Analysis on Technology Implementation and Economic Growth
2023-07-20
来源:年旅网
China-USA Business Review,ISSN 1537・15 14 August 201 1,Vo1.10,No.8,600・609 Analysis on Technology Implementation and Economic Growth Shoji Katagiri,Hugang Han Prefectural University of Hiroshima,Hiroshima,Japan Our model is fundamentally based on Coming and HobOn’s research in 2007.In this paper,instead oftheir concave shape oflearning curve oftechnology implementation,we introduce one ofthe plausible learning curves,the S-shape of learning curve,to investigate the influence of shape of learning curve on the economic growth.The results of the steady state analysis are:Firstly,the higher the values of parameters for level of implementation and costs of implementation and R&D are,the lower the growth rate of potential productivity is;Secondly,the higher the values of parameters for implementation and R&D costs are,the higher the initial level of implementation is.However, under the same conditions,when the value of parameter for implementation level rises,the behaviour of its level is ambiguous;Thirdly,when parameters of implementation level in this paper and the learning rate of growth are suficifently large,the growth rate of productivity of Coming and Hob ̄n’s research in 2007 is larger than the one in this paper.As for the suggestion from this paper,the shape of learning curve of implementation affects the growth rate of productiviy,and eventtually the economic growth rate,therefore we have to deliberately determine the shape of learning curve of implementation in line with empirical results. Keywords:economic growth,technology implementation,inequaliy tIntrOductiOn It is said that worldwide income inequality has widened with progress of ICT(Pilat&W61i,2003;Piflat, 2004;Katagiri,2008 and so on).As to the income difference from global point of view,Anderson(2009) indicates that the income difference widens between countries due to the range of non—tradable goods isolating from the external competition.Acemoglu and Zilobotti(200 1)recognize that many technologies using in LDC have developed in OECD countries and these technologies are designed to optimize the utilization of labor in the industrialized countries.Therefore,the difference of skill supply leads to mismatch between the requirement of technology and skillin LDC.Even when every country can be accessible to the latest technology,this technology mismatch causes the signiifcant differences of total factor productiviy tnd aproduction per capita,that is to say, the source of income inequaliy.Litke Acemoglu and Zilibotti(200 1)j Caselli(2005)indicate that the diference between high.income nd alow—income countries stems from the difference of total factor productivity. Then,from where the difference of total factor productiviy tcomes?Comin and Hobijn(2007)incorporates the eficiency of fproductivity in production of intermediate goods into adoption of the initial implementation level for new technology.It follows that on the equilibrium path,the higher the growth rate of production IS,the 1ower he tinitial implementation level of new technology which firm adopts is,and the growth rate of productiviy tShqii Katagiri,Ph.D.of Economics,Faculy tof Management and Information System,Prefectural University of Hiroshima. Hugang Han.Ph.D.of Electric Engineering,Faculty of Management and Information System,Prefectural Universiy of tHiroshima. Correspondence concerning this article should be addressed to shoji Katagiri.E-mail:katagiri@pu。hiroshima.acjp. ANALYSIS oN TECHNOLOGY IMPLEMENTAT10N AND EC0NoMIC GR0WTH 601 enhances income.In their research,it is assumed that the marginal implementation of learning pattern is decreasing,and the productivity per unit at initial time is high.This implementation level affects total factor productivity,and then its origin is embodied into patent,and new intermediate good can be produced eficientlfy by intermediate goods firm obtaining patents(idea)being created by the fimsr in R&D sector.Although theoretically,Alvarez,Buera and Lucas(2008)and Lucas(2008)refer to the process(how to create ideas)in R&D sector,it is deemed that there are few studies like Comin and Hob ̄n(2007).Furthermore,at present when ICT is closely connected with technology,the quantiy of tproduction(income)is determined by an extent of initial implementation level firm adopts for technology.Then,from the macroeconomic point of view,GDP is determined,and depending on utilization of technology,ultimately the income inequality occurs. From these above,it is thought that the degree of the aforementioned mismatch determines the income difference between the industrialized and industrializing countries,therefore in this paper,we focus on the implementation level for technology.Concretely,in lieu of the technology implementation pattern of Comi and Hobij in(2007),we introduce S-shape of implementation pattern where at the beginning marginal implementation slowly increases,at the middle sharply increases and then gradually decreases,and analyze the growth rate of productiviy atnd optimal initial implementation level and so on. The plan ofthe paper is as follows.The next section describes the model and explores technology adoption. Section 3 defines an equilibrium and steady states and carries on comparative analysis.Section 4 concludes. The Model This model is fundamentally based on Comin and Hobijn’s(2007).Therefore,we succinctly mention the model except the different points from theirs .The economy comprises the following agents;household and ifrms(final good fim,irntermediate good firm and R&D ifm),the populrtaion is normalized to 1.We exclude international transaction. Household Sector The representative household in our model economy is endowed with one unit of time that it inelasticically supplies at each instant.It also owns the capital stock that it rents to firms at a net rate rt.The household selects the path of consumption ct to maximize the present discounted value of the utility flow: 一P s ’cs"ds (1) subject to the standard lfow budget constraint. The resulting optimal consumption path is characterized by the Euler equation. =a(rt一盯) Final Goods Sector (2) Let yt denote the inalf goot output at time f and let yif be the number of units of ith intermediate good used in production at time t.The final goods production function is given by: Yc=( 。。),鼽)百 intermediate good.Given this price,the demand for intermediate good i at time f is given by: (3) where 0<0<1.The market for the final good is perfectly competitive.Letp"be the price ofa unit ofthe fth For the details of the model,refer to Comin and HobOn(2007) 602 ANALYSIS ON TECHNOLOGY IMPLEMENTATION AND ECONOMIC GROWTH -O yc Yit=(] ̄1、 ,(4) Intermediate Goods Producers Each intermediate good is provided by a single producer who owns a patent that ensures his monopoly over the production of the good.Intermediate goods suppliers make two types of decisions as mentioned below- Factor demands and price setting(ifrst type).Intermediate goods are produced using capital St and labor lif that are combined using a Cobb—Douglas technology ofthe form: Y£c=ait垤2 (5) where 0<仅<1.The productivity level with which the/th intermediate good is produced,aif,is time’varying. Each intermediate goods producer hires them at the competitive net rates Wf and rt The the fth intermediate goods producer maximizes the flow profit and we obtaine the following results where 6 denotes the capital deprecimion rate,mc『f=(1/a)(wJ(1一 ))卜“(( + ) ,the marginal cost of production,It"ir,flow profit,and ,, the value offirm. c= (1一 ) c 口nd + :Oa Yit (6) Pct=吉m if-ft=(1一O)pftYic (7) (8) £=(1一日)J P—J d P Yi ds Aggregation over intermediate goods(first type).Aggregation for production functioin is: (9) (1O) (11) Yc=zc f}一 =Zt where the aggregate capital and labor inputs are given by: k k di and lf= lit di=1 Furthermore,the aggregate level oftotal factor productiviy tis given by a CES aggregate ofthe productivity leveI of the intermediate goods: 1一日 1丁 Z£ n 。diJl (12) The aggregate factor demands tuna out to satisfy the same optimality conditions as the factor demands ofthe individual finn in the sense that: wc=o(1一 Yt and r£+ = The aggregate production function representation allows US to rewrite the value of the firm s:a ( 3) t=(1-o)J[e一舯 ㈢而Ysd 0 (14) Technology implementation(second type).We define the implementation level ofintremediate i at time f aS: Xit ㈢而 05) where denotes productiviy ltevel and af the potential productivity level of intermediate i,and 0∈[0,1】一 We ssume alearning leads xi£to evelop according to: 文n=九 ( 一x囊) ANALYsIs 0N TEcHN0LoGY IMPLEMENTATIoN AND ECoN0MIc GR0wTH 603 where n∈[0,1】. >0 is the learning rate and ? eaming patternL’mplieS that: wherecis the time when cur、,ature changes on the way t oa3_GiVen this path of productivity,the value r1 7、 cn。n ( ):(卜 ) 0 t.o芦 。 ∞宫 。 P一舯 ) 。ysds ! ,( 8 deteimDlemented at time f at Xtc being initial implementat on leVe。a Q suu .mrinedby the ex0 0uS learning pro∞ 。 i ~ onlevellementatifthe imp e subsequent patho rtTh e to staher e their learnin g c urvide wFirms o nly dec。fi 叽ediateath口p……啦ue1des矾thde spurbosoeuqucuenwttIy fdueto the u ¨‘ ……‘e…。 ‘ exogenous learning process(see Figure 1。time 0)・ 10 娜 _l3 — 8 o 6 4 2 O time Fig“re』.Implementation curve(example) Next,implementmion cost of good f at time t takes the f0m c …n ( )=(1—8) z7c-a ) 。卜ln(1一 )一 1 (19) Here ̄>0 is the implementation cost p ame ‘s.difference between the value。f the firm and the The oDtimal implementaion level,Xtt nlaxlmlzes the d I erence u‘ n‘u ……… implementation costs: c( cc)一Ctm p 。m (Xtt) (20) Diffrentiating equati。n(20)with reSpect t。 We。btain the f0 。。w ngreSu - 一㈢ btYt 8 (21) P一 ̄;rjdj㈢ With equati。n(21)and equation(22),the Va1ue。fthe fim ris ew in朗 (22) iate equation(1 5)and ANALYSIS ON TECHNOLOGY IMPLEMENTATION AND ECONOMIC GROWTH (1一O)zl bt Yt跏(1+ ) 1 (23) (24) R&D Sector Let gt denote the growth rate ofpotential productivity between the intermediate good developed at t-dt and f.That is: 詈= c technological frontier of t is equal to: 厅一Z /,●\ \ — The cost of inventing an intermediate good whose potential productiviy reprtesents a growth in the f 0 [or gt<0 日l】一l — &。 i【(1一 一O) ̄zl ) ( ) g fc or 0c = &D( ),which can be rewritten as: (25) where >0 is R&D cost parameter.The R&D patent race between innovators brings the growth rate of the potential productivity induced by intermediate good,to the point where the value of the intermediate good equals the R&D cost.That is, 吾 n(1尝)) Equilibrium ,(26) (27) We specify the average potential productivity levelwhich all currently available intermediate good technology can be operated,which equals to: 1-0 \ ,—)/ For the equilibrium conditions and steady state analysiswe transform C Yt,k and if into stationary variables before exploring the transitional dynamics nd sateady state.Just like Comin and Hobijn(2007),we do _ lSO by scaling them by the average productivity trend zc1~.The resulting detrended variables are: ct =Tct,),;=卑, ;=卑,and f;={ z -n z ,(28) Because our aim is to obtain a stationary representation of the equilibrium pathwe detrended our three productiviy tmeasures,a—t,一ztand Zt.This yields the following two detrended productiviy tmeasures: ,0 0 2 (詈)而nnd = )而 respectively. (29) Comin and Hobijn(2007)denotes these measures as potential productivity gap and implementation gap We define a stationary equilibrium in term of the following variables: 嘲 ,Xtt,gt,b 彬 ,z 0 ) Steady State We define the steady state as the equilibrium path where the growth rate of the variables including the ANALYSIS oN TECHNoLoGY IMPLEMENTATIoN AND ECON0MIC GROWTH 605 transformed variables are constant.Let the steady state values of the following variables be r , i , 季, , 爻i, 而,看},and the values at steady state are: 0 、 (彘) ( ctO) k = (30) (31) r=( + 1 ) :——— (32) (33) f(p+lf, )(p+占 _)l ̄'Og+( 0)a-“a ( (意) n b = ) )) (35) (36) (37) (38) 1 1 _} ~ 磊= 而0一z 一。P = oe9 (39) where = +( 一1) 1 and z。and覆。are initial valuas respectively.In steady state,蜀= 而and Xic=1,therefore,the variables c,Xic,z 一,and I£can be omitted at steady state. Next,we focus on the steady state and perform the static analysis.As the Figure 2 of Comin and Hobijn’S (2007),we can c。nifrm that the relati。nship between and is negatiVe <0)from equati。n(33)and 0 Figure 2.Existence of 606 ANALYSIS ON TECHNOLOGY IMPLEMENTATION AND ECONOMIC GROWTH Proposition 1:When the below equation holds,the growth rate of productivity uniquely exists・ = +G一1) I>0 of equation(34)as ・)and differentiate this with respect to ,then: <0 (4o) 一彤 童:}o∞ 协 。 (41) Now.we diff.erentiate the L.H.S.of equation(34)with respect to ,we obtain 1 and we express the R.H・S・ When =0,the R.H.S.of equation(34)is positive.We show these relationships in Figure 2・From Figure 2,we confirm that g exists uniquely。 Static Analysis Here.we menti0n the static analysis for the growth rate of potential productivity in equation(34)・To do so.we apply the implicit function theorem to equation(34).We denote this function s G(a Itfollowsthat: 箬:一d —a <。G(・)、 d,筹:一f <。旦 …d,嚣:一 <。旦 a萄 a蚕 d窜 (42) Therefore,the below proposition is obtained. Propositi0n 2:The higher the values of parameters for implementati。n level ,implememation c。st , and R&D cost are,the lower the growth rate of potential productivity is(see Figure 3,as an example)・ time Figure 3 Implementation curve time. Regarding the static analys1‘s for the implementati。n level of intermediate goods,we。btain the f0ll。wing results: 著=筹+筹塑三0一 面一乏d;t 堡:丝堑>0 d d目d (、4。 3) (44) a6f・ 鬟=武8 +嚣筹=雾+嚣 >o{ a§武a a§ (45) O ANALYSIS 0N TECHN0L0GY IMPLEMENTAT10N AND EC0N0MIC GR0WTH N w t 607 h e O w g Regarding the sign of equation(43)to be ambiguous,we obtain the following result: e n q U 一塑妄嚣塑兮堕芝00;t d‘a厅 ;t a;t ‘ (、 47) O The equation(47)includes many parameters,SO that it is diiculft to determine the signs in the equation. n S a From this,the following proposition is obtained. Proposition 3:The higher the parameters( , )of implementation and R&D costs are,the higher the initial h implementation l is.However,when parameter of initial implementation level/z become large,given d evelO O equation(47),the initial implementation level might either increases or decreases. 塑 < Comparison With n and Hobijn(2007) 0Comi (2007).weexPressthegr。twh rateas 4 and = 0 +( 一1) 1 inC。min andH。bijn(2007).we aSSUme < 0 Next,we compare the growth rate of potential productivity in this paper with the one in Comin and Hobijn 堕叼 that/2<1 .From equation(40),weobtainthat tf,< .We expresstheL.H.S.ofequation(34)and c as ‘) 堕m and Fc(・)and both equations re adefferentiated with respect to and c respectively,then it follows that: < 0 <0 nnd <0 (48) However,we cannot determine which slope is bigger(smaller)for equation(48).Therefore,the below three Cases occur. Case lI O gc < a香 (49)、。 Case 2 OVc(’)一OF(’) Case 3 0 C一百g og (50) ,L 錾 > 09c 4 09 (51) Next,when = c=0, (‘)>F(‘)>0.In Figure 4,Case 1,which hasthe negative different slopes respectively,is shown.The position A and C on horizontal axis色indicate the steady state values obtained in this paper.The position B on the xias indicates the steady state value of gc in Comin and HobUn(2007).From Figure 6 、J 4,in comparison with the value in Comin and Hobijn(2007),depending on the slope of equation(34),the relationship which value is bigger or smaller is determined.In this case,like analogizing with equation(42),the parameter signiifcantly affects this relationship,and when this parameter value is suficifently large(smal1), the growth rate of the potential productiviy tat steady state of Comin and Hobijn(2007)is larger(smaller)than the one obtained here.In Case 2 and Case 3,we obtain the result like the position A,i.e., c> .From this,we obtain the following proposition. Rear to equation(1 45)on page 46 in Mathematical Details of Comin and Hob ̄n(2007) Accoding to Comin and Hob ̄n(2007),generally <1. 608 ANALYSIS 0N TECHN0L0GY IMPLEMENTAT10N AND EC0N0MIC GROWTH 07) O Figure 4.Case 1 the relationship of growth rates. Proposition 4:Under the caSe O gc < dg ,when the parameter of implementation level here andthe parameter of learning rate for Comin and Hobijn(2007)/z,are sufficiently large(smal1),the growth rate of potential productivity in Comin and Hobijn(2007)gc is larger(smaller)than the one in this paper.When ≥ . Based on numerical calculation,we compute the values of initial implementation level in equation(33),and the one in Comin nd aHobijn(2007)。.As a result,the value or finitial implementation level in equation(33) is 0.1 07, nd athe one in Comin and Hobijn(2007) is 0.033.To sum up,the higher value of initial implementation level is necessary for our leaming profile.Finally,regarding the transitional dynamics,SO far we confirm 4 negative and 2 negative eigenvalues in our system.However this should be proceeded with careful attention due to the complexity ofour mode1. Conclusion Instead of the leaming profile in Comin and Hobijn(2007),we incorporate the S—shape learning profile (sigmoid function)into their model,and analyze growth rate of potential productiviy,itnitial implementation level,and ifnal good(incom .We obtain the same results as theirs for many respects.However,aS the different points from them,even if the speed of learning effectiveness accelerates on the way to steady state,the growth rate ofpotential productivity decreases.which is opposite result ofComin and Hobijn’S(2007).This reflects that implementing technology iS expensive due to increase of speed,SO that investment into R&D sector decreases. Moreover,we obtain the result that whether or not the growth rate of potential productivity at steady state in this model is bigger than the one of Comin and Hobijn(200)depends on values of parameters. As the further study.as our final goal iS to analyze the relationship between technological implementation level(or implementation profile)for ICT and income(or income inequality)precisely,more detailed considerations and examinations for the results to be obtained here iS required.In addition to this。social planner Thevalues ofparameterused here 81"e P=0.050,0=0.850, =0.300, =0.017, =216, =115,6=0.050,which are cited from Comin and Hob ̄n(2007),and we assume 0.020. ANALYSIS 0N TECHNOLOGY IMPLEMENTAT10N AND ECoNOMIC GROWTH 609 will be analyzed as a reference economy and in the decentralized economy developed here,optimal tax or subsidy have to be derived to realize the optimal growth rate of potential productivity like in social planner. References Acemoglu,D.,&Dell,M.(2009).Productivity differences between and within countries.NBER Working paper,No.15155. Acemoglu,D.,&Zilibotti,F.(2001).Productivity differences.QuarterlyJournal fEconomiocs,』,6,562—606. Alvarez,F.E.,Buera,F.,&Lucas,R.E.Jr.(2008).Models of idea lfows.NBER Working paper,No.14 135. Anderson,J.(2009).Globalization and income distribution:A speciifc factors continuum approach.NBER Working paper,No. 14643. Caselli,F.(2005).Accounting for cross—country income differences.In A.Phillippe,&N.D.Steven(Eds.),Handbook of Economic Growth(Volume IA,PP.679・741). Comin,D.,&HobUn,B.(2007).Implementing technology.NBER Working paper No.12886. Eeckhout,J.,&Jovanovic,B.(2002).Knowledge spillovers and inequaliy.tAmerican Economic Review,92,1 290-1 307. Katagiri,S.(2008).Analysis of income inequality in global economic environmental change:Using the endogenous growth model including research and development sector.Studies in AppliedEconomics,1,l 19-l38. Lucas,R.E.Jr.(2008).Ideas and growth.NBER Working Paper,No.1 4 l 33. Pilat.D.(2004).The ICT productivity paradox:Insights from micro data.OECD Economic Studies,38. Pilat.D..&W6lh1.A.(2004)./CT and economfc growth New evideneefrom international compar&on.Mimeo.