世界經(jīng)濟千年史

1、世界經(jīng)濟千年史礙2004年第54期（總第4皚70期）岸2004年9月28日荷蘭格阿羅寧根大學(xué)教授、世界經(jīng)濟奧千年史作者Maddiso把n先生在中國經(jīng)濟研究中心萬絆眾樓做了題為哀“暗世界經(jīng)濟千年史啊”拔的講座，集中討論了宏觀經(jīng)濟按計量歷史研究的發(fā)展進程。下扳面是他報告的主要內(nèi)容。壩宏觀經(jīng)濟計量歷史的發(fā)展經(jīng)歷扒了三個歷史階段：20世紀5奧0年代以來的扳“哎收入和財富國際研究協(xié)會時期靶”挨即IARIW時期（Inte扒rnational Ass白ociation for 皚Research in I哎ncome and Wea氨lth）；1820195辦0年的庫茲涅斯研究為代表的靶時期或哀“暗Kuzn

2、etsian時期皚”般；15001820年商人跋資本主義年代。50年代以來敗，宏觀計量的主要目的是提供埃可能的政策選擇以改善國家的盎增長表現(xiàn)以及進行國家間差異礙分析。我們現(xiàn)在有50年代以熬來世界大部分國家的增長和收耙入的官方估計。而在Kuzn巴etsian時代，數(shù)量經(jīng)濟把史學(xué)家在測量世界經(jīng)濟增長和隘量化增長因素方面取得了巨大凹進展。雖然仍有一些領(lǐng)域需要翱改進，但對這個時期經(jīng)濟發(fā)展扮大致輪廓沒有實質(zhì)爭議。相反骯，關(guān)于商人資本主義年代的世柏界經(jīng)濟的描述則存在很大分歧挨，作為代表的是亞當、斯密的笆樂觀看法和馬爾薩斯的悲觀論靶。耙20世紀50年代以來的IA氨RIW時代板雖然50年代以來標準國民經(jīng)柏濟核

3、算體系逐漸被采用，但是斑現(xiàn)存估計中仍有很多問題。使暗用蘇聯(lián)MPS體系（mate扒rial producti癌on system）的國家瓣增長被高估，需要做大量調(diào)整笆。非洲許多新生國家缺少必要阿的技術(shù)和資金進行國民收入增半長核算，雖然國際組織工作在跋一定程度上彌補了這種不足，阿但是非洲數(shù)據(jù)也需要做大量調(diào)氨整。還有1995年以來一些哎高收入國家開始引入享樂指數(shù)瓣（hedonic inde頒x）來反映產(chǎn)品質(zhì)量的改善，氨但是這么做并沒有充分的根據(jù)壩并且會高估經(jīng)濟增長，比如美暗國用新的計量技術(shù)重新進行1巴9291950年國民經(jīng)濟伴核算，結(jié)果使國內(nèi)生產(chǎn)總值(伴GDP)的增長率由2.6%挨增加到3.5%。最

4、后關(guān)于教矮育和知識的計量也還沒有成熟絆理論。稗另一問題是購買力平價轉(zhuǎn)換與挨國家間GDP水平比較。政府凹統(tǒng)計官員會提供真實價格計算凹的總產(chǎn)出和總支出數(shù)據(jù)，經(jīng)濟百學(xué)家、新聞記者和官員們都將矮其作為經(jīng)濟增長和波動的主要皚指標。購買力平價方法與用真擺實價格計量的目的是一樣的，邦即修正價格差異使真實產(chǎn)出和襖支出水平的有效比較成為可能拌。以匯率轉(zhuǎn)換的GDP與以購扒買力平價轉(zhuǎn)換的GDP區(qū)別非跋常大。以購買力平價計算的發(fā)般展中國家的GDP要大大高于懊以匯率轉(zhuǎn)換的GDP，差距可拜能達到3-5倍。同時發(fā)達國暗家GDP以匯率計算則容易被頒高估。更為明顯的一個例子是跋1950年中國和印度以匯率白計算人均GDP分別是$

5、85埃和$175（1990年價格愛），這兩個數(shù)據(jù)低得讓人難以凹置信。那么為何匯率轉(zhuǎn)換還如哎此經(jīng)常的被使用呢？一方面出百于無知和支持自己觀點的需要壩，另一方面許多發(fā)展國家不愿矮意接受購買力平價方法，因為般擔心這會不利于他們申請世行百的優(yōu)惠貸款和資助。這導(dǎo)致了白經(jīng)濟間比較分析的明顯錯誤，絆媒體經(jīng)常稱日本是世界第二大班經(jīng)濟雖然其GDP還不到中國叭的60，英國的政客們也始啊終相信英國的經(jīng)濟規(guī)模比中國艾大。稗18201950年的Ku疤znetsian時代案Simon Kuznets安教授對現(xiàn)代經(jīng)濟增長的研究成罷果，使可以數(shù)量化研究的時間般界限從20世紀50年代推進把到19世紀20年代。關(guān)于這擺段時期我們

6、現(xiàn)有的幾個結(jié)論是阿：加速的經(jīng)濟增長始于182把0年而非Kuznets認為鞍的1760年；經(jīng)濟學(xué)家對西敗歐各國研究表明西歐的崛起是氨同時而非交錯發(fā)生的；這段時八期快速發(fā)展加大了西方與其他哀國家的差距；Kuznets把研究推翻了Kondrati敗eff長期循環(huán)的說法和Sc瓣humpeter循環(huán)發(fā)展論扒點。事實上1820年以來的襖技術(shù)變革不是浪潮式的而是連骯續(xù)推進的，目前有足夠證據(jù)把擺18202001這段時間熬劃分為5個不同的發(fā)展階段。半其中19501973年是一個無可比擬的繁榮的黃金時敗代，這個時期世界GDP年均增長5，世界貿(mào)易年均增長擺8，人均收入有明顯趨同趨拔勢，大部分地區(qū)經(jīng)濟增長都快伴于美國

7、。1973年以后，世阿界經(jīng)濟增長明顯下降，各地區(qū)奧的差異加大，但是基于世界的骯角度，最近這個階段仍然是增懊長表現(xiàn)名列第二的階段。區(qū)分辦領(lǐng)導(dǎo)國家和跟隨國家對于研究八技術(shù)的動態(tài)傳播和趕超過程是傲非常重要的，伴“懊領(lǐng)導(dǎo)扮”敖國家指那些處于技術(shù)前沿的國爸家，而版“版跟隨翱”哀國家指勞動生產(chǎn)率較低的國家安。1500年以來共有4個領(lǐng)安導(dǎo)國家，16世紀的意大利，愛16世紀至拿破侖戰(zhàn)爭時期的按荷蘭，此后的英國和1890般年后的美國。頒對1820年以來英國、美國扮和日本的增長分析，最早考慮傲的是勞動力投入和生產(chǎn)率，戰(zhàn)安后資本成了一個非常重要的增巴長因素，之后又有觀點將安“傲人力資本澳”搬看作生產(chǎn)要素。Denis

8、o礙n擴展增長分析方法指出18笆20年以來英國、美國和日本襖經(jīng)濟增長表現(xiàn)出以下一些顯著邦特點：物質(zhì)資本大量增長，非熬居住建筑和機器設(shè)備增加非常礙明顯，伴隨在機器設(shè)備中加速吧的技術(shù)進步。教育水平大大提芭高，英國提高了8倍，美國和哀日本11倍。人均勞動投入在扒英國和日本下降了40，在俺美國下降了20。對外貿(mào)易芭占GDP的比重在英國由3白增至25，日本0.2至半13，美國2至10。敗自然資源的缺乏對增長并不構(gòu)矮成限制，人均土地在美國下降捌了14倍，英國和日本下降了耙4倍。能源的投入增長比較溫傲和，美國人均增長了3倍，英扳國6倍，日本8倍。礙15001820年的商人癌資本主義年代叭關(guān)于商人資本主義年代

9、（15唉001820）經(jīng)濟表現(xiàn)，艾在18世紀末就已經(jīng)存在兩種懊不同的解釋。亞當、斯密樂觀霸地認為，美洲大陸和新航線的凹發(fā)現(xiàn)給大型經(jīng)濟，國際貿(mào)易和版專業(yè)化的發(fā)展創(chuàng)造了新的機會癌，雖然因為對貿(mào)易的共同限制敗，這些機會不能被充分利用。爸馬爾薩斯則悲觀論則認為，經(jīng)擺濟表現(xiàn)取決于人口增長和固定骯的土地供給間的平衡，技術(shù)進壩步、資本形成和國際貿(mào)易專業(yè)瓣化的因素被忽略，因此只有通扳過災(zāi)難（戰(zhàn)爭、饑荒、疾?。鄄拍軐崿F(xiàn)平衡。這兩種觀點對敖立一直存在。此外悲觀的看法瓣有了一些其他的支持，LeR白oy Ladurie認為法骯國13001720年的經(jīng)疤濟是停滯的，真實工資論者則壩更加悲觀他們有人認為英國1斑820年

10、的生活水平比150哀0年下降了44等等。皚關(guān)于15001820不同搬國家和地區(qū)的經(jīng)濟表現(xiàn)，Ku扮znets在1965年提出凹了關(guān)于西歐人口率和人均GD敖P增長的一個非常有影響的假埃說，認為西歐發(fā)達國家150挨01750年可能的（并且柏或許是最大的）長期人均產(chǎn)出頒年均增長率為0.2%。對上絆述假說驗證表明，西歐150扳01820年人均年增長率骯為0.14，顯著小于Ku傲znets估計，美洲國家整哀體而言人均GDP增長快于西敖歐，非洲南部和北部的發(fā)展差般異比較明顯，亞洲國家整體而挨言收入水平是停滯的，日本的凹人均表現(xiàn)要好于中國和印度，伴但是非常明顯中國這個時期有伴著廣泛的增長，它在人口大量襖增長的

11、情況下生活水平并沒有啊下降并且GDP的增長與西歐翱一樣顯著。把在分析商人資本主義時代的增擺長時，增長分析的方法就不再班適用了。研究表明，人均勞動襖投入、資本和人力資本、知識把在這個時期都有增加，但是非罷常明顯的是全球化在這個時期阿起到了比以往更加重要的作用艾，相對來說全球化在這段時期瓣的作用比在20世紀還要重要敖。西方造船業(yè)和航海的巨大發(fā)昂展，使世界貿(mào)易在1500安1820年增長了20倍。歐胺洲國家還從殖民地，非洲奴隸皚貿(mào)易中獲取了很多利益。美洲斑國家則經(jīng)歷了生態(tài)、技術(shù)和人般口的轉(zhuǎn)型。引進新的農(nóng)作物提芭高了糧食的產(chǎn)量；馬和其他牲阿畜的引入則改善了交通條件；案歐洲各種技術(shù)的引入也有助于骯經(jīng)濟的發(fā)

12、展；歐洲疾病使2/愛3土著居民死亡，加上歐洲人艾和非洲奴隸大量涌入，改變了擺美洲的人口結(jié)構(gòu)。非洲和亞洲辦國家也引入了美洲農(nóng)作物。此擺外還存在大陸間的技術(shù)傳遞，芭歐洲向美洲輸出武器、工具、吧車輛、船和造船技術(shù)、印刷、阿文字、教育和政治經(jīng)濟機構(gòu)，案歐洲采礦技術(shù)在美洲應(yīng)用生產(chǎn)扒出大量金銀供歐洲與亞洲進行拌貿(mào)易，同時期歐洲從亞洲引入叭了紡織和陶瓷技術(shù)。班與亞當斯密和馬爾薩斯關(guān)于商拔人資本主義的對立解釋相對應(yīng)笆，關(guān)于現(xiàn)代化的根源有非常不跋同的看法。有一個學(xué)派認為現(xiàn)阿代經(jīng)濟增長源于工業(yè)革命，而襖之前則是幾個世紀的馬爾薩斯疤停滯。很多人支持這個觀點，班但是這些觀點在根本上是錯誤白的。事實上向現(xiàn)代資本主義過靶

13、渡經(jīng)歷了長時間的準備。首先芭是教育和知識的傳播由于印刷敖術(shù)的發(fā)明和推廣而變革；其次拌造船業(yè)和航海業(yè)到1820年盎也發(fā)生了變革，船只的設(shè)計、埃裝備和天文知識都大大改善，挨有了精確的航海指導(dǎo)等等。這奧些進展都是科研努力的結(jié)果。阿無疑，歐洲的這些發(fā)展是19伴和20世紀經(jīng)濟更快發(fā)展的前伴奏，歐洲的現(xiàn)代化不是一蹴而拔就的。（任麗達、盧鋒整理） -艾-巴-八-愛-扮-八-礙 發(fā)貼時間： 2004-1盎0-17 2:28:56 板 218.31.*.* 頒 八 errantking 等級：貧農(nóng)財產(chǎn)：經(jīng)驗：魅力： 注冊：2003-9-7 文章：55 鑒定：保密 敗-斑-挨-啊-襖-傲-靶- 稗Measurin

14、g and版 Interpreting版 World Econom翱ic Performanc拜e 1500-2001*敖Angus Maddiso按n胺This is a sui巴table occasio暗n for surveyi凹ng the progre礙ss achieved, 皚in the past 6愛0 years, in q巴uantifying wo頒rld economic 凹development, 岸and analysing擺 the causal i拌nfluences whi隘ch determine 笆the pace and 罷pattern of gr

15、翱owth. This wa疤s a major obj百ective of the般 founding fat拌hers of the I吧nternational 岸Association f跋or Research i笆n Income and 班Wealth (IARIW昂). The initia拜tive for crea骯ting an assoc啊iation includ艾ing both acad挨emics and off扳icial statist佰icians came f八rom Simon Kuz阿nets (1901-85啊), the pionee矮r of

16、quantita氨tive economic艾 history. Mil把ton Gilbert (骯1909-79) and 埃Richard Stone挨 (1913-1991) 叭were strategi埃c partners wi拌th enormous i版nternational 扒leverage in c佰reating and d埃iffusing stan挨dard procedur扳es for constr礙uction of com岸parable natio扳nal accounts 跋by official s暗tatistical of扒fices.奧This

17、 paper an阿alyses the de拌velopment of 斑macro-measure凹ment in three八 epochs:安a) for the IA擺RIW epoch, ba盎ck to 1950, t愛he main purpo霸se has been t安o illuminate 靶policy option敗s to improve 愛growth perfor班mance at the 斑national leve癌l and to anal頒yse inter-cou敖ntry divergen吧ce in real in凹come levels t拜o

18、 help devise罷 policies for懊 catch-up. We稗 now have off奧icial estimat八es of growth 稗and levels fo伴r the vast bu笆lk of the wor皚ld economy fr疤om 1950 onwar佰ds.芭b) For the Ku盎znetsian epoc矮h of “modern 爸economic grow板th” back to 1癌820, quantita啊tive economic俺 historians h柏ave made grea瓣t progress in挨 me

19、asuring wo伴rld economic 靶g(shù)rowth and in叭 quantifying 罷the forces de襖termining per艾formance. The昂re is scope f跋or further re敖search to fil鞍l(fā) gaps and cr班osscheck exis壩ting estimate班s, but the br按oad contours 敖of developmen霸t in this per半iod are not u叭nder serious 敖challenge.罷c) for the “m案erchant capi

20、t爸alist” epoch,懊 1500-1820, t跋here are shar艾ply divergent巴 interpretati氨ons about wor把ld, and parti頒cularly Europ氨ean performan奧ce. The dicho伴tomy between 挨positive and 佰negative asse擺ssments start癌ed with Adam 傲Smith and Mal啊thus at the e吧nd of the 18t胺h century. In哎 my view, the骯 evidence for挨 the

21、 Malthusi跋an view is sh凹oddy.盎Historians us霸ually start a鞍t the beginni盎ng of their c搬hronology. Qu鞍antitative ec扒onomic histor礙ians have to 癌work backward哀s from the pr啊esent, procee疤ding from wha斑t is known an板d accepted, t凹o earlier epo案chs where evi罷dence is weak拔er and there 叭is greater re疤lian

22、ce on clu半es and conjec暗ture. 艾*This is the 稗first Ruggles骯 Lecture, del稗ivered at the半 28th IARIW G唉eneral Confer啊ence, Cork, I拌reland August跋 2004I跋(i) Standardi埃sed Estimates襖 of GDP Growt阿h for 1950 on翱wards唉The standardi白sed accounts provide a coh愛erent macroec壩onomic framew哀ork covering 埃the w

23、hole eco笆nomy, crossch佰ecked in thre吧e ways. From 叭the income si罷de, they are 凹the total of 瓣wages, rents 阿and profits. 癌On the demand凹 side, they a壩re the sum of襖 final expend爸itures by con傲sumers, inves皚tors and gove笆rnment. From 頒the productio爸n side, the s霸um of value a哎dded in diffe百rent sector

24、s-澳agriculture, 唉industry and 八services, net伴 of duplicati捌on.阿Milton Gilber般t had been re板sponsible for吧 the official啊 US accounts 邦during the wa白r and from 19瓣50 to 1961 wa盎s head of sta埃tistics and n傲ational accou搬nts in OEEC. 俺The Marshall 癌Plan required翱 criteria for瓣 aid allocati埃on, and NATO

25、笆needed them f翱or its burden霸-sharing exer熬cises. Gilber般t met these r巴equirements b愛y pushing off安icial statist辦ical offices 岸of the 16 OEE半C member coun敗tries to adop岸t the standar岸dised system 案of accounts d阿esigned by Ri跋chard Stone. 扳Stone set up 擺a programme i骯n Cambridge t百o train offic半ial E

26、uropean 案statisticians暗 to implement般 the standard半ised system. 按A set of nati扮onal handbook氨s was prepare凹d to explain 邦the problems 骯of adjustment唉 to the stand矮ardised syste癌m, and a firs百t comparative襖 set of accou礙nts for the 1拌6 countries f辦or 1938 and 1昂947-52 was pu懊blished by OE傲EC in 1954,

27、 w霸ith extensive哀 notes explai敖ning the adju哀stments which敖 had been mad癌e to achieve 霸comparability般.氨In 1953, Ston罷e became chai暗rman of a Uni藹ted Nations c哀ommission whi隘ch establishe唉d a system of絆 accounts for艾 worldwide ap骯plication. Th擺e UN could no隘t exert as mu艾ch leverage o襖n its member 吧c

28、ountries to 芭conform as wa翱s possible in拌 OEEC. The co熬mmunist count般ries used the跋 Soviet SMS (盎system of mat埃erial account扒s) which had 傲a narrower de懊finition of p昂roductive act板ivity (exclud懊ing many serv罷ice activitie絆s), involved 懊serious doubl把e counting (m邦easuring gros巴s output with八out ded

29、ucting胺 inter-sector氨 transfers of絆 inputs) and 頒exaggerated e拌conomic growt藹h. The price 敖system and ta板x-structures 唉were differen罷t from those 熬in capitalist癌 countries, a扒nd measuremen挨t conventions班 gave incenti挨ves to exagge哎rate quality 拔change when n挨ew products w拜ere introduce鞍d. Abram Berg

30、八son (1914-200盎3) pioneered 埃procedures fo埃r re-estimati邦on of Soviet 哀GDP on a basi叭s correspondi礙ng approximat班ely to Wester般n conceptions笆 in coverage,吧 with elimination of doubl吧e-counting, a隘nd repricing 辦on an “adjust懊ed factor cos襖t” basis with壩 imputation f哀or capital co俺sts which wer阿e not

31、conside佰red in Soviet捌 accounting. 敖These correct扳ive procedure笆s were applie礙d to Soviet s藹tatistics by 俺a team of CIA懊 Sovietologis扒ts in Washing哎ton. In New Y八ork, Thad Alt斑on and his co翱lleagues did 伴the same for 岸Bulgaria, Cze案choslovakia, 拜East Germany,懊 Poland, Roma哀nia and Yugos哀lavia. This

32、w翱ork was finan般ced for intel芭ligence purpo骯ses, but was 拔publicly avai絆lable in annu愛al reports to胺 the US Congress (see Madd八ison 1998b).按In the 1990s 挨most of these壩 countries ad澳opted the sta骯ndardised SNA阿 system in pr擺inciple, but implementatio奧n was complic癌ated by the m扒assive change藹 in own

33、ership半, in the leve安l and structu把re of prices,扮 allocation o頒f resources b按etween consum搬ption and inv拜estment, and 版statistical r吧eporting proc藹edures. It wi熬ll take some 俺years before 胺these problem敖s can be full矮y resolved. T斑he IMF contin扒ues to use ex笆aggerated mea罷sures of GDP 擺growth for th

34、懊ese countries扳. As a result礙, it shows a 阿growth in wor扳ld GDP averag扳ing 3.9 perce哎nt a year for捌 1970-200, co唉mpared with m疤y estimate of埃 3.3 percent.扳 For China it靶 shows growth按 averaging 8.半5 per cent a 班year, whereas壩 my adjusted 耙measure shows艾 a growth rat皚e of 6.5 per 耙cent (see Mad絆dison

35、, 2003, 唉p. 231).辦Another area 埃of weakness i百s Africa, whe耙re there was 熬and still is 版a great short盎age of skills岸 and money fo唉r such work i敗n a large num班ber of newly 邦created count凹ries. The gap奧 in estimates斑 of GDP growt哎h was filled 傲in substantia隘l degree by t靶he OECD Devel叭opment Centre把 w

36、hich compil巴ed annual est艾imates of rea邦l GDP growth 跋1950-90 for 5暗1 African cou擺ntries. The C氨entre benefit埃ed from the e伴xpertise of D澳erek Blades, 敖who had been 邦chief statist拔ician in Mala罷wi for eight 板years, and by矮 David Robert耙s who had sim懊ilar experien拌ce in Gambia.八 般A third probl百em in

37、the ass拔essment of GDP growth perf俺ormance in th俺e higher inco唉me countries 哀derives from 氨recent change暗s in measurem案ent conventio埃ns from 1995 霸onwards, invo叭lving adoptio叭n of hedonic 捌indexes to ad擺just for assu拜med changes in quality of 班product, use 襖of chain indi氨ces, and trea埃tment of cons稗

38、umer software絆 as investmen罷t. 頒Hedonic indic絆es are perfec安tly respectab斑le in small d伴oses, but one扮 can be skept安ical about th傲e widespread 埃assumption th癌at quality ch搬anges have be半en so large a柏nd monotonica版lly positive.藹 In the USA, 叭where the swi百tch to hedoni胺cs was most s哎ignificant, t挨he

39、ir net impa般ct was to rai頒se the measur壩ed rate of gr跋owth to a som笆ewhat greater扮 degree than 礙in Western Eu壩rope and Japa疤n. US officia拜l estimates g胺o back to 192拔9, and the ch艾anges in meas暗urement techn捌ique had thei按r biggest imp敗act for 1929-霸50, raising t搬he GDP growth哀 rate for tha奧t period

40、 from啊 2.6 percent 皚a year to 3.5芭. There was n叭o counterpart柏 to this long拌 retrospectiv翱e readjustmen氨t in other co稗untries, and 扮I have contin艾ued to use th半e earlier US 壩official meas熬ure for 1929-昂50 (for reaso氨ns explained 絆in Maddison, 懊2001, p. 138,擺 and Maddison絆, 2003, pp.79壩-80). More th奧

41、an 40 years a藹go, Milton Gi拔lbert warned 骯that such adj耙ustments coul跋d open Pandor按as box: “In 盎the end, they暗 would make i跋t impossible 奧to construct measures of o爸utput and pri氨ce changes th礙at are useful般 to the study疤 of economic 絆growth” (Gilb把ert, 1961, p.藹 287). The da暗nger which ar絆ises fro

42、m an 扮overdose of h板edonics is di版scussed in Ap稗pendix 3.爸Ed Denison (1熬915-1992) exp敗ressed opposi背tion to chang胺es in nationa奧l accounting 艾which treat a佰ccretions of 愛knowledge as 稗investment. H安e considered 拔this a “miscl把assification”藹 which made “暗growth analys辦is chaotic” (八see Denison, 阿1989

43、, p. 10).熬 A major just隘ification for暗 his complain般t was that hi瓣s growth acco愛unts included啊 “human capit安al”, i.e. inc拔rements in th安e quality of 藹the labour fo版rce due to in敖creases in th扒e level of ed藹ucation.翱In fact, the 版only form of 頒knowledge whi俺ch is now tre稗ated as inves胺tment is comp扮u

44、ter software拜. It is odd t白o treat this 霸rapidly depre巴ciating knowl岸edge as inves跋tment, whilst傲 ignoring the百 more durable巴 influence of百 books and education.岸(ii) Purchasi巴ng Power Conv按erters for Cr頒oss-country C絆omparison of 頒GDP Levels 拔Once standard愛ised accounts奧 of real GDP 搬growth in nat擺i

45、onal currenc搬ies had been 白established f胺or all OEEC c岸ountries, the背 next step to扒 facilitate i暗nter-country 疤comparison an阿d multi-count拜ry aggregatio吧n was the dev藹elopment of p盎urchasing pow盎er parity con罷verters (PPPs柏) to measure 襖real GDP leve鞍l(fā)s, rather th氨an relying on百 exchange rat癌e compa

46、rison.胺 The first OE艾EC study, co-拜authored by M般ilton Gilbert傲 and Irving K伴ravis (1916-9靶2). appeared 奧in 1954, a se跋cond in 1958.伴 They compare絆d real expend拌iture levels 扮in 8 OEEC cou芭ntries. A thi扮rd volume by 邦Paige and Bom拔bach (1959) c背ompared real 敖output levels藹 in the UK an昂d USA. Kravis

47、扳 and his coll捌eagues, Alan 岸Heston and Ro礙bert Summers 巴improved the 奧methodology o埃f PPP estimat辦ion in their 礙ICP project a哀t the Univers翱ity of Pennsy柏lvania from 1靶968 onwards.搬The OEEC stud艾ies were bina愛ry comparison阿s between pai爸rs of countri矮es. The three叭 options were絆 i) a Paasche瓣 PPP,

48、with “o靶wn-country” q頒uantity weigh耙ts; ii) a Las笆peyres PPP wi拜th the quanti拔ty weights of霸 the numerair耙e country-the胺 United State板s; iii) a com跋promise geome骯tric (Fisher)叭 average of the first two 靶measures. The矮 correspondin半g measures of骯 real expendi笆ture were: i)哎 Laspeyres co搬mparisons of

49、熬GDP levels ba扒sed on the pr拌ices (unit va拜lues) of the 懊numeraire cou絆ntry; ii) Paa盎sche level co吧mparisons bas傲ed on “own-co跋untry絆”埃 prices (unit澳 values); iii擺) a Fisher ge艾ometric avera叭ge of the two扳 measures. Bi啊nary comparis搬ons, e.g. Ger版many/USA and 氨UK/USA, could斑 then be link啊ed with the

50、 U捌SA as the sta芭r country. Su俺ch star compa皚risons could 吧provide a pro八xy Germany/UK哀 comparison, 擺but it was no芭t “transitive背” (i.e. the r熬esult would n骯ot be identic背al to that de癌rived from a 凹direct German稗y/UK comparis翱on). This was吧 not a great 般drawback for 唉OEEC countrie板s where the i挨nte

51、r-country 稗deviation in 佰performance l皚evels was not皚 too wide. Bu爸t Kravis, Hes啊ton and Summe阿rs were engag邦ed in compari埃sons over a m愛uch wider ran白ge of per cap捌ita income. T啊hey therefore癌 adopted the 鞍Geary-Khamis 盎method, inven頒ted by Roy Ge盎ary (1896-198扒3) and Salem 澳Khamis, which multilate

52、ral班ised the resu隘lts, provided扳 transitivity俺 and other de礙sirable prope般rties. They u阿sed it in con斑junction with伴 the commodit翱y product dum捌my method (CP捌D), invented 捌by Robert Sum巴mers, for fil奧ling holes in奧 the basic da凹taset. Their 版masterpiece w巴as their thir礙d study, the 芭1982 volume W稗or

53、ld Product 疤and Income, w矮hich containe安d estimates f岸or 34 countri愛es (in Africa班, the America礙s, Asia and E奧urope) in 197氨5 prices and 疤international昂 Geary-Khamis吧 dollars. The盎se countries 礙accounted for笆 64 per cent 懊of world GDP 八in 2001.案Table 1 Natur搬e of PPP Conv傲erters to Est傲imate GDP Lev

54、笆els in the Be拔nchmark Year 柏1990擺(billion 1990巴 Geary-Khamis敖 dollars and 拌number of cou柏ntries)翱Europe & Lati擺n Asia Africa絆 World矮W. Offshoots 佰America 俺ICP 15,273 (2爸8) 2,131 (18)翱 8,017 (24) 0白 (0) 25,421 (拌70)扮PWT 59 (3) 71斑 (14) 524 (16壩) 891 (51) 1,胺516 (84)暗Proxies 16 (1隘0) 38 (15) 87阿 (17)

55、 14 (6) 癌155 (48)耙Total 15,349 白(41) 2,240 (4板7) 8,628 (57)奧 905 (57) 27,癌122(202)版Source: Maddi盎son (2003), p啊. 230癌The UN Statis半tical Office 背extended the 擺ICP work and 斑had covered 8拌4 countries b版y 1985. UNSO 瓣then dropped 瓣this endeavou盎r, though som埃e of the regi把onal UN bodie把s continued w藹i

56、th it. The O板ECD recommenc白ed its compar暗isons on a re傲gular basis i熬n 1982. Its l皚atest work co岸vered the 28 版OECD countrie矮s and 20 othe按rs (in Easter壩n Europe, the笆 15 successor胺 states of th百e USSR, and M八ongolia). Ala壩n Heston and 拌Robert Summer胺s produce sho芭rt-hand estim艾ates of PPPs 拜and rea

57、l inco扒me levels for矮 countries fo疤r which full-襖scale ICP typ藹e measures ar按e not availab邦le. As a resu版lt, we now ha案ve reasonably暗 acceptable P壩PP adjusted m哎easures avail岸able for over般 99 percent o氨f world GDP,翱There were th頒ree Eurostat 把estimates (fo百r 1980, 1985 巴and 1993) of 暗PPPs for 22 A板

58、frican countr疤ies, but the 霸results were 壩erratic, and 芭I preferred t胺o use the more comprehensi拜ve and plausi艾ble results o柏f the Penn Wo啊rld Tables (s礙ee Summers an啊d Heston, Oct啊ober 2002). T疤able 1 summarises the natu矮re of the PPP白 estimates I 按used to creat澳e my 1990 ben佰chmark estima礙tes of w

59、orld 疤GDP.半Table 2 World擺s 10 Largest稗 Countries: C氨omparative Ra岸nking, 1950 &伴 2001, at con敗stant矮1990 prices, 佰using 1990 Ge啊ary-Khamis PP阿P converters 邦and 1990 exch霸ange rates靶1950 2001 195唉0 2001艾GDP $billion,擺 with 1990 PP安P conversion 瓣$billion, wit愛h 1990 exchan阿ge rate八USA 1,456 7,9澳66 1,4

60、56 7,96斑6芭China 240 4,5骯70 47 886骯Japan 161 2,6啊25 206 3,358扮India 222 2,0扒03 62 558案Germany 265 1唉,537 337 1,95八1班France 221 1,把258 261 1,491敗UK 348 1,202 皚363 1,253俺Italy 165 1,1絆01 191 1,272氨Brazil 89 990邦 58 638埃Russia 315 79癌1 154 388伴GDP per head 澳1990 PPP $ 19暗90 ER $ 暗USA 9,561 27,捌948 9,561