Foreign Direct Investment decisions and fixed setup costs ...
Foreign Direct Investment and Location Patterns in Developing Countries: A check for set up costs in an “export-platform FDI”?
Sami REZGUI[1]
University of Tunis
Abstract : This paper investigates the impact of set up costs on foreign direct investment (FDI) decisions considering Tunisia as an export platform FDI. After presenting traditional analyses of the determinants of FDI flows in developing countries, I develop a simple model inspired from Razin & ali (2003) where FDI decisions are examined through the simultaneous definition of gravity and participation equations. Using panel data for Tunisian inward FDI covering the period 1990-2001, a gravity equation is estimated by the tobit procedure to correct for error measurements inherent to the use of small or nearly zero positive values of FDI flows as dependent variable. The heckman maximum likelihood is then applied to correct for sample selection bias and check for the existence of fixed set up costs. The results confirm the dependency of the processes of investment decision and the amount of FDI invested. The existence of set up costs is also confirmed when considering foreign investors that are not frequently and enough investing in the Tunisian economy.
JEL classification: F21; F23; C23; C24; O1
Keywords: FDI, set up cost, investment location decision, export platform, gravity model.
1.Introduction
Over the last decade, Foreign Direct Investment (FDI) has been of a major concern for developing countries’ governments. Competition to enhance attractiveness for FDI in order to achieve development objectives (employment, exports, knowledge spillovers) and to limit external debt for financing local investment needs is becoming a rule. However, many economist agree on the fact that developing countries (with the exception of China) and particularly Africa attract abysmal FDI flows. Despite efforts made to enhance macro-economic fundamentals through structural adjustment plans, to liberalise economy and to offer fiscal incentives, FDI flows are still below governments’ hopes. In this paper, I will not concentrate on the traditional theoretical determinants of FDI. The principal objective of this study is to show that beside the “apparent” determinants of FDI attractiveness, many latent variables could influence attractiveness and affect foreign investment decisions.
The empirical literature on FDI has been largely concentrating on macro-economic factors as key determinants of FDI inflows to developing countries. Based on different methodologies, this literature offers a wide range of empirical results but no clear consensus emerges about the relevant variables that could be of great influence on FDI inflows. However, variables such as GDP per capita, financial risks, institutional environment, human factor, exchange rates and openness to trade are the most commonly used variables in models explaining inward FDI (Singh & Jun, 1995).
Inspired from gravity models developed in the context of international trade (Bergstand 1985; Frankel & Romer, 1999), another set of related empirical studies includes geographical characteristics to explain bilateral flows of FDI between developed and developing countries. According to these studies, the traditional location specificity characterised by market seeking and efficiency seeking arguments could not alone be sufficient to explain inward FDI trends. Although the often claimed efficiency and local market conditions guaranteed, it is clearly demonstrated that physical proximity from home foreign investors as measured by geographic distance significantly affect the amount of investments addressed to these countries (Brainard 1993 and 1997; Frenkel & ali 2004). In a dynamic perspective, location choice of investments also depend on agglomeration effects explained both in terms of earlier decisions to invest in a country[2] and in terms of country investors’ profile previously investing in this country[3] (Compos & Kinoshita, 2003; Crozet & ali 2004). Finally, one should postulate that, despite their attractiveness, receiving FDI countries could not influence foreign investment projects. As pointed out frequently in many studies, factors related to business cycles in home countries are expected to play an important role in foreign investors’ decisions. Then, it would be necessary to take in account the proper conditions of the source countries as exogenous variables in order to explain inward FDI.
This brief overview of the literature is intended to summarise some of the major determinants of inward FDI that should be considered in an empirical framework. But, following the new FDI theory, a focus should also be made on foreign investment decision process and particularly the way set up costs could affect these decisions for developing countries. This is the hypothesis I check for in this work taking Tunisia as an example.
The remainder of this paper is structured as follows. In section 2, the set –up costs issue is explained and then integrated in a model adapted for the context of unilateral inward FDI. Section 3 presents the econometric methodology and the Tunisian data employed. Interpretations of the regression results will follow in section 4. Section 5 offers the main conclusions to be derived from the empirical analysis.
2. A brief overview on set up costs in the new FDI theory.
The growing part of multinationals in international trade as observed through the large part of exports and imports linked to their activities was the main stimulus for the construction of a new theory of FDI where patterns of trade are explicitly predicted on the basis of FDI strategies. Within this theoretical corpus, set up costs represents a new perspective for the analysis of FDI determinants. Whether firms are expected to choose between producing and then exporting or displacing production abroad, their choice would be not mechanic. Producing abroad is merely assumed to be motivated by factor endowments and by geographical proximity but additional costs would also be incurred by firms choosing this alternative. How these costs materialise, what is their nature and how could they affect FDI location decision.? We try in this section to shed some lights on the importance of set up costs in recent FDI literature and explore this alternative within a simple model of investment decisions.
2.1 Multinational strategies and set up costs.
Following the theoretical perspectives on Multinationals’ activities, foreign investment decision is explained both by factor endowments and proximity-concentration trade off depending on the nature of multinationals. The former explanation corresponds to the vertical multinationals[4] model (Helpman, 1984) while the latter is assumed for horizontal multinationals models (Markusen 1984; Brainard 1997).
The factor endowments strategy underlying foreign investment decisions considers that FDI patterns are based on classical production factors mainly capital and labour. If FDI are technology based, one would expect them to be concentrated in countries where physical and technological capabilities are abundant. This is the case for a large part of FDI concentrated in the “Triade” (USA, European Union and Japan). One should also expect to see skilled based FDI concentrated in developed countries whereas unskilled based FDI would concentrate in labour force abundant developing countries. In this latter case, large FDI inflows are supposed to benefit African countries but, unfortunately, this is not statistically verified.
As underligned by Lucas (1990), capital would not flow from rich to poor countries because as qualification externalities enhances labour productivity in rich countries, foreign investments will just concentrate within rich countries. But, when the wage cost advantage is invocated, Lucas’s hypothesis would fail in explaining rich countries investments in developing countries. However, as developing countries attract a small part of world FDI, this means that the wage advantage alone could not be sufficient. When added costs are to be incurred, rich countries investors will think before deciding to invest even if wages in host countries are very attractive. The added costs are those concerning labour profiles needed as well as those incurred by training and forming local labour force[5] .
Using a general equilibrium model, Zhang & Markusen (1999) conclude at the existence of a close correlation between FDI inflows and fixed costs incurred through labour requirements. For Caballero & Engel (1999), firms would also be confronted to lumpy investments and accordingly, may incur fixed costs some of them are reorganising costs associated to putting new capital in work. In the context of FDI, fixed costs incurred on capital based investments would be associated to the time needed for foreign investors to readapt the technology to the host country both in terms of skills and technological capabilities.
Theoretical insights on horizontal multinationals consider that the decision to invest abroad depends on the choice between avoiding transport costs (proximity) and benefiting from firm scale economies (or fixed costs advantages) when output is produced in a single place (concentration). The contribution of host country market size to reducing fixed costs is assumed to be important because of the possibility offered to foreign investors to sell their production avoiding transport costs on exports[6]. An equilibrium with a decision to invest abroad (Multinational solution) exist when saving transport costs is possible although additional fixed costs are incurred by firms opening production facilities abroad (Brainard 1997), with a more likely FDI presence from highly productive multinationals (Helpman & ali, 2003).
Inspired by the knowledge capital model (Markusen, 1997, Markusen & ali, 1999), Carr et ali (2001) propose an empirical analysis of FDI strategies combining both vertical and horizontal multinational models. According to the authors, the production volume of foreign affiliates (a proxy for FDI) from one country’s firm in another country depends on characteristics of both countries. These characteristics could be in part linked to factor endowments and in another part associated to the proximity concentration trade off. Two main results are obtained by the authors: trade costs raises affiliates’ production whatever the form of the multinational enterprise (vertical or horizontal) whereas low differences in skilled labour endowment lead to a decrease in affiliates’ sales supporting the knowledge capital model of multinationals[7].
In these models however, FDI inflows seems to be quasi substitutes to direct exports which means that FDI and export would be exclusive strategies. But, as pointed out by Markusen (1997), these strategies could in fact be complementary. For developing countries, FDI and exports would be seen as close complements particularly if we consider foreign investors that use these countries as “export platforms FDI”. As documented recently by Ekholm & ali (2003), part of FDI addressed to some developing countries engaged in free trade areas such as Mexico (NAFTA) is associated to export platform FDI strategies[8]. According to the definition given by the authors[9], most of FDI benefiting to Tunisia could also be considered as such. In table 1, we enumerate the main characteristics of an export platform FDI strategy as described by Ekholm & ali and try to check for them in the Tunisian context[10].
Table 1
Characteristics of an export platform FDI : The case of Tunisia
|Characteristics of an Export platform FDI |Existence |Tunisian context |
|strategy* | | |
| | | |
|Northern foreign firms are interested by low |Yes |Tunisia’s main advantage is labour cost wage |
|wage cost location | | |
| | | |
|Northern foreign firms are interested by | | |
|south country engaged in free trade area |Yes |Tunisia has signed a free trade association agreement with|
| | |the European Union. |
|Demand in the host country is low. | | |
| |Yes |Tunisian’s market size is small and foreign investors are |
| | |not authorised to sell more than 40% of their final output|
| | |on the local market. |
|The host country is engaged in an assembling | | |
|activity |Yes |This is the case for the clothing industry in Tunisia |
* As enumerated in Ekholm & ali (2003)
But even if Tunisia is assumed to be an export platform FDI, would this contribute to increased FDI in this country? We consider in this paper that within the export platform FDI strategy, an opportunity would be offered to Tunisia to attract more FDI provided that set up costs are low. However, this vertical FDI strategy is not necessarily to contribute to more FDI flows to Tunisia with regard to rules of origin [11]. In this paper, only set up costs will be emphasised while rules of origin as trade impediments to FDI will be ignored.
Let’s examine in what follows the nature of set up costs and the way to measure them. I notice first that measures of set up costs differ across the empirical studies integrating this variable. Second, the terms used to define these costs are not always the same. For wheeler and Mody (1992), factors such as infrastructure quality (transportation and communication) could be a proxy for investment costs that affects the expected return on investment and hence influence the pattern of FDI location. In other studies, investment costs are simply associated to an average of several indexes of impediments to invest such as those reported in the World Economic Forum (Brainard, 1997; Carr & ali, 2001; Amity and Waklin, 2003). Using these approximations, Brainard (1997) found that the probability for observing multinationals decreases with investment barriers and low openness to FDI in host countries[12]. Carr & ali (2001) found that investment costs affect FDI particularly when controlling for physical distance.
In fact, when we look at the Global Competitiveness Index (GCI) for some developing countries as reported in the World Economic Forum, one could not systematically conclude at the existence of high barriers to investment in these countries. For example, according to GCI[13] recent measures (2004), Tunisia is ranked 42 loosing its leadership place to the benefit of South Africa. Moreover, when looking at the Business Competitiveness index, Tunisia is ranked 32 just before Portugal and Italy. Barriers to investment should then be interpreted with some caution. Hence, the question would not be to consider whether a country, say Tunisia for example, having low barriers to investments would be automatically attractive as an FDI destination. Instead, it is more interesting to ask, starting from Tunisian FDI data and the new theoretical insights on FDI, why some foreign investors does not invest enough in Tunisia despite low barriers to investment in this country.
As noticed by Brainard (1997) and Helpman & ali (2003), one of the theoretical answer to this question should be placed on the economies of scale perspective particularly at the level of the affiliate to be created. Economies of scale would enhance the probability of FDI because foreign investors would be offered the possibility to spread costs over large scale production and to save transport costs even though production is not necessarily addressed to the local market (export strategy) as is the case when affiliates creation is aimed to develop export platforms. This alternative deserves to be tested in the Tunisian economy for the following reasons : firstly, Tunisian market size is very limited although the free trade agreement signed in 1995 with the European Union is thought to be an opportunity to over pass this constraint and would allow for more FDI flows through an export platform strategy. Secondly, the only foreign investment possibilities that lowers fixed set up costs for FDI reduces mainly to the energy sector through Build Own Operate (BOO) investments and through cement factories privatisation. Finally, some sectors such as transport where economies of scale are clearly evidenced are not freely opened to foreign investors[14].
Investigating empirically the set up costs alternative from the point of view of scale economies is however hardly difficult. In the absence of available data that allows production functions estimates, a measure of these costs would reduce to approximations such those used by Brainard (1997) and consisting in the ratio of advertising expenditures or R&D expenditures on sales[15]. Lacking information on these measures and considering the nature of FDI data available for Tunisia, we prefer to opt for an alternative investigation.
2.2. FDI set up costs and investment decision : a simple model
In a recent article explaining investment patterns in the context of trade openness, Razin & ali (2003) consider that developing countries may face high investment oscillations particularly when scale economies are present within the investment technology. Lumpy adjustment costs incurred on new investments are supposed to take the form of fixed set up costs that lump investments over time. As noticed by the authors, this definition of fixed set up is to be distinguished from the classical cost-of-adjustment specification on the basis of which investment is supposed to be spread over time[16].
In this subsection, we propose a remake of the Razin & ali (2003, 2004) model where fixed set up costs are assumed to affect foreign investment decision. In this model, the foreign investors decision is examined considering the two following steps :
- In the first step, the foreign investor decides how much to invest and his decision depends on the expected profitability of the investment.
- In the second step, the foreign investor decides to invest or not depending on the importance of fixed set up costs.
The two steps are supposed to be correlated: Foreign investors will not decide to invest when fixed costs are high so the level of FDI will be zero or nearly. The model will be adapted to the Tunisian FDI context. In fact, outward FDI from Tunisia to the rest of the world are quit null so it is more appropriate to limit the analyses to inward FDI.
Let’s suppose a foreign investor F deciding to realise an amount of investment in Tunisia. We note I this investment. We distinguish here two categories of foreign investors : those of type M which could be major foreign investors that are usually investing in Tunisia and those of type N that could be new investors or not frequently investing in Tunisia. We will use indexes T and h to designate Tunisia and an other host country respectively. For each type of foreign investor is associated an investment equation :
IN = KN - (1- () KhN (1)
IM = KM - (1- () KTM (2)
In equation 1, KN represents the stock of capital of type N foreign investor which include investment flow realised in Tunisia and KhN is the stock of capital including all investments realised in another host country. In equation 2, KM is the stock of capital of type M foreign investor including the new investment and KTM representing the stock of capital it has already invested in Tunisia in previous periods.
Fixed set up costs, we denote C, are incurred by N et and M type investors. A proportion of these costs (e) are equally supported by N et M and consisting in communication and transport infrastructure supposed to be insufficient in Tunisia. The other proportion of set up costs is attributed to labour requirements and would be positively correlated to the amount of Tunisian labour to be trained and formed (LT). Set up costs on labour requirements are supposed to be unequally supported by N and M. For example investors N would face more labour costs for training and forming local labour than M investors, the latter benefiting from earlier investments they made in formation and training[17].
Total fixed costs are represented by the following equations :
Ci = e + C[(LTi)+] , i = (N,M( (3)
CM < CN (4)
Distinguishing between N and M investors should not be based only on the perceived constraint of set up costs. When observing Tunisian FDI data by country origin of investors, we find that some foreign countries are usually or quite frequently investing whereas other countries are not usually present and enough investing. An agglomeration effect that we traduce here in terms of earlier country presence would also play. For example, French or Spanish investors would be more incited to invest because earlier French and Spanish investors are already present in Tunisia. Nevertheless, Danish investors would be insufficiently incited to invest because historically, Denmark’s investors had rarely invested in Tunisia.
Following this point of view, let’s consider that there is a probability P that M investors invest in Tunisia and a probability p that N investors do so with P > p. Following Razin & ali, we consider that in the presence of set up costs, investors N and M compare the value of the foreign firm when investment are or are not realised. The value of the firm is evaluated in terms of actualised profits including set up costs with w representing wages and A a “hicks neutral” technical progress.
For type M investors, the one period actualised value of the foreign firm when deciding to invest is :
VM+(AT, wT, LTM) = max ([pic]- [e+ C(LTM) + KM - (1- () KTM]( (5)
For type N investors, the one period actualised value of the foreign firm when deciding to invest is :
VN+( AT, wT, LTN) = max ( [pic] - [ e + C(LTN) + KN – (1- () KhN]( (6)
When N and M investors decide to not invest, the value of the foreign firm would be respectively :
For type M : VM- (AT, wT) = Max ([pic] +(1- ()KTM ( (7)
And for type N : VN- (Ah, wh) = Max ([pic] + (1-() KhM( (8)
Equation 8 means that when deciding to not invest in Tunisia, type N investor continue to produce in another host country without making new investments there[18]. Considering equations (5) to (8), M and N investors will decide to invest if :
P [VM+(AT, wT, LTM)] - (1 – P ) [VM- (AT, wT) ] ( 0 (for M investors) (9)
p [ VN+( AT, wT, LTN)] - (1- p) [VN- (Ah, wh)] ( 0 (for N investors) (10)
Equations (9) and (10) are to be considered as participation equations. From these equations, we could define FDI inflows functions according to each investor type :
IM = KM(AT, wT, LTM) - (1- () KTM - [e+ C(LTM)]
if P [VM+(AT, wT, LTM)] ( [(1 – P ) [VM- (AT, wT) ]
= 0 if P [VM+(AT, wT, LTM)] < [(1 – P ) [VM- (AT, wT) ] (11)
IN = KN(AT, wT, LTN) – (1- () KhN - [e + C(LTN)]
if p [ VN+( AT, wT, LTN)] ( (1- p) [VN- (Ah, wh)]
= 0 if p[ VN+( AT, wT, LTN)] < (1- p) [VN- (Ah, wh)] (12)
When including set up costs, the amount of FDI investment flows received by Tunisia depends on equations (9) and (10):
FDI = IM + IN (13)
P[VM+(AT, wT, LTM)] +p[ VN+( AT, wT, LTN)] (
[(1– P)[VM-(AT, wT) ] +(1-p)[VN-(Ah, wh)] (14)
Equations (13) and (14) represents FDI gravity equation and FDI participation equation respectively. The empirical methodology adopted in the next section will rely on an econometric approach that is well adapted to the estimation of these two equations simultaneously.
3. Empirical approach and data description
3.1 Empirical approach
Empirical estimates of FDI determinants has been frequently done using OLS estimates. This econometric strategy well adapted when observed values of the dependent variable are all strictly positive would lead to biased estimators when zero values of the endogenous variable are observed [Amemiya, 1984]. Moreover, ignoring zero values observed for the dependent variable and running least squares just on those who are strictly positive would lead to sample selection bias and in that case, OLS estimators are inefficient.
To characterise the selection bias issue [Heckman, 1979], let’s consider the following two equation model :
Yai = Xai (a + Uai (15)
Ybi = Xbi (b + Ubi (16)
Xai and Xbi are 1 ( Ki vectors of exogenous variables and (j (j = (a, b() is a 1( Kj vector of parameters. A random sampling scheme implies :
E(Uji) = 0 ; E(Uji Uj’i’) = (jj’ , i = i’
= 0 , i ( i’
with h(Uai, Ubi) the joint density of Uai and Ubi
When data is observed for all i ( i = 1,…….,I,), the regression function for equation (15) is :
E (Yai ( Xai ) = Xai (a (17)
Regressing on the sub-sample of available data that is on i = 1,….., Ia with (Ia < I) implies the following regression function :
E (Yai ( Xai, SSR[19] ) = Xai (a + E (Uai ( SSR) (18)
On the basis of the heckman’s characterisation of sample bias, if FDI flows determinants are estimated with OLS in the spirit of equation (15) and considering just the sub-sample of countries between whom FDI flows are observed (non zero flows is the SSR), we obtain unbiased estimators[20] if :
E (Uai ( SSR) = 0 (19)
But ignoring zero FDI flows in a non missing randomly data on this variable raises the selection bias problem. Moreover, ignoring zero FDI flows eliminates the set up cost hypothesis[21]. On an econometric ground, regressing a set of exogenous just to explain the strictly positive FDI flows observed (ybi ( 0) and choosing zero flows as threshold (for Ybi ................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
- mr agard s class home
- globalization and china s economic and financial development
- three perspectives on china s outward foreign direct
- mcdonald s international marketing strategy an analysis
- south africa s experience with special economic zones
- foreign direct investment decisions and fixed setup costs
- investing in resilience focus on agriculture
- land reform in zimbabwe
Related searches
- investment decision and financing decision
- investment decisions in financial manage
- capital investment decisions pdf
- capital investment decisions examples
- investment analysis and portfolio management
- investment decisions in financial management
- investment decisions pdf
- foreign direct deposit canada
- investment analysis and portfolio management pdf
- foreign direct deposit enrollment form canada
- fidelity investment reviews and ratings
- direct investment companies