Spatial Interdependence in US Outward FDI into Africa, Latin America and the Caribbean

• Published date: 01 September 2014
• Author: Uwaoma G. Nwaogu,Michael Ryan
• DOI: http://doi.org/10.1111/twec.12118
• Date: 01 September 2014

Spatial Interdependence in US Outward

FDI into Africa, Latin America and the

Caribbean

Uwaoma G. Nwaogu

1

and Michael Ryan

2

1

Economics, Accounting and Management, Central College, Pella, IA, USA and

2

Department of

Economics, Western Michigan University, Kalamazoo, MI, USA

1. INTRODUCTION

THIS paper extends the previous literature on spatial interdependence and foreign direct

investment (FDI) by examining US outbound FDI into Africa, Latin America and the

Caribbean (LAC). We empirically examine the questions of whether spatial interdependence

matters in these regions for US FDI, and what the prominent motivations are for this invest-

ment.

1

However, we do so with a much broader empirical question in mind: given that the

existing literature on spatial interdependence has primarily focused on developed economies

(with good infrastructure and historically strong/significant trade ties), do spatial linkages exist

between countries where these facts do not exist? In Africa and Latin America, where the

infrastructure lags behind developed hosts and trade ties are less prominent, spatial interde-

pendence may not matter for FDI. This paper explicitly tests these relationships.

The importance of spatial analysis in analysing FDI flows comes from the fact that proxim-

ity matters and what happens in a particular region is interrelated with what is happening in

neighbouring regions. Put differently, geographical units closer to each other ought to exhibit

a higher degree of spatial dependence than those distant from each other. As suggested by

Tobler (1970, p. 236), the First Law of Geography states: ‘Everything is related to everythin g

else, but near things are more related than distant things’. Although the bilatera l gravity

model simplifies the empirical analysis allowing researchers to focus on how certain host and

parent country characteristics interact to affect the host’s inward FDI, it ignores the role

played by neighbouring countries, otherwise known as third-country effects. Ignoring such

effects could pose several problems.

For instance, Blonigen et al. (2007) noted that the recent decade’s empirical work on FDI

determinants relies primarily on a bilateral gravity-type framework. However, from an econo-

metric perspective, third-country effects are important as their omission leads to biased, incon-

sistent or inefficient parameter estimates (Anselin, 1988). These third countries are usually

countries in the immediate proximity to the host country. Furthermore, bilateral FDI analysis

We would like to thank Paul Elhorst for his helpful comments and suggestions on the MATLAB coding.

We also thank our referee, Susan Pozo, Christine Moser, and Rajib Paul for their valuable comments on

the text.

1

Here we consider FDI ‘motivation’ to be vertical, horizontal, export-platform, or complex vertical

FDI. See subsection 2b for a discussion of each strategy. A mixture of FDI strategies might exist when

one uses country- and industry-level data; as a result, one cannot directly test for the existence of one

form of FDI over the other. However, as empirical analysis can capture net effects, Blonigen et al.

(2007, p. 1308) indicate ‘evidence of one dominant form of multinational enterprises activity in the data

is possible given unique sign patters across the various FDI forms’.

©2013 John Wiley & Sons Ltd 1267

The World Economy (2014)

doi: 10.1111/twec.12118

The World Economy

is problematic as it cannot explain newer FDI models such as export-platform FDI (Ekholm

et al., 2007) and complex vertical FDI (Baltagi et al., 2007; Kazunobu and Toshiyuki, 2011).

The use of spatial econometrics allows us to overcome these problems by capturing how the

dependent variable (FDI) for a given spatial unit is jointly determined with that of surround-

ing spatial units, weighted by their geographical distance. For instance, recent empirical mod-

els (Coughlin and Segev, 2000; Baltagi et al., 2007; Blonigen et al., 2007; Garretsen and

Peeters, 2009; Ledyaeva, 2009) have relaxed the two-country assumption by accounting for

spatial interdependence in a gravity framework. In particular, empirical analysis from Baltagi

et al. (2007) and Blonigen et al. (2007) confirm that after controlling for all the relevant loca-

tion determinants, multinational enterprises (MNEs) do take third-country effects into

consideration when deciding on where to do FDI.

If one were to look for a unifying theme among the aforementioned spatial FDI papers, it

would be that the destination countries/regions within each particular study are somewhat

homogeneous. For instance, 17 of Blonigen et al.’s (2007) 35 host countries are European,

2

suggesting strong geographical, infrastructure (rail, air, road) and social ties exist, binding

these countries together economically. In addition, by 2010, 24 are OECD members, with

non-OECD member Singapore ranking 26th in the UN’s Human Development Index.

3

We

also expect strong economic ties to exist between these hosts, even if they are not geographi-

cally close. In fact, the CIA Factbook highlights that in almost every case, the top export

destinations for the European countries in the Blonigen et al. (2007) study are other European

countries (see https://www.cia.gov/library/publications/the-world-factbook/fields/2050.html#

be). In essence, a high level of homogeneity exists among the hosts with regard to economic

development, social development, legal structure and geography. Garretsen and Peeters’

(2009) study is subject to the same concern, as it focuses on Dutch FDI into 18 countries, 14

of which are European (10 of 12 are European when they focus on FDI at the sectoral level).

Coughlin and Segev (2000) examine FDI into 29 Chinese regions and thus have a less

homogeneous set of recipient regions with respect to economic development. However, these

are a geographically unified set of host regions tied together under the Chinese federal

government (and thus use the same currency, legal system, etc.). In each of these cases, the

presence of several unifying ties between hosts would, a priori, reasonably suggest spatial

interdependence exists.

However, the question that naturally arises from this is what if we ‘relax’ some of these

homogeneities and economic ties among the recipient countries. To do so, we focus on US

FDI flows into Latin American/Caribbean (LAC) and Africa. While countries within these

regions share a similar continent, they are far more geographically dispersed than Europe,

which is 1/3 the area of Africa and 60 per cent the size of South America. In addition, we

have greater heterogeneity in our sample with regard to host economic and social develop-

ment. For instance, Blonigen et al.’s (2007) set of 35 countries average 0.822 on the 2011

UN Human Development Index (‘very high human development’), with a standard deviation

in this rank of 0.105. Garretsen and Peeters’s (2009) 18 hosts average 0.890 (0.029 standard

deviation) on this scale, indicating a much more homogeneous set of recipient countries.

In contrast, our set of countries, 68 countries, has an average index of 0.595, with a

2

18 counting Israel, typically considered be more economically aligned with Europe than its own

geographical region.

3

See http://hdr.undp.org/en/statistics/. Blonigen et al.’s (2007) sample of 35 host countries includes only

six countries from LAC and two from Africa.

©2013 John Wiley & Sons Ltd

1268 U. G. NWAOGU AND M. RYAN