Globalization and Growth

By G ENE M. G ROSSMAN AND E LHANAN H ELPMAN

How does globalization affect economic fords those who invent or improve products a
growth? The modern literature on endogenous greater potential market on which to reap returns
growth provides tools and models that are use- even as it subjects them to additional competi-
ful for elucidating some of the mechanisms link- tion from foreign rivals. The incentives for in-
ing international integration with long-run eco- novation may intensify or diminish with integra-
nomic performance. tion, depending on whether the scale effect or
Up until the mid 1980s, studies of growth fo- the competition effect is more powerful. Third,
cused primarily on the accumulation of physi- the integration of world markets has general-
cal capital. But, capital accumulation at a rate equilibrium implications for input prices and rel-
faster than the rate of population growth is likely ative output prices. These price changes affect
to meet diminishing returns that can drive the the cost of innovation as well as the relative at-
marginal product of capital below a threshold tractiveness of alternative directions for indus-
at which the incentives for ongoing investment trial research. Finally, international interactions
vanish. This observation led Romer (1990), Lu- affect not only the incentives for creation of new
cas (1988), Aghion and Howitt (1992), Gross- knowledge, but also those for technological dif-
man and Helpman (1991a) and others to focus fusion, with analogous implications for produc-
instead on the accumulation of knowledge, be tivity growth. Many authors have examined how
it embodied in textbooks and firms as “technol- one or more of these mechanisms operates in
ogy” or in people as “human capital.” Knowl- some specific economic environment. Taken to-
edge is different from physical capital inasmuch gether, the literature offers many theoretical in-
as it is often non-rivalrous; its use by one person sights. Some progress has also been made on
or firm in some application does not preclude its the empirical side, although data and method-
simultaneous or subsequent use by others. The ological impediments have left assessment and
non-rivalrous nature of knowledge suggests in- measurement lagging behind.
creasing returns when output is related to all tan-
gible and intangible inputs, which eliminates the I. International Knowledge Spillovers
inevitability of diminishing returns to the accu-
mulation of some inputs relative to others. The most direct link between globalization
The new models of knowledge accumulation and growth arises when knowledge acquired in
highlight several potential links between inter- one country can be used to facilitate research in
national integration and growth. Research has another. Scientists exchange ideas when they
focused on how the international exchange of meet at international conferences. Knowledge
goods and ideas affects the incentives for knowl- flows in the course of business transactions and
edge acquisition and on the efficacy of inven- in other human interactions. And learning from
tiveness and diffusion. Several mechanisms fea- abroad can occur without personal contact via
ture prominently in the literature. First, integra- publications and reverse engineering. Helpman
tion of peoples and cultures facilitates the flow (2004) reviews a body of empirical research
of knowledge across national borders. Foreign that finds evidence of substantial international
ideas may be useful for inventing new products, knowledge spillovers. At the same time, Coe
for improving existing products, or for produc- and Helpman (1995), Eaton and Kortum (1999)
ing goods at lower cost. Second, integration and others have found that international knowl-
of product markets via international trade af- edge spillovers are far from complete, leaving
room for further integration of the world econ-
Grossman: Princeton University, [email protected].
omy to raise knowledge stocks around the globe.
Helpman: Harvard University and CIFAR, ehelp- Romer (1990) developed a model in which
[email protected]. knowledge accumulated in the course of con-
1
2 PAPERS AND PROCEEDINGS MONTH YEAR

ducting R&D raises the productivity of fu- on sales abroad. This “scale effect” tends to
ture innovation efforts. Grossman and Help- boost the incentive for knowledge acquisition.
man allowed for international knowledge flows, However, in a more global economy, a success-
whereby either the knowledge stock that deter- ful innovator must share the market not only
mines productivity in inventing new products with other domestic firms, but also with those
reflects experience both at home and abroad, that produce abroad. The “competition effect”
or else quality upgrading builds on past re- of globalization presents an offsetting disincen-
search successes in all countries. International tive for knowledge acquisition.
knowledge spillovers tend to accelerate growth Grossman and Helpman (2014) consider a
in all countries, as the cost of further innovation world economy in which inventors develop new
declines in every country with advances made varieties of a differentiated product that are CES
elsewhere. Grossman and Helpman (2014) substitutes. Individuals differ in ability and suc-
posit an arbitrary pattern of partial knowledge cessful innovators draw different technologies
spillovers, whereby research experience in each for producing their varieties.1 The model incor-
country contributes somewhat to R&D produc- porates complementarities between the produc-
tivity elsewhere, but not as fully as it does to tivity of a technology and the ability of the work-
R&D productivity in the country where the re- ers that the firm employs. There are neither fixed
search was carried out. They find in the con- costs of production nor of exporting. In this set-
text of a one-sector model that an increase in ting, the countervailing forces of scale and com-
the extent of spillovers from an arbitrary country petition are quite clear: a reduction in trade costs
to any other raises long-run growth rates every- in some country has no effect on the common
where in a many-country world economy. rate of long-run growth in any of them. The ex-
In much of the literature, the scope for inter- tra profit opportunities that result from greater
national knowledge spillovers is taken as exoge- aggregate demand are exactly offset by the loss
nous. But trade and foreign direct investment of market share to foreign producers.
(FDI) may be conduits for knowledge transmis- The absence of fixed costs is important for this
sion. Firms in an importing country gain ideas conclusion. Baldwin and Robert-Nicoud (2008)
about new products and new techniques from consider an endogenous-growth model with het-
their suppliers. Similarly, firms in an export- erogeneous firms and fixed costs of operation
ing country acquire information by discussing and of exporting, as in Melitz (2003). Then, a
product specifications or receiving ex post feed- decline in trade costs raises the cut-off produc-
back from their customers abroad. And multi- tivity level needed for a firm to survive and re-
national corporations transfer knowledge about duces the cut-off productivity level that leads it
products, processes, and management methods to participate in exporting. The resulting selec-
to their foreign affiliates. This information may tion of more productive firms increases the in-
become available to indigenous firms that ob- tensity of competition in the world market. In
serve their operations or hire their former em- the Baldwin and Robert-Nicoud environment, if
ployees. Indeed, Coe and Helpman (1995) and the extent of international knowledge spillovers
Keller (2010) provide evidence that a country’s remains constant after trade costs fall, the expan-
bilateral trade volume with a particularly partner sion of aggregate demand that results from a fall
helps to explain the extent to which R&D pro- in trade costs is more than offset by growth in
ductivity in the country benefits from the part- the effective competition in the market, leading
ner’s prior research experience. Baldwin et al. to a decline in incentives for ongoing R&D.
(2005) and Keller (2010) find similarly for FDI. Competition effects also can dominate when
costs of innovation fall with accumulated lo-
II. Scale versus Competition cal research experience but there are no interna-
tional spillovers of research knowledge. Feen-
Globalization affords innovators the opportu- stra (1996) considers a world economy in which
nity to exploit their new ideas on a larger stage.
Firms that develop a new product, improve an 1 The model builds on Grossman and Helpman (1991a), but
old one, or find a better production technique allows for heterogeneous firms, heterogeneous workers, and par-
can reap profits not only domestically, but also tial knowledge spillovers.
VOL. VOL NO. ISSUE GLOBALIZATION AND GROWTH 3

two countries develop and produce varieties of the other sector, firms produce a homogeneous
a differentiated product. The cost of innovation good under conditions of perfect competition.
is inversely related to a country’s own cumula- The authors posit that countries have exogenous
tive research experience. For whatever reason, endowments of labor and human capital with
one country has greater incentives for innova- different relative quantities, or else that human
tion than the other and grows faster in the au- capital results from private decisions about ed-
tarky equilibrium. When the countries open to ucation. They assume that R&D is the most
trade, the rapid growth in the number of com- human capital intensive activity in the econ-
petitors in the fast-growing country reduces the omy, followed by production of differentiated
profitability of innovation in the slow-growing products, and lastly by production of the homo-
country, and the gap in their innovation rates geneous goods. In this analysis, international
widens. The consequences for the lagging coun- knowledge spillovers are complete, so inventors
try can be even more severe in a setting with in every country are equally effective at conduct-
multiple industries that differ in their potential ing R&D.
for innovation and productivity growth. Then, Their findings are reminiscent of the
the intensified competition that results from an Heckscher-Ohlin model. The country that
opening of trade can lead the country with has the greater relative endowment of human
lesser incentives for R&D to specialize in in- capital—or the one that has the best educational
dustries that themselves have lesser innovation system—specializes relatively in the creation of
prospects, thereby exacerbating the initial dif- knowledge. In this country, the cost of innova-
ferences between them. Grossman and Help- tion is lowest, because the abundance of human
man (1991a, ch. 8) and Young (1991) make capital makes this factor relatively cheap. With
the further point that, with national knowledge invention comes comparative advantage in
spillovers, history and initial conditions can mat- producing differentiated products, so in the long
ter for the effects of globalization on a country’s run the country with an abundance of human
subsequent growth. They consider the opening capital conducts more R&D, exports differenti-
of trade between two otherwise symmetric coun- ates products in exchange for the homogeneous
tries in which one has an initial advantage in a good, and grows faster. The implications for
sector with potential for knowledge accumula- innovation and growth in the labor abundant
tion. Thanks to its head start, the leading country country are ambiguous: on the one hand,
has a lower cost of innovation, which allows it to knowledge spillovers from abroad augment its
undertake more of this activity while the other productivity in research; on the other hand, its
country does less. In the extreme, a country that comparative advantage in the labor-intensive
would have continued to innovate and grow in activity causes it to allocate more resources to
autarky can be led by competition with a more the production of homogeneous goods. On net,
advanced partner to specialize in activities that either effect can dominate, and so globalization
lack substantial growth prospects. can cause the growth rate to rise or fall.2

III. Innovation in General Equilibrium IV. Technological Diffusion

In a static economy, globalization leads coun- Much research of late has been directed at the
tries to specialize in the activities in which they process of technological diffusion. Ongoing dif-
enjoy a comparative advantage. The same is true fusion, like the creation of new technologies, can
in a setting with endogenous growth wherein
2 Peretto and Valente (2011) tell a similar story about re-
one of the activities that each country undertakes
source abundance. They consider a resource boom in a resource-
is the accumulation of knowledge.
rich country that processes raw materials to generate intermedi-
Grossman and Helpman (1991a) studied mod- ate inputs and also invests and produces differentiated products.
els with two manufacturing sectors. In one sec- If the raw material and labor are complements in the processing
tor, profit-seeking entrepreneurs invest human activity, then the resource boom draws labor into the production
of intermediate inputs and away from innovation and manufac-
capital and labor either to invent new varieties of turing of differentiated products. On the other hand, if labor and
a differentiated product or to push existing prod- raw materials are substitutes in processing, labor devoted to in-
ucts to the next wrung of the quality ladder. In novation and manufacturing grows.
4 PAPERS AND PROCEEDINGS MONTH YEAR

be a source of sustained growth in some envi- further entry. Globalization speeds growth de-
ronments. International integration affects the spite an absence of any scale effect and an ab-
incentives for investment in activities that foster sence of international knowledge spillovers.
diffusion as well as in the productivity of those Alvarez et al. (2014) explore yet another
activities. This provides another link between mechanism that links globalization to diffusion
globalization and growth. in their model of idea flows. They start from
In Perla et al. (2014) heterogeneous firms the supposition that firms learn from those with
continuously face a choice whether to produce whom they conduct business. Each country
a variety of a differentiated product or to search has a current best-practice for producing each
for a better technology. If a firm produces, it can good, à la Eaton and Kortum (2002). Product
pay a fixed cost in order to export. If it elects managers meet others at some exogenous rate.
to try to upgrade its technology, it pays a cost When a meeting occurs, the manager observes
in exchange for a random draw from the dis- the technology of her contact and adopts that
tribution of domestic technologies in use at the technology if it is better than her own. The dis-
time. In this setting, a set of the least produc- tribution of contacts depends upon the distribu-
tive firms opt to search rather than produce. A tion of productivities among active producers. In
fall in trade costs raises the relative profitabil- autarky, the source distribution for the learning
ity of high-productivity firms that exercise the reflects the distribution of productivities in the
opportunity to export relative to low productiv- domestic economy. Trade improves the source
ity firms, that at best sell to the domestic market distribution by replacing some less efficient do-
and face more intense competition there. As in mestic sellers with more efficient foreigners. In
Grossman and Helpman (2014), a fall in trade other words, trade is the vehicle for endogenous
costs is neutral with respect to the incentives for international knowledge spillovers, as in Gross-
knowledge acquisition if the fixed costs of ex- man and Helpman (1991b).
porting are nil. Otherwise, diffusion can accel-
erate or decelerate in response to globalization, V. Concluding Remark
depending on the nature of the cost function for
searching for new technologies. If, for example, The theoretical literature identifies a number
the cost is paid by hiring labor in a setting with a of different potential links between globaliza-
fixed labor supply, then the rise in the real wage tion and growth. Unfortunately, the empirics
that ensues from an opening of trade spells an in- have not kept pace. We still know relatively lit-
crease in the cost of technology adoption and a tle about which mechanisms are operative and
fall in the rate of diffusion. If, instead, the cost is what are their quantitative significance. There
paid in units of the final good, then the widening are several reasons for this. Empirical work
of the gap in relative profitability between high on trade and growth is hampered by a dearth
and low productivity firms encourages a speed- of natural experiments and by the limited num-
up in diffusion. ber of observations we have on what might be
Sampson (2014) tells a related but different considered “the long run.” The cross-country re-
story. In his model, there is free entry by new gression methodology is flawed in this context,
inventors of differentiated products. They draw not only because there are many endogenous
their technologies for producing their inventions variables and few instruments, but also because
from a distribution that reflects the technolo- trade implies that countries’ experiences cannot
gies found among incumbent producers. Sus- meaningfully be treated as independent obser-
tained growth is driven by perpetual improve- vations. Moreover, the relationship between in-
ment of technologies for production, as each tegration and knowledge accumulation ought to
new technology builds on the others. In this vary depending upon the fundamental character-
setting, the positive selection induced by global- istics of a country, including its factor and re-
ization improves the productivity draws for new source endowments and its history. Few empiri-
firms; lower trade costs induces exit by low- cal studies linking growth outcomes to openness
productivity producers and expansion by high- or trade policy have allowed for such a depen-
productivity producers, which improves the dis- dence. although many regressions using cross-
tribution of technology draws and so encourages country data have been computed, they teach us
VOL. VOL NO. ISSUE GLOBALIZATION AND GROWTH 5

little about mechanisms at work. More promis- Grossman, Gene M. and Helpman, Elhanan.
ing have been the studies, such as those reviewed 1991b. “Trade, Knowledge Spillovers, and
by Helpman (2004) and Keller (2010) that shine Growth,” European Economic Review 35(2-
a light on a single, mediating variable, such as 3), 517-26.
the existence and size of international knowl-
edge spillovers. While we may be reasonably Grossman, Gene M. and Helpman, Elhanan.
confident now that such spillovers exist, we still 2014. “Growth, Trade and Inequality,” Na-
do not know very much about how to encourage tional Bureau of Economic Research Working
more of them, or how important they are com- Paper No. 20502.
pared to other factors in determine a country’s Helpman, Elhanan. 2004. The Mystery of Eco-
overall growth performance. And the question nomic Growth. Cambridge MA: Belknap by
of what policies can be used to promote growth Harvard University Press.
in particular countries and settings is far from
settled. Keller, Wolfgang. 2010. “International Trade,
Foreign Direct Investment, and Technology
REFERENCES Spillovers,” In Handbook of the Economics
of Innovation. Vol. 2, edited by B. Hall and
Aghion, Philippe and Howitt, Peter. 1992. “A N. Rosenberg, 793-829. Amsterdam: Elsevier
Model of Growth Through Creative Destruc- B.V.
tion,” Econometrica 60(2), 323-351.
Lucas, Robert E., Jr. 1988. “On the Mechanics
Alvarez, Fernando, Buera, Francisco J. and of Economic Development,” Journal of Mon-
Robert E. Lucas, Jr. 2014. “Idea Flows, Eco- etary Economics 22(1), 3-42.
nomic Growth, and Trade,” unpublished.
Melitz, Marc J. 2003. “The Impact of Trade on
Baldwin, Richard E. and Robert-Nicoud,
Intra-Industry Reallocations and Aggregate
Frédéric. 2008. “Trade and Growth with Het-
Industry Productivity,” Econometrica 71(6),
erogeneous Firms,” Journal of International
1695-1725.
Economics 74(1), 21-34.
Peretto, Pietro F. and Valente, Simone. 2011.
Baldwin, Richard E., Braconier, Henrik
“Resources, Innovation and Growth in the
and Rikard Forslid. 2005. “Multination-
Global Economy,” Journal of Monetary Eco-
als, Endogenous Growth, and Technological
nomics 58(4), 387-99.
Spillovers: Theory and Evidence,” Review of
International Economics 13(5), 945-63. Perla, Jesse, Tonetti, Christopher, and
Coe, David T. and Helpman, Elhanan. 1995. Michael E. Waugh. 2014. “Equilibrium
“International R&D Spillovers,” European Technology Diffusion, Trade, and Growth,”
Economic Review, 39(5), 859-87. unpublished.

Eaton, Jonathan and Kortum, Samuel. 1999. Romer, Paul M. 1990. “Endogenous Techni-
“International Technology Diffusion: Theory cal Change,” Journal of Political Economy
and Measurement,” International Economic 98(5:2), 71-102.
Review 40(3), 537-70. Sampson, Thomas. 2014. “Dynamic Selection:
Eaton, Jonathan and Kortum, Samuel. 2002. An Idea Flows Theory of Entry, Trade, and
“Technology, Geography, and Trade,” Econo- Growth,” unpublished.
metrica 70(5), 1741-79. Young, Alwyn. 1991. “Learning by Doing and
Feenstra, Robert C. 1996. “Trade and Uneven the Dynamic Effects of International Trade,”
Growth,” Journal of Development Economics Quarterly Journal of Economics 106(2), 369–
49(1), 229-56. 405.

Grossman, Gene M. and Helpman, Elhanan.