Foreign direct investment and economic growth in less developed countries: an empirical study of...
Transcript of Foreign direct investment and economic growth in less developed countries: an empirical study of...
This article was downloaded by: [University of North Carolina Charlotte]On: 19 September 2013, At: 05:04Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK
Applied EconomicsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/raec20
Foreign direct investment and economic growth in lessdeveloped countries: an empirical study of causalityand mechanismsMousumi Duttaray a , Amitava K. Dutt b & Kajal Mukhopadhyay ca American Express Company, NY 10285, USAb Department of Economics and Policy Studies, University of Notre Dame, Notre Dame, IN46556, USAc Discover Financial Services, IL 60015, USAPublished online: 11 Apr 2011.
To cite this article: Mousumi Duttaray , Amitava K. Dutt & Kajal Mukhopadhyay (2008) Foreign direct investment andeconomic growth in less developed countries: an empirical study of causality and mechanisms, Applied Economics, 40:15,1927-1939, DOI: 10.1080/00036840600949231
To link to this article: http://dx.doi.org/10.1080/00036840600949231
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
Applied Economics, 2008, 40, 1927–1939
Foreign direct investment and
economic growth in less developed
countries: an empirical study of
causality and mechanisms
Mousumi Duttaraya,y, Amitava K. Duttb,y andKajal Mukhopadhyayc,y,*
aAmerican Express Company, NY 10285, USAbDepartment of Economics and Policy Studies, University of Notre Dame,
Notre Dame, IN 46556, USAcDiscover Financial Services, IL 60015, USA
We examine the causality between foreign direct investment (FDI) and
economic growth for 66 developing countries, taking into account their
interaction with exports and technological change. Time series analysis for
each country is conducted, based on a method introduced by Toda and
Yamamoto (1995) for testing Granger causality in the presence of
nonstationary time series. The main findings of this article are: FDI
causes growth in several of the developing countries, but the mechanism
through which this works differs across countries and reverse causality
from growth to FDI exists for many countries.
I. Introduction
In the last decade or so, less-developed country
(LDC) governments have been trying to attract
foreign direct investment (FDI) to their shores to
foster economic growth, a tendency which is receiving
a boost in the wake of the Asian crisis involving bank
lending and portfolio flows. Academic economists
too, have apparently arrived at a consensus that FDI
has a positive effect on growth. This is in sharp
contrast to the earlier scepticism among mainstream
development economists and dependency theorists
about the ability of FDI by transnational corpora-
tions (TNCs) to promote economic development.1
Empirical support for this favourable view is
provided by cross-country regression equations
which explain differences in growth rates across
countries in terms of FDI measures and other
determinants of growth rates such as the investment
rate and a measure of human capital formation. In
these equations, FDI measures such as the ratio of
FDI flow to output have a statistically significant
positive coefficient, suggesting that FDI has a
favourable impact on the rate of growth
*Corresponding author. E-mail: [email protected] views expressed by the authors are his/her own and do not necessarily reflect those of the affiliated institutions or theiraffiliates.1 The literature generally treats TNC activity and FDI synonymously and we shall do the same in this article. The two areactually the same not identical because TNCs affect development through a range of quasi-FDI and non-FDI activities(Helleiner, 1989, p. 1444).
Applied Economics ISSN 0003–6846 print/ISSN 1466–4283 online � 2008 Taylor & Francis 1927http://www.informaworld.com
DOI: 10.1080/00036840600949231
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
(Blomstrom et al., 1994; de Mello, 1997; Dutt, 1998).2
However regression equations, which incorporateadditional variables such as exports, often implynegative coefficients for the FDI term, or positivecoefficients only under certain circumstances(Balasubramanyam et al., 1996, 1999; Borenszteinet al., 1998; Stocker, 1999). Moreover, this analysissuffers from at least two shortcomings.
First, the positive relation shown by these equa-tions conceals the precise mechanisms by which FDIcan increase growth, knowledge of which is crucialfor designing appropriate policies to enhance thepossible growth-generating effects of FDI, and forassessing the long-run sustainability of the growthprocess. Caves (1982, p. 274) argues that the TNCs’effects on the LDC rate of ‘growth might seem toprovide the ultimate relationship to be investigated.Unfortunately, it may be a rather ineffective focus ofresearch. All the effects of foreign investment . . . canalter the LDC’s growth rate one way or another, andthere is a clear-cut case for pursuing the individualstrands of influence rather than trying to measuresome amalgam of diverse effects, each with its owntime structure of operation’. A few studies haveexamined the relationship between FDI and suchvariables by which it is expected to affect growth,such as technological change (de Mello, 1997, 1999)and exports and capital formation (de Mello, 1999;Stocker, 1999; Greenway and Kneller, 2004).
Second, the occasionally strong positive associa-tion between FDI and growth, suggested by theequations, implies nothing about causality runningfrom FDI to growth. The regression equationstypically take average rates of growth for countriesover a period and the average FDI-income ratio overthe same periods. Since it is possible that FDI isattracted into countries with a favourable growthperformance during the period, causality can runboth ways. Regression equations which have an FDImeasure which relates to a period prior to the periodover which growth rate is averaged, do not show astatistically significant coefficient for the FDI term(Dutt, 1998). Simple bivariate Granger causality testshave been used to study this question (de Mello, 1997;Stocker, 1999). However, in the presence of nonsta-tionary data, these tests provide questionable results.In more recent studies (e.g. de Mello, 1999;Nair-Reichert and Weinhold, 2001; Yabi, 2004),researchers have conducted panel studies on agroup of countries. While these methods generally
overcome the nonstationarity issues by pooling andaggregation, however, results of these studies do notreveal country-specific economic channels and effects.
This article attempts to overcome both theseshortcomings by using a recently developed methodto examine the causality between FDI, growth,exports and productivity for a sample of 66 develop-ing countries. To overcome the second problemof causality, this article uses the method ofToda and Phillips (1993) and Toda and Yamamoto(1995) which improves upon earlier Granger causalitytests to allow tests to be done on nonstationary data.To overcome the first problem, we go beyond thestandard bivariate framework to incorporate addi-tional variables, representing mechanisms by whichFDI and growth can be related. However, since thetime series on FDI for most developing countries isvery short, we have to concentrate on two mechan-isms that have been emphasized in the literature,namely, exports and productivity as a representationof technology.
Studies using time series methods for specificcountries or groups of countries have occasionallyattempted to deal with one, or in a few cases both, ofthe problems mentioned earlier. A number of studiesare available for China. Shan et al. (1999) examinescausality between FDI and growth, but does notexamine mechanisms through which the two affecteach other. Shan (2002) examines a vector autore-gression (VAR) model incorporating a number ofvariables, including exports, for China using quar-terly data, and takes into account the problem ofnonstationarity. It comes closest to our method, butdoes not include a technology variable. Liu et al.(2002) use cointegration framework to do a VARanalysis, but only takes into account trade variables.Yao (2006) uses panel data from Chinese provinces,takes into account the nonstationarity of the data,but does not consider the interaction between FDIand exports and does not examine causality. ForIndia, Chakraborty and Basu (2002) examinescausality issues using a structural cointegrationapproach with a vector error correction mechanismand explores the effects of FDI on unit labour costs,but does not take into account exports. Cuadros et al.(2004) uses a VAR model to examine causality issuestaking into account nonstationarity issues withquarterly data for Argentina, Mexico and Brazil,and also examines the interaction between exports,FDI, foreign income and output levels, but does
2 Earlier work also used stock measures of FDI in addition to, or instead of flow measures, which revealed a negative effect ongrowth. This result was interpreted as showing adverse long-run consequences of FDI on growth. Aside from econometricproblems involved in including the ratio of FDI to GDP in growth rate equations, recent work suggests that stock measuresno longer have a negative impact (Dutt, 1998).
1928 M. Duttaray et al.
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
not examine growth or technology. For Nigeria,Akinlo (2004) deals with nonstationarity issues usingcointegration and error correction models (ECMs) toexamine the effects of FDI on growth, but does notexamine channels by which FDI affects growth anddoes not examine causality issues. Zhang (2001) usescointegration techniques to do causality tests for11 Latin American and East Asian countries, withoutexamining the role of any mechanisms. Aside fromthe specific problems of some of these studies, a casecan be made that the issue of causality whichincorporates different mechanisms should be exam-ined using a unified framework and in a comparableway for a broad range of LDCs. Such an examinationwould allow us to detect general patterns across thesecountries. Our study differs from earlier work inincluding a large set of less-developed countriesacross different continents and in using a co-integra-tion corrected approach which takes into accountthe dynamics of the mechanisms of export andtechnology without imposing any long-term (cointe-grating) relationships.
The rest of this article proceeds as follows.Section II briefly reviews different mechanisms bywhich FDI can affect, and be affected by, growth inorder to choose variables to be included in ourempirical analysis. Section III discusses the data andSection IV the econometric methodology used in ouranalysis. Section V presents our empirical results.Section VI concludes.
II. Conceptual Issues
This section briefly summarizes the reasons forchoosing exports and productivity as the mechanismsfor examining the interaction between FDI andgrowth. We first examine effects of FDI on growth,then the determinants of FDI and finally the mutualdependence of exports and productivity.
The theoretical and empirical literature on thegrowth effects of FDI by TNCs on host countries isenormous.3 According to this literature, the mainchannels through which FDI affects growth arecapital accumulation, components of the balance ofpayments, technological change and industrialstructure.
Regarding capital accumulation, it is claimed thatFDI provides a major source of saving and invest-ment for host countries in situations in whichdomestic saving capacities are low and domestic
investment efforts are weak, as is often the case inLDCs. FDI therefore increases the rate of growth ofcapital in savings – and investment-constrainedeconomies, and is claimed to be less footloose thanother forms of foreign capital inflows. Against this, itis argued that TNCs often raise their resources withinthe host countries without augmenting savings, andFDI inflows have been no more stable than otherkinds of capital inflows (Helleiner, 1984). It is alsopointed out that because of high levels of profitrepatriation (especially if one takes into accountpractices such as transfer pricing) new FDI is oftenless than capital outflows due to profit repatriation.Finally, it is pointed out that FDI often has adeleterious effect on domestic entrepreneurship, thusreducing domestic investment, and also reducesdomestic savings. Neither theory nor empiricalwork have come to a clear conclusion on this issue(Jenkins, 1987, pp. 95–111), but it appears that thecontribution of FDI to capital accumulation inLDCs, as measured by FDI as a ratio of grossdomestic capital formation, is generally quite small,with rare exceptions.
It is argued that FDI improves the balance ofpayments of host countries directly as a resultof capital inflows, and also by increasing exportsand reducing imports of goods which are producedby TNCs in the host country; the rate of growth offoreign-exchange constrained economies is therebyincreased. Against this view, it has been argued – forreasons just mentioned – than when one takes intoaccount profit remittances, FDI often causes a netoutflow of foreign capital. Moreover, restrictiveclauses prevent a rapid growth of exports of theproducts of TNCs, and the need to import materialsand capital goods increases, rather than reducesimports (Jenkins, 1987, pp. 111–14). Empiricalstudies often show negative effects (see Lall andStreeten, 1977; Hood and Young, 1979, for instance),but the results crucially depend on the nature ofcounterfactual assumptions made. There seems to besome evidence that in the early stages of developmentof some of the newly industrialized countries (NICs),TNCs played an important role in expandingmanufacturing exports (Helleiner, 1989, p. 1472). Inaddition to improving the balance of payments, thecontact with world markets that such exports impliesimposes market discipline on exporters, leading togreater efficiency and product innovation, withpossible spillovers to other sectors as well.
Regarding technology, it is argued that TNCs area major conduit for the inflow of foreign
3 For useful surveys, see Lall and Streeten (1977), Hood and Young (1979), Caves (1982), Casson and Pearce (1987),Jenkins (1987), Helleiner (1989) and Lall (1993).
FDI and economic growth in LDCs 1929
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
technology into LDCs. TNCs bring in new technol-ogy unavailable domestically, and spillovers to thehost country occur when domestic workers learn touse the new technology, when domestic firmsemulate them in competing with them by adoptingthe new technology or when domestic firms whichsupply them with inputs learn new technology. Ithas been pointed out that countries open to FDI,such as Malaysia, have benefited from technologytransfers much more than countries which havebeen more restrictive, such as India. Against this, itis argued that costs of technology transfer areexceedingly high, that the technology transferred isinappropriate both in terms of products producedand in terms of factor intensity and that suchtransfers adversely affect the development ofindigenous technological capabilities (Casson andPearce, 1987, pp. 96–107; Jenkins, 1987, Chapter 4)which in fact may reduce the ability of the hostcountry to properly assimilate the foreign technol-ogy (Helleiner, 1989, pp. 1469–70). The importanceof indigenous technological capability developmenthas been stressed in the recent literature ontechnological change, which argues that the earlierdistinction between the two processes of ‘innova-tion’ and ‘diffusion’ is a false one since successfuldiffusion actually involves continuing and incre-mental technical change to modify borrowedtechnology to suit local conditions and to attainhigher standards of production (Bell and Pavitt,1993).
On industrial structure, it is argued that the entryof large TNCs powerful enough to engage inrestrictive practices reduces the degree of competitionin industries in host countries, resulting in problemsrelating to efficiency, technological change andindustrial dynamism, thereby adversely affectinggrowth (Newfarmer, 1979; Jenkins, 1987). However,others argue that the influx of TNCs breaks up themonopoly of domestic firms, and there is alsocompetition amongst different TNCs (Reuber et al.1973; Vernon, 1977). The empirical evidence on theimpact of TNCs on industrial concentration andbarriers to entry is mixed even if one distinguishesbetween at-entry and post-entry effects (Jenkins,1987, pp. 40–4). Moreover, there is ambiguity overthe appropriate meaning of competition in thiscontext (Jenkins, 1987, pp. 44–60). Finally, it isunclear what implications industrial structure has forthe rate of growth of the economy.
FDI can affect growth by other channels as well,for instance by changing income distribution and
through political economy effects. It is argued that itworsens income distribution by creating a taste forproducts which are associated with a cumulativedeterioration in income distribution (Griffin, 1980,Chapter 7), by creating islands of highly paid workersand managers, by increasing capital intensity and byincreasing industrial concentration (as noted earlier).Even if these effects occur (there is considerableconceptual and empirical debate on these issues),their effect on growth depends on the relationshipbetween distribution and growth, on which no clearconsensus exists. They are argued to affect theautonomy of the state in host countries and its abilityto engage in growth-enhancing policies (Hood andYoung, 1979, Chapter 8; Jenkins, 1987, Chapter 8),such as targeting those sectors that are particularlygrowth promoting. This argument, however,assumes that in the absence of TNCs, the state canplay a positive developmental role and is notprevented from doing so by influential powergroups. Moreover, it is not clear that LDC statesare as powerless against the TNCs as is sometimesassumed (Chang, 1998).
If one were to single out a couple of majordeterminants suggested by this literature, exportsand technological change can claim to be theleading candidates. The importance of FDI incapital accumulation is quantitatively too smallfor most LDCs, and the direction of the effects ofFDI on growth through changes in industrialstructure, income distribution and political econ-omy factors are too complex to be unambiguouslyclear. We therefore choose exports and technologi-cal change as the major routes by which FDIaffects growth.
Although FDI can cause growth, causation can runthe other way as well, as is suggested by theexpanding literature on the determinants of FDI indeveloping countries.4 Higher growth in a countryexpands its domestic market more rapidly, and to theextent that FDI seeks markets in host countries, thisincreases FDI. Further, increasing growth can inducehigher FDI by signalling to TNCs that the country isfollowing good policies, and by making possibleimprovements in education and infrastructure whichmakes FDI more profitable in them. Moreover, FDIto a country can be expected to increase afterinvestors know that the country has the potential toexport more and have a technological base whichwarrants advantageous production relocation, so thatexports and productivity increases can also cause FDI(de Mello, 1997).
4 For reviews and recent analysis of the determinants of FDI see Lecraw (1991), UNCTC (1991), Pfefferman and Madarassy(1992), UNCTC (1992), Brewer (1993), and Singh and Jun (1995).
1930 M. Duttaray et al.
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
The mutual dependence of exports, productivityand growth has also attracted widespread attention.Productivity can increase growth in supply con-strained economies by increasing the productivity ofresources such as labour and in demand-constrainedeconomies by increasing investment. In turn, growthcan increase productivity by learning by doing and byallowing more resources to be devoted to educationand research and development. A large literature hasdeveloped to examine the effect of exports ongrowth along the lines discussed earlier (see, e.g.Feder, 1983), and it has also been suggested thatgrowth can cause higher exports (Jung and Marshall,1985). Higher productivity increases competitivenessand can therefore increase exports, while exportgrowth, as discussed earlier, can increase productivityas well.
III. Data
We started with 159 countries that fall into low andlower-middle per capita income economies based onthe World Bank classification. We selected a total of66 developing countries from Asia, Africa, LatinAmerica and the Caribbean for our analysis. Thecountries that are not included in the analysis haveeither incomplete or missing data for the variablesthat we are interested in, or the time span of the datais relatively small. The resulting data set has12 countries from Asia, 30 countries from Africa,21 countries from the Latin American and Caribbeancountries of which 10 are from South America and11 from North America and 3 other island countries.
For each country, we used time series for fourvariables. We measured growth by the growth rate ofreal GDP (G). Since annual figures of populationgrowth rates are not very reliable, we decided tomeasure growth of real total GDP rather than percapita GDP. We measured exports and FDI asexports as a percent of GDP (X), and net inflow offoreign direct invest as a percent of GDP (F). Analternative measure of these variables is rates ofgrowth of real exports and FDI.5 Due to thedifficulties of finding appropriate price deflatorswith which to deflate nominal exports and FDI,and because we believe that taking exports and FDIas a ratio of GDP provides a good measure of the roleof these variables in the economy, we preferred ourmeasure. Finally, we took the logarithm of product-ivity (P) where productivity is defined as the ratio of
real GDP to total labour to be our productivitymeasure. Rates of change in the productivity oflabour are too closely related to rates of GDPgrowth, which explains why we took productivitylevels. We took labour productivity rather than somemeasure of total factor productivity because therelevant data is not available for many countries inour sample. It may be argued that FDI is more likelyto affect productivity in FDI-receiving firms in themanufacturing sector of host countries, but datalimitations for many countries, and the need to takeinto account technological spillover effects to otherfirms and sectors, made us choose productivity in theeconomy as a whole.
All the data were taken from World DevelopmentIndicator 1998 (WDI, 1998). Due to a change in thebase period for later releases of WDI (e.g. in 2000 andlater) we have decided to use the 1998 release data.Table 1 shows the list of countries, the sample periodfor each country and the ratio of FDI to GDP for thethree time periods between 1970 and 1979, 1980 and1989 and 1990 and 1996.
IV. Methodology
The econometric literature on causality tests isgenerally based on the concepts initially developedby Granger (1969) and later popularized by Sims(1972). These causality tests (frequently referred asGranger causality tests) between two variables xt andyt are based on simple OLS regression of one variablext on the lagged values of xt and yt and testing thecoefficients of lagged yt’s to be all zeros andvice versa. In most cases, depending on the causalitydirections, researchers use a bivariate VAR model of(xt, yt) and construct Wald-type tests on coefficientof the appropriate lagged variables. In stationary ortrend stationary VAR systems, these Wald testsfollow chi-squared distribution asymptotically.However, if the VAR systems include variables thatare nonstationary, then the asymptotic chi-squaredcriterion may not be valid.
Toda and Phillips (1993) have developed a limittheory for Wald tests of Granger causality in levelsVAR and Johansen-type ECMs. They have shownthat the asymptotic chi-squared criterion is still validif there is sufficient cointegration between thevariables in the VAR system. They have alsoprovided simulation studies (Toda and Phillips,1994) to further demonstrate the tests in ECMs.
5 These measures are used by Jung and Marshall (1985) for exports and by Stocker (1999) for FDI. These studies, howeverperform simple Granger causality tests.
FDI and economic growth in LDCs 1931
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
Table 1. List of countries, sample periods and average net inflow of FDI as percentage of GDP
Average FDI share of GDP (percent)
Continent Country Sample period 1970–1979 1980–1989 1990–1996
Asia China 1982 to 1996 0.00 0.51 3.93India 1970 to 1996 0.05 0.05 0.30Indonesia 1970 to 1996 0.77 0.37 1.65Jordan 1976 to 1996 0.59 0.93 0.19Malaysia 1970 to 1996 3.02 3.24 6.63Oman 1974 to 1995 1.26 1.52 1.00Pakistan 1970 to 1996 0.13 0.33 0.80Philippines 1970 to 1996 0.30 0.57 1.61Saudi Arabia 1970 to 1995 �0.87 �0.11 0.30Sri Lanka 1970 to 1996 0.15 0.75 0.99Thailand 1970 to 1996 0.57 0.97 1.67Turkey 1970 to 1996 0.17 0.20 0.47
Africa Algeria 1970 to 1996 0.48 0.08 0.02Benin 1970 to 1996 0.57 0.06 0.30Botswana 1975 to 1996 1.98 4.67 �0.25Burundi 1980 to 1996 0.00 0.21 0.07Cameroon 1970 to 1996 0.66 1.23 �0.07Central African Republic 1970 to 1996 0.96 0.61 �0.10Chad 1970 to 1996 1.99 1.55 1.14Congo, Dem. Rep. 1970 to 1996 0.29 �0.14 0.00Congo, Rep. 1970 to 1996 5.62 1.39 0.18Cote d Ivoire 1970 to 1996 1.21 0.56 0.02Egypt, Arab Rep. 1975 to 1996 1.01 2.66 1.27Gabon 1970 to 1996 2.98 1.76 �0.84Gambia, The 1970 to 1995 1.55 0.84 2.13Ghana 1970 to 1996 0.85 0.19 1.61Kenya 1970 to 1996 0.67 0.41 0.22Madagascar 1970 to 1996 0.28 0.16 0.45Mali 1970 to 1996 0.16 0.18 0.38Mauritania 1970 to 1996 �1.06 1.12 0.65Mauritius 1970 to 1996 0.31 0.59 0.71Morocco 1970 to 1996 0.25 0.37 1.23Niger 1970 to 1996 0.89 0.52 0.02Nigeria 1970 to 1996 1.56 1.69 4.98Rwanda 1970 to 1996 0.61 1.00 0.19Senegal 1970 to 1996 0.88 0.22 0.66Sierra Leone 1970 to 1996 1.43 �1.34 0.43South Africa 1970 to 1996 0.50 0.04 0.04Swaziland 1970 to 1996 3.71 4.81 5.49Tunisia 1970 to 1996 1.44 1.82 2.09Zambia 1970 to 1996 �0.66 1.69 2.24Zimbabwe 1977 to 1996 0.00 �0.08 0.38
South America Argentina 1970 to 1996 0.11 0.65 1.34Bolivia 1970 to 1995 �0.17 0.58 2.53Brazil 1970 to 1996 1.13 0.67 0.56Chile 1970 to 1996 �0.35 1.23 2.61Colombia 1970 to 1996 0.36 1.30 2.10Ecuador 1970 to 1996 2.76 0.54 2.20Paraguay 1970 to 1996 1.09 0.28 1.88Peru 1970 to 1996 0.37 0.12 2.50Uruguay 1977 to 1996 0.73 0.52 0.50Venezuela 1970 to 1996 �0.37 0.16 1.65
North America Barbados 1970 to 1995 3.34 0.47 0.65Costa Rica 1970 to 1996 2.26 1.78 3.59Dominican Republic 1970 to 1996 2.07 1.12 2.55El Salvador 1970 to 1996 0.54 0.32 0.23
(continued)
1932 M. Duttaray et al.
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
Recently, Toda and Yamamoto (1995) have
proposed an alternative approach to the testingprocedure suggested by Toda and Phillips (1993).In a more recent applied study on causality of
export on productivity for six industrial countries,Yamada (1998) has applied Toda and Yamamoto’s
approach.We use Toda and Yamamoto’s procedure to test
the pair-wise Granger causality between all four
variables (F, P, X, G) in our system. We define a four-variable VAR(k) model as follows:
yt ¼ cþ�1yt�1 þ � � � þ�t�kyt�k þ ut,
ut � i:i:d: Nð0,�uÞ t ¼ 1, . . . ,T ð1Þ
where yt ¼ ðF,P,X,GÞ0. In this model, the null
hypothesis of noncausality from the ith component
of yt to the jth component can be expressed interms of the (j, i) element of all the coefficient
matrices (�l, l ¼ 1, . . . , k) being equal to zero. Forexample, the null hypothesis of noncausality from
FDI (F) to growth rate of GDP (G) would mean allthe (1, 4) elements of �l in (1) are zeros. If dmax is
the maximum order of integration for all thevariables in yt, then according to Toda andYamamoto’s testing procedure, we estimate (1)
using OLS method for (kþ dmax) lags and useonly the first k coefficient matrices to construct the
Wald test statistics.We use some more notations to define the Wald
test statistics. Let Sj be a unit column vector of
dimension four with unity at the jth place; e.g. we candefine, S3 ¼ ð0, 0, 1, 0Þ
0. Also let Sij ¼ S0j � ðIk � SiÞ0,
where Ik is the identity matrix of order k. Now wedefine the coefficient vector of the four-variableVAR(k) system as p ¼ vecð�Þ where vecð�Þ is a
column vector obtained by stacking the rows of� ¼ ð�1, . . . ,�kÞ. Now let us denote the Wald test
statistics for the null hypotheses of noncausality from
yit to yjt as Wij. Then we have,
Wij ¼ TfSijpg0½Sij�S0ij�Sijp ð2Þ
where � is the ordinary least square estimate of �and � is a consistent estimate of the asymptotic
covariance matrix of theffiffiffiffi
Tpðp� pÞ. Toda and
Yamamoto (1995) have shown that the Wij has
asymptotic chi-squared distribution with k degrees
of freedom when the maximum order of integration
is dmax.For this modified Granger causality test, two
things require specification: (1) the maximum order
of integration of the variables in the system and (2)
the lag length for the VAR specification. The
standard procedure for selecting the order of
integration involves a battery of specification and
unit root tests. The first of these is a test of inclusion
of a constant term, or a trend term, or none of these,
in the simple autoregressive (AR) regression model.
The second is the choice of a lag length for the AR
regression. In this article, we use the Schwarz
Information Criterion (SIC) to select the appropriate
lag length. These two steps are often performed
simultaneously to decide the appropriate specifica-
tion for the unit root tests. Finally, one tests for the
presence of the unit root based on the AR regression.
We use the Augmented Dickey Fuller (ADF) tests to
find the unit root in a series, and determine the order
of integration by testing the unit root again for the
first-difference series. The lag lengths for the VAR
system are similarly chosen by the Hannan–Quinn
Criterion (HQC).This method allows us to examine the direction of
causality between FDI and growth as the mechanism
through which this causality comes about. Consider
the hypothesis that FDI causes GDP growth, which
Table 1. Continued
Average FDI share of GDP (percent)
Continent Country Sample period 1970–1979 1980–1989 1990–1996
Guatemala 1970 to 1996 1.46 1.35 0.75Haiti 1970 to 1996 0.88 0.42 0.06Honduras 1970 to 1996 0.66 0.70 1.35Jamaica 1970 to 1996 3.08 0.22 3.46Mexico 1970 to 1996 0.75 1.03 1.86Panama 1980 to 1996 1.29 0.13 2.45Trinidad and Tobago 1970 to 1996 5.79 1.75 5.56
Other Fiji 1970 to 1996 1.75 2.09 3.85Malta 1970 to 1995 3.10 2.23 3.31Papua New Guinea 1976 to 1996 0.65 3.99 4.32
FDI and economic growth in LDCs 1933
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
is the main focus of interest in this article. Let thenotation ‘x! y’ represent a statistically significant(bivariate) Granger causality between two variables xand y, where x Granger-causes y at 10% or lowerlevel of significance. In our four-variable VARanalysis, there are several different ways in whichF can cause G. First, FDI Granger-causes exportswhich in turn Granger-causes FDI, so that we haveF!X!G; in this case we refer to exports as themechanism through which FDI causes growth.Second, FDI can cause growth through the mechan-ism of productivity growth, so that F!P!G.Third, FDI can cause growth through both themechanisms of export and productivity, so thatF!P!X!G and/or F!X!P!G. Finally,FDI can cause growth through some unspecifiedmechanism (i.e. other than exports and productivity),in which case F!G. In a similar way, we canexamine the causality from growth to FDI and themechanism through which this causation comesabout.
V. Empirical Results
We pre-tested all the four variables for each countryfor the presence of unit roots and its order ofintegration. This required us to perform a total of264 unit root and specification tests. We do notpresent the results in detail but a summary of theresults is as follows. The optimal lag length for the Fseries is found to be three for most of the countries.For a few other countries, the optimal lag lengthvaries from 1 to 6. The P, G and X series have anoptimal lag length of four for most of the countries.Again, there are a few countries for which the lagsvary from two to six. For the AR regressionspecification, a little more than half of the seriesindicate no intercept or time trend. The rest of theseries shows only an intercept in the regression. Noneof the specifications include both an intercept and a
trend term. After performing the ADF unit root testson all the series for all the countries, we find that 25countries have maximum order of integration,dmax¼ 1. Thirty-five countries have dmax¼ 2 andonly 4 out of 66 countries have dmax¼ 0.
Visual inspection of some of the series showsevidence of structural breaks between the years 1970and 1996, but it did not warrant us to perform anyformal tests of unit roots under structural breaks likePerron (1990) or Andrews (1993). Overall, structuralbreaks might have occurred in the series P for Algeriaaround 1989, X for Honduras between 1980 and1984, F for Jamaica around 1980 and X for SouthAfrica around 1980. Further, the series F and X forZimbabwe show frequent regime changes and displaywide variations in G and P.
As mentioned earlier, the lag lengths for the fourvariables VAR systems are determined based on theHQC. The maximum lag length for HQC was set tobe four (given that we have at most 26 years of annualdata for each country). The appropriate lag lengthwas found to be one, for all the countries in ourstudy. Since the data are yearly, a four-variable VARsystem with one lag for each variable provides a goodapproximation of the entire dynamic system for eachcountry in our sample.
We calculated the Wald tests statistic shown in (2),for all possible pairs of variables for each country inour study. In other words, we constructed 12 Grangercausality tests for each of the 66 countries to find theappropriate significant causality and its direction. Wespecify the level of significance to be at most,�¼ 10%, for each Granger causality tests. Thesignificance level for the composite tests of Grangercausality of FDI to GDP growth rate throughproductivity and export channels are found to be �,same as in the individual pair-wise Granger causalitytests.6
The results of our causality exercises, provided inTables 2 through 5, can be summarized as follows:First, of the 66 countries in our sample, we find thatFDI affects growth through some mechanism – i.e. by
6 The argument is as follows: Let the G1 and G2 be two Granger causality tests such that the null hypotheses are, ‘X does notGranger-cause Y’and ‘Y does not Granger-cause Z’, respectively. Also let the level of significance for the tests be:
�1 ¼ PrðX causes YjX does not cause YÞ ¼ PrðA1Þ and
�2 ¼ PrðY causes ZjY does not cause ZÞ ¼ PrðA2Þ
respectively, where A1 and A2 are the critical regions for the tests. Since we are interested in the hypothesis the X Granger-causes Z through Y, we define the composite hypothesis to be G: G1 or G2. Rejecting the null hypothesis in G would imply XGranger-causes Z through Y. For the test G, the composite level of significance is PrðA1 [ A2Þ which has an upper boundPrðA1Þ þ PrðA2Þ ¼ �1 þ �2. In the present case, each of our individual Granger causality tests follows an asymptotic chi-squared distribution. Consequently, PrðA1Þ ¼ Prð� > ��1 ðkÞÞ, and PrðA2Þ ¼ Prð� > ��2 ðkÞÞ, where k is the order of lags in theVAR specification. Since k is same for all individual tests for Granger causalities, this implies that for any �1 � �2, we haveA1 � A2 and A1 [ A2 � A1. Therefore, the level of significance for the composite hypothesis G is � ¼ maxf�1, �2g. In our case,�1 ¼ �2 ¼ � ¼ 10%.
1934 M. Duttaray et al.
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
‘directly’ causing growth, or ‘indirectly’ by causingexports or productivity change which in turn causesgrowth – in only 29 (i.e. in 44% of the countries) ofthe countries. In the remaining 37 countries, FDIdoes not appear to affect growth at all. Second, theshare of countries in which FDI affects growth variesacross continents. It is highest in South America,where it is 70%, and lowest in Africa where it isabout 33%, with North America at about 55% andAsia at 42%.
Third, we are able to identify the precise mechan-isms through which FDI affects growth in a numberof countries. FDI affects growth through theproductivity channel in only seven countries, whichis slightly more than 10% of all countries and justunder a quarter of countries in which FDI affectsgrowth through some mechanism. FDI affects growththrough exports in nine countries, about 14% and31% of all countries and those in which FDI affectsgrowth. In five countries, the effect works throughproductivity and exports, four of which arealready included in the last two categories. Thus weare identify precise mechanisms for only 17 countries,or about 26% of all 66 countries and 59% ofcountries in which FDI affects growth. In theremaining 12 countries in which FDI affectsgrowth, the mechanism must be something otherthan exports or productivity; possible candidatesinclude capital accumulation and technologicalchanges other than those captured by increases inlabour productivity.
Fourth, there are interesting differences acrosscontinents in terms of the mechanism by which FDIaffects growth. In Asia, the only two countries(Jordan and Malaysia) for which we have been ableto identify the mechanism by which FDI affectsgrowth, the mechanism runs through exports. In
Africa, in none of the countries does FDI affectgrowth through the mechanism of productivity. Inthree counties (Cote d’Ivoire, Mauritania and SouthAfrica), it does so through exports alone and inZambia, FDI affects growth through its effects onproductivity and exports. For South America how-ever, five (Argentina, Bolivia, Chile, Ecuador andVenezuela) of the six countries in which wehave identified the mechanism by FDI affectsgrowth, the mechanism is a change in productivity,Colombia being the only country in which themechanism is exports. North America has amore mixed pattern, with productivity being themechanism in two countries (Guatemala andTrinidad & Tobago) and exports in two others(Costa Rica and Panama).
Fifth, we find that reverse causality from growthand growth-related variables to FDI exists for anumber of countries. Growth affects FDI in 20countries (this includes ‘direct’ effects and ‘indirect’effects through exports or productivity), i.e. for over30% of the sample. Other growth related variables –i.e. exports and productivity – affect FDI ‘directly’ in17 and 20 countries. As Tables 4 and 5 show,the number of countries where such reverse effectsare found is 30, which is one more than the number ofcountries for which we found that FDI affectsgrowth.
VI. Conclusions
The main conclusions of our analysis are as follows:(1) FDI affects growth in 29 countries, just 44% ofthe countries in our sample of 66 LDCs. For themajority of the LDCs in our sample, growth does not
Table 2. Summary of effects of FDI on growth and mechanisms of causation
No. of countries (% of countries)
All Asia Africa North America South America
Causality 66 (100%) 12 (100%) 30 (100%) 11 (100%) 10 (100%)F!P!G 7 (10.6%) 0 (0%) 0 (0%) 2 (18.2%) 5 (50%)F!X!G 9 (13.6%) 2 (16.7%) 2 (6.7%) 2 (18.2%) 1 (10%)F!P!X!G 2 (3.0%) 0 (0%) 1 (3.3%) 1 (9.1%) 0 (0%)F!X!P!G 3 (4.5%) 0 (0%) 0 (0%) 2 (18.2%) 0 (0%)F!G 21 (31.8%) 4 (33.3%) 7 (23.3%) 4 (36.4%) 6 (60%)
FDI affects growth 29 (43.9%) 5 (41.7%) 10 (33.3%) 6 (54.5%) 7 (70%)FDI does not affect growth 37 (56.0%) 7 (58.3%) 20 (66.7%) 5 (45.4%) 3 (30%)
Notes: The notation [F! ch!G] implies FDI Granger-causes economic growth via either export channel {ch ¼ X} orproductivity channel {ch ¼ P}. The notation F!G implies FDI Granger-causes growth directly. The other notations implyFDI Granger-causes growth via both the channels in specific order and direction.
FDI and economic growth in LDCs 1935
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
affect FDI at all. (2) In a number of countries, we are
able to identify precise mechanisms by which FDI
affects growth. In 59% of the countries in which
FDI affects growth, we were able to identify whether
the effects are through export or productivity
mechanisms or both. (3) There are interesting cross-
continental variations in the degree to which FDI
affects growth and in the mechanisms through which
Table 3. Effects FDI on growth and mechanisms of causation in LDCs
Asia Africa North America South America Other
No. of countries 12 30 11 10 3
F!P!G Guatemala ArgentinaTrinidad & Tobago Bolivia
ChileEcuadorVenezuela
F!X!G Jordan Cote d’Ivoire Costa Rica Colombia MaltaMalaysia Mauritania Panama
South Africa
F!P!X!G Zambia Guatemala
F!X!P!G Costa Rica MaltaPanama
F!G Indonesia Algeria Guatemala ArgentinaMalaysia Gabon Haiti BoliviaPakistan Kenya Jamaica ChilePhilippines Nigeria Trinidad & Tobago Ecuador
Senegal UruguayTunisia VenezuelaZambia
FDI affects growth Indonesia Algeria Costa Rica Argentina MaltaJordan Cote d’Ivoire Guatemala BoliviaMalaysia Gabon Haiti ChilePakistan Kenya Jamaica ColombiaPhilippines Mauritania Panama Ecuador
Nigeria Trinidad & Tobago UruguaySenegal VenezuelaSouth AfricaTunisiaZambia
FDI does not China Benin Barbados Brazil Fijiaffect growth India Botswana Dominican Rep Paraguay Pap New Guinea
Oman Burundi El Salvador PeruSaudi Arabia Cameroon HondurasSri Lanka Cent Afr Rep MexicoThailand ChadTurkey Congo, Dem R
Congo, RepEgyptGambiaKenyaMadagascarMaliMauritiusMoroccoNigerRwandaSierra LeoneSwazilandZimbabwe
Source: Authors’ calculations.
1936 M. Duttaray et al.
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
it does so. FDI affects growth in a larger proportionof South American countries than in Asia and Africa,and North American countries lie in the middle.Export is a more important mechanism in Asia andAfrica, while in South America productivity growth ismore important. (4) Reverse causality to FDI fromexports, productivity or growth is present in 30 of our66 countries. This suggests that a close correlationbetween growth and FDI should not automatically betaken to imply that FDI causes growth, exports orproductivity change.
We end by commenting on some further worksuggested by our analysis: (1) The presence of ‘direct’effects of FDI on growth, i.e. those not captured by
exports and productivity, suggest that our analysiscould be enriched by incorporating more than two‘mechanism’ variables into it. However, the smallnumber of time series observations makes it difficultto do so at present. (2) Our analysis presents resultsonly about causes, not about directions of effects.This implies that our finding that in 44% of oursample of countries, FDI affects growth overstatesthe case for FDI in promoting growth. Othereconometric procedures need to be used to analysethe signs of the effects. (3) Our analysis opens upthe way for a more systematic examination of thecharacteristics of countries in which growth is morelikely to affect FDI. Do countries need to have a
Table 5. Growth, export and productivity effects on FDI in LDCs
Asia Africa North America South America Other
No. of countries 12 30 11 10 3
Growth affects FDI India Congo, Dem R Dominican Rep Bolivia FijiMalaysia Congo, Rep Haiti ColombiaPakistan Gabon Honduras Ecuadora
Niger Mexico ParaguaySenegal Panama Peru
Venezuelaa
Export affects FDI India Cameroon El Salvador EcuadorPhilippines Cent Afr Rep Haiti PeruSaudi Arabia Chad Honduras Venezuela
Congo, Rep MexicoNiger PanamaNigeriaSierra LeoneZimbabwe
Productivity affects FDI India Burundi Dominican Rep Bolivia FijiMalaysia Congo, Dem R Haiti ColombiaPakistan Congo, Rep Honduras Paraguay
Gabon Mexico PeruNiger PanamaSenegalZimbabwe
Note: a In these countries growth affects FDI through its effect on exports and productivity.
Table 4. Summary of effects on FDI
No. of countries (% of countries)
All Asia Africa North America South America
Reverse causality 66 (100%) 12 (100%) 30 (100%) 11 (100%) 10 (100%)Growth affects FDI 20 (30.3%) 3 (25%) 5 (16.7%) 5 (45.5%) 6 (60%)Export affects FDI 17 (25.8%) 3 (25%) 8 (26.7%) 3 (27.3%) 3 (30%)Productivity affects FDI 20 (30.3%) 3 (25%) 7 (23.3%) 5 (45.4%) 4 (40%)Some effect on FDI 30 (45.4%) 5 (41.7%) 12 (40%) 6 (54.5%) 6 (60%)No effects on FDI 36 (55.5%) 7 (58.3%) 18 (60%) 5 (45.5%) 4 (40%)
FDI and economic growth in LDCs 1937
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
high-enough stock of human stock and high-enoughper capita income levels for FDI to have an effect ongrowth?7 (4) Our analysis also can lead to theexploration of the consequences of different mechan-isms by which FDI affects growth. Is FDI morelikely to have sustained effects on growth if it worksthrough increases in productivity or throughincreases in exports? It is possible to argue, on onehand, that the productivity channel is the moreimportant one if it is believed that productivityincreases are required for sustained growth. On theother hand, it can be argued that productivityincreases can lead to unemployment and growthcannot increase unless export increases to avoidforeign exchange constraints on growth. (5) Therole of TNCs cannot be adequately captured by FDIdata based on financial flows. This is partly becausethe size of capital flows cannot reflect the technolo-gical contributions of TNCs, and also because thedistinction between FDI and portfolio equity invest-ment is based on an arbitrary 10% equity stake rule.Quantitative studies like ours therefore need to besupplemented by case studies addressing qualitativeissues more carefully.
References
Akinlo, A. E. (2004) Foreign direct investment and growthin Nigeria – an empirical investigation, Journal ofPolicy Modeling, 13, 627–39.
Andrews, D. W. K. (1993) Tests for parameter instabilityand structural change with unknown change point,Econometrica, 61, 821–56.
Balasubramanyam, V. N., Salisu, M. and Sapsford, D.(1996) Foreign direct investment and growth in EP andIS countries, Economic Journal, 106, 92–105.
Balasubramanyam, V. N., Salisu, M. and Sapsford, D.(1999) Foreign direct investment as an engine ofgrowth, Journal of International Trade and EconomicDevelopment, 8, 27–40.
Bell, M. and Pavitt, K. (1993) Accumulatingtechnological capability in developing countries,Proceedings of the World Bank Annual Conference onDevelopment Economics 1992, The World Bank,Washington, DC.
Blomstrom, M., Lipsey, R. E. and Zejan, M. (1994) Whatexplains the growth of developing countries,in Convergence of Growth. Cross-National Studies andHistorical Evidence (Eds) W. J. Baumol, R. R. Nelson,and E. N. Wolff, Oxford University Press, Oxford andNew York, pp. 243–59.
Borensztein, E. J., de-Gregorio, J. and Lee, J. W. (1998)How does foreign direct investment affect economicgrowth?, Journal of International Economics, 45,115–35.
Brewer, T. L. (1993) Foreign direct investment in emergingmarket countries, in The Global Race for ForeignDirect Investment (Ed.) L. Oxelheim, Springer-Verlag,Berlin, pp. 177–203.
Casson, M. and Pearce, R. D. (1987) Multinationalenterprises in LDCs, in Surveys in DevelopmentEconomics (Ed.) N. Gemmell, Blackwell, Oxford,pp. 90–132.
Caves, R. E. (1982) Multinational Enterprise and EconomicAnalysis, Cambridge University Press, Cambridge.
Chakraborty, C. and Basu, P. (2002) Foreign directinvestment and growth in India: a cointegrationapproach, Applied Economics, 34, 1061–73.
Chang, H.-J. (1998) Globalization, transnational corpora-tions, and economic development: can the developingcountries pursue strategic industrial policy in aglobalizing world economy? in Globalization andProgressive Economic Policy (Eds) D. Baker,G. Epstein and R. Pollin, Cambridge UniversityPress, Cambridge, pp. 97–113.
Cuadros, A., Orts, V. and Alguacil, M. (2004) Opennessand growth: re-examining foreign direct investment,trade and output linkages in Latin America, Journal ofDevelopment Studies, 26, 167–92.
de Mello Jr., L. R. (1997) Foreign direct investment indeveloping countries and growth: a selective survey,Journal of Development Studies, 34, 1–34.
de Mello Jr, L. R. (1999) Foreign direct investment-ledgrowth: evidence from time-series and panel data,Oxford Economic Papers, 51, 133–51.
Dutt, A. K. (1998) Transnational corporations, directforeign investment, and growth, in TransnationalCorporations and the Global Economy (Eds)R. Kozul-Wright and R. E. Rowthorn, McMillanand St. Martin’s Press, London and New York,pp. 164–91.
Feder, G. (1983) On exports and economic growth, Journalof Development Economics, 16, 59–74.
Granger, C. W. J. (1969) Investigating causal relations byeconometric models and cross-spectral methods,Econometrica, 37, 424–38.
Greenway, D. and Kneller, R. (2004) Exporting andproductivity in the United Kingdom, Oxford Reviewof Economic Policy, 14, 358–71.
Griffin, K. (1978) International Inequality and NationalPoverty, Macmillan, London.
Helleiner, G. K. (1989) Transnational corporations anddirect foreign investment, in Handbook ofDevelopment Economics (Eds) H. Chenery andT. N. Srinivasan, Vol. 2, North Holland,Amsterdam, pp. 1441–80.
Hood, N. and Young, S. (1979) The Economics ofMultinational Corporations, Longman, London.
Jenkins, R. (1987) Transnational Corporations and UnevenDevelopment, Methuen, London.
Jung, W. S. and Marshall, P. J. (1985) Export growth andcausality in developing countries, Journal ofDevelopment Economies, 18, 1–12.
Lall, S. (1993) Introduction: transnational corporationsand economic development, in TransnationalCorporations and Economic Development (Ed.)S. Lall, Routledge, London.
7 This issue has been addressed using standard cross-country regressions by Balasubramanyam et al. (1997) andBorensztein et al. (1998).
1938 M. Duttaray et al.
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013
Lall, S. and Streeten, P. (1977) Foreign Investment, TNCsand Developing Countries, Macmillan, London.
Lecraw, D. J. (1991) Factors influencing foreign directinvestment by transnational corporations in hostdeveloping countries: a preliminary report,in Multinational Enterprises in Less DevelopedCountries (Eds) P. J. Buckley and J. Clegg,St. Martin’s Press, New York, pp. 163–80.
Liu, Xiaohui, Burridge, P. and Sinclair, P. J. N. (2002)Relationships between economic growth, foreign directinvestment and trade: evidence from China, AppliedEconomics, 34, 1433–40.
Nair-Reichert, U. and Weinhold, D. (2001) Causality testsfor cross-country panels: a new look at FDI andeconomic growth in developing countries, OxfordBulletin of Economics and Statistics, 19, 153–71.
Newfarmer, R. (1979) Transnational Conglomerates andThe Economics of Dependent Development, JAI Press,Greenwich, CT.
Perron, P. (1990) Testing for a unit root in a time serieswith a changing mean, Journal of Business Economicsand Statistics, 8, 153–62.
Pfefferman, G. P. and Madarassy, A. (1992) Trends inprivate investment in developing countries, DiscussionPaper No. 14, The World Bank, Washington, DC.
Reuber, G. L., et al. (1973) Private Foreign Investment inDevelopment, Clarendon Press, Oxford.
Shan, J. (2002) A VAR approach to the economics of FDIin China, Applied Economics, 34, 885–93.
Shan, J., Tian, G. and Sun, F. (1999) Causality betweenFDI and growth, in Foreign Direct Investment inChina (Ed.) W. E. Yanrui, Edward Elgar,Cheltenham, pp. 140–56.
Sims, C. A. (1972) Money, income, and causality, AmericanEconomic Review, 62, 540–52.
Singh, H. and Jun, K. W. (1995) Some new evidence on thedeterminants of foreign direct investment in develop-ing countries, Policy Research Working PaperNo. 1531, World Bank.
Stocker, H. (1999) Growth effects of foreign directinvestment: myth or reality?, Department ofEconomics, University of Innsbruck, unpublished.
Toda, H. Y. and Phillips, P. C. B. (1993) Vectorautoregression and causality, Econometrica, 61,1367–93.
Toda, H. Y. and Phillips, P. C. B. (1994) Vectorautoregression and causality: a theoretical overviewand simulation study, Econometric Review, 13, 259–85.
Toda, H. Y. and Yamamoto, T. (1995) Statisticalinference in vector autoregression with possiblyintegrated processes, Journal of Econometrics, 66,225–50.
United Nations Centre on Transnational Corporations(UNCTC) (1991) Government Policies and ForeignDirect Investment, United Nations, New York.
United Nations Centre on Transnational Corporations(UNCTC) (1992) Determinants of Foreign DirectInvestment: A Survey of the Evidence, UnitedNations, New York.
Vernon, R. (1977) Storm over Multinationals: The RealIssues, Macmillan, London.
World Bank (1998) The World Development Indicators,The World Bank, Washington, DC.
Yabi, G. O. (2004) Does direct foreign investmentreally drive growth in developing countries? –Mitigated results of an empirical analyses,Canadian Journal of Development Studies, 17,275–91.
Yamada, H. (1998) A note on the causality between exportand productivity: an empirical re-examination,Economics Letters, 61, 111–14.
Yao, S. (2006) On economic growth, exports and FDI inChina, Applied Economics, 38, 339–51.
Zhang, K. H. (2001) Does foreign direct investmentpromote economic growth? Evidence from East Asiaand Latin America, Contemporary Economic Policy,19, 175–85.
FDI and economic growth in LDCs 1939
Dow
nloa
ded
by [
Uni
vers
ity o
f N
orth
Car
olin
a C
harl
otte
] at
05:
04 1
9 Se
ptem
ber
2013