New Capital Estimates for China

NewcapitalestimatesforChinaCarstenA.HOLZ⁎SocialScienceDivision,HongKongUniversityofScienceandTechnology,ClearWaterBay,Kowloon,HongKongReceived5April2005;receivedinrevisedform13February2006;accepted13February2006AbstractDataonphysicalcapitalareanindispensablepartofeconomicgrowthandefficiencystudies.InthecaseofChina,economy-widefixedassetseriesareusuallyderivedbyaggregatinggrossfixedcapitalformation(netofdepreciation)overtime,andsectoral/ownership-specificseriesbycorrectingthelimitedofficialfixedassetdataavailable.Theseprocedures,tovaryingdegrees,ignorethat(i)grossfixedcapitalformationdoesnotequalinvestment,(ii)investmentdoesnotequalthevalueoffixedassetsnewlycreatedthroughinvestment,(iii)depreciationisanaccountingmeasurethatbearsnonecessaryrelationtochangesintheproductioncapacityoffixedassets,(iv)officialfixedassetdata,whereavailable,incorporatesignificantrevaluationsinthe1990s,and(v)“netfixedassets”donotmeasurethecontributionoffixedassetstoproduction.Thispaperderiveseconomy-widefixedassetvaluesfor1953–2003,correctingfortheseshortcomings.Itusesboththetraditional,cumulativeapproachandanew,sofarunexploredmethodofcombiningeconomy-widedepreciationvaluesandaneconomy-widedepreciationratetodirectlyyieldeconomy-widefixedassets.Thederivedfixedassettimeseriesareevaluatedinacomparisonwitheachotheraswellaswithseriesintheliterature,leadingtotherecommendationofaspecificchoiceoffixedassettimeseries.©2006ElsevierInc.Allrightsreserved.JELclassification:E22;C80;D24;O47;P23;P24;C43;E10;O40Keywords:Capital;Investment;Nationalincomeaccounting;Productionfunctionestimations;Chinesestatistics;Fixedassets;Measurementofeconomicgrowth1.IntroductionStudiesofeconomicgrowthandinquiriesintosector-orownership-specificefficiencylevelsrequiredataonphysicalcapitalstock.Economy-widefixedassettimeseriesforChinaareusuallyChinaEconomicReview17(2006)142–185⁎Tel./fax:[phone].E-mailaddress:[e-mail].1043-951X/$-seefrontmatter©2006ElsevierInc.Allrightsreserved.doi:10.1016/j.chieco.2006.02.004derivedbyaggregatingnetinvestmentovertime.Fixedassettimeseriesforsubsetsoftheeconomyareusuallyobtainedbycorrectingthelimitedofficialfixedassetdatathatareavailable.Yetacloserexaminationoftheseproceduresrevealsmajorshortcomings.Intheliterature,economy-widecapitalvaluesareobtainedbyaccumulatinggrossfixedcapitalformationlessdepreciationovertime,withallvaluesatconstantprices.1Thisignoresthat(i)grossfixedcapitalformationdoesnotequalinvestment,(ii)investmentdoesnotequalthevalueoffixedassetsnewlycreatedthroughinvestment,and(iii)annualdepreciationisanaccountingmeasurethatbearsnonecessaryrelationtochangesintheproductioncapacityoffixedassets.Themarginsoferrorduetothesethreeshortcomingsappearontheorderofapproximately−10%to+45%overestimationofinvestmentexpenditures,20%overestimationofeffectiveinvestment,andupto500%overestimationofdecommissionedfixedassetsbyapplyingdepreciationratherthanscraprates.Insectoralanalysis,of,forexample,industry,wherecurrent-pricefixedassetdataareavailableforsomeownershipgroups,thesignificantrevaluationsofthemid-1990sembeddedinthefixedassetdataareroutinelyignored.2Thisimpliesthatthefixedassetdataofthe1990sareprice-adjustedtwice,onceintheofficialdatathemselves,andthesecondtimebyresearchersthroughtheprocedurestheyhaveadoptedtoprice-“correct”theofficialdata.Somestudies,furthermore,relyonthevariable“netfixedassets,”anaccountingartifactthatdoesnotreflectthecontributionoffixedassetstoproduction.Thispaperfocusesoneconomy-widedata.ItclarifiesthemeaningoffixedassetsinChinesestatisticsandlaysoutthelogicalrelationshipsbetweenfixedassetsandrelevantothervariables.Thevaluesofthesevariablesarederived.Fixedassettimeseriesareconstructedbasedontwodifferentapproaches.Thefirstfollowsthetraditionalmethodofaccumulatinginvestmentovertime.Thesecondreliesonthenationalincome(andproduct)accounts(NIA);economy-widedepreciationintheincomeapproachtothecalculationofgrossdomesticproduct(GDP)isdividedbytheeconomy-widedepreciationratetodirectlyobtainfixedassets.Theoutcomeofthetwoapproachesisseveraleconomy-widerealfixedassettimeseriesforChina.Theseareevaluatedandarecommendationisofferedastowhichseriestochoose.Incontrasttothefixedassettimeseriesconstructedhere,thoseintheliterature,apartfrombeingconceptuallyflawed,tendtoexhibitlowerlong-rungrowthrates,adifferentgrowthpatterninthereformperiod,andanoftenimplausiblerelationshiptoGDPovertime.32.FixedassetdefinitionTheterm“fixedassets”(gudingzichan[heji])inChina'saccountingsystemdenotesthesumof(i)netfixedassets(gudingzichanjingzhi),(ii)correctionstofixedassets(gudingzichanqingli)1See,forexample,GregoryChowandLiKui-Wai(2002),AlwynYoung(2003),WangYanandYaoYudong(2003),orWuYanrui(2004).GregoryChowandLiKui-Waifurtherincludeinventoryinvestment,andWuYanruidoesnotpreciselyspecifytheinvestmentvariableheuses.GregoryChow(1994)relieson“accumulation,”avariableoftheearlierusedMaterialProductSystemthatapproximatelycorrespondstogrosscapitalformation,andseemstoassumeadepreciationrateofzero.2See,forexample,ChenKuanetal.(1988)orGaryJeffersonetal.(1992