patent law harmonization and international trade · patent law harmonization and international...

21
Patent Law Harmonization and International Trade Banri Ito Aoyama Gakuin University, 4-4-25, Shibuya, Shibuyaku, Tokyo, 150-8566, Japan August 8, 2016 Abstract Global harmonization of intellectual property rights is one of the major challenges in multilateral or regional trade negotiations. This study empirically examines the relation between harmonization of patent rights system and international trade ow based on world bilateral trade data during 1995-2005. Aside from methods of previous studies on this topic, this study uses a structural gravity model based on a translog demand system where trade cost elasticity is endogenously determined. The results reveal that the harmonization of patent system measured by the di/erence in the Index of Patent Rights between exporter and importer is negatively correlated with trade share of patent-sensitive industries while there is no correlation in other industries. This result is robust to the model with additional variables that a/ect trade costs. Keywords : Patent Law; International trade; Gravity equation JEL Classifcations : F14, F15, O34 Email: [email protected] Tel: +81-3-3409-3934 Fax: +81-3-5485-0698 1

Upload: nguyenphuc

Post on 06-Aug-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

Patent Law Harmonization and International Trade

Banri Ito�

Aoyama Gakuin University, 4-4-25, Shibuya, Shibuyaku, Tokyo, 150-8566, Japan

August 8, 2016

Abstract

Global harmonization of intellectual property rights is one of the major challengesin multilateral or regional trade negotiations. This study empirically examines therelation between harmonization of patent rights system and international trade �owbased on world bilateral trade data during 1995-2005. Aside from methods of previousstudies on this topic, this study uses a structural gravity model based on a translogdemand system where trade cost elasticity is endogenously determined. The resultsreveal that the harmonization of patent system measured by the di¤erence in the Indexof Patent Rights between exporter and importer is negatively correlated with tradeshare of patent-sensitive industries while there is no correlation in other industries.This result is robust to the model with additional variables that a¤ect trade costs.

Keywords: Patent Law; International trade; Gravity equation

JEL Classifcations: F14, F15, O34

�Email: [email protected] Tel: +81-3-3409-3934 Fax: +81-3-5485-0698

1

1. Introduction

Strengthening protection of Intellectual Property Rights (IPR) is one of the central issue in

free trade negotiations. The agreement on Trade-Related Aspects of Intellectual Property

Rights (TRIPS) in 1994 determined the minimum standard of IPR protection and strength-

ened enforcement procedures backed by the dispute settlement system in WTO. It is expected

that the harmonization of IPR protection accelerate international trade, but there has been

con�ict between developing countries which pursue access to technology and knowledge and

developed countries which favor reinforcement of IPR protection. As a result, this confronta-

tion has made multilateral or regional free trade negotiations little progress. For it remains

even unclear as to whether the harmonization of IPR system spur international trade. There-

fore, policy makers and researchers have long been interested in the impact of global IPR

reform on international trade. This study provides empirical evidence to the debate regarding

how IPR harmonization is related with trade by estimating a structural gravity model.

This paper contributes to the literature that explores the link between IPR protection and

trade (Maskus and Penabarti, 1995; Smith, 1999; Ra�quzzaman, 2002; Ivus, 2010; Awokuse

and Yin, 2010). Unlike previous studies on this topic, the main feature of this study is

to focus on not the strength of IPR protection in importer but the di¤erence in IPR law

between country of origin and destination. Although most studies introduce the strength of

IPR protection in importing countries into the right hand side of gravity equation, it means

that the proportionate e¤ects of a change in IPR protection of importer on bilateral trade

�ows are the same for its all trading partners. This implicit assumption is not likely to hold

when exporters decide to sell their products considering both levels of IPR protection in

2

origin and destination countries. Nevertheless, the di¤erence in IPR law has not been paid

attention to till today. Existing theoretical conjecture between IPR protection and trade is

that strengthening IPR protection in an importing country brings opposite e¤ects on bilateral

trade �ows. These are known as market expansion e¤ects that induces imports from abroad

and market power e¤ects that discourage imports due to strengthened monopolistic power

(Maskus and Penubarti, 1995). This existing framework can be useful to provide insights

on how the disparity in IPR law is related with bilateral trade �ows. Speci�cally, in the

case where IPR protection in destination country is weaker than origin, exported products

will be exposed to a high risk of losing pro�t by counterfeit goods and imitation of their

technologies due to reverse engineering. On the other hand, stronger IPR protection in

destination country than origin country would also hinder exports to the destination as it

implies that monopolistic power of �rms in the destination country is stronger than the origin

country. The market expansion e¤ects are expected to be found in the case where IPR law

is harmonized between bilateral trading partners. As far as the author knows, there has

been no study that intended to exploit the link between the di¤erence in IPR protection and

bilateral trade �ows.

Another feature of this study is characterized by applying a structural gravity model based

on micro foundation to the estimated equation for bilateral trade �ows while most previous

studies rely on nonstructural gravity equations except Awokuse and Yin (2010). Gravity

equations are one of the hot research �elds and recent progress in developing micro-founded

models is signi�cant as seen in the model based on the Ricardian framework (Eaton and

Kortum, 2002), the model that incorporates the conceptual framework of multilateral resis-

tance (Anderson and van Wincoop, 2003), and the model with heterogeneous �rms (Chaney,

3

2008; Helpman et al., 2008; Melitz and Ottaviano, 2008). Based on micro-founded theory,

these models derive structural gravity models with constant trade elasticity. In addition, the

feature of constant elasticity of trade costs is recently relaxed by Novy (2013) that derives a

structural gravity model by applying the general equilibrium translog model into the demand

system. The translog preferences as represented by Diewart (1976) and Feenstra (2003) are

more �exible than constant elasticity of substitution (CES) type preferences, and generate

the gravity model in which trade costs elasticity is no longer constant. More speci�cally,

trade costs elasticity di¤ers according to bilateral trade intensity between the two countries.

In this paper, as the di¤erence in IPR law is assumed to be a trade distortional factor as well

as distance, it follows that the e¤ect of the IPR harmonization varies depending on trade

share in bilateral trade �ows.

The results from the estimation of the structural gravity model reveal that the harmo-

nization of IPR law measured by the di¤erence in the Index of Patent Rights (Park, 2008)

between exporter and importer is negatively correlated with bilateral trade share as pre-

dicted. However, this signi�cant sign is observed in only patent-sensitive industries such as

pharmaceuticals, medical chemicals and medical equipment, which is consistent with earlier

�ndings.

The remainder of this paper is organized as follows. Section 2 presents related studies on

this topic. Section 3 elaborates on theoretical framework and derives the estimatable gravity

equation. Section 4 explains data used in the estimation for the gravity model. Section 5

presents the estimation results of the structural gravity model. Section 6 summarizes and

concludes the study.

4

2. Intellectual property rights and international trade

Strengthening protection of intellectual property rights has both positive and negative im-

pacts on trade. A series of earlier theoretical studies on this topic have argued that the

e¤ects of patent protection in destination countries are inde�nite in the sense that exports

to the countries possibly increase or decrease owing to market expansion e¤ect and market

power e¤ect (Maskus and Penabarti, 1995; Smith, 1999). Previous empirical studies have

focused on the e¤ects of patent protection on bilateral trade, particularly between developed

countries and developing countries. For example, Maskus and Penabarti (1995) found that

there is positive correlation between OECD countries�exports and the strength of patent

protection in destination countries and this relation is especially signi�cant in developing

countries. Smith (1999) who examines the e¤ect of strengthening patent protection in for-

eign countries on U.S. exports reports that U.S. exports increases when destination country

with high threat of imitation strengthens patent protection while it increases in countries

where the threat of imitation is weak. In addition, Awokuse and Yin (2010) examines the

e¤ect of patent protection in China on imports from 36 countries and con�rms its positive

e¤ect that accords with the prior results from the perspective of developed countries. These

previous �ndings suggest that developing countries with room for improvement of patent

system cause market expansion e¤ect by strengthen patent protection whereas developed

countries where imitation risk is low are likely to result in market power e¤ect.

Although most studies have focused on the strength of intellectual property rights protec-

tion in importing countries and observed the market expansion e¤ect in developing countries

in particular, the present study sheds light on the role of the di¤erence in patent protection

5

between origin countries and destination countries in trade among developed countries. For

patent-sensitive products are traded intensively among OECD countries when traded prod-

ucts are divided into patent-sensitive and -insensitive products. Table 1 shows the share of

trade volume over world trade in 2010 for both patent-sensitive products and -insensitive

products.1 Obviously, the share of trade volume for patent-sensitive products is signi�cant

rather than patent-insensitive products, accounting for 61% of the total volume. Neverthe-

less, strengthening patent protection is unlikely to spur bilateral trade according to previous

�ndings that the e¤ect of strengthened patent protection in countries with weak imitation

risk is considered to be marginal due to market power e¤ect while exports from developed

countries to developing countries can be enhanced by the market expansion e¤ect. This

gap raises a question as to whether strengthening patent protection in developed countries

increase bilateral trade between them which is dominant for patent-sensitive products. To

reply to the question, the present study intends to uncover how the harmonization of patent

protection in�uences trade among countries. This paper argues that the both market expan-

sion and market power e¤ects are likely to be in�uenced by not only the strength of patent

protection in destination counties but also that in origin countries. In that case, the distance

in terms of patent system between exporters and importers is expected to be associated with

bilateral trade rather than the extent of patent protection in importers. Even if the patent

law of destination country with low imitation risk is strengthened, the market power e¤ect

may not be necessarily dominant when patent law is harmonized with that of origin country.

On the contrary, it reminds us suggestive of the market expansion e¤ect. However, when

patent protection in destination country is stronger than origin country, exports from the

1The de�nition is explained in Section 5.

6

origin to the destination may slow down owing to the dominance of monopolistic power of

�rms in the destination country. The market expansion e¤ects are expected to be observed

in the case when patent law is closely harmonized between trading partners. As far as the

author knows, very few attempts have been made at such study on institutional distance as

a determinant of bilateral trade in previous literature of gravity models. The exception is

De Groot et al. (2004) who examine the e¤ect of institution on trade �ow using Worldwide

Governance Indicators as a measurement of institutional quality. It is found that having

a similar institutional framework between exporter and importer increases bilateral trade.

In this respect, the present study extends this kind of analysis to the e¤ect of institutional

distance in terms of patent law on trade �ow.

3. Gravity equation under translog demand system

This paper builds on the gravity framework by Anderson and van Wincoop (2003). Specif-

ically, the empirical speci�cation of this study is based on Novy (2013), which derives a

gravity model using the homothetic translog demand system, to explain bilateral trade vol-

umes. Incorporating the translog demand system into gravity framework allows for richer

substitution patterns across varieties owing to its �exibility, compared to CES demand sys-

tem. This feature results in endogenous elasticity of trade with respect to trade costs in

the gravity model (Novy, 2013). Following Diewart (1976) and Feenstra (2003), the translog

expenditure function for country j is expressed as follows:

lnEj = ln (Uj) + �0j +NXl=1

�l ln (plj) +1

2

NXl=1

NXm=1

lm ln (plj) ln (pmj) (1)

7

where Uj is the utility level of contry j with goods l andm. The derivative of the expenditure

share with respect to ln (plj) is expressed as follows:

d lnEjd ln(plj)

= slj = �l +

NXm=1

lm ln (pmj) : (2)

The share of import from country i to country j is descibed as follows:

xijyj=

NXl=Ni�1+1

slj =NX

l=Ni�1+1

�l +

NXm=1

lm ln (pmj)

!(3)

Imposing the market-claearing condition yi =JPi=1

xij, the equation can be rewritten as follows:

xijyj=yiyw� ni ln (tij) + ni ln (Tj) + ni

JXk=1

ykywln

�tikTk

�(4)

where yw is world income and ni is the number of goods of country i, which is de�ned as

ni � Ni � Ni�1. ln (Tj) is a weighted average of logarithmic trade cost and that is the

multilateral resistance term in Anderson and van Wincoop (2003).

ln (Tj) =1

N

NXm=1

ln (tmj) =

NXk=1

nkNln (tkj) (5)

It should be noted that the multilateral resistance term varies across both exporters i and

importers j while the last term varies across only exporters i. In addition, the �rst term in

the right hand side of Eq. (4) is exporter�s attribute, and therefore, in the estimation, the

both �rst and last terms are rewritten as exporter �xed e¤ect �i.

8

�i �yiyw+ ni

JXk=1

ykywln

�tikTk

�(6)

This study follows the literature in this area by modeling the trade cost factor ln tij as

a log-linear function of obseravable trade cost proxies. Speci�cally, ln tij is assumed to be

a function of geographic distance and institutional distance in terms of patent system as

follows:

ln tij = � ln (dist ij) + � (iprij) (7)

where � is distance elasticity of trade costs and � is the coe¢ cient of patent law harmonization

between exporters and importers. The multilateral resistance term can be expressed by the

geographical distance and the patent system distance as lnTj = � lnT distj + �T iprj . From Eq.

(5), the right hand side terms are di�ned as follow:

lnT distj �NXk=1

nkNln (distkj) ; T iprj �

NXk=1

nkN(iprkj) : (8)

Subsituting Eqs. (6)-(8) to Eq. (4), the gravity equation to be estimated is rewritten as:

xijyj= �i � � ni ln (dist ij)� � ni

�ipr ij

�+ � ni ln

�T distj

�+ � ni

�T iprj

�+ "ij: (9)

As a more simple speci�cation, Eq. (9) can be rewritten as the following equation by

9

dividing the measurement of exporters�extensive margin ni.

xijyjni

= �� ln (dist ij)� � �ipr ij

�+ b�i + b�j + vij: (10)

where vij is the error term. b�i denotes �ini that is captured by the exporter �xed e¤ect while b�jdenotes an importer �xed e¤ect de�ned as b�j � � ln �T distj

�+ � ln

�T iprj

�. To complement

the results from Eq. (9), the alternative speci�cation from Eq. (10) is also estimated. The

translog gravity model has a unique feature in the sense that trade cost elasticities are not

constant contrary to standard gravity models based on CES demand function (Novy, 2013).

The trade cost elasticity is derived from Eq. (9) as � = � ni��xijyj

�since it is calculated

as @ ln (xij=yj)�@ ln � ij. The elasticity di¤ers according to the level of trade intensity.

4. Data

The present study uses data of bilateral trade volumes among 119 countries retrieved from

BACI which is the database at the HS 6-digit product-level provided by CEPPI. Regarding

the trade cost factor, the key variable is the measurement of patent system harmonizaiton

between exporter and importer. In this study, the index of patent rights (IPR) is employed

to measure the extent of patent system strength, which has been conducted by Park (2008).

IPR is constructed as the numerical average of the scores for the following �ve categories

pertaining to the protection of patent rights: (1) the coverage of patentability for major

industries, including pharmaceuticals, chemicals, and food; (2) the duration of patent rights;

(3) the strictness of the legal enforcement; (4) the rati�catioons of international agreements

associated with patent protection; and (5) the existence of policies that undermine the im-

10

plementation of patent rights. An index having a higher score represents a country that has

a higher level of patent protection. As IPR is available every 5 years, this study uses the

data for trade volume for the year 1995, 2000, 2005 and 2010. Hence, the size of sample is

30�29�4=3480. The distance of patent system between exporter and importer iprij in Eq.7

is de�ned as the absolute value of di¤erence in IPR beween exporter and importer.

The patent sensitivity is likely to alter the e¤ect of the distance of patent system on

birateral trade. In fact, Ivus (2015) reports that the e¤ect of patent reform on bilateral

trade volume is more pronouced for patent-sensitive products. The distance of patent sys-

tem between exporter and importer is expected to be strongly correlated with biratera trade,

particularly in the patent-sensitive industries. To compare the results according to sensitiv-

ity of patent protection across products, the sample is splitted into two: patent-sensitive

insudtries and patent-insensitive industries. Following Ivus (2015), the present study divides

birateral trade into two groups; patent-sensitive and -insensitive industries using Cohen et

al. (2000) which survey e¤ectiveness of appropriability machanisms for product innovations

at the industry level. First, all products at the HS 6-digit level are class�ed into industries

at the ISIC rev.3 level and then industries where the patent e¤ectiviness score for product

innovations surveyed by Cohen et al. (2000) is higher than the average value of all industries

are de�ned as patent-sensitive industries.2 Geographical distance dist ij as a proxy for trans-

port costs is also controlled as well as contiguousness between exporter and importer. These

information are retrieved from CEPPI database. Data of GDP as a proxy for income level is

2Speci�cally, they are the following 14 industries where the patent e¤ectiveness score is higher than theaverage value, in descending order; 3311 Medical Equipment, 2423 Drugs, 2920 Special Purpose Machinery,3430 Autoparts, 3010 Computers, 2429 Miscellaneous Chemicals, 2800 Metal Products, 3410 Car/Truck,2411 Basic Chemicals, 2910 General Purpose Machinery, 3230 TV/Radio, 3400 Chemicals, 2100 Paper, 2922Machine Tools (Cohen et al., 2000).

11

collected from IMF International Financial Statistics.

For estimating the gravity model in Eq. (9), it is necessary to measure the number of

goods of exporting country ni. Following Novy (2013), the measurement for the extensive

margin proposed by Hummels and Klenow (2005) is employed as a proxy for ni. The extensive

margin is de�ned as follow:

EMij =

Xs2Iij

xkjsXs2Ixkjs

(11)

where Iij is the set of observable product categories where country i has a positive exports

to country j, and k is a reference country that has positive exports to country j in all

product categories I. Hummels and Klenow (2005) calculate the index for 126 countries to

59 importing countries in 1995, based on 5,017 product categories. The reference country

is set as rest-of-world, and therefore EMij indicates the share of rest-of-world�s exports to

country j for products where country i has a positive exports to country j over rest-of-world�s

exports to country j for all products. Taking the geometric mean of country i�s EMij across

all destinations, the extensive margin of country i is expressed as follow:

EMi =Yj2M�i

(EMij)aij (12)

where aij is the weight calculated as the logarithmic mean of the share of country j in the

overall exports of country i. The present study covers aproximately 5,000 products categories

at the HS 6-digit product codes and bilateral trade among 230 countries for the year 1995,

2000, 2005 and 2010. For example, the calculated values of EMi for the 30 OECD countries

in the sample for the gravity equation are presented in Appendix Table. Although the size

12

of sample used for the calculation is di¤erent from that of Hummels and Klenow (2005), the

calculated values are in line with those reported by them.3

5. Estimation results

5.1. Results from translog gravity model

Table 2 displays the basic results from estimation for Eq. (9). The results show the estimated

coe¢ cients from OLS and robust standard errors clustered at the country pair in brackets

to account for the correlation within a country pair. Columns (1) and (2) report the results

based on aggregated trade volume while columns (3)-(5) correspond to the estimation using

the two separated samples in terms of patent sensitiveness. All estimations include exporter

�xed e¤ects and year �xed e¤ects though the results are supressed. To take into account

possible price e¤ects for a country in a speci�c year, columns (3) and (5) report the results

of estimation in which dummy variables for exporter and year are introduced.

Table 2 around here

As expected, the coe¢ cient of ni ln (dist ij) is statistically signi�cant at the 1% signi�cance

level and negative for all the estimations. On the other hand, the results on niipr ij from the

aggregated sample show insigni�cant signs as reported in columns (1) and (2). In contrast,

once the sample is splitted into the two: patent-sensitive and -insensitive, the introduction

of the interaction term of niipr ij and the dummy for patent-sensitive industries generates

the contrastive result that the di¤erence in IPR is negatively correlated with trade share

3For example, Hummels and Klenow (2005) report that the extensive margin index in 1995 is 0.912 forthe U.S., 0.725 for Japan, and 0.786 for Germany while that is 0.232 for New Zealand, 0.163 for Chile, and0.054 for Iceland.

13

of patent-sensitive products while it is positively correlated with the rest of products. This

result indicates that patent law harmonization is important for patent-sensitive products

as predicted. It is remarkable that the coe¢ cients of niipr ij for both patent-sensitive and

-insensitve industries are still signi�cant at the 1% level even after the dummy variables for

contiguous countries, countries with a common language, and exporter-year �xed e¤ects are

included in the model. The signs of contiguous countries dummy and common language

dumy are positive, which are intuisively plausible results.

Dividing the measurement of exporters� extensive margin ni in Eq. (9) leads to the

alternative speci�cation presented in Eq. (10). Table 3 reports the results from the model in

Eq. (10), whereas, in columns (4)-(6), the restriction that all countries have one product (i.e.

ni = 1) is imposed. The results of dummy variables for exporter �xed e¤ects and importer

�xed e¤ects are suppressed to save space. The main result is compatible with the results in

Table 2. What is noteworthy is that the e¤ects of patent harmonization on trade share is

contrastive between patent-sensitive and -insensitive industries.

Table 3 around here

6. Conclusion

Harmonizing patent law is one of the controversial issue in trade liberalization talk. Previous

studies have paid attention to how bilateral trade is related to patent protection in destination

country, but very few researchers have attempted to examine empirically the relation between

trade and di¤erence in patent system. This paper builds on the framework of translog gravity

equation and empirically examine the association between institutional distance of patent law

14

system between exporters and importers, using bilateral trade data of OECD countries for

the year 1995, 2000, 2005 and 2010.

The results of translog gravity model reveal that the institutional distance in terms of

patent law has a contrastive e¤ect on bilateral trade. More speci�cally, as the di¤erence

in patent law increases, import share tends to decrease for patent-sensitive products. From

the results, it is concluded that the e¤ect of patent law harmonization on trade di¤ers in

accordance with patent sensitivity of products. In previous literature, it has been highlighted

that it is important to promote patent protection in destination country to increase bilateral

trade, in the context of WTO-TRIPS implementations in developing countries. The present

study argues that patent law harmonization between exporters and importers is crucial factor

to increase bilateral trade and show that they are strongly correlated with each other only

for patent-sensitive products.

15

A. Appendix table

Table A.1: Extensive margin index by Hummels and KlenowCountry 1995 2000 2005 2010Australia 0.564 0.577 0.645 0.622Austria 0.618 0.705 0.669 0.675Belgium 0.773 0.800 0.811 0.774Canada 0.639 0.758 0.790 0.763Chile 0.180 0.218 0.238 0.253Czech Republic 0.512 0.590 0.636 0.635Denmark 0.619 0.664 0.625 0.600Finland 0.489 0.548 0.541 0.532France 0.850 0.866 0.849 0.811Germany 0.835 0.855 0.862 0.853Greece 0.371 0.408 0.428 0.430Hungary 0.449 0.543 0.561 0.561Iceland 0.102 0.175 0.184 0.194Ireland 0.467 0.556 0.531 0.500Israel 0.359 0.455 0.446 0.455Italy 0.823 0.829 0.821 0.790Japan 0.754 0.763 0.732 0.706Rep. of Korea 0.652 0.721 0.683 0.688Mexico 0.590 0.718 0.698 0.686Netherlands 0.764 0.783 0.816 0.783New Zealand 0.276 0.391 0.417 0.384Norway 0.537 0.606 0.606 0.599Poland 0.399 0.503 0.657 0.665Portugal 0.472 0.539 0.529 0.531Spain 0.723 0.756 0.764 0.751Sweden 0.666 0.695 0.692 0.665Switzerland 0.711 0.729 0.710 0.671Turkey 0.378 0.503 0.610 0.634United Kingdom 0.846 0.882 0.861 0.841United States 0.836 0.899 0.868 0.859

16

References

[1] Anderson, J., & van Wincoop, E. (2003). Gravity with gravitas: a solution to the border

puzzle. American Economic Review 93, 170�192.

[2] Awokuse, Titus O., & Hong Yin. (2010). Does stronger intellectual property rights pro-

tection induce more bilateral trade? Evidence from China�s imports. World Develop-

ment, 38, 1094�1104.

[3] Branstetter, L., R. Fisman, & C.F. Foley (2006). Do stronger intellectual property rights

increase international technology transfer?: Empirical evidence from U.S. �rm-level

panel data. The Quarterly Journal of Economics 121, 321�349.

[4] Chaney, T. (2008). Distorted gravity: the intensive and extensive margins of interna-

tional trade. American Economic Review 98, 1707�1721.

[5] De Groot, Henri L. F., Gert-Jan Linders, Piet Rietveld & Uma Subramanian (2004).

The institutional determinants of bilateral trade patterns. Kyklos, 57(1), 103�123.

[6] Diewert, W.E. (1976). Exact and superlative index numbers. Journal of Econometrics

4, 115�145.

[7] Eaton, J., Kortum, S. (2002). Technology, geography and trade. Econometrica 70, 1741�

1779.

[8] Feenstra, R. (2003). A homothetic utility function for monopolistic competition models,

without constant price elasticity. Economics Letters 78, 79�86.

17

[9] Helpman, E., Melitz, M., & Rubinstein, Y. (2008). Estimating trade �ows: trading

partners and trading volumes. Quarterly Journal of Economics, 123, 441�487.

[10] Hummels, David, & Klenow, Peter J. (2005). The variety and quality of a nation�s

exports. American Economic Review, 95(3), 704�723.

[11] Ivus, Olena. (2010). Stronger patent rights raise high-tech exports to the developing

world? Journal of International Economics. 81(1), 38�47.

[12] Ivus, Olena. (2015). Does stronger patent protection increase export variety? Evidence

from U.S. product-level data. Journal of International Business Studies.

[13] Maskus, K. & Penubarti, M. (1995). How trade-related are intellectual property rights?

Journal of International Economics, 39, 227�248.

[14] Melitz, M., & Ottaviano, G. (2008). Market size, trade, and productivity. Review of

Economic Studies 75, 295�316.

[15] Novy, D. (2013) International trade without CES: Estimating translog gravity. Journal

of International Economics, 89, 271�282.

[16] Park W.G. (2008). International patent protection: 1960�2005. Research Policy, 37,

761�766.

[17] Ra�quzzaman, Mohammed. (2002). The impact of patent rights on international trade:

Evidence from Canada. Canadian Journal of Economics, 35(2), 307�330.

[18] Smith, P. (1999). Are weak patent rights a barrier to U.S. exports? Journal of Interna-

tional Economics, 48, 151�177.

18

Table 1: Share of trade volume (2010)

Non-OECD!non-OECD

Non-OECD!OECD

OECD!non-OECD

OECD!OECD

Patent-sensitive 0.100 0.188 0.102 0.611Patent-insensitive 0.229 0.171 0.245 0.355

Notes: Author�s calculation based on BACI trade data

19

Table 2: Basic results of translog gravity equationDependent variable (1) xij

yj(2) xij

yj(3) xij

yj(4) xij

yj(5) xij

yj

niln(distij) -0.0158*** -0.0158*** -0.00790*** -0.00504*** -0.00508***[0.00211] [0.00220] [0.00105] [0.000710] [0.000725]

niiprij 0.00393* 0.00341[0.00229] [0.00245]

niiprij: patent-sensitive -0.00211*** -0.00236*** -0.00266***[0.000804] [0.000637] [0.000699]

niiprij: patent-insensitive 0.00604*** 0.00579*** 0.00549***[0.00158] [0.00136] [0.00143]

nicontigij 0.0148*** 0.0149***[0.00319] [0.00319]

nicomlangij 0.00427*** 0.00416***[0.00157] [0.00158]

nilnT distj 0.0178*** 0.0163*** 0.00890*** 0.00473*** 0.00296**[0.00330] [0.00397] [0.00165] [0.00102] [0.00123]

niTiprj -0.0161*** -0.0236*** -0.00806*** -0.00496* -0.00867***

[0.00512] [0.00590] [0.00255] [0.00260] [0.00298]niT

contigj -0.0338*** -0.0446***

[0.0126] [0.0128]niT

comlangj -0.00194 0.000515

[0.00312] [0.00320]Exporter dummy Yes No Yes Yes No

Year dummy Yes No Yes Yes No

Exporter-Year dummy No Yes No No Yes

Constant 0.0419*** 0.0545*** 0.0209*** 0.0231*** 0.0339***[0.00575] [0.0175] [0.00287] [0.00261] [0.00466]

Observations 3480 3480 6,960 6,960 6,960R-squared 0.528 0.538 0.423 0.488 0.498Log Likelihood 10605 10642 24298 24713 24782Notes: ***, **, and * indicate signi�cance at the 1%, 5%, and 10% levels, respectively. Robuststandard errors clustered within a country pair (30� 29 = 870 pairs) are in brackets.

20

Table 3: Results from alternative speci�cation of translog gravityDependent variable (1) xij

yjni(2) xij

yjni(3) xij

yjni(4) xij

yj(5) xij

yj(6) xij

yj

ln(distij) -0.00675*** -0.00414*** -0.00415*** -0.00511*** -0.00298*** -0.00298***[0.000810] [0.000595] [0.000603] [0.000691] [0.000482] [0.000488]

iprij: patent-sensitive -0.00267*** -0.00243*** -0.00229*** -0.00197*** -0.00177*** -0.00167***[0.000445] [0.000388] [0.000466] [0.000367] [0.000322] [0.000390]

iprij: patent-insensitive 0.00288*** 0.00312*** 0.00326*** 0.00212*** 0.00231*** 0.00241***[0.000745] [0.000700] [0.000788] [0.000632] [0.000598] [0.000675]

contigij 0.0126*** 0.0126*** 0.0102*** 0.0103***[0.00251] [0.00254] [0.00215] [0.00218]

comlangij 0.00288*** 0.00287*** 0.00246*** 0.00246***[0.00107] [0.00109] [0.000903] [0.000916]

Exporter dummy Yes Yes No Yes Yes No

Importer dummy Yes Yes No Yes Yes No

Year dummy Yes Yes No Yes Yes No

Exporter-Year dummy No No Yes No No Yes

Importer-Year dummy No No Yes No No Yes

Constant 0.0772*** 0.0498*** 0.0476*** 0.0603*** 0.0378*** 0.0359***[0.00950] [0.00711] [0.00683] [0.00826] [0.00603] [0.00573]

Observations 6,960 6,960 6,960 6,960 6,960 6,960R-squared 0.414 0.468 0.475 0.413 0.467 0.475Log Likelihood 22918 23258 23306 24238 24571 24623Notes: ***, **, and * indicate signi�cance at the 1%, 5%, and 10% levels, respectively. Robust standarderrors clustered within a country pair (30 � 29 = 870 pairs) are in brackets. Columns (4)-(6) present theresults from Eq. (10) when the restriction that all countries have one good (i.e., ni = 1) is imposed in themodel.

21