Regional Integration and

Foreign Direct Investment

in Developing Countries 1

Dirk Willem te Velde2 and Dirk Bezemer

July 2004

Abstract

The empirical literature offers little guidance on whether some regions are more successful in
attracting FDI than others. We bring together two differing approaches (detailed descriptions
of regions and studies estimating effects econometrically based on 0/1 dummies) and estimate
a model explaining the real stock of UK and US FDI in developing countries, covering 68
(UK) and 97 (US) developing countries over 1980-2001 and identify the effects of specific
regional investment-related provisions on FDI. We show that i) membership of a region leads
to further extra regional FDI inflows, but the type of regional provisions matters; ii) that the
position of countries within a region matters.

Jel code: F15, F21

Key words: Regional Integration, Foreign Direct Investment, Developing Countries
(including regions in Latin America, Africa and East Asia)

1
This paper forms part of a larger DFID/EC-PREP funded research project on Regional Integration and
Poverty. In a survey paper for the project as a whole, Te Velde, Page and Morrissey (2004) argue that
the attraction of FDI is one way by which Regional Integration can affect poverty because FDI can
increase growth and development when appropriate policies and economic conditions are in place.
Other pathways between regional integration and poverty work through trade, migration and functional
cooperation such as through regional (social) investment funds (see
http://www.odi.org.uk/iedg/Projects/ec_prep1.pdf).
2
Research Fellow, Overseas Development Institute; dw.tevelde@odi.org.uk. We are grateful for
comments made during a seminar at the University of East Anglia on 31 March 2004 and by Sheila
Page and Oliver Morrissey. The UK Department for International Development (DFID) supports
policies, programmes and projects to promote international development. DFID provided funds for this
study (an EC-PREP project on Regional Integration and Poverty) as part of that objective. The views
and opinions expressed are those of the authors alone.
1 Introduction

The purpose of this paper is to understand the relationship between Regional

Integration and Foreign Direct Investment in developing countries. A theoretical and

empirical literature on Regional Integration and FDI has begun to emerge over the

past decade. This has coincided with strong growth in both the number of Regional

Trade Agreements (RTAs) notified to the WTO and the value of FDI in developing

countries. There appears to be a consensus in the literature that RI leads to further

(extra and to some extent intra-regional) FDI. One of the factors often cited is the

increase in the “market size” that follows regional integration.

Chart 1 The number of GATT/WTO notified RTAs in force

200
180
160
140
120
100
80
60
40
Excl. EU-CEEC
20
0
194819511954 195719601963 19661969 1972 197519781981 1984198719901993 199619992002

Source: WTO

However, the empirical literature seems to offer little guidance on whether different

regions are more successful in attracting FDI (unlike an emerging literature on the

effects of certain trade provisions in regions (see e.g. Estevadeordal and Robertson,

2002 on tariffs; and Estevadeordal and Suominen, 2003 on rules of origin) than

others, and if so, why some regions are more successful and crucially whether trade

negotiators can design an RTA in order to have the best possible outcome for FDI.

Empirical studies of FDI and Regional Integration can be divided into:

1
 i) those that describe the investment-related provisions present in a growing

number of RTAs with a prediction of how these should affect FDI (see e.g.

UNCTAD, 1996; Te Velde and Fahnbulleh, 2003); and

ii) those that base their findings on econometric models explaining FDI where

one of the explanatory variables is a “black box” 0/1 dummy or binary

variable describing whether or not a country is a member of a region (see

e.g. Levy et al., 2002).

This paper aims to bring these two approaches together by moving beyond describing

a region as a “black box”3 and conduct empirical research that can help to identify the

effects of specific investment-related provisions in RTAs on Foreign Direct

Investment. We illustrate this by estimating a model explaining the real stock of UK

and US FDI in developing countries, covering more developing countries than

contained in the OECD database often used for such analyses, over the period 1980–

2001. An innovative feature of the analysis is the use a variable that measures the

scope of regional investment and trade provisions in an RTA in addition to standard

explanatory variables.

The structure of this paper is as follows. Section 2 reviews the theory on the

relationship between regional integration and FDI, concentrating on regional trade

and investment rules. Section 3 discusses the econometric evidence arguing that most

econometric studies use a simple 0/1 dummy to describe regions, which offers little

guidance for trade negotiators on whether different types of regions have different

effects on FDI and if so, why. On the whole, econometric studies measure RTAs as
3
One can argue that the term “black box” is not adequate for all studies, because some researchers
discuss provisions within regions but measure these with a 0/1 dummy. Hence, the term black box
refers predominantly to the way regional provisions are measured.

2
black boxes that do or do not exist. In Section 4, we attempt to step outside the black

box of a region and measure trade and investment provisions in RTAs. Section 5

provides new econometric evidence on the effects of Regional Integration on FDI,

focusing on trade and investment provisions in regions and on differences in effects

on FDI amongst members of a region. Section 6 concludes.

3
2 Regional Integration and Foreign Direct Investment: theory

There are various ways through which RTAs can influence FDI and vice versa. We

can distinguish among investment rules, trade rules and other links (see e.g.

Blomstrom and Kokko, 1997; Dunning, 1997a).

Regional Investment rules and FDI

Investment rules govern cross-border investment in the region and usually consist of

rules on treatment and protection of FDI contributing to the “investment climate”.

Investment rules exist in a handful of RTAs4 although they are not as common as

trade rules, particularly amongst the poorer developing countries. Some regions

include voluntary principles (e.g. APEC) while other regions include rules with

effective dispute settlement procedures. Several studies discuss a number of

investment provisions in regional treaties (scope, standard of treatment, performance

requirement, expropriation and dispute settlement mechanisms) and their expected

effects on the volume of FDI. The provisions sometimes apply to regional investors

and sometimes to extra-regional investors.

There is a heated discussion on how investment rules (bilateral, regional and

multilateral) affect investment decisions. Generally, a predictable investment climate

can be in the interest of investors when they were previously disadvantaged by

unpredictable investment conditions. It is not clear whether this would lead to

additional FDI or simply to more comfort for the investor. It is, however, clear that

surveys reveal that investors want a predictable investment climate (e.g. CBI position
4
Investment rules also appear in bilateral trade arrangements (e.g. Singapore-Japan), which are
included here as RTAs if they are notified to the WTO, but more often appear in bilateral investment
treaties.

4
paper for WTO negotiations, EU survey of MNEs), although not necessarily at the

cost of other policy liberalisation (e.g. further trade liberalisation). The predictability

of the investment climate may be enhanced when domestic policies are enshrined or

locked into regional treaties. Much will also depend on existing treatment. If

treatment of existing investors is already good in practice, new (regional) rules may

add little to generating new investment or a better investment climate, other than

offering a little more long-run security. There seems to be no empirical evidence that

addresses the effects of individual investment provisions on FDI.

Regional trade rules and FDI

The elimination of intra-regional tariffs will affect trade vis-à-vis the level of sales by

multinational subsidiaries depending on the importance of transport (e.g. tariff) costs

and plant-level and firm-level costs in setting up multinational subsidiaries (Markusen

and Venables, 1997; Brainard, 1997; Carr et al., 2001). Hence, the type and motive of

investment plays an important role in understanding how FDI is affected by tariffs

and trade (Barrell and Te Velde, 2002). To reflect this, we distinguish between intra-

regional and extra-regional FDI and between horizontal (market-seeking: subsidiaries

selling similar products) and vertical (efficiency and natural resource seeking:

subsidiaries exploiting efficiencies or control over inputs) FDI.

Regional tariff preferences can decrease horizontal (tariff-jumping) intra-regional FDI

because it may now become cheaper to serve the partner country by trade rather than

to establish a subsidiary and incur plant-level costs more than once and firm-level

costs only once. Of course, when firm-level and plant-level fixed costs are zero, there

will be no trade and no concentrated production facility or FDI – just national

5
production. However, on the other hand, regional tariff preferences encourage

vertically-motivated intra-regional FDI, because lower trade costs will provide

incentives to establish international production networks and establish an efficiency-

seeking subsidiary in a partner country that can process imports for re-export.

Extra-regional FDI (the focus of the empirical part of the paper) can also be affected

by declining regional tariff preferences in different ways. First, by lowering tariffs

amongst parties to the RTA, it may become profitable for an extra-regional investor to

avail of an effectively larger market (horizontal market-seeking FDI) from one or

more locations in the region (export platforms). If individual countries of a region are

previously served by trade, this may then raise inward FDI (export platforms or

beachhead locations, see Ethier, 1998). However, if the member countries of a region

were already served through sales of a multinational subsidiary, concentration of

production may occur in one or a few countries in the region, with ambiguous or

negative effects for the volume of extra-regional FDI in each country. The

combination of lower internal tariffs and significant plant fixed costs would lead to a

consolidation of several plants in several members of the region into one or a few

plants, used by the parent to serve the region as a whole. This may also induce FDI

inflows to the most cost-efficient location (usually nearest to the largest market),

possibly at the cost of FDI to other members in the same region. This could be the

case for market-seeking multinationals. The effects of regional trade preferences for

extra-regional vertical (or efficiency-seeking) FDI is likely to be small, though lower

regional preferences may lower costs and raise efficiency in the vertically motivated

subsidiary when it uses inputs from more than one country in the region (regional

enterprises in ASEAN, ANDEAN or SAARC).

6
There are various effects of regional tariff preferences on inward FDI. However, in

the context of developing country regions, where most inward FDI is inter-regional

(even though South Africa is an important investor in SADC), the market size

argument would be the most important, and apart from other factors regional tariff

preferences would tend to raise inward FDI. It must be noted, however, that the

strength of this argument depends on the difference between tariffs applied regionally

and tariffs applied to others on an Most Favoured Nation (MFN) basis. Indeed, the

market size gained as a result of regional integration needs to be the total market size

of the region corrected for initial MFN rates, as this will indicate the maximum scope

of the market size gained. Hence, the higher the MFN rate and the larger the market

size, the more likely it is that market-seeking investors will respond. Accounting for

this, it emerges that regions in Africa such as COMESA and SADC have as much to

offer as ANDEAN not because of their similar market size (second column, Table 1)

but because of similar regional market size corrected for MFN tariffs (final column).5

Table 1 Market size in seven developing country regions

Market size Unweighted Approximate market
(US$bn, 2001) average MFN size “gained” by
tariffs regional tariff
(latest year preferences
available)
NAFTA 11400 8.2 935
MERCOSUR 797 13.6 108
ANDEAN 287 11.2 32
ASEAN 548 7.9 43
SAARC 605 23.6 143
COMESA 181 16.9 31
SADC 171 15.6 27
CARICOM 33 11.6 4
Source: WTO, WDI

5
One can argue that this correction is simplistic: it does not take into account the distribution and peaks
of tariffs, or whether the tariffs are on goods relevant for FDI.

7
Rules of origin constitute another trade rule that can affect location decisions. The

effects of rules of origin (RoO) on investment can vary depending on the type of

investment as well as the interaction with regional tariff preferences. The RoO can

encourage the use of intra-regional inputs diverting away from extra-regional inputs,

even if these were more efficient. However, a stricter and more costly RoO would

stifle intra-regional trade favouring extra-regional imports (which are likely to face

the MFN tariff). The higher the difference between MFN tariffs and regional tariffs,

the higher the incentive to comply with the RoO by importing regionally using good

certificates (see Estevadeordal and Suominen, 2003).

Non-tariff barriers to trade can also affect investment. NTBs include voluntary export

restraints; the threat of imposing EU quotas and using anti-dumping against Japanese

exports motivated the Japanese to set up operations inside the EU. Barrell and Pain

(1999) found that after controlling for relative labour costs and market size, Japanese

investment flows to EC countries over 1980–91 were significantly influenced by anti-

dumping activities taken in the EC.

On balance, RTAs should lead to increased extra-regional FDI, but there are more

ambiguous results for intra-regional FDI. An important reason for the ambiguity of

the effects of trade rules is that MNEs are motivated by exploiting firm-specific assets

(e.g. firm-specific fixed costs) and hence want to enjoy economies of scale and scope,

in addition to simply jumping trade barriers.

8
Other regional initiatives and FDI

There are various other links between RTAs and FDI. Many provisions are region

specific and cannot be easily categorised. For example, provisions other than the trade

and investment rules include free movement of people (CARICOM) and free transfers

of profits which can all facilitate the establishment of intra-regional FDI.

Taking another example, some regions (ANDEAN, ASEAN, MERCOSUR) have

cooperation schemes which aim to establish regional enterprises by promoting joint

ventures. The ASEAN region seems to be one of the most advanced in this area. The

ASEAN Industrial Cooperation scheme (AICO Scheme) seeks to promote joint

manufacturing industrial activities between ASEAN-based companies. More than 100

projects have been selected for special tax and tariff incentives. The ASEAN

secretariat has also begun various activities in the area of investment facilitation, by

providing information through portals, databases, publications and statistics. It can

thus be said that a region can do much more to try to promote investment than design

and implement trade and investment rules. They can put in place the regional

infrastructures (legal, institutional etc.) to deal with investment issues at a regional

level.

Some argue that the effects of RTAs on FDI are not so much about trade and

investment rules, but about the increased predictability of the investment climate by

locking in general reforms (regulation, competition policies, property rights, contract

enforcement, guaranteed access to members’ markets and stable trade policies) in a

wider context. The fact that national policies are “locked” in regional treaties should

give investors additional security in that policy reversals are less likely, reducing non-

9
commercial risk. In practice, this argument would depend on how strong the region is

vis-à-vis individual members. The argument is also related to signalling, in that

signing an RTA signals an intention that can be regarded as favourable to investors;

this would apply equally to intra and extra-regional investors.

Many argue that important effects of RTAs on FDI are dynamic, with competition

creating a more efficient industry and growth, which in turn can affect FDI. Neary

(2001) includes dynamic effects in a theoretical model of describing MNEs. First,

there is the tariff-jumping motive as discussed before: FDI is favoured over exporting

the higher the external tariff and the lower the fixed costs of a new plant. Second, the

export platform motive could affect FDI, as lower intra-regional tariffs would favour

a single plant in the region. Finally, lower intra-regional tariffs would lead to

increased competition from stronger domestic firms and hence lower FDI. On the

other hand, a more efficient private sector can raise efficiency-seeking investment by

becoming efficient regional suppliers, and raise strategic asset-seeking investment.

Blomstrom and Kokko (1997) also argue that regional integration leads to efficiency

gains and higher growth, and thus further FDI. FDI can actually be such a catalyst

through spillovers in terms of technology transfer and other linkages with local firms.

There can thus be long-lasting effects on growth and productivity as opposed to a one-

off effect based on a more efficient allocation of resources. Schiff and Wang (2003)

show that NAFTA imports has raised productivity (between 5.5-7.5%) in Mexico in

the form of imported knowledge stocks, while other imports did have no effects.

10
Apart from trade and investment rules and regional institutions, regions can also

decide to harmonise fiscal and monetary policies. For instance, the Euro area (within

the EU), the UEMOA, and four out of five SACU members (within SADC) have

common currencies. This reduces intra-regional exchange-rate variability and may

reduce cross-border transaction costs, which are amongst the factors contributing to

investment. Because the EU and SADC and SACU are incomplete currency areas,

there should be implications for which parts of the region are influenced.

Spatial distribution of FDI across region

While regional integration can lead to more extra-regional investment for the region

as a whole, this may not lead to more FDI in each individual member country. While

peripheral countries to the EU, such as Ireland, have caught up in terms of

productivity levels with other members of the EU – apparently through trade and FDI

spillovers, there has been a degree of divergence and agglomeration in developing

regions such as the East African Community and the Central American Common

Market, both dating back to the 1950s and 1960s.

An uneven spread of benefits amongst members (Venables, 1999) can be enhanced by

agglomeration effects. Agglomeration effects refer to a spatial clustering of economic

activities. Agglomeration can occur within a county (e.g. cities) or across countries.

Clusters of economic activities can lead to efficiency gains, for instance, because a

pool of specialised support services is feasible owing to economies of scale (e.g.

Porter, 1998). If relocation effects occur within a region, this may lead to efficiency

gains which may reinforce further relocation effects. This would lead to further

divergence or convergence, which could affect the distribution of gains from and

11
ultimately the motives for regional integration processes. On the other hand, as argued

in Ethier (1998) smaller (and possibly poorer – though this is obviously not the case

in regions such as ASEAN) countries may actually have incentives to form a region in

order to attract investment away from other members, particularly extra-regional FDI.

This may be the case when regional tariff preferences allow foreign investors to set up

beachhead locations in a small (or poor) country to serve the entire regional market.

Hence, the spatial distribution of FDI is an empirical question and depends on factors

such as the level of external MFN tariffs, strictness of RoO, market size and

agglomeration effects in individual member countries.

3 Regional Integration and Foreign Direct Investment: econometric evidence

The empirical evidence has begun to address the links between RTAs and FDI. Table

2 provides a review of a few studies tentatively finding that RTAs in most cases boost

extra-regional FDI and in some cases intra-regional FDI. Levy et al., (2002) address

the issue of regional integration and FDI at a basic level, using dummies for regions,

applying the analysis to the OECD database covering 60 countries (thus – excluding

many developing countries). The regressions control for a number of factors and use a

variable for market size. Other researchers have examined individual regions;

Waldkirch (2003) and Monge-Naranjo (2002) for NAFTA, Chudnovsky and Lopez

(2001) for MERCOSUR, and UNCTAD (2003) for several regions.

Dunning (1997b) analysed empirical findings regarding the effects of the formation of

the Internal Market Programme (IMP) in Europe largely on the basis of econometric

studies. He finds that the main dynamic impact of the FDI is through the effects on

12
other determinants of FDI, such as market size, income levels, structure of activity

and agglomeration economies. IMP as an independent variable raised extra- and to a

lesser extent intra-regional FDI but not by as much as other variables. The effects of

the IMP were industry specific, with extra-EC FDI increasing more in FDI sensitive

sector. There was limited evidence that economic activity has become geographically

concentrated as a result of the IMP, although high value-added activities remained

clustered and lower value activities became more dispersed. Finally, there was

complementarity between trade and FDI.

Table 2: RTAs and FDI inflows, selected econometric studies

Study Research question; Explanatory variables Findings

Region, countries and
years; Methodology

Levy, Stein How do RTAs affect the RTA membership, extended market • RTA membership doubles FDI stocks on
and Daude location of FDI? host, extended. market source, average
(2002) capital/worker
FDI from 20 OECD countries distance, market size, bilateral FDI increases upon joining a FTA with:
to 60 OECD/non-OECD trade, inflation • more trade/GDP (openness)
countries, 1982–98 trade/GDP, privatisation • more similar capital/worker
capital/worker, investment • better investment environment
environment, common border, • larger market
common language

Srinivasan Which factors determine US Market size, labour costs • When split by periods (1977–81; 1982–86;
and Mody and Japanese FDI? capital costs,. previous FDI 1987–92), no evidence that IMP increased
(1997) infrastructure (telephone, US and Japanese FDI (but we should bear in
35 OECD and non-OECD electricity), country risk mind that IMP was complete only in 1993)
countries, 1997–92, split out openness
in groups of low-middle, high
income countries; and EEC,
Latin America, East Asia

Brenton et al., Does European integration Population, distance, trade/FDI • Single European Act (1992) and Iberian
(1998) increase FDI? Does it divert agreement dummies, host country enlargement : more FDI but no observed FDI
FDI? Are trade and FDI economic freedom dummies, CEE diversion
substitutes or complements? dummies, host country EU
membership dummy, FDI residual
FDI in and outflows, imports, (in trade regression)
exports for EU and CEEC
countries
Pain and How has intra- and extra EC Sector output, factor costs, currency • FDI determinants differ over sectors
Lansbury FDI by UK and German volatility, corporate finance • IMP introduction boosted FDI
(1996) forms in different sectors conditions, non-tariff barriers (1–3 • IMP redirected UK FDI from US to EC
changed with the introduction scale), IMP dummy, sector
of the Internal Market dummies
Programme (IMP)?

UK and German outward FDI

for seven sectors, 1980/81–92

13
Because the econometric studies use a simple 0/1 dummy to describe regions, there is

little guidance for trade negotiators on whether different types of regions have

different effects on FDI and if so, why. On the whole, econometric studies measure

RTAs as black boxes, which either exist or not, but which do not differ in content.

There is one recent exception. Dee and Gali (2003) examine how “new” trade

provisions in Preferential Trade Agreements affect the patterns of trade and

investment flows. They use gravity models of trade and investment between pairs of

countries over 1988–97. They include two type of indices: i) covering “traditional”

trade provisions regarding agriculture and ii) industrial products and “new age”

provisions covering services and other provisions such as investment rules. The

indices are unweighted averages of scores on sub-categories. They also control for the

usual control variables in gravity equations and include three dummies for each RTA

provision to measure intra-regional effects, extra-regional effects on inward FDI and

extra-regional effects on outward FDI. The traditional trade provisions affected both

intra-regional inward FDI stocks and extra-regional inward FDI stocks in

SPARTECA (investment creation), but only extra-regional outward FDI in the EU

and US-Israel RTA (investment diversion). The new age provisions led to net

investment creation in EFTA, EU, NAFTA, MERCOSUR, SPARTECA, CER, net

investment diversion in AFTA, and no impact in ANDEAN and US-Israel (Tables 4–

7 in Dee and Gali, 2003).

While the study by Dee and Gali has gone some way in understanding the effects of

different provisions in regions on trade and investment flows, many questions relevant

for this paper have remained unanswered. For instance, the study did not include

14
regression with “all RTAs” or RTAs with African countries; it did not include a lot of

developing countries, focusing their attention on RTAs relevant for Australia; it did

not track regional provisions over time –provisions can change over time (as e.g. in

ASEAN); finally, it is not clear whether different types of countries within regions are

affected differently.

4 Stepping out of the “black box”

This section moves beyond describing regions as a black box and describes regions on

the basis of provisions included in protocols. While several papers have included a

discussion of investment and other provisions (UNCTAD, 1996; Page, 2000), none

includes a measured description of trade and investment provisions over time.

Description of provisions in regions

Te Velde and Fahnbulleh (2003) discuss trade and investment provisions across seven

main regions, as well as for each region over time. The following provisions are

compared across regions: investment rules (scope and coverage; National Treatment;

Most Favoured Nation and fair and equitable treatment; performance requirements;

transfers of funds; provisions with respect to expropriation exist; Settlement of

Disputes) and trade rules (Rules of Origin; tariff structures; other relevant provisions).

The comparison yielded some interesting insights: for instance, ANDEAN restricted

FDI in the 1970s but this changed over the 1980s and 1990s. ASEAN has gradually

added more investment provisions over time. NAFTA included quite strong

provisions from its inception in 1994. SADC and COMESA contain weak trade and

15
investment provisions. Generally, regions differ with respect to trade and investment

provisions in two fundamental respects:

• Over time, when regions change or add investment-related provisions

• Across regions, when investment-related provisions differ at one point in time

Table 3 measures trade and investment provisions for 7 regions which are arguably

the most advanced in the developing world regarding the inclusion of investment-

related provisions. The following keys were used:

Investment Index = 0 if not member of group

= 1 if some investment provisions in region (as in COMESA, SADC),
= 2 if advanced investment provisions in region (e.g. improved investor protection in ASEAN)
= 3 if complete investment provisions in region (e.g. Chapter XI of NAFTA)
= -1 if more restrictive provisions (restrictions on foreign investors in ANDEAN in 70s)

Trade Index = 0 if not member of group

= 1 if some trade provisions (e.g. tariff preferences),
= 2 if low MFN, (close to) zero intra-reg tariffs
= 3 if high MFN, (close to) zero intra-reg tariffs

Based on this classification and bearing the theoretical discussion of Section 2 in

mind, a higher value of the index should lead to further (extra-regional) FDI.

Table 3 Regional Integration Index

Investment provisions Trade provisions
1970s 1980s 1990s 1970s 1980s 1990s

NAFTA 0 0 3 (1994) 2 (1994)

MERCOSUR 0 0 2 (1994) 0 0 3 (1991)
CARICOM 0 1 (1982) 2 (1997) 1 (1973) 2 3 (1997)
-1(1970)
ANDEAN 1 (1987) 2 (1991) 1 1 2 (1993)
ASEAN 0 1 (1987) 2 (1996), 3 (1998) 1 1 1
SADC 0 0 1 (1992) 0 0 1 (1992)
COMESA 0 0 1 (1994) 0 0 1 (1994)
Source: Measurement of provisions described in Te Velde and Fahnbulleh (2003); years between
parentheses indicate when certain provisions were announced.

16
Some illustrations

There appears to be hardly any empirical literature that links time varying trade and

investment provisions with inward FDI. Here we illustrate how such as link could be

visualised for extra-regional FDI (see UNCTAD, 2003, for total inward FDI in

ASEAN). Charts 2–4 show how the stock of US FDI as a percentage of GDP evolves

over time as regions add or change trade and investment provisions for ANDEAN

(the sum of Bolivia, Peru, Ecuador, Venezuela and Colombia), MERCOSUR

(Argentina, Brazil, Paraguay and Uruguay) and ASEAN-5 (Indonesia, Malaysia,

Philippines, Singapore and Thailand).

Chart 2 indicates that US FDI into ANDEAN fell in the early 1970s after the

introduction of restrictive investment provisions (especially regarding extra-regional

FDI) contained in Decision 24. It has gradually recovered since the 1990s, when

Decisions 291 and 292 – marking the formal end of the common restrictive policy

towards FDI and the ANDEAN free trade area – were being put into practice. US

FDI rose in MERCOSUR some time after its inception and investment provisions

were included in the mid 1990s (Chart 3). US FDI also seems to have responded to a

gradual increase in trade and investment provisions in ASEAN, the signing of an

investment protection agreement in 1987, the start of the ASEAN Free Trade

Agreement in 1992, improved investor protection in 1996 and the signing of the

ASEAN investment area. The charts suggest that the causation, if any, runs from

investment provisions to FDI.

While the charts control for market size (GDP) of the countries within the region,

other variables are not taken into account. The evidence presented here is therefore

17
only preliminary evidence that FDI is responsive to the type of investment provisions

included in regions – not just whether or not a country is a member of a region. Such

evidence would need to be supported by more formal statistical evidence, to which we

turn in the next session.

Chart 2 US Foreign direct investment in ANDEAN-5

Stock as per cent of GDP
Categena Decision 24 Decisions Andean free
0.15 Agreement 291and 292 trade area

0.10

0.05

0.00
1966 1971 1976 1981 1986 1991 1996 2001

Chart 3 US Foreign direct investment in MERCOSUR-4

Stock as per cent of GDP
Treaty Investment
0.08 of Protocols
0.07 Asuncion
0.06
0.05
0.04
0.03
0.02
0.01
0.00
1966 1971 1976 1981 1986 1991 1996 2001

Chart 4 US Foreign direct investment in ASEAN-5

Stock as per cent of GDP
Inv Signing Improved
0.12 protection AFTA protection AIA
agreement
0.1
0.08
0.06
0.04
0.02
0
1966 1971 1976 1981 1986 1991 1996 2001

5 New evidence on Regional Integration and FDI

The effects of investment-related provisions in regions can be treated more formally

in econometric models.

18
Model set-up

Several determinants of FDI in developing countries are frequently found to be

significant (see e.g. Wheeler and Mody, 1992 and Dunning, 1993): i) the general

potential for viable projects, on the demand side (growth and size of market) and

supply side (skills, infrastructure, financial and technological development); ii) the

domestic regulatory framework within which investment can take place affects

investment decisions (e.g. protection of property rights); and iii) specific factors that

can affect particular projects (availability of project finance, technical assistance,

provision of specific information etc.). A large literature has emerged on determinants

of FDI, of which RI can be one (see e.g. Te Velde 2003).

There are two types of empirical specifications common in the literature on

determinants of FDI. The first is the gravity model explaining bilateral FDI stocks.

Gravity models have recently been based on theoretical foundations (e.g. Harrigan,

2001) and perform well in explaining bilateral trade. Recent advances in

understanding locational decisions, in particular the knowledge-capital model, have

led to the use of gravity models in determining FDI (Carr et al. 2001; Levy et al.,

2002). Gorg and Greenaway (2002) apply the gravity model to bilateral UK FDI

stocks in Central and Eastern European countries, see Greenaway and Milner (2002)

for further detail

We have decided to follow a second approach which is broadly in line with various

authors such as Pain (1997) who applies the methodology to UK FDI in Europe and

the US. We will take a standard FDI model with standard explanatory variables and

19
include an additional variable measuring the degree of implementation of the

investment provisions. In this way we can isolate a separate RTA (provision) effect6

(1) FDI ijt = f ( HOMEijt , HOSTijt , OTHERijt , RTAkjt )

where FDI is the real stock of FDI, i is the home country (here US and UK, and

hence we refer to extra-regional FDI only, except of course US FDI in NAFTA), j is

the host country, t time. HOME country factors can include GDP or interest rates, or

more simply a dummy if it is expected that different source countries react differently.

HOST country factors can include amongst others market size, human capital, and

infrastructure. OTHER may include such variables as distance or shared language.

RTA denote measures of (the sum of) investment-related provisions k in an RTA

applicable in host country j at time t. Rules that are expected to raise FDI (extra,

and/or intra-regional FDI) would appear in the regression with a significant and

positive regression coefficient.

As we indicated above, we cannot expect all countries to be affected by regional

integration in the same way, and hence we include an interaction term between

regional integration processes and the position of the countries within the region

(2) FDIijt = f ( HOMEijt , HOSTijt , RTAjt , RTA jt * POSITION jt )

where POSITION measures the position of country j in a region in three different

ways:

• Real GDP of country j compared to the largest economy in the region at time t.

This tests whether countries of different sizes attract different amounts of FDI. As

6
Dunning (1997b) argues that important effects of RTAs can work through the explanatory variables
and are dynamic. We can control for the regional market size effect, by including it as an explanatory
variable in the regression. However, this is not as straightforward for the other effects on other
explanatory variables, and the variable RTA in the above equations will ultimately pick up such effects.

20
 discussed before, this is an issue with opposing views (Ethier, 1998; Venables,

1999).

• GDP per capita of country j compared to the richest country in the region at time t.

This tests if richer or more productive countries attract more FDI than poorer and

less productive

• Distance of country j from the largest market in the region. This tests whether core

and periphery countries attract different amounts of FDI.

Ideally we would estimate dynamic versions of equations 1 and 2. However, this is

hampered to some extent because we deal with bilateral FDI data containing many

gaps, either for reasons of confidentiality or because it is not measured. The same

applies to some of the explanatory variables, and so it is difficult to use first

differences or dynamic panel data estimators to the most extensive database. While it

is possible to have time continuing variables for a selective group of countries, we

have initially chosen to keep as many countries as possible in the sample. One way to

avoid including a dynamic element is by including time dummies. But, we will also

use a version of an error correction form which can be used to distinguish between

long-run and short-run effects, for a more selected sample of countries:

(3) ∆ ln FDIijt = α ln( FDIijt−1 ) + β ln( HOSTGDPijt−1 ) + χ RTAjt + γ∆ ln( HOSTGDPijt ) + cons + USdum + εijt

Results

We apply versions of equations 1 and 2 to a pooled sample of US and UK FDI in

developing countries over 1980–2001 (see Appendix for a data description). There are

many gaps in the data, with observations per country varying, so it is an unbalanced

panel with a total of 1561 observations. Tables 4 and 5 contain the results of

21
estimation using OLS or GLS estimation. We correct the standard errors for serial

correlation and heteroscedasticity using White’s robust estimator. We approximated

the home country effect by a US fixed effect.

Column I shows an FDI model with standard explanatory variables, including

infrastructure, education and inflation. The coefficients are significant and with the

expected sign, except in the case of inflation which is not significant in this

regression. The column also contains a variable region which has the value 1 if a

country is part of any of the developing country regions (as notified to the WTO and

in force) and 0 otherwise. The coefficient for the variable region is insignificant. This

is not surprising because the variable contains very different regions, some that are

hardly integrated (e.g. CEMAC) and some that are more integrated (e.g. NAFTA).

If we focus on seven regions (NAFTA, MERCOSUR, CARICOM, ASEAN,

ANDEAN, SADC and COMESA) which have included some regional investment

provisions, the coefficient on region7 is significant and positive. The equation in

column II shows that the real stock of FDI is on average 68% higher if countries

become a member of one of the seven regions identified above.

One of the main motivations behind this paper is that one should not expect each

region or each country in the region to have the same capacity to attract FDI as a

result of forming a region. Regions are different with respect to trade and investment

rules and countries within regions also differ. Column III provides a breakdown by

region: relative to being outside one of the seven region, formation of some regions

22
(CARICOM, ASEAN, ANDEAN, NAFTA) attracts additional extra-regional FDI

while this is not true for some others (SADC, COMESA and MERCOSUR).

In the next columns IV and V we explore why different regions attract different

amounts of FDI. We use the indices constructed on the basis of a careful examination

of investment and trade provisions in the seven key regions (see Table 3). Column IV

shows that the coefficient on the variable measuring regional investment provisions is

positive and significant. This implies that regions with more investment provisions

provide UK and US investors with positive signals about how such regions will treat

their investors. The coefficient of 0.41 means that regions with some investment

provisions will raise their real stock of FDI by 41% and increase by a further 41%

(and 82%) if they include further investment-related provisions (i.e. a move on the

index from 1 to 2 will lead to an increase of 41% FDI)7. For instance, ASEAN would

have increased FDI by 123% on average, while COMESA only by 41% because so

far it has fewer investment-related provisions. Column V shows that similar

observations apply to trade provisions – in fact, it is hard to distinguish between trade

and investment provisions because they tend to be announced at the same time (e.g.

NAFTA) although the indices need not have the same value.

The formation of a region does not necessarily lead to an equal distribution across

countries, and some countries may achieve a higher percentage increase in the stock

of real FDI than others. Columns VI-VIII explore some underlying reasons. Column

VI is as column V but includes an interaction term between investment provisions and

the relative size of the country in the region (ratio country GDP to largest GDP in

7
Because the explanatory variable is ordinal one should be careful in interpreting the movement from 1
to 2 and 3. In reality this may go more gradually.

23
region varying between 0 and 1).8 As the coefficient is positive and significant, it

follows that the larger the country relative to others in the region, the more FDI it will

attract on the back of regional integration. This would be consistent with observing

that UK and US investors seek to invest in the largest or larger markets of the region

in order to be closest to where most of the demand is. As an example, US FDI as a

percentage of GDP has increased much more in Argentina (threefold) than in Uruguay

(twofold) after the introduction of MERCOSUR.

Column VII shows that the interaction term with relative GDP per capita in the region

is not significant. This indicates that it is not necessarily poorer countries in a region

that attract less FDI. Finally, column VIII shows that countries that are further away

in distance from the largest economy in the region attract less FDI. This is consistent

with the hypothesis that core countries would attract more FDI than periphery

countries through regionalisation (if not counteracted by other factors, e.g. direct

policy initiatives, as in the Ireland example given above).

Sensitivity analysis

We performed several sensitivity analyses. First, while the regressions in Table 4

included a fixed effect for US FDI, it might be expected that US FDI responds very

differently from UK FDI to all explanatory variables including the variables on

regional integration. Therefore, we ran separate regressions for UK FDI and US FDI

as can be seen from Table 5. We omit regressions with education or inflation, as these

did not appear to give satisfactory results. However we gained more observations.

8
Interaction terms with trade provisions yield similar results.

24
As can be seen from columns 1-2 US and UK investors behave very differently.9

Simple F-tests in a panel context confirm that coefficients on each explanatory

variable are significantly different across source country. Qualitative results are

largely the same. However, the effects of regional integration on UK FDI in one of

the 7 regions are much more equally distributed than US FDI (see the coefficient on

the interaction term), but it is not clear why this is so.

Columns 3 and 4 in table 5 also present separate regressions for UK and US FDI, but

now using a different panel estimator. Whereas previous estimations presented OLS

estimates with robust standard errors, we now present Random Effect Panel data

estimates (these are preferred to Fixed Effects Panel estimates for both the UK and

US; see the Hausman tests at the bottom of the chart). The results are similar, but the

investment provisions variable is insignificant in the US and significant in the UK.

We also explored the use of dynamic specifications (equation 3). Because there are

gaps in the data, the use of first differences does involve an unbalanced panel.

Moreover, in column I (table 6) we take the most simple equation explaining changes

in FDI by changes in host country market size and regional investment provisions in

order to get as many observations as possible. Clearly, the significance and positive

sign of regional investment provisions is robust to using a dynamic specification.

Columns II and III estimate an error correction term for the UK and US respectively.

UK FDI appears to respond particularly well and rapidly to changes in market size

(short-run coefficient is 1.33), US FDI follows market size in the long run (long-run

coefficient is approx 1.2 = 0.05/0.04) US and UK FDI grow between 4 and 11% faster

in countries that become a member of one of the seven regions.

9
Differences amongst source countries can be due to many factors including different sectors, different
hone county factors or differences in host-countries.

25
Finally, we tested for the inclusion of time dummies and other variables, such as

bilateral investment treaties, as well as other estimators such as dynamic fixed effect

panel estimators. However, the effect of the regional variables did not change

substantially.

26
Table 4 Regional Integration and the real stock of US and UK FDI in developing
countries (1980–2001)
ln (FDI) – US and UK Pooled
I II III IV V VI VII VII
Ln (GDP_host) 0.67 0.70 0.65 0.68 0.73 0.67 0.68 0.67
(21.9)** (23.1)** (17.7)** (22.7)** (23.0)** (22.3)** (22.7)* (22.4)**
Education enrolment 0.006 0.003 0.003 0.004 0.003 0.004 0.004 0.004
(4.67)** (2.49)** (2.09)** (2.97)** (2.08)** (2.85)** (2.98)* (3.10)**
Inflation 0.00 -0.00 -0.00 0.00 0.00 0.00 0.00 0.00
(0.30) (-0.33) (-0.30) (0.10) (0.20) (0.39) (0.90) (0.40)
Phonelines per 1000 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003
inhabitants (5.51)** (6.57)** (6.59)** (6.16)** (5.69)* (6.16)** (5.85)** (5.84)**
Roads 0.20 0.17 0.33 0.17 0.08 0.11 0.17 0.15
(4.58)** (3.71)** (7.06)** (3.72)** (1.42) (1.90)* (3.56)** (3.01)**
Region 0.12
(1.00)
Region7 0.68
(7.10)**
SADC -0.37
(-1.65)*
COMESA 0.35
(1.38)
CARICOM 1.31
(8.08)**
ASEAN 1.42
(13.7)**
ANDEAN 1.07
(8.10)**
NAFTA 1.48
(4.08)**
MERCOSUR -0.00
(-0.01)
Regional Investment 0.41 0.17 0.39 0.63
Provisions (6.35)** (1.93)** (4.65)* (7.55)**
Regional Trade Provisions 0.43
(8.45)**
INVPROV*GDPRATIO 0.80
(6.66)**
INVPROV*GDPpcRATIO 0.08
(0.59)
INVPROV*DISTANCE -0.0001
(-3.11)**
US fixed effect 0.60 0.63 0.60 0.62 0.63 0.63 0.62 0.61
(6.48)** (6.95)** (6.56)** (6.84)** (6.98)** (6.96) (6.81)** (6.70)**
No of observations 1521 1521 1521 1521 1521 1521 1521 1521
R-squared 0.43 0.45 0.48 0.44 0.45 0.45 0.44 0.45
Notes: robust standard errors within parentheses, constant omitted from tables
** (*) denotes 5% (10%) significance level

27
Table 5 Differences between UK and US FDI, 1980–2001
Ln (FDI)
US FDI UK FDI US FDI UK FDI

Ln (GDP_host) 0.79 0.40 0.75 0.51

(24.2)** (12.53)** (9.37)** (5.83)**
Phonelines per 1000 0.006 0.002 0.005 0.005
inhabitants (10.6)** (4.48)** (6.67)** (6.11)**
Roads 0.18 0.19 0.22 0.26
(3.20)** (3.35)** (0.74) (1.03)
Regional Investment 0.09 0.35 0.01 0.17
Provisions (0.89) (4.11)** (0.13) (3.00)**
INVPROV*GDPRATIO 1.14 0.46
(7.47)** (3.57)**
No of observations 1161 562 1161 562
No of countries 97 68 97 68
R-squared 0.52
Robust standard errors Yes Yes No No
Hausman-test (RE vs FE): 0.05 0.22
P-value
Estimation method OLS OLS RE- RE-GLS
GLS
Notes: robust standard errors within parentheses for OLS estimations
** (*) denotes 5% (10%) significance level

Table 6 Dynamic specifications for UK and US FDI 1981–2001

∆ln (FDI) –pooled ∆ Ln (UK FDI) ∆ln(US FDI)
∆Ln (GDP_host) 0.61 1.34 0.23
(2.46)** (2.65)** (0.98)
Regional 0.04 0.11 0.04
Investment (2.47)** (3.83)** (2.15)**
Provisions
Ln(GDP_host)-1 0.08 0.05**
(5.44)** (4.70)
Ln(UK FDI) -1 -0.13
(-7.28)**
Ln(US FDI) -1 -0.04
(-5.16)**
US fixed effect -0.01
(-0.31)
No of observations 2024 613 1411
R-squared 0.01 0.10 0.02
Robust standard Yes No No
errors
Estimation method OLS OLS OLS
Notes: robust standard errors within parentheses for first column
** (*) denotes 5% (10%) significance level

28
6 Conclusions

This paper examined the relationship between Regional Integration and FDI in

developing countries. A theoretical and empirical literature on Regional Integration

and FDI has begun to emerge over the past decade which appears to show a consensus

in the literature that RI leads to further (extra and to some extent intra-regional) FDI.

However, the empirical literature seems to offer little guidance on whether different

regions are more successful in attracting FDI than others, and if so, why some regions

are more successful, and crucially whether trade negotiators can design an RTA in

order to have the best possible outcome for FDI.

We argued that it is important to move beyond describing a region as a “black box”

and conduct empirical research that can help to identify the effects of specific

investment-related provisions in RTAs on Foreign Direct Investment. We illustrated

this point on the basis of empirical examples. In particular, we estimated a model

explaining the real stock of UK and US FDI in developing countries, covering 68 (for

UK FDI) and 97 (US FDI) developing countries thus moving beyond analyses on the

basis of the familiar OECD database, over a period 1980-2001, by a number of key

explanatory variables and a variable that measures the scope of regional investment

and trade provisions in key regions.

The new econometric evidence in this paper showed that i) while membership of a

region can lead to further extra regional FDI inflows, the type of region matters for

attracting FDI, i.e. whether or not regions include certain trade and investment

provisions; and ii) that the position of countries within a region matters for attracting

FDI, i.e. that smaller countries and countries located further away from the largest

29
country in the region benefit less from being part of a region than larger countries and

those close to the core of the region (although indirectly smaller countries could gain

from this). We showed that the results were robust to a number of alternative

specifications.

References

Barrell, R and N. Pain (1999) “Trade restraints and Japanese direct investment flows”,
European Economic Review, 43, pp. 29-45.
Barrell, R. and D.W. te Velde (2002) “European Integration and Manufactures Import
Demand”, German Economic Review, 3, pp. 263-293.
Blomström, M. and A. Kokko (1997) “Regional Integration and Foreign Direct
Investment, NBER Working Paper 6019.
Blonigen, B.A., R.B. Davies and K. Head (2002), “Estimating the knowledge-capital
model of the multinational enterprise: Comment”, NBER Working Paper 8929.
Brainard, S. L. (1997) “An Empirical Assessment of the proximity-Concentration
Trade-Off Between Multinational Sales and Trade”, American Economic Review,
Vol. 87, No. 4, pp. 520-544.
Brenton, P., F. Di Mauro and M. Luecke (1998) “Economic Integration and FDI: An
Empirical Analysis of Foreign Direct Investment in the EU and in Central and
Eastern Europe”, Kiel Working Paper no. 890.
Carr, D.L., J.R. Markusen and K. Maskus (2001) “Estimating the knowledge-capital
model of the multinational enterprise”, American Economic Review, 91, pp. 693-
708.
Chudnovsky, D. and A. Lopez (2001) “La inversion extranjera directa en el Mercosur:
un análisis comparativo”, in D. Chudnovsky (ed.), El Boom de Inversión Directa
en el Mercosur, Madrid: Siglo XXI Editoria Iberoamericana, pp. 1-50.
Dee. P and J. Gali (2003) “The trade and investment effects of preferential trading
arrangements”, NBER working paper 10160.
Dunning, J. (1997a) “The European Internal Market Programme and Inbound Foreign
Direct Investment”, Journal of Common Market Studies, 35, pp. 1-30.
Dunning, J. (1997b) “The European Internal Market Programme and Inbound Foreign
Direct Investment”, Journal of Common Market Studies, 35, pp. 189-223.
Dunning, J. (1993) Multinational Enterprises and the Global Economy, Addison-
Wesley Publishing Company.
Estevadeordal, A. and K. Suominen (2003) “Rules of Origin: A World Map”, Paper
presented at the seminar 'Regional Trade Agreements in Comparative Perspective:
Latin America and the Caribbean and Asia Pacific.
Estevadeordal, A. and R. Robertson (2002), “Do Preferential Trade Agreements
Matter for Trade? The FTAA and the Pattern of Trade.”, draft paper December
2002
Ethier, W. J. (1998) “Regionalism in a multilateral world”, Journal of Political
Economy, 106, pp. 1214-1245.

30
G`rg, H, and D. Greenaway (2002) “Is there a potential for increases in FDI for
Central and Eastern European countries following EU accession”, GEP Research
paper 2002//31.
Greenaway, D. and C. Milner (2002) “Regionalism and Gravity”, GEP Research
paper 2002/20.
Harrigan, J. (2001), “Specialization And The Volume Of Trade: Do The Data Obey
The Laws?, NBER Working Paper 8675.
Levy Y.E., E. Stein and C. Daude (2002) “Regional Integration and the Location of
FDI”, IADB Draft.
Markusen, J.R and A.T. Venables (1997) “The Role of Multinational Firms in the
Wage-Gap Debate”, Review of International Economics, 5(4), pp. 435-451.
Monge-Naranjo, A. (2002) “The impact of NAFTA on Foreign Direct Investment
flows in Mexico and the Excluded countries”, draft paper Northwestern University.
Neary, J.P (2001) “Foreign Direct Investment and the Single Market”, Draft paper
CEPR and University College Dublin.
Page, S. (2000) Regionalism among developing countries, MacMillan in association
with ODI, London.
Pain, N and M. Lansbury (1996) “The Impact of the Internal Market on the
Evaluation of European Direct Investment”, London NIESR, mimeo.
Pain, N. (1997), “Continental Drift: European Integration and the Location of UK
Foreign Direct Investment” The Manchester School Supplement. Vol. LXV pp. 94-
117.
Porter, M. (1998) “Clusters and the new economics of competition”, Harvard
Business Review, 76, pp. 77-90.
Schiff and Wang (2003), “Regional Integration and Technology Diffusion. The case
of the North American Free Trade Agreement”, World Bank Policy Research
Working Paper 3132.
Srinisavan, K and A. Mody (1997) “Location Determinants of Foreign direct
investment: An Empirical Analysis of US and Japanese investment” Canadian
Journal of Economics,
UNCTAD (1993) World Investment Report 1993, UNCTAD, Geneva.
UNCTAD (1996) World Investment Report 1996, UNCTAD, Geneva.
UNCTAD (2003) World Investment Report 2003, UNCTAD, Geneva.
Velde, D.W. te (2003) “Policies towards Foreign Direct Investment”, in Wignaraja, G
(ed.), Competitiveness Strategy and Industrial Performance: A Manual for Policy
Analysis, London: Routledge.
Velde, D.W. te and M. Fahnbulleh (2003) “Investment-related Provisions in Regional
Trade Agreements” background paper available from
http://www.odi.org.uk/iedg/projects/ec_prep.html.
Venables (1999) “Integration Agreements: A Force for Convergence or Divergence?”
Proceedings of World Bank ABCDE Conference, Policy Research Working Paper
Series no. 2260, World Bank.
Waldkirch (2003) “The 'New Regionalism' and Foreign Direct Investment: The Case
of Mexico, Journal of International Trade & Economic Development, 2003, 12,
pp. 151-184.
Wheeler, D. and A. Mody (1992). “International investment location decisions: the
case of US firms”, Journal of International Economics, 33, pp. 57-76.
World Trade Organisation (2003) Regional Trade Agreements Notified to the
GATT/WTO and in Force.

31
Data appendix

Foreign Direct Investment

US FDI: US direct investment position abroad on a historical cost-basis, in '000 US$, 90
countries, 1980–2001, see www.bea.gov.uk
UK FDI: Net book value of FDI by UK enterprises, '000 US$, 60 countries,
1981,1984,1986-2001, Business Monitor MA4 (and data obtained directly from CSO).
Variables are deflated by home GDP deflator from the World Development Indicators, and
are in natural logarithm form

Other raw variables

EDU sum of EDUPRIM (School enrolment, primary, % gross),
EDUSEC (School enrolment, secondary % gross) and EDUTERT (School
enrolment, tertiary, % gross)
GDP_USD Gross Domestic Product, current US$
GDPG Annual Change in Gross Domestic Product, percentage
PHONES telephone landlines, # per 1,000 population
ROADS Road network length, kilometres

Transformed variables
GDPpcRATI0 Ratio own GDP/capita to highest GDP/capita within own RTA
GDPRATIO Ratio own GDP to highest GDP within own RTA
DISTANCE Distance to largest market

∆ = change term or first difference operator

A list of countries included and details on data sources are available from the authors.

32