to this article: http://dx.doi.org/10.1080/09603100500118930

    PLEASE SCROLL DOWN FOR ARTICLE

    Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

    This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

    http://www.tandfonline.com/loi/rafe20http://www.tandfonline.com/action/showCitFormats?doi=10.1080/09603100500118930http://dx.doi.org/10.1080/09603100500118930http://www.tandfonline.com/page/terms-and-conditionshttp://www.tandfonline.com/page/terms-and-conditions

  • Testing for market segmentation in

    the A and B share markets of China

    Patrcia Chelley-Steeleya,* and Weihua Qianb

    aFinance, Accounting and Law Group, Aston Business School,University of Aston, Aston Triangle, Birmingham, B4 1AL, UKbEverbright Pramerica Fund Management Co Ltd, Shanghai, China

    Recent research has suggested that the A and B share markets of Chinamay be informationally segmented. In this paper volatility patterns in theA and B share market are studied to establish whether volatility changesto the A and B share markets are synchronous. A consequence of newinformation, when investors act upon it is that volatility rises. This meansthat if the A and B markets are perfectly integrated volatility changesto each market would be expected to occur at the same time. However,if they are segmented there is no reason for volatility changes to occur onthe same day. Using the iterative cumulative sum of squares acrossthe different markets. Evidence is found of integration between the twoA share markets but not between the A and B markets.

    I. Introduction

    For some time it has been argued that the ChineseA and B stock markets have been informationallysegmented, see for example Wo (1997), Sze (1993),or Kim and Shin (2000). Two of the most strikingcharacteristics of this market segmentation havebeen the huge discounts associated with B sharesrelative to A shares, and the finding that there arelead lag relationships between the returns of thefour stock markets.

    Research which has considered market segmenta-tion in the Chinese markets has consistently studiedthe discounts associated with B shares. For example,Fernald and Rogers (1998) showed that B shareinvestors paid about a quarter of the price that Ashare investors paid for shares in Chinese companies.Similar discounts are also noted by Chakravarthyet al. (1998), who discover a discount of 60% on Bshares relative to their A share counterparts. Morerecently, Kim and Shin (2001) have documentedthat the prices of A shares in Shanghai are on averageabout 66% higher than B share prices while Shenzen

    A share prices are about 52% higher than B shareprices.

    As well as heavy discounts in the B share market anumber of studies have drawn attention to leadlagpatterns in the returns of the four Chinese stock mar-kets. Research by Chan (1993) and Chui and Kwok(1998) have shown that there are strong cross auto-correlations between the returns of companies whichhave issued both A and B shares. More recently,Kim and Shin (2000) has used Granger causality teststo describe the nature of the leadlag relationships inreturns. They found that the returns of B shares leadthe returns of A shares and A shares in Shanghai leadShenzen A share returns, although there wasa weakening of this relationship after 1996.

    Both the heavy discount associated with B sharesand the leadlag pattern in returns have been linkedto information segmentation. Chakravarty et al.(1998) argue that the cause of such a huge discounton B shares is information segmentation caused bypoor information processing in the B share market.They argue that foreign investors find it more dif-ficult to acquire and assess information about local

    *Corresponding author. E-mail. P.L.chelley-steeley@aston.ac.uk

    Applied Financial Economics ISSN 09603107 print/ISSN 14664305 online # 2005 Taylor & Francis Group Ltd 791

    http://www.tandf.co.uk/journalsDOI: 10.1080/09603100500118930

    Applied Financial Economics, 2005, 15, 791802

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • Chinese firms because of language barriers, diverseaccounting standards and information barriers withinChina, which prevent the free flow of informationfrom one investor group to another.

    However, information barriers are not onlybelieved to exist in the B share market. Informationdissemination is also believed to be inhibited in theA share market. Chui and Kwok (1998) argue thatjournalists are prevented from reporting financialinformation about firms in a timely way becauseof significant laws which prevent the reporting ofeconomic information. As a result important newsabout the economy is often announced first in neigh-bouring countries such as Singapore or Thailand,allowing foreign investors to benefit from the newsfirst, ahead of A share investors.

    The aim of this paper is to study the extent towhich the A and B stock markets of China are infor-mationally segmented by looking at the correlationbetween volatility changes in each market. Using theiterative cumulative sum of squares (ICSS) procedureof Inclan and Tiao (1994) the dates on which volati-lity shifts to the unconditional variance of returns,took place in the Chinese stock markets are identi-fied. In previous studies researchers have relied heav-ily on information from cointegration tests. The ICSSalgorithm of Inclan and Tiao (1994) however allowsmuch more information to be gauged and can beprovided by cointegration results.

    As shown by Ross (1989) stock market volatilitycan be linked to the arrival of new information, whichincreases volatility as investors utilize new informa-tion that moves prices. If the A and B share marketsare perfectly integrated, so that investors share andutilize the same information set, volatility changes toone market should be reflected elsewhere in othermarkets. However, if markets are segmented so thatinvestors are relying on different information setsthere is no reason to expect volatility shifts in thedifferent markets to occur at the same time.

    It is found that the unconditional variance ofdaily returns for the Shanghai and the Shenzen A andB stock market are nonstationary. All stock marketsare characterized by numerous volatility shifts duringthe sample period. When the dates of the volatilityshifts are compared it is found that the A sharemarkets of Shanghai and Shenzen appear to bereasonably integrated because volatility changes inthese markets are synchronous. However, volatilityshifts to the A and B share markets do not occuron similar dates which indicates that they may beinformationally segmented. It is however noticedthat there is a reduction in segmentation after 1996as there are many more shared volatility shifts from1996 onwards.

    The remainder of this paper is set out as follows,in Section II the Chinese stock market is described,in Section III the ICSS algorithm of Inclan andTiao (1994) is described which will be used to findthe dates on which volatility changes. Section IVdescribes the data and provides some summarystatistics. Section V reports the dates of the volatilityshifts and Section VI discusses the information thatmay have caused volatility changes in the Chinesestock markets. Section VII provides a summary andconclusions to the paper.

    II. Chinas Stock Market

    The Shanghai stock market which opened inDecember 1990 was the first official stock market toopen in China. In July 1991 a second stock marketopened in Shenzen. The Shanghai market tends tospecialize in the listing of state-owned enterpriseswhile the Shenzen market specializes in smallerexport orientated companies, which are often foreignjoint ventures. Dual-listing between Shanghai andShenzen is not allowed.

    Both the Shanghai and Shenzen stock markets canissue two classes of shares, A and B shares. The dif-ference between A and B shares is that until March2001 A shares could only be held by Chinese resi-dents, while B shares could only be held by foreigninvestors or non-resident Chinese nationals. SinceMarch 2001 the B markets have been deregulatedto allow Chinese residents to hold B shares.

    Trading on the A share markets takes place indomestic currency, while trading on the B sharemarket takes place in Hong Kong Dollars inShenzen and US dollars in Shanghai. AlthoughA and B share markets are strictly segmented eachA and B share provides almost identical rights anddividends to shareholders. Although the owners of Bshares are likely to be institutional investors this isnot the case for A shares, traded A shares are heldalmost exclusively by individual investors.

    As Fernald and Rogers (1998) argue the attractionof issuing B shares arises because firms that issue Bshares are entitled to favourable tax treatment andhave more freedom to import and export goods.Issuing B shares can also provide valuable foreigncurrency. Companies are not able to issue B sharesin unlimited quantities as a limit of only 25% oftradable equity can be issued in the B share market.While some companies cannot issue B shares at all.The B share market has grown considerably sinceits start. In 1991 both Shanghai and Shenzen hadless than ten companies that issued B shares but as

    792 P. Chelley-Steeley and W. Qian

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • Su (1999) shows by 1996 each market had issued overa hundred B shares.

    Although there has been a substantial increase inturnover in the A share markets during the 1990strading volume in the B share market has remainedsignificantly below turnover in the A share market,see for example Fernald and Rogers (1998) or Su(1999). Since its introduction the B share markethas traded at a substantial discount to the A sharemarket, see for example, Sze (1993), Wo (1997) orSu and Fleisher (1999) who all document persistentheavy discounts on B shares relative to A shares.This has encouraged deregulation of the B sharemarket and the expectation that A and B marketswill eventually merge.

    Although the B share discount is well documenteda consensus on why it exists has yet to be found.Recently the underperformance of the B sharemarket has been linked to differences in the qualityof information locally and abroad and the rateof information transmission across the differentmarkets.

    Chakravarty et al. (1998) have argued that persis-tent B share discounts exist because foreign investorshave less information about local firms than Chineseresidents. Similarly, Su and Fleisher (1999) suggestthat large discounts to B shares exist because lessinformation arrives in the B share markets than atA share markets. Chui and Kwok (1998) also believethat the quality of information is important butargue that it is foreign investors that have the betterinformation because of the existence of informationbarriers within China.

    III. Detecting Variance Shifts

    To estimate the number of changes in variance andthe point at which each variance shift occurs Inclanand Tiao (1994) suggest a three-step algorithm. In thefirst instance, using the full data set the maximumabsolute value of the Dk series is calculated as

    Dk CkCT

    kT

    k 1, . . . ,T 1

    where, Ck and CT are the mean centred cumulativesum of squares calculated using k and T observationsrespectively. If there are no variance changes overthe sample period then the series Dk oscillates aroundzero but drifts up or down from zero when a varianceshift occurs. If Max|Dk|(n/2)

    1/2 is greater than the

    critical value (1.358 at a 95% level)1 then a possible

    variance change point has been found.

    Once a possible change point cpi has been identified

    after m observations the data should be partitioned

    into two groups spanning t1, . . . , tm1 : tm1, . . . ,T.The Max|Dk|(n/2)

    1/2 statistic is then calculated for

    each of the two new samples. In each of these

    two samples an additional change point could poten-

    tially be identified. This would require a further sub-

    division of the data until all the data has been

    examined in intervals t1 to each change point until

    T is reached and no further change points can be

    found.

    In the third step all NN change points should be

    recorded in order cp1, cp2, . . . , cpNN . Assuming the

    two extreme values are cp0 where t 0 and cpNN1where tT. Each possible change point should berechecked by calculating |Dk|(n/2)

    1/2 for data obser-

    vations spanning alternate change points (cpi : cpi2)

    until the change points cpNN2 : cpNN are reached.If Max |Dk|(n/2)

    1/2 no longer reaches the critical

    value the possible change point should be eliminated.

    This step should be repeated until the number of

    change points found in each pass of the data does

    not change and the change points found are close2

    to those of the previous pass.

    Once all the change points have been identified

    then whether volatility shifts occurred to the A

    share and B share markets at the same time can be

    compared. This allows whether the markets are seg-

    mented or integrated to be discovered. As shown by

    Ross (1989) the variance of price changes is directly

    related to be flow of new information. As new infor-

    mation passes from one stock market to another an

    increase in volatility will be observed as investors

    take advantage of the new information. Kim and

    Verrecchia (1991a, b) argue that the magnitude of a

    volatility change is directly related to the quality of

    information. The better the quality of information

    the greater the price change so the greater the vola-

    tility. This means that large amounts of volatility

    will be caused by news that has a high information

    content. But, news that has already been announced

    will have little or no impact on price changes

    and volatility. If the volatility shifts of the A and B

    share markets are found to be synchronous markets

    must be sharing the same information and be inte-

    grated. In contrast, if the A and B share markets do

    not experience synchronous volatility shifts then the

    markets are likely to be informationally segmented.

    1 n is the number of observations used to calculate Dk.2 Inclan and Tiao (1994) suggest within two data points.

    Testing for market segmentation in share markets of China 793

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • IV. Data and Summary Statistics

    Daily closing index values for the A and B sharemarkets of Shanghai and Shenzen, for the sample

    period 7 October 1992 to 31 December 2002 iscollected from the Taiwan Economic Journal. From

    the closing values a set of daily closing returns arecalculated for each stock market. The focus is on the

    indexes because individual securities will mask theeffect of market-wide information because they

    will be heavily influenced by idiosyncratic risk. Inthis section some summary statistics are examined

    for the full sample period and for three subsampleswhich span the periods 7 October 199231 December

    1996, 1 January 199731 December 2000 and

    1 January 200131 December 2002.Table 1 provides some return summary statistics

    for the four Chinese stock market indexes studiedin this paper. As can be seen by looking at the returns

    of each market the B share indexes provide noticeablylower returns than the A share indexes. During the

    extended sample period the Shenzen B share markethas not even provided a positive return for B share

    investors.By looking at the standard deviation of returns,

    the high volatility of the Chinese markets is obvious,especially in the Shanghai A share market. The high

    volatility of the four markets can also be gaugedby looking at the highest and lowest daily return

    observed during the sample period. For the Shanghai

    A share market both the highest and lowest dailyreturn was almost 100%, which illustrates the poten-

    tial for huge daily price movements that can beexperienced in this market. In the other markets the

    maximum highs and lows experienced are generallylower but still much higher than you would expect to

    observe in developed markets.If Table 1(b) is compared to Table 1(d) which

    provides the summary statistics for the two subper-iods it can be seen that in the A share markets hastended to fall as a move is made from one subperiod

    to another while volatility in the B share marketshas risen.

    Table 2 provides the autocorrelation coefficientsfor the full sample period and for the two subperiods.

    These indicate a tendency towards positive serial cor-relation. In the first subsample the Shanghai A share

    market is characterized by noticeable first order nega-tive serial correlation. Whereas, both B share markets

    are characterized by strong first order positive serialcorrelation. By the second subperiod, there had been

    a significant decline in serial correlation. Pioneeringwork by Dimson and Marsh (1979) and Lo and

    MacKinlay (1990) has shown that serial correlationin index or portfolio returns can be a consequence of

    nonsynchronous trading. The decline in the positiveautocorrelation of the B share market may therefore

    be related to the increase in trading volume that

    Table 1. Summary statistics

    RR% High Low Skewness Kurtosis

    (a): 7 October 199231 December 2002Shanghai A 0.0178 0.0243 0.9140 0.9006 1.7809 29Shanghai B 0.0181 0.0227 0.1238 0.1308 0.4170 4.67Shenzen A 0.0175 0.0241 0.2958 0.1963 1.2068 21.03Shenzen B 0.0254 0.0231 0.1379 0.1669 0.2664 7.41

    (b): 7 October 199231 December 1996Shanghai A 0.0045 0.0349 0.9140 0.9006 1.753 16.75Shanghai B 0.01369 0.0164 0.1238 0.1308 0.365 16.87Shenzen A 0.0307 0.0334 0.2958 0.1963 1.0738 14.72Shenzen B 0.0302 0.0197 0.1379 0.1669 0.2464 21.89

    (c): 1 January 199731 December 2000Shanghai A 0.08 0.0165 0.0869 0.0878 0.0689 6.69Shanghai B 0.0274 0.0259 0.0941 0.0994 0.3330 1.94Shenzen A 0.0675 0.0180 0.0868 0.0869 0.2448 6.31Shenzen B 0.0044 0.0247 0.0955 0.0997 0.5628 3.14

    (d): 1 January 200331 December 2002Shanghai A 0.0834 0.0139 0.0869 0.0878 0.0689 6.69Shanghai B 0.0455 0.0243 0.0941 0.0994 0.3330 1.94Shenzen A 0.0979 0.0148 0.0868 0.0869 0.2448 6.31Shenzen B 0.0587 0.0264 0.0955 0.0997 0.5628 3.14

    Notes: RR% is the daily percentage return, is the standard deviation of returns, High and Low are themaximum and minimum return achieved during the sample period, skewness and kurtosis are theskewness and excess kurtosis statistics.

    794 P. Chelley-Steeley and W. Qian

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • took place at this time between the first and secondsubperiod.

    To gauge the extent to which the markets are seg-mented from each other Table 3 presents the contem-poraneous return correlation coefficient between the

    four markets. These coefficients indicate that there isa significant amount of market segmentation betweenthe A and B markets. During the first subperiod,the correlation coefficient between the returns ofthe Shanghai A share market and its B share market

    Table 2. Daily autocorrelation coefficients

    1 2 3 Q(10) p

    (a): 7 October 199231 December 2002Shanghai A 0.2144* 0.0137 0.0522* 83.30 (0.000)Shanghai B 0.1790* 0.0155 0.0627* 121.39 (0.000)Shenzen A 0.0192 0.0490* 0.0089 24.28 (0.063)Shenzen B 0.1721* 0.0453* 0.0787* 87.30 (0.000)

    (b): 7 October 199231 December 1996Shanghai A 0.2304* 0.0173 0.0537* 72.94 (0.000)Shanghai B 0.3076* 0.0013 0.0015 101.39 (0.000)Shenzen A 0.0259 0.0711* 0.0062 13.72 (0.176)Shenzen B 0.2437* 0.1186* 0.1159* 103 (0.00)

    (c): 1 January 199731 December 2000Shanghai A 0.0335* 0.0459* 0.0638* 15.18 (0.125)Shanghai B 0.1115* 0.0282 0.1064* 32.43 (0.003)Shenzen A 0.0389* 0.0400* 0.0580* 14.35 (0.157)Shenzen B 0.1402* 0.0067 0.0743* 33.43 (0.000)

    (d): 1 January 200131 December 2002Shanghai A 0.0229 0.0414 0.0315 2.25 (0.82)Shanghai B 0.1079 0.0043 0.0044 9.95 (0.077)Shenzen A 0.1525 0.0220 0.0766 39.11 (0.00)Shenzen B 0.0487 0.0289 0.0298 2.17 (0.81)

    Notes: i are daily the serial correlation coefficients at lags one to three. Q(10) is the LjungBox serialcorrelation coefficient for up to lag 10, p is its probability value. * indicates a serial correlation coefficientsignificantly different from zero at a 5% level.

    Table 3. Return cross-correlation coefficients: daily contemporaneous correlation coefficient betweeneach of the four Chinese stock market returns

    Shanghai A Shanghai B Shenzen A Shenzen B

    (a): Sample period 7 October 199231 December 2000Shanghai A 1.00 0.4801 0.7389 0.3719Shanghai B 1.00 0.4674 0.7185Shenzen A 1.00 0.4815Shenzen B 1.00

    (b): Sample period 27 October 199231 December 1996Shanghai A 1.00 0.1830 0.4871 0.1019Shanghai B 1.00 0.1306 0.3526Shenzen A 1.00 0.1802Shenzen B 1.00

    (c): Sample period 1 January 199731 December 2000Shanghai A 1.00 0.4431 0.9615 0.4688Shanghai B 1.00 0.4203 0.7825Shenzen A 1.00 0.4599Shenzen B 1.00

    (d): Sample period 1 January 200131 December 2002Shanghai A 1.00 0.6324 0.9865 0.6263Shanghai B 1.00 0.6341 0.8841Shenzen A 1.00 0.6262Shenzen B 1.00

    Testing for market segmentation in share markets of China 795

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • is only 0.1830. Similarly, the correlation coefficientbetween Shenzen A and B share market returns isonly 0.1802. In the second subperiod the contempora-neous correlations between the markets rises to0.4431 and 0.4599 respectively. During the subperiodwhich coincides with the deregulation of the B sharemarket the correlation between both and A and Bshare markets rises further.

    In contrast, there is much greater comovementbetween the returns of the two A share markets inall subperiods. During the first subperiod the correla-tion coefficient between the Shanghai and ShenzenA share markets is 0.4871. While the correlation coef-ficient between the Shanghai B share market and theShenzen B share market is 0.3526. In the second sub-period, the A share markets have a contemporaneouscorrelation coefficient of 0.9615 which rises to 0.9865in the second subperiod. While the B correlation coef-ficient for B share returns rises to 0.7825 in the secondsubperiod and to 0.8841 in the third subperiod.

    These pairs of correlation coefficients suggestthat the two A share markets and the two B sharemarkets are much more integrated than the A and Bshare markets of a given exchange. This means thatalthough A share investors trade different types ofshares in Shenzen and Shanghai, these traders havemore in common with each other than they do withtheir B share counterparts. Since the fundamentalsof the B share investors are aligned more closelywith the fundamentals of A share traders there mustbe significant information segmentation between Aand B share markets.

    V. Variance Change Points

    In this section the Inclan and Tiao (1994) ICSSalgorithm is applied to allow the date on which thevolatility of each market changed significantly to beestablished. This is useful because as Ross (1989)showed volatility is linked to the flow of information.More recently Kim and Verrecchia (1991a, b, 1997)have shown that the amount of volatility in a marketis related to the quality of the information andthe extent to which the information has been pre-announced. Information that has already beenpre-announced has already been embodied in theshare price, so when it is actually announced pricesand volatility do not change. Since the indexes arewell diversified, market news will be important forboth sets of A and B share indexes. Shared volatilityshocks imply a degree of information integrationbetween markets while informationally segmentedmarkets will display non-synchronous volatilityshocks.

    Table 2(a)(c) showed that the index returns of theChinese markets tend to be serially correlated. Thismeans that the ICSS algorithm should not be appliedto the return series because the asymptotic propertiesof Dk assume that the series being tested is random.Prior to applying the ICSS algorithm it is ensuredthat the return series is random by applying anARMA(1,1) filter to each set of returns. Then theInclan and Tiao (1994) algorithm is applied in thespirit of Aggarwal et al. (1999) who study whethermarket-wide volatility changes for a wide rangeof markets is caused by domestic or world factors.If markets are fully integrated volatility shifts wouldbe expected to occur at about the same time as newinformation is received and acted upon simulta-neously by investors in all markets. If markets arecompletely segmented then volatility changes wouldnot be expected to occur at the same time becauseinvestors would be basing their trading decisions ondifferent information sets.

    Table 4(a)(d) indicates the dates of the volatilitychanges found when the ICSS algorithm is applied toeach market during the sample period. As can beseen in each market there was a considerable numberof volatility shifts, although, typically the B sharemarkets experienced more volatility shifts than theA share markets. For the Shanghai A and B sharemarket there were 14 and 22 respectively. The Shenzenmarket experienced more volatility shifts thanShanghai, its A and B share market had 20 and 28volatility shifts respectively.

    Also noticeable is the increase in the number ofvolatility changes during and after 1996. Prior to1996 there was no more than one volatility changein a year but during 1996 there were two for theShanghai A share market, four for the Shanghai Bshare market, one for the Shenzen A share marketand eight for the Shenzen B share market. The highernumber of volatility shifts continue after 1996.

    None of the volatility shifts occur synchronouslyacross all four markets. This suggests a high level ofmarket segmentation between the markets as in inte-grated markets volatility caused by the release of newinformation would be expected to affect markets atthe time it is released. Looking at the A share marketsonly, it can be seen that many of the volatility shiftsto the Shanghai market are also visible in the Shenzenmarket. This is especially noticeable from 1999onwards when all volatility shifts to the ShanghaiA share market are also experienced by the ShenzenA share market. In all cases the direction of the vola-tility changes are the same in both markets andare generally of a similar magnitude. This suggestsimportant market-wide information is reflectedin both A share markets at approximately the

    796 P. Chelley-Steeley and W. Qian

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • Table 4. Volatility shifts in the share markets

    Date of variance change Dk Interval of shiftIntervalvariance

    % change invariance Dk

    (a): Shanghai A share marketstart31 Dec 1993 1.4969

    31 Dec 1993 31 Dec 199326 May 1995 0.4020 73.14 5.3726 May 1995 26 May 199515 Nov 1996 0.0876 78.21 6.1715 Nov 1996 15 Nov 199619 Dec 1996 0.5410 517.54 4.6019 Dec 1996 19 Nov 199628 Oct 1997 0.1124 79.23 3.7128 Oct 1997 28 Oct 19977 May 1999 0.0297 73.56 5.427 May 1999 7 May 199922 July 1999 0.1588 434.5 4.6022 July 1999 22 July 19998 Sept 1999 0.0246 84.5 1.998 Sept 1999 8 Sept 199928 Oct 1999 0.0689 180.02 1.3828 Oct 1999 28 Oct 199930 Dec 1999 0.0135 80.34 2.1530 Dec 1999 30 Dec 199920 Mar 2000 0.0752 461.15 2.5720 May 2000 20 May 200026 July 2001 0.0138 81.59 1.9326 July 2001 26 July 200124 June 2002 0.0709 412.32 1.6824 June 2000 24 June 20021 Nov 2002 0.0098 86.18 3.581 Nov 2002 1 Nov 2002end 0.0291 196.54 3.58

    (b): Shanghai B share marketstart8 Feb 1993 0.3516

    8 Feb 1993 8 Feb 199322 Mar 1994 0.0803 77.16 2.3822 Mar 1994 22 Mar 199413 Feb 1995 0.0399 50.25 2.5413 Feb 1995 13 Feb 199513 June 1996 0.0183 54.14 3.1913 June 1996 13 June 19961 July 1996 0.1290 604.48 2.0401 July 1996 1 July 199029 Nov 1997 0.0198 84.66 2.3629 Nov 1996 29 Nov 199618 Dec 1997 1.3638 6787.13 6.0218 Dec 1996 18 Dec 199619 Feb 1997 0.0865 93.66 2.8119 Feb 1997 19 Feb 199715 April 1997 0.0134 84.51 2.1915 April 1997 15 April 19975 Jan 1998 0.1277 853.22 1.6005 Jan 1998 5 Jan 199813 Feb 1998 0.5963 367.02 2.2213 Feb 1998 13 Feb 199810 July 1998 0.0677 88.64 3.4810 July 1998 10 July 199811 Sept 1998 0.2167 219.90 2.3911 Sept 1998 11 Sept 199810 Mar 1999 0.0601 72.28 2.6910 Mar 1999 10 Mar 199930 Mar 1999 0.3618 502.31 2.5930 Ma 1999 30 Mar 199921 May 1999 0.1162 67.88 1.4221 May 1999 21 May 199927 Aug 1999 0.5063 393.26 1.8927 Aug 1999 27 Aug 199919 May 2000 0.0748 83.33 4.6919 May 2000 19 May 200025 Oct 2000 0.1756 134.10 2.4225 Oct 2000 26 Oct 200027 Feb 2001 0.0899 48.81 2.4827 Feb 2001 28 Feb 200126 Mar 2001 0.5429 504.08 2.4826 Mar 2001 27 Mar 200131 Jan 2001 0.1608 70.38 2.4831 Jan 2001 1 Feb 2002end 0.0436 72.91 2.48

    (c): Shenzen A share marketstart22 July 1994 0.1753

    22 July 1994 22 July 199410 Aug 1994 2.0430 1065.47 1.6910 Aug 1994 10 Aug 199425 Oct 1994 0.6052 70.38 4.0525 Oct 1994 25 Oct 199417 May 1995 0.0884 85.40 4.6317 May 1995 17 May 199525 May 1995 4.3034 4770.26 3.7025 May 1995 25 May 199522 April 1996 0.0468 98.91 2.4122 April 1996 22 April 19967 Jan 1997 0.2409 414.84% 2.327 Jan 1997 7 Jan 199730 April 1997 0.0716 70.29 3.2230 April 1997 30 April 19977 July 1997 0.3091 331.81 4.287 July 1997 7 July 199729 Oct 1997 0.1014 67.19 3.2629 Oct 1997 29 Oct 19976 Aug 1998 0.0328 67.62 1.676 Aug 1998 6 Aug 199821 Aug 1998 0.2752 738.27 2.8821 Aug 1998 21 Aug 19987 May 1999 0.0248 90.99 2.817 May 1999 7 May 199922 July 1999 0.1855 647.99 3.0122 July 1999 22 July 199928 Oct 1999 0.0420 77.36 2.7928 Oct 1999 28 Oct 199930 Dec 1999 0.0145 65.43 3.3730 Dec 1999 30 Dec 199922 May 2000 0.0834 474.25 4.1122 May 2000 22 May 200028 July 2001 0.0143 82.86 2.52

    (continued )

    Testing for market segmentation in share markets of China 797

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • same time. This is consistent with the results reportedin Section IV which showed an increasing degreeof co-movement between the daily returns of thetwo A share markets.

    For the B share market there are nine volatilitychanges that occur at approximately the same timein both markets. Of these volatility changes all butone are in the same direction in both markets. Theexception is the volatility change which occurs on27 March 1999 and 30 March 1999 in Shanghaiand Shenzen respectively. In Shanghai there is avariance increase but in Shenzen there is a volatilityreduction.

    When considering the A and B share markets thereis much more evidence of market segmentation.Looking at Shanghai first it can be seen thatonly two volatility changes are shared between the

    A share and the B share markets. This suggests ahigh degree of information segmentation across thetwo markets. For Shenzen there are slightly morecases of a shared volatility change but not enoughto give the perception that the two Shenzen marketsare integrated.

    The increase in the comovement between the fourmarkets after 1996 can probably be attributed to theincrease in turnover that occurred after this time.Increased turnover and participation in the stockmarket is likely to increase the information contentof prices. However, it is clear that even after 1996there exists a significant amount of segmentationbetween the A and the B share markets becausemost of the volatility shifts to each A and B sharemarket are non-synchronous. After 2001, when the Bshare market was deregulated there is no evidence

    Table 4. Continued

    Date of variance change Dk Interval of shiftIntervalvariance

    % change invariance Dk

    27 July 2001 28 July 200124 June 2002 0.0812 468.15 2.1124 June 2002 25 June 20021 Nov 2002 0.0106 86.91 2.111 Nov 2 Nov 2002end 0.0403 279.41 2.11

    (d): Shenzen B share marketstart03 Feb 1993 0.0968

    3 Feb 1993 3 Feb 199310 Feb 1993 1.1546 1092.49 1.6910 Feb 1993 10 Feb 199330 Nov 1993 0.0491 95.74 4.0530 Nov 1993 30 Nov 19932 Aug 1994 0.0068 86.24 4.632 Aug 1994 2 Aug 199428 Feb 1995 0.0266 293.87 3.7028 Feb 1995 28 Feb 199528 June 1995 0.0073 72.62 2.4128 June 1995 28 June 19957 May 1995 0.1082 1383.68 2.327 May 1995 7 May 199518 Oct 1995 0.0041 96.19 3.2218 Oct 1995 18 Oct 19957 Nov 1995 0.1088 2539.83 3.264.287 Nov 1995 7 Nov 19955 Feb 1996 0.0089 91.84 3.265 Feb 1996 5 Feb 19964 Mar 1996 0.0465 423.76 1.674 Mar 1996 4 Mar 199626 April 1996 0.0030 93.64 2.8826 April 1996 26 April 199612 June 1996 0.0377 1174.32 2.8112 June 1996 12 June 19961 July 1996 1.3623 3513.69 3.011 July 1996 1 July 199608 Aug 1996 0.0295 97.84 2.798 Aug 1996 8 Aug 199621 Aug 1996 0.2268 669.22 1.6321 Aug 1996 21 Aug 199615 Nov 1996 0.0132 94.19 3.3715 Nov 1996 15 Nov 19967 Jan 1997 0.8462 6322.06 4.117 Jan 1997 7 Jan 199717 Feb 1998 0.1277 84.91 4.3517 Feb 1998 17 Feb 19988 Mar 1999 0.0654 48.74 2.528 Mar 1999 8 Mar 3 199927 Mar 1999 0.1600 144.33 2.1127 Mar 1999 27 Mar 199927 July 1999 0.6250 291.81 2.2327 July 1999 27 July 199921 June 2000 0.1171 81.30 4.0021 June 2000 21 June 199927 Feb 2001 0.0591 49.39 1.8327 Feb 2001 27 Feb 200129 Mar 2001 0.6260 958.58 3.1529 Mar 2001 29 Mar 200131 Jan 2002 0.1698 72.88 3.1531 Jan 2002 31 Jan 200220 June 2002 0.0537 68.35 3.1520 June 2002 20 June 200222 July 2002 0.1646 206.37 3.1522 July 2002 22 July 2000end 0.0289 82.42 3.15

    Notes: Date of variance change is the date at which a significant variance shift was found to exist. Interval of shift is the timeperiod the variance was stationary. Interval variance is the variance of returns during the interval the variance was found to bestationary. The % change in variance is the percentage increase or decrease in variance measured for each interval. At a 95%level of confidence the critical value is 1.358.

    798 P. Chelley-Steeley and W. Qian

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • of an increase in comovement between the A and theB share markets. Volatility shifts to the A and Bmarkets do not become more synchronous after2001. This suggests that segmentation between thetwo markets is caused by more than partitioning ofthe two groups of investors.

    VI. Causes of Variance Shift

    Having established the date on which variancechanges took place the causes of the variance changesin each market are now investigated. To do this allthe financial news reports that were made in theChinese and world press on the days immediatelyprior and subsequent to a volatility change issearched. Because it is not known for certainty howlong it takes for a news item to appear in the press,due to journalistic restrictions that exist in China,all news reports up to tend days prior and subsequentto a volatility shift are studied. To study the newsreports the news archive service provided byNexisLexis is used. Unlike most developed marketsthere are relatively few news reports about theChinese stock markets each day. During some

    months there may be only one or two reports eachmonth, although, a noticeable increase in the numberof news stories which feature the Chinese stockmarket or Chinese companies takes place after 1996.

    Table 5 provides a list of all announcements madein the press which are sufficiently informative to havecontributed to a change in the volatility of the index.It should be noted that not all volatility changescan be explained by information announcements fortwo reasons. First, the coverage of financial news isnot as strong in China as it is in developed marketsso some important news events may not havebeen reported. Secondly, the Chinese stock marketis prone to speculative swings which could cause achange in volatility, but be unrelated to an importantevent.

    Looking at Table 5 it can be seen that regulatorychanges appear to have a significant influenceon volatility. When regulation has been tightened,volatility in the Chinese market has tended to decline.For example, when companies were compelled tomake performance disclosures for the first timein 1993 volatility of both B share markets declinedsignificantly, although, volatility of the A sharemarkets appears to have been unchanged.

    Table 5. (a) News events that appeared when volatility changed: Shanghai A share market

    Date Source Events Reported

    Dec 31, 1993 Xinhua News Agency New Corporate laws introduced requiring all publicshare issues to be underwritten and all sharetransfers to take place through a securities exchange.

    Dec 31

    May 26, 1995 Xinhua News Agency State Council Securities Committee announced quota ofnew share issues would be announced during secondquarter of 1995. Announcement caused some panicamong investors.

    May 24

    Nov 15, 1996 AP Worldstearm Two commercial banks were punished for illegal trading. Nov 15Dec 19, 1996 The Record Two sentenced to death for using public funds to invest

    in stock market.Dec 16

    Oct 28, 1997 Boston Globe Hkdollar/US dollar parity under threat. Hang-Seng fallsby 2%.

    Oct 25

    May 7, 1999Jul 22, 1999Sep 8, 1999 State owned enterprises allowed to invest in stock market

    for first time.Sep 9

    Oct 28, 1999 Xinhua News Agency Insurance companies allowed to invest in stock marketfor first time.

    Oct 28

    Dec 30, 1999 Financial Times Harsh penalties introduced for crimes related tosecurities and futures fraud.

    May 25, 2000 Financial Times Announcement that foreign funds denominated innational currency will be allowed to undertake Ashare stock market investment.

    May 26

    July 26, 2001 Financial Times FTSE launches Chinese A share market indexes. July 26Dec 28

    June 24, 2002 Worldwide Press Government announces end of sale in State ownedenterprises.

    June 25

    Nov 1, 2002 Agence France Presse China to allow foreign firms to buy securities incompanies listed on A share markets.

    Nov 3

    Testing for market segmentation in share markets of China 799

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • Table 5. (b) News events that appeared when volatility changed: Shanghai B share market

    Date Source Events Reported

    Feb 8, 1993 BBC State Council issued new regulations. Feb 5Mar 22, 1994 Xinhua News Agency CSRC announced cancellation of new share issue quota. Mar 13Feb 13, 1995 South China Morning Post Annual results released to shareholders for first time. Feb 10June 13, 1996 Wall Street Journal Dow Jones introduces three Chinese trackers. May 28July 1, 1996 Xinhua News Agency The Shanghai market applied new price index.

    Hong Kong returns to China.Jul 1

    Nov 29, 1996Dec 18, 1996 The record Two people sentenced to death for using public funds

    to invest in stock market.Dec 17

    Feb 19, 1997 Worldwide Dengs death. Feb 19Apr 15, 1997Jan 5, 1998 Singapore Zaobao Fiscal ministry announced new accounting regulations. Jan 1Feb 13, 1998July 10, 1998 Worldwide Press Russian Crisis starts. July/Aug/Sept

    Sep 9Sep 11, 1998Mar 10, 1999Mar 30, 1999May 21, 1999 Xinhua News Agency Announcement that foreign funds denominated in

    national currency will be allowed to undertakeA share stock market investment.

    May 20

    August 27, 1999May 19, 2000 Worldwide Press Taiwan president inauguration. Oct 26Oct 25, 2000 Worldwide Press China SRC makes pledge to open up stock

    markets to foreigners.Feb 27, 2001 Worldwide Press B share markets opened up to domestic residents. Feb 27March 26 Xinhua General News New regulatory system for stock market announced. March 27Jan 31 Xinhua News Market reports describe market as unstable. Jan 31

    Table 5. (c) News events that appeared when volatility changed: Shenzen A share market

    Date Source Events Reported

    July 22, 1994 Xinhua News Agency CRSC announcements did not dispel rumours of tax increases on trading ofstate owned shares.

    Jul 18

    Aug 10, 1994 Daily Telegraph CRSC announced that no new share issues would take place in 1994. Jul 30Oct 25, 1994May 17, 1995 Xinhua News Agency The bond futures market was suspended. May 17May 25, 1995 China Daily State Council Securities Committee announced quota of new share issues

    would be announced during second quarter of 1995. Announcementcaused some panic among investors.

    May 24

    Apr 22, 1996Jan 7, 1997 Worldwide Taiwan stock market falls by 1.6%. Jan 6Apr 30, 1997 Asia Pulse pts Ltd CSRC announced policies for settling problems of listing-companies.(?)July 7, 1997 Worldwide Hong Kong returned to China.

    Shanghai applied new price index.Jul 1

    Oct 29, 1997 Boston Globe Hong Kong dollar/US dollar parity under threat. Oct 25Aug 6, 1998Aug 21, 1998May 7, 1999July 22, 1999 China Daily Uncertainty about interim results reported.

    China Daily Insurance companies allowed to invest in the stock market for the first time.Oct 28, 1999 Financial Times Harsh new penalties introduced for crimes related to securities and futures

    fraud.Oct 27

    Dec 30, 1999 China Daily Insurance companies allowed to invest a greater proportion of their assets inthe stock market.

    Dec 29

    Mar 20, 2000 Mar 18Xinhua News Symposium on stock market regulation and development held.

    May 22, 2000 Financial Times FTSE launches five new indexes based on A share market. May 20July 27, 2001June 24, 2002 Worldwide Press Government announces sale of state owned enterprises to end.Nov 1, 2002

    800 P. Chelley-Steeley and W. Qian

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • Extending the rights of state owned enterprises

    and insurance companies to trade shares appearsto have influenced both A share markets. Since

    investment demand by these two groups couldonly influence the A share market, because they are

    unable to invest in B shares, it is not surprising that

    volatility in the B share markets did not react to thischange.

    Both B share markets were however influenced by

    the decision of the Dow Jones in May 1996 to intro-duce three funds which tracked the B share market.

    This development appears to have increased thevolatility of both the Shanghai and the Shenzen

    Markets considerably. As expected, this event had

    no impact on the A share market as foreign investorsare excluded from trading A shares.

    Political uncertainties have also influenced vola-

    tility. When Hong Kong was returned to China inJuly 1997, and during the time of Dengs death in

    February 1997 volatility declined. Since both theseevents were anticipated long before they actually

    occurred by the time they were officially announced

    much of the news content had already been utilizedby investors. An event that appears to have rallied

    all four stock markets in December 2000 was thereport that China had been the best performingstock market of 2000.

    VII. Summary and Conclusions

    In this paper the extent to which the stock markets ofChina are segmented from each other has been stud-ied. It is found that prior to 1996 there is a strikinglylow degree of comovement between the four Chinesestock markets, but especially between the A and Bshare markets of the Shanghai and Shenzen stockexchanges.

    Using the Inclan and Tiao (1994) ICSS algorithmthe dates of volatility changes in all four stockmarkets are identified. If a market is completelyintegrated volatility changes would be expected tooccur synchronously in all markets. But, if a marketis segmented so that the information sets of thedifferent investors are not shared volatility changescould not be expected to be synchronous.

    Strong evidence of market integration between thetwo A share markets and the two B share markets is

    Table 5. (d) News events that appeared when volatility changed: Shenzen B share market

    Date Source Events Reported

    Feb 3, 1993 S. China Post Company results to be announced for the first time.First announcement of this.

    Feb 1

    Feb 10, 1993Nov 30, 1993Aug 2, 1994 Chicago Tribune The listings of new stocks suspended. Aug 1Feb 28, 1995 Worldwide Barings Bank crashed. Feb 27Jun 28, 1995Jul 5, 1995Oct 18, 1995Nov 7, 1995Feb 5, 1996Mar 4, 1996 S. China Post New futures market regulation introduced. Mar 4Apr 26, 1996Jun 12, 1996 Wall Street Journal Dow Jones introduces three Chinese trackers. May 28Jul 1, 1996 Worldwide Shanghai market applied new price index.

    Hong Kong returned to China.Jul 1

    Aug 8, 1996Aug 21, 1996 Xinhua News Agency Central bank cuts interest rates. Aug 22Nov 15, 1996 Xinhua News Agency Two commercial banks were punished for illegal trades. Nov 15Jan 7, 1997 Fin Times Taiwan stock market falls 1.64%. Jan 6

    Singapore Zaobao New accounting regulations introduced. Jan 1Feb 17, 1998Mar 8, 1999March 27, 1999 Xinhua News Agency Quamnet set up: provides real time prices. May 23Jul 27, 1999Jun 21, 2000Feb 27, 2001 Worldwide Press B share markets opened up to domestic residents. Feb 27March 29, 2001 Gov announces sale of state owned enterprises to end. June 21Jan 31, 20, 2002June 2002July 22, 2002

    Testing for market segmentation in share markets of China 801

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014

  • found. Little evidence is found of market integrationbetween the A and B markets of a given stockexchange.

    When historical news stories are examined in orderto identify what events influenced the volatility of amarket it is found that regulatory changes have hadan important influence on both A and B sharemarkets as have political changes. Another importantcause of volatility changes have been adjustments tothe types of investors participating in the markets.Thus when state-owned enterprises and insurancecompanies were allowed to invest in the A sharemarket for the first time volatility changed.

    References

    Aggarwal, R., Inclan, C. and Leal, R. (1999) Volatilityin emerging markets, Journal of Quantitative andFinancial Analysis, 16, 35668.

    Chan, K. (1993) Imperfect information and cross-autocorrelation among stock prices, Journal ofFinance, 48, 121130.

    Chakravarty, S., Sarkar, A. and Wu, L. (1998) Informationasymmetry, market segmentation and the pricing ofcross-listed shares: theory and evidence from ChineseA and B shares, Journal of International FinancialMarkets, 8, 32555.

    Chui, A. and Kwok, C. C. Y. (1998) Cross-autocorrelationbetween A and B shares in the Chinese Stock Market,Journal of Financial Research, 21, 33355.

    Dimson, E. and Marsh, P. (1979) Risk measurement whenshares are subject to infrequent trading, Journal ofFinancial Economics, 7, 197226.

    Fernald, J. and Rogers, J. (1998) Puzzles in the Chinesestock market, Board of Governors of the FederalReserve System, International Finance DiscussionPapers no. 619.

    Inclan, C. and Tiao, G. (1994) Use of cumulative sumsof squares for retrospective detection of changesof variance, Journal of the American StatisticalAssociation, 89, 91323.

    Kim, O. and Verrecchia, R. (1991a) Market reaction toanticipated announcements, Journal of FinancialEconomics, 19, 273309.

    Kim, O. and Verrecchia, R. (1991b) Trading volume andprice reactions to public announcements, Journal ofAccounting Research, 29, 30221.

    Kim, O. and Verrecchia, R. (1997) Pre-announcementand event-period private information, Journal ofAccounting and Economics, 24, 395419.

    Kim, Y. and Shin, J. (2001) Interactions amoungChina-related stocks, Asia-Pacific Financial Markets,7, 97115.

    Lo, A. W. and MacKinlay, C. (1990) An econometricanalysis of nonsynchronous trading, Journal ofEconometrics, 45, 181211.

    Ross, S. A. (1989) Information and volatility: the noarbitrage martingale approach to timing and resolu-tion irrelevancy, Journal of Finance, 44, 117.

    Su, D. and Fleisher, B. (1999) Why does return volatilitydiffer in Chinese stock markets, Pacific Basin FinanceJournal, 7, 55786.

    Su, D. (1999) Ownership restrictions and stock prices:evidence from Chinese markets, Financial Review,34, 3756.

    Sze, J. (1993) The allure of B shared, The China BusinessReview, 6, 4248.

    Wo, C. S. (1997) Chinese dual-class equities: price differen-tials and information flows, Emerging MarketQuarterly, 9, 4762.

    802 P. Chelley-Steeley and W. Qian

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    13:

    51 0

    9 O

    ctob

    er 2

    014