Managed exchange rates, dual listing, and foreign exchange exposure: the experience of Chinese banks around the financial crisis

Transcription

1 Maaged exchage rates, dual listig, ad foreig exchage exposure: the experiece of Chiese baks aroud the fiacial crisis Mi Ye* ad Elaie Hutso UCD Michael Smurfit Graduate Busiess School Uiversity College Dubl Blackrock, Couty Dubl Irelad. Tel: , Fax: , Published i Joural of the Asia Pacific Ecoomy ( (3, Key words: foreig exchage exposure, bakig sector, duallistig, fiacial crisis, Chia JEL Classificatio: F31, G01, G21 * Correspodig author 1 Electroic copy available at:

2 Maaged exchage rates, dual listig, ad foreig exchage exposure: the experiece of Chiese baks aroud the fiacial crisis Abstract Usig daily equity price data from July 2005 to November 2009, we fid that most of the 14 Chiese listed baks are highly exposed to the RMB/USD exchage rate. By breakig our data period ito two subperiods aroud the fiacial crisis, we fid that Chiese baks were eve more exposed i the postcrisis period, despite the fact that the remibi reverted to a de facto peg agaist the dollar i September This caot be explaied by direct foreig exchage exposure, ad we argue that Chia s baks are subject to substatial idirect exposure as a result of cocers about their loa books i the face of aticipated further appreciatio of the RMB. We also fid that the exchage rate sesitivities of the twi shares of duallisted Chiese baks (those listed i Chia ad Hog Kog are very differet ot oly i magitude but also i sig. We discuss two possible explaatios for this: ivestor setimet ad hot moey iflows ito Chia. 2 Electroic copy available at:

3 1. Itroductio Exchage rate risks icreased substatially after the collapse of the Bretto Woods system of fixed exchage rates i This triggered extesive academic iterest i firmlevel foreig exchage exposure, although most empirical studies have focussed o the exchage exposure of ofiacial firms (see review papers Muller ad Verschoor, 2006, ad Bartram ad Bodar, Followig the failure of several US baks i the 1970s due to heavy foreig exchage losses (Aharoy ad Swary, 1983, a limited umber of studies have examied exchage exposure i fiacial istitutios. 1 Evidece remais sparse, however, o the exposure of baks i developig ad emergig markets. There is also surprisigly little evidece o the exchage rate exposure of firms i coutries with oflexible exchage rate systems. If a currecy is pegged to aother say the dollar it floats agaist other tradig parters currecies, so oe would expect firms exchage exposure to arise as a result of the volatility of say the euro, poud or ye. I oe of the few studies examiig firmlevel exposure i pegged exchage rate systems, Parsley ad Popper (2006 foud that this was ideed the case for ofiacial firms i East ad Southeast Asia. More iterestigly, they also foud that i some coutries, firms were exposed to movemets i the US dollar to which the local currecies were pegged. While this may see couterituitive, it makes sese whe oe cosiders that currecy pegs are i geeral rather shortlived (Klei ad Shambaugh, I this paper, we examie the exposure of the 14 listed Chiese baks to movemets i the RMB/USD exchage rate, durig the period 21st July 2005 to the ed of October, These baks trade i Chia o the Shaghai or Shezhe stock exchages, ad six are duallisted tradig also i Hog Kog. They together accout for over 56 percet of total assets of the Chiese bakig sector as of ed September 2009, 2 ad they are large, fastgrowig ad icreasigly iteratioal i outlook. I 2009, Chia s three largest baks overtook their America ad Europea couterparts to become the largest commmerial baks i the world by market 1 Research o the market risks faced by fiacial istitutios has bee domiated by studies of iterest rate risk (Flaery ad James, 1984; Madura ad Zarruk, 1995; Hirtle, 1997; Elyasiai ad Maser, 1998; Fraser, Madura ad Weigad, Data from the Chia Bakig Regulatory Commissio ( 3

4 capitalisatio (Luo ad Yao, 2009, ad four of the world s 10 largest baks are ow Chiese (The Ecoomist, 10th July, Chia provides a particularly iterestig eviromet i which to study the issue of baks exchage rate exposure i maaged exchage rate regimes. Figure 1 plots the RMB/USD exchage rate from the 1st Jauary 1994 to the ed of October, O 21st July, 2005, the People s Bak of Chia aouced that it would abado the 10 yearold peg agaist the US dollar ad move to a de jure maaged float exchage rate system with referece to a basket of currecies. From July 2005 to mid2008, the RMB was allowed gradually to rise agaist the dollar. This appreciatio (amoutig to more tha 20 percet ceased i mid2008 whe the exchage rate reverted to a de facto dollar peg. We ca compare the exchage rate exposure of the Chiese baks durig two exchage rate regimes a maaged float, ad a de facto dollar peg. To facilitate this compariso we split our sample at 15th September, This break poit is used because it is the time at which the Chiese authorities ceased allowig the RMB to rise agaist the dollar, ad it is also (perhaps ot coicidetally whe the fiacial crisis dramatically worseed followig Lehma Brothers declaratio of bakruptcy. Our empirical work yields two particularly iterestig ad ovel fidigs. First, usig the capital markets approach with daily data, we fid that the Chiese baks were highly exposed to movemets i the RMB/USD exchage rate durig the full July 2005 to October 2009 period. Cotrary to expectatios, however, they were more exposed i the postcrisis de facto peg period tha i the de jure maaged float period. I explorig this surprisig fidig, we ask what is the mai source of foreig exchage exposure i the Chiese baks does it derive directly from foreig currecy assets, liabilities ad trasactios? Or is it idirect, arisig from the effect of exchage rate movemets o, for example, the baks competitive positio or o the ecoomic eviromet? We examie iformatio relatig to direct exchage exposure disclosed i the baks aual reports, ad show that reported exposure to foreig currecyrelated activities caot explai our fidigs o the baks exchage rate sesivity. The exchage rate risk must therefore be largely idirect, ad we suggest that it is primarily the cosequece of cocers about the baks loa quality i the face of the aticipated further appreciatio of the RMB. Give that most 4

5 busiesses i Chia have a strog export orietatio, a appreciatig RMB would egatively affect export performace, ad ivestors would have see a regime of a gradually appreciatig RMB as potetially compromisig the growth of the Chiese export sector. The reversio to the peg agaist the dollar i September 2008 occurred at a time whe Chiese exports fell dramatically as world trade cotracted. Rather tha amelioratig cocers about the robustess of Chiese exporters that a pegged exchage rate should have egedered, ivestors cocers about the Chiese export sector were perhaps exacerbated by ucertaity as to whe the authorities would revert to a regime of RMB appreciatio. Our secod ovel fidig is that for the duallisted Chiese baks those listed i Hog Kog as well as i Chia estimated exchage rate sesitivity has opposite sigs. The Hog Koglisted shares are egatively exposed to exchage rate movemets, implyig that a rise i the RMB is bad ews for shareholders, while the Chialisted shares exchage rate sesitivities are mostly positive. There is cosiderable empirical evidece about what is called the puzzle of the Chiese stock market that ulike for other duallisted stocks where the price i the local market is lower tha i the foreig market, the stock price of Chiese compaies teds to be higher i the local market tha the foreig market. We argue that the Hog Koglisted shares of the duallisted baks better reflect the ecoomic reality of idirect foreig exchage exposure; that is, a appreciatig RMB is ideed bad ews for Chiese baks. We advace two explaatios for the aomalous positive relatio betwee the baks Chialisted stock prices ad the value of the RMB: optimistic ivestor setimet amogst Chiese ivestors, ad hot moey flows ito Chia. The remaider of the paper is orgaised as follows. I sectio 2 we review the related literature, ad i sectio 3 we describe the methodology ad data. Sectio 4 presets our results, followed by a detailed discussio i sectio 5. I sectio 6 we summarise ad coclude. 2. Foreig exchage exposure, exchage rate regimes, ad cross listig: theory ad evidece The foreig exchage exposure of a firm is defied as the sesitivity of its ecoomic value to exchage rate chages (Hekma, Exchage exposure ca be see as 5

6 havig two parts: direct ad idirect. Firms are directly exposed if they have uhedged expected ad future foreig trasactios, assets abroad, or liabilities deomiated i foreig currecies. Idirect exposure (kow also as competitive or ecoomic exposure arises from the idirect effects of exchage rate chages o a firm s competitive positio ad o the ecoomic eviromet. Compared to direct exposure which is relatively straightforward to measure ad maage idirect exposure ca be difficult to detect ad teds to be more complex to maage. A plethora of studies i the iteratioal busiess ad fiace literatures have explored firmlevel foreig exchage exposure, with most focusig o ofiacial firms i developed coutries with floatig exchage rate regimes (two recet largesample iteratioal studies are Doidge, Griffi ad Williamso, 2006, ad Hutso ad Steveso, I cotrast, research o the exchage exposure of fiacial istitutios, ad particularly o fiacial istitutios i emergig markets ad i coutries with oflexible exchage rate regimes, is sparse. 2.1 Exposure uder differet exchage rate regimes Sice the fiacial crises of the 1990s Mexico i 1994, Southeast Asia i 1997, Russia i 1998, Brazil i 1999, ad Turkey ad Argetia i 2001 a umber of researchers have aalysed the implicatios of the exchage rate regime for fiacial stability. Comparig currecy boards, fixed exchage rate ad floatig rate systems, Chag ad Velasco (2000 provided a detailed accout of the possible iteractios betwee bak fragility ad the exchage rate ad moetary regimes. They coclude that fiacial fragility is particularly evidet uder fixed exchage rate regimes. This is attributed to a implicit exchage rate guaratee grated by the govermet, egederig moral hazard (Eichegree ad Hausma, 1999, as fiacial istitutios do ot fully iteralise their exchage rate risks ad are more likely to egage i balace sheet mismatches tha uder floatig rate regimes (Burside, Eichebaum, ad Rebelo, 2001; Scheider ad Torell, They will also be less icetivised to actively hedge exchage rate exposure (Burside, Eichebaum ad Rebelo, Give the cocers about fixed exchage rate arragemets ad fiacial fragility, it is surprisig how little research has bee coducted o the foreig exchage exposure of baks i coutries with fixed or pegged regimes. Burside, Eichebaum ad 6

7 Rebelo (1999 argued that this is due to data limitatios, ad their discussios remai primarily at the theoretical level. I a study examiig the exchage exposure of ofiacial firms (to which the moral hazard problem would equally apply, Parsley ad Popper (2006 examied firms i East ad Southeast Asia. They showed that may firms were sigificatly exposed to chages i the dollar s value durig periods whe these coutries s currecies were pegged to the dollar, ad i Malaysia, the Philippies ad Thailad more firms had statistically sigificat exposure to the dollar with a peg tha without oe. 2.2 Foreig exchage exposure i the bakig sector Studies of exposure i fiacial istitutios i developed coutries have i geeral foud that baks ted to be highly exposed to exchage rate movemets, although the fidigs vary cosiderably betwee studies. I examiig the 48 largest US baks over the period 1975 to 1987, Cho Elyasiai ad Kopecky (1992 foud that oly 20 were sigificatly exposed to exchage rate movemets, but usig a loger time horizo (1975 to 1992 Choi ad Elyasiai (1997 foud that a much higher proportio 83 percet of large US bak holdig compaies were exposed. I comparig the exchage rate sesitivities of US ad Japaese baks returs durig 1986 to 1993, Chamberla Howe ad Popper (1997 reported that 30 percet of US baks were sigificatly exposed, compared to oly 9 percet of Japaese baks. Extedig the sample to iclude more ous baks, Marti (2000 foud that UK, Swiss ad Japaese bak portfolios were sigificatly exposed, whereas US bak portfolios were ot exposed. A few studies have looked at bak exchage exposure i emergig markets. Hahm (2004 ivestigated the exposures of Korea baks i the 1990s whe Korea discarded a multiple currecy basket peggig system ad adopted a semifloatig market average exchage rate system. Hahm foud that oe of his sample baks were exposed to exchage rates before 1994, but 38 percet were sigificatly exposed afterward. He suggested that both direct ad idirect exposures of fiacial istitutios could be substatial i emergig market coutries, sice the quality of fiacial supervisio ad risk maagemet are ot as good as i developed coutries. Wog, Wog ad Leug (2009 examied exchage exposure i Chiese baks ad foud that almost half of Chiese listed baks were sigificatly exposed to chages 7

8 i the USD/RMB exchage rate. I estimatig their exchage exposures, however, Wog, Wog ad Leug implicitly assumed that the duallisted twi shares of baks tradig o both the Chiese mailad ad Hog Kog markets have the same sesitivity to exchage rate chages. As we discuss i the ext sectio, this assumptio may be iappropriate. 2.3 Foreig exchage exposure of dual/crosslisted firms The commolyused capital markets approach to estimatig a firm s foreig exchage exposure has a critical uderlyig assumptio: equity prices fully ad ubiasedly reflect ew iformatio about firm value. May firms aroud the world list i oe or more foreig markets as well as i their home market, ad this has provided a good settig for examiig issues i stock market efficiecy. A questio that has received cosiderable attetio i the efficiecy literature is whether crosslisted shares reflect authetic firm value to the same extet i the two or more markets i which they are listed? If markets are efficiet, stocks of the same compay that have idetical expected cash flow ad risk characteristics should trade at the same price. For crosslisted stocks, however, stock prices i each market may differ due to the existece of owership restrictios, differeces i iformatio asymmetry, market liquidity, elasticity of demad ad risk prefereces, as well as legal, cultural ad laguage differeces. This market segmetatio pheomeo has bee cofirmed i several studies, icludig Bodurtha, Kim ad Lee (1995, Froot ad Dabora (1999 ad Kim, Szakmary ad Mathur (2000. I geeral, studies focusig o listigs i developed markets fid that the price of the stock issued i the foreig market is geerally higher tha the price of the stock i the home market the overseas share premium pheomeo (Eu ad Jaakiramaa, 1986; Hietala, 1989; Stulz ad Wasserfalle, Similar results have bee foud i studies of emergig market firms listig abroad, such as Thailad (Bailey ad Jagtia 1994 ad Mexico (Domowitz, Gle ad Madhava, I cotrast, Chiese Ashares, which are restricted to domestic ivestors, ted to trade at a higher price tha the shares i Chiese compaies that are available to foreig ivestors the the mai oes beig Hshares (listed i Hog Kog ad Bshares 8

9 (listed i Chia. 3 This pheomeo is kow as the puzzle of the Chiese stock market (Ma, 1996; Bailey, Chug ad Kag, 1999; Ferald ad Rogers, A umber of potetial explaatios have bee offered to explai the differece i price betwee A ad Hshares, such as differeces i market liquidity (Wag ad Jiag, 2004, limited ivestmet alteratives for local ivestors (Ferald ad Rogers, 2002; Lee ad Poo, 2005, differig risk attitudes betwee foreig ad domestic ivestors (Zhag ad Zhao, 2003 coupled with cocers over the higher coutry risk of Chia relative to Hog Kog (Wag, 2005, ad differeces i iformatio asymmetry betwee foreig ivestors ad domestic ivestors (Cha, Mekveld ad Yag, Some researchers have also argued that foreig exchage risk is a importat factor, as Hshares are deomiated i Hog Kog dollars while Chiese compaies assets are deomiated i RMB (Ferald ad Rogers, 2002; Wag ad Jiag, 2004; Wag, Implicit i these studies is that the Hshares are more likely tha the Ashares to reflect fudametal value. This was cofirmed by Zhao, Ma ad Liu (2005 who compared stock prices to estimates of value based o various valuatio models. They foud that the Hshare stock prices were ideed more closely correlated with fudametal value tha the Ashares, ad cocluded that the Hshare market was more efficiet. Give the apparet differeces i price discovery ad efficiecy i the Chiese versus Hog Kog stock markets, it is likely that the twi shares i duallisted stocks have differet sesitivities to exchage rate movemets. Whe estimatig foreig exchage exposure for crosslisted firms, assumig away these differeces may result i misspecificatio ad biased estimatio results. 3 Bshares were available oly to foreig ivestors util February 2001, whe the rules were chaged to allow Chiese citizes to trade them if they have the required foreig currecy US dollars for Shaghai Bshares ad Hog Kog dollars for Shezhe Bshares. 4 Wag (2005 ad Lee ad Poo (2005 suggest that with the icreasig itegratio betwee the Chiese market ad the Hog Kog market, ad the deregulatio of overseas ivestmets i Chia, the Ashare ad Hshares are o a course of price covergece. Nevertheless, the price discout pheomeo persists. Of the 61 duallisted compaies (those that have both A shares i the Chia mailad market ad H shares i the Hog Kog market at the ed of 2009, 45 have a A versus H share price spread of more tha 20 percet. 9

10 3. Methodology ad data 3.1 The empirical specificatio Followig prior researchers such as Cho Elyasiai ad Kopecky (1992, Wetmore ad Brick (1994, Choi ad Elyasiai (1997 ad Marti (2000, we use the threefactor model, icorporatig market retur, iterest rate ad exchage rate risk factors: I Rm, t RFt It t t m R, t RFt, [1] where R, t ad RFt are the holdig period rate of retur of the th bak stock from t 1 to t ad the riskfree iterest rate at time t, respectively, ad 10 R RF m, t t is the excess rate of retur of the market portfolio. The other two risk factors, I t ad t, represet the rate of chage i the yield of a riskfree bod from t 1 to t, ad that of the exchage rate, respectively. risks ad measuremet errors., t is a risk compoet for the th bak related to other Followig Wog, Wog ad Leug (2009, we exted this threefactor model as follows. As idicated by Alexader, Eu ad Jaakiramaa (1987, the expected retur of a duallisted firm depeds ot oly o the retur o the domestic market portfolio but also o the retur o the foreig market portfolio. We therefore icorporate both the Chiese market retur ad the Hog Kog market retur i the model for the duallisted baks. Further, as argued by Bartov ad Bodar (1994, there may be a lagged relatio betwee exchage rate chages ad firm value (the lagged respose hypothesis, because of the possible existece of mispricig (Roulstoe, 1999; Hodder, Kooce ad McAally, The lagged effects of the exchage rate are icluded i our estimatio model. We make three importat modificatios to Wog, Wog ad Leug s (2009 model. First, their specificatio for the foreig exchage exposure of the th Chiese bak assumes that the A ad Hshares have the same sesitivity to exchage rate chages. However, as we showed i sectio 3.3, there are substatial differeces betwee the prices of A ad Hshares. We use separate equatios to estimate the exchage rate sesitivity of A ad Hshares. Secod, Wog, Wog ad Leug (2009 used the logterm riskfree iterest rate to calculate the excess rate of retur o the baks shares ad the market portfolio, ad also as the iterest rate risk term. We suggest

11 that these two rates should be differet; a shortterm riskfree rate should be used to calculate the excess returs (Choi ad Elyasia 1997; Parsley ad Popper, 2006, ad a logterm bod rate should be used to cotrol for exposure to iterest rate movemets, because Sog (1994 ad Kae ad Ual (1988 foud that the stocks of commercial baks are more sesitive to chages i logterm tha shortterm iterest rates. Third, Wog, Wog ad Leug (2009 iclude a secod exchage rate term the chage i the RMB exchage rate agaist the Hog Kog dollar. Sice the Hog Kog dollar is fixed to the US dollar i a currecy board arragemet, the RMB/USD exchage rate is highly correlated with the RMB/HKD. Usig both exchage rates i the regressio equatio, therefore, would lead to multicolliearity problems. To icorporate these cosideratios, for duallisted Chiese baks we set up two separate equatios to measure the exchage rate exposure of A ad Hshares. For Ashares: R j A, HK RCH, t RFCH, t RHK, t RFHK, t A A A, CH A,, t RFCH, t j, j 0 t j [2] A, I I CH, t, t For Hshares: R j H, HK RCH, t RFCH, t RHK, t RFHK, t H H H, CH H,, t RFHK, t j, j0 t j [3] H, I I HK, t, t where R, ad A t H R, t are, respectively, the holdig period rate of retur of the th bak stock shares, Ashares (i terms of the remibi ad Hshares (i terms of the Hog Kog dollar, from t 1 to t, ad RF CH, t ad RF HK t, are the shortterm riskfree iterest rate of the Chia local market ad Hog Kog market respectively. Hece, R CH t RFCH, t, is the excess retur of the Chiese market portfolio ad R HK, t RFHK, t is the excess retur of the Hog Kog market portfolio. A, CH market sesitivities of the excess returs of the Chiese bak s Ashares to the excess returs of the Chiese market portfolio ad that of the Hog Kog market portfolio, 11 ad A, HK are the

12 respectively, while H, CH ad H, HK are the market sesitivities of the excess returs of a Chiese bak s Hshares to the excess returs of the Chiese market portfolio ad that of the Hog Kog market portfolio, respectively. We iclude lagged values of the exchage rate chage; t j is defied as the percetage chage of the RMB exchage rate agaist US dollar at the j lag period. Therefore, for the th Chiese bak, the sesitivity of Ashare ad Hshare stock price to foreig exchage rate A, chages are defied as j, j j 0 H ad j, j j 0, respectively. To estimate the iterest rate A, I sesitivity of Ashares ( ad Hshares ( H, I, we iclude the rate of chage i the yield of logterm riskfree bod, I CH, t ad HK t I,, for the Chia local market ad the Hog Kog market, respectively. For baks that are listed oly i Chia, the followig empirical specificatio is adopted: j I RCH, t RFCH, t j, t j ICH, T t CH R, t RFCH, t, j0 [4] 3.2 Data, estimatio method ad summary statistics There are 14 listed Chiese baks, of which 6 5 are duallisted i the Chiese ad Hog Kog stock markets, ad 8 are listed oly i the Chiese mailad market (Shaghai Stock Exchage or Shezhe Stock Exchage. Our data set cotais all 14 listed baks. As suggested by Chamberla Howe ad Popper (1997, movig from mothly data (the data iterval that is most commoly used i the exchage exposure literature to daily data makes it easier to pick up firmlevel exchage exposure. Muller ad Verschoor (2006 poited out that oe might reasoably expect stroger evidece i support of the iclusio of lagged variables whe workig with higher frequecy observatios. For these reasos we use daily data ad we test systematically for the appropriate lag legth. 6 Our data period begis o 21st July 5 Chia Misheg Bak (MSB wet public i Hog Kog o November 26, 2009, ad became the seveth duallisted bak. However, we treat MSB as a locally listed bak because of the short time period sice it issued Hshares. 6 We also remove for each bak the outliers i the daily equity price data, which may arise from sudde chages i market setimet or some special evets of the baks (such as sharp rises i prices i the 12

13 2005, which was whe the Chiese authorities chaged the exchage rate regime to a de jure maaged float from the peg agaist the US dollar, to the ed of October The data start time is later for some baks due to their differet dates of iitial public offerig. The Chiese market portfolio is proxied by the Hu She 300 Idex, which is a weighted idex compiled by Chia Securities Idex Co., Ltd, ad for the Hog Kog market we use the Hag Seg Idex. The shortterm riskfree iterest rate i Chia, RF CH, t, is proxied by the 1year yield o Chiese govermet bods, 7 ad for Hog Kog the shortterm riskfree iterest rate, Iterbak Offered Rate (HIBOR. 8 RF,, is the 6moth Hog Kog HK t We use the fiveyear Chiese govermet bod yield ad the fiveyear yield o Exchage Fud Notes to proxy the logterm riskfree iterest rate i Chia ( I CH, t ad Hog Kog ( I HK, t, respectively. 9 The daily rate of chage of the USD/RMB exchage rate, t, is calculated from the correspodig RMB spot rates. Our data are draw from Bloomberg ad the Hog Kog Moetary Authority. The Ordiary Least Squares (OLS method is employed to estimate foreig exchage sesitivity ad other risk parameters cotaied i Eqs. [2] ad [3] for each duallisted Chiese bak, ad i Eq. [4] for the locally listed Chiese baks. As our estimatio models for the duallisted baks icorporate the returs of both the Chiese mailad portfolio ad Hog Kog market portfolio, multicolliearity may be a problem. However, we fid o sigificat multicolliearity betwee these two market variables. 10 Lastly, we use Almo s (1965 Polyomial Distributed Lag (PDL model first tradig day after IPO, to avoid biased results caused by these outliers. As such, observatios with a excess daily retur lower tha the 1st percetile or higher tha the 99th percetile of the data for each bak are excluded from the sample. 7 We also used alterative riskfree rate proxies: the Chiese iterbak offered rate ad the 3moth ad 1year savigs deposit rates. The results are little chaged whe usig these alterative proxies. 8 We also used other proxies for the HK shortterm riskfree iterest rate: 3moth ad 9moth HIBOR, ad the 1year exchage fud ote rate, but the results remai similar. 9 We also cosidered Chiese govermet bods with differet maturity (10year ad 20year. However, the data for 20year Chiese govermet bods are ot available for our sample period. The result of usig 10year Chiese govermet bod remai similar to that of usig 5year oe. The same holds for. 10 We first calculate the correlatio coeffiet betwee the excess retur of Chia market ad Hog Kog market, which is The, to be more rigorous, we follow the method proposed by Klei ad Nakamura (1962 ad fid that there is o sigificat multicolliearity betwee R CH, t RF ad CH, t 13

14 to avoid multicolliearity problems amog the cotemporaeous exchage rate ad its lagged values. I order to obtai the optimal lag legth, we first ra all possible regressios that use all combiatios of the regressors. Amog the estimated regressio models we foud the optimal lag legth, usig the Akaike (1973 iformatio criterio (AIC, Schwarz (1978 ifomatio criterio (SIC, ad Haa Qui (1979 criterio (HQC, is 6 days for most of our baks, although for some the optimal lag legth is betweee 3 ad 5. As poited out by Cho Elyasiai ad Kopecky (1992, Bartov ad Bodar (1994 ad Atidehou ad Gueyie (2001, the exchage rate exposure parameters may ot be stable durig the whole oberservatio period. To test for the stability of the exposure estimates, as well as to examie the effect of the fiacial crisis ad reversio to a de facto peg agaist the dollar o Chiese baks foreig exchage exposure, we coduct subperiod estimatios by dividig our sample period ito two subperiods at September 15, Ashares ad Hshares Ashares ad the correspodig Hshares are associated with the same rights ad obligatios, ad therefore i theory should have the same price. However, cosistet with the fidigs of the studies of duallisted Chiese shares discussed i sectio 2.3, the two classes of shares trade at substatially differet prices. This ca be see i Figure 2, which depicts daily stock prices for the 6 duallisted Chiese baks. Except for Chia Merchats Bak (CMB, the Ashare stock price is geerally higher tha the correspodig Hshare price. Although a covergece of A ad Hshare stock prices toward the ed of the observatio period is apparet for some of these baks (ICBC, CCB ad BOCs, the AH share premium pheomeo persists durig most of the observatio period. Table 1 presets summary statistics of the twi shares for the 6 duallisted baks: stock returs, tradig volumes ad turover. It is clear that the stock returs of A ad their correspodig Hshares are differet ot oly i magitude but also i sig. Further, the raw tradig volumes of Hshares are much larger tha that of the correspodig Ashares for most of the baks, except CMB. This is due to the fact R HK, t RF. I fact, the multicolliearities amog other idepedet variables are examied usig HK, t the same method, ad o sigificat multicolliearity ca be foud either. 14

15 that there are more shares available i the Hog Kog market tha i the Chia mailad market. We create the variable turover by scalig tradig volume by the umber of shares available to trade i each market; mea ad media turover appears i the two righthad colums i Table 1. The turover of the Ashares is much higher tha that of their couterpart Hshares, which is idicative of greater speculative activity i the Chia mailad markets tha i the Hog Kog market. 4. Results Our fullperiod results ca be foud i Table 2. Pael A presets the results for the duallisted baks (equatios [2] ad [3] ad Pael B for the locallisted baks (equatio [4]. The coefficiet o the Chiese market term R CH t RFCH, t, is strogly sigificat for all of the Ashares ad for the locallisted shares, which largely explais the rather high Rsquareds. The fact that the retur o the Chiese market is ot sigificat for most of the Hshares, whereas the retur o the Hog Kog Market R HK t RFHK, t, is highly sigificat, supports the prior literature (discussed i sectio 2.3 that the pricig process for crosslisted Chiese shares diverges cosiderably betwee the two markets. The last row of the two paels presets the term j,, which summarises overall foreig exchage exposure. This is ot reported for baks with o sigificat exchage rate sesitivity, cotemporaeous or lagged. We fid that most of the Chiese baks both duallisted ad locallisted are sigificatly exposed to exchage rate movemets. Four out of 6 Ashares ad 5 of 6 Hshares are sigificatly sesitive to movemets i the RMB/USD exchage rate respectively; ad for the locallisted shares, this figure is 6 out of 8. The directio of this relatio is positive for the Ashares, but for the Hshares it is egative. A appreciatio of the RMB agaist the US dollar therefore looks like good ews for bak shareholders whe exposure is estimated usig Ashares (ad this also holds i most cases for local bak shares, ad bad ews whe usig the Hshares returs. Cosistet with the lagged respose hypothesis (Amihud, 1994; Bartov ad Bodar, 1994; Walsh, 1994, we fid that several lags of the exchage rate chage are sigificat for may of the baks. The patters of the lag effects differ cosiderably j j0 15

16 betwee the A ad the Hshares. For the Ashares, the cotemporaeous exchage rate coefficiets are either egative or isigificatly positive. By day +3 or +4, they have switched to positive. A similar although weaker patter ca be foud i the localoly listed shares. I cotrast, for most of the Hshares the lag effects die out more quickly ad the sigs remai egative. As the duallisted Ashares ad the 8 locallisted shares are traded i the Chiese mailad market (Shaghai Stock Exchage or Shezhe Stock Exchage while the duallisted Hshares are traded i Hog Kog market, there is clearly somethig uusual goig o i the Hog Kog or Chiese markets that results i baks local shareholders seeig Chiese baks exposed i a opposite way to the same or similar shares tradig i Hog Kog. We discuss this issue i more detail i sectio 5.2. Our subperiod fidigs are preseted i Table 3. Paels A ad B cotai the results for the duallisted baks Ashares ad Hshares respectively, ad Pael C presets the fidigs for the localoly listed baks. We also provide a tabulated summary i Table 4 of our precrisis versus postcrisis fidigs, with Pael A summarisig the sigs o the exchage exposure coefficiets, Pael B the chages i sig of the timevaryig exchage rate effects, ad Pael C the average magitude (absolute value of the exposure coefficiets. The importat fidigs from this aalysis ca be summarised as follows. First, it is clear that there was a substatial chage i bak stocks reactio to exchage rate movemets at the time of the fiacial crisis. The exposure coefficiets for most of the Chialisted bak shares switch from positive before the crisis to egative postcrisis. For BOC, for example, the precrisis exchage exposure coefficiet o its Ashare is 0.79, ad i the postcrisis period it switches to This switchig patter is ot apparet i the Hshare results. As summarised i Pael A of Table 4, for the Ashares, all 6 are positive precrisis, ad postcrisis 4/6 are egative; for the locallisted baks the equivalet proportios are 4/8 ad 6/8. Secod, the precrisis timevaryig effects of exchage rate chages for the Chialisted baks evolve from egative cotemporaeously ad i the early lags (the first oe to three days after a chage i the exchage rate to highly sigificatly positive at lags four to six. This is completely opposite to the postcrisis patter i lags where the effects are strogly egative at lags 4 to 6. To agai use BOC as a example, before the crisis the cotemporaeous ad the first two lag coefficiets for BOC are egative, ad they switch to positive at the 3day lag, icreasig to highly 16

17 sigificatly positive at lags 4 ad 5. Postcrisis, a isigificatly positive cotemporaeous reactio turs egative at the first lag, ad becomes highly sigificatly egative at lags 3 to 5. I cotrast, these patters are ot apparet i the Hshare lags. This switchig behaviour i the lags is summarised i Pael B of Table 4. Third, the exchage rate sesitivity of the Chiese baks icreased dramatically after September This is a particularly iterestig fidig give that durig the postcrisis period the RMB was de facto pegged to the dollar. Exchage rate sesitivity icreased for 4 of the 6 Ashares, 4 of the 6 Hshares, ad 5 of the 8 locallisted shares. Average magitudes, as measured by the absolute value of overall exposure, j j,, j 0 ca be foud i Pael C of Table 4. The most dramatic icrease i exposure is see i the Hshares stock prices the average coefficiet icreased a absolute 175 percet. For the Ashares the icrease is 29 percet ad for the localoly listed baks, 70 percet. 5. Discussio of the fidigs Mailad Chiese stock ivestors seem to see a appreciatio of the RMB as good ews for baks, whereas ivestors i Chiese bak Hshares see a RMB appreciatio as bad ews. Clearly both of these reactios caot be correct i the sese that i efficiet markets, stock prices reflect ews i a appropriate ad ubiased maer. I this sectio we argue that the exchage rate sesitivity of the Hshares is much more likely to reflect the true fudametal foreig exchage exposure of Chiese baks tha the Chialisted shares. I explaiig this we discuss two issues. First, i what directio would Chiese baks theoretically be exposed? I addressig this questio, we discuss whether direct or idirect exchage rate exposure domiates. Secod, why is Ashare exchage rate sesitivity so differet from its Hshare equivalet? 5.1 Is the estimated exposure direct or idirect? Oe of the virtues of the capital markets approach to estimatig firmlevel foreig exchage exposure is that it captures overall (uhedged exposure both direct ad idirect. As we discussed at the begiig of sectio 2, direct exposure affects baks via chagig values of foreig assets ad liabilities, ad as a result of foreig currecy trasactios. To what extet does the baks exchage rate sesitivity reflect direct 17

18 exposure? There are too few Chiese baks to aalyse this issue usig crosssectioal ecoometric techiques (as i Choi ad Elyasia 1997, ad Au Yog, Faff ad Chalmers, However, i order to shed some light o the extet of the Chiese baks direct exposure, we examie iformatio disclosed i their aual reports. We collect data o two importat elemets of the baks iteratioal activities: first, the obalacesheet et foreig currecy positio, which was used by Chamberla Howe ad Popper (1997 to measure balacesheet foreig currecy exposure; ad secod, reported exchage rate gais ad losses 11, which costitutes the best available iformatio o direct exposure. Available data for each bak for the period 2005 to 2009 are preseted i Tables 5 ad Table 5 presets the data o the baks obalacesheet et foreig currecy positio (calculated as foreig assets mius foreig liabilities, scaled by assets. As these figures are i geeral positive, it is clear that most of the baks had a et log positio i foreig currecy durig the period. They were therefore exposed to the RMB iversely that is, the value of their et foreig currecy positio would have falle whe the RMB appreciated. This is cofirmed by the exchage rate sesitivity of et profit; reproduced i Table 6 are the baks ow estimates of the effect o et profit of a icrease i the value of the RMB. Most of these are egative, implyig that a appreciatio of the RMB would have had a egative impact o their et profit durig most of the years from 2005 to The obalacesheet et foreig currecy positio data provide oly a partial perspective o direct exchage exposure. The baks also report foreig exchage gais ad losses, ad these (scaled by total assets ca be foud i Table 7. I most cases, these figures are positive the Chiese baks gaied from foreig exchage trasactios i the period 2005 to Give that their balace sheet exposure suggests that the baks would have suffered losses durig this period of strog RMB appreciatio, these reported foreig currecy gais imply that the balace sheet losses were more tha offset by other foreig currecy activities. BOC ad ICBC provide a additioal set of figures that help illustrate this. These baks separate out balace sheet gais ad losses from overall foreig currecy gais ad losses, ad this data 11 A formal defiitio of exchage rate gais ad losses ca be foud i the otes to Table For bak of Chia (BOC, disclosed iformatio icludes the exposures to the US dollar as well as the euro, the ye ad the poud; for the others, however, data are available oly for US dollar exposure. We report oly US dollar exposures. 18

19 appear i the shaded rows of Table 7. Cosistet with the et log foreig currecy positios show i Table 5 ad the fact that this was a period of strog RMB appreciatio, these baks show foreig currecy losses o balace sheet items, ad yet overall the baks experiece either smaller losses or gais. The balace sheets et log positio i foreig currecy must therefore have bee offset by either hedgig activities or profits from proprietory tradig (or both, because durig this period of RMB appreciatio most of the baks made exchage rate gais implyig a positive relatio betwee the currecy s value ad the baks stock price. I our regressio aalysis we foud positive exchage rate sesitivity for the Ashares ad egative sesitivity for the Hshares; our fiacial statemet data aalysis of direct exposure therefore appears to suggest that the Ashares better reflect ecoomic reality tha the Hshares. There are two strog argumets agaist this coclusio. First, it is ot cosistet with evidece from the literature o duallisted Chiese firms that Hshares are more likely to accurately ad ubiasedly reflect ew iformatio tha the equivalet Ashare (Ma, 1996; Zhao, Ma ad Liu, Secod, as ca be see i Table 7, exchage rate losses ad gais are tiy whe scaled by bak size. Our strog fidigs o the baks exchage rate sesitivity caot be explaied by such small foreig exchage gais. Further, i breakig our observatio period ito two subperiods at 15th September 2008, we fid that the magitude of the estimated exchage exposure for most of the baks icreased dramatically postcrisis. The exchage rate gais ad losses data, however, show that direct foreig exchage exposure did ot icrease substatially after September 2008, ad it actually decreased for some baks. As the icrease i exchage rate sesitivity that we fid i our empirical aalysis caot be explaied by the fiacial statemet data, it must be due to a rise i idirect exposure Idirect effects The extet of a bak s idirect exposure will deped o such factors as the extet of foreig competitio, demad for loas, ad other bakig coditios (Merikas, 1999; Chamberla Howe ad Popper, Idirect exchage exposure might also arise through the effect of exchage rate chages o the quality of local currecydeomiated loas if adverse exchage rate movemets are perceived to affect the 19

20 ability of borrowers to repay (Wog, Wog ad Leug, For the Chiese baks, most ledig activity is to compaies, particularly those that are exportorieted. Chia s baks are therefore idirectly exposed to the competitiveess ad profitability of the maufactured exports sector a sector that is highly sesitive to a risig local currecy. The theoretical relatio is therefore iverse; a rise i the value of the RMB should lead to a decrease i the baks stock prices. I the precrisis period durig which the RMB was allowed gradually to rise agaist the dollar, ivestors would have bee cocered about the effect of a risig RMB o the volume ad pricig of exports ad the kocko effects to the baks cliet compaies. Further appreciatio of the RMB may have bee perceived to precipitate a rise i the probability of firm failure, leadig to a icrease i expected bad loas ad loa loss provisios for the baks. Ivestors may also have bee cocered about the impact of appreciatio of the RMB o the Chiese ecoomy i geeral; a slowig ecoomy would have see a reductio i the speed of growth of future ledig. But why would the foreig exchage exposure of the Chiese baks have icreased so dramatically after September 2008? The fiacial crisis ad its aftermath saw lower cosumer cofidece ad recessio i developed coutries, which for Chia resulted i a dramatic reductio i exports. Betwee the peak i September 2008 ad February 2009, exports had declied by $64.89 billio a sharp pluge more tha 50 percet withi 5 moths. 13 Some iteratioal commetators suggested that Chia was facig its greatest crisis sice its ecoomic reforms bega more tha 30 years ago. 14 This declie, together with cocers about the shorttomediumterm outlook for exports, is probably why the Chiese moetary authorities were reluctat to let the yua move more freely agaist the dollar 15 ad reverted to the dollar peg after September However, while the peg may have kept the dollar prices of Chia s exports steady durig this period, it led to greater ucertaity i foreig exchage markets about whe the Chiese authorities would allow the RMB to start risig aga ad by how 13 Data from the GTA databse (Chia ad Geeral Admiistratio of Customs of the People s Republic of Chia, 14 See, for example, Geoff Dyer, How the fiacial crisis is chagig Chia, Fiacial Times, Oct 23, See Fiace Ad Ecoomics: A Mao i every pocket; The yua goes global, The Ecoomist, September 25, See Chia s currecy glasost has its limits, Wall Street Joural, July 27,

21 much. This ucertaity would have bee heighteed by escalatig tesios betwee Chia ad the US o the value of the RMB. This may well be reflected i a icrease i sesitivity to exchage rate chages amogst Chiese bak stocks, ad it provides a potetial explaatio for our fidigs that the foreig exchage sesitivity of Chiese bak stocks icreased after September There is some evidece that key foreig ivestors had lost cofidece i Chiese baks due to cocers about the possibility of a escalatio of bad debts. I December 2008, Swiss bak UBS sold all of its 3.4 billio BOC Hshares. The foudatio ru by Asia s richest ma, Hog Kogbased Li KaShig, sold 2 billio BOC shares i Jauary 2009, ad aroud the same time Bak of America sold 5.6 billio shares of CCB. Royal Bak of Scotlad has also retrieved its US$2.4 billio ivested i BOC. 17 For ICBC, some shares held by Alliaz ad America Express were sold whe lockup periods 18 o the stocks eded o April 28, While some of these sales may have bee exacerbated by the fiacial distress of the ivestor istitutios, the wave of sellig may still have cotributed to egative setimet about Chiese baks amogst ivestors aroud the world. 5.2 Why is the reactio i duallisted baks so differet i Chia ad Hog Kog? It is clear from the above discussio that the exchage rate sesitivity that we have foud amogst Chiese baks must be due to idirect rather tha direct foreig exchage exposure. As the theoretical directio of exposure is egative, cosistet with the literature o efficiecy i Chiese duallisted stocks (Ma, 1996; Zhao, Ma ad Liu, 2005; Wag, 2005, the Hshare stock price better reflects the ecoomic reality of exchage exposure tha the Ashare price. Why does the Ashare stock price react to exchage rate movemets i the opposite directio? We discuss two possible (omutually exclusive explaatios for this fidig: first, setimet amogst Chiese ivestors, ad secod, hot moey flows ito Chia Ivestor setimet 17 See Pyo MiCha, Chia: Aimig for 8% growth despite the fiacial crisis, SERI Quarterly, Seoul. Apr The Chiese govermet prohibits sellig of stocks held by the cotrollig shareholders for a give period after a compay goes public. 21

22 Our fidig that most locallisted Chiese bak shares are positively exposed to exchage rate movemets may be iterpreted from the perspective of ivestor setimet a otio itroduced by Lee, Shleifer ad Thaler (1991. Bailey (1994 suggested that Chiese domestic ivestors ted to be uduly optimistic ad that this may drive the overpricig of Ashares. Ma (1996 foud that a importat explaatio for elevated Ashare prices is the highly speculative behaviour of Chiese ivestors. Yao, Luo ad Morga (2010 argue that because of limited iformatio disclosure by Chiese listed compaies ad the lack of professioal ivestmet kowledge, combied with euphoria associated with Beijig s 2008 Olympic Games, Chiese ivestors buy shares radomly regardless of the performace of the compay. As we saw i Table 1, stocks o the Chiese mailad markets are much more actively traded tha i Hog Kog, cosistet with the otio that there is cosiderably more speculative activity i Chiese bak stocks i the Chiese markets visávis the Hog Kog Stock Exchage. Takig BOC as a example, its media daily turover i Hog Kog is 0.43 percet compared to a media of 1.53 percet i the Chia mailad market. Our subperiod aalysis sheds some light o ivestor setimet. Before September 2008, almost all of the local shares were positively exposed to exchage rate chages, ad the Hshares exposure coefficiets were i geeral egative. The situatio chaged dramatically after September 2008, whe the exchage rate sesitivity of the Chialisted shares switched to egative, while there was little chage i the patter of exchage exposure for the Hshares. It was clear by September 2008 that the fiacial crisis would precipitate a severe worldwide recessio. Although Chia s bakig sector suffered much less tha baks i other coutries, it is possible that eve Chia s optimistic ivestors became ervous about effects of the aticipated reversal of world ecoomic growth o demad for Chia s goods. As discussed i sectio 5.1, ivestors may have bee particularly cocered about the effects o Chia s massive export sector of further appreciatios of the RMB Hot Moey Aother potetial explaatio relates to hot moey iflows ito Chia from foreig speculators. I the period betwee July 2005 ad September 2008, speculators would have see Chiese moetary authorities allowig the RMB gradually to rise as a sure 22

23 bet profit opportuity 19, ad cosequetly substatial sums of speculative moey flowed ito Chia. Give that Chia s system of capital cotrols makes such speculative activity essetially illegal, there are o offical figures o these flows. However, evidece o the extet of hot moey flow ca be gleaed from data o foreig exchage reserves. Figure 3 plots the quarterly chage i Chia s foreig reserves betwee the secod quarter i 2005 ad the third quarter i Usig various sources of data (as detailed i the otes to Figure 3, we have estimated the proportio of the icrease i reserves i each quarter that ca be attributed to official chaels such as the trade surplus ad foreig direct ivestmet; this is the blue (lower part of the bar. The remaider (i red is uexplaied officially. For most of the period from quarter 2 i 2005 to the secod quarter i 2008 more tha half of the icrease i reserves is uexplaied by official currecy movemets, ad is probably the result of hot moey iflows from abroad. It must be oted that the hot moey pheomeo would ot occur i Hog Kog because the Hog Kog dollar is fixed to the US dollar i a logstadig currecy board arragemet. As the stock market is oe of the few possible destiatios for hot moey, 20 such iflows may ievitably lead to a overheated stock market (Tia ad Ma, 2010, ad push stock prices higher tha their fudametal value. The positive relatio betwee the Chiese baks stock returs ad the value of the RMB may simply be a artefact of this pheomeo; a icrease i the value of the RMB may have triggered greater flows of hot moey ito Chia as speculators had their priors cofirmed that the moetary authorities were still allowig the RMB to rise. The hot moey pheomeo also provides a good potetial explaatio for our fidig that the cotemporaeous reactio to exchage rate chages (ad the first ad secod lags for Chialisted baks i the preseptember 2008 period were egative, reflectig the geuie fears of ivestors that a stroger RMB would have a deleterious effect o Chia s export sector. The lags turig positive at lags 3 to 6 is cosistet with a rise i hot moey flows followig a appreciatio of the RMB. This reversed i the 19 See, for example, Joatha Cheg ad Adrew Batso, The Yua gais favor i Hog Kog, Wall Street Joural (Easter editio, Aug 6, 2008; Michael Pettis ad Loga Wright, Hot moey poses risks to Chia s stability, Fiacial Times, July 13, 2008; Geoff Dyer, Chia risks a scorchig as hot moey flows i at record level, Fiacial Times, Jul 4, Zheg Tuo, a fud maager at Fortis Haitog Ivestmet Maagemet Compay i Shagha says that hot moey ca be ivested i oe of oly two markets: the realestate market ad equity market. (James T. Areddy, Chiese tur to stocks; key idex tops 2000poit mark for the first time i five years, Wall Street Joural (Easter editio, Nov 21,

24 postseptember 2008 period; we foud little cotemporaeous reactio to a chage i the value of the RMB, ad i may cases the loger lags became strogly sigficatly egative. Hot moey flows ito Chia had by the reportedly slowed, 21 ad this is cofirmed by Figure 3 which shows a dramatic dropoff i the uexplaied portio of the chage i foreig reserves durig the last quarter of 2008 ad the first quarter of There are idicatios that hot moey flows have played a role i the rise ad fall of Chia s stock ad real estate markets, 22 ledig support to the hot moey explaatio for our fidig that the sig of the exchage exposure coefficiets for the Chialisted bak stocks switched after September Betwee the ed of 2005 ad the ed of 2007, the Shaghai Stock Exchage Composite Idex icreased more tha sixfold from 1,012 to 6,124. By 17th September 2008 it had falle to a low of 1,929 a drop of 70 per cet from its peak i less tha 10 moths 23. This suggests the possibility that hot moey flows had begu to declie before the Lehma crisis i September I Jue 2008 it was suggested that hot moey had bee redirected ito property. 24 Aother media report i July 2008 suggested that give the rapid declie i the equity market ad some sigs of a slowdow i the property market, the hot moey was beig redirected ito bak accouts 25, earig a iterest rate of over 4 percet o yua deposits compared with 2 percet o dollars. 6. Summary ad coclusios The fiacial crisis revealed the uderlyig fragility of the world bakig sector i the face of severe stress. How to maage the various risks ad to esure the stability of the fiacial system have become top priorities for all coutries. It is geerally held that Chiese baks suffer little foreig exchage exposure due to a almost ivolatile exchage rate uder the logterm dollar peg ad the curret de jure maaged float. Our research results suggest that this is ot the case. Chia s baks are highly 21 See for example, Geoff Dyer, Chia faces heavy capital outflows, Fiacial Times, Ja 11, 2009; ad Chia: risk of capital flight replaces hot moey cocers, Emergig Markets Moitor. Ja 19, See Michael F. Marti ad Waye M. Morriso, Chia s Hot Moey problems, CRS report for Cogress, July 21, Data from Bloomberg database. 24 Capital iflows to Chia: hot ad bothered, The Ecoomist, Ju 28, Loga Wright, Hot moey i Chia: where s it goig ad how s it troublig? Chia Stakes, July 14,

25 exposed to movemets i the value of the US dollar, despite the fact that the RMBdollar exchage rate is heavily cotrolled. This suggests that i the cotext of ecoomic globalisatio, ofreely floatig exchage rate regimes appear to do little to alleviate exposure (Parsely ad Popper, The switch to the de facto dollar peg i September 2008 saw a dramatic icrease i the Chiese baks exposure. As revealed by our aalysis of accoutig data relatig to foreig assets ad liabilities ad cash flow chages due to exchage rate movemets, this caot be explaied by greater direct foreig exchage exposure. We suggest that it is most likely to be explaied by a upsurge i idirect exposure as a result of the risig likelihood of loa losses as exporters struggled to survive i the face of the worldwide declie i demad triggered by the fiacial crisis (Bell ad Chao, We also fid that the reactio to exchage rate movemets of the A versus Hshares of the 6 duallisted baks differ cosiderably, i sig as well as i magitude. We have suggested two (omutually exclusive explaatios for this pheomeo: differeces i ivestor setimet that is, the optimism ad speculative behaviours of Chiese domestic ivestors, ad hot moey flows ito Chia. It is possible that these factors acted together to exaggerate the effect; the torret of hot moey may have icreased the atmosphere of speculatio i the Chiese domestic market, sharpeig the differeces betwee the Chiese ad Hog Kog markets. The Chiese govermet has show a determiatio to deter hot moey iflows, ad is cosiderig cotrols o yua forward trasactios to prevet ivestors from bettig o a risig yua. This may ivolve raisig the miimum requiremet for a bak s foreig exchage holdigs if short forward positios i yua exceed log; the implicatio beig that if forward yua sales to a bak s cliets exceed forward yua purchases, the bak is see as facilitatig bets o the yua s appreciatio. 26 The best ad simplest solutio to hot moey iflows i the loger term is full covertibility ad a free float of the yua. However, together with likely further fiacial liberalisatio ad the icreasig iteratioalisatio of Chia s baks, ew risks may arise i the bakig sector which will complicate foreig exchage risk maagemet 26 See Chia said to cosider cotrols o yua forward trasactios, Bloomberg News, Jue 8,

26 (Bell ad Chao, Improvig risk maagemet capabilities will therefore be a critical ad ogoig task facig the Chiese bakig sector. It is well uderstood that the capital markets approach to estimatig firm or bak foreig exchage exposure picks up ay exposure that is left after hedgig activities have bee udertake (see Bartram ad Bodar, 2007, for a review. I light of the suggestio by Hutso ad Steveso (2010, our fidigs of sigificat foreig exchage exposure costitutes evidece of iadequate hedgig amog Chiese baks. This is perhaps (at least i part due to the fact that derivative ad other markets for risk maagemet products are relatively udeveloped i Chia 27. The implicatio for the regulatory framework is clear; it is importat that baks are allowed greater access to ad freer use of risk maagemet tools. The mooted cotrols o forward cotract holdigs are a step i the wrog directio, although there is also some idicatio that foreig exchage optios are soo to be permitted. 28 Future research might explore i detail the extet ad quality of exchage rate risk maagemet activities i Chia s baks, i order to assess whether their hedgig activities are adequate ad efficiet. Our fidigs have implicatios for baks i other emergig atios ad those i coutries with pegged/maaged exchage rate regimes, especially for coutries with ope ecoomies. First, they should be aware that they may be similarly exposed. Secod, baks maagemet ad regulators should be aware that optimal foreig exchage risk maagemet practice ivolves keepig a watchful eye o idirect as well as direct exposure. 27 Chiese baks ca curretly hedge agaist the risk of exchage rate chages through either RMB trade settlemet or forward cotracts. ( Chia s regulator said to cosider allowig Yua optios tradig by baks, Bloomberg News, Dec 13, The State Admiistratio of Foreig Exchage (SAFE, Chia s currecy regulator, may i the ear future allow RMB optios tradig to help baks ad compaies hedge agaist foreig exchage risk ( Chia s regulator said to cosider allowig Yua optios tradig by baks, Bloomberg News, Dec 13,

27 Refereces Adler, M., ad B. Dumas, Exposure to Currecy Risk: Defiitio ad Measuremet. Fiacial Maagemet, 13, Adler, M., Simo, D., Exchage risk surprises i iteratioal portfolios. Joural of Portfolio Maagemet, 12(2, Aharoy, J., Swary, I., Cotagio effects of bak failures: Evidece from capital markets. The Joural of Busiess, 56(3, Akaike, H. (1973. Iformatio theory ad the extesio of the maximum likelihood priciple. I B. N. Petrov ad F. Csaki (Eds., Proceedigs of Secod Iteratioal Symposium o Iformatio Theory, Alexader, G. J., Eu, C. S., ad Jaakiramaa, S., Asset pricig ad dual listig o foreig capital markets: a ote. Joural of Fiace, 42(1, Allayais, G., Ofek, E., Exchage rate exposure, hedgig ad the use of foreig currecy derivatives. Joural of Iteratioal Moey ad Fiace, 20, Almo, S., The distributed lag betwee capital appropriatios ad expeditures, Ecoometrica, 33, Atidehou R.B., Gueyie, JP., Caadia charterd baks stock returs ad exchage rate risk. Maagemet Decisio, 39(4, Au Yog, H. H., Faff, R., Chalmers, K., Derivative activities ad AsiaPacific baks iterest rate ad exchage rate exposures. Joural of Iteratioal Fiacial Markets, Istitutios ad Moey, 19, Bailey, W., Risk ad retur o Chia s ew stock markets: some prelimiary evidece. PacificBasi Fiace Joural, 2, Bailey, W., Jagtia J., Foreig owership restrictios ad stock prices i the Thai capital market. Joural of Fiacial Ecoomics, 36, Bailey, W., Chug, Y. P., Kag, JK., Foreig owership restrictios ad equity price premiums: What drives the demad for crossborder ivestmets? Joural of Fiacial ad Quatitative Aalysis, 34,

28 Bartov, E., Bodar, G.M., Firm valuatio, earigs expectatios, ad the exchagerate exposure effect. Joural of Fiace, 49, Bartram, S.M., Bodar, G.M., The exchage rate exposure puzzle. Maagerial Fiace, 30(9, Bell, S.K., Chao, H., Prospects for iteratioal baks i Chia: the outlook is mixed. Corporate Fiace Review, 15(1, Bodar, G.M., Getry, W.M., Exchage rate exposure ad idustry characteristics: evidece from Caada, Japa ad the USA. Joural of Iteratioal Moey ad Fiace, 12, Bodurtha, J., Kim, D., Lee, C., Closeded coutry fuds ad US market setimet. Review of Fiacial Studies, 8, Brickley, J. A., Returs ad risks of U.S. bak foreig currecy activities: Discussio. Joural of Fiace, 41, Burside, C., Eichebaum, M., Rebelo, S.T., Hedgig ad fiacial fragility i fixed exchage rate regimes. Europea Ecoomic Review, 45, Chamberla S., Howe, J. S., Popper, H., The exchage rate exposure of U.S. ad Japaese bakig istitutios. Joural of Bakig & Fiace, 21(6, Cha, K., Mekveld, A. J., Yag, Z., Are domestic ivestors more iformed tha foreig ivestors? Evidece from the perfectly segmeted market i Chia. Workig Paper. Chag, R., Velasco, A., Fiacial fragility ad the exchage rate regime. Joural of Ecoomic Theory, 92(1, 134 Cho J. J., Elyasia E., Kopecky, K. J., The sesitivity of bak stock returs to market, iterest ad exchage rate risk. Joural of Bakig & Fiace, 16(5, Cho J. J., Elyasia E., Derivatives exposure ad the iterest rate ad exchage rate risks of U.S. baks. Joural of Fiacial Services Research, 12(2/3,

29 Cho J.J., Prasad, A.M., Exchage risk sesitivity ad its determiats: a firm ad idustry aalysis of US multiatioals. Fiacial Maagemet, 24, Doidge, C., Griff J., Williamso, R., Measurig the ecoomic importace of exchage rate exposure. Joural of Empirical Fiace, 13(45, Domowitz, I., Gle, J., Madhava, A., Market segmetatio ad stock price: Evidece from aemergig market. Joural of Fiace, 52, Domiguez, K. M. E., ad Tesar, L. L., Exchage rate exposure. Joural of Iteratioal Ecoomics, 68, Doukas, J., Hall, P.H., Lag, L.H.P., The pricig of currecy risk i Japa. Joural of Bakig & Fiace, 23, Eichegree, B., Hausma, R., Exchage rates ad fiacial fragility. Natioal Bureau of Ecoomic Research, Workig paper Elyasia E., Maser, I., Sesitivity of the bak stock returs distributio to chages i the level ad volatility of iterest rate: A GarchM model. Joural of Bakig & Fiace, 22, Eu, C.S., Jaakiramaa, S., A model of iteratioal asset pricig with a costrait o the foreig equity owership. Joural of Fiace, 41, Fama, E.F., Frech, K.R., Multifactor explaatios of asset pricig aomalies. Joural of Fiace, 51, Ferald, J. G., Rogers, J. H., Puzzles i the Chiese stock market. The Review of Ecoomics ad Statistics, 84, Flaery, M.J., James, C.M., The effect of iterest rate chages o the commo stock returs of fiacial istitutios. The Joural of Fiace, 39(4, Flood E., Lessard, D.R., O the measuremet of operatig exposure to exchage rates: a coceptual approach. Fiacial Maagemet, Fraser, D.R., Madura, J., Weigad, R.A., Sources of bak iterest rate risk. Fiacial Review, 37, Froot, K.A., Dabora, E.M., How are stock prices affected by the locatio of trade? Joural of Fiacial Ecoomics, 53,

30 Grammatikos, T., Sauders, A., Swary, I., Returs ad risks of U.S. bak foreig currecy activities. Joural of Fiace, 41(3, Hahm, J.H., Iterest rate ad exchage rate exposures of bakig istitutios i precrisis Korea, Applied Ecoomics, 36, Haa, E. J., B. G. Qu The determiatio of the order of a autoregressio. Joural of the Royal Statistical Society, B 41, Hekma, C.R., Measurig foreig exchage exposure: A practical theory ad its applicatio. Fiacial Aalysts Joural, 39, Hietala, P.T., Asset pricig i partially segmeted markets: Evidece from the Fiish market. Joural of Fiace, 44, Hirtle, B.J., Derivatives, portfolio compositio, ad bak holdig compay iterest rate risk exposure. Joural of Fiacial Services Research, 12, Hutso, E., Steveso, S., (2010. Opeess, hedgig icetives ad foreig exchage exposure: A firmlevel multicoutry study. Joural of Iteratioal Busiess Studies, 41(1, Jorio, P., The exchagerate exposure of U.S. multiatioals. Joural of Busiess, 63, 331. Kamil, H., Does movig to a flexible exchage rate regime reduce currecy mismatches i firms balace sheets? IMF Workig Paper. Kae, E.J., Ual, H., Chage i market assessmet of deposit istitutio riskiess. Joural of Fiacial Services Research, 2, Kim, M., Szakmary, A.C., Mathur, I., Price trasmissio dyamics betwee ADRs ad their uderlyig foreig securities. Joural of Bakig & Fiace, 24, Kle L. R., Nakamura, M., Sigularity i the equatio systems of ecoometrics: some aspects of the problem of multicolliearity. Iteratioal Ecoomic Review, 3, Kle M.W., Shambaugh, J.C., The dyamics of exchage rate regimes: Fixes, floats ad flips. Joural of Iteratioal Ecoomics, 75(1,

31 Lee, C. M. C., Shleifer, A. Thaler, R.H., Ivestor setimet ad the closeded fud puzzle, Joural of Fiace, 46, Joseph, L., Poo, J., Covergece of Ashare ad Hshare prices. Workig Paper No. 19, Hog Kog Securities ad Futures Commissio. Luo, D., Yao. S., World fiacial crisis ad the rise of the Chiese baks. Research paper, Uiversity of Nottigham. Ma,., Capital cotrols, market segmetatio ad stock prices: Evidece from the Chiese stock market. PacificBasi Fiace Joural, 4, Madura, J., Zarruk, E.R., Bak exposure to iterest rate risk: a global perspective. Joural of Fiacial Research, 18, Mart A. D., Madura, J., Akhigbe, A Ecoomic exchage rate exposure of USbased MNCs operatig i Europe. Fiacial Review, 34, Mart A. D., Mauer, L. J., Exchage rate exposures of US baks: A cash flowbased methodology. Joural of Bakig & Fiace, 27, Mart A. D., Exchage rate exposure of the key fiacial istitutios i the foreig exchage market. Iteratioal Review of Ecoomics ad Fiace, 9, Merikas A., The exchage rate exposure of Greek bakig istitutios. Maagerial Fiace, 25(8, Muller, A., Verschoor, W., Foreig exchage risk exposure: Survey ad suggestios. Joural of Multiatioal Fiacial Maagemet, 16, Parsley, D. C., Popper, H. A., Exchage rate pegs ad foreig exchage exposure i East ad South East Asia. Joural of Iteratioal Moey ad Fiace, 25, Scheider, M., Torell, A., Balace sheet effects, bailout guaratees ad fiacial crises. Review of Ecoomics Studies, 71(3, Schwarz, G. E., Estimatig the dimesio of a model. Aals of Statistics, 6 (2, Sog, F., A twofactor ARCH model for depositistitutio stock returs. Joural of Moey, Credit ad Bakig, 26,

32 Stulz, R.M., Wasserfalle, W., Foreig equity ivestmet restrictios, capital flight, ad shareholder wealth maximizatio: Theory ad evidece. Review of Fiacial Studies, 4, Stulz, R.M., Williamso, R.G., Idetifyig ad quatifyig exposures. Workig paper. Ohio State Uiversity. Tia, G.G., Ma, S., The relatioship betwee stock returs ad the foreig exchage rate: the ARDL approach. Joural of the Asia Pacific Ecoomy, 15(4, Wag, Y., Duallistig i Hog Kog ad its impact o the performace of A shares. Workig paper, Hog Kog Baptist Uiversity. Wag, S.S., Jiag, L., Locatio of trade, owership restrictios, ad market illiquidity: Examiig Chiese A ad Hshares. Joural of Bakig & Fiace, 28, Wetmore, J. L., Brick, J. R., Commercial bak risk: market, iterest rate, ad foreig exchage. Joural of Fiacial Research, 17(4, White, H., A heteroskedasticitycosistet covariace matrix estimator ad a direct test for heteroskedasticity. Ecoometrica, 48, Williamso, R.G., Exchage rate exposure ad competitio: evidece from the automotive idustry. Joural of Fiacial Ecoomics, 59, Wog, F.M.H., The associatio betwee SFAS No. 119 derivatives disclosures ad the foreig exchage risk exposure of maufacturig firms. Joural of Accoutig Research, 38, Wog, TC., Wog, J., Leug, P., The foreig exchage exposure of Chiese baks. Chia Ecoomic Review, 20, Yao, S., Luo, D., Morga, S., Bak share prices ad stock market itegratio i Greater Chia. Joural of the Asia Pacific Ecoomy, 15(4, Zhag, Y., Zhao, R., Risk uder oe coutry ad two systems : Evidece from class A, B ad H shares of Chiese listed compaies. Review of Pacific Basi Fiacial Markets ad Policies, 6 (2,

33 Zhao, Z., Ma, Y., Liu, Y Equity valuatio i mailad Chia ad Hog Kog: the Chiese AH share premium. HKIMR Workig Paper No

34 Table 1 Descriptive statistics o stock returs, tradig volumes ad turovers of A ad H shares of the 6 duallisted Chiese baks Stock retur (% Tradig volume (millio shares Turover (% Mea Media Mea Media Mea Media BOC_A BOC_H Obs ICBC_A ICBC_H Obs CCB_A CCB_H Obs BOCS_A BOCS_H Obs CITIC_A CITIC_H Obs CMB_A CMB_H Obs Notes: I this table we preset retur ad volume summary statistics o for the 6 duallisted Chiese baks. Stock retur is calculated by the log differece of the daily stock price, ad turover is the ratio of tradig volumes to available shares i the correspodig market. The umber of the observatios varies across idividual baks due to their differet dates of IPO. Observatios that are lower tha the 1st percetile or higher tha the 99th percetile of stock returs, tradig volumes ad turovers are excluded as outliers for each bak. BOC = Bak of Chia, ICBC = Idustry ad Commercial Bak of Chia, CCB = Chia Costructio bak, BOCs = Bak of Commuicatios, CITIC = Chia Iteratioal Trust ad Ivestmet Corporatio, CMB = Chia Merchat Bak. Data sources: Bloomberg database; GTA database. 34

35 Table 2 Fullperiod estimatio results Pael A: Duallisted baks Explaat ory variables Costat Coeffic iets i 1.50 CH R CH, t RFC 0.75 * (31.36 R HK, t RF HK H 0.04 (2.08 I t t t1 t2 t3 t4 t5 t6 I 0, Ashares BOC ICBC CCB BOCs CITIC CM B 0.16 * * 1, , , ( , ( , ( , * (2.54 j j 0 j, 0.49 [1.36] * ( ( * ( ( ( * ( * ( [1.68] * ( ( * ( ( ( ( * ( [1.80] * ( * ( * * ( (1.50 * * ( * * ( ( ( ( * ( * ( ( ( ( * ( * ( * ( [1.99 ] Hshares BOC ICBC CCB BOCs CITIC CM B ( * ( [0.69] ( * ( ( * * [1.06] * ( * ( ( ( ( ( * ( [1.86] * * ( * ( ( ( ( * ( * * * * 1.71 Adj. R N Pael B: Locallisted shares 3.44 [1.55] ( * ( * ( ( ( ( [1.43] Explaatory variables Costat Coefficie ts PDB SDB MSB INB HB BJB NJB NBB i (1.00 (0.57 (0.61 (0.17 (0.09 ( R CH, t RF CH CH, t 0.83* ( * ( * ( * ( * ( * ( * ( * (

36 I t I 0.02 ( ( ( t , ( ( t , ( t , * 1.82 ( t , ( ( * ( ( (0.68 t , ( ( * ( ( * (1.77 t5 0.74* 5, ( ( ( * ( * (2.34 t6 1.03* 6, ( ( * ( * ( * (2.61 j j0 j, 1.16 [2.14] 1.61 [1.08] Adj. R N CH HK I Notes: This table reports the estimated coefficiets of market portfolio (,, iterest rate ( ad exchage rate ( j, variables from estimatig equatio [2], [3] ad [4] o the 14 Chiese listed baks for the period 21 July 2005 to 4 Nov Figures i paretheses are tstatistics, ad White (1980corrected estimates of the stadard errors are i brackets. *,, ad * deote sigificace at the 10, 5, ad 1 percet levels, respectively, for twotailed tests. For each bak, the umber of j lags we specify is the optimal lag legth. Where o sigificat exchage exposure is foud, we do ot report j,. BOC = Bak of Chia, ICBC = Idustry ad Commercial Bak of Chia, CCB = Chia Costructio bak, BOCs = Bak of Commuicatios, CITIC = Chia Iteratioal Trust ad Ivestmet Corporatio, CMB = Chia Merchat Bak, PDB = Shaghai Pudog Developmet bak, SDB = Shezhe Developmet Bak, MSB = Chia Misheg Bak, INB = Idustrial Bak, HB = Huaxia Bak, BJB = Bak of Beijig, NJB = Bak of Najig, NBB = Bak of Nigbo [1.65] 1.97 [2.07] 1.87 [1.72] j [2.08] 36

37 Table 3 Subperiod estimatio results Duallisted Ashares before fiacial crisis after fiacial crisis variables coefficiets BOC ICBC CCB BOCs CITIC CMB BOC ICBC CCB BOCs CITIC i CH,t A, CH 0.66* ( * ( * ( * ( * ( * ( * ( * ( * ( * ( * (37.04 HK,t A, HK 0.13* ( * ( * ( * ( * ( * ( * ( (0.26 A, I * * * (0.25 A, 1.08* 0, ( ( ( * ( (0.57 A, 0.60* 1, ( ( ( A, , ( ( ( ( ( ( A, , ( ( ( ( * ( * ( * * A, 0.86* 4, ( * ( * ( * ( * ( * ( * * A, 1.34* 5, ( * ( * ( * ( * ( * ( * * * 3.17 A, 1.43* 6, (3.36 j j0 A, j, 0.79 [1.40] 3.08 [1.90] 3.56 [1.93] 2.63 [2.12] 2.35 [1.74] 1.73* ( [2.41] [2.73] 3.99 [2.97] 3.89 [3.08] 1.14 [2.26] 37

38 uallisted Hshares before fiacial crisis after fiacial crisis variables Coefficiets BOC ICBC CCB BOCs CITIC CMB BOC ICBC CCB BOCs CITIC H,t K,t i (0.37 H, CH H, HK 0.94* (48.12 H, I 0.04 H, , 0.34 (0.49 ( * ( ( ( * ( ( (1.64 ( * ( * ( ( ( ( * ( (0.60 ( * ( * ( ( (1.21 ( * ( ( * ( ( ( ( ( ( * ( ( ( * ( * ( * 1.91 H, , ( ( ( * H, , * * * ( H, , * * * ( H, , ( * (2.33 H, 1.66* 5, ( * (2.49 H, , (2.23 j j0 H, j, 1.71 [0.96] 2.06 [2.00] 1.91 [1.63] 1.24 [1.34] [2.12] 5.01 [3.85] 4.32 [2.96] 5.25 [4.16] Pael C: Locallisted shares Explaat ory variable s Costat Coeffici ets i before fiacial crisis PDB SDB MSB INB HB BJB NJ B (0.04 CH R CH, t RF 0.79* ( I I t 0.04 ( * ( * 3.51 ( * ( ( * ( ( * ( * * ( * 1.76 * * ( NBB PDB SDB MS B * * ( * 2.58 ( * ( ( * ( * (2.36 after fiacial crisis ( * ( (1.52 INB ( * ( (1.23 H B * ( * (1.81 BJB NJB NB B ( * ( (0.0 1 ( * ( ( * (

39 t t1 t2 t3 t4 t5 t , , ( , ( , ( , ( , ( , (1.58 j j0 j, ( ( * ( [2.27] * [1.38] ( ( ( ( ( ( ( ( 0.31 ( ( * ( * ( * ( * ( [2.58 ] ( ( * ( * ( * ( [2.1 7] Adj. R N CH HK I Notes: This table reports the estimated coefficiets of market portfolio (,, iterest rate ( ad exchage rate ( variables from estimatig equatio [2], [3] ad [4] o the 14 Chiese listed baks for the precrisis period, defied as 21 July 2005 to 14 September 2008, ad the postcrisis period, 15 September 2008 to 4 Nov Figures i paretheses are tstatistics, ad White (1980 corrected estimates of the stadard errors are i brackets. *,, ad * deote sigificace at the 10, 5, ad 1 percet levels, respectively, for twotailed tests. For each bak, the umber of lags we specify is the optimal lag legth. Where o sigificat exchage exposure is j foud, we do ot report. BOC = Bak of Chia, ICBC = Idustry ad Commercial Bak of Chia, CCB = Chia Costructio bak, j, j 0 BOCs = Bak of Commuicatios, CITIC = Chia Iteratioal Trust ad Ivestmet Corporatio, CMB = Chia Merchat Bak, PDB = Shaghai Pudog Developmet bak, SDB = Shezhe Developmet Bak, MSB = Chia Misheg Bak, INB = Idustrial Bak, HB = Huaxia Bak, BJB = Bak of Beijig, NJB = Bak of Najig, NBB = Bak of Nigbo ( ( * ( * ( * * ( [2.73] * * * [3.25] 0.75 ( * * [2.61 ] 0.35 ( * [1.88 ] 0.68 ( ( * [2.40 ] * * [3.47 ] 1.37 * * [4.3 0] 1.07 ( ( * ( [4.24 ] ( ( (1.01 j, 39

40 Table 4 Summary of chages i Chiese baks foreig exchage exposure after the fiacial crisis Pael A: The sig of estimated exchage exposure Ashare (of dual listed baks Local shares Locallisted baks Hshare (of duallisted baks Precrisis 6/6 0 4/8 1/8 1/6 3/6 Postcisis 1/6 4/6 1/8 6/8 1/6 3/6 Pael B: The timevarig effects of exchage rate chages o stock retur Ashare (of dual listed baks Local shares Locallisted baks Hshare (of duallisted baks Precrisis eg. pos. (6/6 eg. pos. (5/8 mixed Postcisis pos. eg. (5/6 pos. eg. (4/8 mixed Pael C: The average magitude of overall foreig exchage exposure Ashare (of dual listed baks Local shares Locallisted baks Hshare (of duallisted baks Precrisis Postcisis % icrease Notes: I this table we provide a summary of the importat fidigs reported i full i Table 3. Pael A j summarises the chage i the sig of overall exposure j, precrisis versus postcrisis, givig the proportio sigificatly positive ad egative; Pael B provides a summary of the chage i the sigs o j the lagged exchage rate coefficiets; ad Pael C summarises the average absolute magitude j, of the exchage exposure coefficiets pre versus postcrisis. j 0 j 0 40

41 Table 5 Obalacesheet et foreig currecy positio as a percetage of total assets Bak 2005[2] 2006[1] 2006[2] 2007[1] 2007[2] 2008[1] 2008[2] 2009[1] Aver. BOC USD Total ICBC USD 2.33 / Total 2.79 / CCB USD / 4.05 Total / 4.29 BOCs USD a / 0.90 a 0.69 a 1.02 Total a / 0.83 a 0.73 a 1.03 CITIC USD / / Total / / CMB USD 0.35 / 3.55 / 1.81 / 0.90 / 1.65 Total 0.33 / 2.69 / 1.31 / 1.35 / 1.42 PDB USD Total SDB USD 0.01 a 0.06 a 0.08 a 0.51 a 0.17 a 0.31 a 1.27 a 1.14 a 0.29 a Total 0.08 a 0.06 a 0.03 a 0.50 a 0.35 a 0.40 a 1.15 a 1.06 a 0.26 a MSB USD a Total a INB USD 0.80 / Total 0.56 / HB USD Total BJB USD / / 0.18 a / 0.12 a 0.23 a 0.15 a 0.19 a 0.18 a Total / / 0.12 a / 0.12 a 0.23 a 0.22 a 0.27 a 0.19 a NJB USD / / / Total / / / NBB USD 0.60 / 0.36 / Total 0.68 / 0.36 / Notes: I this table we report the semiaual obalacesheet et foreig currecy positio, which is balacesheet foreig assets mius balacesheet foreig liabilities. We report the USD as well as total foreig currecy positios for each bak. For some baks, the data o obalacesheet et foreig currecy positio are ot reported; istead we use the data o the et foreig currecy positio i fiacial assets ad liabilities, ad these are marked a. / deotes uavailable data. Data source: the baks aual ad semiaual fiacial statemets. 41

42 Table 6 The effect of a 1% depreciatio of foreig currecy agaist the RMB o et profit (millio yua Bak m m m m BOC / / / / / ICBC / / 889 / CCB / / / / / / / / BOCs / / 361 / 1108 / CITIC / / CMB / / 47 / 87 / 120 / PDB / / 57 / SDB / / / MSB / / 23 / INB / / 21 / HB / / / / 89 / 190 / BJB / / / / / / / / NJB / / / / 3 / 2 2 NBB / / / / / / / / Notes: For ICBC, PDB, SDB ad HB, the appreciatio of foreig currecy meas specifically the appreciatio of US dollar agaist RMB, while for other baks, it meas the simultaeous appreciatio of all related foreig currecy agaist RMB. Most of the baks coduct their sesivity aalysis assumig a 1% chage i the value of the foreig currecy; the exceptios are BOC ad HB (5% ad SDB (8% i 2006 ad / deotes that the data are uavailable. Data source: the baks aual ad semiaual fiacial statemets. 42

43 Table 7 Exchage gais ad losses as a percetage of total assets, semiaual 2005[2] 2006[1] 2006[2] 2007[1] 2007[2] 2008[1] 2008[2] 2009[1] Mea BOC / / / ICBC / / CCB / BOCs / CITIC / CMB PDB SDB MSB INB HB BJB / / / / NJB / / / NBB 9 / Note: I this table we detail the semiaual foreig exchage gais ad losses reported i the baks fiacial statemets. As defied i the aual report of ICBC (the other baks have similar defiitios, exchage gais ad losses icludes the icome from related proprietary foreig exchage busiess, the realised gai or loss ad urealised fair value chage of foreig exchage derivatives as well as the traslatio gais ad losses i foreig assets ad liabilities. The data i the shaded areas are the gais ad losses from foreig assets ad liabilities, which are available oly for BOC ad ICBC. / deotes uavailable data. Data source: the baks aual ad semiaual fiacial statemets. 43

44 Figure 1 The Chiese exchage rate agaist US dollar (Jauary 1994 November Notes: This graph depicts the USD/RMB exchage rate from 1st Jauary 1994 to ed October O 21 July 2005 (marked with the dashed lie, the People s Bak of Chia (Chia s cetral bak aouced that the RMB exchage rate will o loger solely peg to US dollar, but refer to a basket of major currecies with certai weights, correspodig to the actual situatio of our (Chia s foreig ecoomic developmet. 44

45 Figure 2 The equity price of both A ad Hshare for 6 duallisted baks BOC ICBC CCB BOCs CITIC CMB Notes: This figure depicts the stock prices of Aad H shares for 6 duallisted Chiese baks durig the period whe both A ad Hshare prices are available. There time periods for each bak are differet due to their differet dates of iitial public offerig. 45

46 Figure 3 Chage i foreig exchage reserves (billio US dollars Notes: FDI ad trade surplus are the mai sources of foreig exchage reserves (Fiacial Times, 13th July 2008; Emergig Markets Moitor, 19th Ja The blue bar represets the sum of FDI ad trade surplus; the whole bar (the blue plus the red idicates the chage i Chiese foreig reserves; ad the red bar represet those compoets that caot be explaied by movemets of moey via official chaels. Data sources: Miistry of Commerce of the People s Republic of Chia, State Admiistratio of Foreig Exchage, ad Chia Customs Statistics, 46