FDI FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH

Niels Hermes* and Robert Lensink**

(*) Faculty of Management and Organisation; (**) Faculty of Economics, University ofGroningen, PO Box 800, 9700 AV Groningen, The Netherlands.Email: and

The authors thank Gerard Kuper for his comments on an earlier version of the paper.

$%675$&7

FDI may help to raise economic growth in recipient countries. Yet, thecontribution FDI can make may strongly depend on the circumstances in therecipient countries. This paper argues that the development of the financialsystem of the recipient country is an important precondition for FDI to have apositive impact on economic growth. A more developed financial systempositively contributes to the process of technological diffusion associated withFDI. The paper empirically investigates the role the development of thefinancial system plays in enhancing the positive relationship between FDI andeconomic growth. The empirical investigation presented in the paper stronglysuggests that this is the case. Of the 67 countries in data set, 37 have asufficiently developed financial system in order to let FDI contribute positivelyto economic growth. Most of these countries are in Latin America and Asia.

2

󰀔󰀑󰀁,1752'8&7,21

The contribution of foreign direct investment (FDI) to economic growth hasbeen debated quite extensively in the literature. This debate has clarified thechannels through which FDI may help to raise growth in recipient countries. Inparticular, it has been emphasised that FDI may enhance technological changethrough spillover effects of knowledge and new capital goods, i.e. the processof technological diffusion. Yet, as some have argued, the contribution FDI canmake is strongly dependent on the circumstances in the recipient countries.However, empirical studies investigating the relationship between FDI andeconomic growth on the one hand, and the role played by the circumstancesFDI is confronted with whenever it enters a recipient country on the otherhand, are scarce.1

This paper argues that the development of the financial system of therecipient country is an important precondition for FDI to have a positiveimpact on economic growth. The financial system enhances the efficientallocation of resources and in this sense it improves the absorptive capacity ofa country with respect to FDI inflows. In particular, a more developed systemmay contribute to the process of technological diffusion associated with FDI.The contribution of this paper is to investigate empirically the role thedevelopment of the financial system plays in enhancing the positiverelationship between FDI and economic growth.

The paper is structured as follows. Section 2 provides a description ofthe discussion of the contribution FDI can make to increased economicgrowth. The section emphasises the importance of technological diffusion andthe role of FDI, as well as the contribution the financial system can make inthis respect. Section 3 discusses the data and the empirical methodology.Section 4 discusses the outcomes of the empirical investigation. Finally,section 5 provides a summary and concluding remarks.

1

Exceptions are Balasubramanyam HW󰀁DO󰀑 (1996), Borensztein HW󰀁DO󰀑 (1998) and Lichtenbergand van Pottelsberghe de la Potterie (1998). For an overview of the literature on therelationship between FDI and economic growth, see De Mello (1997).

3

󰀕󰀑󰀁)',󰀏󰀁7+(󰀁),1$1&,$/󰀁6<67(0󰀏󰀁$1'󰀁(&2120,&󰀁*52:7+󰀝$󰀁'(6&5,37,9(󰀁7+(25(7,&$/󰀁)5$0(:25.

In the new growth literature the importance of technological change foreconomic growth has been emphasised (Grossman and Helpman, 1991; Barroand Sala-i-Martin, 1995). The growth rate of less developed countries (LDCs)is perceived to be highly dependent on the extent to which these countries canadopt and implement new technologies available in developed countries(DCs). By adapting new technologies and ideas (L󰀑H󰀑 technological diffusion)they may catch up to the levels of technology in DCs. One important channelthrough which adoption and implementation of new technologies and ideas byLDCs may take place is FDI. Empirical research has shown that multinationalcorporations (MNCs) belong to those firms that are technologically veryadvanced and invest heavily in research and development. Their FDI may haveexternal effects on the process of technological change in the host countries.The advanced technology, new varieties of capital goods and managementskills they introduce in these countries may spillover from subsidiaries ofMNCs to domestic firms (Findlay, 1978). This knowledge spillover may takedifferent forms. The following channels through which spillovers from FDImay take place, have been distinguished in the literature (Kinoshita, 1998,pp.2-4; Sjöholm, 1999a, p.560): spillovers through (1) demonstration and/orimitation; (2) competition; (3) linkages; and (4) training.

Spillovers through the GHPRQVWUDWLRQ channel emphasises thattechnologies used by foreign firms are more advanced than those used bydomestic firms, and that these domestic firms may imitate the newertechnologies, which will make them more productive. The same may hold formanagerial practices introduced by foreign firms. The demonstration effectmay take place through direct or indirect contact between firms or throughlabour turnovers from foreign to domestic firms. The more backward thetechnological level in the host country in comparison to the level used by theforeign firms, the more domestic firms may profit from imitating and copyingthese technologies. This appears to reflect the idea of convergence oftechnological skills.

The FRPSHWLWLRQ channel stresses that the entrance of foreign firms

4

intensifies competition in the domestic market. This forces domestic firms tobecome more efficient, which may lead to upgrading existing technology, ordeveloping (or copying) new technologies and management skills.

The OLQNDJHV channel of spillovers stresses the fact that foreign firmsmay transfer new technology to domestic firms through transactions with thesefirms. Such transactions may for example be in terms of the purchase of rawmaterials or intermediate goods. This may lead to intensive buyer-sellerrelations with domestic firms in the host country, and as part of these relationsforeign firms may provide technical assistance and training to local linkagefirms. Moreover, selling to foreign firms may encourage domestic firms toupgrade the production process based on the technical and qualityrequirements demanded by the foreign buyers, increasing their productivity.

Finally, the WUDLQLQJ channel emphasises that the introduction of newtechnologies, and domestic firms copying them, needs to be supported by anupgrading of the human capital available domestically. Domestic firms canonly adopt these new technologies when the labour force is able to work withthem. Therefore, local firms may be stimulated to train their own employeeswhen foreign firms enter the market. This stimulus may be based on one of theother three channels discussed. Thus, perceived opportunities to copy newlyintroduced technologies, increased competition in domestic markets and/or theexistence or development of linkages between foreign and domestic firms maylead to increasing training efforts by domestic firms. This latter point alsomakes clear that in practice it will be rather difficult to separate the fourchannels of spillovers.

In any case, it is hypothesised that the spillover of new technologiesleads to higher productivity of capital and labour in the host country. The maincontribution of FDI is therefore in terms of improving total factor productivity,rather than its contribution in terms of increasing the volume of (physical andhuman) capital.

The next question is what conditions in the host country are importantto maximise the technology spillovers discussed above? In the literature it hasbeen emphasised that the spillover effect can only be successful given certaincharacteristics of the environment in the host country. These characteristics

5

together determine the absorption capacity of technology spillovers of the hostcountry. Thus, FDI can only contribute to economic growth through spilloverswhen there is a sufficient absorptive capacity in the host country. Severalcountry studies have been carried out, providing diverging results on the roleof FDI spillovers with respect to stimulating economic growth. Whereaspositive effects from spillovers have been found for, H󰀑J󰀑 Mexico (Blomströmand Persson, 1983; Blomström and Wolff, 1994; Kokko, 1994), Uruguay(Kokko, Tansini, and Zejan, 1996) and Indonesia (Sjöholm, 1999b), nospillovers were traced in studies for Morocco (Haddad and Henderson, 1993)and Venezuela (Aitken and Harrison, 1999). These diverging results mayunderline the crucial role of certain host country characteristics necessary to letFDI contribute positively to economic growth through spillovers.

Some authors argue that the adoption of new technologies andmanagement skills requires inputs from the labour force. High-level capitalgoods need to be combined with labour that is able to understand and workwith the new technology. Of course, the skills required to work with high-levelcapital goods can be learnt. This, however, demands a minimum educationallevel in order to be able to learn to work with new technologies. Therefore,technological spillover is possible only when there is a certain minimum, or‘threshold’ level of human capital available in the host country (Borensztein, HWDO󰀑, 1998). This suggests that FDI and human capital are complementary in theprocess of technological diffusion.

Other authors argue that the process of technological spillovers may bemore efficient in the presence of well-functioning markets. Under thesecircumstances, the environment in which FDI operates ensures competitionand reduces market distortions, enhancing the exchange of knowledge amongfirms (Bhagwati, 1978 and 1985; Ozawa, 1992; Balasubramanyam, HW󰀁DO󰀑,1996). In terms of the channels of spillovers discussed above, well-functioning markets provide possibilities for competition and linkage effects ofFDI.

Some authors stress that the establishment of property rights – inparticular intellectual property rights – is crucial to attract high technology FDI(Smarzynska, 1999). If intellectual property rights are only weakly protected in

6

a country, foreign firms will undertake low technology investments, whichreduces the opportunities for spillover effects and improvements ofproductivity of domestic firms.

These characteristics may indeed be important to promote the use ofabsorptive capacity of a country with respect to maximising technologyspillovers from foreign firms. Yet, this paper argues that one crucialcharacteristic of the environment in the host country has not been mentioned inthe literature󰀏󰀁L󰀑H󰀑 the development of the domestic financial system. When wereconsider the different channels through which technology spillover may takeplace, it becomes clear that in many cases domestic firms will need to investwhen upgrading their own technology or adopting new technologies, basedeither on a demonstration effect, a competition effect, and/or a linkage effect.The same holds in case they aim at upgrading the skills of their employees (thetraining effect). These investments should be financed, however.

The development of the domestic financial system at least partlydetermines to what extent domestic firms may be able to realise theirinvestment plans in case external finance from banks or stock markets isneeded. Moreover, the development of the financial system also influences theallocative efficiency of financial resources over investment projects. Thus, thefinancial system may contribute to economic growth through two mainchannels (next to providing and maintaining a generally accepted means ofexchange). First, it mobilises savings; this increases the volume of resourcesavailable to finance investment. Second, it screens and monitors investmentprojects (L󰀑H󰀑 lowering information acquisition costs); this contributes toincreasing the efficiency of the projects carried out (see H󰀑J󰀑 Greenwood andJovanovic, 1990; Levine, 1991; Saint-Paul, 1992).2 The more developed thedomestic financial system, the better it will be able to mobilise savings, andscreen and monitor investment projects, which will contribute to highereconomic growth.

Moreover, investment related to upgrade existing or adopt newtechnologies is more risky than other investment projects. The financial system

2

See Levine (1997) or Berthelémy and Varoudakis (1996) for good surveys on the role of thedomestic financial system and its relationship to economic growth.

7

in general, and specific financial institutions in particular, may help to reducethese risks, thereby stimulating domestic entrepreneurs to actually undertakethe upgrading of existing technology or to adopt new technologies introducedby foreign firms. Thus, financial institutions positively affect the speed oftechnological innovation, thereby enhancing economic growth (Huang and Xu,1999). This argument also holds for technological innovation that results fromone or more of the channels of technology spillovers from FDI as describedabove. The more developed the domestic financial system, the better it will beable to reduce risks associated with investment in upgrading old and/or newtechnologies.

Finally, the development of the domestic financial system may alsodetermine to what extent foreign firms will be able to borrow in order toextend their innovative activities in the host country, which would furtherincrease the scope for technological spillovers to domestic firms. FDI asmeasured by the financial flow data may be only part of the FDI to developingcountries, as some of the investment is financed through debt and/or equityraised in financial markets in the host countries (Borensztein HW󰀁DO󰀑, 1998,p.134). Thus, the availability and quality of domestic financial markets alsomay influence FDI and its impact on the diffusion of technology in the hostcountry. This diffusion process may be more efficient once financial marketsin the host country are better developed, since this allows the subsidiary of aMNC to elaborate on the investment once it has entered the host country.

Therefore, in conclusion, FDI and domestic financial markets are

complementary with respect to enhancing the process of technologicaldiffusion, thereby increasing the rate of economic growth. This hypothesis canbe tested empirically, which will be the subject of the next two sections.󰀖󰀑󰀁'$7$󰀁$1'󰀁0(7+2'2/2*<󰀁2)󰀁(03,5,&$/󰀁,19(67,*$7,21The data set used in this paper applies to the 1970-1995 period and contains 67LDCs (see Appendix 2 for a complete list of the countries). For this set ofcountries data is available for all variables used in this study, which means thatthe estimations have been carried out with a balanced data set. Table 1provides basic descriptive statistics for the dependent variable, L󰀑H󰀑 the per

8

capita growth rate (3&*52:7+) and the crucial variable in this study, L󰀑H󰀑gross foreign direct investment inflows as a percentage of GDP 󰀋)',). Bothvariables (and all other variables in this study) are annual averages for the1970-1995 period.

The table shows that 3&*52:7+󰀁and )', are not normally

distributed. The distribution of these variables is skewed. With respect to3&*52:7+, 30 countries have an average growth rate varying between 0 and2 percent, for 13 countries the growth rate is between 2 and 4, for 5 the growthrate is above 4 percent, and for 19 countries the growth rate is negative. For 42countries foreign direct investment as a percentage of GDP is between 0 and 1,for 17 countries it is between 1 and 2, for 5 countries between 2 and 3, and for3 countries between 4 and 5. The largest recipients of foreign direct investmentas a percentage of GDP are Swaziland, Trinidad and Tobago and Malaysia.

The methodology of the empirical investigation follows the

voluminous growth regression literature, which was stimulated by the seminalpaper of Barro (1991). Unfortunately, theory does not provide clear guidanceconcerning the set of variables that should be included in the growth equation.Depending on the aim of the study and the insights and beliefs of the author(s),different explanatory variables have been included and found to be significantin the literature. Recently, some studies have shown that only a few variableshave a robust effect on economic growth (see H󰀑J󰀑 Levine and Renelt, 1992 andKing and Levine, 1993), implying the importance of stability tests. Sala-i-Martin (1997a and 1997b) provides a useful method to test for the robustnessof different variables in explaining economic growth. The empirical analysis inthis paper closely follows his approach. In particular, the regression analysisfor the cross-section of 67 countries is specified as follows:

3&*52:7+󰀁 󰀁αvÃ󰀎󰀁βv󰀏w,󰀁󰀎󰀁β󰁐w0󰀁󰀎󰀁β󰁝󰀏w=󰀁󰀎󰀁H

9

(1),

where ,󰀏󰀁0 and = are vectors of variables and H is an error term. , is a vector ofvariables that are “generally accepted” to be important to explain economicgrowth. 0 is a vector of variables containing the variables of interest in thisstudy. In this study these variables are the log of the FDI to GDP ratio (/)',),and /)', interacted with the log of the private sector bank loans to GDP ratio(/&5('3). /&5('3 is chosen here as a measure of financial development(see below). The vector of = variables contains a limited number of variablesfrom a large set of variables that have been used in the literature to explain percapita economic growth. These variables are used as control variables in theestimations.

The vector of ,󰀁variables contains variables that, according to Levineand Renelt (1992), and King and Levine(1993), have a robust effect oneconomic growth. These variables are: the log of the initial level of thesecondary enrolment rate 󰀋/6(&(15󰀌󰀏󰀁the log of the initial level of GDP percapita (/*'33&), the variable proxying for financial market developmentover the 1970-1995 period (/&5('3) and the log of the investment share inGDP (/,19*'3).

The choice of this vector of , variables needs some further explanation./6(&(15 measures human development and the introduction of /*'33&reflects the process of catch up.

With respect to the choice of the financial development variable, wenote that several variables have been suggested in the literature to measurefinancial development, depending on the specific characteristics of thefinancial system of interest. These variables focus on the size, the efficiencyand/or the relative importance of different financial intermediaries in the totalfinancial system. The problem is that for several of these variables data areonly available for a limited number of countries. Therefore in the analysis thelog of credit to the private sector as a percentage of GDP (/&5('3) is used tomeasure financial development, since for this variable data are available for all

10

countries in the data set. Moreover, this variable is used in several otherstudies (see H󰀑J󰀑 Demirgüç-Kunt and Levine, 1996).3

With respect to /,19*'3, regression models are estimated including

and excluding this variable in the vector of , variables. The reason for this isthat the interpretation of a significant coefficient for a certain variable [depends on whether or not /,19*'3 is included in the regression model. If/,19*'3 is included and the coefficient of variable [ is significant, this isinterpreted as [ affecting growth via the “level of efficiency”. If󰀁/,19*'3 isnot included, it is unclear whether variable [ affects growth via investment orvia efficiency. This distinction is of importance to obtain more informationwith respect to how exactly )',󰀁is related to economic growth.

Table 2 presents the correlation matrix for the , variables,3&*52:7+ and /)',.

󰀗󰀑󰀁5(68/76󰀁2)󰀁7+(󰀁(03,5,&$/󰀁,19(67,*$7,21

The analysis starts by estimating a number of base equations, L󰀑H󰀑 = variablesare not yet included in the regression models. The results of these estimationsare presented in table 3 (without /,19*'3) and table 4 (with /,19*'3). Thesecond column in both tables shows the relevance of including the different ,variables in the model. The tables show that /*'33&󰀏󰀁/6(&(15󰀏󰀁/&5('3and /,19*'3 have a significant impact on economic growth. In the thirdcolumn /)', is added to this equation. This variable does not have asignificantly positive direct effect on economic growth. This confirms the viewthat without additional requirements FDI does not enhance economic growthof a country. Borensztein, De Gregorio and Lee (1998) suggest that FDI is only

3

󰀁In the analysis the log of the average money and quasi money to GDP ratio over the 1970-1995 period󰀁󰀋/0*'3󰀌 has also been used to measure financial development. The estimationresults are very much in line with the results for /&5('3. Results can be obtained from theauthors on request.

11

effective in countries with a high level of human capital. We have tested thishypothesis and the results are presented in column three of tables 3 and 4.Following Borensztein, De Gregorio and Lee (1998) we include the interactiveterm /)',*/6(&(15 in the regression model. If we differentiate the modelpresented in column three of both tables with respect to /)',, we get:/)', = -0.701󰀎0.351󰀍/6(&(15

(model without /,19*'3); and

/)',󰀁 = -0.617+0.289󰀍/6(&(15

(model with /,19*'3).

These outcomes confirm the hypothesis offered by Borensztein, De Gregorioand Lee (1998), L󰀑H󰀑 the growth effects of FDI depend on the level of humandevelopment of a country. The threshold level of /6(&(15 above which/)', has a positive effect on economic growth can be calculated by setting thefirst derivative of the above equations equal to zero. The threshold levels thenequal: (0.701/0.351)=1.997 and (0.617/0.289)=2.135. Since /6(&(15 is thelogarithm of the secondary school enrolment rate, the results imply that /)',(and hence also )',) will have a positive effect on growth in countries wherethe secondary enrolment rate is above 7.4, for model without /,19*'3 and8.5, for model with /,19*'3.

As explained above, the aim of this paper is to empirically investigatethe hypothesis that FDI and domestic financial markets are complementarywith respect to enhancing the process of technological diffusion, therebyincreasing the rate of economic growth. Therefore, the empirical analysisfocuses on the variables /)', and the interactive term /)',*/&5('3, whichrepresent the vector of 0 variables as specified in equation (1). The modelpresented in column four of tables 3 and 4 directly tests the central hypothesis

12

of this paper. The outcomes in the tables show that the interactive term/)',󰀍/&5(' is positive and significantly related to the dependent variable3&*52:7+, whereas /)', alone is significantly negative. This supports theview that FDI only has a positive effect on economic growth if thedevelopment of the domestic financial system has reached a certain minimumlevel. Thus, we find preliminary support for the central hypothesis of thispaper.

It may be argued that the results presented in column four of tables 3and 4 are due to high multi-collinearity between /6(&(15 and /&5('3󰀁(seetable 2). This would mean that the results found are in fact due to the level ofhuman development in a country (L󰀑H󰀑 the hypothesis forwarded by BorenszteinHW󰀁DO󰀑), rather than due to the level of financial development. To furtherinvestigate this issue we estimate a model incorporating /)',,/)',*/&5('3, and /)',*/6(&(15. This model is presented in columnfive of both tables. If we concentrate on the results for the model including/,19*'3󰀁(table 4), the results of the estimation show that /)',*/&5('3remains significant; however, /)',*/6(&(15 becomes insignificant. Theseresults can be interpreted as follows. First, it again confirms the hypothesis thata certain level of financial market development is an important prerequisite forFDI to have a positive effect on economic growth. Second, it suggests that theimportance of a certain level of human capital as a prerequisite for the growtheffects of FDI is at least partly be explained by the existence of a well-developed financial sector. Moreover, the fact that the variable/)',*/&5('3 remains significant in the models where /,19*'3 isincluded suggests that FDI affects economic growth mainly via the level ofefficiency.4

The next step in the empirical analysis is to test the robustness of theabove results. In order to investigate this, we conduct a stability analysis in line

4

󰀁We have also explored the relationship between /)', and /)', interacted with a financialdevelopment variable as exogenous variables and total investment as a share of GDP as theendogenous variable. In line with Borensztein, De Gregorio and Lee (1998) it appears that/)', and /)', interacted with financial market development do not have a robust effect oninvestment levels. This confirms that FDI mainly affects growth via the level of efficiency.

13

with Sala-i-Martin (1997a and 1997b). This stability analysis tests whether thecoefficients for /)', and the interactive term /)',*/&5('3 remain robustafter adding a vector = of a limited number of control variables to the modelspresented in tables 3 and 4. We define a group of 14 variables from which theadditional control variables are taken. These variables are shown to beimportant for explaining economic growth in several other studies. Since weaim at using a fully balanced data set in our analysis, other possibly relevantvariables were not taken into account due to lack of observations. Theadditional variables we take into account in our analysis are $,'*'3(development aid as a percentage of GDP), %$1./ (bank and trade relatedlending as a percentage of GDP), %03 (black market premium), &,9/,%(index of civil liberties), '(%7*'3 (the external debt to GDP ratio), '(%76(total external debt service as a percentage of GDP), (,1)/ (uncertainty withrespect to inflation), (*29& (uncertainty with respect to governmentexpenditures), (;3*'3 (exports of goods and services as a percentage ofGDP), *29&*'3 (government consumption as a percentage of GDP), ,1)/(the annual inflation rate), 35,*+76 (index of political rights), 67',1)/ (thestandard deviation of the annual inflation rate), and 75$'( (exports plusimports to GDP).5 In all estimates discussed below, these variables have beentransformed into logarithmic form.

The stability test starts by determining all possible combinations of alimited number of the above-presented set of 14 variables. We have chosen toperform the stability test by adding combinations of three, respectively fourcontrol variables to the models discussed above. Next, we carry out regressionanalysis including all variables presented in column 4 from table 3,respectively table 4, as well as all possible combinations of three (respectivelyfour) control variables. This means that in case of three additional variables weestimate 14!/(11! 3!) = 364 different specifications of the model presented incolumn 3 of tables 3 and 4 (L󰀑H󰀑 with and without /,19*'3). In case we usefour additional variables the amount of different specifications equals14!/(10!4!) = 1,001.

5

See the appendix for the exact specification and data sources of these variables.

14

After having estimated all different equation specifications, the next stepof the stability test is to look at the distribution of the coefficients of theindividual equations, and calculate the fraction of the cumulative distributionfunction lying on each side of zero. By assuming that the distribution of theestimates of the coefficients is normal and calculating the mean and thestandard deviation of this distribution, the cumulative distribution function(&')) can be computed.

More precisely, if βw is the coefficient for a variable in the specification󰀁Mof the estimated model and σw is the standard error of the coefficient βw, weproxy the mean and the standard deviation of the distribution by:

Σ󰀁βMQ

β󰀁 󰀁

σ󰀁 󰀁

Σ󰀁σMQ.

The number of estimated equations is 364 (in case we add combinations ofthree = variables), respectively 1,001 (when we add combinations of four =variables). In table 5 the mean estimate is presented in the column entitled“COEF” and the mean standard deviation is given in the column entitled“STERR”.

Next, we calculate the fraction of the cumulative distribution functionlying on the right or left-hand side of zero, using a table for the (cumulative)normal distribution. The test statistic we use is defined as the mean over thestandard deviation of the distribution. The column entitled “CDF” in table 5denotes the larger of the two areas. Finally, as an additional stability test, thelast column of the table presents the percentage of all regressions for which thevariable of interest (L󰀑H󰀑 /)', or /)',󰀍/&5('3) is significant at the 95%level.

15

The results presented in table 5 show that the coefficients for /)', andthe interactive term /)',*/&5('3 are very robust. In the models including/,19*'3 as an additional , variable t-values for /)', and /)',*/&5('3are significant at the 95% level in all cases. These results strongly suggest thatFDI enhances economic growth only if domestic financial markets are well-developed, thus supporting the main hypothesis investigated in this paper.

What do the results of the analysis in this paper imply for the countriesin the data set? Like we did before with respect to the hypothesised interactiverelationship between human capital and FDI (Borensztein, De Gregorio andLee, 1998), we calculate the first-order condition of the growth equation withrespect to /)', and set this equation equal to zero. This allows us todetermine the threshold value of /&5('3 above which /)', starts to have apositive effect on growth. Our results suggest that FDI has a positive effect onper capita growth if /&5('3 exceeds 2.50 (when /,19*'3 is excluded fromthe basic model) and 2.53 (when /,19*'3 is included). Since all variableshave been transformed in a logarithmic form, these results imply that &5('3,L󰀑H󰀑 the private sector credit to GDP ratio should be larger than 12 per cent inorder for FDI to have a positive effect on growth. In our data set 37 out 67countries (or 55 per cent) satisfy this threshold value for &5('3. Table 6presents the countries for which the domestic financial system has reached asufficient level of development, L󰀑H󰀑 for these countries FDI contributespositively to economic growth. The table shows that for most Sub-SaharanAfrican countries it appears to be the case that the level of development oftheir domestic financial system is insufficient, so that FDI probably will nothave a positive impact on their economic growth.

16

󰀘󰀑󰀁&21&/86,216

FDI may help to raise economic growth in recipient countries. Yet, thecontribution FDI can make may strongly depend on the circumstances in therecipient countries. Few empirical studies have investigated the relationshipbetween FDI and economic growth and the role played by the circumstancesFDI is confronted with whenever it enters a recipient country. These studiesfocused on the role of human capital available in and the export-orientednessof the recipient country. The original contribution this paper makes is that itargues that the development of the financial system of the recipient country isan important precondition for FDI to have a positive impact on economicgrowth. A more developed financial system positively contributes to theprocess of technological diffusion associated with FDI.

The paper empirically investigates the role the development of thefinancial system plays in enhancing the positive relationship between FDI andeconomic growth. The empirical investigation presented in the paper stronglysuggest that this is the case. Of the 67 countries in data set, 37 have asufficiently developed financial system in order to let FDI contribute positivelyto economic growth. Most of these countries are in Latin America and Asia.Almost all other countries are in Sub-Saharan Africa. These countries havevery weak financial systems and consequently FDI does not contributepositively to growth.

The results of the empirical investigation in this paper provide anumber of policy-relevant conclusions. First, the results contradict the widelyaccepted view that an increase in FDI may important to enhance economicgrowth in Sub-Saharan Africa. This is only true after these countries haveimproved their domestic financial systems. Second, the analysis in this papermay contribute to the discussion on the order of economic liberalisation inLDCs. The outcomes of the empirical investigation suggest that thesecountries should first reform their domestic financial system before liberalisingthe capital account.

The investigation in this paper is a first step into the analysis of the rolethe domestic financial system plays in making FDI contribute to economicgrowth in LDCs. Of course, we would like to determine more precisely how

17

financial system development influences the relationship between FDI andgrowth. One obvious way to proceed is to extend the empirical research, whichmay focus on using alternative indicators, representing different functions ofthe domestic financial system in LDCs.

18

5()(5(1&(6

Aitken, Brian J., and Ann E. Harrison, “Do Domestic Firms Benefit fromDirect Foreign Investment? Evidence from Venezuela,” $PHULFDQ󰀁(FRQRPLF5HYLHZ, 89, 3, 1999, pp. 605-618.

Balasubramanyam, V.N., M. Salisu, and David Sapsford, “Foreign DirectInvestment and Growth in EP and IS Countries,” 7KH󰀁(FRQRPLF󰀁-RXUQDO, 106,1, 1996, pp. 92-105.

Barro, Robert J., “Economic growth in a cross-section of countries,” 4XDUWHUO\\-RXUQDO󰀁RI󰀁(FRQRPLFV, 106, 2, 1991, pp. 407-443.

Barro, Robert J., and JungWha Lee, Data Set for a Panel of 138 Countries,Cambridge Mass., NBER, 1994.

Barro, Robert J., and Xavier Sala-I-Martin, (FRQRPLF󰀁*URZWK, CambridgeMA, McGraw-Hill, 1995.

Berthelemy, Jean-Claude, and Aristomene Varoudakis, “Models of FinancialDevelopment and Growth: A Survey of Recent Literature,” in: Niels Hermesand Robert Lensink, (eds.), )LQDQFLDO󰀁'HYHORSPHQW󰀁DQG󰀁(FRQRPLF󰀁*URZWK󰀝7KHRU\\󰀁DQG󰀁([SHULHQFHV󰀁IURP󰀁'HYHORSLQJ󰀁&RXQWULHV, London, Routledge,1996, pp. 7-34.

Bhagwati, Jagdish N., “Anatomy and Consequences of Exchange RateRegimes,” 6WXGLHV󰀁LQ󰀁,QWHUQDWLRQDO󰀁(FRQRPLF󰀁5HODWLRQV, 10, New York,NBER, 1978.

Blomström, Magnus, and Hakan Persson, “Foreign Investment and SpilloverEfficiency in an Underdeveloped Economy: Evidence from the MexicanManufacturing Industry,” :RUOG󰀁'HYHORSPHQW, 11, 6, 1983, pp. 493-501.

19

Blomström, Magnus, Edward N. Wolff, “Multinational Corporations andProductivity Convergence in Mexico,” in: William Baumol, Richard Nelson,and Edward N. Wolff (eds.), &RQYHUJHQFH󰀁RI󰀁3URGXFWLYLW\\󰀝󰀁&URVV󰀐1DWLRQDO6WXGLHV󰀁DQG󰀁+LVWRULFDO󰀁(YLGHQFH, Oxford, Oxford University Press, 1994.Bo, Hong, Empirics of the Investment-Uncertainty Relationship, unpublishedmanuscript, Groningen, University of Groningen, 1999.

Borensztein, Eduardo, Jose De Gregorio, and Jung Wha Lee, “How DoesForeign Direct Investment Affect Economic Growth?” -RXUQDO󰀁RI,QWHUQDWLRQDO󰀁(FRQRPLFV, 45, 1, 1998, pp. 115-135.

De Mello, Jr., Luiz R., “Foreign Direct Investment in Developing Countriesand Growth: A Selective Survey,” 7KH󰀁-RXUQDO󰀁RI󰀁'HYHORSPHQW󰀁6WXGLHV, 34, 1,1997, pp. 1-34.

Demirgüç-Kunt, Asli, and Ross Levine, “Stock Market Development andFinancial Intermediaries: Stylized Facts,” :RUOG󰀁%DQN󰀁(FRQRPLF󰀁5HYLHZ, 10,2, 1996, pp. 291-321.

Findlay, Ronald, “Relative Backwardness, Direct Foreign Investment and theTransfer of Technology: A Simple Dynamic Model,” 4XDUWHUO\\󰀁-RXUQDO󰀁RI(FRQRPLFV, 92, 1, 1978, pp. 1-16.

Greenwood, Jeremy, and Boyan Jovanovic, “Financial Development, Growth,and the Distribution of Income,” -RXUQDO󰀁RI󰀁3ROLWLFDO󰀁(FRQRP\\, 98, 5, 1990,pp. 1076-1107.

Grossman, Gene M., and Elhanan Helpman, ,QQRYDWLRQ󰀁DQG󰀁*URZWK󰀁LQ󰀁WKH*OREDO󰀁(FRQRP\\, Cambridge MA, MIT Press, 1991.

Haddad, Mona, and Ann Harrison, “Are There Positive Spillovers from DirectForeign Investment? Evidence from Panel Data for Morocco,” -RXUQDO󰀁RI

20

'HYHORSPHQW󰀁(FRQRPLFV, 42, 1, 1993, pp. 51-74.

Huang, Haizhou, and Chenggang Xu, “Institutions, Innovations, and Growth,”$PHULFDQ󰀁(FRQRPLF󰀁5HYLHZ, 89, 2, 1999, pp. 438-443.

King, Robert G., and Ross Levine, “Finance and Gowth: Schumpeter Might beRight,” 4XDUWHUO\\󰀁-RXUQDO󰀁RI󰀁(FRQRPLFV, 108, 3, 1993, pp. 717-737.

Kinoshita, Yuko, Technology Spillovers Through Foreign Direct Investment,unpublished working paper, Prague, CERGE-EI, 1998.

Kokko, Ari, “Technology, Market Characteristics, and Spillovers,” -RXUQDO󰀁RI'HYHORSPHQW󰀁(FRQRPLFV, 43, 2, 1994, pp. 279-293.

Kokko, Ari, Ruben Tansini, and Mario C. Zejan, “Local TechnologicalCapability and Productivity Spillovers from FDI in the UruguayanManufacturing Sector,” 7KH󰀁-RXUQDO󰀁RI󰀁'HYHORSPHQW󰀁6WXGLHV, 32, 4, 1996, pp.602-11.

Levine, Ross, “Financial Development and Economic Growth: Views andAgenda, -RXUQDO󰀁RI󰀁(FRQRPLF󰀁/LWHUDWXUH, 35, 2, 1997, pp. 688-726.

Levine, Ross, “Stock Markets, Growth, and Tax Policy,” -RXUQDO󰀁RI󰀁)LQDQFH,46, 4, 1991, pp. 1445-65.

Levine, Ross, and David Renelt, “A Sensitivity Analysis of Cross-CountryGrowth Regressions,” $PHULFDQ󰀁(FRQRPLF󰀁5HYLHZ, 82, 4, 1992, pp. 942-963.Lichtenberg, Frank R., and Bruno van Pottelsberghe de la Potterie,“International R&D Spillovers,” (XURSHDQ󰀁(FRQRPLF󰀁5HYLHZ, 42 8, 1998, pp.1483-1491.

21

Ozawa, Terutomo, “Cross-Investments between Japan and the EC: IncomeSimilarity, Technological Congruity and Economies of Scope,” in JohnCantwell (ed.), 0XOWLQDWLRQDO󰀁,QYHVWPHQW󰀁LQ󰀁0RGHUQ󰀁(XURSH󰀝󰀁WUDWHJLF,QWHUDFWLRQ󰀁LQ󰀁WKH󰀁,QWHJUDWHG󰀁&RPPXQLW\\, Aldershot, Edward Elgar, 1992, pp.13-45.

Saint-Paul, Gilles, “Technological Choice, Financial Markets and EconomicDevelopment,” (XURSHDQ󰀁(FRQRPLF󰀁5HYLHZ, 36, 4, 1992, pp 763-81.

Sala-I-Martin, Xavier, “I just ran two million regressions,”󰀁$PHULFDQ(FRQRPLF󰀁5HYLHZ, 87, 2, 1997a, pp. 178-183.

Sala-I-Martin, Xavier, I just ran four million regressions, unpublishedmanuscript, Colombia University and Universitat Pompeu Fabra, 1997b.Sjöholm, Fredrik, “Productivity Growth in Indonesia: The Role of RegionalCharacteristics and Direct Foreign Investment,” (FRQRPLF󰀁'HYHORSPHQW󰀁DQG&XOWXUDO󰀁&KDQJH, 47, 3, 1999a, pp. 559-584.

Sjöholm, Fredrik, “Technology Gap, Competition and Spillovers from DirectForeign Investment: Evidence from Establishment Data󰀏”󰀁7KH󰀁-RXUQDO󰀁RI

'HYHORSPHQW󰀁6WXGLHV, 36, 1, 1999b, pp. 53-73.

Smarzynska, Beata K., Composition of Foreign Direct Investment and

Protection of Intellectual Property Rights in Transition Economies,unpublished working paper, New Haven, Yale University, 1999.World Bank, :RUOG󰀁'HYHORSPHQW󰀁,QGLFDWRUV󰀁󰀔󰀜󰀜󰀚, Washington DC, TheWorld Bank, 1997.

22

$33(1',;󰀁,󰀝󰀁/,67󰀁2)󰀁9$5,$%/(6󰀁86('󰀁,1󰀁7+(󰀁$1$/<6,6󰀋VHFWLRQV󰀁󰀖󰀁DQG󰀁󰀗󰀌$,'*'3 %$1./ %03

= development aid as a percentage of GDP

= bank and trade related lending as a percentage of GDP= black market premium, calculated as (black market rate/official rate)-1.

&,9/,% = index of civil liberties

&5(',735 = credit to the private sector as a percentage of GDP'(%7*'3 = the external debt to GDP ratio'(%76 = total external debt service as a percentage of GDP(,1)/ uncertainty with respect to inflation(*29&= uncertainty with respect to government expenditures(;3*'3 = exports of goods and services as a percentage of GDP)', = foreign direct investment as a percentage of GDP*'33&= GDP per capita in 1970

*29&*'3 = government consumption as a percentage of GDP,1)/ = the annual inflation rate,19*'3 = average investment to GDP ratio over 1970-1995 period0*'3 = average money and quasi money to GDP ratio over the 1970- 1995 period

3&*52:7+ = average real per capita growth rate over 1970-1995 period.35,*+76 = index of political rights6(&(15󰀁= secondary school enrolment rate in 197067',1)/ = the standard deviation of the annual inflation rate, calculated

from the inflation figures

75$'(󰀁= exports plus imports to GDP; measure of the degree of

opennessThe source for all variables is World Bank (1997), which is available on CD-ROM, except for %03󰀏󰀁&,9/,%󰀁and󰀁35,*+76. These variables are obtainedfrom the data set created by Barro and Lee (1994). Moreover, (,1)/ and(;3*'3 have been calculated by the authors (see below). The variables from

23

Barro and Lee (1994) refer to averages for the 1970-1990 period. Unlessotherwise stated, all other variables refer to averages over 1970-1995 period.For all variables logarithmic transformations are used.

We need to explanation how the uncertainty variables (,1)/ and (*29&have been constructed. Both variables are constructed by using the standarddeviation of the unpredictable part of ,1)/ and *29&; see Bo (1999) for asurvey of different methods to measure uncertainty. We first specify andestimate a forecasting equation to determine the expected part of ,1)/󰀁and*29&. The standard deviation of the unexpected part of ,1)/󰀁and *29& (L󰀑H󰀑the residuals from the forecasting equation) is used as a measure ofuncertainty. We have used a second-order autoregressive process, extendedwith a time trend, as the forecasting equation:3󰁗󰀁󰀁 󰀁󰀁D 󰀁󰀎󰀁󰀁D!7󰀁󰀁󰀎󰀁D\")',󰁗󰀐 󰀁󰀎󰀁D#)',󰁗󰀐!󰀁󰀎󰀁H󰁗󰀏

where 3󰁗 is the variable under consideration, 7 is a time trend, D is anintercept, D\" and D# are the autoregressive parameters and H󰁗 is an error term.We estimate the above equation for all countries in the data set. By calculatingthe standard deviation of the residuals for the entire sample period for eachindividual country, we obtain the variables (,1)/󰀁and (*29&.

24

$33(1',;󰀁,,󰀝󰀁&28175,(6󰀁,1󰀁7+(󰀁'$7$󰀁6(7

$IULFD󰀝

Algeria, Benin; Burkina Faso; Burundi; Cameroon; Cape Verde; CentralAfrican Rep.; Chad; Egypt; Gabon; Gambia; Ghana; Guinea-Bissau; Coted’Ivoire; Kenya; Lesotho; Madagascar; Mali; Mauritania; Morocco; Niger;Nigeria; Rwanda; Senegal; Sierra Leone; Somalia; Sudan; Swaziland; Togo;Tunisia; Zambia; Zimbabwe

6RXWK󰀁$PHULFD󰀝

Barbados; Costa Rica; Dominican Rep.; El Salvador; Guatemala; Haiti;Honduras; Jamaica; Mexico; Nicaragua; Panama; Trinidad and Tobago;Argentina; Bolivia; Chile; Colombia; Ecuador; Paraguay; Peru; Uruguay;Venezuela

$VLD󰀁DQG󰀁RWKHUV

Bangladesh; China; India; Malaysia; Nepal; Pakistan; Philippines; Sri Lanka;Syria; Thailand; Hungary; Malta; Fiji; and Papua New Guinea

25

26

7DEOH󰀁󰀔󰀝󰀁'HVFULSWLYH󰀁VWDWLVWLFV󰀁IRU󰀁SHU󰀁FDSLWD󰀁JURZWK󰀁DQG󰀁)',

3&*52:7+0HDQ0HGLDQ0D[LPXP0LQLPXP6WDQGDUG󰀁'HYLDWLRQ6NHZQHVV.XUWRVLV0.9380.5296.832-3.1341.9230.7194.120)',0.9980.6934.6980.0030.9891.8246.68627

7DEOH󰀁󰀕󰀝󰀁&RUUHODWLRQ󰀁PDWUL[

Q8BSPXUC

GA9D

GT@8@IS

GDIWB9Q

G8S@9Q

GB9QQ8

Q8BSPXUC

10.210.430.580.48-0.10

10.300.290.380.37

10.270.530.52

10.520.18

10.45

1

GA9D

GT@8@IS

GDIWB9Q

G8S@9Q

GB9QQ8

28

7DEOH󰀁󰀖󰀝󰀁'LUHFW󰀁LQYHVWPHQW󰀁DQG󰀁HFRQRPLF󰀁JURZWK

1

/*'33&

-1.182(-5.15)

/6(&(15

0.891(4.47)

/&5('3

1.562(4.12)

/)',

2-1.238(-5.29)0.882(4.45)1.471(3.97)0.156(1.11)

/)',󰀍/6(&(15

3-1.318(-5.39)1.176(5.20)1.313(3.47)-0.701(-2.14)0.351(2.89)

/)',󰀍/&5('3

0.621(2.85)

&

1.376(1.01)

R2F

0.4619.94

2.127(1.47)0.4615.32

2.253(1.58)0.5014.26

0.943(0.60)0.5114.534-1.180(-5.20)0.740(3.75)1.827(4.59)-1.587(-2.60)

5-1.247(-4.98)0.964(4.09)1.620(4.02)-1.574(-2.68)0.215(1.69)0.429(1.86)1.386(0.85)0.5112.46

Dependent variable:󰀁3&*52:7+󰀑 Amount of observations in all regressions: 67.

29

7DEOH󰀁󰀗󰀝󰀁)',󰀁DQG󰀁HFRQRPLF󰀁JURZWK󰀝󰀁HIIHFWV󰀁YLD󰀁HIILFLHQF\\

1

/*'33&

-1.117(-6.83)

/6(&(15

0.868(5.40)

/,19*'3

2.539(5.56)

/&5('3

0.843(2.46)

/)',

2-1.148(-6.33)0.863(5.43)2.488(5.55)0.807(2.40)0.085(0.65)

/)',󰀍/6(&(15

3-1.220(-6.90)1.106(5.61)2.352(4.82)0.713(1.99)-0.617(-1.84)0.289(2.29)

/)',󰀍/&5('3

0.685(3.45)

&

-4.437(-3.07)

R2F

0.5924.43

-3.910(-2.48)0.5819.47

-3.477(-2.27)0.6118.00

-5.473(-3.44)0.6420.404-1.081(-7.14)0.706(4.64)2.594(5.60)1.171(3.23)-1.839(-3.23)

5-1.113(-6.86)0.805(4.03)2.536(5.25)1.095(2.86)-1.828(-3.28)0.095(0.807)0.599(2.91)-5.125(-3.04)0.6317.36

Dependent variable:󰀁3&*52:7+󰀑 Amount of observations in all regressions: 67.

30

7DEOH󰀁󰀘󰀝󰀁6WDELOLW\\󰀁WHVW

IVH7@S

S󰀕

8P@A

TU@SS

89A

Q@S8

Without /,19*'3 in the base model/)',/)',󰀍/&5('3/)',/)',󰀍/&5('3 364 3641,0011,0010.570.570.580.58-1.348 0.534-1.310 0.5190.5440.1940.5350.1920.9930.9970.9930.9971.0000.9970.9210.983With /,19*'3 in the base model/)',/)',󰀍/&5('3/)',/)',󰀍/&5('3 364 3641,0011,0010.720.720.730.73-1.620 0.615-1.528 0.6030.5050.1730.5000.1721.0001.0001.0001.0001.0001.0001.0001.000180%(5 denotes the number of equations tested.

31

7DEOH󰀁󰀙󰀝󰀁5HODWLRQVKLS󰀁EHWZHHQ󰀁)',󰀁DQG󰀁JURZWK󰀁DQG󰀁WKH󰀁UROH󰀁RI󰀁WKH󰀁OHYHO󰀁RI󰀁GHYHORSPHQWRI󰀁WKH󰀁GRPHVWLF󰀁ILQDQFLDO󰀁V\\VWHP

No positive effect of /&5('3 on relationship between /)',󰀁and 3&*52:7+$)5,&$󰀝Algeria; Benin; Burkina Faso; Burundi; Cameroon; Cape Verde; Central African Rep.;Chad; Gabon; Gambia;; Guinea-Bissau; Cote d’Ivoire; Kenya; Lesotho; Madagascar; Mali;Mauritania; Niger; Nigeria; Rwanda; Senegal; Sierra Leone; Somalia; Sudan; Togo;Zimbabwe/$7,1󰀁$0(5,&$󰀝Guatemala; Haiti

$6,$󰀁$1'󰀁27+(5󰀁&28175,(6󰀝Nepal; Papua New Guinea

Positive effect of /&5('3 on relationship between /)',󰀁and 3&*52:7+$)5,&$󰀝Egypt; Ghana; Morocco; Swaziland; Tunisia; Zambia/$7,1󰀁$0(5,&$󰀝Barbados; Costa Rica; Dominican Rep.; El Salvador; Honduras; Jamaica; Mexico;Nicaragua; Panama; Trinidad and Tobago; Argentina; Bolivia; Chile; Colombia; Ecuador;Paraguay; Peru; Uruguay; Venezuela$6,$󰀁$1'󰀁27+(5󰀁&28175,(6󰀝Bangladesh; China; India; Malaysia; Pakistan; Philippines; Sri Lanka; Syria; Thailand;Hungary; Malta; Fiji

32