FDI, Multinationals and Structural Change in Developing Countries
FDI, Multinationals and Structural Change in Developing Countries
FDI, Multinationals and Structural Change in Developing Countries
#2019-004
Maastricht Economic and social Research institute on Innovation and Technology (UNU‐MERIT)
email: [email protected] | website: http://www.merit.unu.edu
André Pineli
Henley Business School, University of Reading, UK
Rajneesh Narula
Henley Business School, University of Reading, UK
Rene Belderbos
KU Leuven, Belgium, UNU-Merit and Maastricht University, Netherlands
Forthcoming in:
Alcorta, L.,Szirmai, A., Verpagen, B. And Foster-McGregor, N. (eds) ‘New
Perspectives on Structural Change: Causes and Consequences of Structural
Change in the Global Economy’, Oxford University Press, 2020
Abstract
Economic development can be defined as a process in which output growth is
accompanied by qualitative changes in the structures of production and employment.
Can FDI affect this process? This paper looks for answers in two ways. First, it reviews
the extant knowledge about the relationship between MNE activity and economic
development in developing countries. Core theoretical and conceptual issues are
presented and the key findings of both microeconomic (FDI linkages and spillovers)
and macroeconomic (FDI-growth nexus) empirical studies are discussed. The main
message of both streams of literature is that FDI has the potential to catalyse
development, but actual outcomes are contingent on several factors, such as the
absorptive capacity of domestic firms and the level of development of local financial
markets. Second, the paper addresses the relationship between FDI and structural
change more directly, in a cross-country context, using a two-step estimation approach
that is consistent with both theoretical arguments and previous empirical findings which
suggest that the FDI-development nexus is highly country-specific. The results confirm
such heterogeneity and suggest that the interaction between the sectoral concentration of
FDI and the development stage of the country plays a role in determining the
development impact of FDI.
Keywords
foreign direct investment, multinational enterprises, developing countries, economic
development, structural change
JEL Classification
D62, F23, L16, O11, O14, O19, O24
Acknowledgements
André Pineli thanks the CAPES Foundation (Brazil) for the financial support of his PhD
studies at the University of Reading
2
1. Introduction
sector of the economy1. As the traditional sector is viewed as stagnant, this move
depends fundamentally on the ‘pulling forces’ accruing from the modern sector, whose
particularly physical and human capital, and increases in their productivity, which in
The hypothesis that countries could remain indefinitely trapped in a “vicious circle of
poverty” was suggested by Ragnar Nurkse in 1953, based on the observation that low
income leads to low savings, which leads to low investment, that leads to low
productivity, that leads to low income, and so on and so forth (Nurkse, 1953). Foreign
direct investment (FDI) might help countries to break out this vicious circle not only by
growth” strategy (Hirschman, 1958) when a few key investments can stimulate further
investments along the value chain. Increased demand for inputs can enable productivity
1
Perhaps more appealing is Hirschman’s (1958, p. 5) claim that development “depends not so much on
finding optimal combinations for given resources and factors of production as on calling forth and
enlisting for development purposes resources and abilities that are hidden, scattered, or badly utilized”.
3
gains due to higher specialization and increasing returns to scale, thereby benefiting
Markusen & Venables, 1999). In addition, the multinational enterprises (MNEs) may
generate knowledge spillovers to domestic firms (Caves, 1974) and improve the
allocation of resources in host economy by causing the exit of the less efficient
producers (Melitz, 2003; Alfaro & Chen, 2018), thus contributing to enhanced
aggregate productivity. All these elements would impact not only the rate of output
This paper provides an overview of the extant knowledge about the relationship
an introduction to the core theoretical and conceptual issues in the next section, the key
findings of the empirical literature on the developmental effects of FDI are discussed in
Section 3. The main contribution of the paper is presented in section 4, where the
paucity of quantitative evidence on the relationship between FDI and structural change
FDI depends on its sectoral concentration and whether this relationship varies according
to the stage of development of the country – two major suggestions emanating from the
literature review. This is examined in a two-step approach. In the first step, long-run
indicator of structural change. Next, a set of country characteristics and the sectoral
concentration of FDI are employed to explain the heterogeneity observed in the FDI-
structural change nexus across countries. This approach reflects both theoretical
arguments and previous empirical findings suggesting that the relationship between FDI
well as the roles of the sectoral concentration of FDI and the development stage in
growth, it is worth briefly presenting how they view the differential effect of
accumulation. As every dollar of investment has the same effect on growth whatever the
source, there is no specific role to MNEs – not even as a source of technology since
states that “nations are poor because their citizens do not have access to the ideas that
are used in industrial nations to generate economic value”. MNEs are firms that have
the potential to create and transfer knowledge across borders, both intentionally and
made by a domestic firm. Furthermore, endogenous growth models emphasize the roles
human capital externalities (Lucas, 1988; Azariadis & Drazen, 1990) in the process of
economic growth, all of them often related with the presence of foreign MNEs in the
host economy.
Another relevant issue is whether the development impact of FDI is contingent on its
“nature”. FDI projects can be broadly classified into two categories: FDI driven by
also known as the “international flying geese model” – FDI contributes to increased
productivity, while also promoting positive structural change in both home and host
industry invests in a host country with a comparative advantage in the same industry
(Kojima, 1973; 1982; Kojima & Ozawa, 1984). For Kojima (1973; 1982; 2000), this
type of FDI improves the allocation of resources and enhances trade. The “wrong” type
oligopolistic structures.
This comparative advantage-based reasoning has recently been revived by Lin &
comparative advantage, or by defying it4. Lin (2010) argues that an economy’s optimal
2
Albeit being the two most relevant motivations for FDI in developing countries they are not the sole
ones. For a contemporary discussion of FDI motives, see Cuervo-Cazurra, Narula & Un (2015).
3
Besides this macroeconomic approach, the theoretical literature on why MNEs exist revolves around
two microeconomic perspectives. A governance explanation is provided by the “internalization school”
(Buckley & Casson, 1976; Rugman, 1980; Hennart, 1982), which predicts that an MNE will emerge when
a domestic firm internalizes the cross-border market of an intermediate product, after weighting
production costs against transaction, contracting, coordination and monitoring costs of different
governance modalities, ranging from full internalization to pure arm’s length transaction. However, some
influential scholars (Hymer, 1960; Dunning, 1977) argued that simply performing a value-adding activity
overseas is not sufficient to transform a domestic firm in an MNE. The firm’s internationalization must be
underpinned by some type of ownership-specific advantage because when competing in foreign markets,
foreign firms face costs that local competitors do not incur. Thus, the “market power” theory of the MNE
emphasizes the role played by the control or access to proprietary assets (technology, brands, channels of
distribution etc.) in conferring MNEs advantages over its competitors in host countries.
4
For a comprehensive discussion of the role played by factor endowments in economic development, see
Dosi & Tranchero (2019).
6
industrial structure follows a change in its endowment structure. Lin & Monga (2011)
propose that following comparative advantage is the optimal strategy to optimise capital
accumulation5. In the same way as in the flying-geese model, FDI may assist
MNE activity can influence the economic structures of host economies but it is also
5
Lin & Monga (2011) suggest that the state should act to identify new industries in which the country
may have latent comparative advantage, remove the constraints that impede the emergence of those
industries and create the conditions to allow them to become the country’s actual comparative advantages,
with the countries that have preceded them being a useful reference as to which industries might offer
latent comparative advantages. If domestic firms are absent in industries in which the country has latent
comparative advantages, the government could adopt specific measures to attract foreign investors that
may have incentives to relocate their production to lower-cost locations. Nonetheless, Chang & Andreoni
(2016) consider Lin & Monga’s (2011) approach inconsistent because it recommends adhering to
comparative advantage while recognising the need to deviate from it.
6
Lin & Monga’s (2014) evaluation that Latin American countries’ import-substitution strategies failed in
achieving structural transformation because they gave priority to the development of the capital-intensive
heavy industry, when those economies were capital-poor, is quite similar to Kojima’s (2000).
7
The so-called Latin American structuralism (Prebisch, 1949) divided the world into two groups of
countries – the centre and the periphery – which differ from each other in terms of technological
capabilities. In the periphery, there is a reinforcing mechanism linking technological capabilities and
patterns of specialization. Upgrading is obstructed because learning is highly dependent on what the
economic agents produce (Porcile, 2019). The low income-elasticity of the products usually exported by
developing countries would impose a deterioration of their terms of trade vis-à-vis the advanced nations
and bind their rate of economic growth consistent with long-run equilibrium in the balance-of-payments.
The remedy for this situation would be a development strategy less dependent on international trade,
which would focus on the building up of a manufacturing sector that would replace imports by goods
produced domestically.
8
Lee (2013) distinguishes low-income from middle-income countries in respect to the best development
strategy. Specialization according to comparative advantage may be advantageous to low-income
countries but is less suitable for middle-income countries that have already passed initial stages of growth
by technology emulation. In his view, sustained catch-up requires not only an engagement with mature
industries, but also an effort to leapfrog into emerging industries that are new to both advanced and
developing countries.
7
framework (Dunning, 1981; Dunning & Narula, 1996; Narula, 1996; Narula &
Dunning, 2010), states that the quantity and the quality of FDI a country receives (and
sends abroad) changes as its domestic firms accumulate assets that enhance their
capacity to explore economic opportunities and to compete with firms from other
location advantages and the market failures that make hierarchies (internalization) to be
between FDI and economic structure follows five stages that are likely to be observed in
every country but whose transition points, in terms of the country’s level of
development, cannot be determined a priori, since they depend on several aspects, such
each country.
structure is also present in Ozawa (1992), where three sequential stages of economic
factor endowments.
9
FDI and MNE activity are largely regarded as synonyms in the empirical literature. This is because most
statistics collected and published by national governments and multilateral institutions are still based on
the balance-of-payments definition of FDI – which is confined to equity investments, reinvested earnings
and intra-firm loans. However, that equivalence is false since MNEs can engage in cross-border value-
adding activities through arrangements that does not necessarily involve FDI such as those related to the
control of global value chains (GVCs). Since data on other modes of MNE activity is still scarce and,
even when it exists, is hardly comparable across countries, this paper follows the extant literature and
considers FDI and MNE activity as synonymous.
8
Given the lack of quantitative studies relating FDI and structural change, this section
focuses on the main findings of two related streams of literature. The first aims to
analyse how the presence of foreign MNEs affect the domestic actors in an economy,
through linkages, externalities and spillovers, using in most cases firm-level data in a
single-country context. The second deals with the relationship between FDI and
and macro approaches complement each other – while the latter allows the assessment
of the effects of local conditions on the way that FDI foster economic development, the
former is needed to investigate the mechanisms through which FDI affect host
economies.
Considering that MNE affiliates tend to be more productive than their domestic
counterparts, which can be largely explained by their differences in terms of size, assets
and capabilities10, aggregate productivity tends to rise when MNEs gain market share in
host countries (Melitz, 2003). However, such direct effects of MNE presence have
The presence of a foreign MNE in a host country produces external effects on other
economic agents, including affiliates of other foreign MNEs and domestic firms,
10
For a comprehensive review of the empirical literature on the differences between domestic and
foreign-owned firms, see Bellak (2004).
11
The competition effect produces mostly pecuniary externalities, which may affect competitors
(horizontal) as well as suppliers and buyers (vertical), in both product and factor markets. By reducing the
monopoly power of domestic firms in some sectors, the entry of MNEs may contribute to enhance
allocative efficiency (Caves, 1974). Although it may encourage domestic firms to be more efficient, more
competition means fewer opportunities to exploit scale economies, with possible (negative) effects on
sectoral productivity (Aitken & Harrison, 1999). The composite effect on competing domestic firms tends
to be negative. They may be crowded-out by foreign competitors. However, if higher competition
translates into lower prices or increased quality, existing domestic firms in downstream sectors will
benefit, and new ones may enter the market (crowding-in effect).
9
turnover13; iv) and backward and forward linkages14. FDI externalities can be of the
pecuniary type – that is, transmitted through prices in the market – or can constitute
Besides these indirect (and mostly unintentional) effects, domestic firms may be
them. Through these linkages15, MNEs can provide technical, managerial and financial
assistance to their suppliers, for example. As the MNE expects a benefit from this type
of relationship, it has incentives to create and deepen backward linkages. However, the
The hypothesis that MNEs produce significant external effects on domestic firms has
been under intense scrutiny over the last three decades. Empirical findings are quite
spillovers studies (Havranek & Irsova, 2011) shows with some clarity is that the
12
Production techniques and managerial practices used by the MNE may be more efficient than those
used by domestic firms – indeed, the ownership of distinctively superior assets is a necessary condition
for the occurrence of FDI according to the market power theory of the MNE (Hymer, 1960; Dunning,
1977). Their use by the MNE “demonstrates” their superior attributes, and local competitors are able to
observe and imitate them.
13
MNEs train their local employees, who accumulate managerial and technical know-how. This acquired
knowledge leaks from the MNE when workers move to a collocated competitor or start their own firm.
However, MNEs seek to minimise such spillovers often by paying above-market salaries to retain such
employees (Fosfuri et. al., 2001), as has been confirmed over the years in countries such as Mexico and
Venezuela (Aitken et al., 1996), Indonesia (Lipsey and Sjoholm, 2004) and China (Chen et al., 2011).
14
The presence of the MNE may affect local firms that can supply inputs or buy their intermediate
products. The MNE’s production itself increases supply for downstream sectors, possibly bringing prices
down. Furthermore, the increased demand created by the MNE may enable domestic suppliers to benefit
from scale and specialization economies, ultimately benefiting any firm that use the same inputs,
including the MNE’s own competitors. Therefore, backward linkages entail positive horizontal
productivity externalities, besides the more obvious vertical ones (Rodriguez-Clare, 1996; Markusen &
Venables, 1999).
15
Lall (1980) defines linkages as the “direct relationships established by firms in complementary
activities which are external to ʻpureʼ market transactions” (Lall, 1980, p. 204). This definition is
narrower than Hirschman’s (1958) classic definition of linkages as it precludes any non-intentional effect
such as pecuniary externalities.
10
productivity of their local suppliers. There is also evidence that the survival odds of
domestic firms in the Czech Republic and Vietnam (Ayyagari & Kosova, 2010; Kokko
& Thang, 2014) is enhanced by the presence of MNEs in downstream industries. Much
less clear, however, is the effect of foreign MNEs on the productivity of their domestic
Another relevant issue is whether foreign MNEs crowd-out domestic firms (in the
same industry) or, on the contrary, their presence produces a crowding-in effect. The
(Kokko & Thang, 2014), crowding-in in the Czech Republic (Ayyagari & Kosova,
2010), and no effect at all in Turkey (Taymaz and Ozler, 2007; Ferragina, 2014). The
impact seems to vary over time. In the Czech Republic, crowding-out prevails in the
short-run, due to increased competition, but as time allows knowledge spillovers to take
place, crowding-in becomes the norm (Kosova, 2010). However, we do not yet know
on this matter is provided by Munemo (2017). Using country-level data he finds that
tend to be negligible (Irsova & Havranek, 2013). This is not surprising given that
negative and positive spillovers transmitted through different channels are likely to
cancel each other out. In addition, the available evidence does not suggest any
16
It must be underlined that empirical studies on FDI spillovers are (generally) incapable of
distinguishing between pecuniary externalities, knowledge spillovers and (intentional) knowledge
transfers.
17
It must be underlined, however, that the proxy used does not enable identifying the industries the new
domestic firms belongs to.
11
labour issues (Farole & Winkler, 2014). In contrast, the occurrence of export spillovers
has been identified for countries as diverse as Chile (Duran & Ryan, 2014), Venezuela
(Aitken & Harrison, 1999), China (Chen et al., 2013), Poland (Cieslik & Hagemejer,
probably the most important indirect effect of the presence of foreign MNEs in host
economies19. A key issue is whether foreign MNEs behave differently from domestic
firms in terms of backward linkages with the domestic economy. Alfaro and Rodriguez-
Clare (2004) find that MNE affiliates operating in Brazil, Chile, Mexico and Venezuela
source a lower share of their inputs domestically compared to local firms. However, the
value of their domestic purchases of inputs per worker employed is higher than the
same ratio observed among domestic firms, except in the case of Mexico. Jordaan
(2011) finds that MNE affiliates are more supportive of their suppliers than domestic
processes.
The motivation for FDI is one of the key determinants of linkages and spillovers. In
18
In the case of China, the presence of foreign MNEs increases the probability of domestic firms’
initiation on a new export market (Mayneris & Poncet, 2015) and the survival odds of an export market
(Swenson and Chen, 2014). Furthermore, the positive influence seems to be more relevant for penetrating
in “difficult” markets, defined as countries with poorer institutional quality and/or tougher import
procedures (Mayneris & Poncet, 2011).
19
Most studies on FDI linkages are still based on qualitative case studies. If, on the one hand, this
research technique permits a deeper understanding of the mechanisms of linkage formation and its
dynamic evolution, on the other hand it hinders a generalization of the findings since the specificities of
each industry or region are hardly found elsewhere. In turn, the quantitative literature on FDI linkages is
relatively underdeveloped, probably reflecting the difficulty of collecting reliable data in developing
countries. Nonetheless, there are clear limits to quantitative approaches to the study of linkages because it
is difficult to measure the transfer of tacit knowledge. For this reason, this section reviews both
quantitative and qualitative studies.
12
affiliates because they are less dependent on low cost inputs to be competitive
(UNCTAD, 2001). However, if FDI is focused on the domestic market but mainly
motivated by tariff-jumping, and trade restrictions are limited to final products, a likely
other suppliers in home country (Belderbos, 1997; Belderbos & Sleuwaegen, 1998),
thus diminishing the potential for linkage creation. Indeed, Belderbos, Capannelli &
circumvent trade barriers create fewer vertical linkages in the local economy.
The quality of linkages is certainly at least as important as their quantity. The nature
of the relationship between foreign affiliates and domestic agents, that is the extent to
which resources and knowledge are transferred between them, is key to the enhance the
potential for learning, improvement and upgrading (Giroud & Scott-Kennel, 2009).
High levels of local sourcing do not necessarily result in technological learning by local
suppliers (Ivarsson & Alvstam, 2005). For this reason, Merlevede & Schoors’ (2009)
finding that backward vertical spillovers are generally positive in Romania but
significantly larger among export-oriented industries does not cause surprise, even
considering that MNEs in these industries tend to source less inputs locally (UNCTAD,
2001).
Other important issue is the differential linkage impact of different “types” of MNE
affiliates. In the context of four European transition economies, Jindra et al. (2009) find
that the share of inputs sourced locally is positively associated with the affiliates’ levels
of autonomy, initiative and technological capability. Similarly, Giroud & Mirza (2006)
find that MNE affiliates that play a strategic role (R&D or marketing) tend to source
more locally than affiliates that solely run a production plant in four Southeast Asian
countries. In a study of Taiwanese MNEs investing in the US, China and Southeast
13
Asia, Chen, Chen & Ku (2004) find that investors in a producer-driven network are
more likely to build local linkages than their counterparts in buyer-driven networks
because the former have more power to promote innovations in the network. According
to Driffield & Noor (1999), MNEs in the Malaysian electronics sector that employ
significant numbers of local managers and engineers source more locally, what they
attribute to the fact that their superior knowledge of local economy reduces transaction
costs of trading with local firms in comparison to MNEs that employ large numbers of
expatriates. The willingness of MNE affiliates to develop local linkages is also affected
proximity to parent firm/other affiliates, and market power (Altenburg, 2000). The
business culture of the home country also affects the extent and depth of linkages.
Japanese MNEs seem to find it more difficult to establish linkages with domestic firms,
but once they do so these tend to be more intense compared to American MNEs
(UNCTAD, 2001).
The breadth and depth of the linkages forged by MNEs in developing countries – as
well as the extent to which potential spillovers materialize – are also contingent on the
characteristics of the domestic sector. Chief among these seems to be the absorptive
(Narula & Marin, 2003). Spillovers need to be internalized and this is not a costless
process (Narula & Driffield, 2012). As highlighted by Ivarsson & Alvstam (2005), even
embodied elements of technology can only be used at best practice levels if they are
complemented by tacit elements that need to be developed locally, what means that
& Strobl, 2001; Narula & Marin, 2003; Blalock & Simon, 2009; Castillo et al., 2014)
indicates that firms with higher absorptive capacity indeed benefit more from foreign
14
underlined by Pavlinek & Zizalova’s (2016) study on the Czech automotive industry:
although the presence of foreign car assemblers benefits most domestic firms in
most domestic firms from benefiting from direct knowledge transfer from MNEs.
affiliates, and thus limit or enhance the potential for linkages development (Lall &
Narula, 2004). In a study of Japanese electronic MNEs, Belderbos, Capannelli & Fukao
(2001) find that host country’s quality of infrastructure and the size of local components
industry positively affect the extent of backward linkages. Local content requirements
positively affect the level of local procurement, as expected, but do not affect the
institutions that provide, for example, training and quasi-public goods, is a key
determinant of the strength of the linkages between MNEs and suppliers, as they affect
the building of absorptive capacity of local firms (Rasiah, 2003). When such institutions
are absent, MNEs may prefer to integrate vertically or source inputs from abroad.
suppliers in the Indian automotive industry (Okada, 2004). Duanmu & Fai (2007)
emphasise that as technical and managerial techniques are transferred by MNEs to their
local suppliers, the business ideology of the suppliers evolves because MNEs need to
20
Interestingly, high local content requirements exert a deterrent effect on foreign investment by Japanese
MNEs. However, Japanese MNEs are more likely than American MNEs to bring their home country
suppliers with them when investing in a production plant abroad (Hackett & Srinivasan, 1998).
15
Case studies help to clarify the evolutionary nature of linkages between MNEs and
domestic firms. From their study of the electronics industry in China, Duanmu & Fai
(2007) identify three stages of the relationship development: initiation, development and
domestic suppliers’ capabilities and the increase of mutual trust. Moving from the
second to the third stage also involves changes in the motivation of FDI – increasingly
evolutionary pattern appears in Giroud (2007), who finds that, as local suppliers’
Malaysia engage in new joint tasks with MNEs, such as joint design of inputs.
Summing up, it can be said that the presence of foreign MNEs potentially affects
both the macro and micro structures of host economies, although the materialization of
such potential depends on several factors, most of which very context-specific. MNEs’
operations generate demand for inputs, skills and capabilities, thus opening up
can be utilized by domestic actors as inputs in new forward industries. The provision of
generate productivity gains for domestic firms, thus enhancing their competitiveness
and survival odds. The presence of foreign MNEs also induces intra-industry firm
selection and market share reallocation (Alfaro & Chen, 2018), giving rise to
There is a substantial empirical literature on the relationship between inward FDI and
GDP growth. Most studies find a positive correlation between these variables,
export promotion strategy, while developing countries that persisted with import-
substitution strategy did not reap the benefits of FDI in terms of enhanced growth. In
driven mainly by factor prices and, thus, fostered economic efficiency. In inward-
oriented countries FDI, was motivated by trade barriers. Excessive protection led to x-
Another influential study is Borensztein et al. (1998). Their results show a positive
correlation between FDI and GDP growth within a sample of developing countries, but
the size of the effect is dependent on the availability of human capital in the host
economy. This suggests that a country needs adequate absorptive capacity to be able to
benefit from the inflow of superior technologies brought along by foreign investors.
Another factor that seems to moderate the effect of FDI on GDP growth is the level
of financial development of the host country. Alfaro et al. (2004) and Durham (2004)
find that a positive association between FDI and economic growth only takes place
among countries that have reached a minimum level of financial development. This
suggests that potential FDI externalities rarely materialize when local entrepreneurial
21
According to Javorcik & Spatareanu’s (2009) study, credit-constrained domestic firms are hindered
from becoming MNE suppliers in Czech Republic.
17
relationship between FDI and economic growth. Alguacil, Cuadros & Orts (2011) find
that the FDI-growth nexus is stronger among a group of low and lower-middle income
countries when the Economic Freedom index is lower. They justify their finding on the
basis that in these economies the shortage of capital means that FDI is the only option to
increase the rate of accumulation. In these countries FDI would be less likely to crowd-
out domestic investment. Jude & Levieuge (2017) employ several measures of
institutional quality and find a positive association between FDI and GDP growth only
for countries above certain thresholds of institutional quality. However, the effect of
FDI on growth seems to be independent of the levels of political stability and control of
corruption. Harms & Meon (2012) distinguish FDI flows in the form of greenfield
projects from mergers and acquisitions (M&As). While greenfield FDI has a positive
impact on GDP growth among developing countries, M&As has no effect. They fail to
Despite the relevance of the cited studies, a few econometric issues cast doubt on
their findings. Blonigen & Wang (2005) question the adequacy of pooling data from
advanced and developing economies in the same sample, as done by Alfaro et al. (2004)
and Durham (2004), among others22. Studies have also indicated that the causality
between FDI and growth can be mutual (Basu et al., 2003; Li & Liu, 2005; Hansen &
Rand, 2006) or even reverse, as found by Basu et al. (2003) for a group of relatively
closed developing economies. Another problem that is often overlooked, but was
pointed out by Choe (2003), is the overwhelming influence of outliers on the results of
econometric specifications may exert a big influence on the results. Indeed, FDI is
22
When replicating Borensztein et al.’s (1998) study with a sample that includes advanced countries,
Blonigen & Wang (2005) fail to find the positive effects of FDI on growth found in the original study that
included only developing countries.
18
found to exert no effect (Carkovic & Levine, 2005) or even to be harmful to growth
(Herzer, 2012) when country-specific heterogeneous effects are accounted for. Possibly
the most important lesson to be extracted from the extant literature is that the
the conclusion of Kottaridi & Stengos (2010, p. 866-7), who after estimating a semi-
parametric model, affirm that “it appears that the way FDI affects growth differs across
and within countries. The relationship seems to be complex and the impact varies
according to a country’s level of FDI. (…) parameter heterogeneity may exist in the
sense that the effect of a change in a particular variable is not the same. (…) In other
understanding of the development effects of FDI, although the issue of structural change
is seldom addressed directly. The apparent disinterest on the topic has certainly been
influenced by the lack of adequate FDI data at the industry level – or even the sector
level – for a relatively lengthy period and for a reasonable number of developing
acceptable alternative. The main objective of the empirical exercise presented in this
developing countries hosting the activities and under what circumstances. More
specifically, it examines the potential influence of FDI on the move of the labour force
The literature reviewed in section 2 proposes, on the one hand, that FDI can promote
structural change when it conforms to comparative advantages. On the other hand, the
19
is difficult given the data limitations. It would require an FDI database that classifies
production employed. A specific manufacturing plant can require quite different factors
of production depending on the activities that are performed in the country. Upstream
activities such as R&D are intensive in skilled labour, while other manufacturing
available databases provide this level of detail, as investments are only distinguished by
broad industry classifications. Given the available data, the best that can be done is a
sectoral classification distinguishing FDI into the traditional sector, the manufacturing
sector and the non-manufacturing modern sector. Thus, the main hypotheses to be tested
in the following analysis are whether the development impact of FDI depends on its
sectoral concentration and whether this relationship varies according to the stage of
FDI, such as Ozawa’s (1992) framework and the IDP. In addition, the paper builds on
the empirical literature reviewed in section 3.2 to identify other factors that may help to
The analysis is undertaken in two steps. In the first step, a panel time-series method is
in the modern sector. In the second step, a set of variables is employed to explain the
cross-country differences verified in the first step. This procedure seems preferable to a
standard panel data estimation with interaction terms because it is much more in the
spirit of the IDP framework, which highlights the idiosyncratic nature of the
20
relationship between the level of MNE activity associated with a country and its level of
empirical studies (Kottaridi & Stengos, 2010; Herzer, 2012) that allow for heterogeneity
beyond simple interaction terms have shown that the FDI-growth nexus varies
specified based on the usual aggregate production function. With the employment share
of the modern sector replacing output in the left-hand side, this leads to equation 1:
𝛼 𝛿𝑡 𝛽 ln 𝛽 ln 𝛽 ln 𝐻𝐶 𝑢 (equation 1)
The rationale of the model is the following: in dual economies, economic growth
comes from two sources: i) increases in capital/labour ratios and technological progress
in the modern sector; ii) labour force movements from the stagnant traditional sector to
the modern sector. Even considering that additions to the physical capital and human
that the widening of the productivity gap tends to drain factors of production from the
traditional to the modern sector, as pointed out by Lewis (1954)23. Thus, as pulling
factors dominate pushing factors in driving labour out of the traditional sector, it can be
assumed that the same factors that determine the aggregate output level are likely to
determine the level of employment in the modern sector. However, this association is
not automatic. Rapid economic growth can take place with negligible labour
movements – this is usually what happens when a developing country discovers large
23
The historical experience of today’s developed countries indeed indicates that, at least in initial
development stages, technological advancement in the manufacturing sector drives labour movement out
of (traditional) agriculture (Alvarez-Cuadrado & Poschke, 2011).
21
oil reserves24. Thus, the way that FDI and other growth determinants affect the
employment structure depends on how they affect the demand for labour of the leading
The working age population is used as denominator in calculating the modern sector
developing countries. The same variable is used to bring the FDI stocks and domestic
capital stocks to a “per worker” basis. In respect to human capital, the paper follows the
approach proposed by Hall & Jones (1999), in which human capital stock per working
age person is an exponential function of average years of schooling, where the function
ϕ(s) reflects the efficiency of a unit of labour with s years of schooling relative to one
with no schooling, and its derivative is the return to schooling estimated in a Mincerian
regression.
𝐻𝐶 𝑒 (equation 2)
The empirical model also includes country-specific constants and time trends (t) to
measure of technological gap. However, its use in cross-country analysis is even more
controversial than in the firm-level context because the residuals of the aggregate
24
Employment is used instead of output because it better captures the dual economy nature of developing
countries.
22
production function reflects not only technological level, but allocative and productive
4.2 Data
The dataset used in the analysis is comprised by 28 developing countries over the
period 1980-2010. Definitions of the variables as well as their respective sources are
presented in table 1.
HC Human capital per working age person Penn World Table 9.0
Index based on average years of schooling of the WAP and
returns to education
A unit is subtracted from the Penn World Table HC index so
that HC=0 when average years of schooling=0
Given the inadequacy of the data for some key concepts used in the empirical
exercise, a few adaptations are needed. First, what constitutes the traditional sector and
the modern sector of an economy has to be redefined because the classical structuralist
definition (Lewis, 1954) is not reflected in the available statistics, which disaggregate
the definition adopted in this paper – which follows closely Lavopa & Szirmai (2014;
2019), who distinguish industries based on a few key characteristics such as their
potential for scale economies and technological upgrading – the modern sector
construction, transport, storage and communication, finance, insurance, real estate and
business services. In turn, the traditional sector comprises all the economic activity
community, social and personal services. This redefinition clearly bears some degree of
However, as the focus of the empirical exercise is on employment, the distortions are
For the estimation of equation 1, countries’ total physical capital stocks are divided
there is no data on the latter, FDI stocks are used as proxy, while domestic-owned
capital stocks are obtained by subtracting FDI stocks from the total physical capital
stocks. Using this criterion in a conventional growth regression in which the investment
rate belongs to the set of explanatory variables could be problematic because FDI flows
do not necessarily translate into capital formation (Farla, Crombrugghe & Verspagen,
2016). The concept of FDI relates to international financial flows and the respective
24
statistics are drawn from the balance of payments, not from national accounts. Part of
these financial flows is used to acquire existing assets instead of creating new ones.
However, as the explanatory variables of this empirical exercise refer to capital stocks
instead of flows, the potential measurement errors are attenuated because even acquired
4.3 Results
The coefficient of interest in equation 1 is β1. A positive β1 means that a higher level of
(Lavopa & Szirmai, 2014; 2019), a positive β1 can be interpreted as a signal that FDI
As statistical tests suggest that all the variables in equation 1 have a unit root and are
may produce spurious results. Instead, the panel dynamic ordinary least squares
(cointegration) relationship of the variables. Table 2 shows the results of the PDOLS
employment structural change across the sample of developing nations. Doubling the
25
For space reasons, panel unit root tests and panel cointegration tests are not presented here but are
available with the authors.
26
A remarkable advantage of the PDOLS is that it grants more flexibility to account for heterogeneous
cointegrating vectors across countries than other estimation techniques that usually impose a unique
cointegration vector for every country. In addition, by including lead and lag differences of the regressors,
the PDOLS account for serial correlation and endogeneity of the regressors. This feature is important in
the present context because, as underlined by the IDP framework, FDI affects economic structure but is
also determined by it. However, a disadvantage of the PDOLS estimator is the fact that it assumes cross-
sectional independence – what means no correlation between the residuals of different individuals –
except for common time effects (time dummies).
25
FDI/WAP ratio would imply, on average, an increase of half a percentage point in the
MODEMP/WAP ratio.
Table 3 displays the country-specific β1’s. They indicate the existence of marked
differences in the long-run relationship between FDI and employment structure across
countries. In some countries, like Colombia, Malaysia and Thailand, the semi-elasticity
is quite high – an increase of 17% in FDI/WAP would suffice to increase the long-run
MODEMP/WAP ratio in these countries by one percentage point. On the other hand, for
which country features enter as explanatory variables for the (PDOLS regression)
country-specific β1’s (hereafter called BETAs). The choice of variables follows mainly
the empirical literature on the FDI-growth nexus and includes financial development
capital and control of corruption27. Considering that the space for marginal increases in
the MODEMP/WAP ratio is larger the lower the initial level is, the values observed in
1980 are also included in the OLS regression. To test the hypothesis that the
Whenever possible, the variables are averaged for the whole 1980-2010 period, to
better reflect the average conditions faced in the countries. However, in a few cases,
assumptions about the variable’s behaviour over the entire period are needed due to data
limitations. In the case of the variable control of corruption, the indicator for 1996 – the
first year available for all countries in the sample – is used. For the sectoral
concentration of FDI, the procedure adopted is more complex and, certainly, more
controversial – for this reason, the results need to be interpreted cautiously. Since
several countries of the sample do not have FDI statistics disaggregated by sector
covering the period under analysis, the only possible source of this kind of data is FDI
Times group. However, this database started only in 2003, what means that it does not
cover three quarters of the period analysed. In the following estimation, it is assumed
that the sectoral distribution of FDI among the manufacturing sector, the non-
manufacturing modern sector and the traditional sector in the period 1980-2010 was
similar to the distribution of greenfield FDI projects among the same sectors in the
expenditure is reliable, even though for many projects this indicator is not based on
27
The other individual components of the Worldwide Governance Indicators were also tested, with
qualitatively similar results, although not always statistically significant. As they are strongly correlated,
only the results for the variable control of corruption are presented here.
27
Given that there are only 28 observations, a parsimonious model is desirable in order
to preserve degrees of freedom. So, initially, a series of estimations is run with each
estimated with all the variables entering together, differing from each other only by the
FDI variable included. Finally, since the “right” model is unknown, all the possible
selected according to Akaike information criterion (AIC). The results are presented in
table 4.
Table 4 - Determinants of the long-run relationship between FDI and employment structure (β1’s)
model 1 model 2 model 3 model 4 model 5 model 6 model 7 model 8 model 9 model 10
constant 0.005 0.013 -0.005 -0.011 0.003 0.005 0.000 -0.031 -0.009 -0.033
(0.34) (0.45) (-0.47) (-0.87) (0.47) (0.31) (0.00) (-0.84) (-0.19) (-2.15)
initial MODEMP/WAP 0.001 0.103 0.087
(0.01) (0.66) (0.56)
HC -0.004 -0.017 -0.014
(-0.27) (-0.72) (-0.58)
credit/GDP 0.027 0.031 0.032 0.036
(1.26) (1.18) (1.22) (1.55)
trade/GDP 0.034 0.049 0.046 0.039
(1.45) (1.74) (1.64) (1.58)
control of corruption -0.010 -0.027 -0.026 -0.024
(-1.02) (-2.07) (-1.92) (-2.22)
FDI manufacturing 0.001 0.036
(0.03) (0.86)
FDI non-manufacturing 0.011 -0.023
modern sector (0.31) (-0.53)
R-squared 0.00 0.00 0.06 0.07 0.04 0.00 0.00 0.28 0.27 0.25
AIC -108.4 -108.5 -110.1 -110.6 -109.5 -108.4 -108.5 -107.7 -107.1 -112.5
Note: t-statistics in parentheses.
The explanatory power of the simple regressions (1-7) is very low. When other
factors are controlled for, openness to trade and financial development seems to
strengthen the FDI-structural change nexus, but only the former variable is statistically
28
Using the number of projects instead could seem a better alternative since this variable would be less
prone to measurement error. However, as the fixed capital per project tends to vary considerably across
sectors, this procedure would tend to underestimate the real share of the extractive sector, while
overestimating the share of the tertiary sector.
28
significant. In turn, the sectoral composition of FDI seems to have little relevance to
that relationship29. Likewise, the ability of FDI to promote structural change does not
seem to be affected by a country’s level of human capital. This result, however, should
be taken cautiously because the PDOLS regression already included human capital
among the regressors and this may have captured most of its effect on structural change.
The variable control of corruption appears as significant, with a negative sign. Finally,
the initial MODEMP/WAP level does not seem to affect BETAs significantly.
According to AIC, the best fitted model contains three explanatory variables: financial
All the models presented in table 4 implicitly assume that the FDI-structural change
nexus does not differ substantially at different stages of development. However, the
importance of the factors included in the second step of the analysis may change as
indeed suggest that the type of investment a country attracts as well as its development
effect tend to vary as a country becomes richer. To test this hypothesis, models 3 to 7
are re-estimated, with countries divided into two groups, according to their initial
Table 5 - Determinants of the long-run relationship between FDI and employment structure
(β1’s) - Testing the homogeneity of coefficients across subsamples
model 3A model 4A model 5A model 6A model 7A
FDI FDI non-
Sample credit/ trade/ control of
manufactur manufactur
GDP GDP corruption
ing ing modern
initial MODEMP/WAP below median 0.048 0.030 -0.002 0.102 -0.083
(1.08) (0.66) (-0.08) (2.08) (-1.65)
initial MODEMP/WAP above median 0.028 0.037 -0.017 -0.078 0.082
(0.97) (1.25) (-1.21) (-1.77) (1.86)
F-statistic (equality of coefficients) 0.15 0.02 0.46 7.46 6.06
all countries (Table 4) 0.027 0.034 -0.010 0.001 0.011
(1.26) (1.45) (-1.02) (0.03) (0.31)
Notes: t-statistics in parentheses. Coefficients of the group dummies not shown.
29
When the shares of manufacturing and non-manufacturing modern sector in FDI are aggregated in a
single variable, it remains statistically insignificant.
29
statistics cannot reject the hypothesis that both groups of countries have the same
coefficient. Nonetheless, for the FDI variables, the differences between the two groups
In view of these findings, new models, that allow for different effects of the FDI
estimated. Table 6 presents the best fitted models according to AIC. Both models 11
and 12 improves the AIC statistic obtained in model 10. The signs of the FDI
concentration coefficients do not change and their size and significance are not much
impacted by the inclusion of other variables (when compared to models 6 and 7 in table
5). Once again F-statistics reject the equality of the FDI concentration coefficients of the
development, but not at more advanced development stages. The opposite seems to
Controlling for the share of FDI in the traditional sector, in model 13, does not
fundamentally change the results. In comparison to model 10, the inclusion of the FDI
concentration variables increases the size and the statistical significance of the financial
development coefficient but has the opposite effect on the openness coefficient.
30
Table 6 - Determinants of the long-run relationship between FDI and employment structure (β1’s)
model 11 model 12 model 13
constant -0.006 -0.075 0.003
(-0.25) (-2.73) (0.09)
dummy initial MODEMP/WAP below median -0.078 0.095 -0.073
(-3.15) (2.81) (-2.27)
credit/GDP 0.050 0.048 0.051
(2.34) (2.14) (2.33)
trade/GDP 0.033 0.035 0.032
(1.53) (1.54) (1.42)
control of corruption -0.025 -0.021 -0.022
(-2.43) (-1.79) (-1.79)
FDI manufacturing
initial MODEMP/WAP below median 0.126 0.116
(2.97) (2.55)
initial MODEMP/WAP above median -0.072 -0.080
(-1.73) (-1.75)
FDI non-manufacturing modern sector
initial MODEMP/WAP below median -0.102
(-2.24)
initial MODEMP/WAP above median 0.079
(1.69)
FDI traditional sector
initial MODEMP/WAP below median -0.085
(-0.72)
initial MODEMP/WAP above median -0.057
(-0.40)
F-statistic (equality of FDI coefficients) 12.26 9.11 2.91
p-value 0.00 0.01 0.03
R-squared 0.54 0.48 0.55
AIC -119.8 -116.7 -116.8
Notes: t-statistics in parentheses. When FDI in the traditional sector is included ( as in model 13), models 11 and 12 are
econometrically equivalent.
The fact that the cross-country growth literature is plagued by outlier-driven results
is well-known30. Since the empirical model used here is inspired by the aggregate
production function, there is a considerable risk that the results obtained are driven by
12. One-by-one, each country observation is excluded and the models re-estimated with
the remaining countries. It should be underlined, however, that this sensitivity analysis
is merely illustrative because the exclusion of any alleged outlier from the OLS
regression would imply its exclusion from the PDOLS regression as well, a procedure
30
For example, the estimated effect of a bunch of macroeconomic variables such as inflation, openness
and government consumption on growth is demonstrated by Easterly (2005) to be driven by outliers,
usually a few countries with extremely bad policies and negative growth.
31
that would affect the BETAs used as dependent variables in the OLS regression. The
For the FDI concentration variables, it must be underlined, first, that the equality of
coefficients between the two groups of countries is always rejected. Among the
countries at initial stages of development, Nigeria is the only country that seems to
affect the coefficients of FDI concentration substantially. Its inclusion in the sample
weakens the effect. Therefore, the positive association between FDI in manufacturing
and BETAs within this group does not seem to be driven by outliers. In the other group,
Colombia is very influential. When this country is excluded from the sample, the FDI
and Malaysia, in model 11, and Brazil and Mexico, in model 12.
It seems clear that the effect of openness on BETA is strongly affected by the
inclusion of Malaysia in the sample. When this country is excluded, the coefficient
drops to almost zero. Although Brazil and Colombia affect the coefficient in the
opposite direction, even their combined effect is insufficient to counterbalance the effect
of Malaysia. Thus, the actual effect of openness on BETA is likely to be smaller than
In the case of financial development, there seems to be two outliers affecting the
results, pushing into opposite directions. The inclusion of China lowers the coefficient
while the inclusion of Ghana increases it. However, the coefficient of financial
development is always positive and remains significantly different from zero in almost
countries, except for Venezuela. When this country is excluded from the sample, the
None 0.050 (2.34) 0.033 (1.53) -0.025 (-2.43) 0.126 (2.97) -0.072 (-1.73) 12.26 0.00 0.54
Notes: t-statistics in parentheses. Changes in coefficient higher than 20 percent are highlighed in bold.
33
None 0.048 (2.14) 0.035 (1.54) -0.021 (-1.79) -0.102 (-2.24) 0.079 (1.69) 9.11 0.01 0.48
Notes: t-statistics in parentheses. Changes in coefficient higher than 20 percent are highlighed in bold.
4.4 Discussion
The findings confirm that the development effects of FDI are highly country-specific (as
in Kottaridi & Stengos (2010) and Herzer (2012)), thus justifying the choice for the
development of the countries and the type of FDI they receive, as suggested by the
initial development stages, the effects of FDI on the employment structure are larger
stages, the effects are larger when FDI is more concentrated in the non-manufacturing
34
modern sector. This occurs because the development effects of FDI projects are affected
by the endowments and capabilities a country has, which in turn are associated with its
development stage. Although the data only allows a very rough classification of FDI
projects, it can be said that FDI is more likely to promote structural change when there
is a certain alignment of the type of FDI to the stage of development of the country.
The extent to which openness favours a higher effect of FDI on structural change is
quite uncertain. Although the main results suggest a positive effect, sensitivity analysis
shows that they are considerably affected by one outlier. When Malaysia is excluded
from the sample, the effect of openness disappears. However, this result is not
surprising. Indeed, Balasubramanyam et al.’s (1996) early finding that FDI promoted
1970-1985 have not been confirmed by studies that used more recent data such as
Carkovic & Levine (2005) and Herzer (2012), which find no moderating role for trade
The theoretical argument that the materialization of the potential indirect effects of
FDI depends on the level of development of the local financial market is corroborated
by the empirical analysis. Indeed, the FDI-structural change nexus seems to be stronger
the higher the financial development of the country. This result, which is shown to be
Finally, the finding that the FDI-structural change nexus is stronger where the control
of corruption is lower may seem quite odd at first sight as it could suggest that
Studies using both firm-level (Javorcik & Wei, 2009) as well as country-level (Habib &
Zurawicki, 2002; Hakkala, Norback & Svaleryd, 2008) data indicate that corruption has
a detrimental effect on inward FDI. Thus, a possible interpretation for this unexpected
35
finding is that higher levels of corruption may discourage FDI but, once it has taken
place, the returns tend to be higher in less business-friendly environments because the
risk of crowding out domestic investment is lower. Indeed, this resembles the
5. Conclusions
In 1988, John Dunning, a pioneer in the field of international business studies, wrote
that:
of W. W. Rostow’s model of the economic growth process – first presented in the late
theoretical/conceptual area, endogenous growth models made way for differentiating the
incorporated FDI into a stages approach to the process of development. The IDP
(Dunning & Narula, 1996; Narula, 1996; Narula & Dunning, 2010) which described the
Considerably less progress was observed within the literature concerned specifically
with structural change in terms of incorporating MNEs and FDI in a meaningful way.
31
It is also worth to cite a study done by D’Amelio, Garrone & Piscitello (2016), which finds that
(aggregate) FDI promotes access to electricity in Sub-Saharan Africa and the effects seems to be stronger
where the institutions are weaker.
36
Recent contributions such as Lin (2010) still devote a secondary role to MNEs in
structural change. In such a scenario, it is not surprising that empirical studies on FDI
and structural change are almost non-existent. For this reason, one of the objectives of
this paper was to review the extant literature on related issues regarding FDI and
The emergence of new comprehensive datasets in the last three decades enabled
researchers to put many hypotheses about the development effects of FDI under
scrutiny. Old ideas about MNE linkages and spillovers were tested using firm-level
FDI has the potential to catalyse development, but actual outcomes are contingent on
several factors, such as the absorptive capacity of domestic firms and the level of
This paper adds to the empirical literature by providing a more direct assessment of
the relationship between FDI and structural change. The main hypotheses tested were
whether the development impact of FDI depends on its sectoral concentration and
whether this relationship varies according to the stage of development of the country.
reflect both the theoretical proposition (IDP) and the empirical finding that the FDI-
growth nexus is highly country-specific. In the first stage, long-run coefficients relating
FDI to the employment structure were estimated for each country using a panel-times
series method (PDOLS). In the second stage, a set of variables (some of which
borrowed from the empirical literature on FDI and growth) were employed to explain
The results indicate that the FDI-structural change nexus is quite heterogeneous
across countries. FDI is shown to be positively associated with increases in the modern
sector’s share in employment in some countries but negatively associated in others. The
second stage indicates that the degree of matching between the stage of development of
a country and the type of FDI it receives affects the capacity of FDI to promote
associated with structural change. This finding suggests that there are crucial differences
in the ability of countries to provide the capabilities required in these broadly defined
associated with the financial development and the (lack of) control of corruption of the
consider, when formulating their policies to attract MNE activity – which are often
synonyms to subsidies – that the extent to which FDI promotes structural change will
depend on the alignment between the type of investments a country receives and its
stage of development. Some activities may require capabilities that are in short supply
in developing countries, thus reducing the effects of the MNE presence. Initiatives that
contribute to expand and deepen local financial markets may increase the potential
development effects of FDI. Even though FDI is not a sine-qua-non for development, it
The study has some important limitations. The estimations are not underpinned by a
formal theoretical model – instead the empirical specification was borrowed from the
economic growth literature. The full set of variables belonging to the theoretical and
empirical models remains unknown. There are also data-related issues, in particular the
lack of accuracy of the FDI concentration variables and the small sample that could be
used in the analysis. Substantial efforts to improve FDI data, making it comparable
across countries at more disaggregated levels, are essential to advance the research
References
Aitken, B.; Harrison, A. (1999) Do domestic firms benefit from foreign direct
Aitken, B.; Harrison, A.; Lipsey, R. (1996) Wages and foreign ownership: A
comparative study of Mexico, Venezuela and the United States. Journal of International
Alfaro, L.; Chanda, A.; Kalemli-Ozcan, S.; Sayek, S. (2004) FDI and economic growth:
The role of local financial markets. Journal of International Economics, 64, 89-112.
Alfaro, L.; Chen, M. (2018) Selection and market reallocation: Productivity gains from
Alfaro, L.; Kalemli-Ozcan, S.; Sayek, S. (2009) FDI, productivity and financial
Alguacil, M.; Cuadros, A.; Orts, V. (2011) Inward FDI and growth: The role of
496.
policies. In: UNCTAD (Ed). TNC-SME linkages for development: Issues – experiences
Anwar, S.; Nguyen, L. (2011) Foreign direct investment and export spillovers: Evidence
Ayyagari, M.; Kosova, R. (2010) Does FDI facilitate domestic entry? Evidence from
Balasubramanyam, V.; Salisu, M.; Sapsford, D. (1996) Foreign direct investment and
Basu, P.; Chakraborty, C.; Reagle, D. (2003) Liberalization, FDI, and growth in
Belderbos, R. (1997) Antidumping and tariff jumping: Japanese firms’ DFI in the
European Union and the United States. Weltwirtschaftliches Archiv, 133(3), 419-457.
Belderbos, R.; Capannelli, G.; Fukao, K. (2001) Backward vertical linkages of foreign
29(1), 189-208.
Belderbos, R.; Sleuwaegen, L. (1998) Tariff jumping DFI and export substitution:
16, 601-638.
41
Bellak, C. (2004) How domestic and foreign affiliates differ and why does it matter?
Blalock, G.; Simon, D. (2009) Do all firms benefit equally from downstream FDI? The
Blonigen, B.; Wang, M. (2005) Inappropriate pooling of wealthy and poor countries in
empirical FDI studies. In: Moran, T.; Graham, E.; Blomstrom, M. (Eds.) Does foreign
Economics.
Borensztein, E.; De Gregorio, J.; Lee, J. (1998) How does foreign direct investment
Buckley, P.; Casson, M. (1976) The future of the multinational enterprise. London:
Macmillan.
Carkovic, M.; Levine, R. (2005) Does foreign direct investment accelerate economic
growth? In: Moran, T.; Graham, E.; Blomstrom, M. (2005) Does Foreign Direct
Economics.
Castillo, L.; Salem, D.; Moreno, J. (2014) Foreign direct investment and productivity
spillovers: firm-level evidence from Chilean industrial sector. Latin American Business
Chang, H.; Andreoni, A. (2016) Industrial policy in a changing world: Basic principles,
neglected issues and new challenges. Paper presented at the Cambridge Journal of
Chen, T.; Chen, H.; Ku, Y. (2004) Foreign direct investment and local linkages. Journal
Chen, C.; Cheng, Y.; Findlay, C. (2013) Export spillovers of FDI on China’s domestic
Chen, Z.; Ge, Y.; Lai, H. (2011) Foreign direct investment and wage inequality:
Chenery, H. (1977) Transitional growth and world industrialisation. In: Ohlin, B.;
Chenery, H. (1979) Structural change and development policy. New York: Oxford
University Press.
Choe, J. (2003) Do foreign direct investment and gross domestic investment promote
Cuervo-Cazurra, A.; Narula, R.; Un, C. (2015). Internationalization motives: Sell more,
buy better, upgrade and escape. Multinational Business Review, 23(1), 25-35.
D’Amelio, M.; Garrone, P.; Piscitello, L. (2016) Can multinational enterprises light up
Dosi, G.; Tranchero, M. (2019) The role of comparative advantage and endowments in
Driffield, N.; Noor, A. (1999) Foreign direct investment and local input linkages in
Duanmu, J.; Fai, F. (2007) A processual analysis of knowledge transfer: From foreign
enterprise: A search for an eclectic approach. In: Hesselborn, P.; Ohlin, B.; Wijkman, P.
64.
Hyman.
Dunning, J.; Narula, R. (1996) The investment development path revisited: some
emerging issues. In: Dunning, J.; Narula, R. (Eds.) Foreign Direct Investment and
Duran, I.; Ryan, M. (2014) Spillover effects from inward FDI and the exporting
14, 393-414.
Durham, J. (2004) Absorptive capacity and the effects of foreign direct investment and
48, 285-306.
44
Easterly, W. (2005) National policies and economic growth: A reappraisal. In: Aghion,
P.; Durlauf, S. (Eds.) Handbook of economic growth, Vol 1A. Amsterdam: North
Holland.
Farole, T.; Winkler, D. (2014) The role of mediating factors for FDI spillovers in
developing countries: Evidence from a global dataset. In: Farole, T.; Winkler, D. (Eds.)
Making foreign direct investment work for Sub-Saharan Africa: Local spillovers and
Fosfuri, A.; Motta, M.; Ronde, T. (2001) Foreign direct investment and spillovers
Giroud, A. (2007) MNEs vertical linkages: The experience of Vietnam after Malaysia.
Giroud, A.; Mirza, H. (2006) Factors determining supply linkages between transnational
Hackett, S.; Srinivasan, K. (1998) Do suppliers switching costs differ across Japanese
and US multinational firms? Japan and the World Economy, 10, 13-32.
Evidence from Swedish multinational firms. Review of Economics and Statistics, 90(4),
627-642.
Hall, R.; Jones, C. (1999) Why do some countries produce so much more output per
Hansen, H.; Rand, J. (2006) On the causal links between FDI and growth in developing
Harms, P.; Meon, P. (2012) Good and bad FDI: The growth effects of greenfield
investment and mergers and acquisitions in developing countries. Paper presented at the
European Trade Study Group Fourteenth Annual Conference, Leuven, 13-15 September
2012.
Havranek, T.; Irsova, Z. (2011) Estimating vertical spillovers from FDI: Why results
vary and what the true effect is. Journal of International Economics, 85(2), 234-244.
Hennart, J. (1982) A theory of the multinational enterprise. Ann Arbour, MI: University
of Michigan Press.
Herzer, D. (2012) How does foreign direct investment really affect developing
University Press.
Ivarsson, I.; Alvstam, C. (2005) Technology transfer from TNCs to local suppliers in
developing countries: A study of AB Volvo’s truck and bus plants in Brazil, China,
Javorcik, B.; Spatareanu, M. (2009) Liquidity constraints and firms’ linkages with
markets: Firm-level evidence. Journal of International Money and Finance, 28, 605-
624.
Jindra, B.; Giroud, A.; Scott-Kennel, J. (2009) Subsidiary roles, vertical linkages and
44, 167-179.
Jordaan, J. (2011) FDI, local Sourcing, and supportive linkages with domestic suppliers:
Jude, C.; Levieuge, G. (2017) Growth effect of foreign direct investment in developing
economies: The role of institutional quality. The World Economy, 40(4), 715-742.
11, 375-401.
Kojima, K.; Ozawa, T. (1984) Micro and macro economic models of foreign direct
Kokko, A.; Thang, T. (2014) Foreign direct investment and the survival of domestic
Kosova, R. (2010) Do foreign firms crowd out domestic firms? Evidence from the
Kottaridi, C.; Stengos, T. (2010) Foreign direct investment, human capital and non-
Kuznets, S. (1966) Modern economic growth: Rate, structure, and spread. New Haven,
Lall, S.; Narula, R. (2004) Foreign direct investment and its role in economic
volume).
Li, X.; Liu, X. (2005) Foreign direct investment and economic growth: an increasingly
Policy Research Working Papers, no. 5197, Washington, DC: World Bank.
Lin, J.; Chang, H. (2009) DPR Debate: Should industrial policy in developing countries
conform to comparative advantage or defy it? Development Policy Review, 27(5), 483-
502.
Lin, J.; Monga, C. (2011) Growth identification and facilitation: The role of the State in
Lin, J. & Monga, C. (2014) The evolving paradigms of structural change. In: Currie-
Alder, B.; Kanbur, R.; Malone, D.; Medhora, R. (Eds.) International development:
Lipsey, R.; Sjoholm, F. (2004) Foreign direct investment, education and wages in
Markusen, J.; Venables, A. (1999) Foreign direct investment as a catalyst for industrial
Mayneris, F.; Poncet, S. (2011) Entry on difficult export markets by Chinese domestic
firms: The role of foreign export spillovers. CEPII Working Paper n. 2011-32.
Mayneris, F.; Poncet, S. (2015) Chinese firms’ entry to export markets: The role of
is there so little evidence? In: Van Tulder, R.; Verbeke, A.; Strange, R. (Eds.)
13(3), 586-602.
Narula, R.; Driffield, N. (2012) Does FDI cause development? The ambiguity of the
evidence and why it matters. European Journal of Development Research, 24(1), 1-7.
38(3), 263-287.
Narula, R.; Marin, A. (2003) Foreign direct investment spillovers, absorptive capacities
globalization: Lessons from the Indian automobile industry. World Development, 32(7),
1265-1288.
Pavlinek, P.; Zizalova, P. (2016) Linkages and spillovers in global production networks:
16, 331-363.
Rasiah, R. (2003) Systemic coordination and the knowledge economy human capital
307.
Romer, P. (1986) Increasing returns and long-run growth. Journal of Political Economy,
94(5), 1002-1037.
Romer, P. (1993) Idea gaps and object gaps in economic development. Journal of
Rostow, W. (1959) The stages of economic growth. Economic History Review, 12(1), 1-
16.
Swenson, D.; Chen, H. (2014) Multinational exposure and the quality of new Chinese
Taymaz, E.; Ozler, S. (2007) Foreign ownership, competition, and survival dynamics.
Investment Report 2001: Promoting Linkages. United Nations: New York and Geneva.