Re-examination of the Theoretical Model for Global-Flow-of-Funds Analysis*

1. 　I nt r oduc t i on

Theconceptoftheglobal-flow-of-funds,theframework ofanalysis,and theoreticalmodel fortheglobal-flow-of-fundswerebroughtup in 2005.¹⁾　Afterthat,wehavemadesomeimpor- tantadviceaboutthemodeloftheglobal-flow-of-fundsfrom internationalsociety and research seminarduring theseyears.　In thispaper,wewillclarify theconceptoftheglobal-flow-of- fundsand build theframework ofanalysisfortheglobal-flow-of-fundsanalysis.　And wewant to solvethemathematicalstructureofasimultaneousequationssystem in theanalytictheoretic model.　Also,according to theeconomictheory,wewillexaminetheeconomicmeaning of endogenousvariablesand exogenousvariablesin themodel.

Gl oba l - Fl o w- of - Funds Ana l ys i s *

Na n Zha ng**

(Received on October30,2008)

Abstract

Thispapercorrectsthemodeloftheglobal-flow-of-fundswhich wasbuiltin 2005.　Also, itpresentsamorespecificideaoftheglobal-flow-of-funds,and explainstheoutlinesof theoreticalanalysisfrom threedifferentviewpoints,thebalanceofsavings-investment,current balanceflows,and internationalcapitalflow.　Then thispaperconsiderstheissuesthatarisein interpreting and estimating multiple-equation models.　Specialcharacteristicsofmodelof global-flow-of-fundsarealso examined.

Key Words:BalanceofSavings-Investment,CurrentAccountFlows,InternationalCapital Flow,Simultaneous-EquationsModel,Two-StageLeastSquares,Three-StageLeastSquares

* Thisresearch wassupported by thegrants-in-aid forscientificresearch (ScientificResearch C,　 19530228).

** Iwould particularly liketo thank ProfessorYu Bin (theUC Berkeley DepartmentofStatistics)for helping when Iworked in UCB asaVisiting Scholar.And Iwould liketo thank ProfessorPaulRuud (theUC Berkeley DepartmentofEconomics)forthevaluablecommentsabouteconometrictheory. Also,Iwould liketo say thanksforexcellentadviceand alotofnew ideafrom Berkeley Statistics Weekly Seminars.

1)　Nan Zhang,2005,TheGlobalFlow ofFundsAnalysisin Theoryand Application,Mineruvula-shobo.

Theglobal-flow-of-fundsareflow offundsthatrelatesto domesticflowsand international capitalflows.　Theanalysisofglobalflowsoffund isan analysiswhich showsthecharacteris- ticsand thestructurein theflowsoffund,includestheflowsofalldomesticfundswith investment-savings,linkscurrentbalance,and connectsinternationalcapitalflows.

Thefinancialmarketsindicatethedebtsand creditsoffundsasawholeplusthetotalproc- essoffinancialliquidity.　Investigated morecarefully,itemsoffinancialmarketsinclude inflowsofdomesticfunds,overseasfundsby domesticsavingsand creditloansofbankson the sideoffund-sources(fundsinflow).　On theotherhand,fundssplitinto supply offundsto the domesticeconomy and fundsoutflow overseasin fund uses(fundsoutflow).　When theflow of fundsin financialmarketsistied up with theinternationalbalanceofpayments,theoverseas sectorwillbecomefund outflow excess(netcapitaloutflows)ifthecurrentaccountisin surplus.　 Conversely,thedomesticsectorwillbecomefund inflow excess.　Therefore,when thereal economy sideofthedomesticeconomy and overseasisanalyzed underan open economicsys- tem,thebalanceofsavings-investmentofthedomesticeconomy correspondsto thecurrent accountbalance.　According to thedynamicprocessofexternalflow offundsand thedefini- tionalequation ofaSystem ofNationalAccount,from theaccounting identity asfollows,

Wecan gettheequilibrium condition below with through arranging theaboveformula.

Therightsideofequilibrium condition showsCurrentAccount,and theleftsideofequilibrium condition showsthebalanceofsavings-investment,oritisalso called by NetFinancial Investment.　So wecan gettherelationship ofequilibrium condition equationsasfollows.

However,domesticnetfundsoutflow correspond with thecapitalaccountbalance,when therelationship between domesticand overseason thefinancialsideisexamined.　Therefore, relationsbetween thedomesticsavings-investmentbalance,thefinancialsurplusordeficit,the currentaccount,and theoverseasnetfund outflow willbeexpressed in thefollowing structural formulae.

Savings-Investmentand CurrentAccountBalance

(1) TheOverseasIncomeand ExpendituresBalance

(2) Y = + + +C I G EX−IM

S− =I EM−IM

S− =I EX−IM =NFI

S− =I ΔFA−ΔFL=EX−IM

EX−IM=(FO−FI)+ΔFER

TheFinancial-MarketsBalance

(3) Theupperformulaistransformed asfollows:

(4) Netfund supply to overseasbalance

(5) Theconstitution ofthenetoverseasfund flows

(6)

Notes: :financialassetsincrease,　 :financialliabilitiesincrease, EX:exports,　IM:imports,　FO:fund outflow,　FI:fundsinflow,

:Foreign exchangesreserves,　 :domesticfundsoutflow, :overseasfundsoutflow,　 :domesticfundsinflow,

:overseafundsinflow,　 (netoutflow ofoverseasfunds), (netinflow ofdomesticfunds),　CaA:Capitalaccount, DI:Directinvestment,　PI:Portfolio investment,　OI:Otherinvestment

Itisapparentthatthenetoverseasflow offunds correspondswith theCapital

& FinancialAccountin BalanceofPaymentby (4)and Capital& FinancialAccountisconsti- tuted ofFinancialAccountand CapitalAccount,thenetoverseasflow offundsas(6).　Allthe itemson therightofformulamadeinto netvalue,and indicatecourse,composition,and scale theglobalflow offund.　From (1)to (6)webuild theframework ofglobalflowsoffunds.　 And through the(6),weknow thatthechangesoftheglobalflowsoffundsaredetermined by foreign directinvestment,portfolio investment,and otherinvestment.　Thisishow weobtain theanalysisframework oftheglobalflow offundsfrom theabovestructuralformula.

2. 　Mat he mat i c al Pr i nc i pl e of t he Mot e l f or Gl obal - Fl ow- of - Funds

Threetechniquesaregenerally used forjointestimation oftheentiresystem ofequations: Three-StageLeastSquares(3SLS),Generalized Method ofMoments(GMM),and Maximum Likelihood.　ThemodelofGlobal-Flow-of-Fundswasbuiltby 3SLS method.　When Two- StageLeastSquares(2SLS)wasused forjointestimation oftheentiresystem ofequation, 2SLS assumed no correlation between errortermse₁and e₂in simultaneousequations.　On

FO_d+FO_o+ΔFER=FI_d+FI_o

FO_o−FI_o+ΔFER=FI_d−FO_d

NFO_o+ΔFER=NFI_d

(FO_o−FI_o)=DI+PI+OI+CaA

ΔFA ΔFL

ΔFER FO_d

FO_o FI_d

FI_o NFO_o=FO_o−FI_o NFI_d=FI_d−FO_d

(FO_o−FI_o)

theotherhand,3SLS ispresuming thatexistcorrelation between errortermse₁and e₂ofsimul- taneousequations.

Asthepresumed method,2SLS ishow to solvetheequation ofeach ofstructuralequations separately.　Butcompared with thismethod,3SLS isthemethodsforpresuming simultaneous equationsthattakeinto consideration alldirectionsofasimultaneous-equationssystem using thevariance-covariancematrix oftheerrorterm between equations.　Intuition would surely suggestthatsystemsmethod,3SLS isto bepreferred to single-equation methods2SLS.　The estimatorofpresumption thatwasby 3SLS isaconsistentestimator,and when thedisturbance termsofeach structuralequation havecorrelation,itbecomestheestimatorofeffectivepre- sumption moreasymptotically than 2SLS

Webegin with aslightly abstractlinearmodel

(7) WhereYisan observablen´1 random vector,Xisan observablen´prandom matrix,and b isan unobservablep´1 parametervector.　Thea_iareIID with mean 0 and finitevariances²; e israndom errors.　Thisisthestandard regression model,exceptthatXisendogenous,i.e., Xand aaredependent.　Conditionalon X,theOLS estimatesarebiased by

Thisissimultaneity bias.

Thestructuralform ofthemodelis

(8)

ThereareMequationsand Mendogenousvariable,denoted .　And thereareKexoge- nousvariable, ,thatmay includepredetermined valuesof aswell.　Thefirst elementofx_twillusually betheconstant1.　Finally, arethestructuraldisturbances.

In matrix terms,thesystem may bewritten Y_t= +α X_tβ ε+ _t

(X X′ )⁻¹X E′ ( |α X).

γ γ γ β β ε

γ γ

11 1 21 2 1 11 1 1 1

12 1 22

y y y x x

y y

t t m tm t k tk t

t t

+ + ⋅⋅⋅ + + + ⋅⋅⋅ + =

+ 22 2 12 1 2 2

1 1 2 2

+ ⋅⋅⋅ + + + ⋅⋅⋅ + =

+ + ⋅⋅⋅ +

γ β β ε

γ γ γ

m tm t k tk t

m t m t mm

y x x

y y y

#

ttm+β₁mxt₁+ ⋅⋅⋅ +βkmxtk =εtm

y₁,⋅⋅⋅,y_m x₁,⋅⋅⋅,x_k y₁,⋅⋅⋅,y_m

ε_t1,⋅⋅⋅,ε_tm

y y y_{m t}

m m

m m mm

1 2

11 12 1

21 22 2

1 2

[

]

⋅⋅⋅

⋅⋅⋅

⋅

⋅

⋅⋅⋅

⎡

⎣

⎢⎢

⎢⎢

⎢⎢

γ γ γ

γ γ γ

γ γ γ

⎤⎤

⎦

⎥⎥

⎥⎥

⎥⎥

or

　　　　　　　 (9)

Each column oftheparametermatricesisthevectorofcoefficientsin aparticularequation, whereaseach row appliesto aspecificvariable.　Theeconomictheory willimply anumberof restrictionson Gand B.oneofthevariablesin each equation islabeled thedependentvariable so thatitscoefficientin themodelwillbe1.　Economictheory may also imposeothertypesof restriction on theparametermatrices.

To handlesimultaneity bias,economistswould estimate(9)using instrumental-variables regression,also called Two-StageLeastSquares(2SLS).　Themethod requiresan n ´qmatrix ofinstrumentalorexogenousvariables,with n>q ³p.　Thematrix willbedenoted Z.　The and need to beoffullrank,qand prespectively.　Ifq>p,thesystem isover- identified.　Ifq=p,thesystem isjust-identified.　Ifq<p,thecasewhich isexcluded by assuming q³p,thesystem isunder-identified.　Parameterswillnotbeidentifiable.

Fortheexactly identified equation,IndirectLeastSquares(ILS)providesaconsistent estimator.　In theusualcase,theequation to beestimated willbeover-identified,so will havemorerowsthan columnsand cannotbeinverted.　Themethod of2SLS istheusual alternative.　The2SLS method consistsofusing astheinstrumentsforY_jthepredicted values in aregression ofY_jon allthex¢s:

The2SLS estimatoris,thus,

(10)

In thematrix [ ],which has columns,allcolumnsarelinearfunctionsoftheK columnsofX.thereexist,atmost,Klinearly independentcombinationsofthecolumnsofX.　 Iftheequation isnotidentified,then isgreaterthan K,and [ ],willnothavefull column rank.　In thiscase,the2SLS estimatorcannotbecomputed.

When using 2SLS,itisassumed thatthereisno correlation　between thestructuraldistur-

+

[

]

⋅⋅⋅

⋅⋅⋅

⋅

⋅

⋅⋅⋅

⎡

⎣

⎢⎢

⎢⎢

⎢ x x x_{k t}

m m

k k km

1 2

11 12 1

21 22 2

1 2

β β β

β β β

β β β

⎢⎢

⎤

⎦

⎥⎥

⎥⎥

⎥⎥

=

[

]

′ + ′ = ′ y_tΓ x_tΒ ε_t

′

Z X Z Z′

′ X Z_j

ˆ ( )

Y_j=X⎡⎣ X X′ ⁻¹X Y′ _j⎤⎦ =XP_j

ˆ ˆ ˆ ˆ

, ˆ

δ_{j SLS} ^{j j j j}

j j j j

j j j j

Y Y Y X X Y X X

Y y

2 X y

1

= ′ ′

′ ′

⎡

⎣⎢

⎢

⎤

⎦⎥

⎥

′

′

⎡

⎣

⎢⎢

− ⎤

⎦⎦

⎥⎥

ˆ ,

Y X_{j j} M_j+K_j

M_j+K_j Y Xˆ ,_{j j}

bancese₁and e₂in two simultaneousequations.　On theotherhand,Three-StageLeast Squares(3SLS)isthemethod thatexplicitly considerthecorrelation　between thestructural disturbancese₁and e₂in equation-by-equation.　Forestimation oftheentiresystem ofequa- tions,wecan use3SLS.　Theformulaof3SLS isasfollows.

2) (11)

Forthisestimatorto beavalid instrumentalvariable(IV)estimator,Zellnerand Theil(1962) haveeversuggested³⁾thatusing thedisturbanceswhich werecomputed by 2SLS estimatorto estimates_ij.　Thatis,compute2SLS estimatorsd_iin each equation from (11),and calculate vector aftersubstituteitfor(9)formulas,then,calculatetheestimators of . Therefore,the estimatoristhusgiven asfollows.

A.　EstimateP⁴⁾by OLS and compute foreach equation.

B.　Compute foreach equation;then

　　 ⁵⁾ (12)

C.　ComputetheGLSestimatoraccording to (11)and an estimateoftheasymptotic covariancematrix according to (12)using and .

By showing thatitthe3SLS estimatorsatisfiestherequirementsforan IV estimator,we haveestablished itsconsistency,and willusethe3SLS to estimatethemotelofglobal-flow-of- fundsin thenextsection.

3. 　Bui l di ng t he Mode l of Gl obal - Fl ow- of - Funds

Themotelofglobalofflow offundsisbased on IS-LM theory and generalequilibrium theory,refersto theview ofMundell-Flemingmodel,and iscreated by theform ofasimultane- ousequation.　Creation ofthismodelhastwo purposes.　Oneofthepurposesisrelated to the immanentfactorsofmany economicvariables.　Itestimatesthequantity relation between changeofthesavingsand investment,exportand import,externalflow offundsand economic growth in certain country,and therearein reflecting thestructuralchangesoftheflow offunds

ˆ ˆ ( ) ˆ ( )

δ_3SLS =⎡⎣Z′ ∑ ⊗⁻¹ Ι Z⎤⎦ ′ ∑ ⊗⁻¹Z ⁻¹ Ι y

ε_i(i= ⋅⋅⋅1, m) s_ij σ_ij

δˆ_3SLS

Yˆ_j ˆ

δ_j,_2SLS

ˆ ( ˆ ) ( ˆ )

σ δ δ

ij

i i i j j j

y Z y Z

= − T′ −

Zˆ Σˆ

2) ○× isKroneckerproduct.

3) A.Zellnerand H.Theil,1962,“Therr-stage,Least-squares:SimultaneousEstimation ofSimultaneous Equations,”Econometrica,vol.30,pp.54–78.

4) Pisreduced-form coefficientmatrix.

5)Tisobservations.

in amedium-to-long period oftime.

Anotherpurposeisto analyzetheinfluencing mechanism ofan economicpolicy and pol- icy coordination in acertain internationalarea.　Thechangeofinternationaltradeand financial transaction isreflected in global-flow-of-funds.　In thisway,itisnecessary to investigatethe influencesofthefiscaland monetary policieson global-flow-of-funds.　In thispart,weattempt to createthetheoreticalmodelofglobal-flow-of-fundsbased on theanalysisframework of global-flow-of-funds.

3.1　Description ofthemodelbyEconomics

When wewillexplorethestructuralchangeoftheglobal-flow-of-funds,itnotonly need to considerthefactorwhich realeconomy,such assavingsand investment,butalso need to observethefactorofthefinancialmarket,thatinvolving theinterestrates,thechangeofshare price,and theforeign exchangerate,etc.　In theglobalization ofworld trade,wealso need to considertherelationsofinterdependenceoftheinternationaltradeand thefinancialtransaction from theview oftheglobal-flow-of-funds.

According to theanalysisframework which showed theformulafrom (1)to (6)in thispaper, wedesigned aflowchartofthemotelasfigure1.　Figure1 showed themechanism oftheglobal- flow-of-funds.　Theellipsein thefigure1 show theendogenousvariablesand therectanglein thefigure1 show exogenousvariablesin themotel,and thearrow show therelationship between theendogenousvariablesand exogenousvariables.　According to theframework ofanalysisof Section 1,wetry to look into attheglobal-flow-of-fundsfrom threedifferentviewpoints,that is,thebalanceofsavings-investment,currentbalanceflows,and internationalcapitalflow.

Figure1:　Theflowchartforthemotelofglobalflow offunds

In thefirstviewpoint,wedesigned theequation ofSavings-Investment.　Thechangeof DisposableIncome(DI)can influencethesavings(S),themoreDIishigh,and themoresav- ingswillincrease.　Wealso wanted to usefinalconsumption expenditurein theprevious period (C_t–1)to observetheinfluenceon savings.　According to thetheory ofconsumption,the savingsin thisperiod willreduceifconsumption expenditurein thepreviousperiod increased.　

Wealso can know thatonemoreimportantfactorwhich interestrates(R)work upon on the changeofsavings,and iftheinterestrateincreases,thatmeansthemonetary tightening,so the amountofconsumption willreduce.

Thefactorswhich affected investment(I)weresetup aseconomicgrowth (Y),government expenditure(G),and interestrate(R).　WeusetheDY_t–1astheexogenousvariablesofeco- nomicgrowth,thatisthedifferenceofGDP in thepreviousperiod (Y_t–1– Y_t–2).　According to theprinciplesofacceleration ofinvestment,changeofthequantity ofproduction orfinal demand willmakean acceleration in thechangeofinvestmentgrowth,so wecan considerthe plusrelationship between Iand DY_t–1.　J.M.Keynesstudied controlofinvestmentand classi- fied thecasewhich controlofinvestmentispossible,and thecaseofbeing impossible,by con- trolling an interestrate.　Ifin thecasewhich can'tcontrolinvestmentby interest,itneeds governmentto increasegovernmentexpenditureto push investment.　So in orderto explain changeofinvestment,governmentexpenditure(G)ismadeinto an explaining variable,and itis theplusrelation between Iand G.　Also wecan know thatIhasanegativerelation with the interest(R).

In thesecond viewpoint,weneed to connectwith Sequation and Iequation from theview- pointofthecurrentbalanceflows,and derivetheImport(IM)equation and theExport(EX) equation from thesystemsoftheglobal-flow-of-funds.　When wewillanalyzethechangeof theimportand exportin theglobal-flow-of-funds,wedivided into thedemand factor(GDP), thepricefactor(ImportPriceIndex,IPI),thefactoroftheworld economy factor(theworld economy otherthan country which becomescandidateforanalysis,theWGDPcan stand for that),and RealEffectiveExchangeRate(REER)astheexogenousvariables.　And weused theseexogenousvariablesand presumed theeach elasticity regarding theIMand EX.　In the IMequation,wecan know thattheIPIhasanegativerelation with theIM,and theY(=GDP) hasplusrelation between Yand IM.

Thereisthephenomenon in which both ofexternalassetsand foreign liabilitieshavebeen growing largein theglobalflow offundssince2000s.　By looking from thethird viewpoint,in orderto observethecontinuousadjustmentprocessto thebalancefrom theimbalanceofan

internationalcapitalflows,itisnecessary to build amodelnotonly on anetbase,butalso on agrossbasis,wehaveformed theequationsoftheinternationalcapitalinflow and domestic capitaloutflow on thegrossbase.

From thethird viewpoint,weexaminethecausalrelationship oftheinternationalcapital flow.　Asexplanatory variablesoftheinternationalcapitalinflow equations(FI),weused GDP growth rate　in thepreviousperiod (YR_t–1),priceearningsratio (PER),foreign directinvest- ment(FDI),thedifferenceoftheinterestratewhich own country and aforeign country (DR), and dummy variable(D).　According to thetheory offinancialinvestment,theinternational capitalwhetherflow into acountry aremainly decided by two factors,thatis,therisk and return offinancialinvestment.　Wecan know FIhasplusrelation with YR_t–1,PER,and FDI.　 Also weconsiderapositiverelation between DRand FI,becausean interestrategap with afor- eign country willariseiftheinterestrateofown country falls,thecapitalflowswillflow out from theown country to aforeign country.　In otherhand,iftheinterestrateofown country riseand itishigherthan aforeign country,theinternationalcapitalflowswillflow in into this country.

According to formula(6)ofSection 1,sincethecapitalinflow mainly consisted ofFDI, International-Portfolio-Investment(IPI),and externaldebt(ED),wesetup theFDIfunction,the IPIfunction,theEDfunction,etc.from thedynamicprocessoftheglobal-flow-of-funds.

In orderto observetherippleeffectbetween internationalmarketand thefiscaland mone- tary policy ofeach country,weadded thePERfunction,theinterestratefunction (R),and exchangerate(EXR)in thismodelthatbased on themechanism ofprofitability and risk regard- ing to theinternationalcapitalmovement.　Weused R,YR_t–1,EXR,and theratio offoreign liabilities(risk)astheexplanatory variablesofthePERfunction.　TheR,EXRand riskhavea negativerelation with PER,and theplusrelation between YR_t–1and PER.

Weuserealmoney supply (RM),externalnetfinancing (NFI),benchmark interestratesof centralbank (RCB)and YR_t–1astheexplanatory variablesto explain thechangeofR.　Accord- ing to thetheory ofMundell-Fleming,wecan considerRM and NFIhaveanegativerelation with R,and BCBand YR_t–1haveaplusrelation with R.　Becausethecostofafinancialtransac- tion willdecreaseifan amountofmoney supplied increases,an interestratewillbegoing down.　

Also an interestratewillfall,sincedomesticsuppliesoffundswillincreaseifoverseacapital inflowswereraised.

Exchangerate(EXR)isavery importantfactorin theglobal-flow-of-funds,and wesetDR, PER,FIand EXasexplanatory variablesin theEXRfunctions.　Sinceincreasein valueofan

exchangeratediffersfrom theincreaseconceptofothereconomicvariables,firstweneed to clarify theconceptofrevaluation ofan exchangerate.⁶⁾Asthe(a)ofFigure2 shows,the exchangerateisto berevalued,sincetheinflow offoreign capitalwillincreaseiftheinterest rateofitsown country ishigherthan otherforeign country.　So wecan know theplusrelation between theDRand EXR.　And asachangeofstock marketalwaysshows,sinceastock price index movesin thedirection alwayscontrary to an exchangerate,so wecan considerPER hasa negativerelation withEXR.　By observing the(b)ofFigure2,wecan know EXRhasaplus relation between FIand EX.　IftheIScurvewillbegoing down,theexportslump,thecapital inflowswillalso reduceasapaymentwhich exported thegoodsand services.　Then,the exchangeratewilldevalue.

6) Forexample,itwas1 dollar= 97 yen on yesterday,butifitbecome1 dollar= 95 yen today,itwill becalled therevaluation oftheyen,and ifitis1 dollar= 99 yen today,itwillbecalled depreciation in yen.

7) Thisgraph refersto Mundell-Fleming modeland created it(RobertA.Mundell,1968,International Economics).

Figure2:　Determination ofan exchangeratein theglobalflow offunds⁷⁾

In thefunction ofchangesin theforeign-reservesassets(CRA),wesetcurrentaccount (CA),NFI,interestrateofcentralbank (RCB)and theU.S.federalfund interestrate(FFR)as explanatory variablesto explain thechangeofCRA.　CAhasaplusrelation with CRA,because currentbalancesurplusmakesCRAincrease,and currentbalancedeficitdecreasesCRA.　 According to formula(2)in theSection 1 ofthispaper,weknow thatCAand NFIareliketwo sidesofthesamecoin forthechangeofCRA.　IfCA< 0,and NFI= 0,CRAwilldecrease.　 ButifNFI> 0 (netinflow offunds),and itcan covercurrentbalancedeficit,thequantity of CRAwillbekept.　Also,ifNFI> 0,butitcan'tcovercurrentbalancedeficit,CRA willbe subtracted.　So wecan know therelationship between CRAand NFIwilldepend on thechange ofCA.　Wewantto useinterestrateofcentralbank (RCB)aspolicefactorto explain the changeofCRAIftheRCBgo up,itmeansexecution ofrestrictivemonetary policy,and money marketrateswillberaised.　An exchangerateisrevalued by this.　And CRA willdecrease, sinceexportwilldecreaseifan exchangeraterevalued.　So wecan know RCB hasanegative relation withCRA.

Wealso usetheU.S.federalfund interestrate(FFR)asan internationalfactorto explain thechangeofCRA,becauseFFR isavery importantfactorin theglobal-flow-of-funds.　The U.S.isakey currency nation,especially,in ChinaorJapan,morethan halfofforeign reserves arein theU.S.dollar-treasury bond.　According to themarketmechanism,ifFFRgoesup, CRAofthecountry containing theU.S.　Treasury bond willincrease,so therelation ofplus existsbetween FFRand CRA.

In orderto observethedomesticoutflow offunds(FO),weuseCRA,Profitfrom Invest- ment(PI),RCB,FFR,and dummy variable(D)asexplanatory variablesto explain thechange ofFO.　CRAarethedebtsofforeign country in regard to own country,and itcan becomea partofFO,so CRAhasaplusrelation with FO.　PImeansinvestthatincluded foreign direct investmentin foreign countriesand itisgained profitby investment,so wecan know theplus relation betweenPIand FO.　Then,weputFFRasan internationalfactorto explain the changeofFO.　FFRhasaplusrelation with FO,becauseifFFRgo up,theFO willincrease.

3.2　Basicstructureofthemodel

Thereare16 endogenousvariablesand 21 exogenousvariablesin themodeloftheglobal- flow-of-funds.　Thismodelreferred to theprincipleofApplied GeneralEquilibrium model.　 Thismodeltakesthechangein theformation ofexpectation and risk,itwith alag structureor immanentrelevantfactorsofmany economicvariables,and explainsthestateofthefundsflow

in thecontinuousadjustmentprocessto thebalancefrom imbalance,and in amedium-to-long period oftime.　Italso belongsto akind ofdynamicmodel.　A system ofsimultaneousequa- tionsisamodeloffinancialmarketequilibrium,consisting ofthefollows.

Through thosesimultaneous-equationsmodels,wewantto observesystematically thestruc- turalfactorand cyclicfactorin flow-of-funds;how theprofitfactorand therisk factoraffect internationalcapitalflows,and how thechangeofthepattern offundsflow affectdomestic economy growth.　In orderto perform prediction ofafutureflow-of-fundstrend,and thesimu- lation ofthepolicy effectattheend ofthemodel,threedefinitionalequationscalled NetFlow ofFunds,CurrentAccountBalance,NationalIncomeidenticalequation areformed.

StructuralEquations (1) Savings　　 (2) Investment　　 (3) Import　　 (4) Export　　 (5) Capitalinflow　　

(6) Foreign directinvestment　　

(7) International-portfolio-investment　　 (8) Externaldebt　　

(9) Expected stock profit　　 (10) Marketinterestrate　　 (11) Exchangerate　　 (12) Reserveasset　　 (13) Capitaloutflow　 IdentitiesEquation

(14) NetFund flow definition　　 (15) Currentbalancedefinition　　 (16) GDP identicalequation　　

Themodeloftheglobal-flows-of-fundsiscreated from threeviewpoints,thatis, investments-savingsbalance,currentbalanceflow,and internationalcapitalflow.　First, savings-investmentequation isformed from thesideofdomesticsavings-investmentbalance.　 Moreover,webuiltImportand Exportequation thatconnected with savings-investmentequa-

S_t=b11+b DI12 _t+b C13 _t−1+b R14 _t+ε_t1

I_t=b21+b22ΔY_t−1+b G23 _t+b R24 _t+ε_t2

IM_t=b31+b IPI32 _t+b Y33 _t+ε_t3

EX_t=b41+b REER42 _t+b WGDP43 _t+ε_t4

FI_t=b51+b YR52 _t−1+b PER53 _t+b FDI54 _t+b DR55 _t+b D56 _t+ε_t5

FDI_t=b61+b Y62 _t−1+b PI63 _t+b REER64 _t+ε_t6

OPI_t=b71+b rbond72 _t^US+b risk73 _t+b R74 _t+ε_t7

ED_t=b81+b RFL82 _t−1+b CA83 _t+b R84 _t+ε_t8

PER_t=b91+b R92 _t+b YR93 _t−1+b REX94 _t+b risk95 +ε_t9

R_t=b101+b102RM+b103NFI_t+b104RCB_t+b YR105 _t−1+ε_t10

REX_t=b111+b112DR+b113PER_t+b114FI_t+b115EX_t+ε_t11

CRA_t=b121+b CA122 _t+b123NFI_t+b124RCB_t+b125FFR_t+ε_t12

FO_t=b131+b CRA132 _t+b133PI_t+b134RCB_t+b135FFR_t+b136D_t+ε_t13

NFI_t=FO_t−FI_t CA_t=NFI_t+CRA_t Y_t=NFI_t+C_t+ +I_t G_t

tion from thesideoftradeflow.　And wearetrying to show thefeatureoftheglobal-flow-of- fundsin an internationalarea,when assembling astructuralequation.　In orderto follow the continuousadjustmentprocessoftheglobal-flow-of-fundstheinternationalside,webuiltthe internationalcapitalinflow equation and domesticcapitaloutflow equation ofthegrossbase which arefrom simultaneous-equations(5)to (13).　Then,webuiltidentitiesequation which from (14)to (16)forperforming thesimulation ofapolicy.　Each variablenameand classifica- tion in themodelisshown in theTable1.

Table1　ThelistofVariablesand Definition Classification Variablename

Variable

Endogenous ChineseGDP

Y

Exogenous Dif2 oflag ofGDP

Y2

Endogenous GrossSavings

S

Exogenous DisposableIncome

DI

Endogenous GrossInvestment

I

Exogenous FinalConsumption

C

Exogenous lag ofthecapitalstock

K(–1)

Endogenous Export

EX

Endogenous Import

IM

Endogenous Currentaccount

CA

Exogenous RealEffectiveExchangeRate

REER

Endogenous Exchangerates

REX

Endogenous One-yearloansinterest

R

Exogenous ConsumerPriceIndex

CPI

Exogenous Shenzhen B share

PER

Exogenous Changesin ReserveAssets

CRA

Exogenous Errors& omissions

CF

Endogenous Fund outflows

FO

Endogenous Fund inflows

FI

Endogenous Netfinancialinvestment

NFI

Exogenous Economicgrowth rate

YR

Endogenous Foreign directinvestment

FDI

Endogenous Portfolio investment

OPI

Endogenous Otherinvestment(liabilities)

OIO

Exogenous Interestpayment

RFL

Exogenous Liability ratio

Risk

Exogenous Interestratesofcentralbank

RCB

Exogenous Realmoney supply

MR

Exogenous Governmentexpenditure

G

Exogenous Profitfrom Investment

PI

Exogenous FederalFundsRate

FFR

Exogenous TotalofJapan,U S and Euro

WGDP