0% found this document useful (0 votes)

49 views38 pages

Impact of RTAs on Technology Spillovers

This study investigates the impact of deeper regional trade agreements (RTAs) on international technology spillovers using patent citation data. The findings indicate that RTAs enhance technology spillovers on average, with the depth of integration being particularly significant for North-South and South-South country pairs. Additionally, the study reveals that broader economic integration has a greater influence on technology spillovers than technology-specific provisions within RTAs.

Uploaded by

nancyxjaimet75

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

49 views38 pages

Impact of RTAs on Technology Spillovers

Uploaded by

nancyxjaimet75

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

Received: 4 June 2018

| Revised: 10 December 2018

| Accepted: 30 January 2019

DOI: 10.1111/twec.12797

ORIGINAL ARTICLE

Do deeper regional trade agreements enhance

international technology spillovers?

Naoto Jinji1 | Xingyuan Zhang2 | Shoji Haruna3

1
Faculty of Economics, Kyoto University, Kyoto, Japan
2
Faculty of Economics, Okayama University, Okayama, Japan
3
Faculty of Economics, Fukuyama University, Fukuyama, Japan

KEYWORDS
deep integration, patent citations, regional trade agreements, technology spillovers

1 | IN T RO D U C T ION

The last two decades have witnessed a rapid proliferation of regional trade agreements (RTAs).
Although early studies on the economic effects of RTAs focused on their static effects, such as
trade creation and trade diversion (Viner, 1950), more recent studies have addressed the dynamic
effects of RTAs, including technology adoption and technology spillovers (Bustos, 2011; Ederington
& McCalman, 2008; Schiff & Wang, 2003). For example, Bustos (2011) estimates the effects of
Mercosur on the adoption of technology by Argentinean firms. Her results show that a tariff reduction
by Brazil induced statistically significant increases in Argentinean firms' technology spending and
innovation indexes.
RTAs may also enhance technology spillovers across countries (Das & Andriamananjara, 2006).
Although RTAs are primarily aimed at expanding trade in goods by reducing tariffs on imports, many
recent RTAs pursue a deeper integration (Baldwin, 2011). For example, many RTAs now include
liberalisation of investment and harmonisation of intellectual property rights (IPR) protection poli-
cies. Thus, it is expected that RTAs affect the flow of knowledge across countries. In this study, we
investigate this issue empirically.
One approach to measuring technology spillovers, pioneered by Jaffe, Trajtenberg, and Henderson
(1993), is to employ patent citation data (Branstetter, 2006; Hall, Jaffe, & Trajtenberg, 2001; Jaffe &
Trajtenberg, 1999; MacGarvie, 2006; Maurseth & Verspagen, 2002). Patent citations are references to
existing patents included in patent documents. Although the main role of these citations is to delimit
the scope of contributions by the present patent and by previous patents, they can be interpreted to
indicate that the knowledge in the cited patents was “in some way useful for developing the new
knowledge described in the citing patent” (Maurseth & Verspagen, 2002, p. 534). In other words,
patent citations represent a “paper trail” of knowledge flows from the cited patent to the citing patent
(Jaffe et al., 1993). Thus, patent citations can be used as a direct measure of knowledge flows (Hall
et al., 2001). In particular, patent applicants at the United States Patent and Trademark Office (USPTO)

2326 | © 2019 John Wiley & Sons Ltd [Link]/journal/twec World Econ. 2019;42:2326–2363.
|

14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
JINJI et al. 2327

have a legal duty to list the existing patents at the USPTO they cite on the front page of the application
document. Given its effectiveness, we follow this approach to measuring technology spillovers.1 Note
that, in this study, by the term “technology spillovers” we refer to “the process by which one inventor
learns from the research outcomes of others' research projects and is able to enhance her own research
productivity with this knowledge without fully compensating the other inventors for the value of this
learning” (Branstetter, 2006, pp. 327–328). Thus, we distinguish technology spillovers from imitation
or technology adoption.
To the best of our knowledge, only Peri (2005) and Jinji, Zhang, and Haruna (2013) have investi-
gated the effects of RTAs on technology spillovers. Using a sample of 18 countries with 147 subna-
tional regions in Western Europe and North America for the period of 1975–96, Peri (2005) estimates
a gravity‐like model to examine the effects of several resistance factors on patent citations. He finds
that regional, national and linguistic borders have a negative and significant effect on technology
spillovers, but that the effect of “trade blocs” on technology spillovers is statistically insignificant.
However, his study is not comprehensive with regard to analysing the effects of RTAs on technology
spillovers, because he only includes the European Economic Community (EEC)/the European
Community (EC)/the European Union (EU) and North American Free Trade Agreement (NAFTA) as
“trade blocs.” Given that NAFTA came into force on 1 January 1994, and that his sample period ends
in 1996, the trade‐bloc dummy in his analysis may mainly capture the effects of the EEC/EC/EU. On
the other hand, Jinji et al. (2013) find a positive and significant effect of RTAs on technology spill-
overs based on a sample of 103 countries for the period 1990–99. However, the analysis in that study
is still limited, because it includes only nine RTAs.2 In contrast to these studies, we conduct a more
comprehensive analysis of the relationship between RTAs and international technology spillovers by
extending the sample to 114 countries and 125 RTAs. All RTAs notified to the World Trade
Organization (WTO) that came into force by the final year of our sample period are included, as long
as at least two countries in our sample are signatories of the RTA.
The main contribution of this study goes beyond a simple estimation of the average effect of RTAs
on technology spillovers. We focus on the effect of the depth of integration on technology spillovers.
RTAs impose various legal obligations on member countries. As a result, we can evaluate the depth
and nature of the economic integration of RTAs by examining the policy areas covered by the provi-
sions of the agreements, as well as the extent to which these obligations are legally enforceable. To
deal with this issue, Horn, Mavroidis, and Sapir (2010) identify 52 policy areas covered by RTAs of
which either the United States (US) or EC is a member, classifying them into two groups: WTO‐plus
(WTO+) and WTO‐extra (WTO‐X). The WTO+ group includes provisions that fall under the current
mandate of the WTO, but go beyond the commitments at the multilateral level. In contrast, WTO‐X
provisions include issues that fall outside the current WTO mandate. In our analysis, we focus on the
WTO‐X provisions as a measure of the depth of integration of an RTA, mainly because the WTO‐X
areas include both technology‐related provisions, such as IPR and innovation policies, and non‐tech-
nology‐related provisions, such as cultural cooperation and economic policy dialogue. We then con-
struct variables to measure the depth of RTAs from the WTO‐X provisions data and estimate the
relationship between these variables and technology spillovers.

1
We explain the reason why we use patent citations as a measure of technology spillovers in more detail in Section 3.
2
The nine RTAs are those included in the dataset provided by Andrew K. Rose: EEC/EC/EU, United States (US)–Israel FTA,
NAFTA, the Caribbean Community (CARICOM), the Agreement on Trade and Commercial Relations between Australia and
Papua New Guinea (PATCRA), Australia New Zealand Closer Economic Agreement (ANZCERTA), Central American
Common Market (CACM), Mercado Común del Sur (Mercosur) and the Association of Southeast Asian Nations (ASEAN).
|

Using patent application and citation data at the USPTO, we construct a panel of 11,666 pairs of
citing and cited countries/regions from the sample of 114 countries/regions for the 17 years from 1991
to 2007.3 The reason why we use the USPTO data is that the share of non‐domestic residents in all
applications is much higher at the USPTO than at other major patent offices, such as the European
Patent Office (EPO) and the Japanese Patent Office (JPO). Thus, the USPTO data are the best avail-
able to capture cross‐country technology spillovers by patent citations.
Our empirical strategy is as follows. We employ the number of backward patent citations per pair of
countries as a proxy for technology spillovers between the two countries and use this as the dependent
variable in our regressions. With regard to explanatory variables, we include an RTA dummy and an
index of WTO‐X provisions. The RTA dummy captures the average effect of the common RTA mem-
bership on technology spillovers, and the WTO‐X index captures the impact of the deep regional inte-
gration on technology spillovers.4 We first estimate the empirical model for technology spillovers using
the ordinary least squares (OLS). We also estimate the model by a Poisson pseudo‐maximum‐likeli-
hood (PPML) estimator, mainly because our dependent variable (i.e., the number of patent citations) is
count data and there are many zeros in the dependent variable. Moreover, to account for possible het-
erogeneous effects of RTAs and deep integration, we examine heterogeneity in three dimensions: (a)
across country groups (i.e. North–North, North–South and South–South country pairs); (b) across the
nature of RTAs (i.e. free trade agreements [FTAs] and customs unions [CUs]); and (c) the phased‐in
effects with lagged terms.5 We also account for possible endogeneity in the RTA dummy and the
WTO‐X index by employing the instrumental variable (IV) approach. With regard to the impact of a
deep integration, we further investigate whether the technology‐related WTO‐X provisions are more
strongly related to technology spillovers than other WTO‐X provisions. To address this issue, we first
conduct a factor analysis on the WTO‐X provisions data to extract factors that can be used to select
subsets of the WTO‐X provisions. Then, based on the result of the factor analysis, we categorise the
WTO‐X provisions into three subsets. We finally estimate the impact of the WTO‐X indexes that rep-
resent the different WTO‐X provision subsets.
This is the first study of the effects of the depth of RTAs on international technology spillovers.
Our main findings are as follows. First, the estimated coefficient of the simple RTA dummy is positive
and highly significant. This finding is quite robust for different estimation techniques and is consistent
with the result of Jinji et al. (2013), but is different from that of Peri (2005). We do not observe any
significant difference between FTAs and CUs. The estimate of the RTA dummy is also positive and
significant, regardless of the country pairs with different country groups, though it is larger for the
North–North country pairs than for other combinations. In addition, we find the phased‐in effect of the
RTA formation on technology spillovers. The RTA dummies with two‐ and three‐period lags as well

3
Our analysis also differs from those in previous studies (Jinji et al., 2013; Peri, 2005) in terms of the sample period. Although
Peri (2005) covers until 1996, it is important to include the late 1990s and 2000s, because the number of RTAs increased rapidly
in these periods. Jinji et al. (2013) do not cover the 2000s. Note that our sample period ends in 2007 because of the unavailabil-
ity of reliable patent data after 2007.
4
Note that a formation of an RTA by itself does not necessarily facilitate patent applications among member countries in our
data because we use data at the USPTO rather than patent offices in individual countries. Thus, we do not need to worry about
the possibility of such a “facilitation effect” being included in the RTA dummy.
5
The issue of heterogeneous effects of RTAs has recently attracted considerable attention in the empirical studies on the trade
effects of RTAs. For example, Behar and Cirera‐i‐Crivillé (2013) have analysed how the impacts of RTAs on trade differ, de-
pending on whether signatories are developed or developing countries. The heterogeneity in the effects of RTAs on bilateral
trade by the nature of integrations has been addressed by Baier, Bergstrand, and Feng (2014) and Roy (2010). Moreover, the
phased‐in effects of RTAs on trade flows have been analysed by a number of studies, including Baier and Bergstrand (2007)
and Kohl (2014).
|

as the contemporaneous dummy are positive and statistically significant. Furthermore, the estimate
of the RTA dummy generally becomes larger after employing the IV estimations, indicating that the
endogeneity problem causes a downward bias.
Second, we find that the depth of integration influences technology spillovers. The estimated co-
efficient of the WTO‐X index is positive and highly significant. The effect seems to be stronger for
FTAs than for CUs. When the WTO‐X index and the RTA dummy are jointly included as the explan-
atory variables, the estimate of the WTO‐X index becomes insignificant in the full sample and for
the North–North country pairs. However, the WTO‐X index remains positive and significant for the
North–South and South–South integration, even when the RTA dummy is jointly included. This result
suggests that the depth of integration has additional effects on technology spillovers for the North–
South and South–South country pairs but may have no additional impact for the North–North country
pairs. We do not find the phased‐in effect for the WTO‐X index. Estimating with lagged terms, only
the contemporaneous WTO‐X index is significant.
Third, when we examine the content of the legal obligations of RTAs in more detail, we find
that the depth of integration in a broad sense may be more important to enhancing technology spill-
overs than are the technology‐related provisions. Our estimation reveals that the WTO‐X index
coefficients representing a healthy economic environment in a general sense and basic social and
economic conditions tend to be larger than the coefficient of the WTO‐X index representing technol-
ogy‐related provisions. This finding is quite interesting and entirely new. Thus, our empirical study
suggests that the formation of RTAs enhances technology spillovers among members, on average,
and that the depth of integration in RTAs is also important for technology spillovers particularly in
the North–South and South–South integrations. Moreover, a deep integration in a broader sense has
a greater impact on technology spillovers than do the legal obligations directly related to the tech-
nology. A number of robustness checks confirm that this finding does not depend on the specific
empirical strategy that we take.
The remainder of the paper is organised as follows. In Section 2, we discuss the link between
RTAs and technology spillovers. In Section 3, we explain the reason why we use patent citations as a
measure of technology spillovers. In Section 4, we discuss the empirical framework. In Section 5, we
describe the data employed in our empirical analysis, and in Section 6, we present our empirical re-
sults. We conduct a number of robustness checks in Section 7. Finally, Section 8 concludes the paper.

2 | T H E LIN K B E T WE E N RTA S AND

TE CH NO LO GY S P IL LOV E R S

To motivate our study, we first sketch the link between RTAs and cross‐border technology spillo-
vers. Actually, it is expected that RTAs enhance cross‐border technology spillovers through various
channels.
A primary reason why RTAs may potentially facilitate technology spillovers is that RTAs expand
international trade in goods by reducing trade barriers. A number of existing empirical studies have
identified trade in goods as a major channel of technology spillovers (Acharya & Keller, 2009; Coe
& Helpman, 1995; Xu & Wang, 1999). One possibility is that imports of foreign intermediate goods
that embody foreign advanced technology enable the implicit use of that technology in final‐good pro-
duction (Keller, 2004). RTAs also facilitate investment and the movement of persons. As shown by a
number of empirical studies, foreign direct investment (FDI) is another major channel of cross‐border
technology spillovers (Haskel, Pereira, & Slaughter, 2007; Javorcik, 2004; Keller & Yeaple, 2009).
The movement of persons more directly promotes the exchange of knowledge and idea.
|

One may think that trade in goods and FDI are rather indirect channels for the link between RTAs
and technology spillovers. A more direct link between RTAs and technology spillovers may arise
from the depth of integration in RTAs. As discussed in the introduction, many recent RTAs pursue a
deeper integration (Baldwin, 2011). Some RTAs explicitly include provisions that stimulate technol-
ogy spillovers. For example, the US–Peru Free Trade Agreement includes a provision on promotion
of innovation and technological development (Article 16.12), which encourages the engagement in
collaborative scientific research projects and transfer of technology. Thus, RTAs may enhance the
exchange of technological knowledge among member countries because of these technology‐related
provisions.
Horn et al. (2010) propose a framework to systematically measure the depth of RTAs. They define
the content and the legal enforceability of various provisions included in RTAs in terms of two cate-
gories: WTO‐plus (WTO+) and WTO‐extra (WTO‐X).6 As shown in Table 1, 14 policy areas are
classified in the WTO+ category and 38 policy areas are in the WTO‐X category. The WTO‐X

TABLE 1 List of WTO+ and WTO‐X areas in RTAs

WTO+ areas WTO‐X areas

FTA industrial goods Anti‐corruption Health
FTA agricultural goods Competition policy Human rights
Customs administration Environmental laws Illegal immigration
Export taxes IPR Illicit drugs
SPS Investment Industrial cooperation
TBT Labour market regulation Information society
State trading enterprises Movement of capital Mining
Anti‐dumping Consumer protection Money laundering
Countervailing measures Data protection Nuclear safety
State aid Agriculture Political dialogue
Public procurement Approximation of legislation Public administration
TRIMs Audio visual Regional cooperation
GATS Civil protection Research and
technology
TRIPs Innovation policies SME
Cultural cooperation Social matters
Economic policy dialogue Statistics
Education and training Taxation
Energy Terrorism
Financial assistance Visa and asylum
Source: Horn et al. (2010).

6
Orefice and Rocha (2014) use the definitions of Horn et al. (2010) and analyse the relationship between the depth of RTAs and
production networks. Hofmann, Osnago, and Ruta (2018) extend the coverage of the dataset. Using their dataset, Mattoo,
Mulabdic, and Ruta (2017) analyse how deep integration affects trade creation and trade diversion effects of RTAs. Dür,
Baccini, and Elsig (2014) develop the original dataset on deep integration and show that RTAs increase trade flows and that
such positive effect is largely driven by deep integrations.
|

category includes those provisions particularly related to technology, such as IPR, innovation policies,
and research and technology.
Horn et al. (2010) evaluate the extent to which each of WTO+ and WTO‐X provisions is legally
enforceable in RTAs, measuring on a scale of zero to two: The legal enforceability of a provision
takes zero if the provision is not mentioned in the RTA or is not legally enforceable; it takes one if the
provision is mentioned, legally enforceable, but explicitly excluded by a dispute settlement provision;
and finally, it takes two if the provision is mentioned and legally enforceable.
In this study, we investigate whether WTO‐X provisions in RTAs actually enhance technology
spillovers among member countries, and examine which of the WTO‐X provisions are more important
in terms of technology spillovers.

3 | T E CH NOLOGY SP IL LOV E RS AND PATENT CITATIONS

As explained in the Introduction, we use patent citations as a measure of technology spillovers in our
empirical analysis. In this section, we explain the reason why patent citations are a good proxy for
technology spillovers.
In general, the granting of a patent is “a legal statement that the idea embodied in the patent rep-
resents a novel and useful contribution over and above the previous state of knowledge” (Jaffe et al.,
1993, p. 580). Patent citations serve “the legal function of delimiting the scope of the property right
conveyed by the patent” (Jaffe et al., 1993, p. 580). Therefore, “a citation of Patent X by Patent Y means
that X represents a piece of previously existing knowledge upon which Y builds” (Jaffe et al., 1993,
p. 580). Patent applicants are required to provide all references that may affect the patentability of the
invention. In particular, applicants at the USPTO have a legal duty to list the patents they refer to.
Patent citation data are viewed as a good proxy for technology spillovers and hence has been exten-
sively used in the literature (Branstetter, 2006; Hall et al., 2001; Haruna, Jinji, & Zhang, 2010; Jaffe
& Trajtenberg, 1999; Jaffe et al., 1993; Jinji et al., 2013; MacGarvie, 2006; Mancusi, 2008; Maurseth
& Verspagen, 2002; Peri, 2005). The main reason is that from the relationship between a citing patent
and a cited patent we can know a direct flow of knowledge from the cited patent to the inventor of
the citing patent. This is analogous to references in academic journal articles. Unlike references in
academic journal articles, however, “gratuitous” citations are less likely in patent citations, because
citations reduce the scope of contributions by the citing patent (Jaffe et al., 1993).
However, one may argue that patent citations are an incomplete and noisy measure of technology
spillovers (Criscuolo & Verspagen, 2008; Jaffe et al., 1993). Patent citations are an incomplete measure
of technology spillovers, because technology spillovers that can be captured by patent citation data are
only a small fraction of the real spillovers. The reason is that “only a small fraction of research output is
ever patented” (Jaffe et al., 1993, p. 584). In addition, patent citations are a noisy measure of technology
spillovers because citations may be added by patent examiners (Alcácer & Gittelman, 2006; Criscuolo &
Verspagen, 2008; Jaffe, Trajtenberg, & Fogarty, 2000). Citations added by patent examiners are unlikely
to reflect knowledge flows, because inventors may not have been aware of the added citations.7
Despite these limitations, the following two studies have shown that patent citations can be used
as a legitimate measure of technology spillovers. First, based on a survey of USPTO patent inventors,
Jaffe et al. (2000) report that communications among inventors are important and that patent citations
provide an indication of communications. Second, matching a sample of French firms' responses to

7
The share of citations added by patent examiners is lower in the USPTO patents than in the EPO patents (Maurseth &
Verspagen, 2002).
|

a survey and their citations of EPO patents, Duguet and MacGarvie (2005) provide evidence on the
legitimacy of using patent citations as a measure of technology spillovers. They found that patent
citations are significantly correlated with firms' statements about their acquisition and dispersion of
new technologies. In particular, their results indicate that backward citations are positively and signifi-
cantly associated with cooperative R&D, licensing, and mergers and acquisition (M&A).
For the above reasons, we think that we can use patent citations as a legitimate measure of technol-
ogy spillovers.8

4 | E M P IR ICA L F R A ME WOR K

Our primary interest is estimating the relationship between the membership of RTAs and technology
spillovers among RTA members. To investigate this issue, we estimate the following equation:
(
Cijt = exp 𝛽0 + 𝛽1 RTA_Variableijt + 𝛽2 WTOijt + 𝛾1 ln (Pit ) + 𝛾2 ln (Pjt ) + 𝛾3 PROXijt
)
+𝜑t + 𝜂i + 𝜆j + 𝜇it + 𝜈jt + 𝜖ijt , (1)

where Cijt is the number of backward patent citations made by the patents of country i to those of
country j at time t and RTA_Variableijt is RTAijt and/or WTO‐Xijt, where RTAijt is a dummy variable
that takes the value one if countries i and j both belong to a common RTA at t, and zero otherwise and
WTO‐Xijt is the sum of the points of legally enforceable WTO‐X provisions covered by RTAs to
which both countries i and j belong at t.9 Moreover, WTOijt is a dummy variable that takes the value
one if both i and j are GATT/WTO members at t, and zero otherwise. In Equation 1, we include three
control variables: ln (Pit), ln (Pjt) and PROXijt. Variables ln (Pit) and ln (Pjt) are the logarithm values
of the patent stocks for countries i and j, respectively, and PROXijt is an index of the technological
proximity between i and j, as measured by the patent classes. Moreover, a constant term (i.e. β0) and
various fixed effects represented by φt, ηi, λj, μit, and νjt are included when we estimate Equation (1).10
εijt is an error term.
We construct Pit (and Pjt) from the data on patent applications at the USPTO using the following
perpetual inventory method:
Pit = Ait + (1 − 𝛿)Pit−1 , (2)

where Ait is the number of patent applications made by country i at the USPTO during period t and δ
is the depreciation rate. Following convention, we use δ = 0.15 (Hall, Jaffe, & Trajtenberg, 2005).11

8
To check the robustness, we use two alternative measures of technology spillovers in Section 7.4; patents with foreign co‐in-
ventors; and exports of royalties and licensing fees.
9
As explained earlier, each WTO‐X provision is evaluated on a scale of zero to two for individual RTAs.
10
Although we do not include country‐pair fixed effect, we include PROXijt as a country‐pair‐specific variable. In addition, we
use a number of variables related to the country‐pair‐specific characteristics between the two countries as our instrumental
variables, as explained in Section 6.3.
11
When we construct the patent stock data, we use the number of patent applications in 1975 as the initial value of Pit. Since we
use patent application data from 1975 and our sample period begins in 1991, the initial value of Pit has little influence on the
value of Pit in 1990. Note that we estimate the latter value using the perpetual inventory method (Equation (2)). Moreover, in
the estimation, we use patent stock variables with a lag of one period to account for a simultaneity bias. That is, in our estima-
tion, we use ln (Pit−1) and ln (Pjt−1), the values of the patent stocks at the beginning of the year.
|

The index of technological proximity between two countries, PROXijt, is based on the work of Jaffe
(1986). We assume that the “technological position” of country i (j) in research areas in year t can be
characterised by the vector Fit(Fjt). This is a vector of the shares of four‐digit patent classes, according
to the International Patent Classification (IPC), in all USPTO patent applications by country i (j) in
year t. Then, the proximity of i and j in the technology space is computed as an uncentred correlation
between Fit and Fjt:

F�it Fjt
PROXijt = √ .
(3)
√
F�it Fit F�jt Fjt

Note that PROXijt is bounded between 0 and 1, and is closer to 1 if countries i and j are positioned
close to one another.
Tables 13 and 14 show the descriptive statistics of the variables and the correlations among them,
respectively.
In Equation (1), the coefficient of RTA measures the impact of RTAs on technology spillovers
under the assumption that all RTAs have the same effect on technology spillovers. The RTAs repre-
sented by the dummy variable RTA include the 125 RTAs listed in Table 15. In addition, the coeffi-
cient of the WTO‐X index captures the impact of a deep integration in RTAs on technology spillovers.
We expect positive signs for the coefficients of RTA and WTO‐X.
We also estimate the impact of GATT/WTO. The coefficient of WTO represents the impact of
GATT/WTO on technology spillovers if both partners are GATT/WTO members. Since GATT/WTO
membership is expected to enhance trade between trade partners and, hence, enhance technology
spillovers between them, we also expect the sign of the WTO coefficient to be positive.
With regard to the control variables, we expect the signs of the coefficients of both ln (Pit) and
ln (Pjt) to be positive, because the chance of citing a patent increases as the citing and cited countries
have larger patent stocks. The citing country's patent stock, ln (Pit), is considered to reflect its absorp-
tive capacity of technology. The cited country's patent stock, ln (Pjt), represents its potential opportu-
nity of being cited. Then, PROXijt is also expected to have a positive coefficient because countries that
share similar technology have a greater chance of citing each other's patents than otherwise.
We first estimate Equation (1) using the OLS method. In the OLS estimations, we use the loga-
rithm of the patent citations (i.e., ln (Cijt)) as the dependent variable. We include the citing and cited
country fixed effects as well as year dummies or the time‐varying (citing and cited) country fixed
effects. These fixed effects address the issue of the omitted variable bias.
Since our dependent variable, Cijt, is count data and includes many zeros, the OLS estimates of the
log‐linearised model may be biased and inefficient. To deal with this issue, we estimate Equation (1)
by the Poisson pseudo‐maximum likelihood (PPML) estimator, which is recommended by the gravity
literature to deal with the issue of zero trade flows (Head & Mayer, 2014; Santos Silva & Tenreyro,
2006, 2011). As in OLS, we include the citing and cited country fixed effects and year dummies or the
time‐varying (citing and cited) country fixed effects in the PPML estimations.12
Moreover, we address the issue of endogeneity for the RTA dummy and the WTO‐X index by em-
ploying the instrumental variable (IV) approach. We employ two IV approaches, namely, a two‐stage
least squares estimation (IV/2SLS) and a Poisson regression with endogenous regressors (IVP). We

12
Falley (2015) shows that estimating gravity equations by PPML with fixed effects and using an iterative estimation approach
(Anderson & van Wincoop, 2003) lead to the same results.
|

check the robustness of the results using two sets of IVs for each estimation technique. We will explain
more in detail in Section 6.3.

5 | D E S C R I P T IO N OF T H E DATA

This section describes the data we use in our empirical analysis.13 First, the data on patent applications
and patent citations are taken from the April 2012 edition of the Worldwide Patent Statistical Database
(PATSTAT) released by the EPO. We extract the patent statistics of the USPTO from the PATSTAT.
This dataset includes information on the application date, the country name of the assignee, the IPC
patent class, and citations made and received for each patent. In our analysis, we use patent applica-
tion data from 1975 to 2007 and patent citation data from 1991 to 2007. We use the USPTO patent
citation data as a proxy for technology spillovers.14 For example, if a patent applied to the USPTO by
an inventor who is resident in Japan cites a USPTO patent of which the assignee is resident in Germany,
then we count it as a citation made by Japan to the patent of Germany. We interpret it as evidence of
technology spillovers from Germany to Japan.
Let us next explain our sample. The sample includes countries/regions that have at least one patent
application to the USPTO during the sample period. Our sample covers 114 countries/regions (listed
in Table 16). We then construct a panel of 11,666 pairs of citing and cited countries/regions from 1991
to 2007 (17 years).
We construct dummy variables for the membership of the 125 RTAs based on the information
taken from the web page of the WTO.15 All RTAs notified to the WTO that came into force by 2007
and to which at least two countries/regions in our sample are signatories are included in the sample.16
The list of RTAs covered in our sample is shown in Table 15. Data on the content of RTAs are also
taken from the WTO website.17 This dataset was originally provided by Horn et al. (2010) and ex-
tended by the Research division of the WTO. One hundred RTAs are included in this dataset. Thus,
we set the WTO‐X index at zero for 25 RTAs excluded from the WTO's dataset. Kohl, Brakman, and
Garretsen (2016) substantially extend the coverage of RTAs from 100 to 296. However, only four
WTO‐X policy areas are included in their dataset.18 For this reason, we employ their dataset only for
the robustness check.
The information on each country's GATT/WTO membership is also taken from the web page
of the WTO. We also use bilateral gravity variables, as explained in Section 6.3. All of them are
taken from the geographical database provided by the Centre d’Études Prospectives et d'Informations
Internationales (CEPII). Bilateral trade data to construct an instrumental variable, which is also ex-
plained in Section 6.3, are taken from the BACI dataset provided by CEPII.
Descriptive statistics and correlations of the variables are provided in Tables 13 and 14, respectively.

13
The data that support the findings of this study are available from the corresponding author upon reasonable request.
14
Patent applications at the USPTO may include citations of patents at other patent offices, such as EPO and JPO. However, in
our dataset, the cited patents are all those of the USPTO.
15
[Link]
16
We include free trade areas, customs unions, and economic integration agreements, but do not include partial scope
agreements.
17
[Link]
18
With regard to WTO+ policy areas, 13 out of 14 policy areas are included in their dataset.
|

6 | E M P IR ICA L R E SU LTS

6.1 | Baseline estimations

In this section, we report our estimation results. First, the results from the OLS estimations are re-
ported in columns (1)–(6) of Table 2. Estimations with the citing and cited country fixed effects and
year dummies are reported in columns (1)–(3), whereas estimations with country‐year fixed effects
are reported in columns (4)–(6). As shown in the table, all three control variables have highly signifi-
cant coefficients, with the expected signs in all regressions. The coefficients of the log of the patent
stocks in the citing and cited countries/regions (ln (Pi) and ln (Pj)) are positive and highly significant.
This means that pairs of countries/regions holding more patent stocks experience higher technology
spillovers. The index of technological proximity (PROX) also has a positive and highly significant
coefficient in all regressions. Thus, countries holding similar technology exchange more technological
knowledge.
The estimated results concerning the impact of RTAs are also as expected. The coefficient of the RTA
dummy is positive and highly significant (columns (1) and (4)). The depth of RTAs, measured by the
WTO‐X index, influences technology spillovers. That is, the estimated coefficient of the WTO‐X index is
positive and highly significant (columns (2) and (5)). When the RTA dummy and the WTO‐X index are
both included, however, the estimated coefficient of the WTO‐X index becomes insignificant (columns
(3) and (6)). This result may be partly due to the problem of multicollinearity, because the correlation
between the RTA dummy and the WTO‐X index is high (see Table 14). Moreover, when the RTA dummy
is excluded (columns (2) and (5)), the WTO‐X index captures the effect of signing an RTA on technology
spillovers. This is because the WTO‐X index must take the value of zero if the RTA dummy is zero for a
given country pair but the WTO‐X index may be zero even when the RTA dummy takes the value of one.
In contrast, when the RTA dummy and the WTO‐X index are both used as explanatory variables at the
same time, then the WTO‐X index only captures the additional impact of a deep integration on bilateral
technology spillovers. Our estimation results show that the additional impact is statistically insignificant.
We come back this issue in the next subsection by categorising the sample countries into the North and
the South.
We next report the estimation results from the PPML model in columns (6)–(12) in Table 2. Table 2
indicates that the results are qualitatively quite similar to those of the OLS estimation.19 When the RTA
dummy and the WTO‐X index are separately included, the estimated coefficients of the RTA dummy
and the WTO‐X index remain positive and highly significant (columns (7), (8), (10) and (11)). When
the two variables are included together, the estimated coefficient of the WTO‐X index is weakly sig-
nificantly negative (column (9)) or insignificant (column (12)).
Our finding regarding the effect of RTAs on technology spillovers is quite different from that of
Peri (2005). Recall that Peri finds no significant effect of the RTA dummy (which he calls the “trade‐
bloc” dummy). One possible reason why the RTA dummy is not significant in his analysis is that, as
pointed out in Section 1, his RTA dummy mainly captures the effect of RTAs in Europe (EEC/EC/
EU). He uses the USPTO data (as does this study), but the citations in that data are unevenly more
frequent for pairs involving the US. Jinji et al. (2013), who also use the USPTO data, find that the
significance of the EEC/EC/EU dummy depends on the estimation technique. In particular, the EEC/
EC/EU dummy is insignificant for the period 1990–99 when the model is estimated using a negative
binomial model.

19
In the OLS estimations, since the dependent variable is ln (Cijt), observations with Cijt = 0 are excluded from estimation. Thus,
the number of observations is much smaller in columns (1)–(6) than in columns (7)–(12).
TABLE 2 Baseline estimations: the impacts of RTA on technology spillovers

2336
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

|
OLS OLS OLS OLS OLS OLS PPML PPML PPML PPML PPML PPML
RTA 0.263*** 0.285*** 0.333*** 0.339*** 0.244*** 0.288*** 0.250*** 0.275***
(0.037) (0.044) (0.042) (0.048) (0.030) (0.046) (0.027) (0.047)
WTO‐X 0.002*** −0.0005 0.002*** −0.0001 0.002*** −0.0009* 0.002*** −0.0004
(0.0006) (0.0005) (0.0004) (0.0006) (0.0004) (0.0005) (0.0004) (0.0005)
WTO 0.183*** 0.194*** 0.181*** 1.223*** 1.440*** 1.222*** 0.226* 0.178 0.233* 0.722** 0.751** 0.718**
(0.058) (0.059) (0.058) (0.303) (0.295) (0.301) (0.126) (0.123) (0.125) (0.303) (0.327) (0.300)
ln (Pi) 0.372*** 0.376*** 0.370*** 0.599*** 0.615*** 0.596***
(0.025) (0.025) (0.025) (0.033) (0.030) (0.033)
ln (Pj) 0.469*** 0.471*** 0.467*** 0.835*** 0.859*** 0.831***
(0.028) (0.028) (0.028) (0.070) (0.063) (0.070)
PROX 1.700*** 1.711*** 1.703*** 2.085*** 2.096*** 2.086*** 1.840*** 1.988*** 1.819*** 2.677*** 2.880*** 2.657***
(0.084) (0.086) (0.084) (0.120) (0.123) (0.119) (0.415) (0.402) (0.413) (0.349) (0.316) (0.356)
Year FE Yes Yes Yes Yes Yes Yes
Citing Yes Yes Yes Yes Yes Yes
country FE
Cited country Yes Yes Yes Yes Yes Yes
FE
Citng Yes Yes Yes Yes Yes Yes
country‐year
FE
Cited Yes Yes Yes Yes Yes Yes
country‐year
FE
No. of 26,607 26,607 26,607 26,777 26,777 26,777 169,704 169,704 169,704 138,854 138,854 138,854

JINJI et al.
observations
Notes: Estimations are implemented using the Stata command ppml for the PPML model. The dependent variable is ln (Cijt) in OLS and Cijt in PPML. ***, ** and * denote 1%, 5% and 10%
significance levels, respectively. Standard errors clustered by country‐pair are in parentheses. The regressions include the constant term.

6.2 | The heterogeneous impacts of RTAs on technology spillovers

We next analyse the heterogeneous impacts of RTAs on technology spillovers.20 We estimate the
three dimensions of the heterogeneity: (a) across country groups, that is, North–North, North–South
and South–South integrations; (b) across the nature of RTAs, that is, FTAs and CUs; and (c) the
phased‐in effects with lagged terms. In this subsection, we only report the estimation results from the
PPML model.
First, Table 3 shows the results regarding the heterogeneous effects by combinations of the coun-
tries with different country groups. We categorise the countries of the OECD members as the “North”
and countries/regions of the OECD non‐members as the “South.”21 In this table, the estimation results
with a sub‐sample of North–North country pairs are reported in columns (1)–(3), those with a sub‐
sample of North–South country pairs are in columns (4)–(6), and those with a sub‐sample of North—
South and South–South country pairs are in columns (7)–(9). Note that in the estimations of the
North–North sub‐sample, we exclude the WTO dummy because the Northern countries are all GATT/
WTO members in our sample period.
As shown in the table, the RTA dummy and the WTO‐X index are individually positive and highly
significant, regardless of the combination of countries with different country groups. Moreover, an
important finding is that, in columns (6) and (9), the WTO‐X index remains significantly positive even
when both the RTA dummy and the WTO‐X index are included, whereas it is insignificant in column
(3).
The results suggest that the additional impact of deep RTAs on technology spillovers is positive and
significant for the RTAs signed by the combinations of North–South countries/regions and South–South
countries/regions, whereas the additional impact of the WTO‐X is insignificant for the North–North coun-
try pairs. These results seem to be reasonable because the issues addressed by the WTO‐X provisions,
such as IPR protection and labour market regulation, are likely to have been treated properly in Northern
countries, regardless of the formation of RTAs. Consequently, even if those issues are addressed by RTAs,
the WTO‐X provisions have no real impact on technology spillovers among Northern countries. In con-
trast, for North–South and South–South country pairs, the WTO‐X provisions have some real impacts on
technology spillovers even after separating the effect of signing an RTA out.
Second, we run regressions by distinguishing between FTA and CU. The results are reported in
Table 4. In columns (1) and (2), we include an FTA dummy and a CU dummy, while we include inter-
action terms between the WTO‐X index and the FTA dummy and between the WTO‐X index and the
CU dummy in columns (3) and (4). The results indicate that there is no significant difference between
the FTA dummy and the CU dummy. However, the WTO‐X index seems to have a stronger impact on
technology spillovers in FTAs than in CUs.
Finally, following Baier and Bergstrand (2007) and Kohl (2014), we examine the phased‐in effects of RTA
and WTO‐X provisions. The difference between these previous studies and our study is that the data in 5‐year
intervals are used in the previous studies, whereas we use the data in every year. Thus, unlike the previous
studies, our data allow us to investigate the phased‐in effects by shorter intervals. The results are reported in
Table 5. Although the phased‐in effect seems to be present for the RTA dummy, it is not observed for the WTO‐X
index. As shown in columns (1)–(3), when the effect of the RTA dummy is decomposed into lagged dummies
by adding lagged RTA dummies, not only the contemporaneous dummy but also the dummies with two‐ and
three‐period lags are significantly positive. In contrast, with regard to the WTO‐X index, even when lagged
20
We thank an anonymous referee for pointing out this important issue.
21
We take into account the change in the OECD membership over time during our sample period.
2338
TABLE 3 Heterogeneous impacts of RTA on technology spillovers: country groups by North and South

|
(1) (2) (3) (4) (5) (6) (7) (8) (9)

North–South North–South North–South
& & &
North–North North–North North–North North–South North–South North–South South–South South–South South–South
RTA 0.251*** 0.298*** 0.087*** 0.067** 0.137*** 0.123***
(0.030) (0.054) (0.032) (0.033) (0.045) (0.047)
WTO‐X 0.002*** −0.001 0.005*** 0.004*** 0.005*** 0.003**
(0.000) (0.001) (0.001) (0.001) (0.001) (0.001)
WTO −1.760*** 3.970*** −1.755*** 0.750** 0.805*** 0.750**
(0.634) (0.404) (0.633) (0.298) (0.309) (0.299)
PROX 2.802*** 3.004*** 2.761*** 2.260*** 2.392*** 2.296*** 2.224*** 2.423*** 2.249***
(0.349) (0.319) (0.361) (0.157) (0.138) (0.158) (0.151) (0.134) (0.153)
Citng Yes Yes Yes Yes Yes Yes Yes Yes Yes
coun-
try‐
year
FE
Cited Yes Yes Yes Yes Yes Yes Yes Yes Yes
coun-
try‐
year
FE
No. of 11,980 11,980 11,980 58,014 58,014 58,014 124,381 124,381 124,381
obser-
vations
Notes: Estimations are implemented using the Stata command ppml. The dependent variable is Cijt. A sub‐sample of the North–North country pairs is used for columns (1)–(3), the North–South for

JINJI et al.
columns (4)–(6) and the North–South and South–South for columns (7)–(9). ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Standard errors clustered by country‐pair are in
parentheses. The regressions include the constant term.

TABLE 4 Heterogeneous impacts of RTA on technology spillovers: FTA versus CU

(1) (2) (3) (4)

PPML PPML PPML PPML

FTA 0.247*** 0.251***
(0.050) (0.049)
CU 0.239*** 0.249***
(0.041) (0.037)
WTO‐X × FTA 0.012*** 0.010***
(0.003) (0.002)
WTO‐X × CU 0.002*** 0.002***
(0.000) (0.000)
WTO 0.227* 0.201 0.722** 0.758**
(0.125) (0.123) (0.303) (0.326)
ln (Pi) 0.599*** 0.609***
(0.032) (0.031)
ln (Pj) 0.835*** 0.848***
(0.069) (0.066)
PROX 1.837*** 1.917*** 2.677*** 2.809***
(0.415) (0.404) (0.354) (0.326)
Year FE Yes Yes
Citing country FE Yes Yes
Cited country FE Yes Yes
Citng country‐year FE Yes Yes
Cited country‐year FE Yes Yes
No. of observations 169,704 169,704 138,854 138,854
Notes: Estimations are implemented using the Stata command ppml. The dependent variable is Cijt. ***,** and * denote 1%, 5%, and
10% significance levels, respectively. Standard errors clustered by country‐pair are in parentheses. The regressions include the
constant term.

variables are added, only the contemporaneous variable is statistically significant and the estimated coefficient of
the contemporaneous WTO‐X index is quite stable. Therefore, the RTA formation may have the phased‐in effect
on technology spillovers, but the WTO‐X provisions may affect technology spillovers only contemporaneously.
This result seems to be natural because tariffs are often reduced gradually in RTAs but most provisions that are
categorised in the WTO‐X provisions are likely to become effective immediately when RTAs enter into force.

6.3 | Estimations with instrumental variables

In the previous subsections, we found a positive and significant relationship between RTAs and tech-
nology spillovers. However, the RTA dummy and the WTO‐X index that we use as explanatory varia-
bles in our estimations may be endogenous variables. To address the issue of endogeneity, we employ
the instrumental variable (IV) approach with two sets of IVs. Our first IV is the contagion index, as
proposed by Baldwin and Jaimovich (2012) in the context of the “domino effect” of RTAs. According
to the domino theory of regionalism (Baldwin, 1995), the signing of an RTA motivates excluded
|

TABLE 5 Heterogeneous impacts of RTA on technology spillovers: phased‐in effects

(1) (2) (3) (4) (5) (6)

PPML PPML PPML PPML PPML PPML

RTAijt 0.121*** 0.121*** 0.116***
(0.036) (0.036) (0.037)
RTAij,t−1 0.133*** 0.002 0.007
(0.038) (0.022) (0.022)
RTAij,t−2 0.134*** 0.071**
(0.036) (0.029)
RTAij,t−3 0.064*
(0.036)
WTO‐Xijt 0.0017*** 0.0017*** 0.0014***
(0.00048) (0.00049) (0.00054)
WTO‐X ij,t−1 0.00053 −0.00000 −0.00000
(0.00057) (0.000001) (0.000001)
WTO‐X ij,t−2 0.00001** 0.00001
(0.00000) (0.00000)
WTO‐X ij,t−3 0.00000
(0.00000)
WTO 0.731** 0.768** 0.782** 0.762** 0.805** 0.821**
(0.305) (0.307) (0.307) (0.330) (0.334) (0.335)
PROX 2.691*** 2.698*** 2.710*** 2.893*** 2.897*** 2.909***
(0.347) (0.347) (0.345) (0.315) (0.315) (0.313)
Citng country‐year FE Yes Yes Yes Yes Yes Yes
Cited country‐year FE Yes Yes Yes Yes Yes Yes
No. of observations 133,173 127,420 121,154 133,173 127,420 121,154
Notes: Estimations are implemented using the Stata command ppml. The dependent variable is Cijt. ***,** and * denote 1%, 5% and
10% significance levels, respectively. Standard errors clustered by country‐pair are in parentheses. The regressions include the
constant term.

countries to sign new RTAs through the trade diversion effect. Baldwin and Jaimovich (2012) empiri-
cally confirm the presence of the contagion effect in RTA formation. Their idea is that the contagion
effect of country j's RTA memberships on county i can be captured by the contagion index:
∑ Xijt Xkjt
( )( )
ContagRTAijt = RTAjkt , (4)
k∈Ω
Xit Mjt
jt

where Xijt is the bilateral exports of country i to country j, Xit is country i's total exports, Mjt is country
j's total imports, Ωjt is the set of countries with which country j has an RTA in year t, and RTAjkt is a
dummy variable that takes the value one if there is an RTA between j and k at t, and zero otherwise.
Thus, ContagRTAijt indicates the sum of the RTAs signed by country j up to year t, weighted by the
export share of country j in i's total exports and the import share of country k in j's total imports. The
index of ContagRTAijt is likely to be correlated with the state of the RTA formation between countries
i and j, but unlikely to be correlated with patent citations between i and j.
|

The second set of IVs is related to the historical status of country pairs, namely CURCOLij and
SMCTRYij, both of which are taken from Egger, Larch, Staub, and Winkelmann (2011).22 Here,
CURCOLij is a dummy variable that takes the value one if either of countries i and j had been a colony
of the other after 1945, and zero otherwise, and SMCTRYij is a dummy variable that takes the value
one if one country was part of the other in the past, and zero otherwise. As Egger et al. (2011) have
shown, CURCOLij and SMCTRYij are positively correlated with the formation of RTA between coun-
tries i and j. However, the historical status of country pairs is less likely to be correlated with patent
citations, because the technology and the research activity levels are important for the knowledge flow
but the historical relationship of countries is not necessarily important.
With regard to estimation techniques, we employ a two‐stage least squares estimation (IV/2SLS)
and a Poisson regression with endogenous regressors (IVP) as counterparts to the OLS and PPML
regressions that do not control for endogeneity issues. In the IVP, the Poisson model with continuous
endogenous covariates is estimated with the generalised method of moments.23
Table 6 reports the estimated results of the IV/2SLS and IVP estimations. Columns (1)–(4) in
Table 6 report the estimated results using ContagRTA as an IV, and columns (5)–(8) report the results
using CURCOL and SMCTRY as IVs. The IV/2SLS results are reported in columns (1) and (2), and the
IVP results are reported in columns (3) and (4). Similarly, columns (5) and (6) report the IV/2SLS re-
sults, and columns (7) and (8) report the IVP results. With regard to IV/2SLS, Wooldridge's robust
score test of endogeneity rejects the null hypothesis of exogeneity for both RTA and WTO‐X. Thus, it
would be adequate to treat RTA and WTO‐X as endogenous. In addition, the first stage F‐statistics sug-
gest that the null hypothesis of weak instrument is rejected for both sets of instruments (i.e. for
ContagRTA in columns (1) and (2) and for CURCOL and SMCTRY in columns (5) and (6)). Furthermore,
the test of overidentifying restriction indicates that the null hypothesis of CURCOL and SMCTRY
being valid instruments is not rejected for both IV/2SLS (columns (5) and (6)) and IVP (columns (7)
and (8)). Thus, we conclude that the overidentifying restriction is valid for both IV/2SLS and IVP.24
The estimated results in Table 6 indicate that the RTA dummy and WTO‐X index are positively
related with technology spillovers even after the endogeneity issue is addressed, though the estimated
coefficient of the RTA dummy is statistically insignificant in column (7). Thus, our results suggest that
both the formation of RTA and the depth of integration actually matter for technology spillovers.
Moreover, compared with Table 2, the estimates of the RTA dummy and WTO‐X index generally be-
come larger after employing the IV estimations. This implies that the assumption of exogenous RTAs
leads to a downward bias in the estimated coefficients of RTA and WTO‐X.25 This suggests that the
RTA dummy and WTO‐X index are negatively correlated with the error term. The reason is rather in-
tuitive. The number of patent citations, that is, our dependent variable, tends to be much larger for
pairs of developed countries, such as the US, European countries and Japan. In contrast, RTAs have

22
We also employed other bilateral gravity variables, such as Distij and Contigij, as another set of IVs, where Distij is bilateral
great circle distance between i and j, and Contigij is a dummy that equals one if i and j are contiguous. Although the estimation
results using this set of IVs do not qualitatively change, those IVs do not pass or only weakly pass the test of overidentifying
restriction. In addition, it is questionable whether those variables are conceptually uncorrelated with patent citations. Therefore,
we do not report estimation results using those IVs. However, the results are available from the corresponding author upon
request.
23
The IVP estimation is implemented by using Stata's command ivpoisson, where the model is specified by using additive
error terms.
24
With regard to IVP, we only report the results of the overidentifying restriction tests because other tests cannot be performed
for IVP.
25
Egger et al. (2011) report a similar effect of endogenous RTAs in estimations for bilateral trade.
2342
TABLE 6 IV/2SLS and IVP estimates for endogenous RTA and WTO‐X

|
(1) (2) (3) (4) (5) (6) (7) (8)

2SLS 2SLS IVP IVP 2SLS 2SLS IVP IVP

RTA 1.077*** 0.317*** 3.226*** 1.329
(0.180) (0.070) (0.681) (0.944)
WTO‐X 0.018*** 0.005*** 0.060*** 0.011**
(0.003) (0.001) (0.020) (0.005)
WTO 0.172*** 0.233*** 0.245* 0.193 0.079 0.274*** 0.457* 0.225
(0.060) (0.060) (0.130) (0.125) (0.097) (0.082) (0.276) (0.139)
ln (Pi) 0.386*** 0.439*** 0.594*** 0.618*** 0.383*** 0.565*** 0.515*** 0.624***
(0.026) (0.029) (0.036) (0.033) (0.035) (0.077) (0.111) (0.043)
ln (Pj) 0.485*** 0.541*** 0.827*** 0.858*** 0.515*** 0.711*** 0.721*** 0.857***
(0.029) (0.033) (0.073) (0.067) (0.039) (0.095) (0.158) (0.078)
PROX 1.612*** 1.535*** 1.793*** 1.958*** 1.386*** 1.084*** 1.291* 1.893***
(0.087) (0.097) (0.434) (0.423) (0.127) (0.242) (0.726) (0.498)
Test of over identifying restriction
Score chi2(1) [p‐value] 2.01 [0.16] 2.39 [0.12] 0.21 [0.65] 0.01 [0.92]
Test of endogeneity
Robust score chi2(1) 80.04 [0.00] 112.48 [0.00] 69.34 [0.00] 49.33 [0.00]
[p‐value]
F statistics in 1st stage 82.51 [0.00] 67.74 [0.00] 44.96 [0.00] 33.54 [0.00]
[p‐value]
No. of observations 26,597 26,597 169,624 169,624 26,607 26,607 169,916 169,916
Notes: IV/2SLS estimations are implemented using Stata's command ivregress. IVP estimations are implemented using Stata's command ivpoisson. The dependent variable is ln (Cijt) in
IV/2SLS and Cijt in IVP. In (1)–(4) ContagRTA is used as an IV and in (5)–(8) CURCOL and SMCTRY are used as IVs. The test results for overidentifying restriction and endogeneity for “2SLS” are

JINJI et al.
obtained from the regressions with robust errors. ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Standard errors clustered by country‐pair are in parentheses. All regressions
include citing and cited country fixed effects, year dummy, and the constant term.

not been formed between the US and the EU, between the US and Japan, or between the EU and Japan
during our sample period. Therefore, it is understandable that the RTA dummy and the WTO‐X index
are negatively correlated with the error term.

6.4 | The depth of RTAs

So far, we have measured the depth of integration in RTAs by the sum of the legal enforceability
points of all WTO‐X provisions. However, the WTO‐X index may be too broad to capture the impact
of the deep integration for our purpose, because the WTO‐X provisions actually cover various issues.
Only a small number of the WTO‐X provisions seem to be directly related to technology or R&D
activities. To investigate whether some of the WTO‐X areas are more strongly related to technology
spillovers than others, we categorise the WTO‐X areas into some subsets and construct new WTO‐X
indexes that represent such subsets.
To avoid selecting the WTO‐X provisions arbitrarily, we classify them into some subsets by utilising
the factor analysis. The procedure is as follows. We first conduct a factor analysis on the data for the le-
gally enforceable points of the WTO‐X areas.26 The factor analysis enables us to find underlying patterns
in the data and to extract a smaller number of (typically uncorrelated) unobserved variables, or factors,
that could largely explain variations in observed variables (most of which are correlated to each other). In
our data, the observed variables are the legally enforceable points of the WTO‐X areas in various RTAs.
The estimated factor loadings for each WTO‐X area with respect to identified factors indicate how each
WTO‐X area could be characterised by those factors. Then, based on the result of the factor analysis, we
categorise the WTO‐X areas that have a high value of factor loading (namely, factor loading larger than
0.40) to one particular factor into one subset. After this procedure, we create three subsets.
Table 7 reports the results of the factor analysis. Of the 38 WTO‐X areas identified by Horn et al.
(2010), 12 areas are rarely included in RTAs. Thus, we only include 26 WTO‐X areas in our analysis,
which raises the Kaiser–Meyer–Olkin (KMO) measure from 0.58 to 0.63. Only factor loadings larger
than 0.40 are reported in the table. As shown, 10 WTO‐X areas have high factor loading values for
Factor1. Those areas include consumer protection, education and training, health, data protection and
statistics. Thus, Factor1 seems to represent a healthy economic environment in a general sense. For
Factor2, seven areas have high factor loadings, including environmental laws, labour market regu-
lation, agriculture and energy. Hence, Factor2 may represent basic social and economic conditions.
Neither Factor1 nor Factor2 seems to be directly related to technology spillovers. Instead, they reflect
general and basic social and economic aspects of the WTO‐X provisions, and show the fundamentals
promoting trade between countries. In contrast, the WTO‐X areas that have high factor loadings for
Factor3 are competition policy, IPR, investment and movement of capital. Thus, this factor seems to
be more directly related to competition and technology.
Based on the results of the factor analysis, let 𝕏k be a subset of the WTO‐X areas that have a factor
loading higher than 0.40, for Factork, k = 1, 2, 3.27 Then, we construct a new measure of the WTO‐X,
namely Sh(WTO − Xk )ijt. This measure is calculated as the sum of the points of legally enforceable
WTO‐X provisions included in 𝕏k that are covered by RTAs to which both countries i and j belong in
year t, divided by the sum of the total points of all legally enforceable WTO‐X provisions covered by the

26
The source of the dataset is given in footnote 17. As explained in Section 2, each of the 100 RTAs in the dataset has a value
of 0, 1 or 2 for each of the 38 WTO‐X areas, depending on the evaluation of its legal enforceability. Thus, we have 38 variables
with 100 observations in the dataset for the factor analysis.
27
If we indicate each WTO‐X area by the number listed in Table 7, then 𝕏1 = {8,9,11,12,13,14,17,21,22,24},
𝕏2 = {3,6,10,15,16,24,25}, and 𝕏3 = {2,4,5,7}.
|

TABLE 7 Factor loadings of the factor analysis for WTO‐X

No. WTO‐X areas Factor1 Factor2 Factor3 Uniqueness

1 Anti‐corruption –
2 Competition policy 0.443 0.7642
3 Environmental laws 0.672 0.4176
4 IPR 0.654 0.5545
5 Investment 0.579 0.6440
6 Labour market regulation 0.693 0.4466
7 Movement of capital 0.606 0.5634
8 Consumer protection 0.650 0.3643
9 Data protection 0.500 0.5937
10 Agriculture 0.699 0.5000
11 Approximation of 0.596 0.6338
legislation
12 Cultural cooperation 0.782 0.4042
13 Economic policy dialog 0.803 0.2844
14 Education and training 0.523 0.4471
15 Energy 0.616 0.4466
16 Financial assistance 0.535 0.5124
17 Health 0.847 0.2969
18 Illegal Immigration –
19 Industrial cooperation –
20 Information society –
21 Research and technology 0.726 0.3486
22 SME 0.486 0.6227
23 Social matters 0.6774
24 Statistics 0.610 0.490 0.2227
25 Taxation 0.799 0.3836
26 Visa and asylum –
Notes: Factor loadings larger than 0.40 are reported. Estimations are implemented using the Stata command factor, excluding the
areas of anti‐corruption, illegal immigration, industrial cooperation, information society, and visa and asylum, which raise the
Kaiser–Meyer–Olkin (KMO) measure from 0.58 to 0.63. Rotations are performed with the Stata command oblique promax(2).

same RTAs. In other words, Sh(WTO-Xk )ijt indicates the share of the WTO‐X provisions included in 𝕏k
in the WTO‐X index for a pair of countries, i and j, in year t, and takes a value between zero and one. We
then include Sh(WTO-Xk )ijt as RTA_Variableijt in the OLS and PPML estimations of Equation (1).28
The results are reported in Table 8. All the estimated coefficients of Sh(WTO‐Xk) are positive and
significant. Thus, the major three aspects of the WTO‐X provisions are positively related to interna-
tional technology spillovers. Interestingly, the estimated coefficients of Sh(WTO‐X1) and Sh(WTO‐X2)
are larger than that of Sh(WTO‐X3). This means that the general and basic social and economic aspects
28
Since the correlation between Sh(WTO‐X1) and Sh(WTO‐X2) is high, we conduct estimations separately by including either
Sh(WTO‐X1) or Sh(WTO‐X2).
JINJI et al.
TABLE 8 Technology‐related WTO‐X provisions versus other WTO‐X provisions

(1) (2) (3) (4) (5) (6) (7) (8)

OLS OLS OLS OLS PPML PPML PPML PPML

Sh(WTO‐X1) 0.429*** 0.488*** 0.553*** 0.539***
(0.099) (0.113) (0.114) (0.114)
Sh(WTO‐X2) 0.502** 0.756*** 0.632*** 0.594***
(0.204) (0.225) (0.191) (0.198)
Sh(WTO‐X3) 0.137** 0.137** 0.235*** 0.225*** 0.255*** 0.177* 0.273*** 0.220**
(0.058) (0.058) (0.064) (0.063) (0.073) (0.098) (0.074) (0.099)
WTO 0.189*** 0.191*** 1.361*** 1.378*** 0.207* 0.222* 0.741** 0.753**
(0.059) (0.059) (0.297) (0.296) (0.125) (0.125) (0.313) (0.315)
ln (Pi) 0.371*** 0.376*** 0.600*** 0.602***
(0.025) (0.025) (0.031) (0.032)
ln (Pj) 0.467*** 0.472*** 0.840*** 0.838***
(0.028) (0.028) (0.067) (0.069)
PROX 1.717*** 1.718*** 2.107*** 2.107*** 1.933*** 1.847*** 2.790*** 2.691***
(0.085) (0.085) (0.121) (0.121) (0.413) (0.410) (0.338) (0.340)
Year FE Yes Yes Yes Yes
Citing country FE Yes Yes Yes Yes
Cited country FE Yes Yes Yes Yes
Citing country‐year FE Yes Yes Yes Yes
Cited country‐year FE Yes Yes Yes Yes
No. of observations 26,607 26,607 26,777 26,777 169,704 169,704 138,854 138,854

Notes: Estimations are implemented using the Stata command ppml for the PPML model. The dependent variable is ln (Cijt) in OLS and Cijt in PPML. ***,** and * denote 1%, 5% and 10% signifi-
cance levels, respectively. Standard errors clustered by country‐pair are in parentheses. The regressions include the constant term.

| 2345
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|

of the WTO‐X provisions are more important for technology spillovers than the WTO‐X provisions
that are specifically related to competition and technology.

7 | RO BU STN E SS CH E CK S

In this section, we conduct a number of robustness checks to see whether the results in Section 6 are
sensitive to our empirical strategy.

7.1 | Exclusion of the US

First, we check the robustness of our estimated results by excluding the US from the sample. Since
the patent statistics in our analysis are obtained from the USPTO, the number of patent citations made
or received by the US is disproportionately larger than those of other countries. Thus, the inclusion of
the US may distort the results.
The estimated results are reported in Table 9. The estimation procedure is the same as that shown
in Table 2, though we only report the results by the PPML estimations.

TABLE 9 Robustness check: estimation results by excluding the US

(1) (2) (3) (4)

PPML PPML PPML PPML

RTA 0.399*** 0.362***
(0.066) (0.062)
WTO‐X 0.001* 0.001*
(0.0005) (0.0005)
WTO 0.032 −0.020 0.803*** 0.900***
(0.34) (0.094) (0.280) (0.320)
ln (Pi) 0.683*** 0.706***
(0.029) (0.025)
ln (Pj) 0.837*** 0.864***
(0.049) (0.045)
PROX 2.156*** 2.398*** 2.638*** 2.874***
(0.214) (9.36) (0.214) (0.234)
Year FE Yes Yes
Citing country FE Yes Yes
Cited country FE Yes Yes
Citing country‐year FE Yes Yes
Cited country‐year FE Yes Yes
No. of observations 163,142 163,142 112,941 112,941
Notes: The sample excludes all country pairs that include the US. Estimations are implemented using the Stata command ppml for
the PPML model. The dependent variable is Cijt. ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Standard
errors clustered by country‐pair are in parentheses. The regressions include the constant term.
|

Compared to Table 2, it is evident that the coefficient estimates are not qualitatively changed by
excluding the US from the sample, except for insignificant coefficients of the WTO dummy in col-
umns (1) and (2).

7.2 | Controlling for institutional quality

One of the main findings in Section 6.4 is that the depth of integration in a broad sense is more impor-
tant for technology spillovers than are technology‐related provisions in RTAs. We next check whether
this finding is still true even after controlling for institutional quality of country pairs.29
To conduct this robustness check, we use two indexes of institutional quality from the data on the
quality of government (called the QoG dataset) provided by the Quality of Government Institute at the
University of Gothenburg (Dahlberg, Holmberg, Rothstein, Khomenko, & Svensson, 2017). The two
indexes are hf_business and fi_legprop.
The index of hf_business measures the business freedom and is constructed by 10 compo-
nents, such as the procedure and costs of starting a business, the procedure and time of obtaining a
licence, and so on, based on objective data from the World Bank's Doing Business study. This index
takes a value from zero to 100 with the highest freedom at 100. In the data, it actually takes a value
between 20 and 100.
On the other hand, fi_legprop is an index of legal structure and security of property right. The
index consists of judicial independence, impartial courts, protection of intellectual property, military
interference in rule of law and the political process, and integrity of the legal system. This index takes
a value from 0 to 10, where “0 corresponds to ‘no judicial independence’, ‘no trusted legal framework
exists’, ‘no protection of intellectual property’, ‘military interference in rule of law’, and ‘no integrity of
the legal system’ and 10 corresponds to ‘high judicial independence’, ‘trusted legal framework exists’,
‘protection of intellectual property’, ‘no military interference in rule of law’, and ‘integrity of the legal
system’” (Dahlberg et al., 2017, p. 61). In the data, it actually takes a value between 1.88 and 9.62.
The data on hf_business are available from 1995 and the data on fi_legprop are available
from 1970.
Using these indexes, we construct a similarity measure and a dissimilarity measure of institutional
quality for country pairs in the following way:30:
[ ]
hf _bus_simijt = ln (𝚑𝚏_𝚋𝚞𝚜𝚒𝚗𝚎𝚜𝚜it ) × (𝚑𝚏_𝚋𝚞𝚜𝚒𝚗𝚎𝚜𝚜jt ) ,
[ ]
hf _bus_diffijt = ln (𝚑𝚏_𝚋𝚞𝚜𝚒𝚗𝚎𝚜𝚜it )∕(𝚑𝚏_𝚋𝚞𝚜𝚒𝚗𝚎𝚜𝚜jt ) ,
[ ]
fi_leg_simijt = ln (𝚏𝚒_𝚕𝚎𝚐𝚙𝚛𝚘𝚙it ) × (𝚏𝚒_𝚕𝚎𝚐𝚙𝚛𝚘𝚙jt ) ,
[ ]
fi_leg_diffijt = ln (𝚏𝚒_𝚕𝚎𝚐𝚙𝚛𝚘𝚙it )∕(𝚏𝚒_𝚕𝚎𝚐𝚙𝚛𝚘𝚙jt ) .
hf_bus_simijt and hf_bus_diffijt measure similarity and dissimilarity in business freedom between i
and j at t, respectively. Similarly, fi_leg_simijt and fi_leg_diffijt measure similarity and dissimilarity in
legal structure and security of property right between i and j at t, respectively.
The results in Table 10 show that similarity and dissimilarity in institutional quality between two
countries actually matter for technology spillovers. hf_bus_sim is positive and significant in columns (5)

29
We thank an anonymous referee for raising this issue.
30
In the recent literature on trade and institutional quality, the difference in the institutional levels in two countries is used as
bilateral variables of institutional quality (Álvarez, Barbero, Rodríguez‐Pose, & Zofío, 2017; Yu, Beugelsdijk, & de Haan,
2015). We consider that the similarity of institutional quality may facilitate technology spillovers between two countries,
whereas the difference in institutional quality may impede them. Thus, we include both a similarity measure and a dissimilarity
measure of institutional quality in our estimations:
2348
TABLE 10 Robustness check: controlling for institutional quality

|
(1) (2) (3) (4) (5) (6) (7) (8)

PPML PPML PPML PPML PPML PPML PPML PPML
Sh(WTO‐X1) 0.619*** 0.654*** 0.556*** 0.649***
(0.125) (0.112) (0.049) (0.044)
Sh(WTO‐X2) 0.680*** 0.725*** 0.484*** 0.649***
(0.205) (0.184) (0.073) (0.073)
Sh(WTO‐X3) 0.139** 0.196*** 0.023 0.105 0.152*** 0.220*** 0.089*** 0.155***
(0.067) (0.073) (0.089) (0.097) (0.022) (0.026) (0.030) (0.035)
WTO 0.262** −0.206 0.270** −0.167 0.774*** 1.401*** 0.782*** 1.414***
(0.129) (0.194) (0.129) (0.194) (0.215) (0.271) (0.215) (0.276)
ln (Pi) 0.560*** 0.565*** 0.562*** 0.564***
(0.045) (0.054) (0.046) (0.056)
ln (Pj) 0.841*** 0.847*** 0.841*** 0.841***
(0.077) (0.070) (0.079) (0.074)
PROX 2.114*** 2.253*** 2.041*** 2.139*** 2.978*** 3.008*** 2.906*** 2.876***
(0.469) (0.421) (0.452) (0.420) (0.121) (0.123) (0.118) (0.125)
hf_bus_sim 0.121 0.114 2.508*** 2.584***
(0.077) (0.077) (0.333) (0.334)
hf_bus_diff −0.116 −0.117 −1.374*** −1.450***
(0.083) (0.085) (0.299) (0.301)
fi_leg_sim 0.293* 0.294* 1.726*** 1.775***
(0.153) (0.154) (0.218) (0.223)
fi_leg_diff −0.050 −0.051 −3.316*** −3.359***
(0.075) (0.075) (0.217) (0.222)

JINJI et al.
(Continues)

(1) (2) (3) (4) (5) (6) (7) (8)

PPML PPML PPML PPML PPML PPML PPML PPML

Year FE Yes Yes Yes Yes
Citing country FE Yes Yes Yes Yes
Cited country FE Yes Yes Yes Yes
Citing country‐year FE Yes Yes Yes Yes
Cited country‐year FE Yes Yes Yes Yes
No. of observations 110,186 75,318 110,186 75,318 91,494 65,547 91,494 65,547
Notes: Estimations are implemented using the Stata command ppml. (b) The dependent variable is Cijt. ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Standard errors clustered
by country‐pair are in parentheses. The regressions include the constant term.

| 2349
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|

and (7), while it is insignificant in columns (1) and (3). fi_leg_sim is also positive and significant in all of
columns (2), (4), (6) and (8). On the other hand, hf_bus_diff and fi_leg_diff are negative and significant
when country‐year fixed effects are included (i.e. columns (5)–(8)), whereas they are insignificant when
year dummies and country fixed effects are included (i.e. columns (1)–(4)). These results imply that
technology spillovers are enhanced as both countries enjoy business freedom at the similar level or they
have good legal structure and security of property right at the similar degree. By contrast, a difference in
institutional quality (either business freedom or legal structure) may lower technology spillovers.
Importantly, the results in Table 10 clearly show that the findings in Table 8 do not qualitatively
change even when institutional quality is controlled for. That is, the depth of integration in a broad
sense is more important for technology spillovers than are the technology‐related provisions.31 The
only notable difference from Table 8 is that the estimated coefficient of Sh(WTO‐X3) is insignificant
in columns (3) and (4).

7.3 | Extending the coverage of RTAs

In the analysis of the depth of RTAs, the number of RTAs covered in our data is reduced from 125 to
100 due to the availability of data in the data source. As the information on the depth of integration is
not reflected for excluded 25 RTAs, our estimation results in Section 6.4 may be subject to the sample
selection bias.32
As we have mentioned in Section 5, Kohl et al. (2016) developed an updated dataset on the depth
of RTAs by substantially extending the coverage of RTAs to 296. Thus, the sample selection bias may
be mitigated by employing their dataset. However, their dataset covers only four WTO‐X policy areas,
such as capital mobility, competition, environment and labour. Using the notation in Section 6.4,
capital mobility and competition are included in 𝕏3, and environment and labour are included in 𝕏2
. There are no 𝕏1 policy areas in their dataset. Only two out of seven 𝕏2 areas and two out of four 𝕏3
areas are covered. Therefore, although the coverage of RTAs is broader, the number of WTO‐X policy
areas substantially decreases from 38. It is difficult to predict how this trade‐off between the coverage
of RTAs and the coverage of WTO‐X policy areas will affect the estimation results.
In the estimation, we use three variables. First, WTO–Xijt
′
is constructed in the same way as WTO‐Xijt
in Equation (1). However, WTO–Xijt
′
only reflects the points of the four WTO‐X policy areas. Second,
Sh(WTO-Xk′ )ijt, k = 2, 3, is constructed in the same way as Sh(WTO‐Xk)ijt in Section 6.4, except that 𝕏2
includes only environment and labour provisions and 𝕏3 includes only capital mobility and competi-
tion provisions. Since the sum of Sh(WTO-X2′ )ijt and Sh(WTO-X3′ )ijt is equal to one by definition, we
cannot jointly estimate Sh(WTO-X2′ )ijt and Sh(WTO-X3′ )ijt.
Table 11 shows the results by PPML estimations. Columns (1) and (4) indicate that the estimated
coefficient of the WTO‐X′ index is positive and significant, which is qualitatively the same as the
result in Table 2. Columns (2), (3), (5) and (6) show that the estimated coefficients of Sh(WTO-X2′ )
and Sh(WTO-X3′ ) are also positive and highly significant. However, unlike the results in Table 8, the
estimated coefficient of Sh(WTO-X3′ ) is slightly larger than that of Sh(WTO-X2′ ). As argued above, the
results in Table 11 are not directly comparable with those in Table 8 because of the difference in cov-
erage of WTO‐X areas. Thus, although the results in Table 8 may be subject to sample selection bias,
it cannot be concluded that the treatment of this issue changes our main results.

31
We also conduct estimations by using other indexes of institutional quality that are available in QoG dataset, such as hf_in-
vest (investment freedom), wbgi_gee (government effectiveness) and wbgi_rle (rule of law). The results are not quali-
tatively changed by using these indexes.
32
We thank an anonymous referee for pointing out this problem.
|

TABLE 11 Robustness check: extending the coverage of RTA

(1) (2) (3) (4) (5) (6)

PPML PPML PPML PPML PPML PPML

WTO‐X′ 0.031*** 0.032***
(0.004) (0.003)
Sh(WTO-X2′ ) 0.382*** 0.388***
(0.049) (0.042)
Sh(WTO-X3′ ) 0.430*** 0.412***
(0.050) (0.053)
WTO 0.218* 0.214* 0.223* 0.761** 0.774** 0.680**
(0.125) (0.124) (0.126) (0.321) (0.327) (0.286)
ln (Pi) 59.947*** 60.101*** 60.015***
(3.231) (3.214) (3.221)
ln (Pj) 83.585*** 83.736*** 83.925***
(6.910) (6.879) (6.806)
PROX 1.893*** 1.904*** 1.842*** 2.740*** 2.752*** 2.679***
(0.406) (0.400) (0.418) (0.334) (0.326) (0.354)
Year FE Yes Yes Yes
Citing country FE Yes Yes Yes
Cited country FE Yes Yes Yes
Citing country‐year FE Yes Yes Yes
Cited country‐year FE Yes Yes Yes
No. of observations 169,704 169,704 169,704 138,854 138,854 138,854
Notes: Estimations are implemented using the Stata command ppml. The dependent variable is Cijt. Standard errors clustered by
country‐pair are in parentheses. The regressions include the constant term.

7.4 | Alternative measures of technology spillovers

So far, we have used patent citations as a measure of technology spillovers. Although patent citations
are a good measure of technology spillovers, they are not the only measure of technology spillovers.
Thus, it is important to check whether our results are specific to the use of a particular measure of
technology spillovers.33
To address this issue, we employ two alternative measures: the first alternative measure is the data
on patents with foreign co‐inventors, and the second one is the data on exports of royalties and licens-
ing fees. The data source for those two measures is OECD's database ([Link]).
First, let CAijt be the number of patents applied jointly by inventors in countries i and j at year t to
the USPTO. Since co‐inventions represent the international collaboration in innovation, there must be
spillovers of knowledge between co‐inventors. However, there are three limitations to use the patent
co‐application data as a measure of technology spillovers in our analysis: first, unlike patent citation
data, the direction of knowledge flow is unknown in the patent co‐application data; second, since the
data are available only after 1999, the sample period is shortened; and third, the number of sample
countries is also reduced from 114 to 81 by combining our data with the patent co‐application data.

33
We thank an anonymous referee for pointing out this issue.
|

In spite of these limitations, it seems that the patent co‐application data are still a good alternative
measure of technology spillovers.
Second, let RLFijt be the real export value of royalties and licensing fees from country i to country
j at year t. This means that technology spills over from j to i in the form of patent use or licensing.
Following Baier and Bergstrand (2007), the nominal export values are deflated by using exporter GDP
deflators, which are taken from the World Bank's World Development Indicators (World Bank Open
Data: [Link] With regard to the use of RLFijt, the number of sample countries is
not reduced from 114 but the data are available only from 1995.
Now, we replace Cijt in Equation (1) by CAijt or RLFijt and employ the same estimation procedure
as that in Table 8. Here, we only report the results by the PPML model in Table 12. Interestingly,
the results in Table 12 do not qualitatively change from those in Table 8, except that ln (P) becomes
insignificant for both CA and RLF and PROX also becomes insignificant for RLF. In particular, the
results indicate that the estimated coefficients of Sh(WTO‐X1) and Sh(WTO‐X2) are larger than that of
Sh(WTO‐X3) for both CAijt and RLFijt. Therefore, we conclude that our main findings in Section 6.4
are not specific to the use of the particular measure of technology spillovers.

8 | CO NC LUSION S

This study investigated empirically whether RTAs enhance technology spillovers among member
countries. We used patent citation data from the USPTO as a proxy for technology spillovers. We
found that RTAs have a positive and significant impact on technology spillovers in the sense that pairs
of countries belonging to a common RTA have more patent citations. This finding is robust for differ-
ent estimation techniques. Addressing the endogeneity issue using instrumental variable techniques
generally makes the estimates of the RTA dummy coefficient larger. We also investigated whether
technology spillovers among members are larger when countries engage in a deep integration. For
this, we measured the depth of integration by the degree of legally enforceable obligations covered
by an RTA that are outside the current WTO mandate. We then found that the depth of RTAs has
a positive and significant impact on technology spillovers. When the effect of signing an RTA on
technology spillovers is controlled for, the additional effect of the deep RTAs remains significantly
positive for North–South and South–South integrations but becomes insignificant for North–North
integrations and the full sample. Moreover, our estimation results indicate that the legal obligations
related to basic social and economic conditions have a greater impact on technology spillovers than
do technology‐related provisions. We checked the robustness of this finding by conducting a number
of additional estimations and confirmed its robustness.
The primary contribution of this study is to provide a new finding that the depth of regional inte-
gration influences technology spillovers among members of RTAs. In addition, we found the average
effect of RTAs on technology spillovers to be positive and significant. Countries usually form RTAs
to facilitate trade in goods and services among members. If RTAs increase knowledge flows among
members (which our study confirmed), they can expect an improvement in the productivity of do-
mestic firms, as well as an increase in their research and development activities. These effects may be
particularly important for medium- and low‐income countries, in which indigenous firms try to absorb
superior technology from firms in technologically advanced countries.
A number of important policy implications are obtained from our findings. First, we provide evi-
dence that a deep regional integration is particularly important and effective for technology spillovers
between developed and developing countries and among developing countries. Therefore, developing
countries could mostly benefit from technology spillovers by pursuing deep regional integrations.
JINJI et al.
TABLE 12 Robustness check: alternative measures of technology spillovers

(1) (2) (3) (4) (5) (6) (7) (8)

PPML PPML PPML PPML PPML PPML PPML PPML

Sh(WTO‐X1) 2.152*** 2.244*** 1.602*** 1.915***
(0.415) (0.151) (0.332) (0.352)
Sh(WTO‐X2) 2.055*** 2.179*** 2.143*** 2.113***
(0.595) (0.223) (0.511) (0.558)
Sh(WTO‐X3) 1.000*** 0.806** 1.069*** 0.887*** 0.724*** 0.554*** 0.802*** 0.681***
(0.282) (0.321) (0.104) (0.116) (0.125) (0.143) (0.124) (0.138)
WTO 0.438* 0.480** 1.993*** 2.145*** 0.636*** 0.692*** 1.502*** 1.527***
(0.224) (0.239) (0.255) (8.32) (0.137) (0.147) (0.323) (0.322)
ln (Pi) 0.147 0.028 −0.047 −0.070
(0.198) (0.202) (0.142) (0.138)
ln (Pj) 0.449*** 0.477*** 0.108 0.092
(0.099) (0.107) (0.109) (0.119)
PROX 1.304*** 1.219*** 2.008*** 1.888*** 0.422** 0.328 0.419 0.217
(0.342) (0.352) (0.225) (8.05) (0.200) (0.209) (0.348) (0.364)
Year FE Yes Yes Yes Yes
Country FE Yes Yes Yes Yes
Country‐year FE Yes Yes Yes Yes
No. of observations 27,799 27,799 19,473 19,473 136,489 136,489 145,145 145,145
Notes: Estimations are implemented using the Stata command ppml. The dependent variable isCAijt, that is, the number of co‐applied patents by i and j at t in columns (1)–(4) and RLFijt, that is, real
exports of royalties and licensing fees from i to j at t in columns (5)–(8). ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Standard errors clustered by country‐pair are in

parentheses. The regressions include the constant term.

| 2353
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|

Second, according to our results, among various provisions in RTAs, those related to basic social and
economic conditions are more important for technology spillovers than technology‐related provisions.
This finding is quite robust for various measures of technology spillovers. It suggests that deep inte-
grations in terms of basic social and economic conditions can effectively enhance not only economic
relationship among signatories of the RTAs but also technology spillovers among them. On the other
hand, our finding implies that technology‐related provisions in current RTAs may need to be revised
to become more effective in facilitating technology spillovers. Third, it is premature to conclude from
our results that the additional positive effects of deep integrations on technology spillovers are negli-
gible among developed countries. This is mainly because major technologically advanced countries/
regions, such as the US, the EU and Japan, cite each other's patents very frequently but have not
formed RTAs yet, except for the Japan‐EU Economic Partnership Agreement (EPA) signed in 2018.
As we discussed in Section 6.3, this situation seems to cause a downward bias in the estimates of the
RTA dummy and the WTO‐X index when we treat them as exogenous. Taking this into account, fur-
ther analysis is required to see whether there is any difference between deep and shallow integrations
in enhancing technology spillovers among developed countries.
Since our results are based on USPTO data, possible future research could check the robustness of
the results utilising patent statistics from other patent offices, for example, the EPO and JPO.

ACKNOWLEDGMENTS
We thank Masahiro Endoh, Gene Grossman, Kazunobu Hayakawa, Jota Ishikawa, Fuku Kimura, Kozo
Kiyota, Hyun‐Hoon Lee, Sawako Maruyama, Hiroshi Mukunoki, Toshi Okubo, Jee‐Hyeong Park, Andrew
Rose, and participants of Fall 2016 Midwest International Economics Meeting, the 14th International
Conference for the Western Economic Association International, the Kansai Branch workshop of the Japan
Society of International Economics, and seminars at Kyoto Sangyo University and Keio University for their
helpful comments and suggestions on earlier versions of the paper. We also thank three anonymous referees
for their valuable comments and suggestions. We are grateful to Tristan Kohl for generously providing
us his dataset. Hiroaki Sakamoto provided excellent research assistance. Financial support from the Japan
Society for the Promotion of Science under the Grant‐in‐Aid for Scientific Research (B) Nos. 23330081 and
16H03619 is gratefully acknowledged. The authors are solely responsible for any remaining errors.

R E F E R E NC E S
Acharya, Ram. C., & Keller, W. (2009). Technology transfer through imports. Canadian Journal of Economics, 42,
1411–1448. [Link]
Alcácer, J., & Gittelman, M. (2006). Patent citations as a measure of knowledge flows: The influence of examiner cita-
tions. Review of Economics and Statistics, 88, 774–779. [Link]
Álvarez, I. C., Barbero, J., Rodríguez‐Pose, A., & Zofío, J. L. (2017). Does institutional quality matter for trade?
Institutional conditions in a sectoral trade framework. World Development, 103, 72–87.
Anderson, James. E., & van Wincoop, E. (2003). Gravity with gravitas: A solution to the border puzzle. American
Economic Review, 93, 170–192. [Link]
Baier, Scott. L., & Bergstrand, Jeffrey. H. (2007). Do free trade agreements actually increase members’ international
trade? Journal of International Economics, 71, 72–95. [Link]
Baier, Scott. L., Bergstrand, Jeffrey. H., & Feng, M. (2014). Economic integration agreements and the margins of in-
ternational trade. Journal of International Economics, 93, 339–350. [Link]
Baldwin, R. (1995). A domino theory of regionalism. In R. Baldwin, P. Haaparanta, & A. Kiander (Eds.), Expanding
membership of the European Union (pp. 25–47). Cambridge, UK: Cambridge University Press. [Link]
org/10.1017/CBO9780511983368
|

Baldwin, R. (2011). 21st century regionalism: Filling the gap between 21st century trade and 20th century trade rules.
World Trade Organization Staff Working Paper ERSD‐2011‐08. [Link]
d201108_e.pdf
Baldwin, R., & Jaimovich, D. (2012). Are free trade agreements contagious? Journal of International Economics, 88,
1–16. [Link]
Behar, A., & Cirera‐i‐Crivillé, L. (2013). Does it matter who you sign with? Comparing the impacts of North‐South
and South‐South trade agreements on bilateral trade. Review of International Economics, 21, 765–782. [Link]
org/10.1111/roie.12069
Branstetter, L. (2006). Is foreign direct investment a channel of knowledge spillovers? Evidence from Japan's FDI in
the United States. Journal of International Economics, 68, 325–344. [Link]
Bustos, P. (2011). Trade liberalization, exports, and technology upgrading: Evidence on the impact of MERCOSUR on
Argentinian firms. American Economic Review, 101, 304–340. [Link]
Coe, David. T., & Helpman, E. (1995). International R&D spillovers. European Economic Review, 39, 859–887. https://
[Link]/10.1016/0014-2921(94)00100-E
Criscuolo, P., & Verspagen, B. (2008). Does it matter where patent citations come from? Inventor vs. examiner citations
in European patents. Research Policy, 37, 1892–1908. [Link]
Dahlberg, S., Holmberg, S., Rothstein, B., Khomenko, A., & Svensson, R. (2017). The quality of government basic
dataset, version Jan 17. Gothenburg, Sweden: The Quality of Government Institute, University of Gothenburg.
[Link]
Das, G. G., & Andriamananjara, S. (2006). Hub‐and‐spokes free trade agreements in the presence of technology spill-
overs: An application to the Western Hemisphere. Review of World Economics, 142, 33–66. [Link]
s10290-006-0056-x
Duguet, E., & MacGarvie, M. (2005). How well do patent citations measure flows of technology? Evidence
from French Innovation surveys. Economics of Innovation and New Technologies, 14, 375–393. [Link]
org/10.1080/1043859042000307347
Dür, A., Baccini, L., & Elsig, M. (2014). The design of international trade agreements: Introducing a new dataset.
Review of International Organizations, 9, 353–375. [Link]
Ederington, J., & McCalman, P. (2008). Endogenous firm heterogeneity and the dynamics of trade liberalization.
Journal of International Economics, 74, 422–440. [Link]
Egger, P., Larch, M., Staub, Kevin. E., & Winkelmann, R. (2011). The trade effects of endogenous preferential trade
agreements. American Economic Journal: Economic Policy, 3, 113–143.
Falley, T. (2015). Structural gravity and fixed effects. Journal of International Economics, 97, 76–85. [Link]
org/10.1016/[Link].2015.05.005
Hall, B. H., Jaffe, Adam. B., & Trajtenberg, M. (2001). The NBER patent citations data file: Lessons, insights and
methodological tools. NBER Working Paper No. 8498. [Link]
Hall, B. H., Jaffe, Adam. B., & Trajtenberg, M. (2005). Market value and patent citations. Rand Journal of Economics,
36, 16–38.
Haruna, S., Jinji, N., & Zhang, X. (2010). Patent citations, technology diffusion, and international trade: Evidence from
Asian countries. Journal of Economics and Finance, 34, 365–390. [Link]
Haskel, Jonathan. E., Pereira, Sonia. C., & Slaughter, Matthew. J. (2007). Does inward foreign direct investment boost
the productivity of domestic firms? Review of Economics and Statistics, 89, 482–49. [Link]
rest.89.3.482
Head, K., & Mayer, T. (2014). Gravity equations: Workhorse, toolkit, and cookbook. In G. Gopinath, E. Helpman, & K.
Rogoff (Eds.), Handbook of International Economics (Vol. 4, pp. 131–195). Amsterdam, The Netherlands: Elsevier.
Hofmann, C., Osnago, A., & Ruta, M. (2018). The content of preferential trade agreements. World Trade Review. https://
[Link]/10.1017/S1474745618000071
Horn, H., Mavroidis, Petros. C., & Sapir, A. (2010). Beyond the WTO? An anatomy of EU and US preferential trade
agreements. The World Economy, 33, 1565–88. [Link]
Jaffe, Adam. B. (1986). Technological opportunity and spillovers of R&D: Evidence from firms’ patents, profits, and
market value. American Economic Review, 76, 984–1001.
Jaffe, Adam. B., & Trajtenberg, M. (1999). International knowledge flows: Evidence from patent citations. Economics
of Innovation & New Technology, 8, 105–136. [Link]
|

Jaffe, Adam. B., Trajtenberg, M., & Fogarty, Michael. S. (2000). Knowledge spillovers and patent citations: Evidence
from a survey of inventors. American Economic Review, Papers and Proceedings, 90, 215–218. [Link]
org/10.1257/aer.90.2.215
Jaffe, Adam. B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as evi-
denced by patent citations. Quarterly Journal of Economics, 108, 577–598. [Link]
Javorcik, B. S. (2004). Does foreign direct investment increase the productivity of domestic firms? In search of spillovers
through backward linkages. American Economic Review, 94, 605–627. [Link]
Jinji, N., Zhang, X., & Haruna, S. (2013). The effect of regional trade agreements on technology spillovers through
international trade. In S. Egashira (Ed.), Globalism and regional economy (pp. 49–66). Abingdon, UK: Routledge.
Keller, W. (2004). International technology diffusion. Journal of Economic Literature, 42, 752–782. [Link]
org/10.1257/0022051042177685
Keller, W., & Yeaple, Stephen. R. (2009). Multinational enterprises, international trade, and productivity growth: Firm‐
level evidence from the United States. Review of Economics and Statistics, 91, 821–831. [Link]
rest.91.4.821
Kohl, T. (2014). Do we really know that trade agreements increase trade? Review of World Economics, 150, 443–469.
[Link]
Kohl, T., Brakman, S., & Garretsen, H. (2016). Do trade agreements stimulate international trade differently? Evidence
from 296 Trade Agreements. The World Economy, 39, 97–131. [Link]
MacGarvie, M. (2006). Do firms learn from international trade? Review of Economics and Statistics, 88, 46–60.
Mancusi, M. L. (2008). International spillovers and absorptive capacity: A cross‐country cross‐sector analysis based on
patents and citations. Journal of International Economics, 76, 155–165. [Link]
Mattoo, A., Mulabdic, A., & Ruta, M. (2017). Trade creation and trade diversion in deep agreements. World Bank
Policy Research Working Paper No. 8206. [Link]
[Link]
Maurseth, P. B., & Verspagen, B. (2002). Knowledge spillovers in Europe: A patent citations analysis. Scandinavian
Journal of Economics, 104, 531–545. [Link]
Orefice, G., & Rocha, N. (2014). Deep integration and production networks: An empirical analysis. The World Economy,
37, 106–36. [Link]
Peri, G. (2005). Determinants of knowledge flows and their effect on innovation. Review of Economics and Statistics,
87, 308–322. [Link]
Roy, J. (2010). Do customs union members engage in more bilateral trade than free‐trade agreement members? Review
of International Economics, 18, 663–681. [Link]
Santos Silva, J. M. C., & Tenreyro, S. (2006). The log of gravity. Review of Economics and Statistics, 88, 641–658.
[Link]
Santos Silva, J. M. C., & Tenreyro, S. (2011). Further simulation evidence on the performance of the Poisson pseudo‐
maximum likelihood estimator. Economics Letters, 112, 220–222. [Link]
Schiff, M., & Wang, Y. (2003). NAFTA, technology diffusion and productivity in Mexico. Cuadernos de Economía,
40, 469–476.
Viner, J. (1950). The customs union issue. New York, NY: Carnegie Endowment for International Peace.
Xu, B., & Wang, J. (1999). Capital goods trade and R&D spillovers in the OECD. Canadian Journal of Economics, 32,
1258–1274. [Link]
Yu, S., Beugelsdijk, S., & de Haan, J. (2015). Trade, trust and the rule of law. European Journal of Political Economy,
37, 102–115. [Link]

How to cite this article: Jinji N, Zhang X, Haruna S. Do deeper regional trade agreements
enhance international technology spillovers?. World Econ. 2019;42:2326–2363. [Link]
org/10.1111/twec.12797
|

APPENDIX

TABLE A1 Descriptive statistics

Variable No. of Obs Mean Std. Dev. Min Max

Cij 197,792 81.2757 2,249.4170 0.0000 245431
ln (Pi) 182,571 4.5934 3.3166 −1.2439 14.3075
ln (Pj) 182,571 4.5934 3.3166 −1.2439 14.3075
PROX 197,792 0.1030 0.1715 0.0000 0.9868
RTA 197,792 0.0923 0.2895 0.0000 1.0000
WTO‐X 197,792 2.4942 14.8014 0.0000 184.0000
WTO 197,792 0.6223 0.4848 0.0000 1.0000
Sh(WTO‐X1) 197,792 0.0127 0.0711 0.0000 0.5263
Sh(WTO‐X2) 197,792 0.0076 0.0422 0.0000 0.3333
Sh(WTO‐X3) 197,792 0.0365 0.1565 0.0000 1.0000
ContagRTA 197,262 0.0036 0.0169 0.0000 0.6838
Curcol 197,792 0.0003 0.0185 0.0000 1.0000
Smctry 197,792 0.0107 0.1030 0.0000 1.0000
hf_bus_sim 115,823 8.3612 0.2913 6.2841 9.2103
hf_bus_diff 115,823 0.0000 0.3030 −1.6084 1.6084
fi_leg_sim 76,460 3.4698 0.4480 1.5055 4.5196
fi_leg_diff 76,460 0.0000 0.4536 −1.5913 1.5913
CAij 56,862 5.7252 58.2686 0.0000 1,926
RLFij 145,145 19.8592 1612.8670 0.0000 578,499.6
WTO‐X′ 197,792 0.5880 1.8506 0.0000 8.0000
Sh(WTO-X2′ ) 197,792 0.0314 0.1168 0.0000 1.0000
Sh(WTO-X3′ ) 197,792 0.0769 0.2328 0.0000 1.0000
TABLE A2 Correlations of the variables

2358
Cij ln (Pi) ln (Pj) PROX RTA WTO‐X WTO

|
Cij 1.0000
ln (Pi) 0.0893 1.0000
ln (Pj) 0.0898 −0.0364 1.0000
PROX 0.1418 0.5271 0.5271 1.0000
RTA 0.0049 0.1072 0.1072 0.2067 1.0000
WTO‐X 0.0046 0.1684 0.1684 0.2902 0.5357 1.0000
WTO 0.0270 0.2063 0.2063 0.2364 0.0740 0.1217 1.0000
Sh(WTO‐X1) −0.0002 0.1640 0.1640 0.2746 0.5718 0.7703 0.1250
Sh(WTO‐X2) 0.0150 0.0921 0.0921 0.1928 0.5695 0.6237 0.1081
Sh(WTO‐X3) 0.0053 0.1295 0.1295 0.1926 0.7281 0.3345 0.0897
ContagRTA 0.0576 −0.0139 0.3521 0.2013 0.1711 0.1977 0.0620
Curcol −0.0008 −0.0032 −0.0032 −0.0053 0.0088 0.0229 0.0141
Smctry −0.0037 −0.0308 −0.0307 −0.0079 0.1632 0.0328 0.0391
hf_bus_sim 0.0697 0.3629 0.3629 0.4155 0.2213 0.2653 0.1370
hf_bus_diff 0.0001 0.4096 −0.4096 0.0000 0.0002 0.0001 0.0002
fi_leg_sim 0.0700 0.4510 0.4510 0.5265 0.2824 0.2818 0.0700
fi_leg_diff −0.0006 0.4981 −0.4981 −0.0000 0.0001 0.0001 −0.0000
CAij 0.6631 0.1815 0.2159 0.2446 0.0498 0.0593 0.0445
RLFij 0.7309 0.1461 0.1074 0.2005 0.0351 0.0500 0.0405
WTO‐X′ 0.0050 0.1828 0.1828 0.2884 0.6930 0.6475 0.1568
Sh(WTO-X2′ ) 0.0081 0.1587 0.1587 0.2619 0.6260 0.6436 0.1396
Sh(WTO-X3′ ) −0.0005 0.1260 0.1260 0.1982 0.6793 0.3207 0.1201

JINJI et al.
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
JINJI et al.
TABLE A2 (Continued)
Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) ContagRTA Curcol Smctry
Sh(WTO‐X1) 1.0000
Sh(WTO‐X2) 0.6291 1.0000
Sh(WTO‐X3) 0.2921 0.3270 1.0000
ContagRTA 0.1960 0.1373 0.1351 1.0000
Curcol 0.0124 −0.0008 0.0130 0.0168 1.0000
Smctry 0.0297 0.1311 0.0735 0.0277 0.0476 1.0000

Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) hf_bus_sim hf_bus_diff fi_leg_sim fi_leg_diff

hf_bus_sim 0.2526 0.2020 0.1824 1.0000
hf_bus_diff 0.0002 0.0001 0.0002 0.0001 1.0000
fi_leg_sim 0.2924 0.2244 0.2438 0.6430 0.0000 1.0000
fi_leg_diff 0.0001 0.0000 −0.0000 −0.0000 0.6715 0.0001 1.0000

Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) CAij

CAij 0.0299 0.0724 0.0434 1.0000
Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) RLFij
RLFij 0.0266 0.0470 0.0401 1.0000
Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) WTO‐ X′ Sh(WTO-X2′ ) Sh(WTO-X3′ )

WTO‐X′ 0.6399 0.5596 0.5565 1.0000

Sh(WTO-X2′ ) 0.6184 0.5940 0.4989 0.9103 1.0000
Sh(WTO-X3′ ) 0.3591 0.3733 0.5463 0.7588 0.5224 1.0000

| 2359
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE A3 List of RTAs

2360
RTA Name RTA Type RTA Name RTA Type RTA Name RTA Type

|
ASEAN Free Trade Area FTA European Free Trade Association FTA&EIA Mexico–Guatemala FTA&EIA
(AFTA) (EFTA)
Mexico–Honduras FTA&EIA
Armenia–Kazakhsta FTA EFTA–Chile FTA&EIA Mexico–Nicaragua FTA&EIA
Armenia–Moldova FTA EFTA–Croatia FTA Mexico–El Salvador FTA&EIA
Armenia–Russian FTA EFTA–Egypt FTA North American Free FTA&EIA
Federation
Armenia–Ukraine FTA EFTA–Israel FTA Trade Agreement
Australia–New Zealand FTA&EIA EFTA–Jordan FTA (NAFTA)
(ANZCERTA) EFTA–Korea, Republic of FTA&EIA New Zealand– FTA&EIA
Central American CU EFTA–Lebanon FTA Singapore
Common
Market (CACM) EFTA–Morocco FTA Pan‐Arab Free Trade FTA
Canada–Chile FTA&EIA EFTA–Mexico FTA&EIA Area (PAFTA)
Canada–Costa Rica FTA EFTA–Singapore FTA&EIA Panama–Singapore FTA&EIA
Canada–Israel FTA EFTA–Former Yugoslav FTA Panama–El Salvador FTA&EIA
Andean Community CU Republic of Macedonia Pacific Island FTA
(CAN)
Caribbean Community CU&EIA EFTA–Tunisia FTA Countries Trade
and Common Market EFTA–Turkey FTA Agreement (PICTA)
(CARICOM) Egypt–Turkey FTA Pakistan–China FTA&EIA
Central European Free FTA EU–Albania FTA Pakistan–Sri Lanka FTA
Trade Agreement EU–Chile FTA Southern African FTA
(CEFTA) 2006 EU–Croatia FTA Development
Economic and Monetary CU EU–Algeria FTA Community (SADC)

JINJI et al.
(Continues)

RTA Name RTA Type RTA Name RTA Type RTA Name RTA Type
Community of Central EU–Egypt FTA South Asian Free FTA
Africa (CEMAC) EU–Israel FTA Trade Agreement
Common Economic Zone FTA EU–Iceland FTA (SAFTA)
(CEZ) EU–Jordan FTA Singapore–Australia FTA&EIA
Commonwealth of FTA EU–Lebanon FTA Thailand–Australia FTA&EIA
Independent States (CIS) EU–Morocco FTA Thailand–New Zealand FTA&EIA
Chile–China FTA&EIA EU–Former Yugoslav FTA&EIA Trans‐Pacific Strategic FTA&EIA
Chile–Costa Rica FTA&EIA Republic of Macedonia Economic Partnership
Chile–Japan FTA&EIA EU–Mexico FTA&EIA Turkey–Croatia FTA
Chile–Mexico FTA&EIA EU–Norway FTA Turkey–Israel FTA
Chile–El Salvador FTA&EIA EU–Switzerland ‐ FTA Turkey–Morocco FTA
China–Hong Kong China FTA&EIA Liechtenstein Turkey–Syria FTA
Common Market for CU EU–Syria FTA Turkey–Former FTA
Eastern and Southern EU–Tunisia FTA Yugoslav Republic of
Africa (COMESA) EU–Turkey CU Macedonia
Colombia–Mexico FTA&EIA EU–South Africa FTA Turkey–Tunisia FTA
Costa Rica–Mexico FTA&EIA Gulf Cooperation CU Ukraine–Azerbaijan FTA
Dominican Republic ‐ FTA&EIA Council (GCC) Ukraine–Belarus FTA
Central America Georgia–Armenia FTA Ukraine–Kazakhstan FTA
Dominican Republic ‐ FTA&EIA Georgia–Azerbaijan FTA Ukraine–Moldova FTA
Central America–United Georgia–Kazakhstan FTA Ukraine–Former FTA
States Free Trade Georgia–Russian FTA Yugoslav Republic of

Agreement (CAFTA‐DR) Federation Macedonia

|
(Continues)

2361
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2362
TABLE A3 (Continued)

|
RTA Name RTA Type RTA Name RTA Type RTA Name RTA Type

East African Community CU Georgia–Ukraine FTA Ukraine–Russian FTA
(EAC) Israel–Mexico FTA Federation
Eurasian Economic CU India–Singapore FTA&EIA Ukraine–Uzbekistan FTA
Community (EAEC) India–Sri Lanka FTA US–Australia FTA&EIA
EC Treaty CU&EIA Jordan–Singapore FTA&EIA US–Bahrain FTA&EIA
EC(12) Enlargement CU Japan–Mexico FTA&EIA US–Chile FTA&EIA
EC(15) Enlargement CU&EIA Japan–Malaysia FTA&EIA US–Israel FTA
EC(25) Enlargement CU&EIA Japan–Singapore FTA&EIA US–Jordan FTA&EIA
EC(27) Enlargement CU&EIA Japan–Thailand FTA&EIA US–Morocco FTA&EIA
Economic Community of CU Korea Republic of ‐ FTA&EIA US–Singapore FTA&EIA
West African States Chile West African CU
(ECOWAS) Korea Republic of ‐ FTA&EIA Economic and
European Economic Area EIA Singapore Monetary Union
(EEA) Southern Common CU&EIA (WAEMU)
Market (MERCOSUR)
Notes: The listed RTAs are all RTAs notified to WTO that entered into force by 2006 and of which at least two countries/regions in our sample are signatories are included in the sample. RTA types are
free trade agreements (FTAs), customs unions (CUs), and economic integration agreements (EIAs).
Source: The web page of the WTO. [Link]

JINJI et al.
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
|
14679701, 2019, 8, Downloaded from [Link] by Shanghai University, Wiley Online Library on [20/02/2024]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

JINJI et al. 2363

TABLE A4 Sampled countries/regions
No. Country/region No. Country/region No. Country/region No. Country/region
1 Albania 31 Ethiopia 61 Libya 91 Sierra Leone
2 Algeria 32 Finland 62 Lithuania 92 Singapore
3 Angola 33 France 63 Macedonia 93 Slovak Republic
4 Argentina 34 Gambia 64 Madagascar 94 Slovenia
5 Armenia 35 Georgia 65 Malaysia 95 South Africa
6 Australia 36 Germany 66 Mali 96 Spain
7 Austria 37 Ghana 67 Malta 97 Srilanka
8 Azerbaijan 38 Greece 68 Mauritius 98 Sweden
9 Bahrain 39 Guatemala 69 Mexico 99 Switzerland
10 Bangladesh 40 Guinea 70 Moldva 100 Syria
11 Belarus 41 Honduras 71 Morocco 101 Tanzania
12 Belgium 42 Hongkong 72 Nepal 102 Thailand
13 Bolivia 43 Hungary 73 Netherlands 103 Tunisia
14 Brazil 44 Iceland 74 Newzealand 104 Turkey
15 Bulgaria 45 India 75 Nicaragua 105 Uganda
16 Cambodia 46 Indonesia 76 Niger 106 Ukraine
17 Canada 47 Iraq 77 Nigeria 107 United Kingdom
18 Chile 48 Ireland 78 Norway 108 United States
19 China 49 Israel 79 Oman 109 Uruguay
20 Colombia 50 Italy 80 Pakistan 110 Uzbekistan
21 Costa Rica 51 Japan 81 Paraguay 111 Venezuela
22 Croatia 52 Jordan 82 Peru 112 Vietnam
23 Cyprus 53 Kazakhstan 83 Philippines 113 Yemen Republic
24 Czech Republic 54 Kenya 84 Poland 114 Zimbabwe
25 Denmark 55 Korea 85 Portugal
26 Dominican Rep. 56 Kuwait 86 Qatar
27 Ecuador 57 LAOS 87 Romania
28 Egypt 58 Latvia 88 Russia
29 El Salvador 59 Lebanon 89 Saudi Arabia
30 Estonia 60 Liberia 90 Senegal

Trade Agreements and Tech Transfer Impact
No ratings yet
Trade Agreements and Tech Transfer Impact
35 pages
International R&D
No ratings yet
International R&D
28 pages
Iulianna Romanchyshyna - Technical Barriers To Trade in "New Generation" RTAs and in The WTO Agreements - Conflict or Complementarity - Springer Nature (2023)
No ratings yet
Iulianna Romanchyshyna - Technical Barriers To Trade in "New Generation" RTAs and in The WTO Agreements - Conflict or Complementarity - Springer Nature (2023)
277 pages
Understanding Technology Spillovers
No ratings yet
Understanding Technology Spillovers
12 pages
Regional Trade Liberalisation Explained
No ratings yet
Regional Trade Liberalisation Explained
25 pages
Understanding Regional Trade Agreements
No ratings yet
Understanding Regional Trade Agreements
32 pages
Intellectual Property and Technology Transfer
No ratings yet
Intellectual Property and Technology Transfer
13 pages
WIPO-WTO Workshop on Technology Transfer
No ratings yet
WIPO-WTO Workshop on Technology Transfer
13 pages
Intellectual Property and Technology Transfer
No ratings yet
Intellectual Property and Technology Transfer
13 pages
Trade Agreement Violations: ITI Comments
No ratings yet
Trade Agreement Violations: ITI Comments
19 pages
Regional Integration's Impact on Productivity
No ratings yet
Regional Integration's Impact on Productivity
29 pages
Regional Trade Integration and Growth
No ratings yet
Regional Trade Integration and Growth
29 pages
Knowledge Spillovers and Firm Productivity
No ratings yet
Knowledge Spillovers and Firm Productivity
20 pages
Technology Transfer and Its Impacts
No ratings yet
Technology Transfer and Its Impacts
2 pages
Trade Liberalization: RTAs vs WTO
No ratings yet
Trade Liberalization: RTAs vs WTO
1 page
Proximity Effects on Knowledge Spillovers
No ratings yet
Proximity Effects on Knowledge Spillovers
12 pages
Globalization's Impact on Tech Advancements
No ratings yet
Globalization's Impact on Tech Advancements
16 pages
Linking Patents and Economic Data Analysis
No ratings yet
Linking Patents and Economic Data Analysis
32 pages
Technology Spillovers in Major Economies
No ratings yet
Technology Spillovers in Major Economies
31 pages
Trade Agreements and Apparel Rules of Origin
No ratings yet
Trade Agreements and Apparel Rules of Origin
29 pages
WTO and RTAs/FTAs in International Economics
No ratings yet
WTO and RTAs/FTAs in International Economics
4 pages
EU-South Korea FTA Trade Effects Analysis
No ratings yet
EU-South Korea FTA Trade Effects Analysis
41 pages
Technology Transfer
No ratings yet
Technology Transfer
54 pages
WIPO Overview of Patents and Tech Transfer
No ratings yet
WIPO Overview of Patents and Tech Transfer
12 pages
India's Evolving Free Trade Strategies
No ratings yet
India's Evolving Free Trade Strategies
16 pages
Transfer of Technology To Developing Countries: Unilateral and Multilateral Policy Options
No ratings yet
Transfer of Technology To Developing Countries: Unilateral and Multilateral Policy Options
37 pages
RTAs New
No ratings yet
RTAs New
12 pages
Understanding Regional Trade Agreements
No ratings yet
Understanding Regional Trade Agreements
1 page
Technology Transfer in International Trade
No ratings yet
Technology Transfer in International Trade
12 pages
New Regionalism vs. Multilateralism Explained
No ratings yet
New Regionalism vs. Multilateralism Explained
7 pages
Regional Trade Agreements and WTO Impact
No ratings yet
Regional Trade Agreements and WTO Impact
10 pages
RWP12-020 Hausmann PDF
No ratings yet
RWP12-020 Hausmann PDF
60 pages
Overview of Regional Trade Agreements
100% (1)
Overview of Regional Trade Agreements
22 pages
Technology Transfer in South Korea MNCs
No ratings yet
Technology Transfer in South Korea MNCs
16 pages
United Nations Conference On Trade and Development
No ratings yet
United Nations Conference On Trade and Development
25 pages
Technology Transfer and Backflows in Auto Industry
No ratings yet
Technology Transfer and Backflows in Auto Industry
10 pages
Understanding Technological Globalization
No ratings yet
Understanding Technological Globalization
5 pages
Impact of Chinese Imports on Innovation
No ratings yet
Impact of Chinese Imports on Innovation
31 pages
1080 Virtual Inst PDF
No ratings yet
1080 Virtual Inst PDF
132 pages
Indian Trade Regionalism Trends
No ratings yet
Indian Trade Regionalism Trends
28 pages
Regional Integration in The Americas: State of Play, Lessons, and Ways Forward
No ratings yet
Regional Integration in The Americas: State of Play, Lessons, and Ways Forward
32 pages
R&D Spillovers, Patents and The Incentives To Innovate in Japan and The United States
No ratings yet
R&D Spillovers, Patents and The Incentives To Innovate in Japan and The United States
19 pages
Impact of FDI on Technology Transfer
No ratings yet
Impact of FDI on Technology Transfer
12 pages
Knowledge Spillovers and Proximity Analysis
No ratings yet
Knowledge Spillovers and Proximity Analysis
19 pages
Impact of Free Trade Agreements Analyzed
No ratings yet
Impact of Free Trade Agreements Analyzed
97 pages
Understanding Beggar-Thy-Neighbor Effects
No ratings yet
Understanding Beggar-Thy-Neighbor Effects
12 pages
Technology Transfer via Trade Analysis
No ratings yet
Technology Transfer via Trade Analysis
58 pages
Measuring Global Economic Integration
No ratings yet
Measuring Global Economic Integration
26 pages
Understanding Quantitative Restrictions in Trade
No ratings yet
Understanding Quantitative Restrictions in Trade
11 pages
Regional Trade Agreements: 2006 Update
No ratings yet
Regional Trade Agreements: 2006 Update
48 pages
Measuring International Technology Transfer
No ratings yet
Measuring International Technology Transfer
3 pages
Advantages of Regional Trade Agreements
No ratings yet
Advantages of Regional Trade Agreements
29 pages
Focal Firms as Gatekeepers in Innovation
No ratings yet
Focal Firms as Gatekeepers in Innovation
31 pages
WTO and RTA: Global Trade Dynamics
No ratings yet
WTO and RTA: Global Trade Dynamics
10 pages
Factors Affecting The Acquisition of Technology
No ratings yet
Factors Affecting The Acquisition of Technology
6 pages
Impact of Tech Innovation on Trade
No ratings yet
Impact of Tech Innovation on Trade
39 pages
Understanding Technology Transfer Dynamics
No ratings yet
Understanding Technology Transfer Dynamics
21 pages
RTAs
No ratings yet
RTAs
35 pages
Educational Institutions in Nokha, Rajasthan
No ratings yet
Educational Institutions in Nokha, Rajasthan
10 pages
Understanding Fixed and Floating Charges
No ratings yet
Understanding Fixed and Floating Charges
11 pages
Grade 5 Vocabulary and Definitions
No ratings yet
Grade 5 Vocabulary and Definitions
10 pages
Top 15 Real Estate Companies in Ethiopia
67% (9)
Top 15 Real Estate Companies in Ethiopia
5 pages
Teaching Grammar: Rule of Economy
No ratings yet
Teaching Grammar: Rule of Economy
3 pages
Bakels Product Price List 2023
100% (4)
Bakels Product Price List 2023
3 pages
ME5 Series Firmware Release Notes 1.2.0.3
No ratings yet
ME5 Series Firmware Release Notes 1.2.0.3
12 pages
4th Grade Science: Joints & Bones Lesson Plan
No ratings yet
4th Grade Science: Joints & Bones Lesson Plan
2 pages
High-Scoring Global Perspectives Report
No ratings yet
High-Scoring Global Perspectives Report
4 pages
Feeder X3 Spare Parts Overview
No ratings yet
Feeder X3 Spare Parts Overview
7 pages
Evolution of Military Cold Weather Gear
No ratings yet
Evolution of Military Cold Weather Gear
5 pages
NOVATION® Indulge 3340 Fact Sheet
No ratings yet
NOVATION® Indulge 3340 Fact Sheet
2 pages
Grade 9 English Lesson Plan Overview
No ratings yet
Grade 9 English Lesson Plan Overview
23 pages
MSBTE Winter 2024 OOP Question Paper
No ratings yet
MSBTE Winter 2024 OOP Question Paper
4 pages
IMechE UAS'24 Preliminary Design Review
No ratings yet
IMechE UAS'24 Preliminary Design Review
12 pages
Pump Room Block Layout Plan
No ratings yet
Pump Room Block Layout Plan
1 page
Algebra Whole Numbers, Arithmetical Operations, and Properties of Addition and Multiplication
No ratings yet
Algebra Whole Numbers, Arithmetical Operations, and Properties of Addition and Multiplication
3 pages
PLC Latching Logic Explained
No ratings yet
PLC Latching Logic Explained
110 pages
Unit 3 Vocabulary Quiz: On the Go
No ratings yet
Unit 3 Vocabulary Quiz: On the Go
1 page
Magnesium in Obesity, Metabolic Syndrome, 2021
No ratings yet
Magnesium in Obesity, Metabolic Syndrome, 2021
16 pages
Importance of Studying Contemporary Art
No ratings yet
Importance of Studying Contemporary Art
1 page
Machine Learning Overview and Techniques
No ratings yet
Machine Learning Overview and Techniques
120 pages
ICT Connect Magazine - Nov 2010 Issue
No ratings yet
ICT Connect Magazine - Nov 2010 Issue
24 pages
Forensic Age Estimation via Iliac Crest X-Rays
No ratings yet
Forensic Age Estimation via Iliac Crest X-Rays
6 pages
Electric Field Calculations for Point Charges
No ratings yet
Electric Field Calculations for Point Charges
1 page
CCTV Expo 2023: Security Solutions Showcase
No ratings yet
CCTV Expo 2023: Security Solutions Showcase
4 pages
Electron Cloud Lab
33% (6)
Electron Cloud Lab
4 pages
Plant Body Organization: Tissues & Organs
100% (1)
Plant Body Organization: Tissues & Organs
20 pages
Prospecting Techniques by Ben Feldman
No ratings yet
Prospecting Techniques by Ben Feldman
10 pages
VSA Exam Questions Overview 2023
No ratings yet
VSA Exam Questions Overview 2023
11 pages

Impact of RTAs on Technology Spillovers

Uploaded by

Impact of RTAs on Technology Spillovers

Uploaded by

Received: 4 June 2018

| Revised: 10 December 2018

Do deeper regional trade agreements enhance

Naoto Jinji1 | Xingyuan Zhang2 | Shoji Haruna3

2 | T H E LIN K B E T WE E N RTA S AND

TABLE 1 List of WTO+ and WTO‐X areas in RTAs

WTO+ areas WTO‐X areas

3 | T E CH NOLOGY SP IL LOV E RS AND PATENT CITATIONS

6.1 | Baseline estimations

6.2 | The heterogeneous impacts of RTAs on technology spillovers

TABLE 4 Heterogeneous impacts of RTA on technology spillovers: FTA versus CU

(1) (2) (3) (4)

PPML PPML PPML PPML

6.3 | Estimations with instrumental variables

TABLE 5 Heterogeneous impacts of RTA on technology spillovers: phased‐in effects

(1) (2) (3) (4) (5) (6)

PPML PPML PPML PPML PPML PPML

2SLS 2SLS IVP IVP 2SLS 2SLS IVP IVP

6.4 | The depth of RTAs

TABLE 7 Factor loadings of the factor analysis for WTO‐X

No. WTO‐X areas Factor1 Factor2 Factor3 Uniqueness

(1) (2) (3) (4) (5) (6) (7) (8)

OLS OLS OLS OLS PPML PPML PPML PPML

7.1 | Exclusion of the US

TABLE 9 Robustness check: estimation results by excluding the US

(1) (2) (3) (4)

PPML PPML PPML PPML

7.2 | Controlling for institutional quality

(1) (2) (3) (4) (5) (6) (7) (8)

PPML PPML PPML PPML PPML PPML PPML PPML

7.3 | Extending the coverage of RTAs

TABLE 11 Robustness check: extending the coverage of RTA

(1) (2) (3) (4) (5) (6)

PPML PPML PPML PPML PPML PPML

7.4 | Alternative measures of technology spillovers

(1) (2) (3) (4) (5) (6) (7) (8)

PPML PPML PPML PPML PPML PPML PPML PPML

TABLE A1 Descriptive statistics

Variable No. of Obs Mean Std. Dev. Min Max

Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) hf_bus_sim hf_bus_diff fi_leg_sim fi_leg_diff

Sh(WTO‐X1) Sh(WTO‐X2) Sh(WTO‐X3) CAij

WTO‐X′ 0.6399 0.5596 0.5565 1.0000

JINJI et al. 2363

You might also like