Patents Rights and Economic Growth: Empirical Evidence from Middle Income Countries

Transcription

1 Patents Rghts and Economc Growth: Emprcal Evdence from Mddle Income Countres Ata ur Rehman 1*, Sumara Usman 2, Atya Had 3, Dr. Naqv Hamad 4 Ata.fuuast@gmal.com 1,2 Federal Urdu Unversty of Arts, Scence and Technology, Islamabad, Pakstan 3,4 Polcy & Strategc Plannng Unt, Health Department, Punjab, Pakstan Abstract Ths study nvestgates the assocaton between patents rghts and economc growth, n case of selected mddle ncome countres, usng panel ARDL (auto regressve dstrbuted lag) technque. Results under PMG and DFE estmator ndcate that domestc patents contrbute sgnfcantly to GDP of mddle ncome only n short run, whle MG estmator ndcates that patents have no sgnfcant mpact on economc growth both n short and long run. Keywords IPRs, Patents, Economc Growth, R&D. 1. INTRODUCTION One of most mportant characterstc of a modern economy s that technologcal mprovement s crucal for economc development. Modern technology not only mproves factor`s productvty but also boost up the level of output. Among many other factors, one of the most mportant factor that encourages technology s ntellectual property rghts (IPRs). IPRs encourage nvestors and producers to explore new producton technques by provdng them property rghts over ther nventons. The nvestors and nnovators take decson of nvestment dependng upon patents rghts n order to earn hgher profts from ther nventon and these nventons and modern producton technques n turn rase the growth rate of economy. Man objectve of our research s to analyze mpact of resdent patents on economc growth, n lower and upper mddle ncome countres (for tme perod ). Accordng to Park and Gnarte (1997), there exst ndrect co-relaton between patents and economc growth.e. stronger patents rghts enhance R&D and physcal nvestment of an economy, whch n turn boost economc growth. The study s organzed n followng format; after bref ntroducton of study (secton 1), secton 2 consst of lterature revew. Furthermore, model specfcaton and theoretcal framework s dscussed n secton 3, whle secton 4 conssts of results and dscusson and secton 5 comprses of concludng remarks of present study. 2. LITERATURE REVIEW Prevous lterature demonstrates mportance of IPRs n economc growth, and also gves detal about channels through whch IPRs affect growth. Falvey and Foster (2006) and Kumar, 2003 specfy that IPRs affect economc growth both drectly (through encouragng domestc patents) and ndrectly (through dffuson of technology from north). Accordng to Jalles (2010) patents statstcs contrbute postvely to economc growth through promotng novel deas and nnovatons. Helpman (1991) states that although n past R&D data has been utlzed frequently to determne nnovaton, but R&D data only measures nnovatons nputs. Moreover, tme lag s also nvolved between R&D expendtures and output of these expendtures. So, IPRs statstcs (partcularly patents) s used as a proxy for nnovatons (Wang, 2013). Grlches(1990) examne that patents statstcs s consdered as measure of the output of modern technology, whle R&D expendtures are nput of new technology. Although, above lterate ndcate mportance of patents toward nnovaton but on the other sde, Furukawa (2010) and Mnea and Hudson (2013) explore nverted U shape relatonshp between IPRs and nnovaton, and Furukawa (2010) states that IPRs encourage nnovaton n short run, whle n long run IPRs lower nnovaton by depressng the learnng by dong process. Schneder (2004) explore that IPRs encourage nnovaton only n developed country and ndcate that dynamcs of nnovatons and growth are dfferent n developed and developng economes, and IPRs encourage nnovaton only n developed countres. Mnea and Hudson (2013) explore that relaton between IPRs and nnovaton depends upon ntal level of IPRs and GDP per capta level of a partcular country. Even though, above lterature ndcate nverted U shaped relatonshp between patents and nnovaton, but on other sde patents have also contrbuton n ncome of economes. Km.et al., 2009 ndcate that patents statstcs (resdent and non-resdent) contrbute postvely to factor productvty; Moreover, non-resdent patents 137

2 are more sgnfcant than resdent patents. Park (1999); Yang (2006) and Iwasako and Futagam (2011) end up wth result that patents nfluence economc growth ndrectly through encouragng factor accumulaton, R&D and physcal captal. The stronger patents rghts uphold ndustral growth through techncal progress durng tme perod and ,whereas durng year and through rapd factor accumulaton (Hu and Png, 2009).Although, above lterature elaborates that patents encourage economc growth, but patent protecton s consdered more sgnfcant n developed countres than n developng countres. Iwasako (2013) states that heght of patents protecton depends upon rato of publc expendtures to output level of an economy. Lee et al. (2013) and Schneder (2005) argue developng countres are not equpped well to bear cost of patents at present phase of development. Thompson and Rushng (1996) conclude that patents protecton add to economc growth after achevng a certan level of economc growth (Kumar, 2003). Gnarte and Park (1997) and Iwasako and Futagam (2011) explore that sze of research sector, R&D productvty and nfrastructure are also mportant determnant of patents and growth relatonshp. 3. MODEL SPECIFICATION, THEORETICAL FRAMEWORK AND METHODOLOGY Technologcal progress, nnovatons and knowledge are consdered crucal for economc growth of any economy. Classcal economsts (ncludng Adam Smth and others) demonstrated that productvty depend upon savngs, populaton growth and technologcal development. Solow added technology to the producton functon equaton, however, exogenously, whle new growth theores take technologcal progress as endogenous e.g. Romer. P (2002), Grossman and Helpman (1991 a) ntroduced model of endogenous techncal progress. Furthermore, knowledge has an mportant characterstc, that t s non-rval. But on other hand, accordng to Romer (1990) although knowledge s non-rval but ts heterogenety made t excludable. Knowledge s excludable f t s possble to prevent others from usng t, for example, patent laws gve exclusve rghts to nventor over the use of ther desgns and dscoveres. The man purpose of patents also s to encourage nnovaton rate and growth rate of an economy. By keepng n vew above theoretcal background, we wll construct our model. In our study, technologcal progress, labor and captal all factors are combned, whle techncal progress and nnovatons have major contrbuton n growth process. Wrtng Cobb Douglas producton functon. 3.1 Y = Output A = technology K= Captal = margnal productvty of captal L= Labor =margnal productvty of labor In our model technology (A) s affected by total resdence patent granted, IPRs encourage nvestors and nnovators through provdng exclusve rghts to resdence nvestor, whch n turn affect economc growth. Wrtng A n form of equaton.3.2 In equaton 3.2 PR refers to total no of resdence patents granted, whereas X refers to vector other varable, those affect domestc nnovaton and technology.e. human captal and research and development expendtures as % of GDP. Puttng values of X n equaton Puttng equaton 3.3n equaton 3.1, fnal estmable model can be wrtten n followng form..3.4 In equaton 3.4Y refers to output growth and here GDP growth (annual %) s used to estmate output growth. log(pr) s log of total no of resdence patents granted, R&D refers to research and development expendture (% of GDP). Data GDP growth (annual %) and R&D was taken from World Bank, whle data on patents was taken from statstcs on WIPO (world ntellectual property organzaton).hc s human captal ndex calculated as sum of enrollment at all educaton level (.e. prmary, secondary and tertary enrollment) and ths sum was dvded by total labor force. Data on prmary, secondary and tertary school enrollment was taken from UNESCO (Unted Natons educatonal, scentfc and cultural organzaton) statstcs, whle total labor force data was taken from World Bank. 3.1 Methodologcal framework To estmate model of study, we used panel ARDL (autoregressve dstrbuted lag) technque. Pesaran, Shn and Smth (1997, 1999) ntroduced PMG (pooled mean group), MG (mean group) and DFE (dynamc fxed effects) technques, for estmaton of panel ARDL, heterogeneous panel and non-statonary dynamc panel data 1. 1 Dynamc panel data consst of larger no of cross sectonal (N) & also larger no of tme seres observatons (T). Dynamc panel data further assume heterogonous co-effcent for cross sectons unts or group. 138

3 MG technque assumes heterogonous coeffcents both for short run and long run, whle PMG whch allow homogenous coeffcents for long run but heterogonous coeffcents for short run. Lkewse, DFE assumes homogenous coeffcents both n short and long run but estmates panel specfc ntercepts (Rafndad and Yosuf, 2013). Generally, ARDL s wrtten as; Yt = a + g Y, t- j+ b X, t j + et Reparametrng equaton 3.5 n VECM format; p-1 q-1 D - å g y å - j= 1 j= 0 Y q ( 1 b, 1), d ( ) e t = Y...3.6, t - X t- + DY t- j + D X y t j + + t In equaton 3.6 q represent error correcton parameter or speed of adjustment n long run (exstence of long run co-ntegraton relaton), whleb refers to long run coeffcents. Furthermore,g andd are short run coeffcents. By, keepng n vew above methodologcal framework, we wll construct our model n ARDL format. Wrtng equaton n 3.4 VECM format; m Results and dscusson To check order of ntegraton, panel unt root test ((Levn, Ln and Chu Test (test wth common unt root process)) was appled. 4.1 Unt root test Results of unt root test are gven n Table 4.1. P-values are reported n parentheses, t-statstcs are above parentheses values. Results ndcate that Y, PR (Resdent patents), HC (human captal) are statonary at level, as p-value s less than 0.05 and t-statstcs s negatve. R&D s statonary at 1 st dfference. As, our three varables are of I (0) whle other one s of I (0), so panel ARDL technque wll applcable here. Table 4.1 Levn, Ln and Chu Test Varables Level 1 st dfference Integraton order Y (0.0129)* - I(0) PR (0.0427)* - I(0) HC ( )* - I(0) R&D (0.0000)* I(1) P values are n parentheses, t values above parentheses values (* sgnfcant at 5%) 4.2 Lag length selecton After formulaton of ARDL equaton, lag length s selected and for selecton of lag AIC and SC crtera s used. In our model, optmal lag length s 1, because SC value s lesser at lag 1, as shown n table 4.2. So, Schwarz- Bayesan crteron (SBC) s used to choose optmal number of lags for ECM. Table 4.2 Lag length selecton Lags AIC SC HQ * * * * Indcate sgnfcant at 5% level 139

4 4.3 Panel ARDL We can estmate equaton 3.7 usng pooled mean group, mean group and dynamc fxed effects approach. Results of panel ARDL are reported n table 4.3. Table 4.3 Panel ARDL PMG MG DFE Varable Coeffcent t-statstc Prob. Coeffcent t-statstc Prob. Coeffcent t-statstc Prob. Short Run Model EC Constant D1.PR (0.000*) 3.75 (0.000*) 1.60 ( 0.111**) (0.000)* (0.095)** 0.94 (0.348) (0.000)* 1.58 (0.114)** 1.57 (0.116)** D1. HC (0.344) (0.265) (0.228) D1.RD (0.123) (0.580) (0.148) Long Run Model L.Pr (0.657) (0.949) (0.994) L.Hc (0.216) (0.181) (0.468) L.Rd (0.549) (0.267) (0.811) * ndcate Sgnfcant at 5% level, ** ndcate sgnfcant at 10% level Pooled Mean Group (PMG) Table 4.3 (column 2 and 3) demonstrates results of PMG estmator. Upper part of table 4.3 depcts short run model. Results ndcate that n short run only PR(resdent patents) contrbutes sgnfcantly to ncome of mddle at 10% level of sgnfcant and one percentage ncrease n patents leads to 2.14% ncrement n ncome of mddle ncome countres. Furthermore, HC (human captal) and RD (research and development expendture) are nsgnfcant n short run, and have negatve coeffcent n case of mddle ncome countres. Furthermore, for sake of adjustment parameter (restrcted ECM equaton) s calculated. Results are reported n table 4.3. E(-1) s error correcton term or adjustment parameter. Results ndcate that co-effcent of EC (adjustment parameter) s negatve (negatve sgn ndcate convergence) and 75% of dsequlbrum s corrected n one tme perod. Lower part of table 4.3 (column 2 and 3) depcts long run model. Results ndcate that n long run none of varable has sgnfcant effects on ncome of mddle ncome countres. Resdent patents (PR) have no contrbuton towards growth of mddle ncome countres, the reason s that developng countres are not prepared well to acheve benefts from patents at ther present stage of development, accordng to Thompson and Rushng (1996) patents protectons contrbute to economc growth after achevng a certan level of economc growth (kumar, 2003). Moreover, due to weak nfrastructure, market structure, economc polces of developng economes regardng resdence patents and neffcent R&D sector, patents don t contrbute to growth. In long run, coeffcent of research and development expendtures (RD) s negatve because of neffcent research and development sector n developng countres; results are persstence wth Brdsall and Rhee (1993). Moreover, human captal (HC) s also nsgnfcant and results are persstence wth (Mayer, 2001) and Benhabb and 140

5 Spegel (1994) 1 ), means contrbuton of educaton sector towards ncome s not sgnfcant n mddle ncome countres, for the reason that n mddle ncome economes, educaton sector s not so actve, so educaton has no contrbuton towards ncome n mddle ncome countres Mean Group (MG) Table 4.3 (column 4 and 5) reports results of mean group estmator. Above part of table 4.4 depcts short run dynamcs. In short run, none of varable contrbutes sgnfcantly to ncome of mddle ncome countres, as gven n table 4.4. Coeffcent of PR s nsgnfcant n case of MG contrary to PMG technque. Moreover, RD s nsgnfcant wth postve coeffcent opposng to PMG whle coeffcent of HC s also nsgnfcant wth negatve coeffcent. Lower part of table 4.3 (column 4 and 5) depcts long run model. Results ndcate that n long run, PR (resdent patents) contrbutes negatvely towards ncome of mddle ncome countres, contrary to PMG approach. Furthermore, human captal has nsgnfcant mpacts on ncome of mddle ncome countres. Reason of nsgnfcant results s that n mddle ncome economes, educaton sector s not so actve, so educaton has no contrbuton towards ncome n mddle ncome countres 2 and results are persstence wth Hanushek (2013). Moreover, RD (research and development expendtures) has negatve mpact on growth of mddle ncome countres. Adjustment parameter s negatve and shows convergence, means 84% of dsequlbrum s corrected n one tme perod Dynamc fxed Effects (DFE) Table 4.3 (column 5 and 6) demonstrates results regardng DFE estmator. Upper part of table 4.3 (column 5 and 6) depct short run model whle lower part of table demonstrate long run model. Results of DFE estmator are smlar to PMG estmator and none of varable contrbutes sgnfcantly to economc growth n long run. Addtonally, n short run only PR s sgnfcant at 5% level, whle RD and HC have no contrbuton n growth of mddle ncome countres. In long run, although coeffcents of resdent patents and human captal have postve sgn but t have no sgnfcant contrbuton n ncome of mddle ncome countres. Furthermore, coeffcent of HC s negatve n both short and long run. Results demonstrate that speed of adjustment s means 83% of dsequlbrum s corrected n one tme perod. Hausman Test Table 4.6 ndcates results of houseman test for comparson of PMG, MG and DFE estmators. Results demonstrate that PMG s effcent estmator than MG and DFE. As, n both cases null hypothess of effcent PMG estmator s accepted. Hausman test TABLE 4.6 HAUSMAN TEST Null Hypothess Prob>ch ch Results PMG s effcent estmaton than MG under null Hypothess MG s effcent estmaton than DFE under null Hypothess PMG s effcent estmaton than DFE under null Hypothess PMG s effcent estmator. MG s effcent estmator. PMG s effcent estmator. 4.4 Research and Development expendtures Result of R&D are negatve because n developng countres research and development expendtures contrbute to economc after achevng certan level of growth. Furthermore, GDP and ntal level of research and development expendtures also affect R&D and ts postve consequences (Brdsalland, 1993). 1 Benhabb and Spegel (1994) conclude that one reason of negatve sgn s that n developng countres started wth low level of ntal human captal stock. And countres those start wth effcent ntal human captal stock, smaller mprovement n ther educaton lead to larger mprovement n human captal stock. Furthermore, Benhabb and Spegel (1994) recommend usng human captal n level rather than n growth to obtan sgnfcant outcomes. 2 Accordng to Hanushek (2013) Cogntve sklls of the populaton have more mportance than school attanment of partcular country. Hanushek (2013) demonstrates that Cogntve sklls have major role n earnng of any ndvdual and especally n growth of economy and polcy makers should focus on school qualty and ts mprovement. 141

6 Brdsalland (1993) explore that R&D expendtures depend upon ntal stock of R&D and GDP of country because countres havng hgher level of GDP contan lower margnal productvty of physcal nvestment (MPPI). Lower MPPI encourages these countres to nvest n R&D sector n order to lghten the dmnshng margnal productvty of physcal captal. Brdsalland (1993) nvestgate reason for the postve effects of ntal stock of R&D on current level of R&D s that productvty of R&D captal s not dmnshng.e., there are constant or ncreasng returns to R&D nvestment. Brdsalland (1993) further stated that Ths s consstent wth the vew that huge fxed costs are nvolved n R&D nvestment Dependent varable Y (GDP growth) Table 4.7 demonstrates assocaton between research and development expendtures (RD) and GDP growth. Varable RD (research and development expendtures) affects GDP negatvely at level, as coeffcents of RD s negatve. Whle, RD contrbutes postvely to GDP at 1 st and 2 nd dfference as coeffcent of RD (research and development expendtures) s postve at 1st and 2 nd lag of RD. So, nverted U shaped relatonshp exsts between GDP and research and development expendtures. Table 4.7 Relatonshp between Y (GDP growth) and research and development expendtures (RD) Varable Coeffcent Std. Error t-statstc Prob. C * RD RD(-1) RD(-2) R-squared * Sgnfcant at 5% level Dependent varable RD As, dscussed above RD s affect by prevous level, so we compute assocaton between RD and lag value of RD. RD s postvely affected by ts prevous lags both lags are sgnfcant at 5% level as gven n table 4.8. Table 4.8 Relatonshp between RD and lag value of RD Varable Coeffcent Std. Error t-statstc Prob. C RD(-1) * RD(-2) * R-squared * Indcate coeffcents are sgnfcant at 5% level 4.5 DIAGNOSTIC TEST Dagnostc test are reported n table 4.9. Dstrbutons are not normally dstrbuted, as Jarque-Bera test reject null hypothess of normal dstrbutons. Whle, for seral ndependent resduals LM test s employed and error term are not correlated as shown n table 4.9. Moreover, varance of error term s not persstence because we reject null hypothess of persstent varance of error term. For sake of dynamc stablty of UECM, nverse roots are constructed (graphs are reported n appendx; fg A1 and A2). Before dfferencng (fg A1), nverse roots demonstrate nstablty. So, we constructed nverse root agan after dfferencng. After dfferencng (fg A2) nverse roots exhbts stablty as none of root lay on x-axs. 142

7 Table 4.9 Dagnostc tests Dagnostc Tests Test Stats Results Normalty Test Jarque-Bera (0.0000)* Not normal Seral Correlaton Test LM-Stat Lags LM-Stat Prob No seral correlaton Standard Error SE % Auto corelaton DW-statstcs Not exsts * Sgnfcant at 5% level 5. Concluson and polcy recommendatons In ths study, we analyzed mpact of domestc patents on economc growth usng panel ARDL, n case of mddle ncome countres for tme perod Results revealed that domestc patents affect GDP of mddle ncome countres postvely only n short run accordng to PMG and DFE estmator. Whle, MG estmator reveals nsgnfcant results both n short and long run. Ths reveals that present development stage of country and actve R&D sector also play mportant role to obtan encouragng consequences of domestc patents. So, developng countres can obtan benefts from patents through adequate R&D facltes, provdng ncentves for nvestors, buldng adequate nfrastructure and encouragng domestc patents. As, IPRs contrbute to GDP after a certan phase of development, so WTO can resolve ths ssue through allowng flexbltes for developng country regardng IPRs mplementatons. Moreover, proper nformaton regardng patents laws and mplementaton should be provded to nvestors through regular publcaton n order to encourage patents holder n developng countres. REFERENCES Benhabb. J., Spegel, M. (1994). The role of human captal n economc development. Evdence from aggregate cross-country data. Journal of Monetary Economcs, 34, Brdsall, N., Rhee, Y. (1993). Does R&D contrbute to economc growth n developng countres? Polcy research workng paper No. 1221, The World Bank Polcy Research Department. Falvey, R., Foster, N. (2006). The role of ntellectual property rghts n technology transfer and economc growth: Theory and evdence.workng paper of Unted Natons Industral Development Organzaton. Furukawa, Y. (2010). Intellectual Property Protecton and Innovaton: An Inverted-U Relatonshp. Dscusson paper No 1001, School of Economcs, Chukyo Unversty. Gnarte, J. C. and Park, W. G. (1997). Determnants of patent rghts: A cross-natonal study. Research Polcy, 26, Grlches, Z. (1990). Patent statstcs as economc ndcators : A survey. Journal of Economc Lterature, Amercan Economc Assocaton. 28 (4), Grossman, M., Helpman, E. (1991,a). Innovaton and growth n global economy. The MIT press, Cambrdge, MA. Hu, A. G. Z., Png, I.P.L. (2009). Patent rghts and economc growth: evdence from cross-country panels of manufacturng ndustres.department of Economcs, Natonal Unversty of Sngapore. Hanushek, E. A. (2013). Economc growth n developng countres: The role of human captal. Economcs of Educaton Revew, 37 ( ),

8 Iwasako, T. (2013). Welfare effects of patent protecton and productve publc servces: Why do developng countres prefer weaker patent protecton? Scence drect journal, 118, Iwasako, T., Futagam, K. (2011). Patent, protecton, captal accumulaton, and economc Growth.Journal of Econ Theory, 52, Jalles, T. J. (2010). How to measure nnovaton? New evdence of thetechnology growth lnkage. Research n Economcs, Vol (64), Jefferson, G. H., Wu, K., Ca, H., Jang, R. (2013). Trade and ntellectual property rghtsas channels for economc growth. Asa-Pacfc Journal of Accountng andeconomcs, 20(1), Km, T., Maskus, K., and Oh, K-U. (2009). Effects of patent producton on producton n Korean Manufacturng: A panel data analyss. Pacfc economc revew, 14(2), PP Kumar, N. (2003). Intellectual Property Rghts, Technology and Economc Development: Experences of Asan Countres. Economc and Poltcal Weekly, 38(3), Lee, K., Km, J., Oh, J., and Park K-H. (2013). Economcs of ntellectual property n the context of a shftng nnovaton paradgm: A revew from the perspectve of developng. Global Economc Revew: Perspectves on East Asan Economes and Industres, 42(1), Mayer, J. (2001). Technology dffuson, human captal and economc growth n developng countres. UNCTAD/OSG/DP/, Workng paper no. 154.Unted natons conference on trade and development. Mnea, A and Hudson, J. (2013). Innovaton, ntellectual property rghts, and economc development: A unfed emprcal nvestgaton. Journal of World Development, 46, Nabn, M., Nguyen, X., Sgro, P. (2013). On the relatonshp between technology transfer and economc growth n Asan economes. The World Economy journal, 10, Park, W.G. (1999). Impact of the nternatonal patent system on productvty and technology dffuson. In: Lppert, O (eds.) Compettve Strateges for Intellectual Property Protecton. Fraser Insttute, Vancouver, BC, Rafndad, A. and Yosuf, Z.(2013). An applcaton of panel ARDL n analyzng the dynamcs of fnancal development and economc growth n 38 sub-saharanafrcan contnents. Vol (2), Kuala Lumpur Internatonal Busness, Economcs and Law Conference. Romer, D. (2002). Advanced Macro economcs. (Second Edton). McGraw-Hgher Educaton. Romer, P. M. (1990). Endogenous techncal change. Journal of poltcal economy, (Oct part 2): Schneder, P. H. (2004). Internatonal trade, economc growth and ntellectual property rghts: A panel data study of developed and developng countres. Journal of Development Economcs, 78, Wang, C. (2013). The long-run effect of nnovaton on economc growth. School of Economcs, UNSW, Sydney 2052, Australa. Zhou, M., Flesher, B. M., McGure, W.M., Smth, A.N. (2013). Patent law, TRIPS, and economcgrowth: evdence from Chna Asa-Pacfc. Journal of Accountng and Economcs,21(1),

9 APPENDIX Innovaton equaton Fg A1: Before dfferentatng Fg A2: After dfferentatng 1.5 Inverse Roots of AR Characterstc Polynomal 1.5 Inverse Roots of AR Characterstc Polynomal