the attractiveness of countries to foreign direct … · the present paper is concerned with the...

– 113 –

Abstract:

The present paper is concerned with the attractiveness of countries to directforeign investments, that is, the host country characteristics that attract directforeign investment (FDI). It focuses on two types of national characteristics—those that attract inflows of all foreign investment (intermediate products),including FDI, that is, mobility factors; and those that influence the modality ofthese inflows, that is, reflect the preference for FDI rather than other forms offoreign investment or ‘straight (unbundled) imports of intermediate products byindigenous firms (modality factors)’. The paper reports preliminary findingsfrom a study of plausible determinants of FDI inflows into a sample of 25countries.

Keywords:FOREIGN DIRECT INVESTMENT (FDI); ATTRACTIVENESS; INVESTMENTMOBILITY FACTORS; INVESTMENT MODALITY FACTORS.

† Faculty of Business, School of Economics and Finance, Queensland University ofTechnology, Gardens Point Campus, GPO Box 2434, Brisbane QLD 4001.

‡ School of Economics and Management, University College, Australian DefenceForce Academy, University of New South Wales, Canberra ACT 2600; E-mail:[email protected]

We are indebted to Professor Wolfgang Kasper of The University of New South Wales, whoseenthusiasm for the subject and several publications on the attractiveness of countries inspiredour interest in this field of study (see Kasper 1991, 1993). We also wish to thank: Drs On KitTam, Paul Robertson and Paul McGavin for many useful comments on the earlier draft of thispaper; Mr Sam Jegatheswaran for assisting us with the data collection; and Mrs Firzia Pepperfor producing this paper.

Australian Journal of Management, Vol. 21, No. 2, December 1996, © The University of New South Wales

2The Attractiveness of Countries toForeign Direct Investors

bySharon Jackson †Stefan Markowski ‡

AUSTRALIAN JOURNAL OF MANAGEMENT December 1996

– 114 –

1 . Introduction

According to UNCTAD’s 1994 World Investment Report, the global flow offoreign direct investment (FDI) reached US$195 billion in 1993 (Financial

Review 31 Aug. 1994) after peaking at US$232 billion in 1990. As globalisation ofbusiness activity grows, there is considerable business interest in broad brushevaluations of countries as hosts to FDI and in country- and industry-specificstrategic risk appraisals pertinent to FDI decision making. Publications such as TheInstitutional Investor or The World Competitiveness Report (IMD-WEF) andcountless country profiles generated by management consultants, academics,financial institutions, creditworthiness rating agencies and the media attest to thisgeneral interest.

These publications provide information on a broad array of variables that areof interest to businessmen considering overseas production. For example, theannual IMD-WEF survey, The World Competitiveness Report, which is intendedto help investors evaluate country locations and to judge national environments interms of their attractiveness to foreign investment, presents information on over300 criteria of ‘competitiveness’ for 34 countries (eg, IMD-WEF 1991).

On the other hand, there is a well-established academic literature on thedeterminants of FDI. In this literature, testing of hypotheses is usually throughregression analysis, and only a few variables are suggested as plausibledeterminants at any one time. Variables representing market size and growth,labour costs and trade policy usually recur for testing although, more recently,variables such as proximity have been included. We contend however, that whilethese variables may be significant to FDI flows they will also be significant tointernational capital flows that are not direct. To understand FDI flows we need todistinguish between types of foreign investment, or more broadly, between typesof internationally traded intermediate goods and services. Explanation of FDIflows must consider not only factors that determine the mobility of intermediateproducts across national borders, but those factors that influence the modality ofthese flows and that result in firms selecting FDI as the preferred mode.

Our aim in this paper is to take as our starting point the information thatmanagers themselves are selecting as relevant (eg, data from The WorldCompetitiveness Report) and to use discriminant analysis to help identify thevariables from such a list that explain/predict a classification of FDI hosts as high-inflow countries, low-inflow countries and so on. Discriminant analysis allows usto include more variables as potential attractors than is possible with regressionanalysis. We are thus able to ensure that, as well as the usual variables, variablesrepresenting modality are included. Our interest is firstly in the degree ofcorrespondence between this source of information and the literature andsecondly, better identification of modality and mobility influences on foreigncapital flows. Also, our aim is to measure ‘attractiveness’ of countries to foreigndirect investors. A number of business-oriented publications contain composite‘indices’ of country attractiveness to foreign investors, which are weightedaverages of various socio-economic attributes but which cannot be tested againstsome independent measure of investment activity. Since, ultimately, investors votewith their feet, as it were, we selected a testable measure of internationalattractiveness of countries to foreign direct investors.

Vol. 21, No. 2 Jackson & Markowski: FOREIGN DIRECT INVESTORS

– 115 –

A multi-stage screening procedure was adopted by the authors to identifythe variables which best explain differences between countries as hosts to FDI.This procedure explains the structure of the present paper. First, the relevantliterature was scanned to identify the specific variables that had earlier beenhypothesised and tested. Second, a conceptual (theoretical) framework (fordetailed description see Jackson & Markowski 1993), that distinguished betweenthe mobility and modality aspects of FDI flows was used to draw a preliminary listof FDI ‘attractors’. The final list of independent variables was then drawn for thesubsequent statistical analysis by combining the preliminary short list of candidatemeasures with actual variables used by FDI investors. Third, factor anddiscriminant analyses were used to (a) determine if a list of forty plausibledeterminants of FDI flows could be reduced to a smaller number of underlyingfactors; and (b) test which of the short listed variables contributed most to theseparation of sample countries into groups differentiated with respect to levels orshares of FDI received. Various groupings of plausible national attractors of FDIwere identified by means of discriminant analysis. A brief discussion of thesefactors concludes this paper.

The sample of 25 host countries is shown in table 1. Its selection wasdetermined by the availability and completeness of data. However, the samplecontains a reasonable cross-section of developed and developing countries thatare FDI hosts. It excludes non-developed economies, the former and presentCommunist countries (in which FDI inflows have been strictly Government-controlled) and countries openly discouraging inflows of FDI.

2 . Attractors of FDI Inflows—Survey of LiteratureTesting for the determinants of FDI has a long and rich history, and many variableswith varied specifications have been suggested as determinants. Appendix Ashows various specifications of FDI used in the literature. The least ambiguousmeasures of FDI appear to be based on physical activities of affiliates of foreignfirms in a host economy, for example, the volume of output produced by affiliatesor the number of foreign affiliates located in the host country. However, theavailability of such data is very restricted and most common measures of FDI aremonetary. These tend to be derived from the balance of payments statistics as aspecial category of (long term) capital transaction entailing an element ofownership and control of foreign production. Problems arise, though, with regardto the interpretation of such data (see section 3).

The empirical literature concerned with the national characteristics ofcountries of interest to direct international investors can be divided into statistical,in particular econometric, analyses of FDI flows and stocks; case studies ofspecific investment projects at the firm level; and business opinion surveys oflocational criteria used by actual investors. Of these, the two areas of literaturerelevant to this study are statistical studies and business opinion surveys. Theseare briefly discussed in the following sub-sections.1

1. This is not intended to be a comprehensive review of the literature on the plausible determinants of FDIflows (for the latter see UNCTC 1992).


– 116 –

Table 1

Dependent Variables

FDI1

(US$ m)

FDICountryGroups

FDIshare2

(%)

FDIshareCountryGroups

Australia 18,598 3 9.0 2

Austria 2,465 3 2.0 3

Belgium 24,490 2 23.0 1

Canada 14,435 3 4.0 3

Denmark 3,855 3 6.0 3

France 38,771 2 6.0 3

Germany 25,600 2 3.0 3

Ireland 284 3 1.0 3

Italy 11,035 3 2.0 3

Japan 2,070 3 0.1 3

Netherlands 23,790 2 15.0 2

New Zealand 3,563 3 13.0 2

Portugal 5,881 3 13.0 2

Spain 32,771 2 10.0 2

Sweden 9,842 3 8.0 3

Switzerland 10,349 3 6.0 3

Turkey 2,157 3 2.0 3

United Kingdom 81,845 1 16.0 1

USA 124,510 1 5.0 3

Brazil 3,168 3 1.0 3

Indonesia 3,257 3 3.0 3

Korea 2,589 3 1.0 3

Malaysia 7,636 3 21.0 1

Singapore 10,215 3 28.0 1

Thailand 6,233 3 8.0 3

Notes: 1. FDI inflows in US$ million between 1989 and 1991.

FDI = FDI(1989–1991)

2. The 1989–1991 share of FDI in Investment, as defined in theInternational Financial Statistics (IMF; per cent).

FDIshare = FDI(1989–1991)INV (1989–1991)

Source: IMF (1992a, 1992b), various tables.

2.1 Evidence from Statistical Studies

In the empirical literature, the locational determinants of FDI are usually identifiedusing regression analysis to estimate an investment demand function. Withinregression analysis, a common approach has been to test various hypothesesabout the determinants using time series data, most often collected at the national


– 117 –

level but occasionally at the firm level. Three hypotheses in particular have beenselected for repeated testing and appear to represent a core understanding of FDI.These are that FDI responds positively to the size of the host market, and toeconomic growth of the host, and the tariff discrimination hypothesis. The first isincluded on the assumption that FDI occurs only after the host market issufficiently large to allow economies of scale to be captured by source countryfirms. The growth hypothesis is included to allow for the accelerator relationshipbetween demand and investment—growing demand requires a growing capitalstock in order to maintain a constant capital-output ratio and FDI enlarges thehost’s capital stock. The tariff discrimination hypothesis has been stated as ‘toavoid obstacles to trade (ie, tariffs, quotas, transportation costs, environmentalreluctance to import, etc.), foreign investment is undertaken in the country towhich it is difficult to export because of the obstacle’ (Scaperlanda & Mauer1969, p. 561).

The same three hypotheses have been tested many times with time series datausing alternative specifications of the variables (Scaperlanda & Mauer 1969;Schmitz & Bieri 1972; Lunn 1980; Scaperlanda & Balough 1983). The variousspecifications of independent variables used in the literature for testing these andother hypotheses are given in appendix B. Overall, the time series studiesreviewed indicate that market size is a necessary inclusion in the investmentdemand functions, regardless of how it or FDI is defined. Support for the inclusionof market growth is less straightforward. The evidence is weaker andcontradictory and yet there appears to be reluctance to discard the hypothesis.The difficulty is presumed to be in the specification of the model rather than in themodel itself. There is also mixed evidence for the tariff discrimination hypothesis.Again, this is presumed to be the consequence of the difficulty of specification.2

Cross-sectional data allow an expanded set of locational determinants to betested (Veugelers 1991; Schollhammer & Nigh 1984). As well as testing the threebasic hypotheses and for labour costs, a number of variables concerned with theproximity and relatedness of markets and with non-market conditions (such aslanguage, distance between countries, political conditions, the availability ofinfrastructure) can be introduced. These studies indicate that broader social andeconomic variables cannot be ignored.

In the same way, studies conducted at the level of the firm (Lipsey & Kravis1982; Davidson 1980) are valuable in allowing much closer scrutiny of thepatterns and sequencing of FDI flows, and their results confirm the need toconsider more than economic variables when investigating FDI. Language,proximity, cultural similarity are again suggested by their work as explanators forFDI flows.3

2. To be comprehensive, a variable representing tariffs is included in our list of explanatory variablesalthough it could be argued that tariffs are no longer a major issue in international trade. In a moregeneral sense the term ‘tariff’ stands for visible barriers to trade. It is invisible barriers, however, thathave become the more significant barriers to trade.

3. The influence of source country characteristics on flows of FDI have also been studied (Culem 1988;Veugelers 1991; Tallman 1988). The authors reach differing conclusions regarding the value to theirmodels of inclusion of the investing country’s characteristics. The present paper on the attractiveness ofhosts is not itself concerned with source country characteristics and these results are not reported indetail.


– 118 –

In an early attempt to be more comprehensive, Root and Ahmed (1978)presented 44 economic, social, political and policy variables that had previouslybeen identified in the literature and tested as a determinant of FDI. They then usedstepwise discriminant analysis as their statistical procedure, a procedure thatallows a larger number of variables to be put forward for testing than is possibleusing regression analysis. The opportunity to cast a wide net for potentialexplanators, to select within a long list of suggested variables those that are ofespecial significance is very appealing.

2.2. Business Opinion Surveys

Reference has already been made to publications such as the InstitutionalInvestor and The World Competitiveness Report. The data for one third of thevariables in the latter are obtained from business surveys. There are also specificsurvey-based studies of FDI.4

UNCTC (1992, p. 58) observe that the survey studies ‘open up someadditional perspectives on location advantages’. Survey-based studies of FDIdeterminants report a wide range of independent variables. At the most generallevel they include:

• the need to expand, consolidate and protect market share in the host country(for a survey of detailed findings, see UNCTC). This group of influencesconfirms the market size/growth factors identified by statistical investigations(see, for example, Ajami & Ricks 1981; Tejima 1992);

• the presence of tariffs and other barriers to trade in North American andEuropean Community markets as attractors of FDI (see Shepherd, Silberston& Strange 1985; Dunning 1986);

• high transport costs favouring direct production in host countries (seeShepherd, Silberston & Strange; Dunning);

• low unit labour costs as attractors of FDI (see Dunning; Shepherd, Silberston& Strange; Tejima);

• the presence of FDI-oriented incentives in host countries (Shepherd,Silberston & Strange; Tejima);

• political instability; and• access to production facilities, secure energy and raw material inputs, and

human capital in host countries.

Whilst the survey-based studies tend to be less rigorous than the econometricinvestigations, they ‘have the potential to evaluate qualitative factors which arenot readily amenable to the econometric approach’ (UNCTC, p. 2). From our pointof view, they contain a great deal of helpful information of a judgmental naturethat is needed in the process of short listing candidate independent variables forstatistical studies of FDI. In particular, survey-based studies have identifiedvariables which have actually been included by FDI investors in their locational

4. One of the most comprehensive surveys of FDI determinants is the annual survey of Japanese FDIinvestors carried out by the Research Institute of Overseas Investment of the Export-Import Bank ofJapan.


– 119 –

consideration, in contradistinction to various ‘complex’ variables created for usein econometric studies.

Statistical studies that have attempted to distinguish the importance ofvarious host country locational advantages have had limited success (UNCTC1992). One reason for this may be that ‘many studies have muddled the analysisby grouping all inward investment’ (op. cit., p. 58). In this paper we acknowledgethat account must be taken of modality factors and develop our modelaccordingly.

3 . Framework for Screening FDI AttractorsAny scan of potential attractors of FDI is unavoidably eclectic and some broadconceptual framework is needed to identify factors that are likely to influenceinflows of direct foreign investment into particular countries. The broadconceptual framework which is used in this study was first outlined byJ.H. Dunning as ‘the eclectic paradigm of international production’ and has sincebeen refined in several publications (for the most comprehensive presentation ofthe paradigm see Dunning 1988). The paradigm is based on the recognition thatno single theory can satisfactorily encompass the phenomenon of FDI. The utilityof the eclectic paradigm, in our view, lies in the way in which the vast body ofinformation about possible determinants of FDI can be formatted and presented. Itis not a theory of international investment per se but a convenient frameworkwithin which specific theories and testable hypotheses can be developed.

Explanations of the international division of labour can be derived from aconceptual framework based upon the (international) disposition of factorendowments and relative costs of alternative modalities for transactingintermediate products across national boundaries. Firms that enjoy asset-based(ownership) competitive advantages must have reasons for choosing directinvestment, which involves the retention of some property rights in intermediateproducts, in preference to ‘straight’ or ‘unbundled’ exports of such products.5 Byopting for the former mode of transaction in preference to the latter, firmssubstitute their own mode of organisation of exchange for that of the market.Reasons for the intra-firm internalisation of flows of intermediate goods andservices pertain to various aspects of non-marketability of ownership advantagesand/or market-related impediments to the contracting of unbundled intermediateproducts.6 It follows that we need to identify two groups of location-specificcharacteristics:

• determinants of demand for imported intermediate products by differentcountries; and

5. These advantages must be sufficient to offset the cost of setting up and operating a foreign value-addingoperation which exceed costs faced by indigenous or potential producers (Dunning 1988).

6. Dunning (1986) specifically investigates the importance of modality considerations (‘internalisationinfluences’) as determinants of Japanese direct investment in the United Kingdom. The preference forretention of direct control over imported intermediate products is said to be due to:

• difficulties in negotiating satisfactory prices for ownership advantages;• the need to protect intellectual property rights;• the need to control quality in production and distribution; and• flexibility and networking requirements in production and distribution.


– 120 –

• factors pertinent to the choice of the mode of supplying of these products.

Hence, the likelihood of one country becoming the recipient of FDI from anotherdepends on the degree of (dynamic) complementarity between resourceendowments of both countries—to determine the potential flow of intermediateproducts between them, and the presence of firm- and market-related impedimentsto trade to determine the modal split of the flow between FDI and other imports ofintermediate products.

Our interest is only in the attractiveness of host countries and so we abstractfrom the characteristics of source countries, allowing us, as the first stage of thecountry screening procedure, to formalise the eclectic paradigm into the followingMobility–Modality Framework (MMF):

FDIi = αi DIPiIMPi

(1)

where: FDIi is a measure of all recorded FDI in country i;DIPi stands for factors stimulating/enhancing the ith country’s

demand for imported intermediate goods;IMPi represents factors impeding the ith country’s demand for

imported intermediate goods; andαi is the modality coefficient determining the share of FDI in

demand for all imported intermediate products in country i.

Thus, three types of influences are explicitly isolated to determine the hostcountry’s attractiveness to FDI investors:

• gravity factors, DIPi, which stimulate/enhance demand for importedintermediate products from abroad;

• impedance factors, IMPi, which deter such imports; and• modality factors, α i, which determine the preference for FDI as a specific

form of imports.(For details of the model see Jackson & Markowski 1993.)

Equation 1 provides a broad basis for the screening of countries as hosts orrecipients of FDI inflows.7 First, choices had to be made concerning the form ofthe FDI variable. A priori, no one measure is superior; the appropriate formdepends on the particular aspect of FDI that is being investigated. We chose toexplore various aspects of FDI and this paper reports the results for two measuresof FDI.

In the first instance, FDI is measured as the flow into each country listed intable 1 averaged over 1989 to 1991 (FDI). This measure allows us to distinguish

7. Note, though, that FDI, as defined in the international statistics (see below), may involve raising equitycapital or re-investing earnings in the host economy. It is therefore possible, during a particular timeperiod, for FDI to be recorded with no intermediate products crossing borders. As with all forms ofinvestment, a stock-flow representation is needed to take into account indivisibilities or step-wiseadjustments in capital stocks.


– 121 –

between countries on the basis of the size of the nominal inflow. This is of interestin itself, and was also considered to be a necessary preliminary before consideringother measures.

The variable (FDIshare) captures the share of FDI in the host country’sgross capital formation (‘investment’ as defined in the IMF’s InternationalFinancial Statistics; IMF 1992b). This variable normalises the size of the flow forhost country characteristics. It measures the importance of FDI relative to othermodes of investment.

Table 1 contains the set of FDI measures used in this study.8Second, the screening procedure involved the use of a generalised

investment model to identify gravity/impedance factors that are likely to impacton all imports of intermediate products, including inward investment flows, plusthose factors that determine the preference for FDI as a specific form of foreigninvestment (for details see Jackson & Markowski). Table C1 in appendix C showsa selection of plausible measures which may represent each of the gravity,impedance and modality variables. The final listing of plausible determinants ofFDI for subsequent statistical analysis is shown in table C2 of appendix C.Essentially, this is a more specific version of table C1. In compiling table C2, wewere particularly careful to use information which is normally available to andconsulted by FDI investors. Hence the majority of data are derived from IMD-WEF (1991).9

Independent variables listed in table C2 relate to data that were available toFDI investors in 1988. We selected 1988 as a decision year and the subsequentthree year period 1989–1991 as the period over which decisions made in 1988would be implemented. In some cases, FDI investment projects, especially thoseinvolving the development of greenfield sites, take years to gestate. In many cases,though, they may have quite immediate effects. It is therefore reasonable to expectthat a significant proportion of recorded FDI inflows between 1989 and 1991 areoutcomes of investment decisions taken in 1988.

The specific selection of gravity and impedance (modality) variables is quitestraightforward as these can be readily derived from the FDI literature. To widenthe range of variables derived from this literature (almost exclusively concernedwith mobility factors) we selected nine modality variables.10 Our expectation isthat all of these variables would be positively related to FDI. Two variables whichmeasured past flows of FDI (one in dollar amounts, the other as a percentage ofgross capital formation) were taken to represent modality variables for whichinformation was not readily available. That is, we decided to use the past history ofFDI flows as an indication of frictions that make foreign investors retain control

8. Various measures of growth were also investigated. However, these results are not reported in detail inthis paper.

9. With one exception, all data come from published sources. For ‘hard’ (statistical) data we usedpublications of the International Monetary Fund, in particular the International Financial Statistics. Thesource of ‘soft’ (survey-based) data was IMD-WEF (1991).

10. Modality variables are more difficult to identify. As defined, they are those attributes of the hosteconomy that induce the inflow of intermediate products in the form of FDI in contradistinction to otherforms of imports and foreign investment. Subsidies and other preferments extended to FDI investorsprovide reasonably unambiguous modality measures. This is also the case with taxes and othergovernment inputs that specifically discriminate between foreign direct and other investors. Othermeasures are far less clear cut, ie, they are likely to impact on FDI as well as other forms of foreigninvestment.


– 122 –

over their assets. FDI flows will be established where previous frictions exist, sowe postulate a positive relationship between past and present flows. The extent ofinternational (business) alliances and the acceptance of FDI inflows are also likelyto affect FDI more strongly than other inflows of intermediate goods. Theinclusion of border protection, English language capability, confidence in theadministration of justice and the likelihood of expropriation is less clear cut.Ceteris paribus, all these factors may impact on FDI differentially. Borderprotection attracts the establishment of foreign subsidies behind protective walls.English language capability, greater confidence in the system of justice and smallerlikelihood of expropriation reduce costs of operating in a host economy.Exchange rate instability is included on the assumption that direct productionoffers foreign investors more scope for the management of exchange rate risks.

4 . Statistical Analysis and FindingsTwo types of exploratory analysis of the data were carried out by the authors.

1. Factor analysis was intended to show if the list of forty variables could bereduced to a smaller number of underlying factors to be used in subsequentanalysis. For instance, was it possible that the variables such as absenteeism,strike and labour cost represented an underlying attribute such as labourmarket conditions?

2. Discriminant analysis was intended to test which of the variables it wasnecessary to include in order to be able to distinguish between countries bythe level of FDI they recorded. For instance, if we can predict the level ofFDI in Spain without reference to Spain’s terms of trade or import growth,we can exclude these variables from further analysis.

Results of the factor analysis are reported in Jackson and Markowski (1993).Factor analysis did not simplify the initial list of variables to the extent that itwould make an appreciable difference with regard to the results of thediscriminant analysis. More generally, the results confirmed the complexity of thedata set and encouraged an eclectic approach to testing for the determinants.Thus, the present paper reports the results of the discriminant analysis only.

4.1 Discriminant Analysis

Discriminant analysis is a statistical technique used to identify which of a list ofvariables best allocate another variable into two or more categories. In our case, ifthe FDI data are presented so that countries can be classified into, for instance,high-inflow countries, low-inflow countries, etc., we can investigate the 40variables that are potential explanatory variables in order to identify those thatare the best predictors of the countries into their actual FDI categories. That is, thediscriminating variables are those for which we require information in order tosuccessfully predict if a country is a high or low level recipient of FDI.

The technique involved is to form functions from a weighted linearcombination of the potential variables such that the FDI groups are forced to be asstatistically distinct as possible. The weighting coefficients in the functions can bebroadly interpreted as in linear regression in that they identify the variables which


– 123 –

contribute most to the separation of the groups.11 As a check on the success ofthe discrimination, the significant variables only are used to classify the countriesinto FDI group. Successful classification indicates successful discrimination andsuccessful identification of the relevant variables.

The two FDI variables described in the preceding section, are measured indollar or percentage terms. To apply discriminant analysis to FDI flows, thequantitative data must be converted into categorical data, necessitating some lossof information. The conversion consisted of allocating countries into groupsdepending on whether FDI—as variously measured—was more than one standarddeviation above the mean, less than one standard deviation above the mean, lessthan one standard deviation below the mean, and beyond one standard deviationbelow the mean. Table 1 shows the allocation of countries to groups. For example,in table 1 the UK and USA are in the highest group for absolute levels of FDI,Belgium, France, Germany, Netherlands and Spain are in the next highest group,and the remaining countries are in the third group. The countries with the highestproportion of investment from FDI are the UK, Belgium, Malaysia and Singapore.

4.2 Results of the Discriminant Analysis

4.2.1 The Average Level of FDI, 1989–1991 (FDI) With FDI as the categoricalvariable, four discriminating variables were identified at the 10% level ofsignificance using stepwise discriminant analysis.12 Two functions werecalculated, with the first accounting for 87% of the variance in the variables.Coefficients for the first discriminating function and F-values are given below.

VariableStandardisedCoefficient F-value1

Cumulative FDI inflow, 1984–1987 0.79 13.28

Investment as % of GDP, 1987 –0.72 6.43

Extent to which inward trade is unhindered, 1987 0.83 6.42

Availability of skilled labour, 1987 –0.74 3.40

Notes: 1. F-values indicate the change in separation between groups wheneach variable is added or deleted.

There is a strong autoregressive element in FDI. The single best predictor ofobserved levels of FDI is past levels of FDI . This result sits well with themicroeconomic finding of Davidson (1980), who found that firms continue toinvest where they initially invest, and it is only as experience of overseasproduction accumulates that they are willing to venture to new destinations. Atthe country level of analysis, this implies that countries that already attract FDI arethose most likely to continue to do so. This result is also consistent with what weknow of flows of people—migration chains—developing between countries(Hugo 1986; Anjomani & Hariri 1992).

11. The standardised coefficients reported can be interpreted as Beta coefficients in regression analysis. Thesign of the coefficient, however, does not indicate the direction of the relationship.

12. All following results are from stepwise analysis, reported at the 10% level.


– 124 –

The second most significant variable was the relative size of investment.Countries with low investment compared to their GDP level ‘pull in’ FDI. Thecountries with the highest levels of FDI had an average investment share in GDPof 18%, those with least FDI had an average share of 25%. In that sense, foreigncapital (imported savings) compensates for low domestic savings.

The other significant variables are skill levels and the extent to which theflow of goods is unhampered by protection. Our study shows that countries thatmost interfere with the free flow of trade, as perceived by the internationalbusiness community, record least FDI. The tariff discrimination hypothesis is thusnot supported in this work. The skills variable that we have used is harder tointerpret. It is not clear whether it measures cyclical labour market conditions orthe quality of the labour force. The UK and the USA, which are in the highest FDIgroup, are both represented as having low skill levels.

The four discriminators performed well in predicting the classification.Twenty-two of the twenty-five countries (88%) are classified correctly, with theUK, Denmark and Sweden each predicted to be one group lower than theyactually are. Table 2 reports the allocations to FDI country groups predicted fromour analysis and the actual allocations based on 1989–1991 FDI data.

The analysis was repeated excluding cumulative FDI flows as an explanatoryvariable. As well as skills, tariffs and domestic investment as above, thediscriminating variables included GDP levels, productivity and English languagecapability. The presence of GDP, the most significant of the discriminators,indicated that a large market is attractive to FDI, a result consistent with manyearlier studies. Ceteris paribus, English language capability is associated withhigher levels of direct production (perhaps because it makes it easier to recruitsuitable line management and labour force and lowers transaction costs).

4.2.2. FDI as a Percentage of Investment (FDIshare) FDIshare is a relativemeasure of FDI, normalised for the total volume of investment in the host country.As with FDI, there is a strong autoregressive element in FDIshare. The mostprominent discriminating variable is previous share of FDI in investment. Again itappears that high shares of FDI achieved in the past are the best explanation ofpresent high shares of FDI in investment. Coefficients for the first function (whichaccounts for 88% of variance) are:

VariableStandardisedCoefficient F-value1

FDI as % of GCF, 1985–1987 1.15 27.65

Energy resources as % of requirements, 1986 0.23 4.33

Central government debt as % of GDP, 1987 0.34 3.44

Per-capita trade balance in non-energy rawmaterials, 1986

0.59 2.72

Apart from the autoregressive element, significant variables are the energy andnon-energy resource endowment of the host country and the level of debt, which


– 125 –

Table 2

Actual and Predicted Country Groupings

FDI FDIshare

Actualgroup

Predict.group

Prob. ofgroup

member

Actualgroup

Predict.group

Prob. ofgroup

member

Australia

Austria

Belgium

Canada

Denmark

France

Germany

Ireland

Italy

Japan

Netherlands

New Zealand

Portugal

Spain

Sweden

Switzerland

Turkey

UK

USA

Brazil

Indonesia

Korea

Malaysia

Singapore

Thailand

3

3

2

3

3

2

2

3

3

3

2

3

3

2

3

3

3

1

1

3

3

3

3

3

3

3

3

2

3

2*

2

2

3

3

3

2

3

3

2

2*

3

3

2*

1

3

3

3

3

3

3

0.92

0.87

0.98

0.96

0.64

0.98

0.97

1.00

0.77

1.00

0.94

0.97

0.90

0.73

0.90

0.85

0.60

0.97

1.00

0.94

1.00

1.00

0.92

0.99

0.90

2

3

1

3

3

3

3

3

3

3

2

2

2

2

3

3

3

1

3

3

3

3

1

1

3

2

3

1

3

3

3

3

3

3

3

2

2

3*

2

3

3

3

2*

3

3

3

3

1

1

3

0.98

0.97

1.00

0.97

0.86

0.98

1.00

0.91

1.00

1.00

0.88

0.99

0.52

0.78

0.89

0.53

1.00

0.63

0.93

0.91

0.78

1.00

0.98

1.00

0.99

Notes: * = incorrect predictions.

we interpret as an indication of future tax liabilities. In addition to previous sharesof FDI in investment, indebted countries with fairly rich resource endowments arethe ones associated with higher shares of FDI in investment. These are countries,such as Australia, which incurred high debt levels on the strength of their resourcebase collateral. The second highest group of countries in terms of FDI shares arecharacterised by low debt, and abundant energy and natural resources.

These results are strong. A comparison of the actual and predictedclassification of countries reveals that 92% of countries are correctly classified.Portugal and the UK are each predicted to be in a group lower than they actuallyare (see table 2).


– 126 –

The analysis was repeated for FDIshare excluding past shares of FDI ininvestment. The export share of GDP and the investment share of GDP were bothsignificant. A large export share of GDP implies a higher proportion of investmentof FDI types. Labour market conditions are represented by skill levels yet again(in this case skill availability attracts a high FDI share in investment) andabsenteeism. This new formulation of the model was less successful at prediction,with 80% of countries classified to the correct FDI group, compared to 92% in theoriginal specification.

5 . Concluding CommentsThe dominant result of the analysis is that FDI flows to where it was initiallyattracted. ‘Them that have it, gets it’, as it were. Inflows of FDI are clearlyinfluenced by modality factors, although the specific nature of these influences isnot very clear. However, a virtuous circle is created whereby more FDI engendersyet more. This results whether we look at absolute levels of FDI or at the share ofFDI in investment. The pull force exerted by past inflows of FDI, may have limitedpolicy implications in that it does not tell the policy maker what to do to attractinflows of FDI in the first place. But the importance of engendering the virtuouscircle of FDI inflows should not be underestimated.

The initial attraction appears to be dependent on the relative size of domesticinvestment, with FDI attracted to countries where domestic investors were notthemselves vigorously pursuing investment opportunities. For those countrieswith a greater dependence on FDI in relative terms, rich natural resourceendowments are an important attribute of attractiveness. Countries that are rich innatural resources may attract relatively large inflows of FDI despite the lack ofother attributes of attractiveness. This may provide an initial impetus forengendering the virtuous circle of FDI.

Outward orientation appears important in explaining the share of FDI ininvestment. Countries with a large share of FDI in investment are distinguished bya very large share of exports in GDP. This is not true of countries that appear mostattractive if nominal flows only are considered, although in both cases freedomfrom restrictions on international trade contributes to attractiveness. Tariffs do notseem to attract FDI. Our analysis showed that countries characterised by hightariffs and other barriers to trade appear to be less attractive to FDI investors.

Labour costs do not appear as significant in this analysis of levels of FDI.13

More important labour market concerns are the productivity of labour and the skilllevels found in host countries. Governments have long acted as direct providers ofbasic educational services (skills), and it is evident that, in this respect, they have asignificant role to play in attracting FDI.

Our results also show that the inclusion of modality variables in the analysistends to shift the emphasis away from the gravity-impedance variables. Marketsize and growth prospects do not appear to be dominant locational advantages asperceived by FDI investors. Growth of GDP is not presented as significant ineither case and the level of GDP occurs in the first case only when past experience

13. Although cost variables did seem to be important when growth of FDI was considered (not reported herebut see Jackson and Markowski 1993).


– 127 –

of FDI is not considered. With increased globalisation of economic activity,location in a smaller country that offers scope for successful penetration of exportsmarkets may be more attractive than location in a larger but inward-orientedeconomy.

It is the modality factors, however, that are the key to better understandingof the phenomenon of direct foreign investment and further experimentation withmodality factors should enhance the robustness of FDI attractor models. Ourunderstanding of FDI flows will also be enhanced as data becomes more broadlyavailable, for example, as data on physical measures of FDI (such as the number offoreign-owned companies in the host country) are collected on a consistent basisfor large numbers of countries. Also, one would like to distinguish between directinvestment activity in manufacturing as opposed to primary activities and services.Clearly, there are different factors at work when sectoral inflows of FDI areconsidered. Again, the lack of suitable data has not allowed us to undertake suchstudies. As more information is collected, further research should provide betterinsights into factors that determine both volumes and modality of foreigninvestment.

(Date of receipt of final typescript: September 1996.Accepted by Robert E. Marks, Area Editor.)

ReferencesAjami, R.A. & Ricks, D.A. 1981, ‘Motives of non-American firms investing in the United

States’, Journal of International Business Studies, vol. 12, no. 3, Winter, pp. 25–34.

Anjomani, A. & Hariri, V. 1992, ‘Migration stock and the issue of competing andcomplementary flows in United States interstate migration’, Journal of PopulationEconomics, vol. 5, no. 2, pp. 87–100.

Culem, C.G. 1988, ‘The locational determinants of direct investments among industrializedcountries’, European Economic Review, vol. 32, pp. 885–904.

Davidson, W.H. 1980, ‘The location of foreign direct investment activity: countrycharacteristics and experience effects’, Journal of International Business Studies, vol. 11,no. 2, pp. 9–22.

Dunning, J.H. 1986, Japanese Participation in British Industry, Croom Helm, London.

Dunning, J.H. 1988, Explaining International Production, Unwin Hyman, London.

Hugo, G. 1986, Australia’s Changing Population, Trends and Implications, Oxford UniversityPress, Melbourne.

IMD-WEF 1991, The World Competitiveness Report, IMD International /World EconomicForum, Geneva.

IMF 1992a, Balance of Payments Statistics Yearbook, International Monetary Fund,Washington.

IMF 1992b, International Financial Statistics Yearbook, International Monetary Fund,Washington.

Institutional Investor 1988, ‘1988 Country Credit Ratings’, Institutional Investor, March,p. 180.

Jackson, S. & Markowski, S. 1993, The attractiveness of countries to foreign direct investors,Working Paper No. 1/1993, Centre for Studies in Management and Logistics, UniversityCollege, The University of New South Wales, August.


– 128 –

Kasper, W. 1991, ‘Globalization, locational innovation and East Asian development’, inPacific Basin Papers, Federal Reserve Bank of San Francisco, San Francisco.

Kasper, W. 1993, ‘Spatial economics’, in Fortune Encyclopaedia of Economics, Time-Warner,New York.

Kohlhagen, S.W. 1977, ‘Exchange rate changes, profitability and direct foreign investment’,Southern Economic Journal, vol. 44, no. 1, pp. 43–52.

Lipsey, R.E. & Kravis, I.B. 1982, ‘The location of overseas production and production forexport by US multinational firms’, Journal of International Economics, vol. 12, no. 3–4,pp. 201–223.

Lipsey, R.E. & Weiss, M.Y. 1981, ‘Foreign production and exports in manufacturingindustries’, Review of Economic and Statistics, vol. 63, no. 4, pp. 448–494.

Lunn, J. 1980, ‘Determinants of US direct investment in the EEC, further evidence’, EuropeanEconomic Review, vol. 13, no. 1, pp. 93–101.

Lunn, J. 1983, ‘Determinants of US direct investment in the EEC’, European EconomicReview, vol. 21, no. 3, pp. 391–393.

Root, F.R. & Ahmed, A.A. 1978, ‘The influence of policy instruments on manufacturingdirect foreign investment in developing countries’, Journal of International BusinessStudies, vol. 9, no. 3, pp. 81–93.

Scaperlanda, A. & Balough, R. S. 1983, ‘Determinants of US direct investment in the EEC’,European Economic Review, vol. 21, no. 3, pp. 381–390.

Scaperlanda, A.E. & Mauer, L.J. 1969, ‘The determinants of US direct investment in theEEC’, The American Economic Review, vol. 59, no. 4, September, pp. 558–568.

Scaperlanda, A.E. & Mauer, L.J. 1972, ‘The determinants of US direct investment in the EEC:reply’, The American Economic Review, vol. 62, no. 4, September, pp. 700–704.

Schmitz, A. & Bieri, J. 1972, ‘EEC tariffs and US direct investment’, European EconomicReview, vol. 3, pp. 259–270.

Schollhammer, H. & Nigh, D. 1984, ‘The effect of political events on foreign directinvestments by German multinational corporations’, Management International Review,vol. 24, no. 1, pp. 18–40.

Shepherd, D.A., Silberston, A. & Strange, R. 1985, British Manufacturing Investment Overseas,Methuen, London.

Stevens, G.V. 1972, ‘Capital mobility and the international firm’, in The InternationalMobility and Movement of Capital, eds F. Machlup, W.S. Salant & L. Tarshis, ColumbiaUniversity Press, New York & London.

Stevens G.V. 1974, ‘Determinants of investment’, in Economic Analysis and the MultinationalEnterprise, ed. J.H. Dunning, Allen & Unwin, London.

Tallman, S.B. 1988, ‘Home country political risk and foreign direct investment in the UnitedStates’, Journal of International Business Studies, Summer, pp. 219–234.

Tejima, S. 1992, Japanese foreign direct investment in the 1980s and its prospects for the1990s’, Exim Review, Research Institute of Overseas Investment, the Export-Import Bank ofJapan, vol. 2, no. 2, pp. 25–51.

UNCTC 1992, The Determinants of Foreign Direct Investment, a Survey of Evidence, UnitedNations, New York.

Veugelers, R. 1991, ‘Locational determinants and ranking of host countries: an empiricalassessment’, KYKLOS , vol. 44, Fase. 3, pp. 363–382.


– 129 –

Appendix A

Examples of Dependent Variables

Host Country, i, and Source/Investing Country, j.

Dependent Variables Sources and Comments

FDIij—foreign affiliates of country j in country i as aper cent of total foreign affiliates of country j.

EXPij—exports of country j to country i as a per centof total exports of country j.

Veugelers (1991)

Exports regarded as a substitute or complement tolocal production in serving foreign markets. Thus,both FDIij and EXPij are entered as measures offoreign penetration of country i by country j.

FDIij /GNPj—the share of FDI in money terms fromcountry j to country i in the Gross National Product ofcountry j.

Culem (1988)

GNP is introduced to control for the size of theinvesting country, except when the sample coversonly one investing country. Generally, largercountries are expected to invest abroad more thansmaller ones.

Recorded FDIs are pure financial flows. That is,they are neither equivalent to foreign financialinvolvement in domestic industries, nor to thegrowth of the net assets of foreign affiliates, nor tocapital expenditure on fixed assets.

FDIt —direct investment of US firms in their foreignaffiliates in the EEC during period t.

Lunn (1980, 1983)

FDI = change in the current assets of foreignaffiliates of US firms in the EEC plus change inthe net fixed assets of foreign affiliates less changein their net worth and their liabilities owed toforeigners plus a randomly distributed residual.

It—the annual increase in the book value of US directinvestment (total or manufacturing only)orannual plant and equipment acquisitions by foreignaffiliates.

Scaperlanda & Balough (1983)

It = Kit /KTt—where Kit is either total US directinvestment in region i at time torUS direct investment in manufacturing only in regioni at time tandKTt is total US direct investment at time t.

Schmitz & Bieri (1972)

Market shares are used to emphasise shifts in thedistribution of US direct investment.

Total plant and equipment expenditures by all foreignmanufacturing affiliates of US corporations.

Kohllagen (1977)

A sample of 954 products originally produced in USbut production introduced overseas.

Davidson (1980)

Used to derive entry frequencies.

Per capita inflows of non-extractive direct foreigninvestment.

Root & Ahmed (1978)


– 130 –

Dependent Variables Sources and Comments

Foreign investment in US at the firm level. Ajami & Ricks (1981)

Net sales of US owned affiliates—total sales minusimports from the US

Lipsey & Weiss (1981)

Estimated net local sales—net sales multiplied by theratio of local sales to total sales of the affiliate.


Number of foreign owned affiliates. Lipsey & Weiss (1981)


– 131 –

Appendix B

Examples of Locational Determinants of FDI

Host Country, i, Investing Country, j.

Types ofDeterminants

Variables (Expected Sign) Source/Comments

Market Size andGrowth/Size andGrowth ofDemand

GDP Openj—GDP of country i in constantprices corrected for the internationalopenness of i, ie, GDP plus a measure ofGDP for major export destinations from i(+).

Veugelers (1991)

Particularly significant when large marketis also a neighbour.

Lagged real GNPi of the invested country(+).

Annual per cent rate of growth of realGNPi (+).

Growth rate differential between theinvested, i, and the investing country, j(+).

Culem (1988)

A proxy for the lagged sales of foreignaffiliates in country i. It also measuresmarket size.

A proxy for the expected growth indemand (the accelerator principle).

GNPit—GNP of the EEC (+)

(GNPit – GNPit–1) (+)

(GNPit–1 – GNPit–2) (+)

Rate of change in the GNPi growth rate(+).

Lunn (1980, 1983)

GNP used as a suitable proxy for sales.

GNPit is a proxy for market size.

GNPit – GNPit–1 is a proxy for marketgrowth.

GNPit–1 – GNPit–2 captures the acceleratoreffect of sales on FDI.

The acceleration in the growth of GNP isused to capture the unanticipated growthof sales.

Lagged (t–x) sales of foreign affiliates inthe invested country i (+).

Stevens (1972, 1974)

GNPi,t–1 (+)

GNPit – GNPi,t–1 (+)

GNPi,t–1 – GNPi, t–2 (+)

Predicted sales of investing country’smanufacturing affiliates, lagged (+)(sales estimated as a function of GNPi andprivate credit outstandingi).

Change in predicted sales (+).


An accelerator model that uses output as adeterminant of direct investment.

Predicted sales capture anticipated orpotential demand.


– 132 –



Market Size andGrowth/Size andGrowth ofDemand cont’d

GNPit (+)

G1—per cent rate of growth of GNPi (+)

G2 —ratio of per cent rate of growth ofGNPi to that of GNPj (+).

GNPit – GNPi,t–1 (+)

Scaperlanda & Mauer (1969, 1972)

Foreign investment will take place assoon as the market is large enough topermit capturing of economies of scale.

Based on the relation between the level ofaggregate demand and the stock of capitalneeded to satisfy this demand (incrementalK–O ratio).

All measures of market size and growthgive above detrended.

Scaperlanda & Mauer (1972)

GNPi,t–1 (lagged)

(GNPt – GNPt–1) × 100 (+)


Measures market size

Measures market growth

GNPi (+)Per capita income (+)

Joint ranking of above (+)

Davidson (1980)

Measures of market size—relates salesvolume and industry concentrationmotives.

Per capita GDP

GDPi

Extent of urbanisation

Root & Ahmed (1978)

Measures of market size

Large market

New market

Enhanced economic viability

Ajami & Ricko (1981)

Most important reasons for FDI given ina questionnaire.

Results of factor analysis.

GDP (+) Lipsey & Weiss (1981)Lipsey & Kravis (1982)

Measure of the size of the host market—an indicator of economies of scale.

Country/MarketRelatedness

Languageij (dummy, 1 if the same languageis shared between countries i and j,otherwise 0).

Neighbourij (dummy, 1 if a common borderbetween countries i and j, otherwise 0).

Distanceij—ticketed point mileagesbetween the key airports of countries i andj (–).

Veugelers (1991)

Neighbourij in contrast to Distanceij

measures cross-border movements.Generally, common languages and/orborders stimulate foreign penetration.

Distanceij—shortest distance between mainports of US or Germany to relevantcountries (– for US, + for Germany forequation US exports, opposite for thirteencountry explaining exports).


Proxy for both transport and other costssuch as communications.


– 133 –



Country/MarketRelatednesscont’d

Membership of EEC dummy variable. Lipsey & Weiss (1981)

Lower DirectCosts

Unit labour costs in country i, ie, hourlywages divided by labour productivity (–).

Unit labour costs in country i, ie, hourlywages corrected by hourly productivityexpressed in numerative currency (–).

Veugelers (1991)

Found to be insignificant as between theOECD countries studied, there were nobig differences in labour costs andproductivity levels.

The nominal interest rate differentialbetween the invested country i and the ‘restof the world’ (ie, the average of nominalinterest rates among all the countries of thesample), INTi – INTw, (–).

The unit labour cost differential betweenthe investing country, j, and the investedcountry, i, (–).

Culem (1988)

As recorded FDIs are pure financial flows,this variable accounts for the financialnature of recorded FDIs.

This reflects the desirability of transferringlabour-intensive production activities fromhigher to lower wage countries.

Labour cost—wage per worker divided byoutput per worker. Observed wages adjustedfor quality of labour (the average wage wasdivided by the Denison index of earningsdifferentials related to schooling for thatcountry to give the quality adjusted wage);productivity was measured by real GDPdivided by the quality adjusted labour force(labour force figures multiplied by theDenison index).

Lipsey & Kravis (1982)

Cost of Material Inputs (X + Im) /GDPregressed on population and populationdensity. The residual (ROP) is the proxyfor cost of material inputs.

Lipsey & Kravis (1982)

Assumes the price of raw and semi-finished materials will be lower the easierthe access to world markets, ie, prices willbe inversely related to degree of opennessof the host.

Lower IndirectCosts

Gross Fixed Capital Formation in countryi as a per cent of GDPi (+)

Veugelers (1991)

Introduced as a measure of the availabilityof infrastructure supporting the localproduction in i. However, it was foundthat there was insufficient variation ininfrastructional density among the OECDcountries studied.

Asset /CapacityConsiderations

Lagged (t–1 and t–2) net fixed assets offoreign affiliates in the invested country, i(–).

Change in net fixed assets between (t–1)and (t–2), (+).

Lunn (1980, 1983)Stevens (1972, 1974).

A proxy for the adjustment of capitalstocks to desired levels. Direct investmentused to finance capital stock adjustments.


– 134 –



Asset /CapacityConsiderationscont’d

Ratio of exports to imports.

Ratio of commerce, transport andcommunication to a DP.

Root & Ahmed (1978)

Measures import capacity.

Measures extent of infrastructureprovision.

Trade PolicyFactors andMarket Frictions

Tariffij—average tariff rates for allindustrial products between countries j andi (+).

Veugelers (1991)

The share of in per cent (year t) tariffsapplied in the invested country i on(industrial) imports (+).

The lagged (t–1) share of exports from theinvesting country, j, to the investedcountry, i, in GNPj (+).

Culem (1988)

A proxy for the level in tariff barriers andother distortions to free trade.

A proxy for the impact of prior exports oncurrent FDI (eg, defence of foreign marketshares, response to potential protectionistmeasures and other foreign governmentpressures, the consumer preference fordomestic goods or the need to adaptproducts to the local market specificities.

Annual EEC share of total US exportsdivided by ten.

Annual EEC share of total US exports toEEC and EFTA divided by ten.

A dummy variableD = 0 if 1952–1958

I if 1959–1966


Measured as the ratio of US exports to theEEC to exports of EEC countries to eachother.

A measure of product trade between USand either EEC or EEC and EFTA.

Measures the effect of the formation of theEEC. Enters the equationmultiplicatively.

US manufacturing expenditures on plantand equipment (+).

Dummy variable for anticipated exchangerate changes (+).

Kohlhagen (1977)

An alternative to spending abroad is tospend domestically. Based on profitmaximisation hypothesis.

M—US exports to the EEC divided byexports of EEC countries to other EECcountries (–).

∆ M (–)

Scaperlanda & Mauer (1972)

A measure for the tariff discriminationhypothesis.

The stock analogue to the above flowvariable.

Dummy variable scheme(O ≤ D ≤ 1)

D = one minus proportion of the originaltariff rate in existence for that year (+).


Measures the progressive dismantlementof industrial tariffs on EEC trade.


– 135 –



Trade PolicyFactors andMarket Frictionscont’d

The ratio of the investing country’sexports, j, to the invested country, i, to theinvesting country’s exports to the worldless the same ratio from the previous year(–).

Lunn (1980, 1983)

A proxy to measure trade barriers betweeni and j, ie, affecting on exports from j toi.

Dummy variable for mandatory capitalcontrols (–).

Lunn (1980)Kohlhagen (1977)

Dummy variable scheme for mandatorycapital controls(O ≤ D ≤ 1) (–)

Scaperlanda & Balough (1983)Lunn (1983)

Other Time Kohlhagen (1977)

Attractive political climate

Attractive attitude to DFI in investingcountry

Risk avoidance

Access to technology

Ajami & Ricks (1981)

Results of a questionnaire, factor analysis

TrendIntercept shiftMultiplicative term


Net co-operative domestic events in theinvesting country, j, at time t–k(–).

Net co-operative international eventsbetween the USA and the investingcountry, j, at t–k (+).GDP per capita in the investing country, j,at time t–k.GDPj, t–k (+)

Tallman (1988)

FDI inflows into the USA.


– 136 –

Appendix C

Table C1

Plausible Gravity, Impedance and Modality Variables

Variables Possible Measures

Gravity Variables (+)

Risk Free Revenue Market size variables:• GDP• Population• Export share in GDP (a proxy for market size extending beyond

national borders)

Revenue Growth Rate Expected market growth variables:• GDP growth• Population growth• Import growth• Export growth

Input Availability andQuality

Input variables:• Human capital (eg, labour force, education and skills, R&D)• Non-human capital (eg, investment share in GDP)• Factor productivity (eg, labour productivity)• Public and government-furnished goods (eg, transport and

telecommunications, property rights/ security of people andproperty)

• Natural resources (eg, energy resources and non-energy resources)

Impedance Variables (+)

Input Prices Input price variables:• Wage rates• Interest rate differentials (borrowing rate in the host country, i,

relative to ‘world’ borrowing rates)

Cost Inflation Input price inflation:• Labour cost inflation

Taxation Taxation variables:• Corporate profit taxation• Capital gains taxation

Composite Risk Aspects • Measures of instability (eg, country risk rating)• Government debt (future taxation)• Strikes• Absenteeism

Modality Variables (DFI Specific)

DFI-Specific Profit andCapital

DFI-specific tax exemptions/holidays and subsidies gain taxation(negative taxes) (+)

Composite Risk Aspects The likelihood of ex-propriation of foreign interests (–)

Input Availability &Quality

Measures of internationalisation of the host economy:• Participation in international business networks and alliances• Favourable climate for DFI• Foreign (especially English) language capability


– 137 –

Table C2

Plausible Determinants of FDI

Label Source

Gravity Variables

GDP Nominal GDP in US$, 1987 (IMFa; various)

POP Population in million, 1987 (IMFa; various)

GDPG GDP growth (in local currency), 1978–1987, compound rate (%) (IMFa; various)

POPG Population growth, 1978–1987, compound rate (%) (IMFa; various)

PROD Productivity level (GDP/employee), 1987 (IMD, 2.01)

PRODG Productivity growth 1979–1987. Annual % change in real GDP/employee (IMD, 2.02)

GFCF Investment as % of GDP, 1987 (IMFa; various)

EXG Annual compound % change 1981–1987 based on US$ export values(services and merchandise exports)

(IMD, 8.05)

IMG Annual compound % change 1981–1987 based on US$ import value(services and merchandise imports)

(IMD, 8.24)

EXSHARE Exports as % of GDP, 1987 (IMFa; various)

TOT Terms of Trade change 1981–1987 (IMD, 8.14)

NATRES Natural resources 1986, per capita trade balance (IMD, 7.15)

ENERGY 1986, as % of total requirements in tons of oil equivalent (IMD, 7.16)

TEL Telephones per 1000 population, 1987 (IMD, 1.27)

ROAD Road infrastructure adequate (IMD Survey), 1987 (IMD, 1.29)

AIR Air transport adequate (IMD Survey), 1987 (IMD, 1.31)

RD R&D expenditure as % of GDP, 1986 (IMD, 9.05)

ED 1 Public expenditure on education per capita, 1985 (IMD, 5.32)

ED 2 Secondary school enrolment as % of age group, 1985 (IMD, 5.32)

ED 3 Higher education enrolment as % of age group, 1985 (IMD, 5.33)

SKILL Availability of skilled labour (IMD Survey), 1987 (IMD, 5.21)

SECURE Security of people and property (IMD Survey), 1987 (IMD, 10.08)

Impedance Variables

STRIKE Working days lost per 1000 inhabitants, 1984–1987 (IMD, 10.09)

ABSENT Labour absenteeism (IMD Survey), 1987 (IMD, 2.12)

TAX 1 Tax burden as % of GDP, 1987 (IMD, 3.19)

TAX 2 Corporate tax as % of GDP, 1987 (IMD, 6.10)

DEBT Central government debt as % of GDP, 1987 (IMD, 4.03)

RDIFF LIBOR (1987 US$ one year, 7.61%) less Nominal Deposit Rate in a hostcountry (7.61% – x)

(IMFa; various)

LCOST Total hourly compensation for production workers (wage plus on costs) inUS$, 1987

(IMD, 2.04)

∆LCOST Annual percentage change in real manufacturing earnings per hour1981–1987 less annual percentage change in real GDP per employee1979–1987

(IMD, 2.02 &2.07)

RISK Risk rating of countries by Institutional Investor (1988; 100 – x) (Inst. Inv. 1988)


– 138 –

Table C2 cont’d

Label Source

Modality Variables

ALLIAN Extent of international alliances (IMD Survey), 1987 (IMD, 8.32)

ACCEPT Extent to which inflows of DFI are welcome (IMD Survey), 1987 (IMD, 8.35)

FDISHARE

FDI as % of gross capital formation, Average 1985–1987 (IMFb; various)

FDI CUM Cumulative FDI inflow 1984–1987 (US$) (IMFb; various)

TARIFF Degree of natural border protection (IMD Survey), 1987 (IMD, 8.18)

LANG English language capability. Scale 10 (native) – 0 (own est.)

ADMIN Confidence in the administration of justice (IMD Survey), 1987 (IMD, 10.04)

EXPR Expropriation likely (IMD Survey), 1987 (IMD, 10.05)

XR Exchange rate instability 1982–1987 (IMD, 8.16)

Sources: IMD (1991)—tables shown above.IMF (1992a, various tables).IMF (1992b, various tables).Institutional Investor (1988).

the attractiveness of countries to foreign direct … · the present paper is concerned with the...

Documents