Abstract
It is open to question whether the intensified worldwide competition for FDI has reduced its traditionally strong concentration in a few large and relatively advanced host countries. We calculate and decompose Theil indices to track changes in absolute and relative concentration of FDI during the period 1970–2013. We find that both absolute and relative concentration decreased when excluding offshore financial centers from the overall sample. In addition to the narrowing gap between OECD and non-OECD countries, the concentration across non-OECD countries declined for both the absolute and relative measures. This is also true for major subgroups of non-OECD countries. Finally, recent developments indicate that low-income countries are no longer at the losing end of the competition for FDI.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
For details, see UNCTAD’s Investment Policy Monitor, January 2015: http://investmentpolicyhub.unctad.org/Upload/Documents/IPM%20No%2013.pdf (accessed: November 2015). See also Nunnenkamp and Thiele (2013: Fig. 6).
As discussed in more detail below, we will use two different definitions of weights throughout our analysis.
Other frequently used inequality measures, such as the Gini index or the coefficient of variation (CV) do not have this property.
In the case of the Theil index, the sum of these weights is always equal to one so that the within-group component is actually a weighted average of the group-specific inequality measures.
The absolute Theil index is thus given by \(\hbox {T}^\mathbf{I} =\sum _{i=1}^I {x_{i} } \ln \left( {Ix_{i} } \right) \). It is equal to zero (no concentration or perfect equality) if all countries receive the same amount of FDI inflows. It takes its maximal value \(\hbox {T}_{\mathrm{max}}^\mathbf{I} =\ln (I)\) if all FDI goes to just one country.
The (population-weighted) relative Theil index is zero (no concentration or perfect equality) if each country’s share in total FDI inflows is equal to its share in total population. It takes its maximal value \(\hbox {T}_{\mathrm{max}}^\mathbf{I} =\ln (\sum _{i} {\hbox {POP}_{i} /\hbox {POP}_{a} } )\) if all FDI goes to the country (denoted by a) with the smallest population.
Conversely, FDI per capita tends to be relatively low for (very) large countries where international transactions generally play a less important role compared with small countries.
The FDI data are available at: http://unctadstat.unctad.org/wds/ReportFolders/reportFolders.aspx (accessed: April 2015). Population data used to calculate relative weights are mainly also from UNCTAD. They have been augmented by data from the World Bank’s World Development Indicators available at http://data.worldbank.org/ data-catalog/world-development-indicators (accessed: April 2015) and from the IMF’s International Financial Statistics Database (February 2015 Edition).
Note that FDI inflows can be negative, for example, if profit remittances and repayments of loans received from the parent company exceed new equity inflows. Negative values observed after taking four-year aggregates have been set to zero for the calculation of Theil indices. Alternative treatments of negative values, such as the consolidation with inflows from the nearest four-year interval with sufficiently positive inflows, have little effects on the results presented below. FDI inflows smaller than 0.0001 million dollars were treated as zero inflows.
During the observation period several countries have split up or unified as in the case of Germany. To get a balanced country panel, the successor states of the Union of Socialist Soviet Republics and of the Socialist Federal Republic of Yugoslavia as well as of Czechoslovakia were treated as if they existed throughout the whole observation period. Eritrea and Ethiopia are treated as one entity (observation) throughout the observation period. The same is true for Belgium and Luxembourg (for which separate FDI data would be available only from 2002 onwards).
Throughout the subsequent analysis, we define OECD countries according to OECD membership by the end of 1993.
The list of OFCs is mainly taken from the International Monetary Fund (for details, see Zoromé 2007). However, we consider Luxembourg and Switzerland as OECD countries.
For details on the development of FDI inflows for the different subgroups see Table 3 in Appendix.
In deriving Eqs. (6) and (7) we make use of the fact that for subset \(\mathbf{P}\) the weight \({\omega }_{{}_\mathbf{P}}\) from Eq. (3) is equal to 1. As \(\ln (x)\) is not defined for \(x = 0\), we substitute \(x\ln (x)\) by \(\hbox {lim}_{x\rightarrow 0}\, x\ln (x) = 0\). For a similar decomposition in the context of trade diversification see Cadot et al. (2013).
A declining number (or weight) of countries with zero FDI inflows is thus referred to as a lower extensive margin of concentration. This should not be confused with a lower extensive margin of FDI, which generally refers to a rising number of countries with zero FDI inflows in the literature.
The pronounced decline of the extensive margin in 1978–1981 for the relative measure in the right panel of Fig. 2 was associated with China’s opening-up to FDI. Note that the Chinese case illustrates one of the main differences between the absolute and the relative measures of concentration. For the absolute measure, China counts as just one out of 196 countries. For the relative measure, China represents a heavy weight with almost 20% of total population. In the right panel of Fig. 2 China’s opening-up to FDI thus led to a strong decrease in the extensive margin (20% of world population now receive FDI), but also to a notable increase in the intensive margin (20% of world population received still quite low per capita inflows).
The difference between the relative Theil indices in panel b of Fig. 3 can be attributed to two OFC-related developments in the last sub-periods: First, when decomposing the overall index between OFCs and all other countries in our sample, the between-group component increased considerably (not shown). This is due to the above-noted widening gap in terms of per capita FDI inflows in favor of OFCs. Second, concentration strongly increased within the subgroup of OFCs, which together with the OFC’s increasing share of total FDI inflows implies that the OFCs’ contribution to the within component of overall concentration increased as well. These two factors are no longer pushing overall relative concentration upward once OFCs are excluded.
We prefer excluding OFCs from our further analysis since their—limited—effects on overall concentration do not offer relevant insights on whether FDI has become more global in the sense of increasingly involving host countries across the developing world. Mostly, OFCs serve only as stop-over destinations rather than final destinations of FDI and it is generally unknown where FDI flows channeled through OFCs are ultimately used for investment.
The large and heterogeneous group of non-OECD countries will be further decomposed in the next subsections.
At the beginning of our observation period, the 23 OECD members accounted for more than 75% of FDI inflows to all 159 remaining sample countries. In the last sub-period 2010–2013, the 136 non-OECD countries have increased their share to almost 50% (Table 3).
While the relative concentration of inflows across the non-OECD countries, \(\hbox {T}^{\mathbf{N}}\), continued to strongly decrease throughout most of the 1990s, the effect of that decrease on the within-group component of the overall relative Theil index, \(\hbox {T}^{\mathbf{I}\backslash \mathbf{F}}\), was overcompensated after the end of the 1990s by the increasing share of inflows to non-OECD countries, \({ \omega }_{\mathbf{N}}\). As the concentration of FDI inflows across the non-OECD countries was higher than that across the OECD countries, any increase in \({\omega }_{\mathbf{N}}\), ceteris paribus, increased the within-group component of overall concentration.
Both per country inflows and per capita inflows to the OECD were still more than six times higher than those to the non-OECD in the last sub-period, 2010–2013 (Table 3).
More surprisingly, the effect of the host country’s endowment with natural resources proves to be negative, though only at the 10% level. The significantly positive effect of the (accumulated) number of bilateral investment treaties (BITs) is in line with some previous studies. It should be noted, however, that endogeneity tends to be an issue even though we include country fixed effects.
Argentina, Brazil, China, India, Indonesia, Mexico, Russia, Saudi Arabia, South Africa, and South Korea.
Albania, Armenia, Azerbaijan, Belarus, Bosnia & Herzegovina, Bulgaria, Croatia, Czech Rep., Estonia, Georgia, Hungary, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Macedonia, Moldova, Mongolia, Montenegro, Poland, Romania, Serbia, Slovakia, Slovenia, Tajikistan, Turkmenistan, Ukraine, and Uzbekistan.
A more detailed look at the development of per country FDI inflows for the different country groups (Table 3) reveals that the causes for the different peaks in the between-group component were quite different from each other. The strong peak in the mid-1980s was due to a strong increase in inflows to the G20 members that was accompanied by stagnating FDI inflows to the other country groups. By contrast, the increase in the between-group component in the 1990s was mainly caused by weak growth of FDI inflows to the other non-OECD countries, and its increase in the last period was mainly caused by a decrease in FDI inflows to the transition countries.
This is particularly true since the mid-1990s when per country inflows to the G20 have been between 18 and 23 times higher than those to other non-OECD countries, and between 8 and 15 times higher than those to the transition countries (Table 3).
In the last sub-period 2010–2013, the transition countries received about 1000 US$ per capita compared to about 400 US$ per capita for the G20 member countries and 330 US$ for other non-OECD countries.
This neglects the zero concentration for transition countries in the first sub-period which is economically meaningless, however, since there were no FDI inflows to any country in this group at that time.
Because of the low but increasing weight, \({\omega }_{\mathbf{T}}\), of transition countries during most of the observation period, the strong decline in concentration across transition countries had little effect on their contribution to the within-group component, however.
We exclude non-OECD countries of Eastern and Central Asia (EECA) since this group largely resembles the group of transition countries already analyzed in the previous subsection (plus Russia).
FDI inflows to the transition countries of EECA increased from essentially zero before the start of the transition (i.e., up to 1986–1989) to about 15% of all FDI inflows to non-OECD (non-OFC) countries in period 1998–2001.
The main reason for why the exclusion of the EECA countries has a much larger effect on the absolute than on the relative concentration measure is the fact that the weight of the EECA countries is much higher for the first than for the second measure. Together the EECA countries account for about 21% of all non-OECD countries but for only about 6.9% of the non-OECD countries’ aggregate population.
The strong peak in (absolute and relative) FDI concentration observed for the 1982–1985 period resulted from a corresponding peak in the within-group component of concentration only. More specifically it resulted from a strong temporary increase in the contribution of the MENA region to the within-group component, which was due in turn to both a strong temporary increase in the concentration of FDI inflows across the MENA countries (Fig. 6d) as well as a strong increase in the share of FDI inflows going to the MENA countries.
The share of FDI inflows to the SEAP region in total FDI inflows to the four regions increased from around 20% during the 1970s and the first half of the 1980s to about 43% in the late 1980s and about 58% in the early 1990s, before it decreased to about 45% at the end of the observation period (see Table 3).
See previous footnote.
FDI inflows to China, for example, may be high compared to the inflows to other countries in absolute terms, but still be comparatively low in per capita terms. An increase in the relative share of FDI inflows to China may thus increase absolute concentration but decrease relative concentration across SEAP countries.
Despite the decrease in concentration, the contribution of SSA to the within component actually slightly increased after the late 1990s (relative concentration) or the turn of the century (absolute concentration). This is due to the fact that the share of FDI inflows that went to the SSA region (strongly) increased from about 5% in 1994–1997 to more than 9% in the last sub-period (2010–2013). More generally, due to large changes in the shares of the different regions in total FDI inflows (which serve as weights in adding up within-group concentrations to the within-group component) the trends in the absolute and relative concentration of FDI inflows within the different regional groups are only partly reflected in the country groups contributions to the within-group component of concentration.
Only FDI inflows to the MENA group of countries increased even faster during that period (see Table 3).
More precisely, group L comprises 46 countries with low per capita income according to the World Bank’s income classification for the year 2005 (or the closest year for which data are available). Group H comprises 52 countries which the World Bank classifies as lower–middle-, upper–middle- and high-income countries. 1.4 billion people were living in L countries at the end of our period of observation, while 840 million people were living in H countries.
The decline in overall absolute concentration in the 1970s was exclusively due to decreasing concentration across low-income countries, however.
In the second last sub-period (2006–2009) all countries received strictly positive FDI inflows, and in the last sub-period only two out of 46 low-income countries representing slightly more than 2% of the low-income country group’s total population received zero FDI inflows.
The interaction with Infl provides an exception, suggesting that higher inflation discourages FDI in low-income countries.
The same approach is followed in the final step of our empirical analysis when distinguishing between low-income countries, L, and higher-income countries, H, among all other non-OECD countries, R.
To avoid the problem with zero FDI values, one is added to the original FDI values before calculating FDI per capita and before taking the log of the two FDI variables.
GDP, GDPpc, Open, Polcon, and BITs are observed in the initial year of each 4-year sub-period since 1970; Growth, NatRes and Infl are calculated as annual averages during the previous sub-period (1967–1970 for the first sub-period). The data are taken from the World Bank’s World Development Indicators, except for Polcon which is taken from the Polity IV project and BITs which is collected from UNCTAD’s Investment Policy Hub. Polcon ranges from one to seven with higher values representing stricter constraints on the executive.
Alternatively, we consider a dummy variable DumL which is set to one for the low-income countries, L, among all other non-OECD countries, R. Note that DumN and DumL per se cannot be identified since they are absorbed by the country fixed effects.
References
Bickenbach F, Bode E (2008) Disproportionality measures of concentration, specialization, and localization. Int Reg Sci Rev 31(4):359–388
Bickenbach F, Liu W-H, Nunnenkamp P (2015) Regional concentration of FDI in post-reform India: a district-level analysis. J Int Trade Econ Dev 24(5):660–695
Busse M, Königer J, Nunnenkamp P (2010) FDI promotion through bilateral investment treaties: More than a bit? Rev World Econ 146(1):147–177
Cadot O, Carrère C, Strauss-Kahn V (2013) Trade diversification, income, and growth: what do we know? J Econ Surv 27(4):790–812
Kekic L (2009) The global economic crisis and FDI flows to emerging markets. Vale Columbia Center on Sustainable International Investment, Columbia FDI Perspectives 15. New York
Nunnenkamp P, Thiele R (2013) Financing for development: the gap between words and deeds since Monterrey. Dev Policy Rev 31(1):75–98
OECD (2002) Foreign direct investment for development: maximising benefits, minimising costs. OECD, Paris
Stiglitz JE (2000) Capital market liberalization, economic growth, and instability. World Dev 28(6):1075–1086
UNCTAD (2014) World Investment Report 2014. United Nations, New York and Geneva
United Nations (2003) Financing for development. In: Monterrey consensus of the international conference on financing for development. http://www.un.org/en/events/pastevents/pdfs/MonterreyConsensus.pdf. Accessed Jan 2015
Zoromé A (2007) Concepts of offshore financial centers: in search of an operational definition. IMF Working Paper WP/07/87. International Monetary Fund, Washington, DC
Acknowledgements
We thank the editor and the two anonymous reviewers for their useful comments on the earlier version of the paper. We thank Michaela Rank for excellent research assistance.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A1: FDI inflows for different country groups
See Table 3.
Appendix A2: Regression analysis on FDI determinants
We assess whether factors that are widely considered to be major determinants of FDI flows can explain the changes of FDI concentration revealed by the decomposition of Theil indices. In particular, we attempt to explain the “convergence from the top” observed for OECD versus non-OECD countries by the following multivariate regression analysis:Footnote 46
-
(1a)
ln\({FDI}_{\mathrm{jt}}={\upalpha }_{1 }+{\upalpha }_{2}{{\varvec{ X}}}_{\mathrm{jt}}+{{\mu }}_{j }+{{\varepsilon }}_{\mathrm{jt}}\)
-
(1b)
ln\({ FDI}_{\mathrm{jt}}={{\upalpha }}_{1 }+{{\upalpha }}_{2}{{\varvec{X}}}_{\mathrm{jt}}+{{\upalpha }}_{3}({ DumN}_{\mathrm{j}} * {{\varvec{X}}}_{\mathrm{jt}})+{{\mu }}_{\mathrm{j}}+{{\varepsilon }}_{\mathrm{jt}}\)
-
(2a)
ln\({FDIpc}_{\mathrm{jt}}={{\upbeta }}_{1 }+{{\upbeta }}_{2}{{\varvec{X}}}_{\mathrm{jt}}+{{\mu }}_{\mathrm{j}}+{{\varepsilon }}_{\mathrm{jt}}\)
-
(2b)
ln \({ FDIpc}_{\mathrm{jt}}={{\upbeta }}_{1 }+{{\upbeta }}_{2}{{\varvec{X}}}_{\mathrm{jt}}+{{\upbeta }}_{3}({ DumN}_{\mathrm{j}} * {{\varvec{X}}}_{\mathrm{jt}})+{{\mu }}_{\mathrm{j}}+{{\varepsilon }}_{\mathrm{jt}}\)
The dependent FDI variable is defined in million US$ or, alternatively, in US$ per capita of the host country’s population, which corresponds to our measures of absolute and relative FDI concentration.Footnote 47 The vector X consists of the following variables: The host country’s GDP in millions of US$ in constant prices of 2010 (GDP) and the growth in GDP (Growth) reflect the size and growth of host-country markets which are widely supposed to drive market-oriented or horizontal FDI. Host countries with relatively low per capita incomes (GDPpc) may attract cost-oriented or vertical FDI. Vertical FDI is also expected to depend on the host country’s openness to trade (i.e., the ratio of exports plus imports over GDP, Open). The depletion of natural resources (in % of the host country’s GDP, NatRes) captures the host country’s attractiveness to resource-oriented FDI. We include the inflation rate (Infl) as an indicator of macroeconomic instability and the degree of political constraints on the executive (Polcon) to account for the potentially adverse effects of political discretion on FDI.Footnote 48 Moreover, we account for the accumulated number of bilateral investment treaties (BITs) ratified by the host country since BITs may help attract FDI inflows. In addition to these FDI determinants, the basic specification in equations (1a) and (2a) accounts for country fixed effects \({{\mu }}; {{\varepsilon }}\) represents the error term.
In extended specifications of the basic model (equations 1b and 2b), DumN is a dummy variable set to one for all non-OECD host countries.Footnote 49 It is interacted with each variable included in X to assess whether its impact differs between OECD and non-OECD host countries. We also include time dummies, in addition to country fixed effects \({{\mu }}\), in further extensions of the basic model.
Rights and permissions
About this article
Cite this article
Bickenbach, F., Liu, WH. & Nunnenkamp, P. How global is FDI? Evidence from the analysis of Theil indices. Empir Econ 55, 1603–1635 (2018). https://doi.org/10.1007/s00181-017-1346-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00181-017-1346-y