To link to this article:

This article was downloaded by: [Antonios Georgopoulos] On: 19 March 2014, At: 08:15 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK International Journal of the Economics of Business Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/cijb20 Foreign versus Domestic Survival in a Changing Environment Antonios Georgopoulos, Dionysios-Antonios Lalountas & Ioannis- Dionysios Salavrakos Published online: 17 Mar 2014. To cite this article: Antonios Georgopoulos, Dionysios-Antonios Lalountas & Ioannis-Dionysios Salavrakos (2014): Foreign versus Domestic Survival in a Changing Environment, International Journal of the Economics of Business, DOI: 10.1080/13571516.2013.878547 To link to this article: http://dx.doi.org/10.1080/13571516.2013.878547 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content ) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

Int. J. of the Economics of Business, 2014 http://dx.doi.org/10.1080/13571516.2013.878547 Foreign versus Domestic Survival in a Changing Environment ANTONIOS GEORGOPOULOS, DIONYSIOS-ANTONIOS LALOUNTAS and IOANNIS-DIONYSIOS SALAVRAKOS ABSTRACT This paper explores whether and how environmental dynamics can affect foreign and domestic survival. Utilizing a unique longitudinal data set with 420 manufacturing plants created in the protected developing Greek economy (1960 1980), we test these plants ability to survive in the new, integrated environment (1981 2001), when Greece became a member of the European Union (EU). After controlling for time and age effects, we find that environmental dynamics in terms of integration and economic development negatively influence the survival of all tariff-jumping and unskilled labor-intensive plants, regardless of their ownership. However, survival evolution of foreign-owned and domestic plants differs over time depending on the country s degree of economic integration. Specifically, during the shallow integration period (1981 1990), foreign-controlled plants tend to retain a survival premium, which they appear to have acquired in the protectionism era. This means that foreign-owned plants benefited more from external environmental dynamics in terms of tariff protection compared to domestic ones. Nevertheless, in the deep integration period (1991 2001), the declining survival rates tend to converge, and the foreign survival premium completely disappears. Consequently, in the long run, environmental change similarly affects foreign and domestic survival. Key Words: Longitudinal Data; Economic Integration; External Environment; Foreign Survival; Domestic Survival. JEL classifications: C23, F15, F18, F23. The paper expresses the personal views of the authors and does not reflect the views of the European Parliament, the European Commission, or any other formal institution. Furthermore, no confidential information was used for the preparation or writing of this paper. We are grateful to the Editor, Eleanor Morgan, and two anonymous reviewers for their helpful comments. We also thank Vasilios Sogiakas, Department of Economics, University of Glasgow, for his valuable econometric support. Antonios Georgopoulos, Faculty of Business Administration, University of Patras, University Campus, 26504 Rio, Patras, Greece; e-mail: georgop@upatras.gr. Dionysios-Antonios Lalountas, Department of Economics, University of Peloponnese, 22100 Tripolis, Greece & Ministry of Finance, General Directorate of Economic Policy; e-mail: dlalountas@uop.gr. Ioannis-Dionysios Salavrakos, Accredited Assistant European Parliament, 60 Rue Wiertz, 1047 Brussels, Belgium; e-mail: ioannisdionysios.salavrakos@europarl.europa.eu Ó 2014 International Journal of the Economics of Business

2 A. Georgopoulos et al. 1. Introduction The increasing importance and expansion of multinational enterprises (MNEs) in the world economy raises the issue of how foreign survival compares to domestic survival in several local markets (e.g., Bandick 2010; Bernard and Jensen 2007; Bernard and Sjoholm 2003; Görg and Strobl 2003; Kronborg and Thomsen 2009; Li and Guisinger 1991; Mata and Portugal 2002; Taymaz and Özler 2007). The empirical results of the specific survival literature either at firm or at plant level are mixed. In particular, some studies (Baldwin and Yan 2011; Kronborg and Thomsen 2009; Li and Guisinger 1991) find that foreign survival outperforms domestic, though sometimes this decreases over time. 1 Other studies (Mata and Portugal 2002; Taymaz and Özler 2007; Wagner and Gelübcke 2012) produce somewhat neutral results, and they cannot generally locate any clear survival advantage of either group. Furthermore, an emerging stream of literature indicates that there is a disadvantage of foreign survival compared to domestic because of the footloose character of their MNEs 2 (e.g., Alvarez and Görg 2005; Bandick 2010; Bernard and Jensen 2007; Bernard and Sjoholm 2003; Görg and Strobl 2003; Van Beveren 2007). The literature, using models with heterogeneous firms/plants (e.g., foreign MNEs, purely domestic firms, domestic MNEs), initially finds that foreign-owned plants tend to be more efficient than indigenous plants due to their specific characteristics, such as superior capabilities and size (hence, facing lower hazard rates). However, after controlling for these characteristics, it concludes that foreign-controlled plants are more likely to exit the economy. Thus, this literature suggests that important survival differences cannot be explained only by the efficiency differences of plants. Utilizing this suggestion, our paper departs from plant-specific characteristics and focuses on the role of a changing external environment as the primary survival factor, providing some new insights into the survival issue. Environmental dynamics are inextricably linked to aspects of integration, development dynamics, and institutional change (e.g., Benito, Grøgaard, and Narula 2003; Narula 2001; Narula and Dunning 2000). At the level of survival analysis, this means that as the economy develops and its integration proceeds, the chances of plant survival can decrease due to a growing mismatch with the external environment. In this context, our paper explores the important question of whether both foreign-owned and domestic plants created in a growing protected developing economy are similarly prepared to face the challenge of liberalization and integration. To answer this question, we explicitly focus on the survival impact of tariff effect. More precisely, the creation of the single market through the gradual abolishment of tariffs results in rising import competition, increasing survival risk of plants in protected industries. As the integration deepens and tariff protection levels continuously decrease, the hazard of respective plants might increase further, threatening their existence seriously. Thus, during this procedure, it is important to investigate if plant ownership matters. Furthermore, integration dynamics partially overlap with developmental aspects. Hence, we utilize the general dynamic framework of the Investment Development Path (IDP; Dunning 1986; Dunning and Narula 1996; Galan, Gonzalez-Benito, and Zuniga-Vincente 2007; Narula and Dunning 2000), which can help us to understand the atmosphere in which both foreignowned and domestic plants operate in the early development stages and

Foreign versus Domestic Survival in a Changing Environment 3 immediately afterwards, when the new era of integration and liberalization emerges. The stylized form of the IDP describes how the evolution of different stages of a host country alters its attractiveness to foreign and indigenous investors with regard to specific types of plants with, for example, tariff-jumping plants dominating in the first stages (see Dunning 1986; Dunning and Narula 1996; Galan, Gonzalez-Benito, and Zuniga-Vincente 2007; Narula 2001; Narula and Dunning 2000). To our knowledge, this is one of the first studies to investigate the role of environmental dynamics in foreign and domestic survival specifying the particular importance of location advantages that are becoming a sine qua non condition for survival. In particular, a few studies have focused on external dynamics, but they explicitly compare foreign and domestic survival in an economic crisis context (Alvarez and Görg 2005; Godart, Görg, and Hanley 2011). Other studies have examined the impact of environmental change and integration dynamics, but only on foreign survival or the changing roles of foreign units (e.g., Belderbos 2003; Benito, Grøgaard, and Narula 2003; Chung and Beamish 2005; Feils and Rahman 2008; Morgan and Wakelin 1999; Pearce and Papanastassiou 1997). 3 Our paper compares foreign with domestic survival in Greek manufacturing. Despite the country s small size 4 and its peripheral position in Europe, the Greek location is characterized by substantial industrial restructuring and massive closures in the integration era. Hence, it is suitable for the specific investigation. Our unique longitudinal data set captures the survival effect of environmental change, exploring whether 420 foreign-owned and domestic manufacturing plants established in the protected 5 developing Greek economy from 1960 to 1980 survived or closed down in the new environment of integration and liberalization 6 between 1981 and 2001, with 6,079 observations in total. Greece incorporated into the EEC/EU in 1981 and joined the EMU in 2001. However, this is not the whole story. For a more precise analysis of the environmental change, we progressively study survival over the two integration periods, namely the shallow integration period (1981 1990) and the subsequent deep integration period (1991 2001). We propose that during the shallow integration period, foreign-owned plants benefited more from tariff protection compared to domestic. Hence, they tend to retain a survival premium, which they appear to have acquired in the protectionism era. 7 However, in the long term, as development and integration proceed, the survival rates of both groups will tend to converge. Then, any survival premium disappears due to a drastic downgrading of the attractiveness of vulnerable location factors, which affects all plants, independently of their ownership structure. The investigation of this proposition will be the central task of our paper. 2. Modeling Foreign versus Domestic Survival In order to explore how integration and development affects the survivorship of foreign-owned and domestic plants, we address three research issues through a stepwise approach. The first step consists of the investigation of hazard ratios in a standalone framework of environmental dynamics, independently of the plant s ownership status. In this step, we select the main proxies of external change that potentially explain the survivorship risk based on the

4 A. Georgopoulos et al. integration theory and IDP paradigm that offer an appropriate basis for capturing survival effects induced by development and integration dynamics. Our explanatory variables such as TARIFFs (Table 1) can describe the environmental conditions that tariff-jumping plants encountered (which were dominant in the period of protectionism according to the IDP) when they entered the new era of regional integration. In particular, integration gradually decreases tariff protection (e.g., Barrell and Pain 1999; Culem 1988), thus minimizing trade barriers and costs, undermining the survival of tariff-jumping plants (Belderbos 2003). This is the major negative survival effect of the integration process. Moreover, in the dynamic integrated environment, the IDP emphasizes the significance of new qualitative factors (such as new technology, human capital, economies of scale, product differentiation, etc.) at the expense of traditional location advantages. This means that during the new development and integration context, the cost of utilizing unskilled labor (LABOR) 8 rises as a country intensifies the use of the specific production factor. Thus, labor-intensive production gradually becomes less attractive to potential investors and eventually fades out, leading to massive closures of labor-seeking plants (e.g., Bernard, Jensen, and Schott 2006; Dunning 2000; Yamawaki 2004). Furthermore, in a liberalized and integrated environment, new plant-specific location advantages such as product differentiation directly related to horizontal new market-seeking FDI (Caves 1971; Yamawaki 2004) and expressed in relatively high advertising costs (ADVERT) play a positive role in plant survival. Foreign as well as domestic plants increasingly invest in product differentiation in order to survive, as product heterogeneity can lower the exit effect (Colantone, Coucke, and Sleuwaegen 2008). In addition, the reduction in trade costs results in both exports and imports increasing (OPEN; Melitz 2003). Intensified import competition increases the probability of closure of the least efficient producers (e.g., Colantone and Sleuwaegen 2010). In turn, those plants that manage to survive may become more competitive over time (Colantone and Sleuwaegen 2010). Moreover, in the dynamic integrated environment, the significance of technology may increase. Thus, plants employing advanced technology are less likely to close (Colombo and Delmastro 2001; Doms, Dunne, and Roberts 1995), and under specific circumstances may become channels of local and international technology diffusion (Veugelers and Cassiman 2004). In addition, plants located in high-tech industries (TECH) 9 can benefit from external economies, synergies, clusters, and agglomeration effects, showing strong survival chances (Pennings and Sleuwaegen 2000; Yamawaki 2004). Overall, we use the above explanatory variables as an integrated set of proxies for the explanation of survivor risk, taking into account environmental dynamics and changing location advantages. Additionally, we employ the plant-specific control variable current SIZE, the industry-specific concentration variable CONCE, and three dummies as proxies for the plant foundation in three different sub-periods: FOUND67, FOUND6872, and FOUND7375. Hence, we control for age, exploring different policy regimes within the protectionism era (see Table 1). In order to investigate the explanatory power of all these factors on plant survival, we use the Cox model. We consider that the specific model is appropriate to create a general overview as regards the main survival determinants of external environmental dynamics as follows:

Foreign versus Domestic Survival in a Changing Environment 5 Table 1. Variables Variable Definition Rationale Literature DURATION The difference between the year of establishment and the year of closure or the last year of investigation, 2001 FOREIGN Dummy variable that takes the value of 1 for foreign plants and 0 for domestic ones Benefits of foreignness vs. costs of foreignness SIZE Current plant size; labor force (ln) from 1960 to 2001 Liability of smallness vs. economies of scale ADVERT The contribution of advertising expenditure to sales for each plant, from 1960 to 2001 TARIFF Nominal protection rate for each industry, from 1960 to 2001; NACE, 4-digit level LABOR The evolution of labor unit cost for each industry, from 1960 to 2001; weighted by the percentile share of labor costs in total operating costs; NACE, 4-digit level (in the statistics, we use LABOR 100) CONCE The four-plant assets concentration share, for each industry, from 1960 to 2001 TECH Dummy variable that takes the value of 1 if the industry is technology intensive and 0 otherwise OPEN The sum of import penetration ratio (ratio of imports to apparent domestic consumption, i.e., domestic production + import export) and export orientation ratio (ratio of export to domestic production) for each industry, from 1960 to 2001 Product differentiationnew market seeking market proximity Tariff-jumping market seeking Belderbos 2003; McCloughan and Stone 1998 Baldwin and Wang 2011; Delios and Beamish 2001; Kronborg and Thomsen 2009; Mata and Portugal 2002; McCloughan and Stone 1998; Mitchell, Shaver, and Yeung 1994; Shaver, Mitchell, and Yeung 1997 Belderbos 2003; Colombo and Delmastro 2001; Mata and Portugal 2002; McCloughan and Stone 1998; Pennings and Sleuwaegen 2000; Yamawaki 2004 Caves 1971; Pennings and Sleuwaegen 2000; Taymaz and Özler 2007 Barrell and Pain 1999; Belderbos 2003; Culem 1988 Labor seeking Bernard, Jensen, and Schott 2006; Culem 1988; Mold 2003 Market power Colombo and Delmastro 2001; Li 1995; Mata and Portugal 2002; McCloughan and Stone 1998; Shaver, Mitchell, and Yeung 1997; Yamawaki 2004 Efficiency oriented Pennings and Sleuwaegen 2000; Yamawaki 2004 Intensification of competition Colantone and Sleuwaegen 2010; Harris and Li 2011 (Continued)

6 A. Georgopoulos et al. Table 1. (Continued) Variable Definition Rationale Literature FOUND67 Dummy variable that describes the period of rapid economic growth (1960 1967) FOUND6872 Dummy variable that characterizes the period of dictatorship (1968 1972) FOUND7375 Dummy variable that defines the period of the oil and Cyprus crisis (1973 1975) The decision on entry depends on economic/ political stability The decision on entry depends on economic/ political stability The decision on entry depends on economic/ political stability Kronborg and Thomsen 2009; McCloughan and Stone 1998 Kronborg and Thomsen 2009; McCloughan and Stone 1998 Kronborg and Thomsen 2009; McCloughan and Stone 1998

Foreign versus Domestic Survival in a Changing Environment 7 kðt=xþ ¼k 0 ðtþ expðx 0 bþ (1) where k 0 ðtþis the baseline hazard its functional form is unspecified and /ðxþ is the systematic hazard function, for which we adopt the usual exponential specification, that is, /ðxþ = exp(x 0 b), is the matrix with the explanatory variables, and b is the vector of parameters to be estimated. The second step of our analysis is the investigation of possible survival differences between domestic and foreign-owned plants. Throughout the above-mentioned model parameterization, we have adopted a neutral survival framework. However, the IDP suggests the co-existence of plants of different ownership status and examines their dynamic evolution during the development investment path of a specific host country. Moreover, literature on foreign and domestic survival using models with heterogeneous firms/plants emphasizes important differences between ownership structures as regards efficiency, flexibility, and so on (e.g., Alvarez and Görg 2005; Bandick 2010; Bernard and Jensen 2007; Bernard and Sjoholm 2003; Görg and Strobl 2003; Van Beveren 2007). Therefore, plant heterogeneity may imply different survival chances for plants of different ownership. Survival asymmetry is in line with the conventional industrial organization approach of FDI, indicating that foreign-owned plants are likely to have more access to specific proprietary knowledge (Dunning 2000) compared to domestic, which would be expected to lower the likelihood of closure (benefits of foreignness). Hence, we extend the model, introducing a dummy variable (FOREIGN) that accounts for the ownership of plants under investigation. The dummy variable takes the value of 1 for the case of a foreign plant and 0 otherwise. By application of the extended model, we can directly examine if a survival premium for foreign-owned plants exists, and if the explanatory power of the selected proxies differ significantly between foreign-owned and domestic plants. Thus, the existence of a survival premium is tested through the significance of the coefficient of the ownership dummy variable, according to the following model: kðt=xþ ¼k 0 ðtþ expðx 0 b þ FOREIGN b F Þ (2) where FOREIGN = 1 for foreign plants and 0 otherwise. Furthermore, we adopt several interaction terms between the ownership dummy (FOREIGN) and the selected proxies in order to test the significance of the explanatory variables conditional on the origin of the plants under examination: kðt=xþ ¼k 0 ðtþ expðx 0 b þ FOREIGN b F þ X 0 FOREIGNb interaction Þ (3) For illustrative purposes, we plot the standard Kaplan Meier survival curves for foreign-owned and domestic plants. The corresponding figure portrays the estimated survival probability as a function of plant age. In this way, we shape a general picture of how the variable ownership influences plant survival. Our third and final research aim is the investigation of the explanatory power of integration dynamics on foreign versus domestic survival, given that countries are not able to jump from nonintegration to deep integration

8 A. Georgopoulos et al. instantaneously. Thus, survival effects of integration may take place gradually in different stages and affect competitiveness and survival of plants with different ownership in a different manner. More specifically, at the level of European integration, the literature (Benito, Grøgaard, and Narula 2003; Narula 2001) makes a clear distinction between two subsequent periods of the unified European environment: the shallow integration period (essentially involving the reduction of tariff barriers) and the deep integration period, including different common economic policies applied in the framework of the Maastricht agreement (such as industrial, trade, and monetary policies). Drawing on these scholars, we exogenously determine the two corresponding periods for Greece characterized by the corresponding policy mix and the integration level achieved (i.e., the first integration period from 1981 to 1990 and the subsequent integration period from 1991 to 2001). 10 We introduce a time variable, which accounts for the deep integration period, 1991 2001, as follows: kðt=xþ ¼k 0 ðtþ expðx 0 b þ FOREIGN b F þ FOREIGN Iðt [ 10Þb F1 Þ (4) In this model, we aim to explore any potential interaction between the ownership scheme and the most important explanatory variables (such as TARIFF, LABOR, and ADVERT) with respect to the deep integration and the shallow periods, as shown below: kðt=xþ ¼k 0 ðtþ expðx 0 b þ FOREIGN TARIFF b F2 þ FOREIGN LABOUR b F3 þ FOREIGN TARIFF I(t [ 10) b F5 þ FOREIGN LABOUR I(t [ 10) b F6 Þ ð5þ where Iðt [ 10Þ: is an indicator variable equal to 1 if the condition is true (t > 10; i.e., during the deep integration period) and 0 otherwise. Similarly, the indicator variable for the shallow integration period becomes I(t 10). Finally, it should be noted that the choice of the specification for the baseline hazard is based on empirical grounds. More specifically, we consider a general specification containing a standard set of explanatory variables. Using this particular specification, we re-estimate the model under different parametric specifications. Then it is possible to estimate the baseline hazard under different parametric hypotheses. Therefore we compare the shape of the estimated baseline hazard, with the empirical (smoothed) hazard. Overall, as regards our modeling strategy, we avoid the application of parametric approach models, which suggest specification of the distribution of survival time. We believe that such models are not appropriate to explain survival evolution for two reasons. First, no theoretical background exists that can justify the adoption of a specific distribution for survival time. Second, if one were adopted, a wrong parametric hypothesis (e.g., normal distribution) would lead inevitably to incorrect estimates of the parameters of interest. Perhaps for the aforementioned reasons, the use of parametric models in survival analysis is very limited.

Foreign versus Domestic Survival in a Changing Environment 9 3. Data and Descriptive Statistics For the creation of our sample, we exploited the comprehensive business database of the leading Greek market research company Icap Hellas that publishes a wide range of financial and other business data. The database contains all manufacturing plants with full data, such as address, location, year of establishment, management, product groups, and industrial sector. Within this database, we identified the foreign plants, deriving specific information concerning ownership and other plant-specific characteristics from the official lists provided by the Foreign Chambers of Industry and Commerce based in Greece. We explicitly selected those foreign-owned plants with a minimum labor force of 20 individuals as the initial size, and a foreign participation in the share capital of at least 30%, unchanged over time, with a low dispersion. We excluded mergers, financial distresses, and bankruptcies. 11 Thus, we created a database of 210 foreign-owned plants that are all important foreign units established during the 1960 1980 period (protectionism period). These plants continued to operate in the integration period (1981 2001). The specific plants either terminated their operations during this period or survived until its end. We used a matched sample methodology (see also Kronborg and Thomsen 2009; Mezias 2002). The objective was to approximate a controlled experiment by matching a treatment group (foreign-owned plants) to a control group (domestic plants) with similar observable characteristics. By matching them and looking at their average difference, we evaluated the pure effect of nationality. According to the suggestion of the recent survival literature (Kronborg and Thomsen 2009; Mata and Portugal 2002), we kept in mind that finding comparable domestic plants may not be an easy task. In particular, domestic and foreign-owned plants may have quite different strategies with respect to the industries they enter. Moreover, the size of plants exhibits a marked increase as plants age. Hence, after a systematic evaluation, we matched the Table 2. Industry breakdown of the selected plants Industry NACE-4 Digit Level Foreign Plants Domestic Plants All Plants Foods 16 16 32 (7.6%) Beverages 7 7 14 (3.3%) Tobacco products 4 4 8 (1.9%) Textiles 15 15 30 (7.2%) Clothing/leather 12 12 24 (5.7%) Printing/publishing 6 6 12 (2.9%) Paper 9 9 18 (4.3%) Petroleum products 5 5 10 (2.4%) Chemical products 45 45 90 (21.4%) Rubber products and plastics 9 9 18 (4.3%) Nonmetallic minerals 13 13 26 (6.2%) Primary metals/ metal products 17 17 34 (8.1%) Machinery/electrical appliances 36 36 72 (17.1%) Transportation/shipping 7 7 14 (3.3% Other industries 9 9 18 (4.3%) Total 210 210 420 (100.0%)

Table 3. The evolution of survival of the foreign-owned and domestic plants Foreign-Owned Plants Domestic Plants Year Survivors 1 Closures per Year Cumulative Closures Failure Rates 2 Survivors 1 Closures per Year Cumulative Closures Failure Rates 2 Protectionism 1960 1980 210 0 0 0 201 0 0 0 Integration I 1981 210 0 0 0.0 210 0 0 0.0 1982 210 0 0 0.0 209 1 1 0.5 1983 210 0 0 0.0 208 1 2 1.0 1984 209 1 1 0.5 206 2 4 1.9 1985 208 1 2 1.0 202 4 8 3.8 1986 207 1 3 1.4 197 5 13 6.2 1987 206 1 4 1.9 191 6 19 9.0 1988 205 1 5 2.4 184 7 26 12.4 1989 203 2 7 3.3 176 8 34 16.2 1990 200 3 10 4.8 166 10 44 21.0 Integration II 1991 194 6 16 7.6 158 8 52 24.8 1992 186 8 24 11.4 151 7 59 28.1 1993 176 10 34 16.2 142 9 68 32.4 1994 164 12 46 21.9 137 5 73 34.8 1995 151 13 59 28.1 130 7 80 38.1 1996 137 14 73 34.8 125 5 85 40.5 1997 125 12 85 40.5 121 4 89 42.4 1998 119 6 91 43.3 114 7 96 45.7 1999 116 3 94 44.8 110 4 100 47.6 2000 114 2 96 45.7 107 3 103 49.0 2001 113 1 97 46.2 105 2 105 Total 113 97 97 46.2 105 105 105 50.0 Plants under operation by the end of the year. Cumulative closures in relation to the total number of plants (210), in percent. 1 2 10 A. Georgopoulos et al.

Foreign versus Domestic Survival in a Changing Environment 11 210 foreign-owned plants with domestic twins in the same industry and size class (total sample 420 plants). Foreign-owned plants showed a relatively high concentration in a few industries such as chemicals, metal products, and electrical machinery and appliances (46.6%; Table 2). This is in agreement with the findings of previous studies (e.g., Caves 1996; Kronborg and Thomsen 2009) that MNEs mainly operate in oligopolistic industries characterized by economies of scale and scope. 12 Subsequently, following the survival evolution of all these plants, our research demonstrates that 97 foreign-owned and 105 domestic (202 plants in total, 48%) ceased to operate during the investigation period (Table 3. Furthermore, we observed that in the shallow integration period, domestic plants revealed higher failure rates compared to foreign-owned units, indicating a foreign survival premium. However, in the deep integration period, the failure rates tended to converge, and the specific premium almost disappeared (Table 3 and Figure 1). Our analysis took place at plant level. 13 Both nationalities of plants constitute single-plant operations in the Greek context, and plants classified as domestic do not belong to Greek MNEs. 4. Empirical Findings Sample correlations between the independent variables and descriptive statistics are shown in Table 4. The correlation coefficients are relatively low, and no serious multicollinearity problems are detected in the regression estimation (all Pearson coefficients <0.7). To begin with, we also used initial size. However, we decided to exclude it due to the high correlation with current size. Furthermore, the largest variance inflation factor (VIF) was 2.0, which is much lower than the multicollinearity threshold of 10 (Neter et al. 1996). To address the first aspect of our analysis, we applied a stratified Cox model to explain the survival hazard ratios of plants (model 1, Section 2). In this model (Table 5, column 1), the FOREIGN variable was stratified, given our attempt to test the impact of environmental dynamics on the survival of all manufacturing plants independently of their ownership. The parameterization and the specification of the applied model imply a neutral effect (i.e., an increase of a unit on an explanatory variable does not affect hazard) whenever the estimated coefficient (in form of hazard ratios) is statistically significant and equal to unity. Values above it would imply a positive relationship between the variable under investigation and the survival risk and vice versa. At first, with the exception of SIZE, CONCE, and the first and third foundation variables, we observed that all variables were statistically significant at the 1% level (p-values are given in parentheses) and had mostly the expected sign. More precisely, an increase in tariffs by one unit increased the chances of survival by 10%. Similarly, rising advertising costs as well as more openness strengthened survival. In turn, an increase in labor costs increased the closure risk by 2.3 percentage points. Hence, these findings are in accordance with the assumptions and findings of recent literature, as presented in Section 2. Counter to expectations, location in technology-intensive industries had a negative impact on survival, as plants operating in the high-tech sector faced an approximately 100% higher risk than those operating in the conventional sector. Even at first sight, somewhat surprisingly, a more systematic investigation showed that this result essentially confirms the hypothesis of the

12 A. Georgopoulos et al. Figure 1. Number of plant closures. technological factor, since in Greece, during the period under examination, high-tech industries had a clear competitive disadvantage in the world markets. Hence, corresponding plants revealed inevitably decreasing survival prospects. The second research aim focused on the survival effects of ownership. Figure 2 displays the Kaplan Meier survivor curves of foreign-owned and domestic plants respectively, showing that during the second decade of a plant s life, domestic closures were exceptionally high compared to foreign closures, although this difference decreased over time (here we examine the duration of a plant s life, irrespective of the year of plant establishment). Hence, foreign-controlled plants had different chances of survival in relation to domestic plants. To compare the survival functions of the two groups of plants (Figure 2), we applied two nonparametric tests for homogeneity: the Wilcoxon test and the log-rank test. The value for the Wilcoxon test was 17.39, suggesting different survival functions of foreign-owned and Greek plants at the 1% level of significance. Thus, foreign-owned plants exhibited higher survival rates than domestic plants. At the same time, the value for the log-rank test was 5.60, indicating no survival differences between the two groups of plants at the 1% level of significance. Since the Wilcoxon test focuses on earlier survival differences whereas the log-rank test emphasizes later ones, we can conclude that foreign survival premium decreased substantially toward the end of the observation period. Also, in Figure 3, the estimated ln( ln) survival curves for both plant groups intersect at the end of the period under investigation. To examine this issue further, we ran model 2 (Section 2), with the inclusion of the FOREIGN dummy variable. Specifically, Table 5, column 2, shows the existence of a foreign survival premium of about 48%, which is statistically significant at 1%. Moreover, the established explanatory power of the previous step of our analysis did not seem to change quantitatively with the incorporation of the ownership variable. Our third research aim was to explore long-term survival effects of integration and development dynamics strongly connected with different regulation regimes of the Greek industrial and business framework. Therefore, we

Foreign versus Domestic Survival in a Changing Environment 13 Table 4. Means, standard deviations, and Pearson correlations between the independent variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Mean 0.509 4.300 3.245 9.909 12.13 36.13 0.393 56.12 0.208 0.165 0.192 Standard deviation 0.499 0.621 3.275 12.33 17.08 22.76 0.488 28.81 0.406 0.372 0.394 Min 0.000 3.401 0.000 0.000 1.340 10.00 0.000 0.013 0.000 0.000 0.000 Max 1.000 5.991 12.50 95.00 124.2 97.00 1.000 177.0 1.000 1.000 1.000 FOREIGN (1) 1.000 SIZE (2) 0.313 1.000 ADVERT (3) 0.14 0.225 1.000 TARIFF (4) 0.03 0.062 0.251 1.000 LABOUR (5) 0.019 0.018 0.246 0.207 1.000 CONCE (6) 0.176 0.058 0.187 0.062 0.071 1.000 TECH (7) 0.180 0.011 0.101 0.011 0.148 0.210 1.000 OPEN (8) 0.156 0.042 0.187 0.192 0.614 0.082 0.184 1.000 FOUND67 (9) 0.239 0.321 0.142 0.108 0.083 0.035 0.088 0.052 1.000 FOUN6872 (10) 0.134 0.035 0.005 0.027 0.060 0.037 0.013 0.087 0.229 1.000 F0UN7375 (11) 0.040 0.046 0.036 0.059 0.115 0.114 0.097 0.161 0.250 0.217 1.000 Number of observations: 6,079.

14 A. Georgopoulos et al. Table 5. Regression results The determinants of closure rate Variables (1) (2) (3) (4) (5) (6) Stratified Model Standard Cox Regression I(t 10) Survival Premium Shallow Period I(t 10) Survival Premium Shallow Period Interactions Same as in (4) with ADVERT I(t > 10) Survival Default Deep Period TARIFF 0.8996302 (0.005) *** 0.9132898 (0.011) ** 0.8974225 (0.003) *** 0.89814 (0.004) *** 0.8602 (0.001) *** 0.860623 (0.03766) ** LABOR 1.023243 (0.000) *** 1.022691 (0.000) *** 1.023101 (0.000) *** 1.02326 (0.000) *** 1.02389 (0.000) *** 1.02389 (0.000) *** TARIFF FOREIGN LABOR FOREIGN ADVERT FOREIGN 1.400 (0.001) *** 1.4674 (0.000) *** 1.054834 (0.423) 1.0001 (0.960) 1.0018 (0.712) 0.99563 (0.877) 1.5544 (0.025) ** 0.487147 (0.399) ADVERT 0.4404032 (0.000) *** 0.4402376 (0.000) *** 0.4416167 (0.000) *** 0.438686 (0.000) *** 0.3602 (0.000) *** 0.360239 (0.05502) * FOREIGN 0.5274379 (0.001) *** 0.6502 (0.040) ** 0.552045 (0.033) ** 0.412 (0.015) ** 0.162958 (0.042) ** SIZE 1.047911 (0.781) 1.057732 (0.737) 1.07262 (0.674) 1.065638 (0.706) 1.0904 (0.610) 1.090451 (0.610) CONCE 0.9964428 (0.416) 0.996518 (0.425) 0.9962367 (0.390) 0.996090 (0.371) 0.995 (0.300) 0.995433 (0.300) TECH 2.01112 (0.000) *** 2.030445 (0.000) *** 1.993521 (0.001) *** 2.032224 (0.000) *** 1.979 (0.001) *** 1.979326 (0.001) *** OPEN 0.9866515 (0.001) *** 0.987394 (0.002) ** 0.9869975 (0.001) *** 0.986823 (0.001) *** 0.985 (0.001) *** 0.985746 (0.001) *** I(t) FOREIGN 0.2931 (0.011) ** 0.394 (0.315) 2.533219 (0.315) TARIFF 0.6839 (0.001) *** 0.7188 (0.002) ** 1.39113 (0.002) ** FOREIGN I(t) LABOR FOREIGN I(t) ADVERT FOREIGN I(t) 0.975 (0.372) 0.993 (0.827) 1.00624 (0.827) 0.313 (0.17) 3.190974 (0.170) FOUND67 0.9660046 (0.908) 2.104003 (0.943) 0.9347101 (0.822) 0.9627443 (0.900) 0.944 (0.850) 0.9440832 (0.850) FOUND6872 2.081077 (0.002) ** 2.104003 (0.002) ** 2.07356 (0.002) ** 2.111412 (0.002) ** 2.075 (0.002) ** 2.075122 (0.002) ** FOUND7375 1.299914 (0.222) 1.303238 (0.215) 1.25466 (0.289) 1.337991 (0.175) 1.2860 (0.244) 0.286067 (0.244) Log likelihood 638.384 745.546 741.901 739.809 733.90 733.90 Note: Parameter estimates in hazard ratio form; p-values in parentheses. *** p < 0.01; ** p < 0.05; * p < 0.1.

Foreign versus Domestic Survival in a Changing Environment 15 Figure 2. Non parametric survival estimates. Figure 3. Test of proportional hazards assumption. enriched our methodology to capture time effects resulting from the dynamic succession of different integration stages that may influence the evolution of foreign survival premium. More specifically, we introduced a control variable that accounted exogenously for the two periods (i.e., the shallow and the deep integration periods), one treating the interactive terms 14 between ownership and time periods.

16 A. Georgopoulos et al. The baseline refers to the domestic plants operating in the deep period. As we can see from Table 5, column 3 (which corresponds to model 4 in Section 2), the HR of the ownership variable is <1 and statistically significant. In other words, there is a foreign survival premium, which is about 35% (i.e., 1 0.65). However, foreign plants operating in the shallow period reveal more chances of survival of about 71% (1 0.29) compared to domestic plants operating in the deep period. Therefore, with regard to time effects (shallow vs. deep period), we find that the premium attributed to shallow period is about 36%, which is calculated as the risk difference 0.71 0.35 = 0.36. In other words, this is the net period effect, indicating that plants operating in the shallow period enjoy about a 36% greater chance of survival compared to the plants operating in the deep period (regardless of their ownership status). At the same time, the effects of TARIFF and LABOR remain statistically significant and do not change quantitatively relative to the previous results. The same applies to the remaining explanatory variables of the model. Furthermore, we investigated the time effect within the variables TARIFF, ADVERT, and LABOR. To achieve this, we used interactions of these variables with the variable FOREIGN and additional triple interactions, that is, TARIFF FOREIGN I(t 10) and LABOR FOREIGN I(t 10). The econometric results are presented in Table 5, column 4. Regarding the LABOR variable, which seems to have a negative impact on survival, we observed that there were no nationality effects (the corresponding coefficient 1.001 was insignificant) and no time effects (the corresponding coefficient 0.975 was insignificant). Next, we found that tariff protection boosted plant survival. The tariff effect depends on both ownership structure and specific time period. This means that the positive tariff effect on survival does not apply to foreign-owned plants during the deep period, given that the HR of interaction term TARIFF FOREIGN is >1. However, the triple interaction term TARIFF FOREIGN TIME is highly significant (at 1% level). This indicates a foreign survival premium of about 32% (1 0.683) during the shallow integration period due to the high tariff protection compared to domestic plants operating in the deep period. The net period effect associated to tariff protection is about 72%, obtained by considering both the overall (the time and the ownership effect) and the ownership effect. Finally, it is worth noting that the coefficients of the other variables have not changed significantly. To strengthen and extend these findings, we included the variable ADVERT in our empirical analysis (see columns 5 and 6 of Table 5). The corresponding results are essentially the same as those presented above, while the only difference is the definition of the baseline. More specifically, the results given in column 5 are noteworthy for three main reasons. First, the net period effect associated with tariff protection remains statistically significant and shows little change with the addition of variable ADVERT. Second, in the shallow period, foreign survival premium (I(t) foreign) loses its statistical significance after the inclusion of additional variables, which inactivates the positive survival impact of the interaction term FOREIGN TIME (compare column 3 with column 5). Third, while ADVERT has a positive impact on survival, the interaction of ADVERT with the ownership variable has a significant negative impact on survival compared to the domestic plants in the deep period. For the shallow period, the

Foreign versus Domestic Survival in a Changing Environment 17 corresponding interaction term is insignificant. Therefore, there are no time effects connected with the variable ADVERT. Overall, the main empirical results can be summarized as follows. First, we generally observed a foreign ownership premium. Second, we identified an overall time effect, which means that as the integration deepened (moving from the shallow to the deep period), the chances of survival for all plants decreased. Third, there was a strong time effect associated with tariffs, which remained unchanged under different model specifications. 5. Conclusion In this paper, using a unique longitudinal data set, we compare the evolution of foreign and domestic survival in Greek manufacturing, taking into account the effects of integration and development dynamics. In particular, in the new institutional environment of increasing openness and integration, we tested the ability of 420 foreign-owned and domestic plants created in the protectionism period (1960 1980) to survive under conditions of environmental change that Greece underwent from 1981 to 2001. Our results indicated a negative survival evolution for all tariff-jumping and unskilled labor-intensive plants and a positive survival evolution on new market-seeking plants, showing a steadily increasing product differentiation, regardless of their ownership nature. Furthermore, we concluded that survival evolution in the face of environmental dynamics does not mean different things for foreign-owned and domestic plants in the long run. The ultimate consequences are in fact quite similar, but they unfold along a different period. In particular, the analysis of time effects revealed that, in the shallow integration period, the foreign-owned plants retained their survival premium (which they had already established in the protectionism era) over domestic ones, though both exhibited a shrinking survival rate. In this context, the tariff effect mainly explained the maintenance of that premium, indicating that foreign-owned plants benefited more from environmental dynamics compared to domestic ones. However, in the deep integration period, the survival rates of both kinds of plants converged significantly, and toward its end were almost equal. 15 Advertising intensity could not reverse the average negative trend in foreign survival. Nevertheless, our results support the view that the relatively high efficiency level of foreign plants acquired at the early development stages can contribute to the extension of their survival premium for a specific period, despite the adverse evolution of existing location factors. To summarize, we consider that our study adds value to the survival literature by giving a plausible explanation of why and how the foreign survival premium might decrease over time. Specifically, our research methodology appears to be suitable for the exploration of survival in emerging economies, which enter a new era of liberalized and integrated markets. Furthermore, our findings may have implications for industrial dynamics, suggesting that in the new environment, new qualitative location factors (such as human capital, product differentiation, etc.) should be created for the attraction of new FDI (as also suggested by the dynamic framework of IDP). This policy is necessary, as in a dynamic environment, internal learning effects in unskilled labor-intensive and tariff-jumping plants tend to be relatively limited. Another policy

18 A. Georgopoulos et al. implication arises from our main finding concerning the convergence of foreign and domestic survival. 16 Thus, it seems that in the long term, there is no reason to suggest specific government efforts targeted differently to foreignowned and domestic plants. However, in the first years of integration, domestic plants need more support to adjust to the new conditions. Comparative survival dynamics require further investigation, given the specific integration and development path of each country. Since different environmental contexts may produce asymmetric survival effects, future research on survival should exactly specify the nature and the characteristics of the external influential factor. Specifically, a simple demand drop in a specific industry may exert a quite different influence on survival compared to a cyclical economic crisis with expected outcome, or to a nonpredictable global crisis, or even to a new institutional environment altering the internal structure of the whole production system of the economy. The exact specification of the environmental context and the precise analysis of the degree of vulnerability of the dominant types of plants under operation may help researchers to capture the survival consequences for domestic and foreign-owned plants. In this context, time differences cannot be explained only by regional integration dynamics, but also by overall globalization effects or time effects that are subject to structural change over time. Notes 1. In particular, Li and Guisinger (1991) find that domestic entrants are more likely to fail than foreign ones. Also, Baldwin and Yan (2011) conclude that foreign-owned plants have much lower failure rates than domestic ones, but their survival rates are more sensitive to changes in tariffs. Kronborg and Thomsen (2009) find that foreign-owned units have a survival premium compared to domestic ones, but this declines in the long term as a result of globalization and new foreign entry. 2. The footloose character of MNEs is expressed in the enhanced ability to shift production to other locations when the conditions in the host country deteriorate (e.g., external shocks in terms of a drop in domestic demand, economic crisis, etc.). Nevertheless, some relevant concerns to this argument are also mentioned by corresponding studies. Specifically, the high heterogeneity of firms indicates that not all MNEs may be footloose, and not all plants have the same ability to absorb adverse economic conditions and survive. Moreover, barriers to exit may differ. Foreign-owned plants may be on average more skill- and capital-intensive than domestic ones. Thus, they might face higher sunk costs due to a major commitment (e.g., Bandick 2010; Van Beveren 2007). Furthermore, if the shock is only temporary and concerns a few of their products, foreign MNEs could reduce production, shifting production excess within their transnational production system, without necessarily exiting the market (downsizing instead of closure; e.g., Alvarez and Görg 2005; Bernard and Jensen 2007). 3. Many of them exclusively investigate how regional integration reformulates the roles of foreign affiliates in the single market in form of restructuring, downgrading or upgrading, and relocation of activities and product lines within the multinational production system (Benito, Grøgaard, and Narula 2003; Feils and Rahman 2008; Morgan and Wakelin 1999; Pearce and Papanastassiou 1997). 4. In 2007, Greece had a population of 11 million residents, while its GDP per capita in constant 2000 prices was 23,318 USD (OECD statistics database). 5. Tariffs were the most important instrument for the protection of the domestic infant industry. Due to their removal in the integration period, the share of imports in the domestic consumption more than doubled from 23% in 1980 to 51% in 2000 (Bank of Greece). 6. We consider liberalization and the integration period synonymous throughout this paper. Especially in the case of Greece, liberalization of the national economy is strongly connected with its gradual incorporation into the regional and global markets.