How Does Arithmetic Demystify Growth Success? The Taiwanese Case. Shu-shiuan Lu* University of California, Los Angeles October 2006

Transcription

1 How Does Arihmeic Demysify Growh Success? The Taiwanese Case Shu-shiuan u* Universiy of California, os Angeles Ocober 26 Absrac This paper sudies he produciviy improvemen in economies experiencing rapid srucural ransformaion, i.e. resource reallocaion from a low o a high produciviy secor. In paricular, i measures he produciviy growh ha is aribued o reallocaing labor from he farm o he non-farm secor in Taiwan beween 1951 and 23. Examining produciviy under a wo-secor raher han a one-secor framework reduces he measured produciviy growh by 28% in erms of oal facor produciviy. In oher words, he resuls show ha resource reallocaion, especially ha of labor, from a low o a high produciviy secor plays a subsanial role in accouning for he significan produciviy improvemen in an economy experiencing rapid srucural ransformaion. Keywords: Source of Growh, Secoral Transformaion, and Taiwanese Economy JE Classificaion: O11, O41, O53 *I am exremely graeful o my advisor, ee Ohanian for advice and suppor. I also hank Andrew Akeson, Harold Cole, Mahias Doepke, Roger armer, Gary Hansen, Giovanni Peri, Kenneh Sokoloff, and he paricipans a he UCA Proseminar in Moneary and Macroeconomics for helpful commens. All he remaining errors are mine. Please direc correspondence o Shu-shiuan u, 8283 Bunche Hall, Box 95l477, os Angeles, CA shusanlu@ucla.edu Webpage: hp://shusanlu.bol.ucla.edu 1

2 1. Inroducion This paper uses a muli-secor concep o idenify he role of resource reallocaion from low produciviy o high produciviy secors o accoun for he rapid economic developmen of Taiwan. I employ simple arihmeic o analyze economic growh under a wosecor model concep raher han a one-secor concep. Moreover, I demonsrae ha in he second half of he 2 h cenury efficien reallocaion of resources resolves par of he mysery of high produciviy growh in Taiwan. While he lieraure on growh shows remendous work on he adjusmen of capial and labor qualiies, one sill canno exclude he fac ha improvemens in produciviy are imporan in explaining he success of Taiwan 1 and is abnormally high rae of convergence o he developed world. Jorgenson and ip (21), Maddison (21), and oung (1995) provide abundan numerical resuls of produciviy growh pos Second World War and enable crosscounry comparison for produciviy. However, he produciviy improvemen in Taiwan is comparaively high and remains a puzzle under heir careful accouning. The focal poin of his paper is o show how using simple arihmeic o look ino produciviy growh under a muli-secor framework can elucidae an undersanding of he growh miracle in Taiwan. The macroeconomic daa show ha he percenage of he populaion working in he agriculural secor fell sharply. Meanwhile, he overall economy grew rapidly. This raises he quesion of wheher rapid growh and srucural ransformaion are conneced. To address his quesion, I begin by comparing produciviy growh across Eas Asian, ain American and OECD counries. The resuls show ha hose counries ha grew quickly also experienced rapid labor reallocaion from he farm o he non-farm secor. Using Taiwan as a deailed case sudy, I show ha applying simple arihmeic o examine produciviy under a wo-secor raher han a one-secor framework reduces he measured produciviy growh by 28% in erms of oal facor produciviy. The res of he paper is organized as follows: irs, I review he empirical lieraure on he comparison of produciviy growh across economies and show ha Taiwan had a high produciviy growh rae relaive o he res of he world. Second, I show he arihmeic 1 Though oung (1995) emphasized ha faser facor accumulaion raher han oal facor produciviy in Taiwan explains why Taiwan able o cach up wih he developed economies from 1966 o 199, he growh rae of oal facor produciviy remains comparaively higher han he res of he world for he same period, especially when look ino he sub-periods: I will ariculae his poin in deail in secion 3. 2

3 framework I adop o capure he imporance of efficien reallocaion of facors in accouning for growh in erms of average labor produciviy and oal facor produciviy. Third, I show he empirical evidence ha he economies wih rapid growh, e.g. Japan, Souh Korea, and Taiwan, experienced rapid srucural ransformaion along heir growh pah. Then, I ake Taiwan as he case sudy for his paper, analyze how srucural ransformaion affeced is growh from 195 o 22, and show how he growh rae in oal facor produciviy reflecs in par a move of facors. In he las secion, I offer some concluding remarks. 2. ieraure Review for Produciviy Growh across Counries Produciviy can refer o eiher average labor produciviy (AP) or oal facor produciviy (TP). I clarify wo conceps of TP here: one is o view TP as he residual of he quaniy of facor inpus (broad TP); he oher is o view TP as he residual of all he explainable facors (narrow TP). The laer implies ha TP is an unexplainable erm. Maddison (21) conducs a cross-counry sudy on AP in erms of GDP per person employed in 199 US$. He shows ha Taiwan is one of he fases growing economies in erms of AP. Table I replicaes he able in he book 2. As can be seen, he average labor produciviy in Taiwan in 1998 was 13.7 imes ha of 195. This number is higher han any of he seleced economies in Europe, Wesern offshoos, ain America and Asia. urhermore, Souh Korea ranked he second. The corresponding value in 1998 was imes ha of 195. inally, Japan ranked he hird and he corresponding value in 1998 was 8.79 imes ha of 195. Therefore, Taiwan, Souh Korea and Japan have a faser produciviy improvemen han all oher economies in erms of average labor produciviy; among hem, Taiwan is he fases. Table I: Average abor Produciviy in 199 Inernaional Dollars /195 Canada 2,311 35,32 43, Unied Saes 23,615 4,727 55, rance 11,214 31,91 5, Germany 9,231 26,623 4, Ialy 8,739 25,661 42, Japan 4,511 23,634 39, Unied Kingdom 15,529 26,956 4, I do no repor he same daa for Hong Kong and Singapore since hey have missing daa for 195 and

4 Souh Korea 2,516 8,689 28, Taiwan 2,569 11,924 35, Argenina 12,538 21,349 25, Brazil 5,6 12,111 14, Chile 1,316 1,316 26, Colombia 6,492 12,22 16, Mexico 7,685 18,399 2, Peru 6,17 12,685 1, Venezuela 23,792 37,856 26, (Maddison 21, p. 349 &35) Barro and Sala-I-Marin (23) summarize he lieraure on TP growh. They summarize works from Jorgenson and ip (21), Elias (199) and oung (1995). Their able shows ha he TP growh in Taiwan (1966 o 199) is relaively high compared wih he res of he world. Table II replicaes he ables from Barro and Sala-I-Marin (23), and Jorgenson and ip (21). The able summarizes he TP growh raes for OECD, ain American and Eas Asian counries. As can be seen, Taiwan, Souh Korea, Japan, and Hong Kong grew faser han he oher economies. Since Hong Kong is a ciy-sae, I choose Souh Korea and Japan as fas growing economies in addiion o Taiwan 3. Table II: Growh in TP from 194 o 199, Assored Economies OECD Counries Canada rance Germany Ialy Japan UK US OECD Counries Canada rance Germany Ialy Japan UK US ain American Counries Argenina Brazil Chile Colombia Mexico Peru Venezuela Eas Asian Counries Hong Kong Singapore Souh Korea Taiwan Table III & IV replicae he ables in oung (1995) which show he sub-period produciviy growh for Taiwan and Souh Korea. The numbers in brackes are he percenage conribuion of growh in labor, growh in capial, and growh in TP o he oupu growh 3 In addiion, if we look a he TP growh for OECD counries for he period 196 o 1989 raher han he period 196 o 1995, Ialy also has high TP growh compared wih oher OECD counries. inally, when one compare across ain American economies, Brazil and Mexico have higher TP growh han he oher economies wihin he group. 4

5 rae. As is shown, he oal facor produciviy is high for Taiwan during and In addiion, he oal facor produciviy for Souh Korea (198 o 199) is high as well. In oher words, hough oung carefully adjused qualiy improvemens on facor inpus, he conribuion of TP o growh remains as high as 42% in Taiwan (beween 198 and 199). Therefore, i is clear from his work ha he fas growing Eas Asian economies do realize some high produciviy growh in addiion o rapid capial and labor accumulaions. Table III: oung s Table for Taiwanese Growh Accouning (1995, p. 661) Period G_ G_K G_ G_TP (1.%) (4.21%) (29.29%) (3.63%) (1.%) (36.49%) (48.79%) (14.56%) (1.%) (26.71%) (3.73%) (42.31%) (1.%) (33.63%) (38.73%) (27.66%) Table IV: oung s Table for Korean Growh Accouning (1995, p. 66) Period G_ G_K G_ G_TP (1.%) (28.18%) (64.52%) (6.49%) (1.%) (41.76%) (49.35%) (9.3%) (1.%) (42.11%) (38.27%) (2.%) (1.%) (58.57%) (38.8%) (2.15%) (1.%) (31.56%) (4.31%) (28.24%) (1.%) (26.34%) (49.73%) (24.3%) (1.%) (39.5%) (43.68%) (16.5%) 5

6 In sum, in erms of produciviy, Taiwan, Souh Korea and Japan have been growing rapidly relaive o he res of he world. Among hem, Taiwan s growh is he fases. Consequenly, I would like o use Taiwan as he main case sudy for his paper and supplemen my sudy wih Souh Korea, Japan and Ialy o suppor he argumen ha here is a posiive relaionship beween rapid growh and fas srucural ransformaion. 3. The Arihmeic rameworks for Idenifying Reallocaion Effec in Produciviy Growh In he following secion, I will decompose produciviy growh (in erms of AP and TP) ino wo pars: one is he conribuion of efficien resource reallocaion o produciviy growh; and he oher is he conribuion of produciviy improvemen o growh. The conribuion of efficien resource reallocaion measures he proporion of growh ha resuls from reallocaing resources, such as labor or capial, from a less producive secor o a more producive secor. Such a shif of resources can conribue o growh in oal oupu wihou any produciviy improvemen in eiher high or low producive secors. On he oher hand, he conribuion of produciviy improvemen measures he proporion of growh ha resuls from echnological improvemen assuming no reallocaion of resources. 3.1 Decomposiion of Average abor Produciviy Average abor Produciviy (AP) is defined as oal oupu divided by oal employmen. When AP is expressed in erms of a wo-secor framework, AP is composed of produciviy in he farm secor muliplied by he share of farm labor plus produciviy in he non-farm secor muliply by he corresponding share of labor in ha secor. The decomposiion is shown as follows: q + = + q = N N + q = N N (1 ) + N + N N where N + : arm Secor; N: Non-arm Secor; : Oupu; : abor; q: Relaive Price (Price of non-farm oupu equals one) N + N 6

7 7 Therefore, Equaion 1: [ ] ( ) + + = ) (1 1 1) ( 1) ( N N N N N N N N N N N N q q q q q q A B Z C D D C B A X where X = reallocaion; A= % change in non-farm average labor produciviy B = iniial non-farm average produciviy relaive o he aggregaed C= % change in farm average labor produciviy D = iniial farm average produciviy relaive o he aggregaed = % οf labor in farm secor Z = % of labor in non-farm secor As a consequence, he aggregaed average labor produciviy incremen ha resuls from he reallocaion effec is X[AB-CD], whereas he incremen ha resuls from produciviy improvemen is he sum of D(C-1) and ZB(A-1). 3.2 Decomposiion of Toal acor Produciviy Toal facor produciviy (TP) can also be represened as he weighed average of TP in he farm and he non-farm secors. Under a wo-secor model, aggregaed TP is composed of α, β, TP in farm and TP in non-farm secor.

8 TP Define : K q K q = K + K + N + N + N = θ 1 θ θ 1 θ K N α K + (1 α) K β + (1 β ) where α % of capial in farm secor β % of labor in farm secor Thus, Equaion 2: TP q = θ ( α K ) ( β ) = TP α β θ 1 θ 1 θ α β θ + TP N 1 θ + (1 α ) θ [(1 α ) K ] [(1 β ) ] θ (1 β ) N 1 θ 1 θ (1 α ) θ (1 β ) 1 θ In oher words, I can decompose he TP ino wo erms: TP of he farm secor and TP of he non-farm secor, weighed by he percenage of he facors in each secor. Therefore, o obain he TP wihou reallocaion effec (from pure produciviy improvemen), one could se α and β consan and equal o he iniial value (α and β ) over ime. or he case if here were no labor reallocaion, one could se β consan and equal o he iniial value β. Similarly, for he case if here were no capial reallocaion, one could se α consan and equal o he iniial value α. inally, for he case if here were no TP improvemen in he farm secor, one could se TP consan and equal o he iniial value TP ; for he case if here were no TP N improvemen in he non-farm secor, one could se TP N consan and equal o he iniial value TP N. inally, for he case if here were no TP improvemen, one could se boh TP and TP N respecively and consanly equal o TP and TP N over ime. 8

9 4 Conribuion of Efficien Resource Reallocaion o Produciviy Growh In secion 2, I show ha Japan, Souh Korea and Taiwan experienced relaively high produciviy improvemen afer he Second World War. To argue ha rapid developmen coincides wih rapid srucural ransformaion, I compare he speed of srucural ransformaion of Japan, Souh Korea and Taiwan from 196 o 22 wih he experiences in he U.S. and Canada in his secion. irs, I show he speed of ransformaion in he U.S. and Canada. Second, I compare he speed of srucural ransformaion of various economies afer he Second World War wih hisorical daa from he U.S. and Canada 4.1 The Speed of Transformaion: U.S. and Canada The percenages of employmen in he agriculural secor in he U.S. and Canada from 188 o 22 are shown in able V. or he Canadian daa, hose saring from 1881 o 194 are from he census of Canadian Hisorical Saisics; hose saring from 195 o 196 are from he annual daa of Canadian Hisorical Saisics; and hose saring from 197 o 22 are from he Inernaional abor Organizaion. As can be seen, i ook he US and Canada roughly he same lengh of ime, 5 o 6 years, o experience farm secor employmen dropping from 23% o 3%. Table V: Percenage of Employmen in Agriculures, , US and Canada US ear % ear % 22.74% 2.8% 12.15% 8.3% ear % 4.4% 3.39% 2.86% 2.56% 2.55% Canada ear % 48.1% 45.8% 4.2% 34.3% 32.7% ear % 28.8% 25.8% 22.89% 13.25% ear % 9.9% 7.28% 5.63% 4.4% 3.91% The oher measuremen o show he percenage of labor force in he farm secor is by he percenage of he economically acive populaion in he agriculural secor. The values for 9

10 he U.S. and Canada are abulaed in Table VI. The daa for he US is from 188 o 199 and aken from he US Bureau of abor Saisics repored in he Saisical Absrac of he Unied Saes 23; whereas hose for Canada is from 195 o 199 and aken from he Inernaional abor Organizaion. According o he publicaion of he Inernaional abor Organizaion, The economically acive populaion comprises all persons of eiher sex who furnish he supply of labour for he producion of economic goods and services as defined by he Unied Naions sysems of naional accouns and balances during a specified ime-reference period. Two useful measures of he economically acive populaion are he usually acive populaion measured in relaion o a long reference period such as a year and he currenly acive populaion or equivalenly he "labour force" measured in relaion o a shor reference period such as one week or one day. usually acive populaion may be subdivided as employed and unemployed in accordance wih he main aciviy The labour force or "currenly acive populaion" comprises all persons who fulfil he requiremens for inclusion among he employed or he unemployed as defined. (hp:// Table VI: Percenage of Economically Acive Populaion in Agriculures, US ( ) and Canada ( ) US ear % 5.6% 42.84% 37.56% 31.55% 27.4% 22.1% ear % 17.57% 12.32% 6.64% 4.29% 3.48% 2.84% Canada ear % ear % 19.84% 13.19% 7.79% 6.66% 3.37% As can be seen, i ook he US and Canada 4 o 5 years o experience heir economically acive populaion in he farm secor drop from 23% o 3%. In sum, he US and Canada display roughly similar speeds of srucural ransformaion from a sociey mainly focused on agriculure o indusry and services. Therefore, I ake he 1

11 number of years he US or Canada ook o ransform from one level of labor in agriculure o anoher as he sandard number of years of ransformaion required for an economy. 4.2 Measuring he Speed of Transformaion In order o measure how fas economies ransform, I compare each economy s srucural ransformaion wih ha of he US and Canada. I do no compue a direc annual rae of srucural ransformaion because he process of ransformaion is likely o be nonlinear over ime: one would expec ransformaion o be more rapid in early sages of developmen and slow down as he economy maures. To overcome his issue, I ake he following seps. irs I obain he percen of farm labor for each of he analyzed economies in 196 and Second, I compare hese values o he daa for he US and Canada and compue he number of years i would have aken for he US or Canada o underake he idenical ransformaion (boh in level and percen change) in farm labor. Third, I divide he number of years he US or Canada ook o achieve he ransformaion by he number of years he analyzed economy ook. I erm his raio as he raio of sandard o acual. This raio represens how fas an economy ransforms. or example, if i akes he U.S. 5 years o have he percenage of farm employees o drop from 2% o 1% and if i akes Japan 25 years o drop from 2% o 1%, I argue he srucural ransformaion in Japan is wice he speed of he U.S. The corresponding raio of sandard o acual equals wo. Therefore, he higher he raio, he faser an economy ransforms from an agriculural economy o an indusrialized economy. The resuls 5 are in Table VII. In view of he employmen in he farm secor, Japan, Ialy 6, Souh Korea and Taiwan experienced he fases srucural ransformaion. These economies also experienced higher TP growh in Table II. Noe ha I did no repor he employmen daa for ain American counries because employmen daa are no a good measuremen o compue he speed of srucural ransformaion in hese economies. Insead, an economically acive populaion raher han employmen serves as a beer measuremen. Tha is because employmen is 4 Excep where indicaed in able VII. 5 The employmen daa for he developed economies are obained from he US Bureau of abor Saisics, oreign abor Saisics webpage. The daa for S. Korea are from he Inernaional abor Organizaion, and he daa for Taiwan is from is Agriculural Saisic Absrac locaed on he web. 6 Ialy also has high TP growh during , as shown in able II, OECD Counries,

12 subjec o business cycles and he unemploymen rae is relaively high and volaile in ain America. Table VII: Percenage in arm, in erms of Employmen, OECD Counries, S. Korea and Taiwan Equivalen Sandard Counries Raio of Sandard ear Acual Japan 29.51% 4.49% 7 ears 1.67 rance 22.48% 3.66% 45 ears 1.7 Germany 13.78% 2.52% 55 ears 1.3 Ialy 32.47% 3.53% 8 ears 1.9 Neherlands 1.25% 3.1% 3 ears.71 Sweden 15.75% 2.47% 55 ears 1.31 Counries % in farm (dae) Unied Kingdom 4.73% (196) Souh Korea 48.6% (1974) Taiwan 46.5% (1965) % in farm (dae) 2.67% (1983) 8.9% (23) 6.56% (24) Equivalen Sandard ear Raio of 3 ears ears ears 2.44 Sandard Acual In view of he economically acive populaion 7 from 195 o 199, Korea and Japan s percenage in agriculure dropped more han 4%, which was he larges decline among 45 counries 8. (The able for 45 counries is repored in Appendix 1.) Japan dropped from 48.83% of populaion in agriculures o 7.28%; such a ransiion ook he U.S. approximaely 7 years. Similarly, Korea had 76.9% of he employed in agriculures and dropped o 18.11% by 199, which is a process ha ook he U.S. more han 7 years. 7 I collec he daa from he Inernaional abor Organizaion. Taiwan did no repor economically acive populaion. Therefore, I use he Souh Korea Case, which I have daa for boh measuremens, o show he wo measuremens, employmen and economically acive populaion can boh show he speed of ransformaion is fas for he growh miracles in Japan, Souh Korea and Taiwan. 8 Taiwan does no repor economically acive populaion. 12

13 inally, wihin he group of ain American economies, Brazil, Colombia and Mexico are he op hree economies wih he bigges percenage of change in erms of percenage of he economic acive populaion in he farm secor. Excep Colombia, which was no included in Table II, Brazil and Mexico are also he economies wih relaively higher TP growh in he region. In sum, compared wih he group consising of OECD, Asian and ain American counries, Japan, Souh Korea and Taiwan experienced relaively high produciviy improvemen afer he Second World War. Meanwhile, all of hese economies experienced rapid srucural ransformaion. Therefore, economies wih higher produciviy growh also show relaively fas srucural ransformaion. Among hem, Taiwan experienced a fas rae of indusrializaion in he second half of he 2 h cenury. One key feaure of indusrializaion is a shif of labor from an agriculural (farm) o a non-agriculural (non-farm) secor. Based upon he coinciding rapid economic growh experiences and rapid efficien labor reallocaion, I would like o use Taiwan 9 o show ha, once he resource reallocaion of labor is aken ino accoun, he rapid growh in he economy becomes less myserious (in secion 5 and 6). 5. The Economic Srucural Transformaion in Taiwan The daa used in his secion is downloaded from DaaSream, World Penn Tables and assored Taiwanese governmen web pages and saisical daa books. The period covered is 1952 o 23 (1952 is he earlies I can obain he daa). Someimes he daa may sar a laer ime or end a earlier ime due o he availabiliy of he daa. The srucure in his secion is as follows: irs, I show he srucural ransformaion in Taiwan in erms of GDP and employmen. Then, I show he reason why i makes sense o conduc he analysis under a muli-secor framework raher han a one-secor framework. 5.1 Srucural ransformaion in GDP igure I shows he secoral GDP relaive o he aggregae in erms of percenage from 1952 o 24. As can be seen, he percenage of GDP ha comprises he agriculural secor is 32.22% in 1951 and falls o 1.74% in 23; in addiion, he percenage aribued o he 9 I use Taiwan as my example because he Taiwanese labor produciviy improvemen (boh AP and TP) is he mos significan 9 among he hree economies. 13

14 manufacuring secor is 19.69% in 1952, reaches is peak (47.11%) in 1986 and drops back o 29.54% in 24. inally, he percenage aribued o services remains around 46%, sars o akeoff rapidly in 1987, and reaches 68.72% in 24. In oher words, he GDP share in agriculure has declined monoonically since 1952, whereas he aggregaed share of he oher wo secors (manufacuring and services) increased in imporance o he economy during his period. 5.2 Srucural ransformaion in employmen igure II shows he secoral employmen relaive o he oal employed populaion in erms of percenage from 1952 o 24. As can be seen, he percenage of he employed working in he agriculural secor is 56.1% in 1952 and falls o 6.56% in 23; in addiion, he percenage of he employed working in manufacuring secor was 16.9% in 1952, reaches is peak (42.8%) in 1987 and drops back o 35.21% in 24. inally, he percenage of he employed working in services is 27% in 1952 and reaches 58.23% in 24. In oher words, he number of he workers in agriculures has sharply declined since 1952, whereas he aggregaed number of workers in he oher wo secors has increased. In sum, Taiwan experienced rapid indusrializaion from he 195s o he end of he 2 h cenury. During his period, in erms of labor and oupu, he service secor became he larges secor in Taiwan. 5.3 How Srucural Transformaion Maers: One Secor vs. Muliple Secors orm 1952 o 23, Taiwan ransformed from an agriculural sociey o an indusrialized economy; hen o a sociey where he majoriy of he labor force was employed in he service secor. If he produciviy differenials among he hree secors are very large, a muli-secor analysis wih a leas hree secors is beer able o capure he ransiion dynamics. A he oher exreme, if he produciviy among secors is close, a muli-secor analysis is redundan and he efficien ransfer of resources (in paricular, labor) from one secor o he oher does no maer for growh. I compare he produciviies of agriculure, indusry and services in erms of average labor produciviy and marginal labor produciviy. If he produciviy differenial among any of hese secors is small, he wo secors can be aggregaed ino one. On he oher hand, if he 14

15 produciviy differenial among any of hese secors is large, wo of hem mus be viewed as disinc secors. igure III shows he labor produciviy index, which is aken from he Produciviy Saisics provided by Direcorae-General of Budge, Accouning and Saisics Execuive uan (DGBAS.) I ransformed he index so ha he base year is The index is in erms of average labor produciviy, which is defined as Real Gross Domesic Produc. abor inpu( abor hour,. Employed) As can be seen, from he produciviy relaive o he base year, indusries and services have a produciviy index close o each oher over ime, whereas he farm secor is furher apar from he oher wo. In oher words, in view of average labor produciviy, I combine indusry and service secors as a non-farm secor and he agriculural secor as a farm secor. Now I urn o look a produciviy in erms of marginal labor produciviy. Presumably, if he marginal labor produciviies are no equal across secors, laborers have a endency o move from one secor o he oher; such a shif leads o higher growh since he resources are moved from a less producive secor o a more producive secor. To measure marginal labor produciviy, I esimae he per worker wage rae in each secor, assuming ha he wage rae could be a proxy for marginal labor produciviy in each secor. I look a he cos side insead of he producion side so ha I do no need o assume he form of he producion funcion. The wage raes for he indusries and services are aken from he Earning and Produciviy Saisics provided by Direcorae-General of Budge, Accouning and Saisics Execuive uan (DGBAS.) The wage rae for agriculural workers is my esimae. I esimaed he real wage for he farm worker as farm family income from agriculural aciviies divided by he number of farmers per farm household. The original daa for agriculures are from he arm Saisics. igure IV shows he esimaed yearly income for laborers in each secor from 1973 o 23. As can be seen, he annual wage income for workers in indusries and services are roughly he same and increase hand in hand, whereas ha for workers in agriculures remains consan over he period 1996 o 23, excep ha i shows a rend of growh afer igure V shows he relaive raios for he wages among secors. As can be seen, he wage raio beween indusry and services is roughly around one (i is 1.26 in 1979 and declines 15

16 o 1.12 in 23). On he oher hand, he raio of annual wage in indusries o agriculures is 3.97 in 1986 and 2.81 in 23. The wage gap beween farm and non-farm secors is huge compared wih he wage gap beween indusries and services. inally, he raio of non-farm o farm aains he highes level in he lae 198s, and drops o he level roughly equal o ha in he lae 197s. However, i sill remains above 2.5. Therefore, he produciviy in erms of he real wage, which is a proxy for marginal produciviy and average labor oupu, shows a widening gap in absolue erms beween farm and non-farm secors. Given ha he labor produciviy of he agriculural secor is smaller han ha of he non-agriculural secor, indusrializaion enables he economy o reallocae resources from a less producive secor o a more producive secor. Such a shif resuls in higher labor produciviy in erms of arihmeic, even wih no produciviy improvemen in farm or nonfarm secors. Therefore, invesigaing he overall economic growh a he aggregaed level may significanly ignore he reallocaion effec and overesimae he conribuion of non-facor inpus. Therefore, based upon he analysis in his secion, o illusrae he impac of fas indusrializaion on he aggregaed growh, I use a wo-secor framework farm and non-farm (including indusries and services) o sudy he effec of ransferring resources from a low produciviy secor o a high produciviy secor, in paricular, he economic ransiion in Taiwan. 6. Empirical Analysis wih a Two-Secor ramework: he Taiwan Case In his secion, I use a wo-secor framework and he equaions derived in secion 3 o show how he efficien use of resources, reallocaed from a low produciviy secor o a high produciviy secor, amplifies he produciviy growh in erms of average labor produciviy and oal facor produciviy. I use Taiwan as he case sudy. irs, I show he accumulaed produciviy differenial beween farm and non-farm secors widened over he period of sudy ( ). Second, I use he decomposiion formula for average labor produciviy (as in Equaion 1) o demonsrae he how efficien labor reallocaion conribues o he produciviy growh analyzed under a one-secor framework. inally, I use he decomposiion formula for oal facor produciviy (as in Equaion 2) o simulae he produciviy growh wih six cases, fixing cerain variables- e.g. percenage of facors in he farm secor or TP. 16

17 6.1 AP Differenial beween farm and non-farm secors igure VI shows he average labor produciviy differenial beween farm and nonfarm secors from As can be seen, for he farm secor, he AP in 23 was 6.28 imes wha i was in 1952; whereas for he AP in he non-farm secor, he value in 23 was 1 imes wha is was in In oher words, he non-farm secor and he farm secor have differen raes of growh in erms of AP. Consequenly, ransferring labor from a farm o a non-farm secor akes advanage of his unbalanced rapid growh more han an economy wihou any srucural ransformaion. 6.2 Percenage conribuion from he reallocaion effec o AP growh igure VII shows he 1-year moving average of he percenage conribuion o growh from he reallocaion effec and pure produciviy improvemen in erms of AP growh. As can be seen, he impac of produciviy improvemen on growh followed a downward sloping rend from he beginning and reached is firs valley in 1974 (from 84% o 56%). aer on, he conribuion of he reallocaion effec o growh remains a a relaively high level compared o he early 195s, flucuaes around 4%, and hen reaches is zenih in 1992 (49%). In oher words, he reallocaion effec is more significan in he laer period of my sudy. I repor he yearly value in Appendix 2. The percenage conribuion from produciviy improvemen and he reallocaion effec are lised in he able. As shown, he conribuion of reallocaion effec begins o dominae ha of produciviy improvemen more frequenly in he 198s and 199s: The reallocaion effec dominaes he produciviy improvemen for wo ou of 27 years (wih average conribuion of 3% from 1953 o 1979), whereas in eigh ou of 26 years he reallocaion effec dominaes he produciviy improvemen wih an average conribuion 4% from 198 o 23. Therefore, one can conclude ha he labor reallocaion effec becomes imporan when here is a big labor produciviy gap relaive o he iniial aggregae labor produciviy. Tha is one of he reasons why he reallocaion effec in he laer period of he sudy is significan, alhough he shif is more rapid in he earlier periods. I seems ha he reallocaion of resources may accoun for he relaively high aggregaed TP growh in he 198s. I could be ha human capial does no accumulae fas enough o cach up wih he produciviy gap. Thus, he produciviy differenial is no fully refleced in he earlier period. 17

18 By he 198s, enough human capial was buil up o exploi he produciviy gap. As a resul, a iny fracion of he labor ransferred ou of he farm secor was responsible for a bulk of he produciviy growh. The srong reallocaion effec in he 198s resuled from he fac ha he produciviy gap widened during his ime, as shown in igure VI. Such a resul is consisen wih oung s (1995) findings ha he TP growh in Taiwan remained high in he 198s afer aking ino accoun qualiy adjusmen. Therefore, when using AP as he measuremen for produciviy, he labor reallocaion effec is significan in accouning for high produciviy growh in Taiwan, especially in he 198s. 6.3 Percenage conribuion from he reallocaion effec o he TP growh Wih respec o TP growh, I demonsrae he conribuion of efficien reallocaion of resources in a manner differen from he previous secion because he decomposiion of he reallocaion effec lacks a cerain degree of freedom o divide he produciviy growh ino reallocaion effec and produciviy improvemen. Using Equaion 2, I ran six simulaed cases and I summarize he cases in able VIII. Table 8: The six cases for he simulaion Assumpion Abbreviaion Case 1* The TP in daa (he real TP) TP Toal Case 2 No labor reallocaion No Case 3 No capial reallocaion No K Case 4 No TP improvemen No TP Case 5 No TP improvemen in he farm secor No TP Case 6 No TP improvemen in he non-farm secor No N TP *: This case akes ino accoun all he TP improvemen and facor reallocaion (equivalen o he real daa). The furher apar he case value is away from he real daa (Case 1), he greaer is he conribuion of he absen facor in he case. igure VIII shows six cases. As can be seen, he TP acquired from Case 3: no K or Case 5, no TP does no reduce he Case 1: TP oal much. However, he figure shows in Case 4: No TP or in Case 6: No N TP, he TP grows in he opposie direcion he TP declines raher han he increases, whereas he real TP (TP oal) rises. 18

19 inally, as in Case 2, no, he resuling TP is more han 28% off he TP oal. In oher words, efficien facor reallocaion, especially shifing labor from a less producive secor o a more producive secor has a significan posiive impac on he arihmeic produciviy improvemen. In shor, reallocaing labor from he farm o he non-farm secor does increase he calculaed TP growh on op of he produciviy improvemen due o echnological advances. Therefore, aking ino accoun he reallocaion effec would clear par of he mysery of he rapid growh in Taiwan 7 Conclusion In his paper, I measure produciviy improvemen under a wo-secor raher han a one-secor framework, moving away from he one-secor framework ha radiional growh accouning is based upon. In he firs secion, I showed empirically ha produciviy improvemen is fas in hose low-income economies converging o he developed economies afer he Second World War e.g. Japan, Souh Korea and Taiwan. Meanwhile, hese economies experienced rapid srucural ransformaion, swiching from agriculural economies o indusrialized socieies. Therefore, he paper linked he fas srucural ransformaion wih high produciviy growh and found ha srucural ransformaion is significan in accouning for he growh miracles. The arihmeic decomposiion of he wo definiions of produciviy - average labor produciviy and oal facor produciviy shows ha he effec of shifing labor from a low produciviy secor o a high produciviy secor has an undeniable effec in accouning for he high produciviy obained from he analysis based on he one-secor framework. Therefore, I conclude ha he arihmeical high produciviy improvemen is aribued o he combined resuls of he srucural ransformaion of he economy and some echnological advancemen. Secoral ransformaion magnifies he growh aribued o echnological advancemen. In oher words, he mysery of high growh in produciviy can be resolved when aking ino accoun he effec of efficien reallocaion of producion facors, especially labor, on an economy experiencing ransformaion from an agriculural o an indusrialized and a pos-indusrialized economy. Noe ha I do no argue ha he efficien resource reallocaion guaranees rapid economic growh bu raher explains i. 19

20 The decomposiion exercise leads direcly o one possible area for furher research: o sudy wha causes economic ransformaion. One possible explanaion saes ha produciviy differenials and he resuling wage differenial among secors creaes an incenive for labor o move from a low produciviy secor o a high produciviy one. or example, adopion of new echnology in he non-farm secor creaes jobs in his secor, increasing he facor reurn in he non-farm secor relaive o he farm secor. Thus, laborers are willing o leave he farm and work in he more producive secor. Based on he neoclassical model, such a ransiion should be rapid. However, he empirical daa show ha he produciviy gap among he secors does no decrease immediaely. Perhaps his is due o a mechanism or a fricion ha slows down he immediae closing of he produciviy gap beween he farm and non-farm secor. Thus, esablishing a model ha includes such a fricion (e.g. model wih ime o build human capial) can explain: why he Eas Asian growh successes experienced higher produciviy growh han he oher successes (e.g. he U.S. and Canada). Moreover, i provides an explanaion for why some economies experience he same srucural ransformaion bu a a differen pace wih he resul of lower produciviy growh. 2

21 Reference Barro, Rober J., and Sala-I-Marin, Xavier. (24) Economic Growh 2 nd Ediion, Massachuses Insiue of Technology Jorgenson, Dale W. and ip, Eric. (21) Whaever Happened o Produciviy? in Hulen, Charles R., Dean, Edwin R., and Harper, Michael J. New Developmens in Produciviy Analysis, The Universiy of Chicago Press Maddison, Angus. (21) The World Economy, A Millennial Perspecive Developmen Cenre Sudies, OECD oung, Alwyn (1995), The Tyranny of Numbers: Confroning he Saisical Realiies of he Eas Asian Growh Experience Quarerly Journal of Economics Saisical Absrac of Naional Income in Taiwan Area he Republic of China, Saisical absrac of he Unied Saes 23. Bureau of Saisics, Treasury Deparmen, USA Earning and Produciviy Saisics hp://win.dgbas.gov.w/dgbas4/bc5/earning/h456e.asp Agriculural Saisical Absrac in Taiwan Area he Republic of China, hp://sa.coa.gov.w/dba_as/as_roo.hm Hisorical Saisics of Canada, hp:// Comparaive Civilian abor orce Saisics, 1 Counries, , US Bureau of abor Saisics, oreign abor Saisics fp://fp.bls.gov/pub/special.requess/oreignabor/flslforc.x 21

22 Appendix 1: IO repor for economic Acive populaion, (Uni: %) ear Change Ausralia Belgium Canada Denmark rance Germany Ialy Neherlands New Zealand Porugal Spain Sweden Swizerland Unied Kingdom Unied Saes Greece Hungary Iceland Israel Ireland Poland Turkey ain American Counries Argenina Brazil Chile Colombia Cosa Rica Cuba Ecuador Honduras Panama Peru Mexico Uruguay Puero Rico Asian Counries Hong Kong, China Japan Korea, Republic of Singapore China India Indonesia Malaysia Philippines Thailand

23 Appendix 2: AP ear Produciviy Improvemen Reallocaion Effec ear Produciviy Improvemen Reallocaion Effec % 5.723% % % % 14.81% % 46.58% % 21.39% % % % % % % % 2.668% % 49.18% % % % % % % % 74.17% % % % % % % % % % 16.94% % 25.86% % % % % % 2.92% % % % % % % % % % % % 5.759% % 17.88% % % % % % % % % % % % 5.41% % % % % % % % 25.81% % % % 4.767% % % % % % % % % % % % 6.998% % % % % % % 23

24 Secoral GDP o he aggregaed, % Agriculures Indusries Services Source: Agriculural Saisiics igure I: Secoral GDP o he aggregaed, Percenage of labor in each secor o oal employed % year Agriculure Indusry Service Source: Agriculural Saisics igure II: Percenage of abor in Each Secor relaive o he oal employed, abor Produciviy Index, in value, consan price a 1996, Index, 1981= Agriculures Indusries Services ear Source: DGBAS, Produciviy Saisics igure III: Produciviy Index,

25 early earning, consan a 1996 price, NT$ 1996 pric Income per farmer Income per worker in Indsury Income per worker in Service ear igure IV: early Earning Per Worker in Each Secor Wage rae raio among secors 5. Raio indusry/farm services/indusries nonfarm/farm ear igure V: Wage Raio among Secors Average abor produciviy consan 1996 NT$ arm Non-farm ear igure VI: Average abor Produciviy in he arm and Non-farm Secors 25

26 % Conribuion of AP from produciviy improvemen and reallocaion (1 year moving average) 9.% 8.% 7.% 6.% 5.% 4.% Produciviy Improvemen Reallocaion Effec 3.% 2.% 1.%.% ear igure VII: Percenage Conribuion of abor Reallocaion Effec on AP Growh igure VIII: TP decomposiion 26