Upload
lemien
View
214
Download
0
Embed Size (px)
Citation preview
Journal or Scientific & Industrial Research
Vol. 62 . M arch 2003, pp 157- 167
An Empirical Analysis of the Inter-industrial Spillover Effect of Information and Communications Technology on Cost and Labor - The Case of Korea
Gwangman Park and Yongt,1e Park'
Seou l National Uni vers it y( SI'JU), Dcpariment or Industri al Engineering, San 56-i, Sh i ll lln-Dong, Kwanak-Gu, Seoul 151-742, Korea
Received: 23 August 2002; accepted: 27 December 2002
Thi s research cmploys an cmpirical data set or Korean industry and examines the spi ll over impac t o r diffusion of information and cO lllmunications (IC) technology on oth er illliust ri al secto rs in terms of labor and COS I. To thi s end. IC industry is di vided into machinery and serv ice sectors, the n()tion or R&D stock is operationally defined, and th L: input-output tahl e is used 1<' gaugL: the inter-industrial !low of technoiLl.;ica l kno wledge, Then the cost funct ion and labor pri ce function are employed to app ly a regression model. Amongst others the resLi lt 0 1' empiri cal analys is shows that IC machinery and service ex hibit quite oppos ite pattern in terms of lahor elleet in (hat IC mach inery sector renders lahor- subst itution effect. whereas service sector exerts labor-c reation efrec\' The cost efl'eet. however. is mixed among indu>trial SL:cto rs and thu s unc l e~lr to draw a conclusion, In addition, there appears a trade-olT hetween labor effect and cost cl'iCcl.
Introduction
In the knowledge-based economy and information soc iety the inter-industrial flow of tec hnolog ical knovv ledge exerts a substanti al impact on producti vity and labor demands , Since diverse indust ries are interrelated th rough a user-supplier network , tech nolog ical knowl edge generated in one industry may be employed by , in either embodied or disembodied way, another industry, and vice versa, In turn the inter-industrial diffusion of technological knowledge brings about changes in labor structure and productivity of related industri es, Amongst others, informati on and communicati ons (IC) technology accounts for the major pipeline of network in that it intermed iates the knowl edge fl ow across industries, It is also the most pervas ive technology in terms of speed of penetrati on and range of app li cation,
In fact , it is found th at IC tec hnology cluster of industries has pl ayed a major ro le among OEeD countries in th e generati on and diffusion of new technologies with increas ing importance over time" 2
* Corresponding au thor Tel: 82-2-il80-X358. e-mail : park yt C!!l eybernel.snu ,ac,kr
and the World Bank ' s experience in support of Ie projects in the developing countries also suggests that there is a vast potential for using IC [ 0 accelerate the process of economic developmenr' , Specificall y the report argues that IC is important in the devel oping countries because it enab les global trade and production , alleviates in format ion povert y, enhances competiti veness, improves public sec tor management , and promotes environmental-fri endl y deve lop ment.
Yet, one of the barri ers to impede the diffusion of IC is the lack of awareness of the potential benefi ts of IC adopt ion with respect to productivity and employ ment. It is genera ll y postulated that the dissemination of IC renders, both pos iti ve (labor-creation and/or cost- reduction) anJ negati ve (labor-substituti on and/or cost-increase) , effects on other industries , Along thi s line the sp ill over effects of IC on other industri al sectors descrve in-dept h analysis, especially among countries in As ia Pacific region that have been ac ti ve in introducing IC technology,
The attempts to gauge the tec hnology spillovers are by no means a novelty, Tracing back to the sell1in al work of Solow4 the literature on economi c effect of R&D is abundanr'- '2, By and large, these studi es
158 J SCI IND RES VOL 62 MARCH 2003
applied e ither g rowth accounting approach , production fun ction approach or cost fun ct ion approach and inquired into the R&D spillover effect on economic growth 13. However, s ince the me thods of measurement ha ve no t yet been reached a cred ible consensus, ind ividua l studies are varied in te rms of underl ying theory and data used . Furthermore the results are country-specific and/or pe riod-spec ific and therefore hard to genera li ze. A long thi s stream, some vari ati ons and ex tension s were made by atte mpting to di stingui sh R&D spi 1I 0vers on severa l di mensions: on the diffe rence between di sembodied R&D and embodi ed R&D, on the difference between manufacturing sector and service sector, and on the difference between own R&D and borrowed R&D.
Recentl y the unusuall y long ex pansion of the US economy, as we ll as uneven performance of othe r countries, ha ve attracted renewed attention to the source of growth . A lthough severa l studies over the past few yea rs have conc luded that there ex ists no s ing le facto r that ex plains the growth performance, they agreed that a new driving force of growth be IC techno logy. In summary, many empirica l stud ies indicate that the c riti ca l ro le of IC to economic pe rformance has inc reased over time and , amo ng advanced economies, the US has not been the so le country benefitin g from the pos iti ve e ffec t of IC
. I ' I~ I " capita In vestment . ' .
However, it should be noted that a substantia l di ve rgence sti II ex ists across countri es in te rms of re lati ve pe rformance and the impact of IC plays in a different contex t among nationa l systems. T he ev idence is mi xed and the view is indec isive in that the growth pe rfo rmance of severa l countries has been s luggish in spite of intensive in vestment in Ie. This phenomenon is espec ia ll y true among the deve loping economi es3
. A nothe r theme of inqu iry is about the re lati ve importance between IC-produc ing sector and IC-usi ng sec to r. Some studies attributed the improve ment in productivity to IC-prod uc ing sector, whe reas othe rs argued the effici ency ga in s f rom IC-us i ng, particularl y servi ce sector, contri buted to
growt h.
A lth ough prev ious research is large in vo lume, several important Issues remain unex pl ored o r necess itate furthe r anal ys is . T hi s research attempts 10
address these resea rch needs by inves ti gating the case of Korea. Specifically , compared wi th past studi es the
mai n objective and contribution of this research is three-fold. First , thi s research is more ~; pec ifi c in that it defines spill over effect as producti vity improve ment and empl oyment , rather than as overall g rowth performance. Secondl y the current researc h exa mines the inte r-industri a l effect, rather than the mac ro-economic effec t. Thirdly, this research divides IC indu stry into two ma in sub-sec tors, IC-machine ry and IC-service, and contrasts the differe nce in te rms of diffusion effec t.
Thi s pape r compri ses fo ll ow ing contents. First , we propose a map of sp ill over effec t tha t show how both positive impact and negat ive im p<lc t are spec ifi ed in te rms of productivity and employment. T he map serves as a building block for ensuing anal ys is. T hen, we ex plain about the deta il ed research methodology w ith respec t to operati ona l definiti on o f techno logica l knowledge, measurement of inte r-i ndus trial kno \-vledge f low, and fo rmul at ion and manipulation of cos t function model. Next , we dwe ll on th result s of empirica l anal ys is . He re, indi vidua l in d ustries a re located on the map depending on the type o f spill over effec t and the di ffe rence between IC -mac hinery and IC-service a re compared. Finall y, some implications o f current research and prospecti ve i ~s l1 es of future research are presented.
Map of Spillover Effect
The spectrum of lC indu stry is so w ide and each techno logy area may be he terogeneous in te rms o r both techni cal and econo mic characte ri sti cs. Hence, as menti oned before, we divide IC industry in to IC-machine ry sec tor and IC-serv ice sector, assuming that these two su b-secto rs are cons ide rabl y diffe re nt to each o ther in te rms of knowl edge contents and flow pattern. In order to measure the inte r-industri a l spi 1I 0ver effect, other indust ri es are c lass i fi ed into 17 diffe rent sectors, based on the Korea Standard Industry C lass ificati on ( KS IC) of 1990, as li sted in Table I.
Then , we propose a map of spi ll over effect that consis ts of nine ce ll s. As shown in F igure I the map comprises two main dimens ions, cost and labo r, and each di mens ion is composed of three leve ls, pos iti ve, neutral, and negati ve. Then, the re appear four maj or areas of interes t, labo r-c reation area, labor-substituti on a rea, cost-reduc ti on area , and cost-inc rease area . The purpose of the map is to locate respec ti ve industries 0 11 the corres ponding areas. For in stance. in dustry A is
PARK & PARK: INTER-INDUSTRI AL EFFECT OF ICT ON COST & LABOR 159
Table I - Li st of nineteen industry sectors
Code Industry Code Industry Code Industry
IT machinery 7 Chemi cals 13 Transport ati on eq ui pment
2 IT servicc 8 No n-ferrous 14 Apparatus
:1 Min ing 9 Iro n/steel 15 Other manu facturing
4 Food 10 Mctals 16 Electri c/gas
5 Tex til e/apparcl II Machinery 17 Construction
6 Wood 12 Electronics 18 Trade
19 Transportation/warehousing
Labor substituti on area
Increase in cost
No cost e ffec t
Decrease
Cost (+) Labor (-)
in cos t Cost (-) Labor (-)
Dec rease in labor demand
Cost (+)
Neither labor demand nor cost
e ffect area
Cost (-)
No labor demand effect
Cost (+)
Lab?!J +)
Labor (+)
............. \ . ...... ...
................ > ..... Cost (-)
Labor (+)
Increase in labor demand
Labor creati on area
Cost reducti on area
Figure I - Map of spillover effec t
located on labor-creati on and cost-reduction cell , it means th at the di ffu sion of IC technol ogy exerts positive impact on industry A . However, if industry B is located on labor-creati on and cos t-increase cell , it indicates th at the diffusion of IC techno logy makes pos iti ve impact in terms o f labor but negati ve impact in terms of cos t. By doing so, we may be ab le to identify spec ific effec ts of IC technology on individual industries.
Methodology of Empirical Analysis
Operational Definition of Technological Kllowledge: R&D Stock
It has long been considered an intractabl e to task to gauge the amount of technological knowledge and , as a remedy, many proxy measures have been used. A lthough vo luminous previ ous researc h has empl oyed vari ous proxy measures, it might be imposs ible to
160 .I SCI INO RES VOL 62 MARCH 200]
select the unique best proxy indicator, since respect ive indicators have their own advantages and di sadvantages. Among others the fo ll owing three measures. R&D human resources, patents, and R&D stock. are frequentl y adopted as proxy indicators of tec hnological knowledge.
R&D HIIIIWIl Reso llrces
In the input side the amount of R&D human resources is ofte n used as a proxy in dicator for the capac it y of kn owledge base. The rati ona le for using R&D human resources is based on the assumpti on th at the more human resources a finn or a nati on possesses, the more techn ological capability it maintains. It i5; also assu med tha t the amount of R&D investment (ex pendi ture) is proporti onal to the nUlllber of R&D personnel. To illu st rat e, Leoncini et a/. I
() have adopted R&D person nel of eac h industry as a proxy measu re for in novati on capacity. Park and Kiml 7 have also used R&D human resources in measuring the fl ow of tec hnological kn owledge ac ross industries. Thi s indicator, although simp le and useful , has limitat ion and shortco mings in th at technologica l inllova ti on dra ws on ideas from various sources other than forma l R&D human resources .
P(l{ c lllS
In the output side the amount of patents is the most frequ ent measure adopted in survey practi ce and empirica l research. The rati onal e is due to the hypothes is th at technolog ica l advances are best conceived in terms of re levant events or inventi ons, whi ch, in turn , are registered as patents. Patent s meet exp lic it crite ria of originalit y, technica l feasibility, and commercial wonh lx. Patents also have ad vantages in terms of availabilit y of database, scope of cove rage, and variety of informati on. Thus, patent s have long been considered ev idence of innovati ve perforillance and generally preferred to other output indica tors . In addi ti on, patents ha ve been used to es tilllate the
1 1 fl 0 I . . d . . I ') 10 tee lno ogy . ows ana t le lr Impact on pro~ uctl vlty . - .
Patent s, howeve r, are also subject to critica l drawback since Illany in venti ons and tec hnologica l knowledge are not patented and patents, if registered, may diller from one another wi th respect to qualitative signifi cance. Moreover the index of patented inventi ons is merely a list of blueprint s ava ilable that gives littl e information on innovati ve va lue and ~o ll1ll1e rc i al use}l. Patent s should therefore :1t best be
considered a partial measure or tec hnolog ical knowledge, applicable onl y in th ose indus trial sectors where patenting is common prac tice22
.
R&D Stock
In the throughput side, R&D ~ toc k refl ec ts the cumul ati ve amount of techn ologica l knowledge that a firm or an industry possesses at a certa in point in time. The cumulati ve stock is obtained throuoh R&D
b
in ves tment , technology import , and other inputs. Several pas t studies empl oyed the no tion of R&D stock, as one of the inputs of th e producti on function , and gauged the cumulat ive amount in an attempt to est imate the rate of return to R&D in ves tment ~.23 In additi on, R&D stock may indicate the future potential to develop new products or processes '). In thi s research , R&D stock is empl oyed as proxy measure of tec hnological knowledge, since it is considered a more comprehensive measure, vis-a-vi s R& D personnel or patents . In general, R&D stock is defin ed as follow s:
TS, = TF, + ( I - 8)TS'_I'
1/
TF, = I~L , RD' _i ' ; :::: 1
where,
TS, : R&D stock of year I , TF, : Suppl y of a
new technologica l kn owledge or year t,
RD, : R&D in vestment of year t, Pi : Time lag
distribution,
8 : Deprec iation rate of R&D stock.
Despite it s re lati Ve ad vantages, R&D stock is more co mpl ex and thus is more difficult to quantify' . h may be noted that R&D ex pend iture i ~ subject to pri ce chan ge, R&D stock beco mes obsolete over time, and R&D output takes time until comillerc ia li zat ion. Therefore, as shown in the above deflllition, such additional parameters as R&D deflator, deprec iati on rate of R&D stock , growth rate of R&D investment , and time lag need to be estimated. Since the estimati on of these parameters is beyond the scope of current research, we consu lt with previous researc h. For R&D deflator and depreciation rate, we basica lly follow the
PARK & PARK: INTER- INDUSTRIAL EFFECT OF ICT ON COST & LABOR 16 1
guideline of OECD2-1. For time lag the estimates of
Korea Industri al Techn ology Assoc iati on ( 1984- 199 1) are app li ed25
Inter-indusTrial Technological Kn owledge Flow: 110 Tohle
Once R&D stock is adopted as the proxy measure of knowledge amou nt the nex t task is to aau ae the inter-industri al flow of tec hn ologica l b b
knowledge. In the current research the input/output (I/O) table26 is used to defin e the inter-industri al fl ow pattern and to draw the knowledge fl ow matri x. Since the embodied knowledge is di sseminated through the pu rchase of machinery and components , some industri es act as suppli ers or se ll ers of intermediate and capital goods , whereas other industries act as users or buyers, both constituting a tec hn ology diffusion network . The knowledge fl ows across i nciu stri es are measured, based on the coeffi cients of the Leonti ef inverse. Thi s approach has been adopted in several earlier stu dies I. 2. 5. 10.27 . The coefficients embrace both di rect flo w (from industry i to industry j) and indirect fiow (from industry i to industry .i via industry k) .
Thus, a transformat ion matrix T is constructed where X means diagonal matrix of industri al output ,
[/ - A"rl denotes domestic Leontief in verse and D
indicates di agonal matrix of final demand :
T = X - I[/ -A" r I D.
Now, by combining technologica l knowledge amount (R&D stock) and knowledge flow (VO table), the inter-industrial know ledge fl ow matrix is const ructed. The knowledge fl ow matri x F is expressed as foll ows:
F = TS . B = [I i ] ,
where f' : amount of interna l knowledge stoc k .1/
of indu stry i , and
t·· : amoun t of knowledge flow from indu siry i . '}
to industr j.
Note that the above matri x takes both disembodied and embodied knowl edge into account. The R&D stock may represent the embodied knowledge channel whereby tec hn ologica l knowledge spread~ through research activity while I/O table may
reflect the embodi ed channel whereby knowledge is di sseminated through the purchase of machinery. equipment, and components across industri es.
Measurement of Effect: Cost Funct ioll (lild Labor Price Function
By compu ting the above inter-industri al knowledge flow matri x, we can gauge the spillover amount of Ie technology to other industrial sectors. Then the next step is to measure the spill over effect of IC technology on producti on cost and labor demand of other industrial sectors . To thi s end , we empl oy the fo ll owing (trallslog) cost function model. Note that the cost fun cti on modelR is adopted, i n ~ t ead of' the production fu nction model, since current research defines change in productivity as change in cost, not as change in output amount.
lilC = ~O + ~ W ·I n w + ~ " .I n c+~ )' · ]ny +
~ ! . In R + ~ we . In w . In c + ~ wy . In w . In y
+ p cy . Inc· In y + ~ WI . In w . In R +
~ LT ·I n c- In R + ~ ) I . In y . in R
+(InR + ~ WS · ln w ) · ( ~ Ill· lnM + ~ , · lnS)+
~ es . In c· ( ~ !Il . In M + ~ s . In S)
subj ect to f3 11' + f3 , = I , f3 11 '" = 0 , [31\'\ + f3 = 0 " '
. " ( I )
where, C : producti on cost, II ' : labor pri ce, c : capital price, v : output,
R : techn ologica l kn ow ledge stock of each industry,
M : technologica l know ledge stock diffused by IC machinery, and
S : technologica l knowledge stoc k d iffllsed by IC service .
The constrai nts in Eq. ( I) could be de ri ved from the homogeneoll s property of each fac tor price. Fu rthermore, usi ng Shephard' s lel11ma. we cou Id derive the followi ng share or the labo r in tota l co~; t by partial differenti ati on of the above cost fun ction by labor price:
162 .I SCIINO RES VOL 62 MARCH 2003
L = ~ w + ~ wy . In y + ~ W I • In R +
~ ws · (~ 1l 1 · lnM + ~s · lnS). .. . (2)
Then the coefficients of the two functions , Eqs ( I ) and (2), are es timated by appl ying the seeming ly unre lated regression2x
. As the final step the spi ll over e ffects are computed by measuring the subsequent Eqs (:' -6), and each indi cates the e las ticity:
Cost e las ti c ity due to IC machinery:
d ln(C/c)=R . [lnR +R . . In(w / c) ] . () In M I-' m I-' ws
. . . (3 )
Labor demand e last ic ity due to IC machine ry :
aL a In M = ~m . ~ ws ' ... (4)
Cost e las ti c ity due to IC service:
a IneC / c) = R . [In R + ~ , . In (w / c)]. d inS 1-' , \\ s
... (5)
Labor demand e last ic ity due to IC serv ice:
dL a InS = ~ s . ~ IV S •
. .. (6)
Note that a negative va lue of cos t e lastic ity means cost-reduction effect, whereas a pos iti ve va lue indicates cost-increase e ffect. On the contrary, a positive va lue of labor e lastic ity represents labor-creati on effect, while a nega ti ve va lue implies labor-substituti on effect.
Dala
The set of empirical data covers IC and ot her manufac turing sectors of Korea during 1977-1 997. Specifically, R&D ex penditure data are from the annuals of Report on the Survey of R&D in Science and Tec hnology (MOST/C) and input-ou tput data are due to the Input-Output Table·
10. The reference pe ri od
is o f parti cular impo rtance in case of Korea in that IC technology began to be adopted since the late 1970s and actively diffused across industries up until 1990s.
Results of Analysis
Ie Machill elY
Regarding the economi c impac t of IT machinery the estimated coeffi c ients are summari zed in Tabl e 2 and the map of sp ill over effect is shown in Figure 2.
Among others, a noteworthy find ing is that the diffu sion of IC machinery has caused decrease in labor de mand. For a lmost a ll industri a l sectors the labor demand e lasti city turn s out to be negat ive, the reby indicating the ex istence of labor-subst ituti on e ffect. This fact sugges ts that. irrespect ive of indus tri es, firms
Tah le 2 - Elasticit y of IC machinery
Industry (Code) Labor dem,lIld Cost elasticity Industry (Code) Labor demand Co;. t elas ti city
Mining 0) -0.0373 (0. 11 95) Machinery ( I i) (-0.0048) (00 125 )
Food (4) -0.06_2 -0.0774 Electronics ( 12) -0.1 86 1 -(JOI5Sl
Tex tile/Apparel (5) (00086) (0.0247) Transport ( 13) -0.0381 (() IS74 )
Wood (6) -004Sl2 (0 1007) Apparat ll s ( 14) (-0 .0335) ! -(J.G3 1 0)
Chemicals (7) 0.0632 0.1 586 Other manuf. ( 15) (-00076) (00 156)
on - ferl"llll s (X) (-0.0033 ) (-00362) Elec./Gas (16) ( -0.(227) 1-0.2 176)
I ro n/Stee l (Sl) (-00032) (0.0244) Construct ion ( 17) -0. 1753 0.03 18
M ew!s ( 10) (00044) (0 0272) Trade (18) -0.0 167* 0. 1327 *
Trans/Storage ( 19) (0.0573 ) (-0. 1 (iSlO)
* significance level o f 0. 1 : () : insign i fi cant
PARK & PARK: INTER-INDUSTRIAL EFFECT OF ICT 0 COST & LABOR 163
Increase in cost
Decrease in cost
@ -O.17.'i3
o.m 18
@ -O.O I67*
O. IY27 *
CD -00:173
CD -0.O4l)2
@ -(Uml -00622
0 -00774
@ -O.186 1 12
-0.OI.'i9
Decrease in labor demand
no : Industry code
0 0.0632
0.1.'i 86
Increase in labor demand
3. Mining
4. Food S. Tex tile/apparel 6. Wood 7. Chemicals 8. Non-ferrous 9. [ron/stee l 10. Metals II . Machinery 12. E lectroni cs 13. Transport 14. Apparatus IS. Other manuf. 16. E lec./gas 17. Constructi on
a : Labor demand effect caused by the knowl edge of IT mac hinery b : Cost e lasti c it y caused by the knowledge of IT machinery
The va lues of a & b are signifi cant at sig. leve l of 0.05 . But, * is signifi cant at sig. leve l of 0. 1.
Figure 2 - Map ur spill over elTccl due 10 IT mac hi nery
have install ed IC machinery to replace and/or reduce human resources and , overall , reduction in non-IC labor has been greater th an increase in ]C machinery operati on labor. ]n that regard , the notion of sk ill -biased techno logical change is hi ghli ghted' i. The notion postulates that the diffu sion of IC tec hnology tends to rep lace low-ski II , routine jobs but may create high-skill , complex jobs. The empiri ca l resu lt implies that the diffusion of ]C machinery may bring about biased chan ge toward Jaw-sk ill human resources. The only excepti~n with statis tica l significance is chemi ca l industry. This is probab ly because chemi ca l indust ry is a process-ori ented industry that is cap ital-intensive,
rather than labor-intensive, with littl e room for replacement of manual labor but with some roo l11 for employment of IC machinery operation labor.
However the directi on of spillover effec t is indec isive in terms or east e lasti c ity. Ou t of 17 industries , II industries ex hibit pos iti ve value, whereas six industries show negati ve va lue. Thu s, although there ex ist so me variati ons, a maj ority of industri es have experi enced increase in cost due to adoption of IC machinery. Howeve r the elasti c ity coeffi cienis are not significant in man y industri es. Evidentl y the purchase of Ie machinery req uires substanti al amollnt of initial in vestment. whil e
164 J SCI INO RES VOL 62 MARCH 2003
cost-.-;aving effect takes some time to occur. Therefore, it may be too ea rl y in time to draw a conclusi on on whether IC machinery exerts negati ve impact on cost or not.
Another fact deservi ng attention is that there appears no industry that b nefits both pos iti ve effects, namely increase in labor demand and decrease in cost. On the contrary, there fou nd some indu5.t ri es in those ce ll s where e ither cost-reducti on but labor-dec rease or cost- increase but labor-increase . Thi s finding impli es thaL in the case of IC machinery, there lIlay ex ist trade-off between cost effect and labor e ffect.
Tel eCOl11l71UIl i ca rioll s S(' r vice
Rega rding the economic impact of IT service the es timated coeffi cients are li sled in Table 3 and the map of spillover effect is shown in Figure 3.
It is qu it e signi fica nt to know th at the diffu sion of Ie service exerts pos iti ve impac t Oil labor demand. For almost all industrial sectors the labor demand elasticity turns out to be positive, ex hibiting the labor-c reat ion effect. Thi s result is quite oppos ite to thar of IC machinery. However the phenomenon may not be surprr sll1g. si nce the install at ioll and man agement. of IC servic' necess itates new labor fo rce . Furthermore, firms may adopt IC service not to rep lace human resources but to enh,ince the t'ft'i cicncy of management system. Conseq uentl y, net Increase in Ie service labor force, without subst itutin g non-IC labor force, may result in positive impact on empl oyment. Once again the not ion of sk ill-biased tec hn ologica l change is fo und in Ie service where the diffus ion of IC service may cause biased change toward hi gh-skill expert s.
However, like the case of IC Illach i nery th e direct ion of spillover effect is mi xed in terms of CO:-.I e lasti ci ty. Out of 17 industries , ten indll s(ri es ex hibit negati ve va lue, whereas seven in d ~I s tri es show pos itive value. Thus, on the contrary to IC machinery, it seems that a maj ority of industries have ex pe ri e n c~cI decrease in cost due to udoption of IC service. However, if we note as to the number of industries with stati stica l significance, more indust ries show increase in cos t. rather than decrease. Thus, like the case of IC machinery, a conclusive judgment is reserved for the cost impact of IC servic~.
Interestin gly, there appea r more industri es, v is -~I -v i s IC machinery, that benefit both pos iti ve effects, increase in labe l' demand alld decrease in cost. This finding may suggest that. although nOI ev ident Ihe potential benefit of IC se rvice is greuter than that of IC mac hinery .
Table 4 contrasts the ~;p i Il over effect between IC machinery and IC service.
Conclusive Remarks
The study attempted to meas ure the inter-industri al spillover effect of IC technology. Spec ifi ca ll y, auth ors divided IC industry into two main sub-sec tors, IC-machinery and IC-service, and con trast the diffe rence in terms of diffu :-. ion effecr. The diffusi011 effect was gauged with respect ro two factors , cost factor and labor factor.
The most signifi cant findin g is that the economi c impact of techn ology diffusi on ex hibil s considerab ly different, even opposite, pattern between iC machinery and IC service. The diffu .·ion of IC mac hinery to other industrial sectors tends to resulr in
Tahk 3 - E l a~t i c it y or IT service
Inuuslry (Couc) Llbor uemanu Cost elast icity Industry (Code) Labor demand Cost elasticil y
Mining (3) O.04X4 (-0. 1285) Machinery ( I I ) 0.0539 (-02572)
Food ( .. 1) 0.08! 8 -0.03 12 Electron ics ( 12) o 1362 O.OX97
Tcxlil e/atJparcl (5 ) 0.0420 -00302 Transport ( 13) 0.0212 n.0305
Wood (6) (0.0262) 0.0 1 S6 " AppJratlis ( (4) 0. 1041 (-0 1617)
Chemicals (7) (0.02 14) (· 0074 1 ) Othermanlli'. ( I S) -0.0 127 n. ll07
lon-ferrou s (X) D.n I 05 0.0124 * Elec./gas ( 16) 0.02 16 * O. I WlX "
I ron/steel (9) 0.0 188 0.0240 Constrllc lion ( 17) 0.1819 -0. 1230
Metals ( 10) (0.0 153) (-00242) Trade ( I X) 0.0037 (-0. 1546)
Trans./siorage ( 19 ) 0.3592 (-OX627)
" sign i ficance leve l of 0. 1 : ( ) : insignificant
.... ..
PARK & PARK : INTER-INDUSTR IAL EFFECT OF ICT ON COST & LABOR 165
Increase in cost
Decrease in cost
@ -0.0 127
0. 11 07
Decrease in labor demand
0 0.0 156*
no : lndustry code
CD 0.0 105 0 O.0 ' HX
0.0 124* (UJ240
@ O I:l62 @ O.02 12 12
0.m{lJ7 0.0305
@ O'(J2 16*
16 0 183X*
0 O.04H4 @ OO53lJ
@ O. IO-11 @ O.fJO:l7
@ 03592
0 O·0X ' X
-00:1 12 CD 0.0-120
-0.0:102
@ 0 18 1lJ 17
-0. 12:10
Tncrease in labor demand
3. Mining
4. Food 5. Tex ti Ie/appare l 6. Wood 7. Chemi ca ls 8. Non-ferrous 9. Iron/s tee l 10. Metals I I. Machinery 12. Electroni cs 13. Transport 14 . Apparatus IS . Other manu f. 16. Elcc./gas 17. Const ructi on
a : Labor demand effect caused by the kn ow ledge of IT service b : Cost elastic ity caused by the knowledge of IT service
The va lues of a & b are significant at sig. level of 0.05. But, * is significant at sig. leve l of 0. 1.
Figure 3 - Map of spillover effect due to IT service
labor-substituti on effect since firms may install IC machinery to replace !ow-sk ill , routine jobs, whereas IC service brings about labor-creati on effect, as firms empl oy hi gh-s kill ex perts for service operation . The ,~ec tora l characteri stic.12 has to be taken into account in developing IC-related technology and industry policy. However the spi ll over effect is mixed in terms of cost elasti city. The adopti on of IC machinery and/or
service requires initial in ves tment but cost-sav ing effect is yet uncl ear across industrial sectors . Therefore, it may be too earl y to reach a conclusive judgment.
Although this paper provides va luable insights, it is at the same time subj ect to a couple of limitat ions. First the result of analys is may be country-spec ific. The empirical database is restricted to the case of
166 J SCI IND RES VOL 62 MARCH 20m
Table 4 - Compari son or spillover erfeets between IC machinery and I C servi ce
Labor demand Cost elasti city
Code Industl"y (Code) IC IC IC Ie
machinery service machinery service
3 Min ing +
4 Food
5 Tex til e/apparel
6 Wood +
7 Chemicals ++ ++++
X Non- rerrous + +
9 I roll/s leel + +
10 Metal,
II Machinery ++
12 Electroni cs +++ ++
13 Transport,Hion equipment
+ +
14 Apparatus +++
15 Other ll1anuracturing
+++
16 Eleclri c/gas + ++++
17 Conslruclion ++++ +
IX Trade + +++
19 Transportation/ warelml!s i ng
+++++
Notes: I - : -0.0- -0.05 , - - : -0.05--0. 10. - - - : -0. 10--0.15,
----: -O.IS--O.20, 2 + : O.O- U.05. ++: 0.05 - 0. 10. +++ : n.1 0-0. 1 5, ++++: O. I S-n.20, +++++ : 0.20-
Korea and thus the result must be interpreted so. If the analys is is extended to an internat ional compari son among countries in As ia Pacific region, different res ults may be produced. Secondl y the proxy measures empl oyed may reflect other fac tors than tec hnological knowledge now. If other measures and/or functional models are applied the results may be different to some ex ten t. These limitations, however, would not hamper the val idit y and utility of the research, si nce the main purpose of thi s study is not to measure the exact degree of effect , magnitude of elast ic ity, but to examine the direction of e ffect , and sign of e las tic ity. It is our belief thal changes in databa'ie, proxy measure, or functi onal model may ca use changes in magnitude of elastic ity but not in sign of e la, ticity. Nevertheless. we suggest that more extensive research be conducted in the future
by extending and changing database and analyti ca l mode ls.
Acknowledgement
This research was partially supported by Nati onal Research Lab program of Mini stry of Science and Technology of Korea.
References
Sak urai N, loannidi s R & Papaconslantilh'lI G. The impaci or R&D and technology difrusion on producl;vity growth: evidence ror 10 OECD counlri es in the j 970s and 19RO,. STI Working Pal'er ( 1996) ,
2 Bernstein J. The st ruciure or Canadian inter-i ndustry R&D spi ll overs. and Ihe rate of return 10 R&D . .I fI/{! EUiI/ . 37 (March. 1989) 3 15.
3 Hanna N, Guy K & A rnold E. The difrusion or informaiion technology: experience or industrial counl ri es and less()n~ 1'01'
develuping countries. World 8allk DisCIISS Pall. 2S I ( 1995).
4 So low R. Technica l change ane! the aggregale production runcl ion, Rei ' /::('011 Sial , (August , 1(57) 3 12.
5 Terl eckyj . E.rfeci of R&D Oil Ill e !lmriIiClil'il1' groll 'llI oj' illdllslries: An ex"loralorr .1'1 lid\' (Nation:tl Planning Association, Washington DC) 1974.
6 Jurgenson D. Accoullling 1'01' capiwl. in Callila/' e./ficll·IIC1 · and growlII. ed ited by G von Furstenberg (I3al l in,t!er, Cambridge) 1980.
7 Scherer F, Using l inked patelll and R&D dal:t 10 measure interindustry technology !lows, in R&D 1III/elliS alld
"rodllC/il'ill ', ediled hy Z Gri li ches (Ch icago Uni versit y Press. Ch icago) 1984.
R Rernstein J & Nadri M, Research alld developmenl ,llle! intrainduslry spillovers: an empi ri cal app lic:ltioll of dynamic (Iuali ty. Rei ' Ecoll SIOI. 71(2) ( 19)1<) ) 249.
9 Goto A , & Suzuki K. R&D capilal , ral e or rClUrn on R&D illvestlllen i and sp il lovel' or R&D in Japancse m:llll1racturing. Rev Emil Sial, 71 (4 ) ( 19X9) 555.
10 M ohnen P. New technologies ane! inler-indll slry spiliovc rs. STI Rev, 7 ( 1989).
II Grili ches Z. Producli vily. R&D and the data l·onstrailli. Alii ....l /::('0 11 Rei ', X4 ( 1992) I .
12 Nadiri M. Innovalioll s and Icch!loiogical spillo vers. NIIU? Workill g Paper. 4423 ( 1993).
13 Boskin M & Lau L , Capilal . lechnol( 'gy. ,lIld cconomic growth. in Tcclllloiog\' alld Ihl' 11'{'(dlll (~( I/{:Iioll .\. ediled by N Rosenherg 1'1 al .. (S lallrOre! Univcrsi ty Pre,s. Siallrord ) I CJ<J2 .
14 Papaconslanlillou G. Saku ri:i N & Wy(' ~()rr A. Embodied tcchnology di l'fusion: an empirical anal VS IS for 10 OECD countri cs, aECD Wo rf.. ill g Palwr I V (X) ( 19%).
15 Papaconstanlinou . G. Sakurai . N & Wyckoff A. Domestic illld i lllcrnational producl- cmbodied R&D dl rru sion. Res Po!. 27 ( 1998) 30 1.
16 Leoncini R. Maggioni M & Monlrcssor S. Illlersecl mi,l l innovalion !lows and national lechnologi c.ll syslem network
PARK & PARK: INTER- INDUSTRIAL EFFECT OF ICT ON COST & L ABOR 167
analysis for comparing Italy alld Germany, Res Po l . 25 ( 1996) 4 !5 .
17 Park Y & Kim M , A taxonomy of industri es hased on knowledge now stru cture, Techl/ol AI/a l SImI M al/age, 11(4)
( 1999) 54 1.
18 Ku znets S. Innovati ve ac ti vit y: problems o f definition and measurement. in Th e Rale al/d direcliol/ o/' iI/ ve l/li ve
aCl i vin', edi ted by R Neison (Princeton Universit y Press, J) 1962.
19 Scherer F, Inter-industry technology fl ows in the United States. Res Pol , II ( 1982) 227.
20 Evenson R & Puttnam J. Th e Yale- Cal/oda POICI/I floll'
c(!II cordol/ ce (Economic Growth Center, Yale Uni vers it y) 1988.
2 1 Saha l D. Pe7llel'lls o/' techI/o log ical inl/ovalirm
(Addi son-Wes ley. London ) 1981.
22 Tijssen R. Glohal and domesti c utili za ti on of industri al relevan t science: palent ci tat ion anal ys is of science-technology interacti ons and knowledge !lows, Res
Pol. 3() (200 I ) 35.
Gri li ches Z. R& D and the producti vit y siowdown, Alii £COI/ ReI'. 7()(2) ( 1980) 434.
24 OECD, Th e lIl eaSllrelllel/1 o/,sci el/ l if ic ol/r/l f'Chl/ical (fClil'ilr:
F rascal i lI1a llllo l (Pari s) 198 I .
25 Korea Industri al Technology Associati on (KITA). While
paper 0 1/ il/dllslrialtechl/ology (Seoul , Korea) AllIlual s.
26 Leon tief W, Th e slmC/ilre o/, A lllericCIII ('co l/ (l/I/r / 919- 1939 (Oxford University Press. NJ ) 195 I.
27 Davis A, Techllology i l/ l !:l/silV of us Olllfill1 {(I/d lrad,' (US Department o f Commerce, Internati ona l Trade Admini strati on) 1982.
28 Zellner A , An efll eient mcthod of estimating seeming ly unrr:: lated regression of aggrega ti on bias. J Alii S[((I Assoc. 57 ( 1962) 500.
29 Ministry of Sc ience and Technology (MOST). Reporl 0 1/ Ihe
sltrvey {If R& D ill sci !!l1ce (md lechl/o logy (Seoul , Korea) Annu als.
30 Korea Bank. /l/fllIl -Olllfilil lah/e (Seou l. K orc~l ) 1983. 1987. 1990, 1993. 1995.
3 1 Hill L & Brynjolfsson E, informati on technology alld in terna l firm organi zati on: an exploratory analysis. J Mal/age II/ /, Srst. 14 ( 1996) 8 I.
32 Pav itt K. Sectoral pallerns of technical change: towards a taxanomy and a theory, Res Po l, 13 ( 1984) 343.