FDI in the EU periphery: A Multinomial Logit Analysis of Greek firm strategies 1

Transcription

1 FDI in the EU periphery: A Multinomial Logit Analysis of Greek firm strategies 1 Louri, Helen Department of Economics Athens University of Economics and Business 76 Patission Street Athens Greece lourie@aueb.gr Papanastassiou, Marina Department of European and International Economic Studies Athens University of Economics and Business 76 Patission Street Athens Greece marinap@aueb.gr Lantouris, John Department of Economics Athens University of Economics and Business 76 Patission Street Athens Greece ABSTRACT Outward FDI is a recent activity of Greek firms. It is mostly directed towards the Balkan countries and forms an alternative strategy to exports. An analysis of the decision-making process of Greek firms before undertaking any type of internationalization strategy is attempted using a multinomial logit model on firmlevel data. Borrowing capacity, labour intensity, sales growth rate, relative firm size and acquired familiarity with foreign markets contribute to the strategic decision of investing abroad. Keywords: Internationalization, FDI, MNE s. JEL codes: F21, F23.

2 FDI in the EU periphery: A Multinomial Logit Analysis of Greek firm strategies 1. INTRODUCTION Foreign Direct Investment (FDI) has been encouraged in Greece since the early 1950s, in order to revive and expand the country's industrial base. The growth rate of inward FDI flows has been positive throughout the last 40 years, broken only by the periods and (Bank of Greece, 1998). Heavy - Smithian -type industries, such as chemicals, basic metals and transportation, attracted the majority of FDI flows between In the 1980s and 1990s labour intensive Heckscher-Ohlin -type industries (Ozawa, 1994) such as textiles, food and drink as well as consumer electronics were the main recipients of FDI flows. As can be seen in Table 1, until the early 1990s Greece was a traditional recipient of FDI flows starting to invest abroad later and more hesitantly than the other peripheral European Union (EU) economies. The opening up of Eastern European markets in the 1990s offered new investment opportunities to Greek firms. The majority of Greek outward FDI is concentrated in the Balkan region, although significant projects have taken place in other Eastern European countries, in certain EU countries, such as Portugal, or even in the U.S.A. Other emerging markets outside Europe, including China, Nigeria and Egypt have also attracted Greek FDI more recently. In a revised version of the Investment Development Path (IDP) theory, Dunning and Narula (1996) identify five main stages of development at a country level. These stages are distinguished by the different growth rates in the stocks of inward and outward investment which eventually underline the net outward investment (NOI) position of a country. Greece could be classified as a stage three country. The key characteristics of this stage include the deterioration of comparative advantage in domestic labour intensive industries (increase in domestic wages exceeding productivity) and the advent of outward investment in stage one and stage two countries which possess cheap labour and/or natural resources. Outward 1

3 investment in this case could be either market seeking or resource seeking which could eventually suggest the creation of export-processing zones in the host economy (Dunning, 1993, p.57). Efficiency or asset-seeking motives do not seem to influence Greek FDI decisions at this initial phase of their Balkan expansion. Data provided by the Ministry of National Economy (1995), for selected Balkan countries, show that in 1995 Greek investment was ranked second in Albania after Italy s, while in Bulgaria Greek companies had already invested more than 33 million dollars. Three subsidiaries in food and beverage production accounted for 64% of total Greek invested capital in that country. Greece was the leading investor in Bulgaria, in terms of number of companies, and the fifth in terms of invested capital. In Romania, Greek investment was calculated to be about 32 million dollars (for the period ). Greece s participation in Romania s inward FDI was ranked 14 th in terms of number of companies and 13 th in terms of invested capital. All three countries have attracted Greek companies in Heckscher Ohlin -type industries. In this paper we seek to investigate the determinants of alternative expansion strategies, i.e. exports versus FDI, stressing the importance of firm-specific characteristics. On the novel side of the research is the use of a new set of such characteristics based mostly on financial information and its test on evidence produced by newly investing Greek firms in the Balkan area. The analysis is based on a sample of 292 firms and covers the period CONCEPTUAL FRAMEWORK Firms expand abroad through either trade or FDI in order to capitalise on specific (intangible or tangible) assets they hold, relative to the cost of doing so, as it is affected by sectoral or locational considerations. Rational firms select their expansion strategy according to its risk adjusted expected return (R). Thus, a firm i will opt for strategy j, j J IFF R ij >R ik, where k J but j k. The risk adjusted expected return is determined by different sets of factors depending on firm-specific characteristics, sectoral characteristics as well as location characteristics (Caves, 1996). Dunning s Ownership-Location-Internalization (OLI) 2

4 paradigm, (Dunning 1977; 1988; 1993) has also emphasised that the returns to FDI, and hence FDI itself, can be explained by the competitive-ownership advantages of firms (O), indicating who is going to produce abroad and for that matter, other forms of international activity (Dunning, 1993, p.142), by locational factors (L) influencing the where to produce (p.143) and by the internalisation factor (I) that addresses the question of why firms engage in FDI rather than license foreign firms to use their proprietary assets (p. 145). There are a series of empirical studies that use firm-level characteristics to investigate the determinants of FDI per se. Grubaugh (1987) tests the determinants of US FDI at a sectoral level. He finds that size tends to favour multinationality, a result compatible with Horst s (1972) seminal study on firm determinants of FDI. In addition, intangible forms of ownership characteristics, such as R&D intensity, tend to increase the probability of a firm expanding its production abroad. As Grubaugh argues, these findings support Hymer s hypothesis about the importance of size and product diversification in encouraging a firm to become a multinational. Juhl (1979) examines the sectoral patterns of German FDI in Less Developed Countries (LDCs). His results confirm the importance of size as the main determinant of such a strategic decision. Juhl s findings also suggest that human capital-intensive industries tend to relocate the second and third stage of the product cycle, i.e. the mature and standardized stages, to LDCs in order to take advantage of the lower labour costs. However, Hoesch (1998) in a study on German investment in Central and Eastern Europe (CEE) found that it is small German firms (in employment terms) that tend to penetrate CEE markets through FDI instead of other developed European markets. Jeon (1992) noted that Korean firms in Heckscher-Ohlin type industries tend to pursue FDI in LDCs. He also confirmed his hypothesis that, as wage costs rise in labour intensive industries, there is an incentive to take advantage of cheaper labour costs in LDCs through FDI. 2 A smaller number of studies try to analyse the alternative methods of servicing foreign markets, i.e. exports, licensing or FDI. While most are based on L factors 3, there are few studies that stress the significance of O advantages in selecting the optimal expansion strategy. Buckley and Pearce (1979) emphasise the role of size, 3

5 arguing that large firms tend to service foreign markets through FDI rather than trade. Kravis and Lipsey (1982) place special emphasis on particular characteristics of the parent companies, arguing that high-skilled capital intensive parents transfer, through FDI, the low skilled part of their production to low-wage countries. Lipsey and Weiss (1981) agree that there are factors that simultaneously affect investment and trade (p. 498) but their research focuses mainly on the impact of US manufacturing subsidiaries on trade. In a subsequent paper Lipsey and Weiss (1984) test the relationship between FDI and exports of US firms using both parent and affiliate level characteristics Among the parent characteristics, size (in terms of sales) and innovativeness are included. Their results show that size positively determines exports, while FDI and exports do not substitute for each other. In a recent work by Pfaffermayr (1996) the relationship between Austrian outward FDI and exports as well as their common determinants (p. 502) are examined. His general conclusion suggests no substitutive effect between these two alternative entry strategies. R&D intensity appeared to have the same positive effect on both FDI and trade, while labour and capital intensity positively influenced FDI and exports respectively. In this paper we concentrate the analysis on the parent-specific characteristics that determine the strategic choice of firms between FDI and exports. Using a sample of Greek firms investing in the Balkans, we aim at estimating a probability for the expansion choice made by each independent firm based on its own specific characteristics. Location factors are not stressed since only a homogeneous area (the Balkans) is studied. The significance of the sectoral aspect is also reduced, since only homogeneous, consumer goods producing sectors are taken into account. Both homogeneity constraints are tested. Three alternative expansion strategies are distinguished, i.e. exports to the Balkans, FDI to the Balkans and finally any other strategy indifferent to the Balkan region. The novel O characteristics introduced include a set of variables that represent financial-asset advantages: borrowing capacity, (long and short term), liquidity and profitability; size advantages: relative size of firms in terms of total assets; human capital advantages: a productivity measure; advantages related to firm prospects: growth rate of sales; and finally, advantages related to the geographical diversification 4

6 of the firm: export intensity and the number of foreign markets the firm is already servicing through trade. 3. DATA AND ECONOMETRIC MODEL Information on Greek FDI, for the period, in the Balkan area has been collected through a variety of sources: = a questionnaire was sent to 120 firms possibly involved in various investment projects in the area in the period according to the daily economic press. Almost one third of them answered. Only 10 were involved in FDI. All were relatively large firms (by Greek standards) and produced consumer goods. = a Ministry of Development report (1998) on Greek investment in the Balkans in the 1990s. = a Ministry of National Economy report (1998) on Greek investments in Albania subsidised by the National Development Law. = reports of Greek embassies in the Balkans about Greek FDI in the area = company accounts = information on exports to the Balkans or alternative strategies, taken from the annual ICAP publications on Greek industry. = sectoral data, found in the annual publications of the Confederation of Greek Industries. The O characteristics of the firms were instrumented by their one to two-year lagged values. The assumption is that such strategic decisions take time to be realised. For the financial characteristics, the average of the last two years was used (in order to have a more representative picture of the firm), while size (in terms of total assets), exports and geographical market diversification were obtained with a one-year lag. Since all firms involved in FDI were large (by Greek standards) both in terms of employment and sales, our sample includes all firms in the FDI active sectors with employment as great as the average of their respective sector (fluctuating between 50 employees in clothing and 300 in tobacco). Hence FDI firms were compared against all similar (in terms of size) firms in their sectors. The total number of firms in our 5

7 sample was 292, 26 of which were involved in (Balkan) FDI and 61 in (Balkan) exports. The remaining 205 firms were indifferent to either strategy in the Balkan area. Firms were further grouped according to the relative size of their export activity. Those that exported more than 50% of their sales (108 firms) were considered as export oriented, while the rest (184 firms) were mostly domestic market servicing. This distinction is based on the assumption that different intensities of export orientation may suggest different strategic responses to the firms O characteristics 4. The sectors examined are food and drink, tobacco, textiles, clothing and furniture. These sectors are important components of the Greek industry, as they represent 44% of its sales (in 1995) and also constitute the core of Greek outward FDI. The sectoral distribution of the firms is presented in Table 2. Clothing firms are relatively more active in terms of FDI in the Balkans, with food following. Food firms are relatively more active in exports to the Balkans with textiles following 5. For the econometric analysis of the data, unordered multiple choice models are most relevant (Greene, 1997, pp ). The alternative strategies are 0, 1 and 2 with 1 being the choice of a firm to export in the Balkan area, 2 to undertake FDI in the same area and 0 to follow any other strategy different from the first two, marked as the «indifference» choice in our model 6. The three alternative choices depend on a set of characteristics, w, and are not ordered. They just represent different reactions and not ordered choices, where 0<1<2. The return of strategy j that firm i expects is given by R ij = β wij + ε ij (1) If strategy j is chosen (the choice Y i =j being a random variable) it must be because R ij is the largest among the alternative returns. Thus Prob(R ij >R ik ), k j is the highest. In the general case w ij =[z ij,x i], where x i contains characteristics of firm i not depending on choices, while z ij contains the attributes of the choices varying across 6

8 choices and firms. In our case only x i is relevant, hence a multinomial logit model is suitable. If the J disturbances (strategic choices) are independent and identically distributed following the Weibull distribution F( ε ij ) = e εij and we normalize, assuming β 0 =0, the following probabilities result β x j i e Prob(Y=j) = j 1+ e k= 1 β kxi Prob(Y=0) = 1+ 1 j β e k x i k= 1 The multinomial logit is endowed with simplicity, but the independence of the irrelevant alternatives (which follows from the assumption of independent disturbances) has been considered as a major disadvantage of the model. Hausman and McFadden (1984) have proposed a statistic that measures the validity of the hypothesis, based on the idea that if a subset of the choice set is irrelevant, including it in the model altogether will lead to inefficiency, but not to inconsistency. However, if the choice sets are dependent, which is contrary to the assumption of independent disturbances, the estimates will be inconsistent. The statistic proposed to test the validity of the independence assumption is 1 2 χ = β β ^ ^ ^ ^ ^ β ^ V β r f r V f r f β^ β^ where r stands for the estimators of the restricted model, f for the estimators of the full model and V r ^ ^ and V f stand for the var-cov matrices of the restricted and full 7

9 versions respectively. The statistic is asymptotically distributed as χ 2 with degrees of freedom equal to the number of the characteristics used as independent variables. Eliminating one after the other the choices of exports, FDI, and the third one comprising all the rest, Hausman's specification test failed to reject the null hypothesis, that the difference in the coefficients is not systematic. The associated χ 2 values were 1.447, 0.725, and , respectively, while the critical value at the 10% level with 7 degrees of freedom is Hence the independence of irrelevant alternatives hypothesis cannot be rejected and the use of multinomial logit analysis is not inappropriate. 4. RESULTS As can be seen in Table 3 (the variables are explained in Appendix 1), firms involved in FDI (total sample statistics) differ in terms of borrowing shares, relative strength in their respective industry, growth potential and export intensity. They are also more labour intensive. Export oriented firms carry a higher debt burden than the rest, while their relative size in terms of total assets is smaller. They are also less labour intensive and more geographically diversified 7. Three sets of multinomial logit regressions were estimated using as dependent variables the different types of strategies and as independent variables a list of firmspecific characteristics, as identified above. The results of the analysis for the aggregate sample and the two sub-samples are displayed in Table 4. In the first multinomial logit estimation the effects on the probability of exports to the Balkans versus the probability of indifference (P 1 /P 0 ) are presented. The statistically significant coefficients include a positive relationship with GROWTH for the export oriented firms. A possible interpretation may be that the most dynamic (in growth terms) export oriented firms are the ones that consider exports to the Balkans more preferable than an indifference strategy. Such a strategy is likely to include high start-up costs (e.g. to establish distribution networks, investigate trade regimes, etc.) and hence demand more dynamic followers. The negative relationship with EXPORT shown by all types of firms may suggest that the higher their exporting intensity the less interested firms are to export to the Balkan market, probably because of different 8

10 tastes in the region combined with low incomes and hence low demand. Alternatively, it may suggest that firms inexperienced in international markets (with low share of exports in sales) turned their attention to the Balkan area after 1990 apparently because of its recent opening favoured by geographical proximity. This behaviour is compatible with the theory of internationalisation (Johanson & Vahlne, 1977; 1990). The geographical dispersion of exports (GEOGR) affects positively the choice of exporting to the Balkans, since, the more familiar firms are with different foreign markets, the easier it may be for them to expand to new neighbouring markets. The negative (but statistically weak) sign of LABOUR intensity could indicate some qualitative differences in the demand patterns of the local Greek market and that of the Balkans. Alternatively, it may be an indication of different degrees of labour quality embodied in different products with Greece serving as the location of both the innovative and the standardised stages of the product (Vernon, 1966). The second set of estimations presents the effects on the probability of FDI strategy over indifference (P 3 /P 0 ). The variables that come out as significant are different from the ones affecting the export strategy. TDEBT (heavily influenced by short-term debt as can be seen in Table 3) has a negative and a positive sign for the domestic market servicing and export oriented firms respectively, indicating the different financing strategies pursued by these groups. Export oriented firms probably have easier access to short-term financing, based on their export potential. The higher their short-term financing ability, the more likely they are to prefer FDI, while domestic market oriented firms are more risk-averse declining to favour FDI when their short-term debt is high. The positive sign for LDEBT for the aggregate and domestic market servicing sample, indicating the need to establish a solid financial background before proceeding with such a demanding strategy as FDI, is supported by Hennart (1986). SIZE and GROWTH affect the FDI choice positively underlining that large, dynamic firms are more likely to prefer expanding their production base in the Balkans. Finally, there is a positive sign for LABOUR for the export oriented firms, as cheap labour is one of the Balkan location advantages. The positive sign for GEOGR in the aggregate and domestic market oriented samples suggests that the more familiar with foreign markets firms are the more likely they are to undertake FDI. 9

11 The third set of estimations examines the probability of FDI with exports to the Balkans as the comparison choice (P 3 /P 2 ). Actually, the χ 2 of the estimated equation is equal to the χ 2 produced by the restriction that the coefficients of the first two equations are identical. Such a hypothesis is rejected. The estimated negative relationship with TDEBT reinforces the argument that domestic oriented firms are financially apprehensive. The higher their total debt, the less likely they are to engage in FDI versus exports. In contrast, the positive sign of LDEBT indicates that they are more likely to adopt a FDI strategy, the more carefully balanced their financial structure (as indicated by their long-term borrowing) is. The SIZE effect confirms the familiar result that the larger the firm, the more likely it is that it will prefer FDI to exports. The positive GROWTH effect adds a dynamic dimension to size as a determinant of FDI preference. LABOUR intensity favours FDI in agreement with the argument that the more labour intensive production in stage three countries is, the more likely they are to initiate outward investment in cheap labour countries. Resource seeking motives may cause this response. The positive EXPORT effect suggests that the more firms export, the more likely they are to turn to FDI in the Balkans because a)they are more familiar with foreign tastes and habits, and b)they may want to create export platforms there from which to satisfy other foreign markets. The last argument combines easily with the negative coefficient of EXPORT in the first set of estimations (P 1 /P 0 ). Firms may not be interested in satisfying the low demand of the Balkan market but in using their local plants as export platforms. Yet the more geographically diversified exports are (GEOGR) the less likely firms are to prefer Balkan FDI to exports. The general explanatory ability of the model is satisfactory, as the model likelihood test ratio shows. The predictions (Appendix 2) are adequate given how unbalanced our sample has been (Greene, 1997, pp.891-3) 8. Predictions improve significantly when the sample is disaggregated in export and market oriented firms, hence becoming more homogeneous. 10

12 5. CONCLUSIONS In this paper the internationalisation process followed by firms of a peripheral EU economy, as displayed by the expansion strategies of Greek firms in the Balkan area in the 1990s, was examined. The novelty of the approach was the stress of the role played by firm-specific characteristics in determining the probabilities of each alternative strategy. Despite the limited variety of sectors in our sample, the empirical findings are in line with the theoretical suggestions. Taking into account the restrictions of the data, the adopted model predicted reasonably well the observed firm choices. In the 1990s Greek firms increased markedly their shares of outward FDI in the Balkans through a rather careful consideration of their financial and market structures, apparently being urged by the loss of local comparative advantage. An obvious example is the significant positive effect of the long and medium term borrowing capacity of firms engaged in FDI in contrast to the negative effect of shortterm borrowing. The solid market basis is noticed in the positive effect of relative firm size as well as the growth rate of sales. In addition, the more intense the acquired familiarity with foreign markets through exports, the more likely the undertaking of FDI was found to be. Labour intensity, an old local comparative advantage, was found to affect positively the FDI choice versus exports. New low labour cost locations are preferred. The estimated different strategic reactions of the domestic market catering and export oriented firms (the two subgroups examined separately) may underline the resolution of the former to engage more intensely in FDI, while the latter are more indifferent between exporting or investing, possibly preferring a home production based expansion strategy. Bearing in mind the novelty of the subject, the future research agenda could expand both the sectoral coverage and the firm-specific information used to explain multinational expansion and other related strategies. 11

13 NOTES 1 The authors acknowledge support from a TMR grant on Foreign Direct Investment and the Multinational Corporation (FMRX-CT ). 2 For a thorough review of the literature concerning the determinants of FDI see United Nations (1992) and Dunning (1993). 3 Veugelers (1991) provides a comprehensive analysis of the role of L characteristics. 4 This hypothesis was supported by the estimations and the improvement in the correctly predicted strategic choices. 5 The result of a non-parametric test showed no statistically significant relationship between the variables sector and the types of penetration strategy. 6 There was no information available on licensing agreements. Hence such a choice is not taken into account. 7 Only the variables included in the estimations are presented. Other variables such as liquidity and profitability, size in terms of production value as well as growth in terms of employment among others, not found to be significant or being correlated with the explanatory variables, are not shown. 8 Our predictions classify FDI correctly in 31-39% of the cases, an improvement by up to 100% when the two subgroups are evaluated separately in contrast to when they are evaluated together, showing the importance of estimating for strategically homogeneous groups. REFERENCES BANK OF GREECE (1998) Unpublished data on Greek foreign direct investment, Athens. BUCKLEY, P. and PEARCE, R. (1979) Overseas production and exporting by the world s largest enterprises: a study in sourcing policy, Journal of International Business Studies 10, CAVES, R. (1996) Multinational enterprise and economic analysis, Cambridge: Cambridge University Press. CONFEDERATION OF GREEK INDUSTRIES (1991-6), The Greek industry , Athens. DUNNING, J. (1977) Trade, location and the multinational enterprise: A search for an eclectic approach in OHLIN, B., HASSELBORN, P. and WIJKMAN, P. (eds.), The international allocation of economic activity, London: Macmillan. DUNNING, J. (1988) Explaining international production, London: Macmillan. DUNNING, J. (1993) Multinational Enterprises and the global economy, New York: Addison Wesley. DUNNING, J. and NARULA, R. (1996) The investment development path revisited: 12

14 Some emerging issues in DUNNING, J. and NARULA, R. (eds.), Foreign direct investment and governments: Catalysts for economic restructuring, London: Routledge. GREENE, W.H. (1997) Econometric analysis, Upper Saddle River, NJ: Prentice Hall. GRUBAUGH, S.J. (1987) Determinants of direct foreign investment, Review of Economics and Statistics, 69, HAUSMAN, J. and McFADDEN, D. (1984), A specification test for the multinomial logit model, Econometrica, 52, HENNART, J.F. (1986) Internalisation in practice; early foreign direct investments in Malaysian tin mining, Journal of International Business Studies, 17, HOESCH, D. (1998) Foreign direct investment in Central and Eastern Europe: Do mainly small firms invest? Discussion Paper No. 50, IFO Institute for Economic Research, Munich. HORST, T. (1972) Firm and industry determinants of the decision to invest abroad: An empirical study, Review of Economics and Statistics, 54, ICAP ( ) Industry, annual publications, Athens: ICAP Pulications. JEON Y-D. (1992) The determinants of Korean Foreign Direct investment in manufacturing industries, Weltwirtschaftliches Archiv, 128, JOHANSON, J. and VAHLNE, J.-E. (1977) The internationalisation process of the firm- a model of knowledge development and increasing foreign market commitment, Journal of International Business Studies, 8, JOHANSON, J. and VAHLNE, J.-E. (1990) The mechanism of internationalisation, International Marketing Review, 7, JUHL, P. (1979) On the sectoral patterns of West German manufacturing investments in less developed countries: the impact of firm size, factor intensities and protection, Weltwirtschaftliches Archiv, 115, KRAVIS, I. B. and LIPSEY, R.E. (1982) The location of overseas production for exports by U.S. multinational firms, Journal of International Economics, 12, LIPSEY, R.E. and WEISS, M.Y. (1981) Foreign production and exports in manufacturing industries, Review of Economics and Statistics, 63, LIPSEY, R.E. and WEISS, M.Y. (1984) Foreign production and exports of individual firms, Review of Economics and Statistics, 66,

15 MINISTRY OF NATIONAL ECONOMY (1998) Report of Greek investment in Albania (mimeo), Athens. MINISTRY OF DEVELOPMENT (1998) Report on Greek investment in the Balkans, (mimeo), Athens. OZAWA, T. (1994) Japanese MNCs as potential partners in Eastern Europe s Economic Recovery in P. BUCKLEY and GHAURI, P. (eds) The economics of change in East and Central Europe, Burlington Mass: Harcourt Academic Press. PFAFFERMAYR, M. (1996) Foreign outward direct investment and exports in Austrian manufacturing: substitutes or complements, Weltwirtschaftliches Archiv, 132, UNITED NATIONS (1992) The determinants of foreign direct investment: A survey of Evidence, New York: United Nations. UNITED NATIONS (1998) World investment report, Geneva: United Nations. VERNON, R. (1966) International investment and international trade in the product cycle, Quarterly Journal of Economics, 80, VEUGELERS, R. (1991) Locational determinants and ranking of host countries: An empirical assessment, Kyklos, 44, pp

16 Table 1. Inward and outward FDI flows as a percentage of gross fixed capital formation by country Country World Inward Outward European Union Inward Outward Greece Inward Outward Portugal Inward Outward Spain Inward Outward Ireland Inward Outward Year (annual average) Source: United Nations, World Investment Report 1998 (adapted from table B.5, page 385)

17 Table 2. Sectoral distribution of firms by types of market penetration strategies EXPANSION STRATEGY SECTOR INDIFFERENCE EXPORTS FDI TOTAL to the Balkans To the Balkans FOOD BEVERAGE TOBACCO TEXTILES CLOTHING FURNITURE TOTAL

18 Table 3. Descriptive Statistics: Variable means and Standard deviations (s.d.) Total sample Variable Description Means (s.d.) Total Indifference Exports FDI (292 obs.) (205 obs.) (61 obs.) (26 obs.) T DEBT total Debt/total assets (0.322) L DEBT M. & L.-term Debt/total assets (0.093) SIZE Relative (total assets) size (3.621) GROWTH Growth of sales (0.226) LABOUR no. of employees / sales (0.090) EXPORT Exports/sales (0.393) GEOGR Geographical diversification (0.276) (0.365) (0.093) (2.112) (0.206) (0.092) (0.414) (0.244) (0.188) (0.084) (4.892) (0.200) (0.046) (0.205) (0.230) (0.174) (0.101) (6.380) (0.336) (0.129) (0.430) (0.216) Firms serving the domestic (Greek) market (exports<50% of sales) T DEBT total Debt/total assets (0.375) L DEBT M. & L.-term Debt/total assets (0.092) SIZE Relative (total assets) size (4.229) GROWTH Growth of sales (0.209) LABOUR no. of employees / sales (0.051) EXPORT Exports/sales (0.133) GEOGR Geographical diversification (0.288) (184 obs.) (117 obs.) (54 obs.) (13 obs.) (0.448) (0.094) (2.280) (0.218) (0.052) (0.135) (0.254) (0.196) (0.086) (5.091) (0.183) (0.047) (0.129) (0.232) (0.145) (0.069) (7.795) (0.215) (0.037) (0.092) (0.151) 17

19 Export oriented firms (exports 50% of sales) (108 obs.) (88 obs.) (7 obs.) (13 obs.) T DEBT total Debt/total assets (0.202) L DEBT M. & L.-term Debt/total assets (0.095) SIZE Relative (total assets) size (1.903) GROWTH Growth of sales (0.253) LABOUR no. of employees / sales (0.126) EXPORT Exports/sales (0.160) GEOGR Geographical diversification (0.246) (0.212) (0.092) (1.822) (0.186) (0.122) (0.147) (0.230) (0.107) (0.079) (3.054) (0.288) (0.039) (0.179) (0.229) (0.138) (0.125) (1.284) (0.422) (0.168) (0.151) (0.222) 18

20 Table 4. Multinomial logit estimation for the FDI/Export decision All Firms Domestic Market Oriented Firms, XPR<50% Export Oriented Firms, XPR 50% Variable Coefficient (t-statistic) Coefficient (t-statistic) Coefficient (t-statistic) Exports: (P 1 /P 0 ) CONSTANT (-3.823) a (-2.940) (-0.871) T DEBT (-0.505) (-0.891) (0.571) L DEBT (0.718) (0.908) (0.721) SIZE (0.494) (0.707) (0.215) GROWTH (0.919) (-0.492) (2.274) b LABOUR (-1.346) (-1.527) (0.657) EXPORT (-3.609) a (-1.621) (-2.306) b GEOGR (6.542) a (5.676) a (2.343) b Equation (p=0.0000) (p=0.0000) (p=0.0000) chi2 * FDI: (P 2 /P 0 ) CONSTANT ( (-2.183) 5.066) a b (-2.425) b T DEBT (-0.314) (-2.168) b (2.012) b L DEBT (2.617) a (3.058) a (0.348) SIZE (3.039) a (2.565) a (1.798) c GROWTH (3.838) a (1.932) c (3.102) a LABOUR (0.947) (-0.305) (1.993) b EXPORT (0.916) (-1.521) (-0.320) GEOGR (1.892) c (2.813) a (1.036) Equation (p=0.0000) (p=0.0000) (p=0.0005) chi2 * 19

21 All Firms FDI : (P 2 /P 1 ) Variable Coefficient (tstatistic) Domestic (Greek) Market Serving Firms, XPR<50% Export-Oriented Firms, XPR 50% Coefficient (t-statistic) Coefficient (t-statistic) CONSTANT (-1.756) c (-0.364) (-1.036) T DEBT (-0.010) (-1.851) c (0.776) L DEBT (1.414) (2.279) b (-0.556) SIZE (2.816) a (2.396) b (1.005) GROWTH (2.660) a (2.161) b (-0.436) LABOUR (1.709) c (0.601) (0.136) EXPORT (3.512) a (-0.740) (1.942) c GEOGR (-3.019) a (-0.852) (-1.870) c Equation (p=0.0000) (p=0045) (p=0.0162) chi2 * log likelihood model chi2 * (p=0.0000) (p=0.0000) (p=0.0000) Pseudo-R Sample size a: p<0.01, b: p<0.05, c: p<0.10. Two-tailed test. * According to Likelihood Ratio test. 20

22 APPENDIX 1: Description of variables TDEBT LDEBT SIZE GROWTH LABOUR EXPORT GEOGR = average total debt/total assets in the last two years = average long- and medium term debt/total assets in the last two years = total assets of firm/(total assets of industry/number of firms in the industry) in the last year = average growth rate of sales in the last two years = number of employees/sales (in millions) in the last year = value of exports/sales in the last year = log of the number of different geographical markets the firms is active in (Greece plus up to 10 alternative exporting destinations) in the last year Correlation coefficients between the variables for the full sample T DEBT L DEBT SIZE GROWTH LABOUR EXPORT GEOGR T DEBT 1,000 L DEBT 0,0523 1,000 SIZE -0,0547 0,0894 1,000 GROWTH 0,0358-0,0182 0,0623 1,000 LABOUR -0,0770 0,1356-0,2412-0,1141 1,000 EXPORT 0,0879-0,0245-0,1802-0,0076 0,2751 1,000 GEOGR 0,0600-0,0700 0,3733 0,0509-0,3042-0,0962 1,000 APPENDIX 2: Predictions 21

23 full sample Predicted Indifference Exports FDI total Indifference actual Exports FDI Total correct classified FDI: 19,23% exports: 57,38% indifference: 92,20% exports<50% Predicted Indifference Exports FDI total Indifference actual Exports FDI Total correct classified FDI: 38,46% exports: 61,11% indifference: 85,47% exports 50% Predicted Indifference Exports FDI total Indifference actual Exports FDI Total correct classified FDI: 30,77% exports: 57,14% indifference: 100,0% 22