New exports matter: FDI, export discoveries and growth in MENA countries, Journal of International Trade and Economic Development,

New exports matter: Discoveries, FDI and growth, an empirical assessment for MENA countries

Published in The Journal of International Trade and Economic Development, 20(4), 507-533

Dalila Nicet-Chenaf, GREThA CNRS 5113, University of Bordeaux, France Eric Rougier, GREThA CNRS 5113, University of Bordeaux, France

(Corresponding Author)

Abstract Export diversification has become a priority goal for the development of the MENA countries. In this paper, we aim at measuring both the effects of exports’ diversification on growth in MENA countries and the way new exports and FDI interact with each others in the process of growth. Although the effects of FDI on growth have been scrutinized by numerous studies up to now, the effects of diversification and discoveries in export have only very recently been assessed. But no one has made explicit the way FDI and export discoveries interact in the growth process. A model is estimated by the system-GMM and we provide robust evidence that export discovery and FDI stimulate GDP growth in our sample of countries, and that FDI does not necessarily have the same effect on growth according to the level of discovery of the country. We also show that the joint positive effect of new exports and of imports suggest that technological spillover from import but also from the integration to global value chains are likely to occur in our sample of countries.

Keywords: Export diversification, FDI, Growth, MENA, GMM system

JEL classification codes: F1, O11

1

Introduction

Although the view that FDI can spur economic growth is now widely shared among

policy makers1, the tempered results of empirical studies underlie the idea that FDI effects are

not automatic and depend closely on the characteristics of each host country and on the nature

of each FDI2. Moreover, the absorptive capacities of domestic firms and receiving economies

have been shown to be preconditions for incorporating the benefits of FDI externalities, so

that the FDI impact can be non-significant, negative or positive according to the economic,

institutional and technological circumstances of the host country3. It has been successively

shown that absorption capacities are triggered by the rise in the level of education

(Borensztein et al., 1998; Lipsey, 2000), a narrowing of the gap with the technological

frontier (Lipsey, 2000; Xu, 2000; Görg & Greenaway, 2004; Li & Liu, 2005), a higher level

of financial development (Hermes & Lensink, 2003; Alfaro et al., 2004), an economy more

oriented towards exportations (Balasubramanyam et al., 1996; Bende-Nabende et al., 2000;

OECD, 2002), a bigger macroeconomic stability (Prüfer & Tondl, 2007) or better

infrastructures and institutions (Olofsdotter, 1998; Bénassy-Quéré et al., 2005; Busse &

Groizard, 2006; Prüfer & Tondl, 2007).

Yet, whereas theoretical linkages are obvious, the existing literature has ignored the role

of diversification in the FDI-growth nexus4. On the one hand, the diversification in output or

1 De Gregorio (1992) and Blomström et al. (1994) provided evidence that FDI is three times more « efficient » than local investments, notably because of its ability to stimulate the domestic accumulation of capital (crowding-in effect) and via the externalities that are related to a higher content in terms of organization and technologies (spillover effect). 2 Hence, the empirical relationship that exists between economic growth and foreign direct investment is not entirely devoid of ambiguity. Studies based on aggregate data show that FDI can have aggregate effects on growth for a developing economy, but their results remain weak and contradictory since they are very sensitive to the choice of model and data on FDI as argued by UNCTAD (1999) and Ram and Zhang (2002) 3 Beyond the absorptive capacities, the spillover effects of FDI on productivity depend on many factors such as the degree of spatial concentration of the activities, the size and the export capacity of domestic firms and the characteristics of FDI. See Moran et al. (2005) or Crespo and Fontoura (2007) for recent surveys. 4 A recent exception is Prüfer and Tondl (2007) who show with Bayesian Averaging Methods that structural change measured by the shift in the sector structure of exports has no effect on growth in Latin America, alone or together with FDI. They also confirm “the importance of certain institutional and infrastructure factors for growth” while they” have to reject the findings of other studies on the influence of macroeconomic factors and

2

trade has been linked to the level of development (Imbs & Warcziag, 2003; Klinger &

Lederman, 2004, 2006a, 2006b; Koren & Tenreyro, 2007a, 2007b) On the other hand, the

effects of trade diversification on growth also been assessed (De Pineres & Ferrantino, 2000;

Al-Marhubi, 2000; De Ferranti et al., 2002; Lederman & Mahoney, 2007; Hesse, 2007).

Moreover, the attraction of vertical or platform FDI and the integration into global value

chains is now at the very centre of the development policies of the Middle East and North

African (MENA) countries. The reason is that the imports of new intermediate inputs and the

new exports of processed products which are related to the vertical or platform investments

are expected to have favourable spillover effects on growth for the host country. But sector-

based analyses have underlined that the effectiveness of productive spillover are conditioned

by a series of factors among which the density of the links between subsidiaries of foreign

firms and local firms, be them partners or competitors5. Now the domestic firms of the

MENA countries are very often weakly integrated with foreign affiliates and the effective

outcome of vertical FDI must actually be questioned in that peculiar context.

Up till now, no one has tried to assess the way FDI and new exports interact in the process

of growth by now. Our paper is a contribution aimed at filling this gap in the empirical

literature. The purpose of this paper is to test the joint effects of new exports and FDI on GDP

growth for a sample of countries characterized by relative weakness in these dimensions.

Consequently, we assume in this paper that the features of export diversification of the

host economy are a crucial dimension of its absorptive capacities. The pioneering works by

Chenery (1979) or Syrquin (1989) showed how structural change in output is at the root of the

development process and how it also concerns trade. At an early stage, the risks towards

growth and stability of an excessive concentration of the primary products exports have been

human capital in general in LA” (Prüfer and Tondl, 2007: 24). But they say nothing about export diversification and discoveries. 5 The other factors are the degree of training and skill of the local labour, or the technological and organizational capacities of the local firms. See Moran (1998) and Moran et al. (2005) for surveys.

3

underlined by Prebish (1950) or Singer (1950). But the effect of export diversification goes

far beyond the logic of portfolio recently emphasized by Acemoglu and Zibilotti (1997) or

Kalemli-Ozcan et al. (2003) or Levchenko and Di Giovanni (2008)6. We suppose that it is

also a vector of long-term growth as it produces new complementarities between new and

former activities. Consequently, we argue that diversification can trigger the spillover effects

of FDI on growth which depend, among other things, upon the density of the production

system (Moran, 1998; Moran et al., 2005).

Another motivation of this work is that the methods used to assess the aggregate effects of

FDI on growth are generally not convenient for this purpose because the estimation is often

biased by a problem of endogeneity of explanatory variables, of which FDI can be considered

as a major one7. Borensztein et al. (1998) use an interactive term between FDI and education

in order to determine how the impact of foreign investments on growth is influenced by the

level of human capital in the economy. Using the same model, Hermes and Lensink (2003)

and Alfaro et al. (2004)8 try to assess the effects of financial liberalization on the relationship

between FDI and growth. The problem is that they all fail to address the dynamic nature of

the underlying model and the necessity to estimate it with convenient estimators. Carkovic

and Levine (2005) have recently shown that the assessment of the effect of FDI on growth is

very sensitive to the choice of estimator, particularly when the system difference-level

estimator is used instead of the OLS estimator. We carry on with their approach and we

estimate our dynamic model of growth successively with fixed and random effects estimators,

and with the Generalized Method of Moments (hereafter GMM) (Arellano & Bover, 1995;

Blundell & Bond, 1998).

6 They modernize the ideas of the pioneers and show that export diversification helps to limit the effects of the deterioration of exchange terms on the trade-generated incomes in the framework of the portfolio theory. 7 A small number of surveys have taken these difficulties into account. They propose to analyse these relations within the framework of a simultaneous equation model (Bende-Nabende et al., 2000; Li & Liu 2004) or through procedures of Bayesian Averaging (Prüfer & Tondl, 2007). 8 Starting from two different models, Hermes and Lensink (2003) and Alfaro et al. (2004) end up with an equivalent econometric specification.

4

In this econometric framework, we look at the joint role of FDI and diversification

towards economic growth. We start from a specification similar to Carkovic and Levine

(2005), but we incorporate new variables that were up to now seldom used in the aggregate

analysis, such as the degrees of diversification or of concentration of exports, and the number

of « discoveries » in new exports.

Our sample focus on the Middle East and North African countries (hereafter MENA

countries) which after many years of a relative closing on foreign investments, have moved

towards more active strategies of attracting FDI since the end of the eighties9. Middle East oil

producers have been excluded from the sample because the high level of concentration of

their export structure upon extractive industries does not produce the incitements to

diversify10.

After a brief overview of the links between FDI, growth and diversification in MENA

countries (section 1), we discuss the notions of trade diversification and export discovery and

specify the way they interact with FDI in the process of growth (section 2). Model, estimation

methods and data are described in section 3, and the section 4 presents the empirical results

and the comments. Then, section 5 concludes.

1. FDI, diversification and growth in MENA countries: failed expectations

At the end of the nineties, the MENA region under-achievement in terms of FDI attraction

started to be highlighted by different studies. Petri (1997) underlines the poor performance of

the countries of the region in attracting FDI by comparing it to the higher performances of

countries with similar « fundamentals ». During the nineties, FDI represented an average of

0,9% of the GDP in MENA countries, against 2,5% in African countries, 3,8% in Eastern

9 These strategies were reaffirmed during the nineties while an increasing number of empirical investigations showed that FDI could have beneficial effects on the growth of developing countries. 10 This concern has recently been expressed by some of these oil producing countries but it’s too early to get significant trends in the past data.

5

Asia and 4,5% in Latin America (Sekkat, 2004). A few years later, and despite a fast increase

of the FDI inflows to Tunisia, Morocco and Egypt, such a weakness in attracting investment

was still underlined by the studies of Iqbal and Nabli (2004), Chan and Gemayel (2004),

Sekkat (2004) or Daniele and Marani (2006). In addition, Noland and Pack (2007) or Iqbal

and Nabli (2004) also showed that the degree of integration to the global production chains is

very limited in spite of the closeness to the European market. Moreover, Haddad and Harrison

(1993) then Harrison (1996) found very few empirical evidence of the existence of spillover

towards local firms, even though the joint-ventures in Morocco are on average more

productive than local firms11.

Cross-sectional studies have not produced particularly consistent evidence about what

explain both the poor performance of MENA countries in attracting FDI and the weakness of

the spillover effect from FDI. Furthermore, the weakness of the legal and administrative

environment in the MENA countries has been underlined as a major hinderance to growth by

the recent surveys of Alessandrini (2000), Daniele and Marani (2006), Chan and Gemayel

(2004), Bénassy-Quéré et al. (2005) or Sekkat (2004) that all underline the necessity of

furthering the reforms in that field.

A few empirical studies have tried to explain why FDI does not spur growth significantly

in MENA countries. The limited capacities of absorption of MENA countries compared to

other developing countries are put forward to explain the weak effects of FDI on growth by

Sekkat (2004) or Elmawazini (2007). As regards openness to trade and to foreign investment,

it appears in the literature that the rise in the integration of the MENA countries into the world

economy has not generated much effect on their growth up to now. Trade openness is then

11 Harrison (1996) even suggests that in Morocco, FDI effects on productivity might have been negative in the short term because of the consequences in terms of production scale of the loss of local market shares for domestic firms. More recently, Bouoiyour and Akhawayn (2005) have shown on a panel of Moroccan industries that FDI has significant spillover effects on the productivity of labour. Furthermore, they give evidence that these effects are proportional to the technological gap between foreign subsidiaries and local firms and increase together with the openness of the sector to exportations.

6

positively associated with FDI inflows in MENA countries, but it does not contribute

significantly to produce spillover effects towards productivity and growth (Haddad &

Harrison, 1993; Harrison, 1996; Sadik & Bolbol, 2001).

According to the calculation by Sadik and Bolbol (2001), inflows of FDI in MENA

countries actually affect growth through the accumulation of capital and not through

aggregate productivity gains12. FDI received by Egypt, Jordan and Tunisia had positive effect

on productivity for specific activities only (energy and textile for Tunisia, energy and services

for Jordan, and sectors that are highly protected against competition in Egypt) with arguably

limited influence of technological transfers13.

If FDI can spawn spillover effects for specific sectors only, developing such sectors is a

condition for a higher return from inflows of foreign investment in terms of productivity.

Consequently, if the structures of output and trade probably matter for FDI spillover to occur,

trade openness is not a convenient measurement of the way integration to the world economy

can trigger growth. Essentially, what is at stake is the ability for growth is to discover new

exports and to diversify the structure of production.

Yet, an unpublished survey of trade diversification for five MENA countries [Egypt,

Lebanon, Jordan, Morocco and Tunisia] from the World Bank has recently put forward the

small progress of these countries towards the diversification of their production and export

during the last twenty years. Furthermore, their exports are generally characterized by a high

concentration since the four biggest export sectors represent 75% of the exports for Egypt,

Tunisia, Jordan and Morocco, against 57% for South-East Asia and 49% for Eastern Europe

countries. Their export structure is also very dependent on natural resources – agriculture and

12 FDI does not give a significant explanation to the growth in Morocco, Oman and Saudi Arabia. Sadik and Botbol (2001) explain this result by the presence of internal factors that are not controlled for in regressions (for Morocco, influence of climatic hazards on harvests, for Oman and Saudi Arabia, influence of the oil prices). 13 Sadik and Bolbol (2001) point for example that the progress in efficiency recorded in Tunisia during the eighties is rather linked to the intensification of competition associated with the presence of foreign firms than to real transfers of an advanced technology.

7

food (Morocco, Jordan), oil and gas (Tunisia, Morocco, Jordan), fertilizers (Jordan, Morocco)

or low skilled works such as textile (Tunisia, Morocco, Jordan) – and the relative weight of

their medium or high technology exports remains very modest – 21.2% on average in the five

countries against 55% for the new European countries and South-East Asia. World Bank

(2006: 26) underline for example that the Morocco’s export discovery levels are below those

related to the same level of income per capita, and also below the levels of its competitors

(China, Romania, Turkey). Their diagnostic is that the very low levels of private investment

that limits the growth of GDP for Morocco are due to a series of market and policy

distortions14 that reduce the incitement to innovate and self-discover for entrepreneurs. The

statistical study shows in particular that diversification and competitiveness of exports are too

weak, and thus binding for growth and structural transformation. Their conclusion is that an

acceleration of growth could be drawn by exportations and their diversification, and that the

whole economic policy of Morocco must be oriented towards the incitement to discover new

tradable products. The need to enter into a strategy of export diversification towards services

and new dynamic activities has been underlined for Morocco and for the other MENA. Their

ability to go over from traditional exportations to exportations with a higher value added (Lall

et al., 2005) or higher productivity content (Hausmann et al., 2007) is presented as a key for

economic growth (World Bank, 2006: 63)15.

What appears is that low degrees of FDI attraction and its weak impact on growth as well

as a low level of diversification are key features of the MENA countries. Therefore, reaping

positive rewards of FDI likely depends on what the country exports and on its capacity to 14 The first ones refer to the stiffness of the work regulations and its high cost, a too heavy taxation that burdens the firms’ profits and the income of skilled workers, and the non-adaptability of both the commercial system (anti-export bias joint to a very high level of protection against importations) and the exchange rate system (fixed rate and risk of over-valuation). 15 This ability was arguably at the very heart of the strategies of export incentives applied by Asian economies. The examples of Taiwan, South Korea or Chile are invoked to underline the strategic importance of « fundamentals »: a stable macroeconomic environment, pro-market policies and an active industrial policy for the sector-based incitement of exportations and the mobilization of savings and investments towards these sectors. The difference is that it does not necessary mean industrialization today. On that point, see Hausmann et al. (2007).

8

renew its export beyond traditional comparative advantages. Simultaneously, FDI could

arguably be a key factor in the way diversification or discoveries actually impact growth. In

the following section, we examine the effect of export diversification and discoveries on

growth, and we show how FDI can have an effect on it.

2. Discoveries, FDI and growth: how do they interact with each other?

A number of theoretical works have provided explanations about how diversification

towards much more complex goods can stimulate aggregate productivity and growth16. They

provide useful guidelines to assess the way FDI impact on the diversification-growth

relationship. From the seminal works by Prebish (1950) and Singer (1950) to the more recent

papers by Bertinelli et al. (2006) or Levchenko and Di Giovanni (2008), diversification of

exports has firstly been perceived as a way to stabilize the revenues from export in the long

run in front of highly elastic demands and very volatile market prices17. The diversification of

export is usually seen as an endogenous process whose driving force is the decision taken by

producers to invest in diversified activities in order to stand on the optimal risk-returns

frontier (Acemoglu & Zibilotti, 1997; Kalemli-Ozcan et al., 2003). But diversification then

looks like an optimal strategy open to countries that have both a disposable capital to invest

and adequate opportunities to invest it (Koren & Tenreyro, 2007b)18. In a general setting, FDI

could thus favour both diversification and growth by increasing at the same time the quantity

of capital available for investments – as long as the crowding effects on domestic investments

16 The effect of diversification on growth has been recently examined by a few empirical works. De Pineres and Ferrantino (2000), Al-Marhubi (2000), De Ferranti et al. (2002), Lederman and Mahoney (2007) or Hesse (2007) give evidence that, generally speaking, the diversification of exports has a robust positive effect on the increase of GDP per capita. Herzer et al. (2006) reach the same result for the Chile case. 17 This volatility in export prices and volumes has been aggravated by the attendance of China to the world markets, producing in the same time more competition in terms of prices and volumes on the textile markets, and huge price movements on the market of raw materials, creating instability for income from trade and for growth in several developing countries (Kaplinsky, 2006). 18 They argue that consequently, the poorest countries with low savings and investment should specialize in a small number of low-risk sectors in order to stand at their optimum.

9

remain limited – and the opportunities for investment through the backward and forward

linkages and via the imitation likely to come together with the settlement of foreign firms.

As trade diversification eases the stabilization of the revenues from export, it

contributes to long run growth while it enables firms to plan investment, preserve an import

capacity and create new tradable activities. But exporting firms have also higher productivity

levels because they generally use technologies that are more advanced and they use their

resources in a more efficient way (Melitz, 2003)19. Moreover, the shifts in the structure of

exports from lower productivity to goods associated with a higher productivity and a more

dynamic demand enable resources to be used more efficiently and consequently generate

higher economic growth (Hausmann et al., 2007)20.

Vertical FDI may directly contribute to these shifts as they introduce the production of

new goods with generally very elastic demand in export markets in countries that generally

did not produce them before21. The notion of discovery is helpful to understand what is at

stake with the diversification of exports in a developing economy. Talking about discoveries

means that the shifts towards new activities and exports are not risk-less and that firms need

incentives to enter in new markets or to innovate. Hausmann & Rodrik (2003) have shown

that the key for structural change in a developing economy is the creation of incentives to

disclose the production costs of new activities so that this knowledge can give rise to

additional private investments in these fields. An increase in the number of discoveries22

unveils the information about the production costs of a larger variety of products and prompts

to make private investments that go in the direction of a larger diversification of production

and export. 19 They also have lower costs because they take advantage of the scale effects of the global market. 20 For Hausmann et al. (2007), the higher productivity goods are those that are produced and exported by the countries with a higher income per capita. 21 Ultimately, Hausmann et al. (2007) point that the existence of externalities associated with the process of export discovery draws paths of cumulative comparative advantage that are not entirely determined by endowments. 22 They define « discoveries » as products whose export value progressed from less than 10.000 USD in 1993 to over 1 million USD between 2000 and 2002.

10

The point is that in developing countries, export diversification actually means new

products are produced and exported (Carrère et al., 2007) and these new products or exports

often call for new production techniques. The new knowledge attached to new exports can be

seen as a non-rival asset that can be spread without limitation towards former or newer

industries and feed the productivity gains (De Pineres & Ferrantino 2000; Feenstra & Kee

2004). Moreover, exports with higher technology content generally ease the shift towards new

specialization throughout the product space (Hausmann & Klinger, 2007)23, and are

significantly correlated with higher future growth as evidenced by Hausmann et al. (2007).

This idea is not new actually. While Young (1991) showed that a higher learning by

doing’s content of export contributes to a higher productivity increase, Stokey (1991) added

that higher investment in human capital could allow for the production of more sophisticated

goods, and consequently spur growth. It has also been argued by Chuang (1997) that learning

triggers the production and export of new and refined goods, and that it generates a demand

for technology that spurs the imports of even more sophisticated goods24. The result is a

process of technological upgrading and diffusion through the diversification of the structure

of exports and imports that ends up in a sustained growth.

In that context, vertical FDI might contribute directly to the upgrading of exports and

to the effect of trade on growth. But as openness to FDI might also raise the returns from

investing in human capital because of the technological spillover effects, it might increase the

quality of goods that are produced and exported and the rate of growth of income per capita as

well. Furthermore, FDI also generates the risky direct investment and the knowledge about 23 Hausmann and Klinger (2007) have argued the thickening of the product space creates new opportunities for discoveries or changes of specialization by shifting from one activity to another one, i.e. by moving the factors of production towards the new activities that are close to the ones that were already produced by the economy in terms of complementary factors. It also helps to reduce the cost of export discoveries since close assets need similar combinations of the private and public capital that are available in the economy (Hausmann & Klinger, 2006). As it provides the developing economy with new skills and other production factors that enable the product space to be denser near its productive capacities, FDI might enhance the ability of the host economy to innovate and diversify its structure of production and export so as to sustain growth in the long run (Hausmann & Klinger, 2007). 24 See An and Iyigun (2004) and Hausmann et al. (2007) for recent empirical evidence.

11

the costs and benefits of a given activity that the host economy’s actors need to get

information on « what the country is good at producing » (Hausmann & Rodrik, 2003: 606).

Indeed, FDI can be a vector of knowledge about the costs of a new activity in countries where

entrepreneurs are too sparse and risks associated to innovation are too high because of poor

institutions. The knowledge of private costs for a new activity generates externalities that are

a potential source of increasing yields for other entrepreneurs, as long as the public incentives

allow the innovators to seize a part of the social profit of their discovery (Hausmann &

Rodrik, 2003). But it can happen that this effect is limited if the economy is poorly diversified

and offers low opportunities for investments. This is especially true when FDI is concentrated

on primary extraction activities, or when it does not create linkages towards the domestic

firms.

Then, FDI is supposed to play a key role in the relationships between diversification,

discoveries and growth since it introduces both new equipment goods25 and the backward and

forward linkages that allow for a denser product space and ease the diversification towards the

new exports that will generate a further increase of GDP. Lastly, rather than a formalized

technology, FDI and intermediate goods involve a tacit technology that must be adapted by

the receiving environment (Nelson, 2000; Evenson & Westphal, 1995). The discovery process

can then be the unexpected result of the adaptation of the imported technologies associated

with FDI to the local conditions.

On our side, we assume that a diversified economy offers a larger variety of

complementary factors (Hausmann & Rodrik, 2003; Hausmann & Klinger, 2007) that lowers

the introduction costs of new technologies involved in the intermediate goods linked to FDI as

well as it increases the productivity of these semi-finished goods. Thus, a superior degree of

25 It has been evidenced by Borensztein et al. (1998) and Hermes and Lensink (2003) who show that in a model of endogenous growth à la Romer (1990), FDI introduces new varieties of intermediate goods in an economy and consequently increases growth if human capital (Borensztein et al., 1998) or financial development (Hermes & Lensink, 2003) are sufficient to reduce the introduction costs of new technologies and increase the yields of the new equipments.

12

diversification in the production and exportations or a rise in export discoveries can spur the

spillover effects of FDI between firms and industries. The dual question is: does FDI

contribute more to growth when the host economy is more diversified and provides a denser

product space with a higher variety of complementary factors, or when it is poorer and less

diversified? That is what we want to test in the next section for a sample of MENA countries.

3. Model, data and estimators

In this section, we carry on with the empirical model of Borensztein et al. (1998) or

Hermes and Lensink (2003), whereby the GDP growth per capita is explained by the stock of

human capital, the initial level of income per capita, the foreign direct investment and all

other variables that usually influence growth, some of them through the channel of FDI26. In

Borensztein et al. (1998), an interactive term between FDI and schooling helps to determine

the way the impact of foreign investments on growth is influenced by the level of education in

the economy. Using the same type of interactive variables, Hermes and Lensink (2003) or

Alfaro et al. (2004)27 gauge the effects of financial liberalization on the relationship between

FDI and growth. The problem is that they all fail to address the dynamic nature of the

underlying growth model and the necessity to estimate it with the GMM estimators (Blundell

& Bond, 1998) as shown by Carkovic and Levine (2005).

We consider the following regression equation:

Log (Yit / Yit-1) = α LogYit-1 + Xit Ψ + β FDIit + Zit Π + fi + εit (1)

where Yit is the GDP per capita for the country i at time t, Xjt and FDIit denote respectively

the number of export discoveries and FDI, Zjt stands for the set of control variables at the time

t for the country i, and fi and εit represent respectively the country fixed effect and the

26 Public spending, black market premium, political instability, political rights, financial development, inflation rate, quality of the institutions. See Barro and Sala-I-Martin (1995 : Chap 12). 27 Starting from two different models, Hermes and Lensink (2003) and Alfaro et al. (2004) end up with an equivalent econometric specification.

13

specification error. As α is the speed of convergence, the equation (1) is equivalent to the

equation (2) with θ = (α +1):

Log Yit = θ LogYit-1 + Xit Ψ + β FDIit + Zit Π + fi + εit (2)

But estimating the equation (1) or (2) for a panel of countries with specific individual

effects causes problems of correlation between the lagged endogenous term and the

specification error term (Holtz et al., 1988; Arellano & Bond, 1991). A solution is to use the

GMM to control for endogeneity and to get convergent estimators. According to Arellano and

Bond (1991), it first consists in getting the first-order difference of equation (2) so as to

remove the fixed effect as in equation (3) below28:

ΔLog Yit = Growthit = θ ΔLogYit-1 + ΔXit Ψ + β ΔFDIit + ΔZit Π + Δεit (3)

By construction, the difference in error term (εit – εit-1) is correlated with the difference in

the log of GDP per capita (Yit-1 – Yit-2) in equation (3). The next step consists in using

instruments for T ≥ 2. In generalizing the GMM, Arellano and Bond (1991) suggests to

instrument (Yit-1 – Yit-2) by all available lags on the lagged endogenous variable in level, and to

instrument (Xit-1 – Xit-2) and (Zit-1 – Zit-2) by their value in level lagged by one period or more.

The Sargan test is subsequently used to assess the validity of the instruments. However,

according to Blundell and Bond (1998), when the dependent variable and the explanatory

variable are continuous, the lagged levels of the variables are not reliable instruments for the

first-order difference equation (3) and the GMM-system must be used. This method consists

in stacking the model in difference with the model in level. From then on, we add up the

instruments for regressions in level that are the lagged differences of the related variables. We

thus use the exogenous variables of the (yit-2, yit-3,…,y it-n), (xit-1, xit-2,…, it-n) and (it-1, Zit-2,…, Z

28 The term ΔX denotes the first-order difference of X

14

it-n ) types as the instruments for equations in first-order difference while the variables in

difference Δyit-1, Δxit-1 and ΔZit-2 are the instruments of the equations in level29.

The baseline equation estimated is equation (2). In this equation, the log of income per

capita is regressed on explaining variables, including the lagged value of that income per

capita. In equation (3), the variation of the log of income per capita is regressed on explaining

variables, including the lagged value of the variation of the income per capita. The variation

in the growth of GDP per capita is explained by FDI, export discoveries and a set Z of control

variables. This set includes the variables that control for initial conditions in the augmented

standard model of growth (Durlauf et al., 2005) together with a set of standard policy or

international integration variables. A growing number of papers test for the robustness of an

ever larger series of explanatory variables of growth and give evidence that some of them

remain robust in explaining growth whatever the form of the model (Sala-i-Martin et al.,

2004; Ciccone & Jarocinski, 2007, Prüfer & Tondl, 2007). Moreover, Ciccone and Jarocinski

(2007) have shown that the growth regression results are very sensitive to the source of the

data and that theoretical priors are necessary to avoid false interpretation of the results. We

have selected variables that are consistent with the underlying model of growth that we have

used to understand the linkages between diversification, FDI and growth in the previous

section30.

Accumulation of physical capital is measured by the investment rate (Investment), human

capital is measured by the high school’s registration rate (Schooling)31, and labour by the

growth of the working aged population (Labour). The state of transport infrastructure

29 These instruments are valid only under the assumption of a non correlation between exogenous variables and non observed individual effects E(xit,fi) = 0.

30 Data sources and descriptive statistics are reported in Appendix 1. 31 The data on the number of years at high school from Barro and Lee (2000) are not available on an annual basis and are incomplete for several countries of our sample.

15

(Infrastructure), the degree of development of communication networks (Communication)32,

the degree of development of the banking system approximated by the ratio of the domestic

credit supplied by the bank sector on GDP (Credit), and trade openness as measured by

exports and imports of assets and services (Export and Import) are included as policy

variables33. Under certain circumstances, we will see that Import can be considered as a proxy

for technological transfer from trade and integration to the global value chains (Coe et al.,

1997). The FDI variable (FDI) measures the net inflows of FDI in percentage of the GDP.

We also introduce the following controls for the effect of the quality of macroeconomic

policies on growth: the annual inflation rate (Inflation), the weight of public expenditures (in

% of GDP) (Government Spending) and the growth of M2/GDP (M2). The flow of public

international aid (Aid) is also included as a proxy for the expenses in investment promotion

(Harding & Javorcik, 2007).

Diversification is traditionally measured by a Gini or a Herfindhal index computed on the

distribution of exports by sectors but recent studies have used an index of discovery which is

a counting of the new exports. But these notions are not similar as diversification can remain

even in the presence of discoveries if former exports disappear while discoveries happen34.

Following Imbs and Warcziag (2003), but with a focus on export structure, a series of

empirical analyses (Klinger & Lederman, 2004; Hausmann et al., 2007, and Carrère et al.,

2007) has shown recently that the relationship between development and export

32 The infrastructure indicator is measured by the size of the roads network (measured by the number of tarred roads in percentage of the total) and the quality of the electricity network (given by the losses on the electrical network). The more the indicator's value is close to 1, the more the infrastructures are developed. The communication indicator is measured by the number of telephones per 1.000 inhabitants, the number of personal computers per 1.000 inhabitants and the number of persons equipped with Internet. Data is from the World Bank. 33 As in Romer (1986, 1990) and Lucas (1988), endogenous growth models show that firms can benefit from technological externalities produced by the diffusion of knowledge between firms and activities. From this perspective, anything allowing a better circulation of the information (infrastructure, financial development) and easing technological transfers (trade openness, FDI) can be considered as deciding factors for growth and for discoveries and diversification as well. 34 For a discussion of that point and on the statistical accuracy of the notion of discoveries, see Carrère et al. (2007).

16

diversification is non-linear and hump-shaped35, whatever the metric for diversification that is

used.

The number of discoveries, the diversification index and the concentration index are from

the UNCTAD database. The number of discoveries is measured by the absolute annual

growth in the number of exported products (in absolute value) at the three digits level of the

CTCI-3. However, only the products having a value higher that 100.000 USD or accounting

for more than 0.3% of the country's total exports are included (Discovery). The diversification

index is a variant of Finger-Kreinin's indicator on the similarity of the trade structure

(Appendix 1), whose value lies between 0 and 1. This index gives the measurement of the

way a country's structure of exports differs from the worldwide one. The closer to 1 the index

is, the stronger the divergence is. The concentration index of a country's exports set is

measured by the Herfindahl-Hirschmann's index whose value is between 0 and 1 (see

Appendix 2). The closer to 1 is the index, the stronger is the integration.

The figure (1) exhibits the evolution of the diversification and concentration index and of

discoveries for the countries of the sample. Export diversification, concentration and

discoveries present somewhat flat patterns of evolution for the majority of the countries of the

sample. Figure 1.c shows that the pace of export discoveries is different among the countries

of the sample, even if it does not significantly accelerate during the period36.

Figure 1. Diversification (1.a), concentration (1.b) of exports, and discovery (1.c) for eight MENA countries: 1994-2005

35 Hausmann and Rodrik (2003) show that economic development eases export discoveries up to the low levels of average incomes (income per capita between 4.200 and 5.500 USD), before the relationship becomes adverse between discoveries and further development. The empirical findings of Carrère et al. (2007) support this Hump-shaped pattern. Using both a standard Herfindhal diversification index and a “discovery” measurement of the changes in the export structure, they confirm that pattern. 36 This difference probably explains why the variance in discoveries contributes to explain significantly the variance in growth rates, whereas the variance of the diversification or the concentration of export fails to do it in the results of the forthcoming section.

17

Interactive terms are further introduced in the models, defined by the amount of FDI

inflows times either the discovery variable (Disco*FDI), the integration variable (Con*FDI)

or the diversification variable (Diver*FDI).

Our sample consists in eight MENA countries over the period 1995-2005 (Algeria, Egypt,

Israel, Jordan, Lebanon, Morocco, Tunisia, and Turkey). They have been chosen according to

the availability of annual data.

4. Results, comments and sensitivity analysis

Results for the four models estimated are reported in table 1. A first estimation in level

(column 1) includes individual fixed effects that control for the heterogeneity of the sample

and the omission of explaining variables37. The model is also estimated in level with random

effects (column 2). The Fisher test (F(7, 64) = 9.63) indicates that the specific individual

effects are significant, while the Hausman test suggests that the fixed effects model must be

preferred to the random effects model (chi2(16) =65.08). But these models do not correct

neither for the endogeneity of the right hand side variables, nor for the potential correlation

between the regressors and the specific individual effects. From then on, estimation by the

GMM-difference (column 3) is computed in order to control for the above bias, and GMM-

system estimation (column 4) is used to reintroduce the fixed effects. The variables that are

systematically instrumented are the lagged GDP per capita, FDI, investment and exports38.

Both the GMM models give suitable results from an econometric point of view. The validity

of the instruments is confirmed in both cases by the Sargan/Hansen test39. Ultimately, the

37 The Fisher Fk, N(t-1) - k, statistics reported at the head of the column (1) rejects the null hypothesis that all the β coefficients estimated equal zero. 38 Endogenous variables were determined through the following test: each variable (FDI, investment then export) is regressed against all other explanatory variables, the residues are retrieved and the growth equation is regressed by adding each of the estimated residual terms separately. If the residues are significant, then the variables are considered as endogenous. 39 The χ2(46)=57.25 and χ2(51)=39.6 statistics are in both models inferior to the fractile of the χ2 law at respectively 46 and 51 degrees of freedom

18

Arellano and Bond’s test (Arellano & Bond, 1991) indicates40 the absence of autocorrelation

for the residuals εit from the models in GMM since the z statistic calculated is inferior to the

1.64 threshold.

Table 1: Dependent variable: annual GDP growth for 1995-2005. Fixed Effects (FE), Random effects (RE), Difference-GMM and system-GMM estimations

Observations : 88 Groups : 8

(1) F(16, 64) = 59.55

Prob>F=0.00

(2) Wald chi2(16) =

51921.39 Prob > chi2 = 0.00

(3) Wald chi2(16) =

978.93 Prob > chi2 = 0.00

(4) Wald chi2(16) =

3899.97 Prob > chi2 = 0.00

GDPit-1 .1876** (1.87)

.9401*** (6.87)

.0.1322 (1.36)

.0.9815*** (22.28)

School .006*** (3.68)

.0002 (0.40)

.0059*** (4.00)

-0.0007 (-0.66)

Investment 0.0615* (1.74)

.0618*** (2.72)

.0112* (1.75)

.0346** (1.98)

Labour 9.91E-09** (2.08)

1.05E-08*** (3.67)

1.66E-08*** (3.39)

5.22E-09* (1.71)

FDI .0014** (2.04)

.001** (2.40)

.0022** (2.07)

.0026** (2.29)

Discoveries .0002* (2.88)

.0002** (2.22)

.00003 ** (2.16)

.00007* (1.78)

Disco*FDI -1.95E-06** (-1.94)

-7.00E-09*** (-2.31)

-1.23E-08** (-2.52)

-2.25E-08*** (-2.34)

Export .1215* (2.83)

.0099 (0.54)

0.0591 (1.22)

.0194 (0.45)

Import 2.19E-9* (2.12)

1.38E-12 (1.57)

.147*** (3.03)

3.95E-12*** (3.38)

Inflation -.0003** (-1.95)

0.00004 -0.40)

-.00027** (-2.31)

-.00028* (-1.80)

M2 -.00089* (-1.14)

-.00052 (-1.70)

.00085 (-1.55)

-.0011** (2.06)

Government Spendings

-.0110 (-4.07)

-1.19E-08* (-1.72))

-.0095 (-3.58)

.0046 (-1.29)

Infrastructure .3187*** 5.62)

.0027 (0.06)

.2037*** (2.99)

.0638 (0.65)

Credit .00037** (1.96)

.00025 (0.92)

.00037 (0.85)

.0012** (0.044)

Aid -.0003 (-.09)

.0003 (.0.48)

.00005 (-0.12)

.0003 (0.39)

Communication .0712** (2.23)

.0393 (1.04)

0.0274 (0.97)

0.009 (0.17)

Constant 1.6108 (1.60)

-1.1130** (-2.36)

-.0592 (1.08)

-.8634 (-1.00)

Fisher test for fixed F(7, 64) = 9.63 40 The z statistic is asymptotically normally distributed and the null hypothesis of no second order autocorrelation is rejected if this statistic is superior to 1,64 in absolute value. Otherwise, we accept the hypothesis.

19

effects Hausman test for RE against FE

Prob > F = 0.00

chi2(16) =65.08 Prob>chi2 = 0.0000

J test (Sargan) for overidentifying restrictions41

chi2(46) = 57.25 chi2(51) =39.6

AR(2) test for zero autocorrelation in first–difference errors

Z=0.212

Z=0.8767

Note: (***) significant at the 1% level; (**) significant at the 5% level; (*) significant at the 8% level

For the four models reported in the Table 1, the coefficient estimated for the core

variables of the Solow augmented model (X set and lagged GDP) are of the expected sign and

highly significant (Initial GDP, Investment, Labour), even the Schooling variable but for

equation 1 and 2 only 42. Note that the coefficient for global convergence is computed as (θ-1)

= α is significant and negative for all models43 (the value is -0.4809 for the system-GMM

model).

Higher levels of FDI and discoveries are generally associated with faster growth of GDP

per capita. Note that neither the concentration nor the diversification indexes are significant44.

Since the coefficient for discoveries is significant and positive, it suggests that new exports

enhance growth for MENA countries. This positive effect can occur through the stabilization

of the growth paths or from export resources as highlighted by the export portfolio models

(Bertinelli et al. 2006; Levchenko & Di Giovanni 2008). But it can also stem from an increase

of the aggregate productivity of the economy due to the presence of technological spillovers

41 Under the null hypothesis that all instruments are exogenous, J is distributed as a chi-square with m-r degrees of liberty is the number of instruments minus the number of endogenous variables. 42 As a general rule, econometric surveys conclude that this variable is not significant when it is measured by the high school enrolment rate. It is however difficult to obtain better data on a yearly basis. 43 In the four models, it is possible to conclude on the existence of a global convergence between countries. In order to do this, it is necessary to recalculate (θ-1) = α, as well as the student-test associated with the coefficient (tθ = (θ -1)/( θ’ standard error)). Then we observed the coefficients are in all cases significant and negative. It indicates a global convergence for the countries. For the FE model, tα = (0.1876325- 1)/0.10009879 = -8.11, the RE model tα = (0.9401736-1)/0.0254982 = -2.35, the GMM model tα = (0.1322084-1)/0.0971758 = -10.15, and for the GMM-system model tα = (0.0981537-1)/0.0440606 = -2.27. 44 We only report results for discoveries in the Table 1.

20

from the new export to other activities. The positive sign of FDI can be seen as the

measurement of the net effect on growth, beyond the crowding-in effects on domestic

investment.

Table 1 gives also strong evidence that the interactive variable « Deco*FDI » has always a

significant and negative sign45. This means that the higher the number of discoveries is, the

less income growth is responsive to FDI, or alternatively that more FDI inflows tend to

disconnect growth from the number of discoveries. There are two explanations in the case of

MENA countries. The first one is common in the studies concerning the MENA countries.

The FDI received by these countries is alleged to generate very limited spillover effects

because they are either isolated from the domestic productive network, as it is the case for

investments in raw materials, or oriented towards the domestic market with a weak integration

to the global value chains, as it is the case for the mergers and acquisition operations due to

privatizations (Sadik & Bolbol 2001). Vertical or platform investments in Special Economic

Zones do not necessarily produce spillover if they are isolated from the domestic production

system of the host economy (Moran et al, 1998)46. Another explanation is that more

diversified economies enjoy larger benefits from FDI because their domestic market is larger

and the domestic demand is more diversified.

However, the significant and positive coefficient for import47 could be seen as a first

element of answer. As the share of imports and FDI from countries closer to the technological

frontier was high for the countries of our sample at the beginning of the time period48, this

result suggests that technological spillovers could have occurred from imports of production 45 When the model is only estimated with the FDI and discovery variables, the FDI coefficient happen to be significant but negative, indicating a negative role of FDI in the growth process. As soon as the FDI variable is combined with the interactive variable (decofdi), the FDI coefficient becomes positive again while the coefficient of the interactive variables is negative. 46 Prüfer and Tondl (2007) have recently shown that vertical FDI does not produce any gain in productivity growth in Latin America whereas horizontal investment has more beneficial effects on host economies. 47 Even if Export is not significant in the estimations, the discoveries are positive and significant factors of growth. 48 Around 65% of the imports of the 8 countries of our sample came from the USA, Canada, UE and Japan in 1995 (Author’s calculation from the Chelem trade database, CEPII)

21

and consumption goods (Coe & Helpman, 1995; Coe et al., 1997) for the countries and period

of our sample. But the correlation between overall imports and specific imports of

intermediate goods is large (0,863), so that it also gives credit to the idea that the integration

of MENA countries into global value chains through FDI and new exports could have started

to significantly impact growth, albeit this result is contrary to former studies (World Bank

2007; Noland & Pack 2007; Iqbal & Nabli 2004). But the result that new exports and imports

of intermediate goods are simultaneously significant in spurring growth for the countries

under study provides some support for this idea and for the argument of Melitz (2005) that

LDCs mostly receive vertical FDI generating imports of intermediate inputs and the export of

new processed products.

In order to explain the critics of Carkovic and Levine (2005) against the results of

Borensztein et al. (1998), Melitz (2005) remarks that controlling for trade openness necessary

lessen the impact of foreign investments on growth for LDCs as they mostly receive vertical

FDI that are strongly dependant upon low trade barriers for the imports of intermediate inputs

and the export of processed products. We show that controlling for discovery in export

provides interesting result regarding the way FDI affect growth. This point is of special

relevance here as soon as a few MENA countries, including the biggest receivers, target this

kind of foreign investments by settling special economic zones dedicated to exports of goods

they did not produce or trade before.

Results reported in table 1 also show that macroeconomic stability (Government

Spending, Inflation) is significant in explaining growth for MENA countries. Extensive

monetary policy and inflation have a negative influence on growth as it hampers the

competitiveness of exported goods and then reduces the growth rate. As regards fiscal policy,

its effect on growth is mixed as expected. The coefficient for Government spendings is non

22

significant in most of the estimations, but Infrastructure, largely public investment-led,

positively affects growth in two models (FE and GGM models). The development of the

banking system has also a positive influence for FE and GGM-system models in so far as it

enables a larger financing of the investment projects. This way, we meet up with the results of

the surveys indicating that the development of the banking and financing system plays a

significant role in the way FDI affects growth (Hermes & Lensink 2003, Alfaro et al., 2004).

Similar results can be found in the study by Prüfer and Tondl (2007) which uses a Bayesian

Averaging Method and insists on the key role of infrastructure and policies that reduce

structural uncertainty in developing economies, albeit it rejects the influence of

macroeconomic stability in the explanation of growth of Latin American countries.

Lastly, neither Aid nor Communication equipment do affect growth significantly in the

level and system-GMM model. As shown by Harding and Javorcik (2007), the bilateral or

multilateral public aid can be interpreted like a proxy of the setting up of agencies for the

promotion of exportations and investments. Indeed, developing countries generally benefit

from aid inflows aiming at co-funding the setting up of promotion agencies and supporting

their actions. Our results show that in the context of the countries of our sample, the

importance the provision of infrastructure is more decisive for growth than the supply of

information about the conditions of foreign investment designed to attract the investments and

make them more efficient (Charlton, 2003; Morrisset & Andres-Johnson, 2004). The

proximity with Europe endows these countries with a physical power of attraction for FDI,

and infrastructure and macroeconomic ease the positive effects of growth to come out. Policy

implications are that MENA countries should pursue their effort to discover new comparative

advantages, even if they don’t correspond to manufactures or goods with high technology

contents. The technological content of trade (Lall et al., 2005) and the income level of trade

23

(Hausmann et al., 2007) provide useful labels of the quality of import and export which can

be profitably used in further empirical works in order to corroborate this implication.

A shortfall of our model is that it does not allow looking into the effect of FDI on

diversification and of diversification on growth in the data. But, simply regressing

diversification or discovery on FDI does not produce significant results for the countries of

our sample. The results of robustness checks have not been reported in the paper. Table 1

gives the results for the estimation of the most complete models. The change from one

estimator to another one does not create any significant instability in the value of the

estimated coefficients. A rise in the size of the panel when Lebanon is included does not

modify significantly the results. Moreover, control variables (aid, communication,

importation, M2) have been included step by step without either modifying significantly the

estimations. Testing for outlier (Algeria) does not significantly change the results either.

Conclusion

Export diversification has become a priority goal for the development strategies of the

MENA countries that want to go beyond some excessively strong specializations on raw

materials and finished goods for which prices and demand are rather unstable. Diversification

must favour at the same time the domestic and foreign investment and induce some

endogenous structural changes creating development. In this paper, we aimed at measuring

the effects of diversification of exports on growth in MENA countries. We show that export

discoveries are more decisive for growth than trade diversification per se. The issue was also

to test the hypothesis along which FDI do not necessarily have the same effect on growth

according to the level of diversification or discovery. Within the framework of a growth

model estimated by the GMM system method, we show that while FDI and new exports

favour the growth in MENA countries, higher levels of the latter decrease the effects of FDI

24

on growth. The step forward of this work will be to control both for the quality of import and

export and the nature of FDI to assess the joint effect of trade and investment on growth.

Appendix 1: Sources and units of the variables Variables* Sources GDPppa per capita WDI Constant (2000) School enrolment (secondary) WDI % Investment WDI Millions USD Labour WDI Millions FDI UNCTAD Millions USD Discover UNCTAD annex Export WDI Millions USD Import WDI Millions USD Inflation WDI Index M2 WDI % Gov. Spending WDI % Credit WDI % Aid WDI % Communication WDI Index of tree variables : fixed and mobiles

phones (for 1000); internet users; personal computers (for 1000)

Infrastructure WDI Index of two variables : paved road (%) ; electric power transmission and distribution losses (million kWh)

* All the variables in level are in a logarithmic form in the model.

Appendix 2

Definition of the Diversification Index

The values of the diversification index lie between 0 and 1.It gives information about the degree to which the export structure of a given country divert from the worlds’ export structure. A value close to 1 means that the export structure of the country strongly divert from the world’s trade one. The index S for a country i is a variant of a Finger-Kreinin index (Finger and Kreinin, 1979) and is a given by :

n Σ | hij- hi| Sj = i=1

2

where hij = share of the product i in the total of exports of the country j, and hi = share of the product i in the total of the world’s exports

Definition of the Concentration Index

25

The values of the concentration index lie between 0 and 1. It gives information about the degree of exports concentration of a given country. A value close to 1 means that the export structure is concentrated. The index H for a country i is a variant of the Herfindahl-Hirschmann normalized and is a given for country j by :

√ Σ (xi /X)2 - √(1/n) Hj = i=1 1 - √ (1/n)

where xi is the export value for the product I, and X is the sum of the xi .

Appendix 3: Regressions results for the Diversification and the concentration indexes

Dependent variable: annual GDP growth for 1995-2005. Difference-GMM and system-GMM

estimations

Observations : 88 Groups : 8

(Difference-GMM) Wald chi2(16) =

374.55 Prob > chi2 = 0.00

(system-GMM) Wald chi2(16) =

3425.25 Prob > chi2 = 0.00

(Difference-GMM) Wald chi2(16) =

426.52 Prob > chi2 = 0.00

(system-GMM) Wald chi2(16) =

3467.17 Prob > chi2 = 0.00

GDPit-1 0.5443*** (3.65)

0.9313*** (19.59)

0.3889** (2.76)

0.9432*** (19.93)

School 0.0054*** (3.18)

-0.0004 (-0.46))

0.0060*** (3.59)

0.0004 (0.45)

Investment 1.81e-12 (0.97)

1.69e-9** (2.07)

9.39e-13 (0.49)

0.00005* (1.96)

Labour 4.52e-09 (0.59)

1.02e-0.8** (1.93)

1.50e-08** (2.10)

1.23e-08** (2.22)

FDI 2.68e-04** (2.12)**

6.51e-6** (2.30)

0.0001* (2.07)

0.00005** (2.56)

Concentration -0.5339 (-1.14)

0.1172 (0.98)

- -

Diversification - - 0.1895 (1.29)

0.0955 (0.51)

Concentration*FDI 0.5174 (1.13)

0.0008*** (0.11

- -

Diversification*FDI - - 0.00001 (0.35)

0.00007 (1.38)

Export 3.32e-6** (2.33)

1.31e-12 (0.45)

4.35e-8** (2.22)

2.41e-12 (0.84)

Import 4.05e-12** (2.12)

2.97e-6** (2.57)

4.82e12** (2.61)

1.06e-12** (2.43)

Inflation -.0001 (-0.58)

-0.0008 (-0.35)

-0.0001 (-0.47)

0.00008 (0.30)

M2 -0.0005 (-1.07)

-0.0009* (-2.01)

-0.0005 (-0.98)

-0.0009** (-2.78))

Government Spendings

-.0097** (-3.13)

-0.0029 (-0.86)

-.0121*** (-4.30)

-0.0040 (-1.20)

Infrastructure 0.1062 (1.32)

0.0194 (0.22)

0.1786** (2.33)

-0.010 (-0.13)

Credit 0.0002** (1.97)

0.0011* (2.37)

.0003 (0.71)

0.0010** (2.05)

26

Aid -.0006 (-1.33)

-0.0001 (-0.23))

-0.0005 (-1.12)

-0.00006 (-0.11)

Communication 0.0213 (0.43)

0.0354 (0.64)

-0.0180 (-0.40)

0.054 (0.98)

Constant 3.417*** (3.38)

.5539* (1.84))

4.373*** (4.57)

0.4101 (-1.06)

J test (Sargan) for overidentifying restrictions49

chi2(42) = 38.32 chi2(52) =39.81 chi2(43) = 48.58 chi2(52) =47.96

AR(2) test for zero autocorrelation in first–difference errors

Z=0.315 Z=0.723 Z=0.421 Z=0.856

Note: (***) significant at the 1% level; (**) significant at the 5% level; (*) significant at the 8% level

References Acemoglu, Daron, and Zilibotti F. 1997. Was Prometheus Unbound by Chance? Risk,

Diversification and Growth. Journal of Political Economy 105 (4): 709-751. Al-Marhubi, F. 2000. Export diversification and growth: An empirical investigation. Applied

Economics Letters 7: 559-562 Alessandrini, G. 2000. FDI in the MENA region. Third Mediterranean Development Forum

Cairo 5-8 March 2000. Alfaro, Laura, A. Chandra, Selim Kalemli-Ozcan, and S. Sayek. 2004. FDI and economic

growth: the role of local financial markets. Journal of International Economics 64(1): 89-112.

An, G. and Murat Iyigun. 2004. The Export Skill Content, Learning by Exporting, and Economic Growth. Economics Letters 84(1): 29-34.

Arellano, M. and S. R. Bond. 1991. Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations, Review of Economic Studies 58(2),: 277-297.

Arellano, M. and O. Bover. 1995. Another Look at the Instrumental-Variable Estimation of Error Component Models. Journal of Econometrics 68(1): 29-52.

Balasubramanyam, V., M. Salisu, and D. Sapsford. 1996. Foreign direct investment and growth in EP and IS countries. Economic Journal 106(434): 92-105.

Barro, Robert J., and Xavier Sala-i-Martin. 1995. Economic Growth. McGraw Hill. Barro, Robert J., and J. W. Lee. 2000. International Data on Educational Attainment: Updates

and Implications, Centre for International Development, Working Paper No. 42, Harvard University.

Bénassy-Quéré, Agnès, Coupet, Mayliss, and Thierry Mayer. 2005. Institutional determinants of Foreign Direct Investment, CEPII Working Paper N° 2005-05, CEPII, Paris.

Bende–Nabende, A., J.L. Ford, and J.R., Slater. 2000. The Impact of FDI and Regional Economic Integration on The Economic Growth of The ASEAN-5 Economies,1970-1994 : A comparative Analysis in a Small Structural Model. In J.L.Ford (ed.),

49 Under the null hypothesis that all instruments are exogenous, J is distributed as a chi-square with m-r degrees of liberty is the number of instruments minus the number of endogenous variables.

27

Finance, Governance and Economic Performance in Pacific and South East Asia, London: Edwar Elgar.

Bertinelli, L., V. Salins, and E. Strobl. 2006. Export diversification and price uncertainty in Sub-Saharan Africa and other developing countries: a portfolio theory approach, mimeo, Ecole Polytechnique, Paris X University.

Blomström, M., R.E. Lipsey, and M. Zejan. 1994. What explains developing country growth? NBER Working Paper 4132, National Bureau of Economic Research, Cambridge, Mass.

Blundell, R., and S. R. Bond. 1998. Initial Conditions and Moment Restrictions in Dynamic Panel Data Model. Journal of Econometrics 87: 115-143.

Borensztein, E., J. De Gregorio, and J.W. Lee. 1998. How does foreign direct investment affect economic growth? Journal of International Economics 45 (1): 115-135.

Bouoiyour, J., and A. Akhawayn 2005. Labour productivity, technological gap and spillovers: Evidence from Moroccan manufacturing industries. The African Finance Journal 7(2).

Busse, M., and J.L. Groizard. 2006. Foreign direct investment, regulations and growth, World Bank Policy Research Working Paper 3882. The World Bank, Washington, DC.

Carkovic, Maria, and Ross, Levine (2005). Does foreign direct investment accelerate economic growth? in Moran T.H., Graham E.M. and Blomström M. (eds.) Does Foreign Direct Investment propote development ? Peterson Institute for International Economics.

Carrère, Céline, Olivier Cadot, and Vanessa Strauss-Kahn (2007). Export diversification: what’s behind the Hump? CEPR Discussion Paper No DP6590. CEPR, London.

Chan, K.K., and Gemayel E.R. 2004. Risk instability and the pattern of FDI in MEN, IMF Working Paper. International Monetary Fund, Washington D.C.

Charlton, Andrew. 2003. Incentive Bidding for Mobile Investment: Economic Consequences and Potential Responses. OECD Development Centre Working Paper 203. Paris, OECD.

Chenery, Hollis. 1979. Structural Change and Development Policy. New York: Oxford University Press.

Chuang, Y. 1997. Learning by doing, the technology gap, and growth. International Economic Review 39(3), 697-721.

Ciccone, Antonio, and Marek, Jarocinski, 2007. Determinants of Economic Growth: Will Data Tell?, CEPR Discussion Papers 6544, C.E.P.R. Discussion Papers, Centre for Economic Policy Research, London.

Coe, D. T., and E. Helpman. 1995. International R&D Spillovers. European Economic Review 39: 859–887.

Coe, D. T., E. Helpman, and W. Hoffmaister. 1997. North South R&D Spillovers. The Economic Journal 107 (440): 134-149.

Crespo, N., and M. P. Fontoura. 2007. Determinant factors of FDI spillovers – What do we really know? World Development 35 (3): 410-25.

Daniele, V., and U. Marani. 2006. Do institutions matter for FDI? A comparative analysis for the MENA countries. MPRA Paper No. 2426. Munich Personal Repec Archive, Munich.

De Ferranti, D., G. E. Perry, D. Lederman, and W.F. Maloney. 2002. From natural resources to the knowledge economy : trade and jobs quality, Washington, DC: The World Bank.

De Gregorio, J. 1992. Economic growth in Latin America. Journal of Development Economics 39 (1): 59-84.

28

De Pineres, S.A.G., and M. Ferrantino. 2000. Export diversification and structural dynamics in the growth process: The case of Chile. Journal of Development Economics 52: 35-91

Durlauf, S. N., P. A. Johnson, and J. R. W., Temple. 2005. Growth econometrics. In Handbook of Economic Growth, ed. Philippe Aghion and Stephen N. Durlauf, volume 1A. Amsterdam: North-Holland.

Elmawazini, K. 2007. Do MENA countries benefit from the technology diffusion from FDI? Science,Technology and Sustainability in the Middle East and North Africa (GTSD - MENA) 3 (1).

Evenson, R., and L.E. Westphal. 1995. Technological change and technology strategy. In Handbook of Development Economics, ed. J. Behrman and T.N. Srinivasan, volume 3A. Amsterdam: North-Holland.

Feenstra, R. and H. L. Kee. 2004. Export variety and country productivity, Policy Research Working Paper Series 3412. The World Bank, Washington D.C.

Finger, J. M., and M. E. Kreinin. 1979. A measure of ‘export similarity’ and its possible uses. Economic Journal 89: 905-12.

Görg, H., and D. Greenaway. 2004. Much Ado About Nothing? Do Domestic Firms Really Benefit from Foreign Direct Investment? World Bank Research Observer 19 (2): 171-197.

Haddad, M., and A.E. Harrison. 1993. Are there positive spillovers from foreign direct investment? Evidence from panel data from Morocco. Journal of Development Economics 42 (1): 51-74.

Harding, T., and B. Javorcik. 2007a. Developing Economies and international investors: Do Investment Promotion Agencies bring them together? World Bank Policy Research Working Paper 4339. The World Bank.

Harrison, A. 1996. Determinants of effects of foreign direct investment in Cote d’Ivoire, Morocco and Venezuela. In Industrial Evolution in Developing Countries, ed. M.J. Roberts and J.R. Tylbout. New York: Oxford University Press.

Hausmann, Ricardo, and Bailey Klinger. 2006. Structural Transformation and Patterns of Comparative Advantage in the Product Space. Mimeo. Harvard University.

Hausmann, Ricardo, and Bailey Klinger. 2007. The Structure of the Product Space and the Evolution of Comparative Advantage. CID working paper 146, Harvard University.

Hausmann Ricardo, and Dani Rodrik. 2003. Economic development as self-discovery. Journal of development Economics 72, 603-633.

Hausmann, Ricardo, D Hwang, and Dani Rodrik. 2007. What you export matters. Journal of Economic Growth 12 (1): 1-25.

Hermes, N., and R. Lensink. 2003. Foreign direct investment, financial development and economic growth. Journal of Development Studies 40 (1): 142-163.

Herzer, D., F. Nowak-Lehman, and B. Siliverstovs. 2006. Export-Led Growth In Chile: Assessing The Role Of Export Composition In Productivity Growth. The Developing Economies vol. 44 (3): 306-328.

Hesse, H. 2007. Export diversification and economic growth. Mimeo. The World Bank. Holtz-Eakin D., W. Newey, and H.S. Rosen. 1988. Estimating Vector Autoregressions with

Panel Data. Econometrica 56: 1371-1396. Imbs, Jean, and Robert Warcziag. 2003. Stages of diversification. American Economic Review

93 (1): 63-84. Iqbal, F., and M. K. Nabli. 2004. Trade, foreign direct investment and development in Middle

East and North Africa. Paper prepared for the Conference: The Middle East and North Africa Region: Challenges of growth and globalization, Organized by the International Monetary Fund, Washington DC, April 7-8, 2004.

29

Kalemli-Ozcan, S., B.E. Sorensen, and O. Yosha. 2003. Risk Sharing and industrial specialization: regional and international evidence. American Economic Review 93 (3): 903-18.

Kaplinsky, R. 2006. China and the global terms of trade. IDS Bulletin 37 (1): 43-53. Klinger, Bailey, and Daniel Lederman. 2004. Discovery and development: an empirical

exploration of new products. World Bank working paper 3450. Washington DC: The World Bank.

Klinger, Bailey, and Daniel Lederman. 2006a. Diversification, innovation and imitation inside the global technological frontier. World Bank working paper 3872. The World Bank.

Klinger, Bailey, and Daniel Lederman. 2006b. Innovation and exports portfolio. World Bank working paper 3983. The World Bank.

Koren, M., and S. Tenreyro. 2007a. Volatility and development, The Quarterly Journal of Economics 122 (1): 243-287.

Koren, M., and S. Tenreyro. 2007b. Technological diversification. Proceedings Federal Reserve Bank of San Francisco, issue Nov.

Lall, S., J. Weiss, and J. Zhang. 2005. The “sophistication” of exports: A new trade measure. World Development, 34(2): 222-237.

Lederman, D., and W. F. Maloney. 2007. Trade Structure and Growth. In Natural Resources: Neither Curse Nor Destiny, ed. D. Lederman and T.N. Srinavasan, Palo Alto: Stanford University Press.

Levchenko, A., and J. Di Giovanni. 2008. The risk content of exports: a portfolio view of international trade. Mimeo. The International Monetary Fund.

Li, X., and X. Liu. 2005. Foreign Direct Investment and economic growth: An increasingly endogenous relationship. World Development 33 (3): 393-407.

Lipsey, R. 2000. Inward FDI and Economic Growth in Developing Countries. Transnational Corporations 9 (1): 67-95.

Lucas, Robert 1988. On the Mechanics of Economic Development. Journal of Monetary Economics 22: 3-42.

Melitz, Marc J. (2003) The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity. Econometrica LXXI (2003): 1695–1725.

Melitz, Marc J. (2005) Comment to Carkovic Maria, and Levine Ross (2005). Does foreign direct investment accelerate economic growth? in Moran T.H., Graham E.M. and Blomström M. (eds.) Does Foreign Direct Investment propote development ? Peterson Institute for International Economics.

Moran, T., 1998, Foreign direct investment and development: the new policy agenda for developing countries and economies in transition, Institute for International Economics, Washington, D.C.

Moran, T. H., Graham E. M. and M. Blomström (2005) Does Foreign Direct Investment Promote Development? Peterson Institute for International Economics, Washington, D.C.

Morisset, J., and K. Andrews-Johnson. 2004. The effectiveness of promotion agencies at attracting Foreign Direct Investment. FIAS occasional Paper N°16. Washington D.C., Foreign Investment Advisory Service.

Nelson, R. R. 2000. On technological capabilities and their acquisition. In Science and Technology: Lessons for Development Policy, ed. R. Evenson and G. Ranis, Boulder: Westview Press.

30

Noland, M., and H. Pack. 2007. The Arab Economies in a Changing World. Institute for International Economics. June 2007.

Olofsdotter, K. 1998. Foreign direct investment, country capabilities and economic growth. Weltwirtschaftliches Archiv 134(3): 534-547.

OECD. 2002. Foreign Direct Investment for development: maximising benefits, minimising costs. Organisation for Economic Cooperation and Development, Paris.

Petri, P. A. 1997. The Case of Missing Foreign Investment in the Southern Mediterranean. OECD Development Centre Working Papers 128. Organisation for Economic Cooperation and Development, Paris.

Prebisch, Raul. 1950. The Economic Development of Latin America and its Principal Problems. Economic Bulletin for Latin America 7: 11-22.

Prüfer, P., and G. Tondl. 2007. Does it Make a Difference? Comparing Growth Effects of European and North American FDI in Latin America. Proceedings of the German Development Economics Conference, Göttingen 2007.

Ram, R., and K.H. Zhang. 2002. FDI and economic growth: evidence from cross-country data for the 1990s. Economic Development and Cultural Change 51 (1): 205-214.

Romer, Paul M. 1986. Increasing Returns and Long Run Growth. Journal of Political Economy 94: 1002–37.

Romer, Paul M. 1990. Endogenous Technological Change. Journal of Political Economy 98: S71-S102.

Sadik, A. T., and A. A. Bolbol. 2001. Capital Flows, FDI, and Technology Spillovers: Evidence from Arab Countries. World Development. 29 (12): 2111-2125

Sala-I-Martin, Xavier, Gernot Doppelhofer, and Ronald Miller, 2004. Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach, American Economic Review, 94(4): 813–835.

Sekkat, K.. 2004. FDI inflows to the MENA region: an empirical assessment of their determinant and impact on development. FEMISE Research n°FEM21-15.

Singer, Hans W. 1950. US Foreign Investment in Underdeveloped Areas: The Distribution of Gains between Investing and Borrowing Countries. American Economic Review 40: 473-485.

Stokey, N. 1991. Human capital, product quality, and growth. Quarterly Journal of Economics 106: 587-616.

Syrquin, M. 1989. Patterns of structural change. In H. Chenery and T.N. Srinavasan, (eds.), Handbook of Economic Development. Amsterdam: Elsevier Science Publishers.

UNCTAD. 1999. World Investment Report: Infrastructure for Development. New York and Geneva: United Nations.

World Bank. 2006. Fostering Higher Growth and Employment with Productive Diversification and Competitiveness. Morocco Country Economic Memorandum, vol.1 and 2. The World Bank, Washington DC.

World Bank. 2007. Export diversification in Egypt, Lebanon, Jordan, Morocco and Tunisia. Mimeo. The World Bank, Washington DC.

31

Xu, B. 2000. Multinational Enterprises, Technology Diffusion, and Host Country Productivity Growth. Journal of Development Economics 62(2): 477-93.

Young, A. 1991. Learning by doing and the dynamic effects of international trade. Quarterly Journal of Economics 106(2): 369-405.