EXCESSLIQ UIDITYQ Q Q ANDINFLATIO NO O O INCH INA:H H H 2001-2010by
BijieJia
Athesissubm itted totheGraduateFacultyofAuburnUniversityinpartial fulfillm ent ofthe
requirem ents fortheDegreeofMasterofScienceinEconom ics
Auburn, Alabam aAugust 4, 2012
Keywords: inflation, excessliquidity, real GDP,m oneysupply, pricegap, twinsurplus
Copyright 2012byBijieJia
ApprovedbyHenryThom pson , Chair, ProfessorofEconom ics
JohnD. Jackson, Co-Chair, ProfessorofEconom icsHyeongwooKim , AssociateProfessorofEconom ics
ii
Abstract
SincetheChinesegovernm ent adm inistrated theReform andOpening-uppolicydesignedtowardstim ulating China's econom y in1979, Chinasteppedintoaperiodof rapiddevelopm ent.
Acceleratedgrowthof thetradesurplus andforeigncapital inflows ledtothecreationof excessliquidity, whichraisedinflationarypressures inthe econom y. Inadditiontoother factors, the
persistenceof Chineseinflationm ay belargelyattributabletoexcessliquidity. Thishypothesis isexam ined withthe?pricegapm easure ?inaregressionm odel that includestheEngle-Grangertest for
conintegration. Estim ation isbasedonavailablequarterlydatafortheperiodfrom 2001to2010.
iii
TableofContents
Abstract...................................................................................................................................................ii
Introduction............................................................................................................................................ 1
CHAPTER1. IntroductionofTwinSurplusinChina.........................................................................2
CHAPTER2. IntroductionofExcessLiquidity...................................................................................4
CHAPTER3. ModelingInflationasaFunctionofExcessLiquidity................................................ 7
CHAPTER4. OtherReasonsforInflationinChina..........................................................................12
CHAPTER5. Conclusion....................................................................................................................13
REFERENCES.....................................................................................................................................14
APPENDIX:.........................................................................................................................................16
1
Introduction
Theobjectiveof this thesis istoanalyzetheim pact of excessliquidityfor the2001-2010
periodonChineseinflation. Thisthesisism ade upof 5chapters. Chapter 1dealswiththeso-
called"twinsurplus"andhowit wasform ed. InChapter1analysisisgiventothereasonsbehind
excess liquidity. Estim ation m ethods thenfollowinChapter 2. Throughits m easurem ent we
noticethat when?toom uch m oney chasestoofewgoods?excessliquidityarises. InChapter 3,
weuseEngle-Ganger two-stepm ethod toprovideevidencethat excessliquiditycausedinflation.
After m odeling therelationshipbetweenexcessliquidityandinflationinChina, inChapter 4we
discuss relatedissues suchas the exchange rate, interest rate, andfoodsupplyshortage. In
Chapter5, wedrawtheconclusionandsuggest futureresearch.
2
CH APTERH H H 1.IntroductionofTwinSurplusinChina
TheChineseeconom y hasbeengrowingrapidlysincethecentral governm ent put inplace
m easures towardsliberalizingitsm arkets. In1979, theChinesegovernm ent devisedthenotionof
aSpecial Econom ic Zone(SEZ) tobeestablishedinthecoastal cities; its chief aim was to
stim ulate thecoastal areas. Foreigninvestm ent inSEZscouldenjoytariff reductions andother
preferential policiesdesignedtoattract foreigninvestm ent.
After theestablishm ent of SEZs, increasinginflows of foreigncapital brought alongthe
developm ent of China'seconom y, especiallyinthecoastal cities. Over theyears, foreigntrade
andforeigndirect investm ent (FDI) inChina hadbeenexpandingrapidly. Because China
typicallylikestokeepalargeforeignexchangereserve, Chinaisfacingatwinsurplusonboth
current andcapital accountsintheir balanceof paym ents. Thecurrent account surpluswas$12
billionin1990and$249.9billionin2006. Since2004, FDIinChinakept increasingannuallyat a
rateofover$60billion. Chinahasoneofthehighest levelsofFDIintheworld. SeeFigure1.
At thebeginningof thereform ist liberalizationpolicies, ForeignExchangeReserves(FER)
inChinaweresm all; it wasevennegativeat -$1.3billionin1980. From 1981to1989, China's
FERstayedat alowlevel. TheaverageFERperyearinChinaduringthisperiodwasonlyabout
$4.8billion, but it wasstill enoughtocoverasm all dem and forim ports.
Bythe1990s, FERincreasedrapidlybecauseof thedevelopm ent of international tradeand
increasedinflowsof foreigninvestm ent. Theaverageannual growthrateduring1991-1996was
m ore than60%. After2000, becauseoftheaccelerationofthetradesurplus, FERhadbeenrising
form ost ofthedecade.
3
Accordingtoanannual report from theNational Bureauof Statisticsof China(NBSC), the
rapidgrowthinim ports that beganin2001waslargelyduetoChina?s joiningtheWorldTrade
Organization(WTO). Duringthesam e period, exportshadbeenexpandingat afaster pacethan
im ports. FERhadexceeded$600billionin2004. October2006wasm ilestone forChinawhenits
FERexceeded$1trillion.
Finally, after 2006anduntil Septem ber 2008, foreignexchange reserves inChina had
reached$1.9trillion; inChineseper capitaterm s, this is roughlyequivalent to$1,500when
m easured against 2008populationlevels. BySeptem ber 2011, FERreachedto$3.2trillion.
China'sbalanceofpaym ents hasshownsurplusesinboththecurrent andcapital accounts. This
has arousedconcerns onthepart of theChinesem onetary authorityabout theriseof excess
liquidityanditspressuresonconsum er priceinflationinChina.
4
CH APTERH H H 2. IntroductionofExcessLiquidity
Higher exportsthanim ports over yearsinChinam ake for atradesurplusinthecurrent
account. Inaddition, lowdom estic consum ption weakens the velocityof m oney. Increased
ForeignDirect Investm ent FDIhasacceleratedthecapital inflowinChina.
Zhang(2009) notes that accum ulation has built upabundant liquidityfor China's dom estic
m arket. TheChineseauthoritiesstartedtoexpresstheir seriousconcernthat liquidityexpansion
inChinaisout of control, especiallywhenconsideringthepotential transm ission from excess
liquidityintoinflation.
Let usfirst considerthequantityequationofm oney
YPVM ?=? . (1)
Mdenotesstockof m oney, Vdenotesthevelocityof m oney, Yrepresentsoutput, andPisthe
pricelevel. Equation(1) describes arelationshipthat sim ply states that the stockof m oney
m ultiplied bythenum ber oftim es am oney unit that isusedforpurchasinggoodsandservices(V)
equalsreal output m ultiplied bythepricelevel.
If wewant totakeafurther lookintohowm oney supplychanges afect theeconom y's
inflationandoutput, (1)couldberewrittenas:
)ln()ln( YPMV ?=? . (2)
Then, weget
pyvm ?+?=?+? . (3)
Thesm all lettersdenotelogarithm s, and?denotesthefirst differences. Solving(3) for m? , we
canget
5
vpym ???+?=? . (4)
According to(4), m oney supplychanges equal the output difference plus the price level
differencelessthechangeinthevelocityofm oney.
AccordingtoPolleit andGerdesm eier (2005) for them easurem ent of excessliquidity, the
price gapm easurem ent for excess liquidityis understandable. Theyreference the workof
Hallm an, Porter, andSm all (1991) whodefinelong-term andshort-term pricelevel equations.
Thelong-term priceequationas (10) m easures theequilibrium pricelevel withthelong-run
m oney velocityandoutput level; likewise, theshort-term (or actual) pricelevel as(11) isbased
onthelong-runm oney velocitytrendandthepotential output level
*** tttt yvmp ?+= (long-term ) (5)
tttt yvmp ?+= (short-term ). (6)
*tp denoteslong-term equilibrium pricelevel, alsocalledpotential pricelevel; tp isfortheshort-
term pricelevel. Thedifferencebetween tp and *tp iscalledthepricegap,
)()( *** tttttt yyvvpp ?+?=? . (7)
Thepricegap )( *tt pp? couldbedecom posed intotwoparts: )( *tt vv? and )( * tt yy ? . If )( *tt pp? is
negative, it suggests that therewill beanupwardpressureoninflationinthefuture. When
tp< *tp, actual priceislower thanpotential price, it indicatesthat therewill beinflationlater;when
tp> *tp , actual priceis higher thanpotential price, sothat wewill expect adecreasepressureoninflationinthefuture.
Accordingto(5),
*** ttt ympv +?= (8)
Substitute *vin(7)with(8)
6
)()]([ **** tttttttt yyympvpp ?++??=? (9)
Sim plify (9), weget
tttttt ypvmpp ??+=? )( ** . (10)Equation(10)couldberewrittenas
YPMVPP ** // = . (11)where
)( *ttt pvm ?+ denotesthem oney supplyandtydenotesthereal output. Accordingto(10),
excessliquiditywill onlyhappenwhenthereistoom uch m oney inthem arket andtoofewgoods
at the sam e tim e. Moneyis the source of the price gap, andthus the im pact oninflation.
YPMV */ isthem easurem ent ofexcessliquidityhere.
Excessliquidityislikelytobeanim portant factor behindtheswift increaseininflationthat
Chinahasseeninrecent years. Inconclusion, excessliquidityoccurswhencurrent pricelevel is
greaterthanthepotential long-term pricelevel sincetoom uch m oney chasestoofewgoods.
7
CH APTERH H H 3. ModelingInflationasaFunctionofExcessLiquidity
Ruffer andStracca(2006)confirm that excessliquiditywasauseful indicator ofinflationary
pressureandsuggest that it isuseful tocom pare broadm oney supply(M2) withnom inal GDP
(NGDP) toestim ate theexcess liquidityof acountry. Theyestim ate avector auto-regressive
(VAR) m odel for 15countries. Yang(2010) investigates theinflationdynam ics andeffect of
excess liquidityinChina, andrevealedthat thequasi-m oney (M2) is them ain forcebehind
China'sinflation.
Excessliquidityhappenswhenthereistoom uch m oney chasingtoofewgoods, whichm eans
excessliquidityoccurswhenthegrowthof them oney supplyisfaster thanthat of GDP. Under
thisassum ption, weintroduceanindexof excessliquidity, theMarshallianKindex(MK). MK
indexisdescribedasthedifferenceof growthof m oney supplyandnom inal GDP. Wecanalso
usetheratioof them as(m oney supplyM2)/(nom inal GDP) todescribethelevel of howm uch
m oney supplym ay exceedoutput.
However, accordingtoYang(2010), thestandardMKindexisanindexof current levels, if
wewant toexploreitslong-term effect oninflation, wem ust doalog-transform ation of theMK
indexsothat wecanusethedifferencebetweenm oney growthrateandnom inal GDPgrowthrate.
Then, wecanseethat therelationshipbetweenexcess liquidityandinflationdepends onthe
coefficient oftherelativechangeMKindexinthem odel ofdescribinginflation.
Thetim e periodofourm odel isfrom 2001to2010. WithreferencetotheChineseeconom y,
however, onlythe data after 1992couldbeused. Yang(2009) pointedout that the m arket
dom inant pricem echanism inChinaworkedsince1992whenpricereform hadcom pleted.
8
Thelim itation herefor usingquarterlydataisthat therem ay beseasonality. Wetherefore
needtodom ake bothinflationindicator (CPI) andexcessliquidityindicator (ln[(M2)/(GDP)])
de-seasonedandstationary. Beforeweset upthecom pleted m odel, consider theexplanationof
som e im portant variablesinthism odel.
Theexcessliquidityindicator, alog-transform ation MKindexshouldbeexpressedas
EL=lnMKindex=lnNGDPM2 . (12)
Am ain m easurem ent of priceistheconsum er priceindex(CPI) that describesthegrowthrateof
pricelevel as inflationrate. Here, weusetheCPI for m ultiple item s for inflationindicators
becausewewouldliketodiscusstheim pact ofexcessliquidityongeneral inflation.
Weset uparegressionm odel heretotest therelationshipbetweeninflationandexcess
liquidityinChinafrom 2001to2010. From Figure1wecanseethechangeofinflationbasedon
consum er priceindexalongwithexcess liquidityinChinafrom 2001to2010(un-seasoned
quarterlydata).
Figure1suggeststheexcessliquiditym ay system atically afect theconsum er priceinflation
inChinaover them ost recent decade. Alinear regressionm odel will test for therelationship
betweenexcessliquidityandinflationinChina. Beforethat, weapplyX12-ARIMA(runbyE-
views7)m ethod torem ove theseasonalityfrom boththeoriginal CPI andELdata; andthen, we
will test forthestationaryoftheseprim ary variableseries.
WeuseAugm ented Dickey?Fullertest (ADF) test. ADFtest istestingfor whether therehas
aunit root. If thereisaunit root, weaccept thenull hypothesisthat theseriesisnon-stationary,
otherwisewereject thenull hypothesis. ForCPI, wehaveitsAR(1)process
ttt ecpiaacpi ++= ?110 . (13)
9
Weexpressthefirst differenceofCPIforADFtest as
tpj jtjtt cpicpiaacpi ???= ?? +?++=? 1110 . (14)
Im plying theADFtest for tcpi and tcpi? respectively, accordingtotheestim ation inTable2:
tcpi isnot stationary; tcpi? isstationary. Also, theresidual term isWNsincetheARCH(1) testinTable2islessthan1.6andDurbin-hisbetween-1.96and1.96.
ForEL, wehaveitsAR(1)process
ttt eelaael ++= ?110 . (15)Also, weexpressthe1
st differenceofELfortheADFtest
tpj jtjtt elelaael ???= ?? +?++=? 1110 . (16)
Im plying theADFtest for equation(15) and(16) respectively, accordingtotheestim ation in
Table2, tel? isstationary. Also, theresidual term hereisWN.
Sinceboth tcpi? and tel? arestationary, weset upasim ple regressionm odel asinflationis
regressedonexcessliquidity
ttt eelaa +?+= 10? . (17)
t? ism easured by tcpi? , tel? denotes growthrateof excessliquidity. AccordingtotheOLSestim ation (seeTable3), theestim ate of(17)is
tt el?+= 60.260.0? . (18)(0.17) (0.54)
TheEngleandGranger?stwo-stepm ethod generatesaregressionm odel basedonalong-term
relationshipofinflationandexcessliquidity.
Asim ple regressionequationbasedonnon-stationaryandde-seasonedvariablesis
10
ttt elaacpi ?++= 10 . (19)Evenif thevariables arenon-stationary, thelinear com bination of thesetwovariables m ay be
stationary, i.e. )0(~1 Ielacpi tt ? , and1ahereisunique. Ifthelinearcom bination herebetweencpi
andel isstationary, tcpi and tel arecointegrated. Table4istheOLSestim ation for(19).
Theerror correctionm odel (ECM) introduces inform ation from (19) into(17). Theerror
correctionregressionrequirestheresidual term in(19)isstationary. Checktheresidual term t? in
(19)withDFtest withnoconstant term intheEngle-Grangerregression
ttt ea +=? ?11?? . (20)SeeTable5, theEGtest (t-stat) is significant sincet <
T? =-2.67, and 1?<0, sothat t? isstationary. Thereisapresum ption that if
1?<0, the ttst ea ++= ?11)1( ?? isAR(1) stationary. Theresidual term in(20)isWN. Sothat
tcpi and tel inequation(19)arecointegrated.Theresidual inequation(19) is not WNbut stationary, suggestingthat residuals in(19) are
relatedtoeachother. Since 1?<0in(20) bytheEGtest, wecouldderiveaerror correction
m odel (EMC)
tcpi? = tttt aelaa ??? ++?+= ?1210 . (21)ReferringtoTable6, theOLSestim ation showstherepresentationof(21)is
128.003.162.0 ???+==? ttt elcpi ?? . (22)(0.28) (0.61) (0.03)
Astotheresidual term t?, sinceDWisapproachingto2andit passedtheARCHtest, it isWN.
Since t?and tel? arecointegratedinthiserrorcorrectionregressionas(22), thechangeofCPI
in(21)not onlydependsonthedifferenceof EL, but alsoonthelong-rundynam ic equilibrium
CPI. The residual term 1?t? brings the relationshipam ong levels intothe error correction
regression(21).
11
Thelong-term im pact from excessliquidityoninflationcouldbem easured bythesum of
error correctionadjustm ent andtransitory. Theerror correctionadjustm ent is denotedbythe
negativeproduct ofcoefficient oferrorcorrectioncoefficient in(22)andthecoefficient of tel? in
(18). It istheadjustm ent in tel? towardstothedynam ic equilibrium .
Errorcorrection=-(-0.28)?2.60=0.73. (23)(0.03) (0.54) (0.02)
Thetransitoryterm isexpressedbythecoefficient of tel? in(22), sothat
Transitory=1.03. (24)(0.61)
Then, thelong-term im pact isderivedas:
Long-term im pact =errorcorrection+transitory= 0.73 +1.03=1.76 (25)
(0.02) (0.61) (0.61)
Accordingtoequation(25), thelong-term im pact from excessliquidityoninflationwill be1.76,
explainedas1percent increaseinthegrowthrateof excess liquiditym ay cause1.76percent
increaseininflationasalong-term im pact. Thestandarderror for thelong-term im pact derived
withtheruleoferrorpropagation 5.22 )( ??? ?????? +=??= .
12
CH APTERH H H 4.OtherReasonsforInflationinChina
Econom ists havebeenstudyingtheRMB?s under-valuationproblem asoneof theprincipal
reasonsbehindincreasedinflationarypressuresinChina.
Intheory, whenthereisinflation, central bankscouldcontrol them oney supplybyapplying
tighteningm onetary policy. Thestandardpolicyistoraisethebenchm ark interest rateinorderto
absorbexcessm oney incirculationor sell bonds. Priceswill therebydecreaseandcurrencywill
appreciate.
Figure1showstheinflationchangeinChinain2001-2010. Wenoticethat intheyears2006
and2007, China'sCPIkept increasingsteadily, until it reachedapeakof113.16inthe2ndquarter
of year 2008. Duringthisperiod, thesteepriseinoil pricesin2007hadbeendrivingupvery
largebillsneededinChina(Chinaisoneof theworld'slargest oil-consum ing countries). Also,
foodpricesinChinawereincreasingrapidlybecausetheglobal agricultural com m odities' prices
wereincreasingat asteeppacesince2003forstructural andcyclical reasons. Inparticular, prices
ofm eat andporkstayedinahighlevel, becauseChinesepeoplehaveporkveryoftenintheirdiet
andit wouldtaketim e toadjust toothertypesofm eat. Moreover, theSichuanEarthquakeandan
unexpectedwinter storm in2008createdahugeim pact ontransportationandpower supplyin
China, thus drivingupthe prices of som any fooditem s, for exam ple, pork, poultry, eggs,
vegetableoil anddairyproducts.
13
CH APTERH H H 5.Conclusion
InflationinChinahasbeenaconstant m acroeconom ic problem sincethetim e thereform ist
liberalizationpolicywasput intopracticebytheChinesegovernm ent in1979. Withtherapid
increaseof foreignexchangereservesandforeigndirect investm ent inflows, includingatightly
controlledexchangerate, excessliquidityhasbeenaproblem .
Atwinsurplus has shownupinChina's national accounts andhas rem ained over recent
decadesduetoaconstant international tradesurplus andstrongrates of inflowof investm ent.
EventhoughtheChineseCentral Banktriedtoraisethebenchm ark rate5tim es in2011, it does
not seem tohavethedesiredeffect onsterilizingtheexcessliquiditybecauseit hasattractedm ore
foreigninvestorstobuybondsinChina.
Accordingtothepresent resultsthereisasignificant im pact from excessliquidityoninflation.
Thispositiveeffect betweenexcessliquidityandinflationisstatisticallysoundandeconom ically
m e aningful. Macroeconom ic theorysuggestsinflationisrelatedtom oney supplyandthecapacity
of potential output. Resultssuggest that thepricegapasthem easurem ent of excessliquidityis
viable.
14
REFERENCES
Cole, Wayne(2012), Analysis: Oil creepstowardtopof Asia'seconomicworrylist, accessedonline.
Coibion, Olivier, YuriyGorodnichenko, andJohannesF. Wieland(2010), intheNBERWorkingPaperSeries, "TheOptim al InflationRateinNewKeynesianModels."
Flem ing, J. Marcus(1962). "Dom estic financial policiesunderfixedandfloatingexchangerates".IMFStaffPapers9: 369?379. ReprintedinCooper, RichardN., ed(1969). International Finance.
NewYork: PenguinBooks.
Friedm an (1960), Aprogram form onetary stability, NewYork: Fordham UniversityPress.
Hallm an, Porter, andSm all (1991), "IsthePriceLevel TiedtotheM2MonetaryAggregateintheLong-Run?", Am erican Econom ic Review, p. 841-858.
Mundell, Robert A. (1963). "Capital m obility andstabilizationpolicyunder fixedandflexibleexchangerates". CanadianJournal of Econom ic andPolitical Science29(4): 475?485. Reprinted
inMundell, Robert A. (1968). InternationalEconomics. NewYork: Macm illan.
Polleit, ThorstenandGerdesm eier, Dieter(2005), "Measuresofexcessliquidity", HfB?BusinessSchool ofFinance& Managem ent, workingpaperNO.65
Ravallion, Martin(2010), "Price Levels andEconom ic GrowthMakingSense of the PPPChangesbetweenICPRounds", PolicyResearchWorkingPaper5229.
RuferandStracca(2006),"What isglobal ExcessLiquidity, anddoesit m atter?", workingpaperseriesNO.696, EuropeanCentral Bank
15
Siklos, PierreL. andZhang, Yang(2010), "IdentifyingtheShocksDrivingInflationinChina",PacificEconom ic ReviewP204-223.
Thom pson, Henry(2011), International Economics: Global Markets andCompetition(3rdEdition), WorldScientificPublishing, Chap.12-13.
Thom pson, Henry(2011), "AnIntroductiontoTim e SeriesRegression"
Yiping, Huang, WangXunandHuaXiuping(2010), "What Determ ine China'sInflation", ChinaCenterforEconom ics Research, P8
Yang, Jisheng(2010), "Expectation, ExcessLiquidityandInflationDynam ics inChina", HigherEducationPressandSpringer-Verlag, P412-P418
Zhang, Chengsi (2009), "ExcessLiquidity, InflationandtheYuanAppreciation: What CanChinalearnfrom Recent History", TheWorldEconom ics , Vol.32, Issue7, P998-P1018
"Tim e-series Econom etrics: ConintegrationandAutoregressiveconditional Heteroskedasticity".Advancedinform ation ontheBankof SwedenPrizeinEconom ic SciencesinMem ory of Alfred
Nobel, 8October2003
Jia, Yueqing(2010), "Lewisturningpoint andChina'sFDIprospects", accessedonline.
"VARModel forMultivariateTim e Series", accessedonline.
"China'sforeignexchangereserves, 1977-2011", accessedonline.
"China'sTradePerform ance", accessedonline.
16
APPENDIX:
FIG .1:G G G CPIandELinChina:2001-2010
*Source: IFS, OECD, Econom ics Forum ofRenm in University(Database), Econom ics StatisticsDatabaseserviceprovidedbyEconom y Watch.com , Author'scalculation.
17
TABLE1: ADFtestforCPIandEL
Residual Termt-stat
tcADF+ ARCH(1) Durbin-h
tcpi? -3.90** -0.78 0.61
tel? -4.78*** -0.55 1.56
tcpi -3.06 0.93 3.67
tel 1.40 -0.53 0.45Note:"***","**"and"*"denotesignificant at 1%, 5%and10%significant levelsrespectively. Autom atic-based onAIC,
m axlags=9.
TABLE2:O LSO O O estimationfor(20)
0a tel?
t? 0.60(0.15) 2.60(1.51) R2.27ARCH(1).13DW1.08
18
TABLE3:Newey-WestStandardErrorsadjustmentfor(20)
0a tel?
t? 0.60(0.17) 2.60(0.54) R2.27ARCH(1).8DW1.92
TABLE4:O LSO O O estimationfor(22)
0a tel
tcpi -12.42(16.57) 95.84(13.73) R2.56ARCH(1)3.81
DW.33
TABLE5:EG G G G Test(25)
1?t?
t?? -0.34**(0.11) R2.18DW2.02ARCH(1)-0.55
F-stat .33EG-3.05
*Note: t?=-2.67, for50observationsat 90%confident level. EGtest usedt-stat ofcoefficient of 1?t? .
TABLE6:O LSO O O estimationforECM(26)
0a tel? 1?t?
t? 0.62(0.28) 1.03(0.61) -0.30(0.04) R2.42DW2.06ARCH(1)1.01