Download Technology Trade

Technology Trade
José L. Groizard
University of the Balearic Islands√
December, 2007
Abstract
This study addresses the question of why some countries import more R&D
intensive goods than others. Using a panel data set of countries for the period
1970 to 1995, results indicate that trade openness, domestic investment, FDI
and the quality of intellectual property rights (IPR) systems positively affect
technology imports. However, the higher the percentage of the workforce with
primary studies and the more developed the country, the lower technology
imports are.
JEL Classifications: F10, F13, O33, O34
Keywords: R&D, technology diffusion, absorptive capacity, complementarities, intellectual property
rights, technology imports.
√
Department of Applied Economics, University of the Balearic Islands, Ctra de Valldemossa km
7,5, 07122 Palma de Mallorca, Spain. Tel.: +34-971-172-784; fax: +34-971-172-389. E-mail:
[email protected]
I. INTRODUCTION
Differences in technology levels account for a major portion of cross-country income and
growth disparities (Klenow and Rodríguez-Clare, 1997; Hall and Jones, 1999; Caselli,
2005). While industrialized economies have the advantage in terms of innovation, the
majority of the world, which operates below the technological frontier, adopts technology
developed by other countries. This fact brings to backward economies an important
advantage because imitation and adaptation of new technologies is less expensive and
risky than creating them (Gerschenkron, 1962). Importing foreign technology is, then, a
mechanism with which to increase growth and reduce the income gap across countries.
Technology crosses borders via a variety of formal and informal channels, such as
contracts for technology transfer, trade, foreign direct investment (FDI), the migration of
skilled labour and imitation. A significant amount of recent evidence has supported the
idea that trade is an important channel in technology diffusion both among developed
countries (Coe and Helpman, 1995) and between developed and developing countries
(Coe, Helpman and Hoffmaister, 1997, Busse and Groizard, 2007)) and among different
potential channels, trade is perhaps the most consistent one (Keller, 2004). Hence,
understanding the technology adoption process through trade is an open debate.
The factors that influence technology diffusion through trade and the role of
development in this process are less straightforward. For instance, there are several
studies showing that stronger intellectual property rights (IPR) encourage trade (e.g.
Maskus and Penubarti, 1995 and Falvey et al., 2006), but there is no clear evidence that
IPRs encourage technology imports. Additionally, human capital, domestic investment or
2
openness are somehow perceived in the literature as factors enhancing technology
diffusion, however evidence on their respective contributions is scarce.
This study investigates the determinants of technology imports in a cross-section
of countries between 1970 and 1995, using panel data techniques. Our goal is to
investigate the precise role played by different factors that exert a potential influence on
technology adoption decision-making. Results suggest that domestic and foreign
investment positively influences high-tech imports. Moreover, importer countries may
increase technology inflows improving the quality of IPR systems. Other factors exert a
positive effect, such as exports of government expenditure. In contrast, countries with a
human capital base strongly skewed towards unskilled workers, import less technology
from abroad, while skill-biased economies tend to import more. Finally, GDP per worker
negatively affects technology imports, suggesting that the role of R&D intensive imports
in fostering productivity growth changes with the development process.
The study is organized as follows. The second section reviews the theoretical and
empirical literature on the determinants of technology adoption through trade. The third
section examines the technology gap and describes the data used to test the main
hypothesis. The fourth section presents the model, the estimation techniques and
discusses the results. Finally, section five summarizes the main conclusions of the study.
II. BACKGROUND
The neoclassical trade model emphasizes the role played by factor endowments and
technology levels to predict trade pattern. Leamer (1984) estimated a Heckscher-OhlinVanek (HOV) model and found support for the idea that countries import relatively larger
amounts of those goods that are intensive in the factor relatively scarce in the country.
3
However, Trefler (1995) and Harrigan (1997) have evaluated that HOV performs poorly
in a cross-section of countries. Nevertheless, when the assumption that technology is the
same for all the countries the model gains in explanatory power. In the latter studies there
is no role for factor complementarities in the determination of trade patterns, because
they ignore the case of trade in intermediate goods and product differentiation.
Neoclassical theory neglects the idea that countries tend to import more of those
goods in which the agents have a special advantage to use them more efficiently.
Surprisingly, there is a lack of theoretical literature concerning the specific determinants
of technology imports. The only two exceptions are Eaton and Kortum (2001) and Caselli
and Wilson (2004). The former proposes a comparative advantage model with
heterogeneous capital goods, in which imports of capital goods (as a share of GDP)
depend on investment rate and trade cost parameters that capture home-bias and other
geographic factors1. The latter proposes a model of investment based on heterogeneous
forms of capital, suggesting that shares in different types of capital goods in investment
depend on their intrinsic efficiency (embodied technology) and how much they are
complementary to other inputs, the abundance of which may vary across countries.
On the other hand, the literature on technology diffusion points out that
technology is specific to a particular combination of inputs (Atkinson and Stiglitz, 1969;
Basu and Weil, 1998) making its adoption more difficult, even with non-existent
adoption costs. Human capital is one of the most important factors likely to restrict
imports of R&D intensive goods, because the implementation of foreign technology
1
For instance, Eaton and Kortum (2001) find that a country’s level of human capital (measured by
average years of schooling) is associated with lower trade barriers, and that once human capital is
accounted for, tariff barriers remain an insignificant trade cost parameter.
4
might require skills and capabilities that are embodied in workers. Other studies
emphasize that technology developed in advanced countries is more likely to be skillbiased (Acemoglu and Zilibotti, 1999), and, consequently, technology adoption and
human capital specific investment needs to be undertaken simultaneously.
Empirical studies on the role of complementarities on trade are scant. Several
studies have detected that human capital plays a positive role in a country's absorptive
capacity. These findings suggest that the imitative ability acquired through trade is also
likely to be constrained by certain complementary factors. For instance, Falvey et al.
(2006) explain that intellectual property rights affect trade according to level of
development, imitative ability (human capital) and the size of the importing country. On
the other hand, the empirical evidence provided by Caselli and Wilson (2004) suggests
that imports of equipment are positively related to the R&D content of imports interacting
with a time trend, outward FDI, remoteness and property rights. But no clear evidence
has been found detecting any influence of human capital levels, intellectual property
rights or per capita income. However, a negative and robust inward FDI effect has been
found, suggesting that FDI flows skew the composition of capital stock towards low-tech
equipment. By looking at different types of capital goods, Caselli and Coleman (2001)
found that imports of computers increase according to human capital stock size and the
level of openness to trade, suggesting that technological spillovers are transmitted
through trade.
In summary, the ambiguity that arises in the literature concerning the role played
by openness, domestic and foreign investment, IPR and human capital in the process of
importing technology has driven this research.
5
III. DATA AND MEASUREMENT
A fundamental question approached in this study is how to measure technology.
Technology is defined as the quality of capital and intermediate goods used in the
production process at any given time. Technology adoption, hence, is investment
dedicated to updating and upgrading the quality of installed capital. Capital is a
heterogeneous composite of different kinds of goods, such as machinery and instruments.
To our knowledge there is no international data on disaggregate production or investment
for a large number of countries and periods and no index to compare capital quality
across countries2. Furthermore, one advantage of this definition of technology is that
while the production of capital goods is highly concentrated in a few countries, as Eaton
and Kortum (2001) have shown the imports of capital and intermediate goods emerge as
a reasonable proxy for technology adoption investment.
In the absence of detailed international production data on capital and
intermediate goods, the analysis has focused on the trade flows of machinery, instruments
and fine chemical products because these sectors are R&D intensive in the developed
world3. The theory suggests that a country with comparative advantage in the production
of knowledge intensive goods should show a significant presence of such products in the
trade pattern. The revealed comparative advantage index (RCA) in product i and country
j is described by the expression
2
Existing data matching production and trade data cover a shorter period of time (1980-1997) and
is restricted to a lesser number of countries. See Caselli and Wilson (2004) for an explanation.
3
This taxonomy is based on Pavitt (1984). In Table A1 in the Appendix, the reader can find a brief
description of the capital and intermediate goods selected according to R&D intensity.
6
(1)
RCAij =
∑
X ij / ∑ X ij
X ij / ∑
∑
X ij
where X ij is the value of R&D intensive exports (in US dollars) in country j at a given
time. A value of 1.1 shows that country j’s technology exports relative to its total exports
are 10 percent higher than the share of world technology exports relative to total world
exports. Hence, an RCA index above one for a country can be interpreted as a
technological advantage, and conversely, an RCA below one can be interpreted as a
technological disadvantage.
Figure 1 shows worldwide technological change since 1965, as revealed in the
export pattern among country groups. In 1965, the technology gap between the top 8
R&D providers and the developing economies was tremendous, but has diminished
significantly in the period studied. Furthermore, the data reveals a technology gap
between technological leaders and OECD economies4. During these years, the ability to
export technology to world markets has been concentrated in OECD countries as a sign
of technological superiority and in particular, among a small group of top 8 R&D
performer countries.
INSERT FIGURE 1
These types of goods are by definition R&D intensive in R&D performing
countries. Figure 2 shows a strongly positive relationship between RCA in technological
products and R&D expenditure. The larger a country's investment in R&D, the greater its
4
The data for OECD is an average that includes the top 8 R&D performer countries, when we
exclude them the gap also is magnified. We plotted OECD average, since the aim of this exercise was to
compare technology gaps between developing economies and technological leaders and not between
technological leaders.
7
technological advantage is in the export pattern5. The data clearly shows that
technological advantage and R&D effort are positively related in a cross-section of
countries, which means that this type of product embodies technology that is based on
R&D activities at country level. Since data on R&D stock for several years and multiple
countries is not available we have only considered countries with an RCA index above
one in the initial year as technology leaders. The group of countries selected are
Germany, Finland, France, Japan, Sweden, Switzerland, United Kingdom and USA6.
IV. MODEL AND ESTIMATION
The goal of this study is to estimate the determinants of R&D intensive goods imports
exported by the top 8 technology leaders. We used the following model
(2)
log M
jt
=α+X
jt
β+u j +vt +ε jt
jt
where M is per worker technology imports, measured in current US dollars in country j
jt
and time t. X is a set of macroeconomic and institutional explanatory variables that
j
capture the social capability to absorb advanced technology, u is a fixed country effect,
v
t
jt
is a set of time dummies, and ε is an error term identically and independently
distributed across countries and time.
Technology is adopted because agents find that expected profits overcome
expected costs. Although this decision is taken on the demand side, expected costs and
5
Thailand and Mexico are clear outliers, since neither of these economies would be expected to
show a technological advantage in the RCA index, due to their low level of R&D expenditure. However,
they still have an RCA index greater than one due to the fact that they are importers and assemblers of
technological components.
6
Eaton and Kortum (2001) found similar trade patterns comparing more recent production data of
capital goods.
8
benefits are affected by the supply side and by the institutional environment. At national
level, the demand for technology imports is constrained by an adequate supply of
complementary inputs, such as skilled workers. To control for this issue we have
introduced several educational variables relating to workforce. On the one hand, we
constructed an aggregated measure of human capital based on the adult population's years
of education, taken from Hall and Jones (1999).7 On the other hand, we disaggregated
human capital at different levels of education according to attainment measures
calculated by Barro and Lee (2000). There are also relatively important potential
technology adopters, such as domestic companies, the government and multinational
companies. To control for these we included three proxies, domestic investment,
government consumption and FDI inflows8.
There are several factors that affect technology adoption from the supply side.
Rosenberg (1972) emphasized three factors, such as the novelties introduced into
technology after creation, the invention of new uses for technology and the provision of
complementary inputs by technology providers. At country level we controlled for these
factors by including investment per worker as an independent variable.
Finally, another set of control variables were introduced to take the institutional
and policy environment into account. The increasing literature on intellectual property
rights and trade concludes that a stronger IPR system has two opposite effects (Maskus
and Penubarti, 1995). On the one hand, there is a positive market expansion effect, since
companies should be encouraged to export into that market because protection reduces
7
Hall and Jones (1999) measured human capital by imposing a discontinuous linear function of
schooling year per worker.
8
Data extracted from WDI (2005).
9
the risk of piracy and protects company profitability in that market. However, there is a
market power effect of negative sign, since stronger IPR protection reduces the ability of
domestic companies to imitate and this increases the market power of the exporter firm,
which might lead to reduced sales ex-post in that market. The final result will depend on
the relative importance of market power and market expansion. To measure IPR we
introduced an index of patent protection ranked from less (0) to more (5) patent
protective legal systems as calculated from Ginarte and Park (1997).
According to the conventional view, trade yields static gains because exporting
the goods a country produces more efficiently, allows the importation of more of those
goods the country produces more inefficiently. In other words, exports are what an
economy pays for its imports. More recent theories bring attention to trade openness as a
channel of knowledge regarding spillovers9. According to this view, more open
economies tend to benefit more in terms of access to public world knowledge, which
might lead to dynamic gains from trade. To control for these issues we introduced exports
per worker as a measure of trade openness10.
We added GDP per worker as a control variable, since there are possible omitted
factors determining technology imports not clearly captured by the variables already
described. The data was obtained from an updated version of the Summers and Heston
(1991) data set.
9
Grossman and Helpman (1991) provide an excellent overview of models.
10
Data on exports comes from Easterly and Sewadeh (2005).
10
Several X variables, such as education and the intellectual property rights
protection index are not available on a yearly basis. Therefore, a panel with a crosssection of a maximum of 79 countries, from 1970 to 1995, at five year intervals was used.
Table 1 displays summary statistics of the variables used in the empirical analysis.
Each variable is represented as an average for the countries and periods employed. In
1970, technology imports represented 12 US dollars per worker while total exports
represented 280 US dollars per worker, that is 23 times larger than the former. However,
in 1995, the average country imported technology to the sum of 136 US dollars and
exported 1876 US dollars both in per worker terms. This means that during this period
technology imports grew at a faster rate than exports for the country sample.
[INSERT TABLE 1 ABOUT HERE]
A major concern would be the potential bias arising from reverse causality in
several explanatory variables. For instance, technology imports increase GDP per worker
rather than the other way around. Since GDP per worker is not a variable of interest and
only a control variable, we minimized the endogeneity issue by lagging it five years.
Other variables with similar problems are domestic investment and exports. It is likely
that technology imports and domestic investment and imports and exports are determined
simultaneously because they may have similar underlying factors. To deal with this issue
we introduced them into the regression analysis as predetermined, lagged five years.
4.1 Results
Table 2 summarizes the basic results derived from the total sample estimation. Each
column represents a different specification of the model. The first one includes the main
set of predictors, the second one shows the role of education level variables, the third is
11
an extension to decompose the influences of trade openness in technology imports, and
the forth and fifth are robustness checks.
Each specification was estimated by fixed effects (FE) and random effects (RE)
including year time dummies. The dummy coefficients have been omitted due to space
restrictions. A test of FE versus RE was performed to evaluate the efficiency and
consistency of the estimators. The Chi-squared statistic and its p-value are displayed at
the bottom of the table, and as we can see the RE estimators have been discarded in all
cases11.
[INSERT TABLE 2 ABOUT HERE]
In column 1, six out of seven predictors have a significant effect on technology
imports at conventional levels: income, investment, government consumption, IPR
protection index, FDI and exports. All the variables have the expected sign with the
exception of the non-significant Hall-Jones human capital index, which is discussed
below.
In column 2, human capital was broken down into three different levels:
percentage of the labour force with completed primary education and no more,
percentage of the labour force with completed secondary education and no more, and
percentage of the labour force with completed tertiary education. The effect of the
educational levels on technology imports is negative at primary level and null at
secondary and higher levels of education. The remaining variables maintain the same sign
11
The Sargan-Hansen test is a test of overidentifying restrictions that is designed to generate non-
negative test statistics. See Arellano (1993) for a precise explanation.
12
and significance with the exception of the FDI and IPR protection index which are not
significant.
Column 3 shows the estimates of the model once openness to trade is divided into
manufacturing and non-manufacturing exports and the composition of the labour force
education is accounted for. The manufacturing export coefficient is negative and
insignificant and the non-manufacturing export coefficient is positive and significant at
the 1 percent level. When splitting exports, all the other predictors are significant at
conventional levels with the exception of income per worker and educational attainments
at the secondary and tertiary levels. Moreover, IPR and FDI turned out to be positive and
significant.
The specification in column 4 checks the robustness of previous results from
different measures of labour force education. Existing data allows discrimination between
the number of individuals that start a certain level of education but who do not finish it.
We employed measures of portion of individuals attaining some level of education (i.e.
primary, secondary and tertiary). Comparing columns 3 and 4 we can see two changes in
the significance level of coefficients. On the one hand, the percentage of individuals with
some secondary education affects technology imports negatively. On the other hand,
income becomes significant while IPR loses its significance level.
Finally, the specification in column 5 replaces income per worker with shares of
agriculture and industry in GDP. The only noticeable change is that exports of
manufactured goods are now significant, while the significance levels and magnitude of
the other estimated coefficients are maintained.
13
4.2 Discussion
Why do some countries import more R&D intensive goods per worker? According to the
literature, a broad set of variables has been tested as potential explanations. Rising
domestic investment, government consumption, exports of non-manufacturing goods,
FDI inflows and improving the institutional Intellectual Property Rights (IPR) system
stimulate technology imports. Meanwhile, increases in income per worker and in
attainment of basic education seem to constrain access to embodied foreign technology.
The estimated elasticity of their respective contributions vary slightly across
specifications, but, in general, parameters seem to be stable.
Economies investing more resources domestically also invest more in foreign
technology. The estimated elasticity suggests that increasing domestic investment by 1
per cent results in an increase in imports of R&D intensive goods of 0.5 per cent, five
years after. This variable, basically aimed at capturing the effort of the local economy to
provide supply side stimulus for the adoption of new technology, such as the adequate
provision of complementary physical capital inputs or expenditure to adapt foreign
technology to local conditions. Moreover, level of public expenditure also raises the
imports of technology, meaning that the government is an important technology adopter,
either acting directly or generating favourable conditions relating to the level of public
consumption.
When computing the effect of a general measure of human capital, the elasticity
becomes negative but insignificant. When splitting the human capital effect into three
educational categories, economies with a low percentage of primary educated workers
tend to import more technology, while countries with a higher percentage of tertiary
14
educated workers tend to import more R&D intensive goods12. This suggests that
unskilled workers and technological goods are substitutes, an idea which is consistent
with the international skill-biased technological progress evidenced by Berman and
Machin (2000) and Gancia and Epifani (2007). A striking implication of these findings is
that low income countries needing foreign technology to catch-up with the leading
economies, but which do not have enough human capital are stuck in a form of poverty
trap. In other words, policies aiming at extending primary education at the expense of
tertiary education will perpetuate this state.
A robust link exists between exports and technology imports. In the first two
specifications, total exports have a positive elasticity estimated with great precision. A 1
percent increase in exports per worker generates an increase in technology imports
ranging from 0.22 to 0.26 percent five years after. Disaggregating between manufacturing
and non-manufacturing exports results shows that non-manufacturing exports stimulate
technology imports, and that manufacturing exports have no clear impact on imports of
R&D intensive goods.
These findings can be interpreted as evidence of both, gains from trade, and
international knowledge spillovers linked to trade. On the one hand, trade theory
establishes that all countries experience welfare gains when they specialize according to
comparative advantage. One of these welfare enhancing channels is that exporting
countries are able to import those goods that are difficult to substitute by domestic
products, such as R&D intensive goods. The fact that countries that export more
manufactured goods per worker tend to import less technology and countries that export
12
However, the effect was measured with a low level of precision.
15
more non-manufactured goods import more technology, is conceivable within the
comparative advantage framework. On the other hand, there is a great deal of new
theoretical13 and empirical14 literature stressing the role of international trade as a channel
for knowledge and technology diffusion around the world. Trade allows the flow of ideas
beyond national borders and learning from the experience of others in the production and
use of new technology. So, when economies are open to trade they are not simply trading
products; they are establishing information channels that are extremely useful for the
transfer of free knowledge. The kind of knowledge that flows from other countries is
likely to result from experience rather than from R&D investment, so it is a complement
to the knowledge embodied in the capital and intermediate goods traded. The interaction
between embodied and disembodied knowledge has to be carefully considered because it
may be the origin of innovative processes in early stages of development. This is possible
because domestic and foreign knowledge is more complementary when differences in
development are wider, that is when the presence of non-manufactured products in the
export pattern is larger. This outcome is similar to that found by Caselli and Coleman
(2001) in the case of computers.
Multinational companies are also complements in the adoption of technology
because they demand R&D intensive goods. A 1 percent increase in FDI raises imports of
R&D intensive goods by 0.04 per cent. The magnitude of this elasticity is far below the
contribution of domestic investment to technology imports which is around 0.5.
13
Grossman and Helpman (1991) provide a good discussion on a collection of theoretical models.
14
Coe and Helpman (1995) for OECD countries, and Coe et al. (1997) for Non-OECD countries,
among others.
16
Another important determinant of technology imports is the protection of
intellectual property rights. Our findings suggest that increasing the level of patent right
protection may increase technology imports. This implies that incentives to export
technology are stronger when institutions guarantee appropiability from technology
providers. This empirical finding is in sharp contrast with other study results which never
found that the market power effect existed in R&D type goods (Caselli and Wilson,
2004).
Finally, by increasing GDP per worker, economies tend to reduce imports of
technology. This is a sensible result since a higher level of development means an
increased ability to produce more sophisticated goods and as a consequence to facilitate
the use of domestic technology. Several studies have documented the role of trade, and
R&D intensive trade as a channel for increasing the level of GDP per worker. Results
reported in this study suggest that the role of trade in promoting technology diffusion is
greater the lower the level of development is in the country.
V. CONCLUSION
In this study some macro-determinants of technology adoption in a cross-section of
countries between 1970 and 1995 have been explained. Technology imports have been
measured in terms of R&D intensive goods imports exported by top R&D performing
countries. The findings suggest that countries can increase imports of R&D intensive
goods through domestic and foreign investment, increasing human capital levels, opening
up the economy to trade and improving the IPR system. However, economies tend to
substitute imports of foreign technology by increasing the number of non-skilled
individuals into the workforce and exporting manufactured goods. Moreover, increasing
17
the level of development tends to reduce technology imports. The cross-country
variability in the factors that enhance or restrict technology adoption through imports also
implies that those factors will affect investment in R&D intensive sectors.
References
Acemoglu, D. and F. Zilibotti (2001). “Productivity Differences”, Quarterly Journal of
Economics, 116(2): 563-606.
Arellano, Manuel, (1993).“On the testing of correlated effects with panel data,” Journal
of Econometrics, Vol. 59, 1-2: 87-97.
Atkinson, A.B. and J.E. Stiglitz (1969). “A New View of Technological Change”,
Economic Journal, 79: 573-78.
Barro, R.J. and J.W. Lee (2000). “International data on educational attainment: updates
and implications”, CID Working Paper 42, Harvard University, Cambridge.
Bassu, S. and D.N. Weil, (1998). “Appropriate Technology and Growth”, Quarterly
Journal of Economics, 113(4): 1025-54.
Berman, E. and S. Machin (2000). “Skill-biased technology transfer around the world”,
Oxford Review of Economic Policy, 16(3): 1245-1279.
Busse, M. and J.L. Groizard (2007) “Technology Trade in Economic Development”, The
World Economy, forthcoming.
Caselli, Francesco (2005). ‘Accounting for Cross-Country Income Differences’, in P.
Aghion and S. Durlauf (eds.), Handbook of Economic Growth (Amsterdam:
North-Holland).
Caselli, F. and W.J. Coleman II (2001). “Cross-country technology diffusion: the case of
computers”, American Economic Review, 91: 328-335.
Caselli, F. and D. Wilson (2004). “Importing Technology, Journal of Monetary
Economics, 51, 1: 1-32.
Coe, D.T. and E. Helpman (1995). “International R&D Spillovers”, European Economic
Review, 39(5): 859-887.
18
Coe, D.T., A.W. Hoffmaister and E. Helpman (1997). “North-south R&D Spillovers”,
Economic Journal, 107: 134-149.
Diwan, I. and D. Rodrik (1991). “Patents, appropriate technology, and North-South
trade”, Journal of International Economics, 30: 27-47.
Easterly, W. and M. Sewadeh (2005), Global development network growth database,
World Bank, (http://www.worldbank.org/research/growth/GDNdata.htm)
Eaton, J. and S. Kortum (2001), “Trade in Capital Goods”, European Economic Review,
45(7): 1195-1235.
Falvey, R., N. Foster and D. Greenaway (2006). “Trade, Imitative Ability and Intellectual
Property Rights”, University of Nottingham Research Paper No. 2006/25.
Fink, C. and C.A. Primo-Braga (2005). “How stronger protection of intellectual property
rights affects international trade flows?”. In Fink, C and K.E. Maskus (eds.),
Intellectual Property and Development: Lessons from Recent Economic Research,
The World Bank, Washington, DC.
Gancia, G. and P. Epifani (2007). “Skill bias of world trade”, The Economic Journal,
forthcoming.
Gerschenkron, Alexander (1962). Economic Backwardness in Historical Perspective, The
Bellknap Press, Cambridge.
Ginarte, J.C. and W. Park (1997). “Determinants of patent rights: a cross-national study”,
Research Policy, 26(3): 283-301.
Grossman, G. and E. Helpman (1991). Innovation and Growth in the Global Economy,
Cambridge, MIT Press.
Hall, R.E. and C.I. Jones (1999). “Why do Some Countries Produce so Much More
Output per Worker than Others?”, Quarterly Journal of Economics, 114, 2: 83116.
Harrigan, James (1997). “Technology, Factor Supplies and International Specialization:
Estimating the Neoclassical Model”, American Economic Review, 87(4): 475-94.
Keller, Wolfgang (2000). “Do Trade Patterns and Technology Flows Affect Productivity
Growth?”, World Bank Economic Review, 14 (1): 17-47.
19
Keller, Wolfgang (2002). “Trade and the Transmission of Technology”, Journal of
Economic Growth, 7: 5-24.
Keller, Wolfgang (2004) “International Technology Diffusion”, Journal of Economic
Literature, 42 752-782.
Klenow, P.J. and A. Rodríguez-Clare (1997). ‘The Neoclassical Revival in Growth
Economics: Has It Gone Too Far?’. In B. Bernanke and J. Rotemberg (eds.),
NBER Macroeconomics Annual (Cambridge, MA: MIT Press).
Maskus, K.E. and M. Penubarti (1995). “How trade-related are intellectual property
rights?”, Journal of International Economics, 39: 227-248.
Pavitt, Keith (1984). “Sectoral patterns of technical change: towards a taxonomy and a
theory”, Research Policy, 13: 343-373.
Rosenberg, Nathan (1972). “Factors Affecting the Diffusion of Technology”,
Explorations in Economic History, 10(1): 3-33.
Summers R. and A. Heston (1991). “The Penn World Table (Mark 5): an expanded set of
international comparison”, Quarterly Journal of Economics, 106, 1991, 327-368.
Treffler, Daniel (1995). “The Case of Missing Trade and other Misteries”, American
Economic Review, 85: 1029-46.
UNCTAD, UN COMTRADE. Online Access to the Trade Database of the UNCTAD,
Internet Posting: http://comtrade.un.org/, 2005.
UNESCO (1999). UNESCO Statistical Yearbook 1999, Paris.
World Bank (2005). World Development Indicators 2005, CD-ROM, Washington.
20
Table A1: Technological goods
SITC Code rev.1
Description
541, 553
Medicinal, pharmaceutical products and fine chemical
products
7111, 7112, 7113, 7114, 7115, 7116, 7117, 7118
Various machinery and equipment (excluding internal
combustion engines)
722, 7231, 7249, 726, 729, 734
Specialized machinery for several industries (excluding
machinery for manufacturing paper and food processing)
861, 862, 864
Specialised machinery and instruments for particular
industries
9510
Other goods
Source: Standard International Trade Classification (SITC), revision 1 and ECLAC (1996).
21
Figure 1: Revealed Comparative Advantage in R&D inventive goods
1.6 0
1.4 0
1.2 0
1.0 0
0 .8 0
0 .6 0
0 .4 0
0 .2 0
0 .0 0
1965
1995
OCDE
No-OCDE
Top 8
Figure 2: Technology Advantage and R&D Effort, 1995
JPN
2
Revealed Comparative Advantage
SGP
IRL
GBR DEU
MEX
USA
KOR
SWE
FRA
FIN
1
THA
ITA
AUT
CHN
CZE
HUN
0
BRA
PRT ESP SVK
POL
EST
LTU HRV
ROM
KGZ
TUN LVA
ARG
NZL
MDA
KAZ
TUR
BGD
SEN
CHL
ECU
EGY
KWT
MDG
UGA
0
1
DNK
NLD
ISR
SVN
CAN
BEL
NOR
ISL
2
3
R&D expenditure (% of GDP)
22
4
Table 1: Summary Statistics
Variable
Obs
Mean Std. Dev.
Log of Technology Imports per worker
329
4.198
1.654
Log de GDP per worker
329
9.072
0.962
Log Investment per worker
329
7.336
1.496
Government Consumption (% GDP)
329
15.124
5.956
Intellectual Property Rights Protection Index
329
2.571
0.921
Log FDI per worker
329
3.452
2.216
Log of human capital
329
0.626
0.300
Primary Education Completed
327
14.493
8.223
Secondary Education Completed
327
9.818
9.214
Higher Education
329
8.005
8.494
Log Exports per worker
329
6.747
1.756
Log Manuf. Exports per worker
328
4.829
2.603
Log Non-Manuf. Exports per worker
328
6.364
1.611
Note: all variables refer to country averages for the period considered
23
Min
-0.499
6.737
3.939
3.135
0.000
-3.878
0.000
1.200
0.100
0.000
2.194
-1.301
1.531
Max
8.106
10.530
10.040
43.406
4.860
8.318
1.224
45.100
49.300
50.300
10.779
10.105
10.067
Table 2: Results
(1)
-0.388
[0.127]***
0.538
[0.060]***
0.019
[0.008]**
0.168
[0.075]**
0.032
[0.018]*
-0.614
[0.380]
0.260
[0.061]***
Log GDP per worker
Log Investment per worker
Government Consumption (% GDP)
Intellectual Property Rights Protection Index
Log FDI per worker
Log of Human Capital
Log Exports per worker
(2)
-0.322
[0.127]**
0.570
[0.059]***
0.020
[0.008]**
0.119
[0.075]
0.027
[0.018]
Log Non-Manufacturing Exports per worker
-0.011
[0.004]***
-0.001
[0.005]
0.016
[0.018]
Secondary Education Completed
Higher Education Completed
(4)
-0.208
[0.115]*
0.509
[0.059]***
0.019
[0.008]**
0.122
[0.075]
0.042
[0.018]**
-0.037
[0.026]
0.245
[0.062]***
-0.012
[0.004]***
-0.002
[0.005]
0.014
[0.018]
-0.040
[0.026]
0.261
[0.060]***
(5)
0.481
[0.066]***
0.027
[0.010]***
0.146
[0.080]*
0.035
[0.019]*
0.215
[0.063]***
Log Manufacturing Exports per worker
Primary Education Completed
(3)
-0.182
[0.116]
0.560
[0.059]***
0.020
[0.008]**
0.133
[0.075]*
0.037
[0.018]**
Some Primary Education
-0.054
[0.025]**
0.173
[0.055]***
-0.012
[0.004]***
0.000
[0.006]
-0.002
[0.024]
-0.012
[0.004]***
-0.014
[0.005]***
-0.005
[0.011]
Some Secondary Education
Some Higher Education
Agriculture share (% GDP)
Industrial share (% GDP)
Constant
0.550
[0.889]
328
79
0.83
Observations
Number of countries
R-squared
0.108
[0.879]
326
79
0.84
-1.008
[0.844]
325
78
0.84
-0.009
[0.913]
325
78
0.84
-0.010
[0.006]
0.002
[0.007]
-1.541
[0.561]***
286
73
0.82
63.031
71.746
72.698
76.439
62.963
Sargan-Hansen Statitic (χ2)
p-value
0.000
0.000
0.000
0.000
0.000
Standard errors in brackets. Significance levels: * significant at 10%; ** significant at 5%; *** significant at 1%. The
dependent variable is the log of technology imports per worker. The following variables are lagged five years: GDP per
worker, Investment per worker, Total exports per worker, Manufacturing Exports per worker and Non-Manufacturing
Exports per worker. All the specifications are estimated using the fixed effect estimator and include a set of time dummy
variables.
24