Imports and Intellectual Property Rights on Innovation in China Angus C. Chu

Guobing Shen

Xun Zhang

October 2017 Abstract In this study, we develop an open-economy R&D-based growth model with two intermediate production sectors that use domestic and foreign inputs, respectively. We …nd that strengthening intellectual property rights (IPR) has a positive e¤ect on innovation in the sector that uses domestic inputs but both positive and negative e¤ects on innovation in the sector that uses foreign inputs. We test and con…rm these theoretical results using an empirical analysis of matching samples that combine Chinese provincial IPR data with industrial enterprises database and customs database.

JEL classi…cation: F43, O31, O34 Keywords: Intellectual property rights; imports; knowledge spillovers; innovation Chu: China Center for Economic Studies, School of Economics, Fudan University, Shanghai, China. Email: [email protected]. Shen: Institute of World Economy, School of Economics, Fudan University, Shanghai, China. Email: [email protected]. Zhang: Institute of World Economy, School of Economics, Fudan University, Shanghai, China. Email: [email protected]. The authors are grateful to Zhao Chen, Yuichi Furukawa, Xiaohuan Lan, Hwan Lin, Ming Lu and Yibai Yang for their insightful comments and helpful suggestions. Shen acknowledges …nancial support from Chinese National Social Science Foundation Key Project (grant 15AZD058) and Chinese Ministry of Education Project of Key Research Institute of Humanities and Social Sciences at Universities "The Study on Innovation Protection and Chinese Firms’Promotion of Foreign Trade Competitiveness under the Industry Production Network".

1

1

Introduction

Intellectual property rights (IPR) serve as an important policy tool for stimulating innovation and economic growth. Seminal studies, such as Nordhaus (1969) and Judd (1985), and many subsequent studies assume that strengthening IPR stimulates innovation. However, some recent studies …nd that IPR may sti‡e innovation.1 In this study, we use a growththeoretic model to show that the e¤ect of IPR on innovation depends on how IPR a¤ects the spillovers of knowledge and confront our theory with empirics. Speci…cally, we develop an open-economy R&D-based growth model with two intermediate production sectors that use domestic and foreign inputs, respectively. Then, we use the model to explore the e¤ects of IPR on knowledge spillovers and innovation. Our results can be summarized as follows. In the sector that uses domestic inputs, strengthening IPR has a positive e¤ect on innovation. However, in the sector that uses foreign inputs, strengthening IPR has both positive and negative e¤ects, where the latter is due to IPR suppressing knowledge spillovers from imports. We test these theoretical results using an empirical analysis of matching samples that combine Chinese provincial IPR data with industrial enterprises database and customs database. Our regression results con…rm that IPR indeed has the usual positive e¤ect on innovation but also a negative interactive e¤ect on innovation via imports, which is consistent with the above-mentioned suppression e¤ect of IPR on knowledge spillovers from imports. However, importing …rms still have better innovation performance than non-importing …rms, which is consistent with our theoretical model. Finally, strengthening IPR …rst enlarges and then reduces the di¤erence in the innovation performance between importing and nonimporting …rms, which is also consistent with our theoretical model. This study relates to the theoretical literature on innovation and economic growth. The seminal study in this literature is Romer (1990). Subsequent studies use variants of the R&D-based growth model to explore the e¤ects of IPR; see for example Lai (1998), Li (2001), Goh and Olivier (2002), Grossman and Lai (2004), Furukawa (2010), Chu and Pan (2013), Iwaisako and Futagami (2013), Yang (2013), Cozzi and Galli (2014), Zeng et al. (2014), Lin (2015), Huang et al. (2017) and Saito (2017). The current study di¤ers from these studies by exploring a novel channel through which strengthening IPR causes a negative e¤ect on innovation by suppressing knowledge spillovers from imports of intermediate inputs. The study also relates to the empirical literature on the determinants of innovation. For example, Goldberg et al. (2010) use …rm-level data to show that imported intermediate inputs increase product innovation in India. Chen and Puttitanun (2005) consider crosscountry panel data whereas Hu and Png (2013) consider industry-country panel data, and both studies …nd that strengthening IPR increases innovation. Recent studies explore channels through which IPR a¤ects innovation. For example, Ang et al. (2014) show that IPR stimulates innovation by improving …rms’ external …nancing ability. Naghavi and Strozzi (2015) …nd that IPR interacts with international migration to encourage domestic innovation by creating an environment that transmits knowledge acquired by emigrants. The current study complements these studies by exploring the e¤ects of IPR on innovation in China via imports. 1

See for example Lai (1998), Goh and Olivier (2002), Furukawa (2010), Chu and Pan (2013), Iwaisako and Futagami (2013) and Saito (2017) for theoretical studies and also Ja¤e and Lerner (2004), Bessen and Meurer (2008) and Boldrin and Levine (2008) for evidence.

2

The rest of this study is organized as follows. Section 2 presents the theoretical model. Section 3 discusses the empirical framework. Section 4 shows the regression results. Section 5 concludes. Data description is relegated to the appendix.

2

Theoretical model

We extend the small-open-economy growth model2 in Grossman and Helpman (1991) into multiple production and R&D sectors. Also, we assume that one sector uses domestic inputs to produce di¤erentiated products, whereas the other sector uses foreign inputs.

2.1

Household

The representative household has the following utility function: Z 1 U= e t (ln Cy;t + ln Cz;t )dt,

(1)

0

where > 0 is the discount rate. Cy;t is the consumption of a domestic …nal good chosen as the numeraire.3 Cz;t is the consumption of an imported …nal good from abroad.4 Its price pz;t is exogenous and grows a constant rate, which can be positive, zero or negative. The asset-accumulation equation is A_ t = rt At + wt l

pz;t Cz;t .

Cy;t

(2)

At is the amount of assets. rt is the interest rate.5 l denotes labor. wt is the wage rate. From standard dynamic optimization, the optimality conditions are C_ y;t = rt Cy;t

,

Cz;t = Cy;t =pz;t . 2

(3) (4)

A small open economy may not fully capture China. However, our small-open-economy model simply assumes the terms of trade to be exogenous, and more importantly, the terms of trade neither a¤ect the equilibrium allocation of R&D labor nor the equilibrium growth rates of technologies. 3 Domestic …nal good can be consumed by the household, used to produce intermediate inputs or exported abroad. 4 Foreign …nal good can be consumed by the household or used to produce intermediate inputs. 5 Here we assume …nancial autarky under which the domestic …nancial market is not integrated to the global …nancial market; otherwise, the interest rate rt would be determined by the exogenous global interest rate. This assumption seems reasonable given capital control in China. Under the assumption of …nancial autarky, it can be shown that the asset-accumulation equation ensures balanced trade.

3

2.2

Domestic …nal good

Domestic …nal good is produced by the following aggregator:6 Yt = (Xtd )0:5 (Xtf )0:5 ,

(5)

where Xtd is an intermediate good that uses domestic inputs and Xtf is an intermediate good that uses foreign inputs. Pro…t maximization yields the following conditional demand functions for Xtd and Xtf : Yt Xtd = , (6) 2Ptd Yt

Xtf =

2Ptf

,

(7)

where Ptd and Ptf are the prices of Xtd and Xtf respectively.

2.3

Intermediate goods

Intermediate good i 2 fd; f g is produced by Xti

=

(Lit )1

Z

nit

[xit (!)] d!,

(8)

0

where Lit denotes domestic production labor and xit (!) denotes domestic or foreign di¤erentiated inputs. Pro…t maximization yields the following conditional demand functions for Lit and xit (!): wt = (1 )Pti Xti =Lit , (9) pit (!) = Pti (Lit )1

[xit (!)]

1

,

(10)

where pit (!) is the price of xit (!).

2.4

Domestic di¤erentiated inputs

Domestic di¤erentiated inputs xdt (!) are produced by domestic …nal good with an one-to-one technology. The pro…t function is d t (!)

= pdt (!)xdt (!)

xdt (!) = Ptd (Ldt )1

[xdt (!)]

xdt (!).

(11)

The monopolistic price is pdt (!) = minf ; 1= g, where < 1= . As is common in the 7 literature, due to incomplete patent protection , the monopolist cannot charge too high a price; otherwise, an imitator will produce xdt (!). The amount of pro…t for ! 2 [0; ndt ] is d t (!)

=(

1)xdt (!) =

1 Ptd Xtd = ndt

1 Yt 2ndt

d t,

(12)

Our results are robust to Yt = (Xtd ) (Xtf )1 ; derivations are available upon request. For simplicity, we focus on = 0:5. 7 See for example Li (2001), Goh and Olivier (2002), Iwaisako and Futagami (2013) and Yang (2013). 6

4

where the second equality uses symmetry in (8), (10) and pdt (!) = . The balanced-growth value of an invention is vtd (!) =

r

d t (!) gd

1 Yt 1 2ndt + gnd

=

where g d and gnd are the steady-state growth rates of

2.5

d t

vtd ,

(13)

and ndt respectively.

Foreign di¤erentiated inputs

Foreign di¤erentiated inputs xft (!) are produced by foreign …nal good with an one-to-one technology. The pro…t function is f t (!)

= pft (!)xft (!)

pz;t xft (!) = Ptf (Lft )1

[xft (!)]

pz;t xft (!).

(14)

The monopolistic price is pft (!) = min f ; 1= g pz;t , where < 1= . Once gain, due to incomplete patent protection , the monopolist cannot charge too high a price; otherwise, an imitator will produce xft (!). The amount of pro…t for ! 2 [0; nft ] is f t (!)

=(

1)pz;t xft (!)

1 Ptf Xtf

=

=

nft

1 Yt 2nft

where the second equality uses symmetry in (8), (10) and pft (!) = growth value of an invention is vtf (!) =

r

f t (!) f

g

1 Yt

=

2nft

where g f and gnf are the steady-state growth rates of

2.6

1 + f t

gnf

f t,

(15)

pz;t . The balanced-

vtf ,

(16)

and nft respectively.

R&D for domestic di¤erentiated inputs

The innovation process in the sector that uses domestic inputs is n_ dt = ktd Rtd ,

(17)

where Rtd denotes domestic R&D labor in sector d. The productivity of Rtd is given by ktd = ndt , which captures knowledge spillovers as in Romer (1990). Free entry yields n_ dt vtd = wt Rtd , ndt vtd = wt .

2.7

(18)

R&D for foreign di¤erentiated inputs

The innovation process in the sector that uses foreign inputs is n_ ft = ktf Rtf , 5

(19)

where Rtf denotes domestic R&D labor in sector f . The productivity of Rtf is given by R nf ktf = nft (1 + ft ), where ft = pz;t 0 t xft (!)d!=Yt is the value of imports (for producing di¤erentiated inputs) as a ratio to output. This speci…cation is consistent with Grossman and Helpman (1991) who also assume that knowledge spillovers arise from trade.8 Imposing symmetry and using (7) and (15), one can show that ft = =(2 ) and ktf = nft + nft = =2 and nft = captures an additional knowledge spillover e¤ect from imports. where In this case, patent protection reduces knowledge spillovers because a larger markup reduces the demand for imports. Thus, although entrepreneurs are able to appropriate foreign technologies, this foreign knowledge spillover e¤ect is decreasing in . Free entry yields n_ ft vtf = wt Rtf , (1 + = )nft vtf = wt .

2.8

(20)

Equilibrium labor allocation

The resource constraint on labor is Rtd + Ldt + Rtf + Lft = ltd + ltf = l, where ltd

Rtd + Ldt and lf

(21)

Rtf + Lft . Substituting (6), (9) and (13) into (18) yields Ld =

1 1

( + Rd ),

(22)

which together with (21) implies that steady-state equilibrium Rd is 1

Rd =

1

ld

,

(23)

where ld is endogenous. Substituting (7), (9) and (16) into (20) yields Lf =

1 1

1+ =

+ Rf

,

(24)

which together with (21) implies that steady-state equilibrium Rf is Rf =

1

1

lf

1+ =

,

(25)

where lf is endogenous. To solve for ld and lf , we use (6), (7) and (9) to obtain Lf = Ld ,

(26)

which together with (22) and (24) implies lf = 8

+

+ ld .

See Coe and Helpman (1995) for empirical evidence that trade a¤ects international spillovers.

6

(27)

Combining (21) and (27) yields ld ( ) = +

lf ( ) =

1 2

l

1 2

l+

+

+

,

(28)

,

(29)

which show that stronger patent protection leads to a reallocation of labor from sector f to sector d because suppresses knowledge spillovers from imports in sector f .

2.9

Equilibrium growth rates of technologies

The steady-state equilibrium growth rate of ndt is gnd

n_ dt = Rd ( ) = ndt

1

1

ld ( )

,

(30)

+

which is increasing in . Intuitively, stronger patent protection increases pro…t, which in turn increases R&D in sector d. Furthermore, this positive e¤ect is strengthened by the reallocation of resources from sector f to sector d. Proposition 1 summarizes this result. Proposition 1 The growth rate of technology in the sector that uses domestic inputs is increasing in patent protection . Proof. Use (30). The steady-state equilibrium growth rate of nft is gnf

n_ ft nft

= (1 + = )Rf ( ) = (1 + = )

1

lf ( )

1

,

(31)

which can be increasing or decreasing in patent protection . Intuitively, stronger patent protection increases pro…t, which is a positive e¤ect on R&D in sector f . However, stronger patent protection also has a negative e¤ect on knowledge spillovers and R&D in sector f . Furthermore, this negative e¤ect is strengthened by the reallocation of resources from sector f to sector d. Therefore, the overall e¤ect of patent protection on the growth rate of technology in sector f is ambiguous.9 Proposition 2 summarizes this result. Proposition 2 The growth rate of technology in the sector that uses foreign inputs can be increasing or decreasing in patent protection . Proof. Use (31). 9

See also Goh and Olivier (2002), Iwaisako and Futagami (2013) and Saito (2017), who explore other channels through which patent breadth has ambiguous e¤ects on innovation.

7

Finally, taking the di¤erence between the growth rates of nft and ndt yields gnf

gn

1

gnd =

2

l+

+

+

(32)

> 0.

+

The growth rate of technology is higher in sector f than in sector d due to the additional knowledge spillovers from imports in sector f . Furthermore, it can be shown that gn is …rstly increasing and eventually decreasing in . If we consider ! 0, then gn is explicitly an inverted-U function in . Proposition 3 summarizes these results. Proposition 3 The growth rate of technology is higher in the sector that uses foreign inputs than in the sector that uses domestic inputs. The di¤erence gn in the growth rates is …rstly increasing and eventually decreasing in patent protection . Proof. Use (32).

3

Empirical model

From Wooldridge (2006), the basic form of the logit model is10 Pi = P (yi = 1jZ i ) = F (Z i ; ) =

exp( 0 + Zi ) , 1 + exp( 0 + Zi )

(33)

where Pi is the probability of …rm i having an innovation. F (Zi ; ) is the cumulative distribution function of the logistic distribution. Manipulating (33), we obtain ln

Pi 1 Pi

=

0

+ Zi ,

(34)

where Zi denotes a vector of explanatory variables. We consider IPR and imported intermediate inputs as two main explanatory variables on innovation. To analyze how IPR a¤ects importing …rms’ innovation, we introduce an interaction term between IPR and imports. We specify our empirical model as follows: ln

Pit 1 Pit

=

0

+

1 IP Rpt

+

2 IN Tit

+

3 IN Tit

IP Rpt + Zit +

p

+

j

+

t

+ "it (35)

where ln(Pit =(1 Pit )) is the log odds of …rm i achieving innovation at time t and Pit = P (N EWit = 1jZit ). N EWit denotes innovation of …rm i in year t de…ned as whether …rm i produces new products in year t.11 If it does, then N EWit = 1; otherwise N EWit = 0. Explanatory variables include IP Rpt , IN Tit , IN Tit IP Rpt and other control variables Zit . IP Rpt denotes the log level of IPR in province p of China at time t. IN Tit is a dummy 10

Our …ndings in Table 1 and Figure 1 are robust to the probit model. Results are available upon request. We measure innovation of …rms by a dummy variable of new products to correspond to the theoretical model. From the law of large numbers, the probability of each …rm having a new product corresponds to the growth rate of products in its sector. 11

8

variable of whether …rm i imports intermediate inputs in year t. If it does, then IN Tit = 1; otherwise IN Tit = 0. p is the province …xed e¤ect. j is the industry …xed e¤ect. t is the year …xed e¤ect. "it is the error term. f 1 ; 2 ; 3 g respectively capture the e¤ects of IPR on innovation, knowledge spillovers from imports, and the interaction between IPR and imports. First, 1 captures the direct e¤ect of IPR on innovation, which corresponds to Proposition 1. According to Proposition 1, 1 should be positive indicating that IPR has a positive e¤ect on innovation. Second, 2 captures whether importing …rms bene…t from knowledge spillovers. If importing …rms bene…t from knowledge spillovers, then 2 should be positive. Third, 3 captures the e¤ect of IPR on knowledge spillovers of importing …rms, which corresponds to Proposition 2. According to Proposition 2, 3 should be negative indicating that IPR hinders knowledge spillovers from imports. Finally, Proposition 3 implies P (N EWit = 1jIN Tit = 1; IP Rpt ) > P (N EWit = 1jIN Tit = 0; IP Rpt );

(36)

i.e., importing …rms have better innovation performance than non-importing …rms. Other explanatory variables Zit include the proportion of …rm exports in total output (EXP ), foreign capital share (F OR), the log of …rm age (AGE), the log of total factor productivity (T F P ) and the log of …rm size measured by employment (SIZE) that may a¤ect enterprise innovation. In addition, we also control for the log of per capita GDP at the provincial level (IN COM E) and the log of the Her…ndal index computed from the 4-digit Chinese Industry Classi…cation system (HERF ). With the entry of China to the WTO in 2001 and the requirements of the TRIPS Agreement, the strengthening of IPR in China during that period has an exogenous nature. Therefore, we consider data from 2000 to 2007.12 We present a description of the data and summary statistics in the appendix.

4

Regression results

Table 1 shows the regression results. From columns (1) to (3), IPR contributes to innovation at 10% signi…cance level whereas imports IN T contribute to innovation at 1% signi…cance level. In column (4), we include an interaction term between IPR and imports of intermediate inputs. Two …ndings emerge. First, the coe¢ cients of IPR and imports have the same sign but become more signi…cant compared to those in columns (1) to (3). Second, the interaction coe¢ cient is negative at 5% signi…cance level, which means that IPR has a signi…cant negative e¤ect on importing …rms’innovation. These results are consistent with Propositions 1 and 2. 12

We consider data up to 2007 due to the incompatibility of data in the Chinese industrial enterprises database from 2008 onwards.

9

Table 1: Regression results Variable IP R

N EW (1) 2:342 (1.274)

IN T IN T

(2)

0:475 (0.090)

(3) 2:346 (1.278) 0:476 (0.090)

0:505 (0.054) -0.022 (0.151) 0:425 (0.141) 0:191 (0.036) 0:168 (0.037) 1.268 (2.368) 0:131 (0.018) -17.804 (23.653) yes yes yes 883793

0:504 (0.054) -0.017 (0.151) 0:432 (0.142) 0:199 (0.035) 0:164 (0.037) 1.764 (2.716) 0:132 (0.017) -25.700 (28.121) yes yes yes 883793

IP R

SIZE EXP F OR TFP AGE IN COM E HERF Constant Province …xed e¤ect Industry …xed e¤ect Year …xed e¤ect Observations

0:543 (0.060) 0.136 (0.166) 0:248 (0.124) 0:212 (0.035) 0:166 (0.037) 1.769 (2.710) 0:135 (0.017) -25.914 (28.033) yes yes yes 883793

(4) 2:805 (1.290) 3:038 (1.159) 1:738 (0.779) 0:504 (0.054) -0.009 (0.154) 0:421 (0.138) 0:198 (0.035) 0:165 (0.037) 1.672 (2.664) 0:132 (0.018) -25.447 (27.793) yes yes yes 883793

N otes: The brackets are the standard errors clustering at the province level. Signi…cant at *10%, **5% and ***1%.

10

From column (4), we calculate the probability of innovation at each level of IPR. As shown in Figure 1,13 with the improvement of IPR, importing and non-importing …rms both experience higher innovation probability, but the innovation probability of importing …rms is higher than that of non-importing …rms. However, with stronger IPR, the gap between the two types of …rms …rst widens and eventually narrows due to the negative e¤ect of IPR on knowledge spillovers from imports. These …ndings are consistent with Proposition 3.

Figure 1: Firm innovation probabilities computed from the logit model

5

Conclusion

This study develops a small-open-economy R&D-based growth model to explore the di¤erent e¤ects of IPR on the innovation of importing and non-importing …rms. We test our theoretical results from the model using an empirical analysis of matching samples that combine Chinese provincial IPR data with industrial enterprises database and customs database. Our study shows that IPR has an overall positive e¤ect on innovation in China. Furthermore, importing …rms have better innovation performance than non-importing …rms; therefore, the government should continue to encourage international trade. 13

Stata’s margins command is used to calculate the marginal e¤ect of IPR on innovation probability. IPR is measured in log.

11

6

Appendix

6.1

Intellectual property rights

Shen (2010) use the method in Ginarte and Park (1997) to construct an annual measure of intellectual property rights (IPR) at the country level in China. We use his data from 2000 to 2007. As for the level of IPR in each province, we use information on the level of administrative protection and the level of judicial protection as follows. First, we use two indicators to measure IPR at the administrative level. (1) The importance of provincial government’s emphasis on IPR. As in Ang et al. (2014), we use the number of articles on the protection of IPR in the newspapers of provincial authorities divided by the total number of articles in the newspapers of each province as a measure of this index. The higher the index, the more emphasis on the protection of IPR by the provincial government. (2) The degree of administrative protection of provincial patent o¢ ces. As in Wu and Tang (2016), we use the annual number of patent disputes as a ratio to the cumulative number of patent licenses in China Intellectual Property Rights Yearbook (2001-2008)14 to calculate the administrative protection level of the State Intellectual Property O¢ ce in each of the 31 provinces in China (equal to one minus the ratio of the number of annual patent disputes to the cumulative number of patents granted). A larger value of the indicator is associated with more e¤ective administrative protection by the provincial patent authority. Second, we measure the provincial intellectual property judicial protection by two indicators. (1) Provincial judicial protection situation. Data on the protection of producer rights is from Fan et al. (2011). Based on the fairness of law enforcement and the e¢ ciency of law enforcement agencies, this indicator measures the legal environment in each province in di¤erent years. (2) Whether the courts take the “three-in-one" trial in intellectual property cases. If the courts at all levels in a province have announced the “three-in-one" trial in intellectual property cases in a given year, then the variable is set to 1 for the year and beyond, otherwise 0. We consider this variable because China’s intellectual properties are protected by both administrative protection and judicial protection. Wang and Lv (2016) consider IPR in Guangdong province and …nd that the “three-in-one" trial model has a signi…cant role in promoting …rm innovation by improving the quality and e¢ ciency of trials in courts. Ginarte and Park (1997) measure IPR by taking the arithmetic average of IPR subindicators to compute their aggregate index. However, the arithmetic mean may not fully re‡ect the di¤erence in the relative importance of the IPR sub-indicators. Wu and Tang (2016) use principal component analysis to measure the enforcement of IPR in Chinese provinces. Principal component analysis converts a number of related indicators into a representative comprehensive indicator by dimensionality reduction. We use this method to synthesize the national protection of IPR, the provincial administrative enforcement and the provincial judicial protection to form our provincial IPR index. The eigenvalues of our …ve principal components are 1.7702, 1.0297, 0.9325, 0.7078 and 0.5598. Given that the …rst three principal components have accumulated 74.7% of the information, we construct the IPR index of each province with the …rst three principal components. This IPR index at the provincial level eliminates the overlap of information between intellectual property legislative 14

Data in 2000 is based on the annual statistical report of the State Intellectual Property O¢ ce.

12

protection at the national level, administrative enforcement and judicial protection at the provincial level.

6.2

The matching procedure of industrial enterprises database and customs database

First, we follow Brandt et al. (2012) in cleaning up the Chinese industrial enterprises database and constructing panel data. Second, we match Chinese industrial enterprises panel data and customs database. We begin by matching …rm name, followed by zip code and the last seven digits of phone number, and …nally zip code and legal person name. Accordingly, we combine industrial enterprise panel data with custom information. Then, in order to unify the import product information, we reduce product data from HS 8-digit to HS 6-digit. Also, we unify the annual HS code to HS1996 standard according to BEC classi…cation to identify and calculate import information of intermediate products at the …rm level. We delete observations with missing data on total assets, employments, total output, total assets less than the current assets, …xed assets, or accumulated depreciation less than the depreciation of current year. We follow Brandt et al. (2012) and Kee and Tang (2016) to exclude …rms with less than 8 employees, age less than one and the total value of imported intermediates greater than the total amount of intermediate inputs. Finally, we match IPR data with industrial …rms-customs matched panel based on the province-year dimension. The resulting panel includes data at the …rm-level from the manufacturing sector (i.e., classi…cation code 13 to 42 from the 2-digit Chinese Industry Classi…cation system) in China’s 31 provinces in year 2000-2007, except for 2004 because data on new products in 2004 is missing.

6.3

Summary statistics

Table 2 provides the summary statistics of the variables used in the empirical analysis. Table 2: Summary statistics Va ria b le NEW IN T IP R IP P 1 IP P 2 IP P 3 IP P 4 IP P 5 S IZ E EXP FO R TFP AG E IN C O M E HERF

D e …n itio n A d u m m y va ria b le o f n e w p ro d u c ts A d u m m y va ria b le o f im p o rtin g inte rm e d ia te in p u ts T h e lo g o f inte lle c tu a l p ro p e rty rig hts in d e x -p rin c ip a l c o m p o n e nt a n a ly sis Inte lle c tu a l p ro p e rty le g isla tive p ro te c tio n A d m in istra tive p ro te c tio n o f p rov in c ia l p a te nt o ¢ c e s P rov in c ia l g ove rn m e nt’s e m p h a sis o n inte lle c tu a l p ro p e rty p ro te c tio n T h e p ro te c tio n o f p ro d u c e r rig hts fro m Fa n et al.(2 0 1 1 ) A d u m m y va ria b le o f “th re e -in -o n e " tria l T h e lo g o f …rm siz e m e a su re d by e m p loy m e nt T h e p ro p o rtio n o f …rm e x p o rts in to ta l o u tp u t Fo re ig n c a p ita l sh a re T h e lo g o f to ta l fa c to r p ro d u c tiv ity T h e lo g o f …rm a g e T h e lo g o f p e r c a p ita G D P a t th e p rov in c ia l le ve l T h e lo g o f th e H e r…n d a l in d e x

O b s. 883793 883793 883793 883793 883793 883793 883793 883793 883793 883793 883793 883793 883793 883793 883793

M ean 0 .0 8 9 9 0 .1 4 2 1 1 .4 5 3 3 4 .3 6 1 4 0 .9 9 8 7 0 .0 0 5 6 5 .2 4 2 4 0 .3 3 8 8 4 .8 0 0 8 0 .1 5 6 6 0 .0 7 0 4 3 .7 2 8 9 1 .9 0 8 5 9 .7 9 4 9 -4 .4 8 0 8

S td . D e v . 0 .2 8 6 1 0 .3 4 9 2 0 .1 1 9 5 0 .3 2 9 1 0 .0 0 1 1 0 .0 0 4 4 2 .0 7 6 7 0 .4 7 3 3 1 .0 4 6 8 0 .3 3 0 0 0 .2 3 7 6 0 .9 8 3 8 0 .9 3 5 0 0 .5 8 3 3 1 .1 1 8 3

M in 0 0 -0 .9 2 8 2 3 .4 0 .9 6 9 7 0 -0 .4 6 0 2 .0 7 9 4 0 0 -6 .5 2 4 7 0 7 .9 1 6 5 -6 .5 8 1 9

N otes: S a m p le is lim ite d to sta te -ow n e d e nte rp rise s a n d a ll o th e r …rm s w ith sa le s a b ove 5 m illio n R M B in th e m a nu fa c tu rin g se c to r.

13

M ax 1 1 1 .7 0 0 3 4 .5 3 1 0 .0 1 6 8 10 1 1 2 .1 4 5 1 1 9 .7 2 8 4 .6 7 2 8 1 1 .0 1 0 9 -0 .1 2 2 1

References [1] Ang, J., Y. Cheng, and C. Wu, 2014. Does enforcement of intellectual property rights matter in China? Evidence from …nancing and investment choices in the high-tech industry. Review of Economics and Statistics, 96, 332-348. [2] Bessen, J., and M. Meurer, 2008. Patent Failure: How Judges, Bureaucrats, and Lawyers Put Innovators at Risk. Princeton University Press. [3] Boldrin, M., and D. Levine, 2008. Against Intellectual Monopoly. Cambridge University Press. [4] Brandt, L., J. Van Biesebroeck, and Y. Zhang, 2012. Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97, 339-351. [5] Chen, Y., and T. Puttitanun, 2005. Intellectual property rights and innovation in developing countries. Journal of Development Economics, 78, 474-493. [6] Chu, A., and S. Pan, 2013. The escape-infringement e¤ect of blocking patents on innovation and economic growth. Macroeconomic Dynamics, 17, 955-969. [7] Coe, D., and E. Helpman, 1995. International R&D spillovers. European Economic Review, 39, 859-887. [8] Cozzi, G., and S. Galli, 2014. Sequential R&D and blocking patents in the dynamics of growth. Journal of Economic Growth, 19, 183-219. [9] Fan, G., X. Wang, and H. Zhu, 2011. NERI INDEX of Marketization of China’s Provinces 2011 Report. (Chinese) Economic Science Press. [10] Furukawa, Y., 2010. Intellectual property protection and innovation: An inverted-U relationship. Economics Letters, 109, 99-101. [11] Ginarte, J. C., and W. G. Park, 1997. Determinants of patent rights: A cross-national study. Research Policy, 26, 283-301. [12] Goh, A.-T., and J. Olivier, 2002. Optimal patent protection in a two-sector economy. International Economic Review, 43, 1191-1214. [13] Goldberg, P., A. Khandelwal, N. Pavcnik, and P. Topalova, 2010. Imported intermediate inputs and domestic product growth: Evidence from India. Quarterly Journal of Economics, 125, 1727-1767. [14] Grossman, G., and E. Helpman, 1991. Trade, knowledge spillovers, and growth. European Economic Review, 35, 517-526. [15] Grossman, G., and E. Lai, 2004. International protection of intellectual property. American Economic Review, 94, 1635-1653. 14

[16] Hu, A., and I. Png, 2013. Patent rights and economic growth: Evidence from crosscountry panels of manufacturing industries. Oxford Economic Papers, 65, 675-698. [17] Huang, W., C. Lai, and P. Chen, 2017. International R&D funding and patent collateral in an R&D-based growth model. International Review of Economics and Finance, 51, 545-561. [18] Iwaisako, T., and K. Futagami, 2013. Patent protection, capital accumulation, and economic growth. Economic Theory, 52, 631-668. [19] Ja¤e, A., and J. Lerner, 2004. Innovation and Its Discontents: How Our Broken System Is Endangering Innovation and Progress, and What to Do About It. Princeton University Press. [20] Judd, K., 1985. On the performance of patents. Econometrica, 53, 567-586. [21] Kee, H. L., and H. Tang, 2016. Domestic value added in exports: Theory and …rm evidence from China. American Economic Review, 106, 1402-1436. [22] Lai, E., 1998. International intellectual property rights protection and rate of product innovation. Journal of Development Economics, 55, 133-53. [23] Li, C.-W., 2001. On the policy implications of endogenous technological progress. Economic Journal, 111, 164-179. [24] Lin, H., 2015. Creative destruction and optimal patent life in a variety-expansion growth model. Southern Economic Journal, 81, 803-828. [25] Naghavi, A., and C. Strozzi, 2015. Intellectual property rights, diasporas, and domestic innovation. Journal of International Economics, 2015, 96, 150-161. [26] Nordhaus, W., 1969. Invention, Growth, and Welfare: A Theoretical Treatment of Technological Change. The MIT Press. [27] Romer, P. 1990. Endogenous technological change. Journal of Political Economy, 98, 71-102. [28] Saito, Y., 2017. E¤ects of patent protection on economic growth and welfare in a twoR&D-sector economy. Economic Modelling, 62, 124-129. [29] Shen, G., 2010. Nominal level and actual strength of China’s intellectual property protection under TRIPS agreement. Journal of Chinese Economic and Foreign Trade Studies, 3, 71-88. [30] Wang, H., and T. Lv, 2016. Judicial protection of intellectual property and enterprise innovation: Quasi-nature experiment based on Guangdong province intellectual property case. (Chinese) Management World, 10, 118-133. [31] Wooldridge, J., 2006. Introductory Econometrics: A Modern Approach (3rd ed.). Ohio: Thomson/ South-Western, pp. 596-597. 15

[32] Wu, C., and D. Tang, 2016. Intellectual property rights enforcement, corporate innovation and operating performance: Evidence from China’s listed companies. (Chinese) Economic Research Journal, 11, 125-139. [33] Yang, Y., 2013. Optimal patent policy, research joint ventures, and growth. Economics Letters, 118, 381-384. [34] Zeng, J., J. Zhang, and M. Fung, 2014. Patents and price regulation in an R&D growth model. Macroeconomic Dynamics, 18, 1-22.

16