Boston, USA, August 5-11, 2012

Transcription

1 Poster Sesson # Tme: Monday, August, 202 PM Paper Prepared for the 32nd General Conference of The Internatonal Assocaton for Research n Income and Wealth Boston, USA, August -, 202 New Frameworks for Measurng Global-Flow-of-Funds: Fnancal Stablty n Chna Nan Zhang For addtonal nformaton please contact: Name: Nan Zhang Afflaton: Hroshma Shudo Unverst, Japan Emal Address: zhang@shudo-u.ac.jp Ths paper s posted on the followng webste:

2 New Frameworks for Measurng Global-Flow-of-Funds: Fnancal Stablty n Chna Nan Zhang Abstract In ths paper we am to buld a statstcal montorng system for measurng global-flow-offunds usng concepts and theoretcal frameworks of global-flow-of-funds. Frstly, we nspect the nfluence of global-flow-of-funds and the contnual growth of macro economy on the stablty of fnancal systems, and buld a statstcal montorng system for global-flow-of-funds whle referrng to the FSIs frameworks of Internatonal Monetary Funds. Secondly, we dynamcally lnk real economcs wth fnancal economcs, and combne domestc flow of funds wth nternatonal captal flows, to buld a composte ndex (CI) that reflects the rsks nvolved n external flow of funds. Thrdly, we create a Chnese fnance stress ndex that corresponds well wth the current status of Chnese external flow of funds. Fourthly, we expand the emprcal analyss based on above statstcal methods, and brng up future ssues for dscussons. Key Words: Statstcal framework, Fnancal Soundness Indcator, Systemc rsk, Fnancal Crss, Composte Index, Fnance Stress Index. Introducton Treasury Internatonal Captal provded by the U.S. Department of the Treasury brought Chnese flow of funds to the world s attenton by ndcatng that as of March, 20 Chna holds almost. trllon U.S. dollars worth of U.S. Treasury bonds. At the end of 2006, the Chnese foregn exchange reserves reached.07 trllon U.S. dollars, the largest n the world. Then by March 20, the Chnese foregn exchange reserves rapdly ncreased to 3.04 trllon U.S. dollars. As U.S. dollar deprecaton contnues, the rsks accompanyng the massve Chnese foregn exchange reserves hang lke the Sword of Damocles. Realty asks that we objectvely observe the drecton of Chnese external flow of funds, assess the rsks n possessng massve foregn exchange reserves for the development of the Chnese economy, and contemplate ways to deal wth the Chna-U.S. mrror-mage n the scope of nternatonal flow of funds. There are dfferent explanatons for the causes of mbalance that exsts n the Chna-U.S. economy, but they all share one common vewpont. That s, governments should strengthen 2

3 ther supervson on nternatonal flow of funds to prevent future fnancal crses. Chna joned the Internatonal Monetary Fund (IMF) n December 6, and by agreeng to the IMF s Artcle VIII General Oblgaton of Members, has realzed the lberaton of ts current accounts. However, the Chnese government has not fully opened ts captal fnancal market, and lberaton s lmted to only part of the captal accounts. As the Chnese economy becomes more nfluental n the world s economy, Chna's external flow of funds become ncreasngly ntegrated nto global-flow-of-funds. Therefore, an nevtable task for the Chnese economy s to supervse and operate ts captal accounts to nternatonalze the Chnese fnance. A statstcal montorng system to establsh fnancal stablty s mperatve to accomplsh ths task. Based on the suggestons from IMF, People s Bank of Chna publshed the Chna Fnancal Stablty Report n 200 for the frst tme, and whch contnues to today. Meanwhle, IMF has publshed Fnancal Soundness Indcator (FSIs) n 2006 and also the Global Fnancal Stablty Report to reflect on world fnancal stablty. In addton, IMF has strengthened ther research on the early warnng system (EWS) for future fnancal crses. FSIs take advantage of three basc computatonal frameworks: SNA system, nternatonal accountng standards, and Basel Commttee on Bankng Supervson, to buld 2 core ndces and 27 encouraged ndces to reflect fnancal stablty. However due to dfferng analytcal goals, there are two shortcomngs n FSIs for t to reflect fnancal structural problems and to assess fnancal rsks. Frst, snce FSIs focus on the sum of sectors, t cannot provde statstcal bases for the argument that economc structural changes affect fnancal stablty. Secondly, FSIs focus on fnancal stablty, but lacks feedbacks on fnancal rsks caused by real economcs as well as factors from overseas. The am of ths paper s to buld a statstcal montorng system for global-flow-of-funds usng concepts and theoretcal frameworks of global-flow-of-funds. Frst, nspect the nfluence of global-flow-of-funds and the contnual growth of macro economy on the stablty of fnancal systems, and analyze FSIs frameworks as well as Treasury Internatonal Captal. Secondly, buld a statstcal montorng system for global-flow-of-funds from a theoretcal framework of nternatonal flow of funds. Thrdly, dynamcally lnk real economcs wth fnancal economcs, and combne domestc flow of funds wth nternatonal captal flows. Then buld ) a composte ndex (CI) that reflects the rsks nvolved n external flow of funds and 2) a Chnese fnance stress ndex to focus on the current status of Chnese external flow of funds. Fourthly, based on the above statstcal methods, expand the emprcal analyss, and brngs up future dscussons. 2. About the Fnancal Stablty Statstcs Gude One of the lessons learned from the 7 Asan fnancal crss s the lack of statstcal nformaton for montorng fnancal rsks. Amng to buld a global fnancal stablty statstcs reportng system, the Statstcs Department at IMF organzed a conference of statstcans n 3

4 related felds n, whch was the start of FSIs. In 2000, the Statstcs Department at IMF conducted statstcal surveys on fnancal stablty among partcpatng countres. Based on the surveys, IMF Executve Board ntroduced the FSIs statstcal system n 200. In the followng years 2002 to 2004, IMF drafted the Gude, whch s a gudebook for establshng and promotng the FSIs system. After careful consderaton and solctng vews on ts draft, IMF publshed the Gude on ts webste n July 2004 and as an offcal document n March Startng July 200, FSIs data have been sent to 62 engagng countres through IMF s database for use. IMF encourages ts partcpatng members to establsh FSIs, to allow comparson of FSIs data among themselves and to strengthen ts supervson on ndvdual countres and the nternatonal fnancal system. The goals n establshng FSIs are the fnal assessment and supervson of the strength and vulnerablty of fnancal systems. Borrowng from the basc computatonal frameworks of SNA system, Internatonal Accountng Standard (IAS), and Bankng Supervson Gude (Basel Commttee on Bankng Supervson), FSIs share commonalty wth the above three n sector classfcaton, but at the same tme dffer sgnfcantly n three ways due to dfferent analytcal goals. One s that ther treatment of sector nformaton s dfferent from the nterest of commercal accountng and bankng supervson approached n ndvdual enttes, but the FSIs framework, lke the SNA system, focuses on aggregated sector nformaton. Second s the dfference n statstcal range of focus. SNA system focuses on the entre macro economy, t embrace symmetrc recordng of flows and postons wthn and across sectors, but commercal accountng and supervsory approaches do not (because of the focus on the ndvdual entty). The FSIs framework favors a symmetrc recordng of flows and postons wthn the sector, to avod dstortons n the sector data, but not necessarly among sectors, because the type of data requred dffers by sector. Thrd s the dfference n observatonal targets. SNA targets nclude all economc actvtes, whereas the FSIs framework, adopt summaton and merger to avod the double countng of captal and actvty, whch resembles frameworks n commercal accountng and bankng supervson. Table contans 2 core and 27 encouraged sets n FSI. Core sets are applcable to all countres, and encouraged sets can be adopted accordng to each country s needs. Core sets cover captal adequacy, asset qualty, earnngs and proftablty, lqudty and senstvty to market rsk. Encouraged sets cover depost takers, other fnancal corporatons, nonfnancal corporaton sector, and four sectors n households, as well as market lqudty and real estate markets whch are very nfluental to fnancal stablty. Further, as rapd development n nformaton technology and new fnancal products takes place, nternatonal captal flow s seeng ntensve change n magntude and speed. As a result, montorng the fnancal system pressure and volatlty of nternatonal captal crculaton becomes more mportant, servng a vtal role as a regulator n captal management n FSIs. IMF, Fnancal Soundness Indcators, 2006, p4. 4

5 Table. Fnancal Soundness Indcators: The Core and Encouraged Sets Depost takers Captal adequacy Asset qualty Earnngs and proftablty Lqudty Senstvty to market rsk Core Set Regulatory captal to rsk-weghted assets Regulatory Ter captal to rsk-weghted assets Nonperformng loans net of provsons to captal Nonperformng loans to total gross loans Sectoral dstrbuton of loans to total loans Return on assets Return on equty Interest margn to gross ncome Nonnterest expenses to gross ncome Lqud assets to total assets (lqud asset rato) Lqud assets to short-term labltes Net open poston n foregn exchange to captal Encouraged Set Depost takers Captal to assets Large exposures to captal Geographcal dstrbuton of loans to total loans Gross asset poston n fnancal dervatves to captal Gross lablty poston n fnancal dervatves to captal Tradng ncome to total ncome Personnel expenses to nonnterest expenses Spread between reference lendng and depost rates Spread between hghest and lowest nterbank rate Customer deposts to total (non-nterbank) loans Foregn-currency-denomnated loans to total loans Foregn-currency-denomnated labltes to total labltes Net open poston n equtes to captal Other fnancal corporatons Assets to total fnancal system assets Assets to gross domestc product (GDP) Nonfnancal corporatons sector Total debt to equty Return on equty Earnngs to nterest and prncpal expenses Net foregn exchange exposure to equty Number of applcatons for protecton from credtors Households Household debt to GDP Household debt servce and prncpal payments to ncome Market lqudty Average bd-ask spread n the securtes market Average daly turnover rato n the securtes market Real estate markets Real estate prces Resdental real estate loans to total loans Commercal real estate loans to total loans Source: IMF, Fnancal Soundness Indcators Complaton Gude, March 2006 FSIs cover several aspects n fnancal stablty. In a fnancal system, captal strength s an mportant factor for every organzaton, especally as a buffer for any unexpected losses. In montorng fnancal stablty, we should also consder other asset structure, qualty, and exposure to fnancal rsks. Income and expendture nformaton are also crucal. If an entty

6 cannot produce enough ncome, t s never fnancally secure and stable. For non-fnancal corporatons, the focal pont s ther ablty to pay ther labltes and to fnance debt. In short, fnancal stablty ndex ams to montor changes n savngs (exposure to rsks) and flow, because they are able to reflect changes n vulnerablty n fnancal sectors and to help assess fnancal sectors ablty to handle stress and rsks. 3. New frameworks for global-flow-of-funds statstcs (GFFS) GFF relate to domestc flows and nternatonal captal flows. The analyss of GFF shows the characterstcs and the structure n the flows of fund, ncludes the flows of all domestc funds wth nvestment-savngs, lnks flow of funds wth current balance, and connects nternatonal captal flows. 3. The Mechansm of Global-Flow-of-Funds The fnancal markets ndcate the debts and credts of funds as a whole n addton to the total process of fnancal lqudty. Investgated more carefully, tems of fnancal markets nclude nflows of domestc funds, overseas funds by domestc savngs, and credt loans of banks on the sde of fund-sources (funds nflow). On the other hand, the fnancal markets splt nto supply of funds to the domestc economy and overseas sector as fund-uses (funds outflow). Fgure s a fgure showng the mechansm of global-flow-of-funds among three countres: A, B and C; an nternatonal fnancal market; and an nternatonal organzaton. The economy of three countres conssts of the balance of savngs-nvestment whch reflects real economy actvty, and the fnancal market whch ndcates the fnancal crculaton of funds. As a balance between domestc economy and overseas sector n each country, real economy (savngs-nvestment balance) serves as a current balance, and the fund loan balance n the fnancal market serves as the captal balance from an nternatonal vewpont. The current transacton and captal transacton of each country are connected nternatonally, and a part of captal transacton lnks to an nternatonal fnancal market, IMF, the World Bank, etc., that are formed by fund-rngs n the global-flow-of-funds. On Fgure, the excess savngs correspondng to current balance surplus occur n the country of captal suppler (n country A), causng fnancal assets to see net ncrease n fnancal sector. Fnancal market receves nflow of the funds from domestc and overseas sector, and supples funds to the two sectors smultaneously. In the fund loan balance of country A fnancal market, the net nflow of funds corresponds to excess savngs balance n the domestc sector, and the net outflow of funds (ncludng the change n foregn exchange reserves) corresponds to the current account surplus n the overseas sector. That s, the net ncrease of the fnancal assets whch offsets the excess of domestc savngs s balanced wth 6

7 the net nflow of funds from the domestc sector n the fnancal market. And the external clamable assets formed correspondng to the current balance surplus becomes the net outflow of the funds from fnancal market as supply of overseas lendng. ( EX > IM ) ( EX < IM ) ( EX < IM ) A country B country C country I < S I > S I > S FI FI d o CRA < > FO FO d o FI FI d o CRA > < FO FO d o FI FI d o CRA > < FO FO d o Internatonal fnancal market Internatonal organzaton Fgure the Mechansm of Global-Flow-of-Funds Notes: FI d : domestc nflow of funds, FO d : domestc outflow of funds, CRA : Changes n Reserve Assets FI o : overseas nflow of funds FO : overseas outflow of funds o On the other hand, n a captal mporter such as country B or C, the current balance defct s lnked wth the domestc excess of nvestment (savngs defct) and the net ncrease of the fnancal lablty n the fnancal sector. In the fnancal market, there s an excess of credt wth the domestc excess of nvestment, and current account defct s fnanced by the net nflow of funds (captal balance surplus) from overseas. Therefore, on the funds account balance wth the sectors of the domestc-and-overseas n the fnancal market of country B and C, t becomes a net outflow of funds wth the domestc sector, and the net nflow of funds wth the overseas sector. The net nflow of funds from the overseas sector becomes a source of funds for the domestc sector that attempts to keep a balance of credt. Moreover, the net outflow of funds nto the domestc sector n the fnancal market wll serve as over-borrowng for the domestc sector, also known as the net ncrease of fnancal lablty. In ths way, an nternatonal captal movement from a surplus country of current balance to a defct country arses. The flow of captal goes drectly between two natons: from a surplus country to a defct country, or, also arses n many countres va the nternatonal fnancal market, IMF, the World Bank, etc. ndrectly. These nternatonal funds wll be managed by an agency of an ntergovernmental publc base or part of the World Bank, although most of them arse by factors such as the pursut of nterest dfferental or captal gan and rsk averson 7

8 through a market mechansm. Anyway, nternatonal captal movement wll be fnanced wth the balance on current account from the standpont of Balance of Payment n each country, and wll perform the functon of nternatonal fnancal ntermedaton from an excess-savngs country to a defct-savngs (excess-nvestment) country from a global standpont. Moreover, when a captal suppler country s a key-currency country (lke the U.S.), ths country has acheved the functon whch supples nternatonal lqudty. Thoroughly observng the flow of funds, the moblty (nternatonal lqudty and domestc money supply) s an ntegrated system n global-flow-offunds whch connect major power economes, because the flow of funds between countres are connected wth domestc flow of funds as an ntegrated whole n each of the relevant countres. 3.2 New Framework for Global-Flow-of-Funds Statstcs When the flow of funds n fnancal markets s ted up wth the balance of payments, the overseas sector wll have fund outflow excess (net captal outflows) f the current account s n surplus. Conversely, the domestc sector wll have fund nflow excess. Therefore, when the real economc sde of the domestc and overseas economy s analyzed under an open economc system, the balance of savngs-nvestment of the domestc economy corresponds to the current account balance. Accordng to the dynamc process of external flow of funds and the defntonal equaton of a system of natonal account, the accountng dentty becomes as follows, Y = C + I + G + EX - IM Equlbrum condton s obtaned through arrangng the above formula. S - I = EM - IM The rght sde of equlbrum condton s the current account, and the left sde of equlbrum condton s the balance of savngs-nvestment, or net fnancal nvestment. Hence, equlbrum condton equatons are reached as follows. S I = EX IM = NFI However, domestc net funds outflow correspond wth the captal account balance when we examne the fnancal relatonshp between domestc and overseas. Therefore, relatonshps among the domestc savngs-nvestment balance, the fnancal surplus or defct, the current account, and the overseas net fund outflow wll be expressed n the followng structural formulae. Savngs-Investment and Current Account Balance S I = FA FL = EX IM () Overseas Income and Expendtures Balance EX IM = ( FO FI) + CRA (2) Regard r t FI t as nterest payment of external debt, and putcra = FRA t FRAt, (2) changes to (3) 8

9 ( t t t t t t t t = EX IM ) + ( FI FO ) + ( r FI ) + ( FRA FRA ) 0 (3) ' GFF crss can be shown as follows when FRA mnmum receved was set to FRA ( EX t IM t ) + ( FIt FOt ) + ( rt FIt ) + FRAt < FRA' (4) Where S : gross savngs, I : gross nvestments, FA: change n fnancal assets, FL : change n fnancal labltes, EX : exports, IM : mports, FO : fund outflow, FI : fund nflow, FRA: foregn reserves asset Formula () shows the relatonshp among the savngs-nvestment and domestc fund flows and current account. When S > I, FA > FL s captal surplus, CA >0, and mean current account s a surplus(em > IM). On the other hand, when S < I and FA < FL, there exsts fnancal defct n the flow of funds wth CA < 0, and mean current account s defct (EM < IM). It s clear that global flow of funds and nternatonal flow of goods behave lke the head and tal of a con formula (2). When current account s n surplus, captal account can be n defct (FO > FI, net outflow of funds). Or an ncrease n foregn reserves can cause a country s external clam to ncrease. Otherwse, when domestc nvestment s larger than domestc savngs, current account s n defct, and external debt can only ncrease through nflow of funds (FO < FI, captal balance s n surplus), or decreasng foregn reserves as a compensaton for the current account defct. By changng formula (2), we see consttutes of foregn reserves whch show that the smultaneous exstence of current balance and captal balance surplus wll ncrease foregn reserves. In other words, when current balance s n surplus, net outflows should also be n surplus, or foregn reserves wll ncrease. As a result, foregn reserves ncrease rapdly, whch leads to a systematc problem n external flow of funds. Based on formula (2), man factors that affect the change n foregn reserves are current balance and captal balance. Snce the change n current balance s due to the balance of savngs-nvestment, change n foregn reserves actually depends on the change n the structure of flow of funds. Formula 4 shows several possbltes for crss to take place n nternatonal flow of funds. One s when current account defct s too large (IM > EM) for pre-foregn exchange reserves to handle. Second comes from changes n stock market returns, market nterest rates, and foregn exchange rates, whch cause short-term captal outflow to be dramatcally larger than nternatonal captal nflow. In ths case, a shortage n foregn exchange reserve for handlng domestc captal needs may trgger a currency crss. Thrd s crss n external debt payments due to current account defct and captal account defct. The fourth case s when exchange rates rapdly fluctuate, causng currency to go through sgnfcant apprecaton or deprecaton and eventually leadng to systematc crses n current account, captal account, external debt payments, and so on.

10 In order to test fnancal stress and make early warnngs for systematc fnancal crss through GFF, we need a new statstcal framework that corresponds to the operatonal structure of GFF, whch becomes the foundaton of a statstcal montorng system. Ths statstcal framework must reflect dynamc changes n economc enttes and fnancal statstcs, and also lnk domestc flow of funds wth nternatonal captal movement. Four aspects of external flow of funds should be montored: ) ndcate any nfluence on current accounts from changes n economc structure whch causes savngs-nvestment mbalances, 2) ndcate rsks n nternatonal captal flow caused by surplus or defct of domestc funds, 3) ndcate shocks to nternatonal captal flow caused by mbalance n current accounts, and by nternatonal largescale captal nflows or outflows, 4) ndcate causes of changes n foregn exchange reserves and pressure of fnancal stablty from rapd changes n foregn exchange reserves. 3.3 Selecton of Indexes for Montorng System for GFF In buldng the computatonal rules for a statstcal montorng system for nternatonal flow of funds, four basc computatonal frameworks can be referred: SNA system ndex, Basel Commttee on Bankng Supervson, IMF fnancal stablty statstcal system, and Treasury Internatonal Captal provded by the U.S. Department of the Treasury. Accordng to the dynamc structure of nternatonal flow of funds (see formula -4), we desgned a statstcal montorng system for GFF (see table 2). Out of the four observatonal stages n nternatonal flow of funds, we frst pcked factors that affect savngs-nvestment balance (S-I): dsposable ncome, fnal consumpton, net savngs rato, CPI, market nterest rates, GDP and government expendture. As factors that affect domestc fund surplus or shortage ( FA- FL), we pcked sectoral fnancal surplus or defct to GDP, net outflow of funds to GDP, spread between hghest and lowest nterbank rate, regulatory captal to rsk-weghted assets, lqudty asset to total assets, lqud asset to short-term labltes, non-performng loans to total gross loans, resdental real estate loans to total loans, real estate prces. As factors that affect external trade flows (EX-IM), real effectve exchange rate, consumpton expendture of U.S., total GDP of man areas, mport prce ndex, Chnese GDP, trade balance to GDP. As factors that affect external flow of funds (FO-FI), we pcked economc growth rate, spread between overseas and domestc nterests, nterest rates of central bank, foregn currency-denomnated loans to total loans, government bonds yelds of the US, holdng US government bonds to total FRA, return on equty, change n reserve assets, proft from nvestment, net foregn exchange exposure to equty, FDI, large exposure to captal, net open poston n foregn exchange to captal. 0

11 Table 2.Global-Flow-of-Funds Statstcs S I = FA FL = EX IM GFFS Source Source Net savngs rato SNA GDP SNA Dsposable ncome FOF Government Expendture FOF CPI CBQSB Fnal consumpton FOF Spread between lendng and depost rate FSIs Spread between hghest and lowest nterbank r FSIs Net outflow of funds to GDP IFS Lqud asset to total assets FSIs Sectoral fnancal surplus or defct to G FOF Lqud asset to short-term labltes FSIs Regulatory captal to rsk-weghted asset FSIs Resdental real estate loans to total loans FSIs Nonperformng loans to tatal gross loans FSIs Real estate prces FSIs Trade balance to GDP BOP Consumpton Expendture of the US BEA Real effectve exchange rate IFS Total GDP of man areas IFS Yuan Change rate BOP Import Prce Index GAC CA = ( FO FI) + CRA GFFS Source Source Economc growth rate, SNA Return on equty FSIs Spread between overseas and domest FSIs Changes n reserve assets BOP Interest rates of central bank CBQSB Proft from nvestment BOP Foregn-currency-denomnated loans to t FSIs Net foregn exchange exposure to equty FSIs Government bonds yelds of the US FRB FDI BOP Holdng US government bonds to tot FSIs Large exposures to captal FSIs Federal funds rate of the US TIC Net open poston n foregn exchange to ca FSIs We can use the statstcs observaton system of GFF to measure long-term structural problem. There are endogenous varables and 6 exogenous varables n the model of GFF (Zhang, 2008). Ths model takes the change n the formaton of expectaton and rsk, t wth a lag structure or mmanent relevant factors of many economc varables, and explans the state of the funds flow n the contnuous adjustment process to the balance from mbalance, and n a medum-to-long perod of tme. It also belongs to a knd of dynamc model. A system of smultaneous equatons s a model of fnancal market equlbrum, consstng of the follows. However, ths model can t reflect the short-term wave moton or fnancal stable stuaton of external flow of funds, and can t observe the shock to the fnancal pressure by rapd change of overseas fnancal markets, ether. Moreover, the model doesn t have a functon whch emts the early alarm relatve to generatng of fnancal crss when we want to know that how the proft factor and the rsk factor affect nternatonal captal flows, and how the change of the pattern of funds flow affect domestc economy growth. Therefore, t needs to create the ndex system whch shows the short-term trend of GFF, and fnancal stress ndex as a contnuous research.

12 4. Create a Trend Index for GFF Because external flow of funds fluctuate fluently due to nternatonal flow of funds, nterests, stock prces, foregn exchanges, so the GFF model doesn t reflect necessarly a country s shortterm equlbrum n flow of funds, but s able to make predctons to the future progress and to smulate effects of fnancal polces from dynamcally observng the md- or long-term change trend. Besdes, n desgnng the model, we have to pay attenton to domestc factors, changes n nternatonal market, and also benefts and rsks nvolved n foregn captal flows. Due to these concerns, n buldng GFF model we am to observe the causal relatonshp of each economc varable and structural problem n the system of external flow of funds. However ths model cannot reflect recent captal flow and fnancal stablty, statstcally speculate for potental fnancal crss, nor can t statstcally descrbe the fnancal stress caused by a dramatc change n a fnancal market. That beng sad, we attempt to buld a GFF trend composte ndex (CI) system and a fnancal stress ndex. 4. Purpose and the Bass of Selecton FOF GDP Fgure 2. FOF and GDP wth common trends of change n Chna (00 mllon) Sources: The People s Bank of Chna Quarterly Statstcal Bulletn. Fgure 2 ndcates that Flow of Funds and GDP share a smlar trend, whch s our actual bass for buldng GFF trend ndex. GFF trend ndex refers to Dffuson Index. Under a statstcal montorng system (see Table 2) that reflects the changes of GFF, we select n knds of tme seres ndcators excludng seasonal varatons, rregular varatons and trend changes. We call the adjusted tme seres ndex as DI (Dffuson Index) and t s defned by n+/n, when movng n the extended drecton compared wth the frst half and settng the number of seres to n+. Expressed as percentages, DI bascally uses monthly data. The man focus s tmelness, sustanablty, and correspondng equvalence wth GFF changes. The concept of correspondng equvalence consders whether or not an ndex can reveal the amount along wth the drecton flow 2

13 of funds and can reflect the dynamcs of domestc and the nternatonal fnancal markets. More mportantly, correspondng equvalence requres that the change of ndex over tme exhbts the change of fund flows amount, drecton and the fnancal market volatlty. Based on dfferent movements n tme by dfferent ndces, we can classfy GFF ndex as Leadng Index, Concdent Index, and Laggng Index. Based on above defnton on the dffuson ndex, we can observe changes n those ndces. Among the concdent ndces, when any ndex ncreases by 0%, we consder t as an expanson phase of flow of funds, or economc development. On the other hand, when the ncrease s less than 0%, we may see t as a shrnkng phase of flow of funds, or economc depresson. By the selecton crteron of the above ndex, the system of the trend ndex n global flow of funds s shown n Table 3. Table 3.Composte Index for GFF Leadng Index Return on equty Real estate prces Real exchange rate CPI Government Bonds Yelds of the US Spread between lendng and depost rates Resdental real estate loans to total loans Spread between overseas and domestc nterests Spread between hghest and lowest nterbank ra Concdent Index Net savngs rato Trade balance Regulatory captal to rsk-weghted assets Net outflow of funds Sectoral fnancal surplus or defct Changes n reserve assets Laggng Index Holdng US government bonds Net open poston n foregn exchange to capta Proft from nvestment Dsposable ncome Fnal consumpton 4.2 The method of the ndex creaton Set the value of any ndex at tme = t as ( t)( =,2,, n), and ts rate of change s y { y ( t) y ( t d) }/ y ( ) r ( t) = t () 3

14 Durng a perod=d, we may set d=, comparng wth a month before, or d=3, 3 months before. To smplfy our explanaton, set d=. When DI at any one pont can be wrtten as n DI( t) = { sgn( r ( t)) + } (6) 2n = Where sgn s a sgn functon, defned as sgn( x ) = 0 ( x < 0) ( x = 0) ( x > 0) From formula (6) we see that DI s the average value n each ndex sgn(r). sgn s an ncreasng functon, and DI changes n the same drecton as r. Hence we know that DI can reflect changes n flow of funds, and also changes n dfferent economes. However, DI only reflects one type of change, ncrease or decrease n varable seres, and does not show the magntude. To compensate for ths, we can use composte ndex, whch s bult by four stages as below. About an ndvdual seres y (t), a symmetrcal change rate (t) r as follows r ( t) = y ( t) y ( t ) { y ( t) + y ( t ) }/ 2 (7) But, when y (t) s 0, negatve, or a fracton, r ( t) = y ( t) y ( t ) 2 Calculate the average, standard devaton of r (t) r ( t) = T t ' t = t r ( t) t s ( t) = ' T t = t { } 2 r ( t) r ( t) ' Where T=60(fve years), t = t T + s the tme before fve years. We get standardzaton. 2 z ( t) = ( r ( t) r ( t)) / s ( t) (8) 3 By each Leadng Index, Concdent Index and Laggng Index, compound an ndvdual ndex as follows and calculate an average (composton) change rate ν (t). ν ( t) = r( t) + s( t) z( t) () When consderng t as the number of the ndces whch consttute n 4

15 n n n r( t) = r ( t), s( t) = s ( t), z( t) = z ( t) n n n = = = 4 CI (t) s set by standard tme to 00, and an average change rate s calculated one by one 2 + ν ( t) CI( t) = CI( t ) (0) 2 ν ( t) The above formula can rewrte as follows. ν ( t) = CI( t) CI( t ) { CI( t) + CI( t ) }/ 2 () Changes n v(t) s the average change n r(t), or a change rate n Flow of Funds. Therefore, change rate of CI(t) s the same as v(t). When CI ncreases (decreases), flow of funds ncreases (decreases) and the economy s gong up (or fall). Fnancal Stress Test n GFF Analyss After nvestgatng changes n GFF and fnancal stablty usng CI, we further explore fnancal stress usng Fnancal Stress Index (FSI) that s made by GFF statstcal montorng system. The am of the stress test s to test stress usng the system and assess any potental pressure on fnancal stablty. Please refer to artcles below: KLR Model (Kamnsky, Lzondo & Renhart, 6,), FR Model (Frankel & Rose,6), STV Model (Sachs, Tornell & Velasco,6), DD Model (Demrgüç-Kunt, A. & E. Detragache,8), HK Model (Hakko and Keeton, ), LMY Model (Illng, M. and Y. Lu, 2003), RSSI Model (Rav Balakrshnan, Stephan Dannnger, Selm Elekdag, and Irna Tytell, 200). Stress test and fnancal stress ndex have dfferent functons. Stress testng, as a predctve tool, ams to assess the mpact of unexpected shocks such as the 2008 Fnancal Crss on the soundness of GFF. On the other hand, tests such as CI use ndces that reflect current fnancal condtons. Stress testng and CI play dfferent but complementary roles n survellance.. Establshng Chnese FSI The FSI addresses the weakness nherent n models that use EWIs by mprovng the reference varable. In partcular, the FSI s contnuous, of hgh frequency (daly), and covers the equty markets, bond markets, foregn exchange markets, and the bankng sector. Therefore, t s sutable for analyzng fnancal stablty n hghly developed countres wth numerous systemcally mportant fnancal markets and nsttutons. Fnancal stress s a contnuous varable wth a spectrum of values, where extreme values are called a crss. Stress ncreases wth expected fnancal loss wth rsk (a wdenng n the dstrbuton of probable loss) or uncertanty (lower confdence about the shape of the dstrbuton of probable loss). Stress s the

16 product of a vulnerable structure and some exogenous shock. Fnancal fraglty descrbes weaknesses n fnancal condtons and/or n the structure of the fnancal system. (Illng and Lu, 2003). Based on the characterstcs of Chnese flow of funds and the tmelness and sustanablty of GFFS data, we selected the varables below to buld the fnancal stress ndex. Ⅰ Spread between overseas and domestc nterests Ⅱ Return on equty (Shangha comprehensve ndex) Ⅲ Holdng US government bonds to total FRA Ⅳ Net open poston n foregn exchange to captal Ⅴ Exchange Market Pressure Index (EMPI) erj µ er CRAR rj j µ EMPI t = σ σ erj CRARj CRARj (2) er: the change rate of exchange rate n monthly CRAR: the change rate of reserve asset n monthly Table 4.Varable selecton and the feature of Chna s FSI Varable s code Varables Perod of data Mean Standard Error A Spread between overseas and domestc nterests Jan.2004-March B Return on equty Jan.2004-March C Holdng US government bonds to total FRA Jan.2004-March D Exchange Market Pressure Index (EMPI) Jan.2000-March Included observatons: 6 Because we are unable to obtan the monthly data of net open poston n foregn exchange to captal, we used 4 varables n Table 4 to buld the Chnese FSI. Varable A stands for the stress factors on global flow of funds from spread between overseas and domestc nterests. Varable B s for return on equty, usng Shangha composte ndex to reflect stress on captal market. Varable C s Chnese holdng US government bonds to total FRA, used to test the rsks nvolved n Chnese holdng the US bonds. Varable D, exchange market pressure ndex, shows the stress on Chnese fnance from Chnese exchange rate apprecaton and ncrease n foregn exchange reserves. In formula (2), μ and σ ndcate the average and standard devaton respectvely of er 6

17 and CRAR sequence. These four varables are taken from monthly data from January 2004 to March 20, and we have calculated the average and standard devaton respectvely for each 4 varables. Standardzng varable A, B, C, D based on Table 4, we obtan standardzed varables ZA, ZB, ZC, ZD wth consstent unts. Then we sum these standardzed varables to obtan FSI wth GFF n perspectve. FSI = Z + Z + Z + Z t A B C D (3) FSI Fgure 3. The changes of FSI Fgure 3 ndcates changes n Chnese FSI from January 2004 to March 20. There are 0 months wth FSI > 0, 46 months wth FSI < 0. Of the entre observng perod, FSI remaned relatve stable from January 2004 to June However snce then tll August 200, FSI was n a volatle range, especally snce August 2007 t sld from 0.33 to -. n August 2008 due to B (return on equty), C (holdng US government bond to total FRA), and D (exchange market pressure ndex). Then FSI rose to 4.47 n January 200, and reached ts maxmum of 6.4 n June 200. These changes reflect massve stress on Chnese fnance from the US subprme mortgage crss n the th quarter whch led to the fnancal crss n May Recognton of Fnancal Stress Perod From prevous studes we know that there are 3 ways to recognzng stress perod. ) When FSI s at. tmes or twce the hstorcal averages we should consder t as the stress perod; 2) Lookng at the longtudnal hstory, when FSI s above a crtcal value wth ndex average(for example, 0%), government should pay attenton to t as a stress perod; 3) look at fnancal stress data durng fnancal crses n other countres, and when one country s FSI s above those 7

18 reference data cauton should be used. Ths paper uses the frst method above n detectng stress perod. In a normal dstrbuton we have P( x µ 2 σ) = 0.4, meanng that t s statstcally sgnfcant for FSI to reach a value twce of ts hstorcal value. P( FSI µ > 2σ ) = 0.0 (4) FSI FSI Hence we have ( FSI µ ) > 2σ, and FSI FSII t FSI ( FSIt µ = 2σ FSI FSI ) () Hence, FSII t s our ndex for recognzng fnancal stress perod. In formula (), when FSII t > 0, we can consder ths perod as stress perod and approprate polces should be taken. When < FSII t < 0, fnancal stress s at a normal range, and when FSII t < fnancal stress s relatvely low. FSIIt Fgure 4. The changes of FSIIt Fgure 4 ndcates changes n fnancal stress perod recognton ndex from January 2004 to March 20. From ts trend we see that there s a systematc stress to Chnese fnancal stablty from global flow of funds stress, domestc captal market stress, stress from holdng US bonds, exchange rate apprecaton, and ncrease n foregn exchange rates. From January 2004 to August 2008, the stress level was relatvely low, but from August 2008 FSIIt was gong up and has been greater than 0. In 200 and afterwards, FSIIt contnues gong up and t has become a hstorcal peak prce n May to August, 200. That s, fnancal stress has been ncreasng and governmental attenton should be pad. 8

19 6. Concludng Remarks Ths paper constructs a statstcs observaton system of GFF that s based on the mechansm of GFF, referencng FSIs that s released from IMF. It (GFFS) s promsng for not only buldng the econometrcs model for long-term analyss usng ths statstcs observaton system, but also made the trend ndex and fnancal stress ndex. We can test the strength of the shock to the fnancal system by the synthetc element of external flow of funds and usng the trend ndex and fnancal stress ndex, and also can observe the stuaton of short-term fnancal stablty. By the vewpont of global flow of funds, we concluded some ponts that fnancal stress n Chna s ncreasng after the U.S. fnancal-crss n 2008, through observng the changes n fnancal stress ndex. It turns out that perod of especally May, 200 to September 200 s the tme whch the Chnese polcy authortes should look out for. However, two subjects stll reman n ths research. One s a problem of the practce to DI and CI ndex. Although ths paper examned the methodology of how to create DI and CI by the vewpont of methodology, t hasn t done the actual tral calculaton yet. In fact, ths needs to make a database for makng DI and CI becomes a necessty. The other one s creatng the measurement model whch can estmate FSI. Due to restrcton n word lmt, ths paper only conducted descrptve-statstcs analyss on the stuaton of Chna fnancal pressure usng FSI. However, based on FSI, t s necessary to buld a model whch can explan change of FSI and to make the effectve early warnng system n order to expand the functon of FSI. References Abad A (2003). Early warnng systems: A survey and a regme-swtchng approach. IMF Workng Paper 03/32. Washngton DC: Internatonal Monetary Fund, February, pp. Andrew Bern and Catherne Pattllo (). Are Currency Crses Predctable? A Test, IMF Staff Papers, Vol. 46, No. 2, pp Balakrshnan, Dannnger, Elekdag,and Tytell (200). The Transmsson of Fnancal Stress from Advanced to Emergng Economes, IMF, Workng Paper No. 0/33. Crag S. Hakko and Wllam R.Keeton (200). Fnancal stress: What Is It, How Can It Be Measured, and Why Does It Matter? Federal Reserve Bank of Kansas Cty n ts Journal Economc Revew, Volume, Issue (Q II), pp. -0. Economc and Socal Research Insttute (ESRI, 200), Reexamnaton of the Economc Trend Index ( n Japanese ), Natonal Prntng Bureau. Frankel and A.Rose, Curreney(6). Crashes n Emergng Markets:An Emprcal Treatment, Journal of Internatonal Economcs, No.4.pp

20 Gracela L.Kamnsky,Saul Lzondo and Carmen Renhart (8). Leadng Indcators of Currency Crss.IMF Staff Papers, Vol.4, No., pp.-48. Illng, M. and Y. Lu (2003). An Index of Fnancal Stress for Canada, Bank of Canada Workng Paper , pp.-. IMF(2006). Fnancal Soundness Indcators Complaton Gude. Jeff Frankel and George Saravelos (200). Can Leadng Indcators Assess Country Vulnerablty? Evdence from the Global Fnancal Crss Harvard Kennedy School. Kamnsky, Gracela L. and Carmen M. Renhart (). "The Twn Crses: The Causes of Bankng and Balance-of-Payments Problems." Amercan Economc Revew, 8(3), pp Mattheu Bussere and Marcel Fratzsvuer (2002). Towards a New Early Warnng Ststem of Fnancal Crses, European Central Bank Workng Paper, No.4, Nan Zhang (2008). Global-Flow-of-Funds Analyss n a Theoretcal Model, Quanttatve Economc Analyss, Internatonal Trade and Fnance, Kyushu Unversty Press. Roberto Cardarell & Alessandro Rebucc (2007). Chapter 3. Exchange Rates and the Adjustment of External Imbalances, World Economc Outlook- Spllovers and Cycles n the Global Economy-, IMF, pp Roberto Cardarell, Selm Elekdag, and Subr Lall (200). Fnancal Stress, Downturns, and Recoveres, IMF Workng Paper, WP/0/00, pp.-4 Sachs, Jeffrey, Aaron Tornell, and Andres Velasco (6a). "Fnancal Crses n Emergng Markets: The Lessons from," Brookngs Papers on Economc Actvty vol. 27 (), NBER Workng Papers 76, pp Sachs, Jeffrey, Tornell, Aaron and Andres Velasco (6b). "The Mexcan Peso Crss: Sudden Death or Death Foretold?" Journal of Internatonal Economcs, vol. 4(3-4), pp Steven Radelet and Jefrey Sachs (8).The Onsef 0f the East Asan Fnancal Crss.NBER Workng Paper No Strauss-Kahn, Domnque, (2008), Letter from IMF Managng Drector Domnque Strauss-Kahn to the G20 Heads of Governments and Insttutons, November. 20