Growth Accounting and Productivity Title Industrial Sectors in Korea ( Author(s) Pyo, Hak K.; Rhee, Keun-Hee; Ha, Bo

Transcription

1 Growth Accountng and Productvty Ttle Industral Sectors n Korea ( Author(s) Pyo, Hak K.; Rhee, Keun-Hee; Ha, Bo Ctaton Issue Date Type Techncal Report Text Verson publsher URL Rght Htotsubash Unversty Repostory

2 H-Stat Dscusson Paper Seres No.171 Growth Accountng and Productvty Analyss by 33 Industral Sectors n Korea ( ) Hak K. Pyo Geun-Hee Rhee Bongchan Ha June 2006 Htotsubash Unversty Research Unt for Statstcal Analyss n Socal Scences A 21st-Century COE Program Insttute of Economc Research Htotsubash Unversty Kuntach, Tokyo, Japan

3 Growth Accountng and Productvty Analyss by 33 Industral Sectors n Korea ( ) June 30, 2006 Hak K. Pyo, Keun Hee Rhee, and Bongchan Ha* *Professor, Dvson of Economcs, Seoul Natonal Unversty, (E-mal) pyohk@plaza.snu.ac.kr ; Senor Researcher, Korea Productvty Center ; Researcher, Insttute of Economc Research, Seoul Natonal Unversty

4 1. Introducton In recent years, both growth accountng and productvty analyss by sector have attracted renewed attenton on a global scale. The decomposton of nputs nto captal (K), labor(l), energy(e), and ntermedate materals(m) for detaled ndustry-level analyss of productvty growth was frst appled to the post-war US Economy by Jorgenson, Gollop, and Fraumen (1987). The basc KLEM-methodology has been extended to cover eght European countres by the European KLEM proect as explaned n Tmmer (2000) and Canada-Japan-USA database. In addton, the number of decomposed nputs has ncreased to nclude mported goods (I) and servces (S) formng the framework of KLEMS methodology as done by Forsgerau and Sorenson (1999). Earler studes along the lne of the KLEM approach n Korea such as Kwon and Yuhn (1990), of whch man nterests were lmted to estmatng elastcty of substtuton and productvty growth, usng data n manufacturng sector only or value-added accountng. More recent emprcal works such as Km and Hong (1997), Pyo(2001), Tmmer and van Ark(2000), and Rhee(2001) have also appled value-added accountng. The earler excepton of applyng gross-output accountng was Km and Park (1985) but t was also lmted to manufacturng sector. Therefore, t would be desrable for us to attempt a growth accountng of gross output n a consstent framework of the KLEM approach. In addton to growth accountng, the nternatonal comparson of productvty among countres requres a consstent database for purchasng power party (PPP) at ndustral level by each country. The purpose of ths paper s to dscuss the applcaton of the KLEM-methodology to the Korean database and present the result of gross output growth accountng and productvty analyss. After a revew of avalablty of data, we have constructed the detaled database of consstent output and nputs data for the Korean economy for the perod of The data n constant prces are avalable wth 1995 as the base year and the decomposton of data s as detaled as specfed by the summary of the frst ICPA proect by Kuroda (2001) and qute comparable to the Japanese database. The paper s organzed as follows. Secton 2 examnes database on Korea ncludng gross output, four nputs and value-added from natonal accounts and Input-Output Tables and other sources of data. Secton 3 presents the result of growth accountng and productvty analyss. The last secton concludes the paper. 2

5 2. Database for Korea ( ) 2.1 Gross Output Data from Natonal Accounts and Input-Output Table Natonal accounts by the Bank of Korea (1999, 2004) reports annual seres ( ) of gross output, ntermedate consumpton, GDP, ndrect taxes, consumpton of fxed captal, domestc factor ncome, compensaton of employees, and operatng surplus of 21 ndustres ncludng 9 manufacturng ndustres and 3 sub-sectors of government servces n current prces followng 1993 UN System of Natonal Accounts. The Bank of Korea has also publshed Input-Output Tables snce Its most recent 2000 Input-Output Table s the 19 th Table. The detaled descrpton of Input-Output Tables durng s summarzed n Table 1. The Table for 1995 has 402, 168, 77, and 28 ndustral sectors n basc, small, medum, and large classfcatons, respectvely. Therefore, the estmaton of tme seres Input-Output Tables followng those methods descrbed n Kuroda (2001) would be requred f we have to estmate KLEM model wth more than 21 ndustral classfcatons snce Input-Output Tables are avalable only n selected years. We have attached the reclassfcaton of IO tables n Table 2 and 3. For the present study, we have generated gross output and value-added by 33 ndustres through RAS method. The generated annual data of both gross output and value-added have been adusted to match aganst Natonal Income Accounts whch do not contan both ndrect tax and subsdy. Snce RAS method s senstve to the benchmark year s value of the I/O coeffcents, we have used the I/O Table n the closest year as benchmark value. We have used V-Table to generate commodty prces by 33 sectors and then used the generated commodty prces to estmate output prces by 33 sectors. 3

6 Table 1. Input-Output Tables n Korea ( ) 1. Transacton Tables at Producers Prces (number of sector classfcaton) Year Basc Small Medum Large * Extended I-O tables wth sector classfcaton of the precedng man I-O tables. 2. Lnked Input-Output Tables (number of sector classfcaton) Year small medum Large Supportng Tables (1) Employment Table: * 1995* 1998* 2000* (*ncludng employment matrx) (2) Fxed Captal Formaton Table: U-Table and V-Table (21 ndustres ncludng 9 Manufacturng ndustres) Annual U-Table at 1990, 1995, or 2000 constant prces and annual V-Table n both current and constant prces for are avalable n 1994, 1999, 2004 Natonal Accounts. Sources : The Bank of Korea Input-Output Tables( ) CD-ROM 4

7 Table 2. Reclassfcaton of IO Tables nto 33 sectors (basc classfcaton) 33 Sectors Classfcaton n IO Table(1983) Classfcaton n IO Table(1985) 1. Agrculture 2. Coal Mnng 3. Metal and Mon-metal 4. Ol and Gas 5. Constructon 6. Food 7. Textle 8. Apparels 9. Lumber and Wood 10. Furnture 11. Paper alled 12. Prntng, Publshng, alled 13. Chemcals 14. Petroleum products 15. Leather 16. Stone, Clay, Glass 17. Prmary metal 18. Fabrcated machnery 19. Machnery 20. Electrcal machnery 21. Motor 22. Transportaton equp. 23. Instrument 24. Rubber and Msc. Plastc 25. Msc. manufacturng 26. Transportaton 27. Communcaton 28. Electrc utlty 29. Gas and water utlty 30. Trade 31. Fnance 32. Other Prvate servce 33.Publc servce 1~38 1~ ~39 40~54, 56~58 40~44, 46~ ~ ~342 59~98 52~91 99~118, 121, 123~125 92~112, 117~ ~120, 122, ~116, 123, ~133, 135~ ~131, 133~ , 237, , 238, ~ ~ ~ ~ ~172, 174~ ~ ~ ~ ~ ~128, ~ ~ ~217, 228~ ~218, 229~ ~227, 233~236, 238~242, 244~ ~228, 234~237, 239~243, 245~ ~ ~ ~ ~ ~ ~ ~291, 296~ ~295, 300~ ~ ~ , 195, 197~ ~185, 196, 198~ ~ ~314, ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~375, 378~ ~377 5

8 33 Sectors Classfcaton n IO Table(1990) Classfcaton n IO Table(1995/1998) 1. Agrculture 2. Coal Mnng 3. Metal and Mon-metal 4. Ol and Gas 5. Constructon 6. Food 7. Textle 8. Apparels 9. Lumber and Wood 10. Furnture 11. Paper alled 12. Prntng, Publshng, alled 13. Chemcals 14. Petroleum products 15. Leather 16. Stone, Clay, Glass 17. Prmary metal 18. Fabrcated machnery 19. Machnery 20. Electrcal machnery 21. Motor 22. Transportaton equp. 23. Instrument 24. Rubber and Msc. Plastc 25. Msc. manufacturng 26. Transportaton 27. Communcaton 28. Electrc utlty 29. Gas and water utlty 30. Trade 31. Fnance 32. Other Prvate servce 33.Publc servce 1~34 1~30 35~36 31~32 39~50 35~45 37~38 33~34 325~ ~329 51~93 46~88 94~109, 113~115 89~104, 111~ ~112, 116, ~108, ~ ~ , 238, ~ ~ ~ ~ ~ ~ ~ ~ ~ ~121, 123~ ~110, 114~117, ~ ~ ~212, 223~ ~198, 209~ ~222, 228~237, 289, 241~ ~208, 214~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~368, 378~ ~351, 360~369, 372~ ~ ~371 6

9 33 Sectors Classfcaton n IO Table(2000) 1. Agrculture 2. Coal Mnng 3. Metal and Mon-metal 4. Ol and Gas 5. Constructon 6. Food 7. Textle 8. Apparels 9. Lumber and Wood 10. Furnture 11. Paper alled 12. Prntng, Publshng, alled 13. Chemcals 14. Petroleum products 15. Leather 16. Stone, Clay, Glass 17. Prmary metal 18. Fabrcated machnery 19. Machnery 20. Electrcal machnery 21. Motor 22. Transportaton equp. 23. Instrument 24. Rubber and Msc. Plastc 25. Msc. manufacturng 26. Transportaton 27. Communcaton 28. Electrc utlty 29. Gas and water utlty 30. Trade 31. Fnance 32. Other Prvate servce 33.Publc servce 1~30 31~32 35~45 33~34 312~328 46~86 87~102, 109~ ~106, ~ ~ ~ ~ ~ ~ ~108, 112~115, ~ ~196, 207~ ~206, 212~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~351, 361~371, 374~ ~373 7

10 Table 3. Reclassfcaton of Natonal Accounts nto 33 Sectors Natonal Accounts 33 Sectors 1. Agrculture, Forestry and fshng 1.Agrculture 2. Mnng and Quarryng 2. Coal Mnng 3. Food, Beverage and Tobacco 6. Food 4. Textle and Leather 7. Textle 3. Metal and Mon-metal 4. Ol and Gas 8. Apparels 15. Leather 5. Wood, Paper, Publshng and Prntng 9. Lumber and Wood 6. Petroleum, Coal, and Chemcals 13. Chemcals 11. Paper alled 12. Prntng, Publshng, alled 14. Petroleum products 24. Rubber and Msc. Plastc 7. Non-Metallc Mneral Products except Petroleum and Coal 16. Stone, Clay, Glass 8. Metal, Fabrcated Metal Products 17. Prmary metal 18. Fabrcated machnery 9. Machnery and Equpment 19. Machnery 10. Transport Equpment 21. Motor 20. Electrcal machnery 23. Instrument 22. Transportaton equp 11. Furnture and Other Manufacturng Industres 10. Furnture 25. Msc. manufacturng 12. Electrcty, Gas and Water 28. Electrc utlty 29. Gas and water utlty 13. Constructon 5. Constructon 14. Wholesale and Retal Trade, Restaurants and Hotels 30. Trade 15. Transport, Storage and Communcaton 26. Transportaton 27. Communcaton 16. Fnance, Insurance, Real Estate and Busness Servces 31. Fnance 17. Communty, Socal and Personal Servces 32. Other Prvate servce 18. Producers of Government Servces 33.Publc servce 8

11 2.2 Measurement of Captal Stocks The success of late ndustralzaton by newly ndustralzng economes could not have been made possble f both the rapd accumulaton of captal and ts changng dstrbuton among sectors were not realzed n ther development process. However, t s dffcult to dentfy these factors emprcally because the tme seres data of captal stocks n fastdevelopng economes by both types of assets and by ndustres are not readly avalable. The lack of nvestment data for a suffcently long perod of tme to apply the perpetual nventory estmaton method was the man cause of the problem. However, the Natonal Statstcal Offce of the Republc of Korea has conducted naton-wde natonal wealth survey four tmes snce Korea s one of a few countres whch have conducted economy-wde natonal wealth surveys at a regular nterval. Snce the frst Natonal Wealth Survey (NWS) was conducted n 1968, the subsequent surveys were made n every ten years n 1977, 1987, and 1997, respectvely. Snce such regular surveys wth naton-wde coverage are very rare n both developed and developng countres, an analyss on the dynamc profle of natonal wealth seems warranted to examne how natonal wealth n a fast growng economy s accumulated and dstrbuted among dfferent sectors. The estmaton of natonal wealth by types of assets and by ndustres was made by Pyo (1998) and updated n Pyo (2003) by modfed perpetual nventory method and polynomal benchmark-year estmaton method usng four benchmark-year estmates. The latter study modfes and extends the earler one n two respects. Frst, the result of 1997 NWS has been released n 1999 so that we can make use of addtonal benchmark-year estmates. Second rebasng the estmates of captal stocks from 1990 prces to 1995 prces seems nevtable because Bank of Korea has re-based ther natonal accounts accordngly. 1) Natonal Wealth Survey n Korea In Natonal Wealth Survey (NWS), the gross captal stock (GK) was evaluated by multplyng the purchase prce of the fxed tangble asset by the approprate prce ndex by types of assets whch have been compled and prepared by the Bank of Korea datng back to the year 1910 as follows: GK = P PI t t where GK s the value of gross captal stock of asset evaluated at a certan benchmark year m (m>t), P t s the purchase prce of the asset n year t, and PI t s the prce ndex to reflect the value of the asset n year t. In other words, the gross captal stock s supposed to reflect the 9

12 repurchase value or reacquston value of the fxed tangble asset. Second, the net captal stock (NK) was deduced by multplyng the gross captal stock by the resdual cost rato as follows: n N t t NK = GK S ( S ) = P PI ( ) P P t where S s the value of survved assets, N s average servce lfe of the asset and n s the number of years elapsed. In other words, a proportonal deprecaton method s adopted. The value of survved assets s assumed to be 10 percent of the purchase value when t reaches the assumed average servce lfe and 1 percent of the purchase value when t reaches double the assumed average servce lfe. Suppose for example, the average servce lfe of a certan asset such as personal computer s 5 years and t was purchased three years ago at the prce of 3,000 dollars. In addton, assume that the nflaton rate of the computer prce snce the purchase year s 20 percent. Then the followng calculatons can be made: t n N GK = 3, = 3,600 (1) NK = 3, ( ) (2) 3,000 Korea s NWS assumes shorter average lves than BEA(1993). For example, BEA s are years for resdental structure and years for nonresdental structure whle Korea s NWS assumes average servce lfe of years for resdental buldngs and 8-60 years for nonresdental buldngs. For automobles, Korea s assumed average lfe (4-5 years) s shorter than BEA s (10 years). The shorter average lfe and, therefore, hgher deprecaton rate are typcal n the process of late ndustralzaton. 2) Estmaton of Net and Gross Captal Stock (1) Estmatng Method for In prncple the exstence of four benchmark year estmates of gross and net captal stocks makes t possble for us to apply the polynomal benchmark year estmaton method. In Pyo s earler studes (Pyo 1988, 1992, and 1998), he estmated proportonal retrement rates and deprecaton rates both by types of assets and by ndustres based on the polynomal equatons. When we appled the polynomal benchmark year equaton to estmate the proportonal retrement rates for the sub-perods of and , most of estmates became negatve 10

13 ncludng the average economy-wde retrement rates (-3.0% for and 3.1% for ) except other Constructon(0.6%) and Transport Equpment(3.4%) n and Nonresdental Buldng(0.9%) n Therefore, followng Pyo (1998), we have appled the polynomal benchmark year estmaton method to estmatng deprecaton by types of assets only. Thus we have generated net stocks by types of assets frst for the perod of and then, dstrbuted them over dfferent sectors of ndustres by usng nterpolated ndustral weghts between the respectve benchmark years. We have decded to estmate net captal stock frst and then to estmate gross captal stock by usng nterpolated net-gross converson ratos for the followng two reasons. The basc reason s due to the fact that the margn of predcton error from the polynomal benchmark year equaton turns out to be larger wth gross captal stock than wth net captal stock as had been observed n Pyo (1992). (2) Estmatng Method for Snce we have decded to use estmates of 1968 NWS as the frst benchmark year estmates, we have to estmate captal stocks for the perod of usng the perpetual nventory equaton backward. We frst deflated the net stock data of 1968 NWS n current prces nto those n 1995 prces usng mplct GDP deflator of captal formaton. Then, we have estmated captal-output coeffcents by ndustres by regressng net captal stock n 1995 constant prces on real GDP by ndustres and a lnear tme trend varable durng the perod of n order to generate estmate of net stocks by ndustres durng n 1995 prces. We have also used the 1968 NWS weghts of dfferent types of assets for the perod of In order to estmate net stocks by ndustres for the perod of , we have used both 1968 NWS and 1977 NWS to estmate nterpolated ndustral weghts by each type of assets. Then, for the perod of , we have used both cumulated weghts of captal formaton by ndustres from old Natonal Accounts by the Bank of Korea (1984) and ndustral weghts of 1968 NWS. (3) Estmatng Method after 1997 Natonal Statstcal Offce of Korea has decded to termnate Natonal Wealth Survey by 1997 and to swtch from drect estmaton to ndrect estmaton of natonal wealth followng the method of BEA and OECD. The cost of such drect natonal wealth survey has ncreased sgnfcantly as the sze of natonal economy has expanded consderably. In addton, some of the partcpatng nsttutons such as Kookmn Bank for unncorporated busness enterprses have been prvatzed so that Natonal Statstcal Offce alone can no longer afford natonal 11

14 wealth survey. Japan had termnated ts Natonal Wealth Census n 1970 for almost the same reasons. Therefore, for the perod after 1997 whch s the last natonal wealth survey, we have to estmate captal stocks by a modfed perpetual nventory method usng 1997 NWS as benchmark estmates. Frst, we estmate net stocks by type of assets n constant prces by usng the deprecaton rates estmated from the perod of and dstrbute them across ndustres usng both ndustral weghts n 1997 NWS and those n subsequent Mnng & Manufacturng Census and Surveys and Wholesale and Retal Surveys. In the long run, the estmated deprecaton rates by type of assets may need to be updated and revsed by the mcro data-based studes. Second the generated net stocks by type of assets and by ndustres have to be converted nto gross stock by usng the net-gross converson rato of 1997 NWS for the tme beng. But ultmately we may need further studes on the trend of net-gross converson rato by type of assets and by ndustres and the average asset lfe. 3) Reconclaton wth Database of Pyo (2003) Snce the database of Pyo (2003) covers 10 broad categores of ndustral sector together wth 28 sub-sectors of Manufacturng, t can be reclassfed and reconcled wth 33-sector classfcaton for the ICPA proect. Assumng that the flow of captal servce s proportonal to captal stock, we used the average captal stock of two years as the captal servce 1. In order to make qualty adustments to the captal nput data, we have taken the followng steps: (1) Followng Kuroda(2001), we defne the captal servce of asset n ndustry as K ( t) = b ( t) A ( t) = 1,.., n; = 1,... m 1 = b ( t) [ A ( t) + A ( t 1)] 2 (3) where b (t) denote the proporton of the -th asset type on the -th sector s total captal servce nput A (t) whch s the average of unweghted sum over all assets durng the t-th and (t-1)th perod. 1 We could not use the formula of Kuroda and Nomura (1999) because nvestment data n Natonal Income Accounts are classfed ether by asset type or by ndustry but not by both. 12

15 (2) The growth rate of captal servce nput s defned as [ ln A ( t) ln A ( t 1) ] ln K ( t) ln K ( t 1) = + [ ] v ln b ( t) ln b ( t 1) l = 1,2,..J (4) where v s the average share of an ndvdual component n the value of property compensaton. The frst term on the rght sde s the change of the quantty of captal servce and the second term s the change of the qualty of the captal servce. The growth rate of the qualty of captal was very small n comparson to the growth rate of the quantty of captal. There was no substantal change n the structure of captal n Korea durng ) Measurement of Captal Input Prce Followng Jorgenson, et.al (1987) and Tmmer (2000), the aggregate ndex of captal servces over the dfferent types of assets n -sector( (t) ) can be assumed as a translog functon of the servces of ndvdual assets ( K A (t) ) as follows: [ ln A ] ( t) ln A ( t 1 ln K ( t) ln K ( t 1) = v ) where weghts are gven by the average shares of each type of captal n the value of property 1 compensaton, v = [ v ( t) + v ( t 1) ] 2 p ( t) K ( t) and v ( t) = wth P (t) the rental p ( t) K ( t) prce of captal servces from asset type. In order to apply the above aggregaton formula, t s necessary to mpute the rental prces of captal servces. In the absence of taxaton, Hall and Jorgenson (1967), Jorgenson, Gollop and Fraumen (1987) and Jorgenson and Yun (1991) have derved the followng formula for mputng the rental prce of captal servces from asset type : P ( t) = { r( t) + δ Π( t)} q ( t 1) (5) where r(t) s the rate of return, (t) s the acquston prce of nvestment good wth q [ q( t) q ( t 1)] Π ( t) = whch s the rate of nflaton n the prce of nvestment good. The q ( t 1) 13

16 nomnal rate of return after tax s usually assumed to be the same for all assets n an ndustry so that r(t) does not have subscrpt. The acquston prces of each asset n dfferent ndustres are not usually avalable and, therefore, nvestment deflators are frequently used as substtutes for the acquston prces. But nvestment deflators n Natonal Accounts are avalable ether by types of assets or by ndustres not by both. Estmates of deprecaton rates n Pyo(2003) are also avalable ether by types of assets or by ndustres not by both. In case of Korea, there have been studes by Yun and Km (1997) and Won and Hyun(2000) on the estmaton of effectve margnal tax rates followng methods developed by Kng and Fullerton (1984) and Jorgenson and Landau (1993). Snce they have used the above formula for cost of captal, they must have generated the mputed cost of captal. But ther estmates are based on whole ndustres not for each ndustry and even those estmates are not avalable. Faced wth lack of data and consstent estmates for the varables to mpute rental prce of captal n each ndustry, we have adopted the followng approach. In order to get captal nput prces dfferent for both each asset and each ndustry, we have slghtly changed Eq (5) to Eq (5), whch s the formula of the captal nput prce for both each asset and each ndustry : P ( t) = { r ( t) + δ Π ( t)} q ( t 1) (5) In Eq (5) we have assumed that the prce of nvestment asset ( ), the rate of deprecaton of asset ( δ ), and the nflaton rate of nvestment asset ( ) are dentcal across all ndustres. But we have assumed that the rate of return can be dfferent n each ndustry. The applcaton of the Eq. (5) requres data on the rate of return by ndustry ( ), the acquston prce of nvestment asset ( q ), and the rate of deprecaton of asset ( δ ). Because we do not have data for the nomnal rate of return for each ndustry but for the nomnal value of captal servces summed over all types of assets n th ndustry, we estmated the rate of return for each ndustry, r (t), by usng the equalty between the nomnal captal ncome n th q Π r ndustry ( CI ndustry ( ) and nomnal value of captal servces summed over all types of assets n th P ( t) K ( t) ). From ths equalty we estmated the rate of return for each ndustry, 14

17 r (t) as follows: CI ( t) = P ( t) K ( t) = { r ( t) + δ Π ( t)} q ( t 1) K ( t) CI r ( t) = ( t) { δ Π ( t)} q ( t 1) K q ( t 1) K ( t) ( t) (6) Usng these nomnal rates of return of each ndustry( (t) ), we were able to calculate the rental prce of captal servces of each asset and ndustry by addng the deprecaton rate and subtractng the nflaton rate of captal, and multplyng the result to the prce of captal. r 2. 3 Measurement of Labor Input In order to measure labor nput for KLEM model, we have to obtan both quantty data of labor nput such as employment by ndustres and hours worked and qualty factors such as sex, educaton and age. Both avalablty and relablty of labor statstcs n Korea have mproved snce But the measurement of labor nput by ndustres cannot be readly made because the statstcs of employment by ndustres are not detaled enough to cover 33 sectors. Followng the characterstcs of labor nput descrbed n Kuroda (2001), the sources of labor statstcs are presented n Table 4. Economcally Actve Populaton Yearbook by Natonal Statstcal Offce reports the number of employment, unemployment, not-economcally-actve populaton and economcally actve populaton by 10 categores of age group (15-19, 20-24, 25-29, 30-34, 35-39, 40-44, 45-49, 50-54, 55-59, and 60 over). Employment by ndustres s avalable n 9 broad categores of ndustres:(1) Agrculture, Forestry and Fshng (2) Mnng (3) Manufacturng (4) Constructon (5) Wholesale, Retal, Restaurants and Hotel (6) Electrcty and Gas, and (7) Transportaton, Storage and Communcaton and (8)Fnance, Insurance, Real Estate, and Busness Servce (9)Other Servces. More detaled classfcatons of employment wll have to rely on Employment Table, whch s publshed as a supportng table to Input- Output Table. But t s avalable only every fve year when man Input-Output Tables are 15

18 publshed. Mnng and Manufacturng Census (Survey) by Natonal Statstcal Offce also report employment statstcs but t s lmted to mnng and manufacturng only. Unemployed persons by gender and educatonal attanment are also avalable from the same source. Report on Monthly Labor Survey by Mnstry of Labor publshes monthly earnngs and workng days of regular employees by 12 broad categores of ndustres. Survey Report on Wage Structure by the same mnstry reports wages by 6 categores of occupatonal classfcaton n old seres ( ) and now reports 9 new categores n new seres ( ): (1) Senor Offcals and Managers (2) Professonals (3) Techncans (4) Clerks (5) Servce and Sales Workers (6) Sklled Agrculture and Fshery Workers (7) Craftsmen and Assembler (8) Plant and Machne Operator and (9) Other Laborer. Nomnal and real wage ndex are also avalable from Report on Monthly Labor Survey by Mnstry of Labor. Table 4. Characterstcs of Korean Database for KLEM Model (1)Gross output Number of Sectors -Natonal Accounts 21 -Input-Output Table (1980,1985,1990,1995,2000) U-Table and V-Table Annual tables ( ) 22 (2)Captal stock (captal servces) -Gross and Net Stock data: Natonal Wealth Survey (1987)(1997) 5 types of assets -Captal Formaton data: Natonal Accounts(1999, 2004) -Annual Gross and Net Stock by Industres and by Types of Assets and 10 large ndustres and Estmated Deprecaton Rates are avalable n Pyo (2001) wth 28 sub- -Net stock and Investment data: Mnng and Manufacturng Census manufacturng Mnng and Manufacturng Survey ndustres - Types of assets: resdental buldng, nonresdental buldng, other constructon, transport equpment, machnery equpment (3)Labor nput (hours worked) 18 types -Survey Report on Wage Structure (Mnstry of Labor) -Economcally Actve Populaton Survey (NSO) -sex (male/female) -educatonal attanment (mddle school and under, hgh school, and college and/or above) -age classes 2 types 3 types 3 types 16

19 (16~34, 35~54, and 55 above) (4)Intermedate nputs -Natonal Accounts -Input-Output Table(1980, 1985, 1990, 1995, 2000) -U-Table and V-Table Annual Tables( ) -Energy Input; sectors 2, 4, 14, 28, and 29 For the present study, we have obtaned the raw data fle of Survey Report on Wage Structure from the Mnstry of Labor and Economcally Actve Populaton Survey from Natonal Statstcal Offce for the perod of The data are classfed by two types of gender (Male and Female), three types of age (16-34, 35-54, and 55 above), and three types of educaton (mddle school and under, hgh school, and college and/or above) and, therefore, there s a total of 18 categores of labor. Snce the raw-data fle of the Survey Report on Wage Structure contans more detaled ndustral classfcaton than that of the Economcally Actve Populaton Survey, we calculated the quantty of labor from the Economcally Actve Populaton Survey and the qualty of labor from the Survey Report on Wage Structure. Ths enables us to nclude self-employed labor as well as to use more detaled data. However, snce the Survey Report on Wage Structure does not nclude the Agrculture and Government sectors, we had to use the average value of the entre economy for the qualty measure of these two sectors. In order to make qualty adustments to the employment data, we have taken the follow steps: P Ll 1) Defnng as wage rate for -sector and l-type category of labor, the share of labor ncome by l-type category of labor n -sector can be expressed as; p L v (7) Ll Ll l = pll Ll The average weght of -sector and l-type labor ncome durng the perod of (t-1) and t can be generated as; [ v ( t) + v ( 1) ] 1 vll = Ll Ll t (8) 2 2) In order to make a qualty adustment to labor nput data, we have further decomposed labor 17

20 nput of -sector and l-type as follows: L ( t) d ( t) M ( t) H ( t) l = l (9) where d l denotes relatve weght of workng hours of l-type n -sector. In other words, L l (t) measures labor nput of l-type labor n -sector. average workng hours of -sector respectvely. M and H denote the employment and 3) Fnally, the growth rate of -sector labor nput has been computed as follows: ln L ( t) ln L ( t 1) = [ ln M ( t) ln M ( t 1) ] + [ ln H ( t) ln H ( t 1) ] + v Ll [ d ( t) ln d ( t 1) ] l ln = 1,2,..33 (10) where the frst bracket on the rght hand sde measures change n employment, the second bracket measures change n average workng hours, and the thrd bracket measures the change n qualty of labor through change n weghted workng hours. The average growth rate of the qualty of labor s 1.33% and t explans about 42% of the growth rate of labor. It s a relatvely hgh proporton n comparson to the proporton of the qualty of captal. 2.4 Measurement of Energy Input and Materal Input In order to separate energy nput from ntermedate nput, we have decomposed ntermedate nputs nto two nput categores followng ICPA crteron. For ths purpose, we have used IO Tables and dentfed 5 sectors (sector 2, 4, 14, 28 and 29) as energy nput sector and the remanng 28 sectors as materal nput sector. 2.5 Deflators for Gross Output and Inputs The 21-sector gross output data by Bank of Korea s natonal accounts are avalable only n current prces. For the perod after 1985, we have used V Table n both constant and current prces to generate mplct gross output deflators by sector. For the perod before 1985, we have used Lnked IO Table n constant prces to generate mplct gross output deflators by sector for 1985 and nterpolated the data for For the deflators of energy nput and materal nput, we 18

21 have used the same sources of data; V Table for the perod after 1985 and Lnked IO Table before The basc characterstcs of KLEM database n Korea ( ) n 1995 prces s presented n Table 5. Durng the perod of , Korea s gross output has grown at the average annual rate of 7.95 percent. Four nputs have grown at the rate of 9.36 % (K), 3.15 % (L), 5.28 % (E), and 8.47 % (M) respectvely. 2.6 Input Shares Regardng shares of nputs, we have used Compensaton of Employees n Gross Domestc Product and Factor Income by Knd of Economc Actvty n natonal accounts and Operatng Surplus to generate relatve share of labor nput and captal nput respectvely n total valueadded and then adusted them nto shares n total gross output. We have dvded the amount of energy nput and materal nput by gross output to generate shares of energy nput and materal nput respectvely. 3. Growth Accountng and Productvty Analyss We have conducted a growth accountng and a productvty analyss for Korea ( ). We have qualty-adusted the nput data. In other words we have used educaton, sex and age for labor nput and ndvdual assets data for captal nput to consder the changes of the nput shares. Table 5. KLEM data n Korea ( ) (1995 prces) Year Gross output K L E M (bll. won) (bll. won) (100,000hour) (bll. won) (bll. won)

22 growth(%) ) Captal(K) and labor(l) are the values wthout qualty. 2) The growth rates are the average growth rates. 3) The growth rates of captal and labor nclude both growth rates of quantty and qualty. In Table 5, we present average growth rates of gross output and four nputs n 1995 constant prces. The gross output of whole ndustres has grown at an average annual rate of 8.04 percent whle captal(k), labor(l), energy(e), and materal nput(m) have grown at the rate of 9.36 percent, 3.15 percent, 6.68 percent, and 8.65 percent respectvely durng the perod. The average estmated shares of four nputs were 0.20( v ), 0.20 ( v ), 0.08 ( v ) and 0.52 ( v ) respectvely. We have estmated total factor productvty based on both gross output growth accountng and value added growth accountng. K L E M 3.1 Gross Output Growth Accountng The economy-wde growth rate of TFP durng the pre-crss perod ( ) has been estmated as 0.20 percent as shown n Table 6. And the growth rate, durng the post-crss perod ( ), has been estmated as n 1.48 percent. After the economc crss n 1997, the growth rate of gross output has been declned, and the growth rates of nput factors lke as captal, labor, energy and materal have also reduced from those before the crss. Accordngly, the growth rate of TFP durng the post-crss perod has been relatvely hgher than that of the pre-crss perod. The relatve contrbuton of TFP to gross output growth (22.29%) ncrease n the post-crss perod seems to reflect structural adustments made after the fnancal crss of Idle capactes were cut-off and the level of employment was also drastcally reduced. 20

23 Table 6. Gross Output Growth Accountng of TFP n Whole Industry n Korean Economy <average growth rate(%)> Captal nput Labor nput gross output total captal qualty of captal quantty of captal total labor qualty quantty energy materal TFP of of labor labor 84-' ' ' contrbuton rato to output growth(%) 84-' ' ' * The growth rates of nputs are weghted by each share of ncome. We can examne the relatve contrbuton rato of the nput factors to the output growth. The contrbuton rato to output growth durng the post-crss perod has been reduced from the precrss perod; captal (18.22%), labor(2.56%), and materal(45.48%), wth the excepton of energy(11.60%). Hence, the contrbuton rato of TFP to output growth was ncreased up to percent durng the post-crss perod. The economy-wde growth rate of TFP has been estmated as 0.55 percent durng the entre perod of our analyss, , and the contrbuton rato of TFP to output growth s 6.84 percent, whch s of rather nsgnfcant magntude. The relatve magntude of contrbuton to output growth s n the order of materal, captal, labor, TFP and energy. The qualty of captal has not affected the growth of output at all, but the qualty of labor has affected the growth of output about 3.36 percent. The total factor productvty growth n gross output growth accountng s lower than that wthout qualty adustment n nput data. We are reconfrmng both Krugman's(1994) proposton and emprcal fndngs by Young(1994) and Lau and Km(1994). Secondly, the growth rate of TFP n manufacturng s gross output durng the pre-crss perod ( ) has been estmated as 1.21 percent as shown n Table 7. And the growth rate of TFP durng the post-crss perod ( ) has been estmated n 1.73 percent. The pattern 21

24 of the growth rates of TFP s smlar to the entre economy. But quantty of labor nput n manufacturng has been reduced durng the post-crss perod. It reflects a drastc structural adustment n Korea s labor market after the fnancal crss of As a consequence, the contrbuton rate of labor to output growth has become negatve. In manufacturng, the growth rate of TFP has been estmated as 1.36 percent durng the entre perod of The contrbuton rato of TFP to output growth s estmated as percent. Table 7. Gross Output Growth Accountng of TFP n Manufacturng < average growth rate(%)> Captal nput Labor nput qualty quantty quantty gross total total qualty energy materal TFP of of of output captal labor of labor captal captal labor 84-' ' ' Contrbuton rato to output growth(%) 84-' ' ' * The growth rates of nputs are weghted by each shares of ncome. Table 8 Gross Output Growth Accountng of TFP n Servces <average growth rate(%)> Captal nput Labor nput qualty quantty quantty gross total total qualty energy materal TFP of of of output captal labor of labor captal captal labor 84-' ' '

25 Contrbuton rato to output growth(%) 84-' ' ' * The growth rates of nputs are weghted by each shares of ncome. Thrdly, the growth rate of TFP n servce sectors durng the pre-crss perod ( ) has been estmated as percent as shown n Table 8. Durng the pre-crss perod the growth rate of captal n servce sector (3.08%) has been much hgher than that n manufacturng (1.17%), so ts growth mght have affected adversely on the overall effcency n servce sector. But the growth rate of TFP durng the post-crss perod ( ) has changed to 1.04 percent. The growth rates of both nput factors and gross output have slowed down durng the post-crss perod. As a result the growth rate of TFP has been estmated as 0.10 percent n servce durng the entre perod. The relatve contrbuton of TFP on output growth n servce has become lower than manufacturng. And the relatve magntude of the contrbuton to output growth s n the order of materal, captal, labor, energy and TFP. 3.2 Value Added Growth Accountng Next, we have analyzed the growth of TFP by a value-added growth accountng. In Table 9 the economy-wde growth rate of TFP durng the pre-crss perod ( ) has been estmated as 0.37 percent. And the rate durng the post-crss perod ( ) has been estmated as 0.71 percent. As a whole the growth rate of TFP s estmated to be 0.46 percent. So the pattern of TFP growth from value-added accountng s smlar to that from gross output growth accountng. But the absolute levels of TFP growth rates are lower n value-added growth accountng than those n gross output growth accountng. Agan, the effect of qualty change n captal nput was almost neglgble. But the growth and contrbuton rato of labor qualty to value-added are consderable. The contrbuton rato of labor qualty to the growth of value added s about 10 percent as a whole. The contrbuton of total labor to the growth of value added s composed of two factors, labor qualty (9.88%) and labor quantty (13.13%). The economy-wde contrbuton rato of captal to growth of value added s 70 percent, that of labor s 23 percent, and that of TFP s 7 percent. In value added context the role of captal n producton s therefore greater than labor as shown n Table 9. 23

26 Table 9 Value Added Growth Accountng of TFP n Whole Industry Value Added total captal labor Captal Input <average of log growth rate(%)> Labor nput qualty of quantty qualty of total labor captal of captal labor quantty of 84-' ' ' Contrbuton rato to output growth(%) 84-' ' ' ) The growth rates of each nput are weghted by ts share of ncome. 2) Value added s deflated by 1995 prce. TFP 3.3 Sectoral Growth of TFP In manufacturng, the sectors such as electrcal machnery, machnery, motor, rubber, and nstrument have showed relatvely hgher growth rates of TFP durng the entre perod of as shown n Fgure 1 and Table 10. Among them the electrcal machnery sector whch was based on IT technology has shown hgher growth of productvty. But the labor ntensve sectors such as Apparels, Msc Manufacturng, Food, and Fabrcated metal have shown lower growth rates of total factor productvty. 24

27 Fgure 1 Sectoral Growth of Total Factor Productvty <average growth rate(%)> 1) GOTFP : TFP by gross output growth accountng 2) VATFP : TFP by value added growth accountng 25

28 Table 10 Sectoral Growth of TFP (Gross Output Growth Accountng: ) <average of log growth rate(%)> Code Industry Gross captal labor output nput nput energy materal TFP 1 Agrculture Coal mnng Metal non-metal Ol and gas Constructon Food Textle Apparels Lumber and wood Furnture Paper alled Prntng, publshng alled Chemcals Petroleum products Leather Stone, clay, glass Prmary metal Fabrcated metal Machnery Electrcal machnery Motor Transportaton equp Instrument Rubber Msc. manufacturng Transportaton Communcaton Electrc utlty Gas utlty Trade Fnance and real estate Other prvate servce Publc servce Total ) Cap tal and labor are adusted wth qualty, and weght ed by each share of ncome 26

29 On the other hand, the sector of Communcaton whch has been strongly related wth IT technology has recorded the hghest growth rate (7.96%) of TFP among servce. But the sectors such as Constructon, Publc Servce, Trade, Gas utlty, Other Prvate Servce, Transportaton and etc. have produced negatve growth rates of TFP. Therefore, we can see that the man sectors for productvty growth have been the IT-related sectors. Korean economy has heavly nvested n IT sectors on a full scale snce 1995 (Table 11) as analyzed n Ha and Pyo (2004). Table 11 The nvestment n IT sector Year IT Investment (bllon won) Growth(%) , , , , , , , , , , *Investment s real value n 2000 prces *Source: Bank of Korea ( 3.4 Productvty Trend and Captal-Output Coeffcent We have generated a seres of labor productvty; gross output per men-hour (GO/L). Durng the entre perod of , the growth rate of economy-wde labor productvty s estmated as 6.19 percent, 8.71 percent n manufacturng and 4.37 percent n servce(table 12). As expected, the growth of labor productvty n manufacturng has been greater than that of servce sector. It may reflect product and process nnovaton n manufacturng. Durng the perod of the declne of labor nput has affected the growth of labor productvty n manufacturng. The trends n labor productvty are shown n Fgure 2. We have also generated a seres of captal-output coeffcents; captal-gross output 27

30 coeffcents(k/go) as shown n (Table 13). The economy-wde captal-output coeffcent has grown at average rate of 1.53 percent. The economy-wde captal-labor rato has grown at 7.72 percent. Table 12. The Growth of Labor Productvty <average growth rate(%)> perod gross output labor nput labor productvty whole ndustry 84-' ' ' manufacturng 84-' ' ' servce 84-' ' ' * Labor nput s total workng hours Fgure 2 Labor Productvty(GO/L) <1000won/h> *GO=gross output, L=total workng hours 28

31 Table 13 Trends n Productvty and Captal-Output Coeffcents n Korea( ) Year Labor productvty Captal-output coeffcent Captal-labor rato GO/L K/GO K/L (1000 Won/h ) (1000 Won/h) Average growth rate(%) Note: 1) Labor(L) s measured n terms of total workng hours. 2) The growth rates are the average of logarthm growth rates. 29

32 Fgure 3. Captal-Output Coeffcent/Captal-Labor Rato 4. Concluson In the present paper, we have constructed a Korean database of gross output, GDP, and four nput seres durng the perod of n the framework of KLEM Model. Accordng to a gross-output growth accountng, the Korean economy has recorded a sgnfcant growth of gross output at the average rate of 8.04 percent durng the perod of On the other hand, captal(k), labor(l), energy(e), and materal nput(m) have grown at the rate of 9.57 percent, 1.85 percent, 6.68 percent, and 8.65 percent respectvely durng the perod. The average estmated shares of four nputs were 0.20( v K ), 0.20 ( v L ), 0.08 ( v E ) and 0.52 ( v M ) respectvely. As a consequence, the total factor nput and total factor productvty has ncreased at the average annual rate of 7.49 percent and 0.55 percent respectvely. Therefore, the relatve contrbuton of total factor productvty to gross output growth s estmated to be only 6.84 percent whch s of rather nsgnfcant magntude. We thnk that the qualty adustment n labor nput has played a role n estmatng lower growth rate of total factor productvty. We are reconfrmng both Krugman's(1994) proposton and emprcal fndngs by Young(1994) and Lau and Km(1994). We also have noted that there was a dscernable structural turnng pont after the 1997 economc crss n Korea: both captal-gross output coeffcent and captal-valueadded coeffcent has started to fall after

33 References Bank of Korea, Natonal Accounts, (1984, 1994, 1999, 2004) Bank of Korea, Input-Output Tables ( ) CD-ROM Bureau of Economc Analyss, Fxed Reproducble Tangble Wealth n the Unted States, , Washngton, D.C.:U.S. Government Prntng Offce, 1993 Hall and Jorgenson, Tax Polcy and Investment Behavor, Amercan Economc Revew, 57:3, June, pp , 1967 Jorgenson, D.W., F. M. Gollop and B.M.Fraumen, Productvty and US Economc Growth, Cambrdge MA: Harvard Unversty Press, 1987 Jorgenson, D.W. and Landau, Ralph, Tax reform and the Cost of Captal, The Brookngs Insttuton, Washngton, D.C., 1993 Jorgenson, D.W. and Yun K.Y., Tax Reform and the Cost of Captal, Oxford, Oxford Unversty Press, 1991 Kng, Mervyn and Don Fullerton, The Taxaton of Income from Captal, Unversty of Chcago Press, 1984 Km, Kwang Suk and Sung Duk Hong, Accountng for Rapd Economc Growth n Korea, , Korea Development Insttute, KDI Press, Seoul, Km and Park, Sources of Economc Growth n Korea, , Seoul: KDI Press, 1985 Krugman, Paul, The Myth of Asa s Mracle. Foregn affars, November/December 1994 Kuroda, M., The Internatonal Comparson of the Productvty among Pan-pacfc Countres, mmeo, 2001 Kuroda, M., and K. Nomura, Productvty Comparson and Internatonal Compettveness, Journal of Appled Input-Output Analyss, Vol. 5, 1999 Kwon, J.K, and Kyhyang Yuhn, Analyss of Factor Substtuton and Productvty Growth n Korean Manufacturng , Korean Economc Development, Greenwood Press, Westport, CT, pp , Lau, Lawrence J. and Jong-Il Km, The Sources of Growth of East Asan Newly Industralzed 31

34 Countres, Journal of Japanese and Internatonal Economes, Pyo, Hak K., Estmates of Captal Stock and Captal/Output Coeffcents by Industres: Korea, , Internatonal Economc Journal, summer 1988 Pyo, Hak K., A Synthetc Estmate of Natonal Wealth of Korea, , KDI Workng Paper No.9212, Korea Development Insttute, Seoul, 1992 Pyo. Hak K, Estmates of Fxed Reproducble Tangble Assets n the Republc of Korea, , KDI Workng Paper No.9810, Korea Development Insttute, Seoul, 1998 Pyo, Hak K., Economc Growth n Korea ( ): Along-term trend and Perspectve, n Seoul Journal of Economcs, Vol.14 No. 1, Seoul, sprng 2001 Pyo, Hak K., Estmates of Captal Stocks by Industres and Types of Assets n Korea ( ), Journal of Korean Economc Analyss, Panel for Korean Economc Analyss and Korea Insttute of Fnance, Seoul 2003 Rhee, Keun-Hee, Total Factor Productvty Analyss n Korean Manufacturng, Korea Productvty Center, 2001(In Korean) Tmmer, M.P Towards European Productvty Comparsons usng the KLEMS Approach- An Overvew of Sources and Methods,, Gronngen Growth and Development Centre and The Conference Board, 2000 Tmmer, M.P and van Ark, Captal Formaton and Productvty Growth n South Korea and Tawan: Realsng the Catch-up Potental n a World of Dmnshng Returns, Gronngen Growth and Development Centre Unversty of Gronngen, 2000 Won Yunh and Jn Kwon Hyun, Margnal Effectve Tax Rates n Korea: ", Workng Paper 00-02, Korea Insttute of Publc Fnance,2000 Young, Alwyn, Lessons from the East Asan NICs : A Contraran Vew, European Economc Revew, papers and proceedngs, 1994 Yun and Km, Margnal Effectve Tax Economcs, 1997 (n Korean) Rate of Corporate Investment n Korea, Publc 32

35 33

36 Appendx Table A-1. Gross Output n 33 Sectors Table A-2. GDP n 33 Sectors Table A-3. Captal Stocks n 33 Sectors Table A-4. Labor Input n 33 Sectors Table A-5. Energy Input n 33 Sectors Table A-6. Materal Input n 33 Sectors Table A-7. Average Shares of Captal Inputs Table A-8. Average Shares of Labor Inputs Table A-9. Average Shares of Energy Inputs Table A-10. Average Shares of Materal Inputs Table A-11. Captal Prce Index Table A-12. Captal Input Prce Table A-13. Energy Prce Index Table A-14. Materal Prce Index Table A-15. Prce Index of Commodtes from V-Table Table A-16. Prce Index of Gross Output Deflators by 33 Industres Table A-17. Growth rate of the qualty of captal nput Table A-18. Growth rate of the qualty of labor nput 34

37 Table A-1. Gross Output n 33 Sectors (Unt : mllon won) (1995 constant prce)

38 Table A-2. GDP n 33 Sectors (Unt : mllon won) (1995 constant prce) 36

39 Table A-3. Captal Input n 33 Sectors (Unt : bllon won) (1995 constant prce) 37

40 Table A-4. Labor Input n 33 Sectors (Unt : Hour) 38

41 *The sum of the labor nput of ndustres 26 and 27 durng the perod of 1984~1992 s allocated n proporton to the year Table A-5. Energy Input n 33 Sectors (Unt : mllon won) (1995 constant prce) 39

42 Table A-6. Materal Input n 33 Sectors (Unt : mllon won) (1995 constant prce) 40

43 Table A-7. Average Shares of Captal Inputs (Unt : %) 41

44 Table A-8. Average Shares of Labor Inputs (Unt : %) 42

45 Table A-9. Average Shares of Energy Inputs (Unt : %) 43

46 Table A-10. Average Shares of Materal Inputs (Unt : %) 44

47 45

48 Table A-11. Captal Prce Index (1995=100) 46

49 Table A Captal Input Prce (Buldng and Structure) 47

50 Table A Captal Input Prce (Transportaton Equpment) 48

51 Table A Captal Input Prce (Machnery) 49

52 Table A Captal Input Prce (Total Asset) 50

53 Table A-13. Energy Prce Index (1995=100) 51

54 Table A-14. Materal Prce Index (1995=100) 52

55 Table A-15. Prce Index of Commodtes from V-Table (1995=100) 53

56 Table A-16. Prce Index of Gross Output Deflators by 33 Industres (1995=100) 54

57 Table A-17. Growth rate of the qualty of captal nput (Unt : %) 55

58 Table A-18. Growth rate of the qualty of labor nput (Unt : %) 56