Does firm heterogeneity affect foreign market entry and exit symmetrically? Empirical evidence for French firms

Journal of Economic Behavior & Organization 85 (2013) 35–47

Contents lists available at SciVerse ScienceDirect

Journal of Economic Behavior & Organization journal homepage: www.elsevier.com/locate/jebo

Does firm heterogeneity affect foreign market entry and exit symmetrically? Empirical evidence for French firms Dirk Engel a,b,1 , Vivien Procher c,b,d,∗ , Christoph M. Schmidt b,e,2 a

University of Applied Sciences Stralsund, Zur Schwedenschanze 15, 18435 Stralsund, Germany Rheinisch-Westfälisches Institut für Wirtschaftsforschung (RWI), Hohenzollernstr. 1-3, 45128 Essen, Germany c Jackstädt Center of Entrepreneurship and Innovation Research, University of Wuppertal, Gaußstraße 20, 42119 Wuppertal, Germany d Ruhr Graduate School in Economics (RGS Econ), Germany e Ruhr-Universität Bochum, Fakultät für Wirtschaftswissenschaft, Lehrstuhl für Wirtschaftspolitik und angewandte Ökonometrie, Universitätsstraße 150, 44801 Bochum, Germany b

a r t i c l e

i n f o

Article history: Received 17 August 2010 Received in revised form 29 September 2012 Accepted 30 October 2012 Available online 16 November 2012 JEL classification: F23 D21 L21 C25

a b s t r a c t This paper studies the internationalization behaviour of French companies using more than 330,000 observations for three two-year intervals. We analyse the ‘symmetric’ role of productivity and other major firm attributes to characterize companies that enter into and exit from foreign markets. High levels of productivity are documented to be characteristic of companies deciding to engage in exporting or foreign direct investment (FDI). However, there does not seem to be a significant correlation between productivity and divestment decisions. Moreover, companies with corporate shareholders are more likely to intensify their international engagement and to retain their cross-border activities. Finally, high levels of short-term and long-term debt tend to increase the likelihood of entry into a more intense international engagement. © 2012 Elsevier B.V. All rights reserved.

Keywords: Foreign markets Entry Exit Exporting FDI

1. Introduction The on-going process of economic globalization implies increased competition and shorter product life cycles, thereby making it more difficult for firms to succeed in international markets. Consequently, serving foreign markets is not an irreversible decision, with more market entries often accompanied by an increase in the number of withdrawals in subsequent periods (see e.g., Dunne et al., 1988; Hopenhayn, 1992). Among others, Bernard et al. (2003), Melitz (2003) and Helpman et al. (2004) developed theoretical models to link firm heterogeneity with different modes to serve foreign

∗ Corresponding author at: Jackstädt Center of Entrepreneurship and Innovation Research, University of Wuppertal, Gaußstrasse 20, 42119 Wuppertal, Germany. Tel.: +49 202 439 5012; fax: +49 202 439 2989. E-mail addresses: [email protected] (D. Engel), [email protected] (V. Procher), [email protected] (C.M. Schmidt). 1 Tel.: +49 38 31 45 66 05; fax +49 38 31 45 66 04. 2 Tel.: +49 201 81 49 227; fax +49 201 81 49 236. 0167-2681/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jebo.2012.10.016

36

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

markets. A key prediction of all these contributions is a ‘symmetric pattern’ of productivity: More productive firms have a higher propensity to enter foreign markets as well as to continue foreign market activities. Most empirical articles have focussed on the determinants of foreign market entry (e.g., Bernard and Jensen, 1995; Bernard and Wagner, 2001). In contrast, so far determinants of downward changes are analysed with respect to export activity by Wagner (2008) and Ilmakunnas and Nurmi (2010). Both papers show that firms realizing a productivity level below a certain threshold level will exit the market in the next period, whereas firms with productivity levels above this threshold will stay in the market. Thus, a symmetric pattern of productivity to explain foreign market participation by exporting is clearly confirmed. Since divestitures are quite common but mostly neglected by research (see Benito, 1997, 2005 for further explanations), we know little about whether the symmetric pattern of productivity is also valid for serving the market by a foreign affiliate. In this paper we make a first attempt to answer this question empirically. For the purpose of comparing findings for exporting and investing abroad, we consider these two modes to serving foreign markets in our model. Furthermore we argue that the company network and strategic control is of far-reaching importance to assess entry and exit in foreign markets. A multi-unit firm with several domestic subsidiaries might find it easier to establish a foreign affiliate than a single-unit firm, whereas these specific network and learning effects might be less relevant for the export decision. With respect to market exit, the embeddedness in a company network spurs competition between affiliates, which in turn might increase the propensity for each affiliate to be divested. Here a firms’ network would influence entry and exit via foreign affiliates asymmetrically. These considerations suggest that the blanket assumption that major firm characteristics always exhibit a symmetric pattern across different modes of internationalization is unwarranted. Based on the existing literature and with specific reference to symmetry we derive assumptions on the potential influence of major characteristics for the internationalization of firms. Our hypotheses are tested empirically on the basis of a large sample of more than 335,000 observations for French firms from the manufacturing and service industries. Our data allows us to distinguish between three modes of internationalization – purely domestic activities, exporting and FDI – covering three two-year intervals 2000–2002, 2002–2004, and 2005–2007, respectively. We apply a time-pooled multinomial probit (MNP) model to estimate the upward and downward transition probabilities of companies across these three categories. From a theoretical point of view we have a two-stage decision process to serve foreign markets. At the first stage a firm makes the decision of going abroad, followed by choosing a specific destination in stage two. This paper clearly focuses on the decision at the first stage and the distinctive attributes of upward and downward changers, whereas Mayer et al. (2010) and Procher (2011) among others mainly address the role of firm and foreign market characteristics to explain the location choice. Our findings clearly show a symmetric pattern of productivity to explain the decision to start and to continue exporting. The mode of serving foreign markets through foreign affiliates is mostly characterized by an asymmetric pattern of firm’s attributes. In line with existing studies, high-productive firms are more likely to invest abroad, however, firms with foreign affiliates do not exhibit significantly higher productivity levels than divesting enterprises. Moreover we detect a symmetric pattern of ownership. Companies with corporate shareholders achieve a significantly higher probability to enter and to stay in foreign markets. The remainder of the paper is organized as follows: drawing on various economic and business theories, Section 2 discusses the role of major determinants of firm’s internationalization behaviour. Section 3 describes the data, the econometric approach and variables employed. Estimation results are presented and discussed in Section 4. Section 5 concludes. 2. Determinants of foreign market participation: a brief literature review Starting from the stylized fact that exporting firms are different from non-exporting firms (e.g., Bernard and Jensen, 1995), a rich body of theoretical and empirical contributions shed light on the relationship between the characteristics of the firm and firm’s internationalization behaviour. 2.1. Productivity Seminal theoretical papers from Hopenhayn (1992), Bernard et al. (2003), Melitz (2003) and Helpman et al. (2004) suggest that productivity differentials between firms or plants play a central role in explaining market turbulence, namely the entry and exit of firms to serve domestic as well as foreign markets. Hopenhayn (1992) models a long-run equilibrium for an industry with competing firms, whereas Melitz (2003) extends Hopenhayn’s model by focussing on monopolistic competition and export participation of firms. Bernard et al. (2003) models firm heterogeneity with the assumption of Bertrand competition in a Ricardian model to incorporate several countries. In sum these theoretical papers predict that more productive firms have a higher propensity to enter foreign markets. In contrast, firms realizing a productivity below a certain threshold level will exit the market in the next period, whereas firms with productivity levels above this threshold will stay in the market. Bernard and Jensen (1995) are among the first to show that exporters already have a higher productivity than nonexporters in the period before they enter export markets (see Wagner, 2007 for a comprehensive survey). Higher productivity offers the chance to realize higher profits and thus to receive sufficient compensation for any non-recoverable sunk costs. Less empirical research has been conducted on the relationship between productivity and exiting export markets.

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

37

Wagner (2008) was the first to test the Hopenhayn model, concentrating on the export market participation. He applies a Kolmogorov–Smirnov test to compare the entire distributions of labour productivity for export starters, export stoppers and continuing exporters. In line with Hopenhayn’s model the distribution of productivity for continuing exporters stochastically dominates the distributions of export starters as well as stoppers. Ilmakunnas and Nurmi (2010) estimate a discrete-time proportional hazard duration model for Finish plants and detect a symmetric pattern of productivity for entering and staying in export markets. Helpman et al. (2004) extend the theoretical model by incorporating FDI as an additional mode to serve foreign markets. The authors suggest that the mode chosen by firms reflects their productivity level: only the most productive firms become multinational enterprises (DI), whereas firms with intermediate productivity enter foreign markets via exports (DX). The least productive companies produce only for the domestic market (D). This postulated productivity ranking was confirmed for firms from various countries (see Greenaway and Kneller, 2007 for a survey of related studies). Among others, Boddewyn (1979) and Duhaime and Grant (1984) point out that poor economic performance of subsidiaries plays an important role in the divestment decision of parent companies, though multivariate analysis to test this hypothesis is lacking. Recent studies either focus on plant level attributes to explain plant closures, or on corporate level characteristics to determine divestment decisions. With respect to the former, Watts and Kirkham (1999) detect that U.K. plants to be closed display a significantly lower labour productivity than comparable plants with ongoing business activities. The study of Bernard and Jensen (2007) confirms this finding for the U.S. Addressing the role of corporate level characteristics, Haynes et al. (2003) demonstrate that poorly performing corporations divest to a larger extent than high-productivity corporations. In an earlier study, however, Haynes et al. (2000) failed to detect this effect. Following from the above discussion we hypothesize a symmetric pattern of productivity to explain both upward and downward changes of firms choosing their mode to serving foreign markets. 2.2. Size, age and financial constraints Beside firm productivity, the industrial organization literature addresses the role of size and age for the internationalization decision of firms (e.g., Geroski, 1995; Caves, 1998 summarizes much of this research). Roberts and Tybout (1997) discuss the impact of experience and detect a positive effect of firm age on export participation. Bernard et al. (2003) argue that firm size has a positive effect on export entry, which is mostly confirmed in empirical studies (e.g., Wagner, 2003). Other recent papers focus on the role of financial constraints. Chaney (2005) builds upon the model of Melitz (2003) and argues that financially constrained firms have some disadvantages in entering foreign markets. The importance of financial restrictions may increase with the level of sunk costs and thus have a major impact on the mode of internationalization. A company can fail to finance its internationalization because of a cash shortage even if it is profitable. Companies with low cash flow and high liabilities might face internal problems in financing their international growth strategy. Moreover, debt obligations can also influence the cost for external credit since creditors usually prefer a high liquidity ratio in order to reduce their short-term risk exposure. Consequently, an upward change in the mode of internationalization and continued foreign market presence is then positively and symmetrically affected by the firm’s financial health. Recent empirical evidence, however, is rather inconclusive (see Lancheros and Demirel, 2012 for an overview). Manova (2012) analyses sectoral trade flows between countries and detects that export participation and sectoral export volumes decrease with the level of financial frictions. This finding is confirmed in pooled regressions at the firm level for highly-developed countries (see e.g., Greenaway et al., 2007 for the U.K., Bricongne et al., 2012; Stiebale, 2011 for France). Stiebale (2011) also estimates a dynamic specification, controlling for time-invariant unobserved heterogeneity. Here the relationship between financial performance and the propensity to export as well as export volume estimates loses its significance. Thus, small variations in financial performance matter to a lesser extent for the foreign market participation than achieved levels of financial health. 2.3. Multi-unit characteristics and ownership structure In addition to companies’ economic performance and internal resources, many studies point out that multi-unit and multinational characteristics can also affect a firm’s mode of internationalization (e.g., Bernard and Jensen, 2007). The reasoning that multi-unit and management-led firms with corporate or financial shareholders show a different internationalization pattern are mainly drawn from the resource-based view (e.g., Wernerfeldt, 1984) and research on the strategic fit between affiliates and corporate headquarters (e.g., Boddewyn, 1979, 1983; Benito, 2005; Filatotchev et al., 2008). Operating multiple interdependent entities requires the adaptation of organizational structures and the coordination of production processes. Hence, firms which already have a domestic (or foreign) subsidiary benefit from learning effects which in turn might lower the organizational barriers and costs for a new (foreign) affiliate (e.g., Amit, 1986). Multi-unit firms have privileged access to foreign market related intelligence within their established networks (see e.g., Gupta and Govindarajan, 1994 on intra-company knowledge flows) to achieve a more efficient foreign market entry. With respect to market exit decisions, multi-unit firms have more possibilities of re-allocating resources from one location to another and thus, they are more likely to properly adjust sales and employment levels than single-unit firms (e.g., Bernard and Jensen, 2007). Birkinshaw et al. (2005) point out that subsidiaries of multinational companies need to continuously defend their position within the parent company’s network of affiliates due to existing internal and external competitive forces. Consequently, we expect an asymmetric role of multi-unit characteristics: Those firms may have a significantly higher

38

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

propensity to invest abroad and, if they are already engaged in foreign markets, a lower propensity to continue exporting or producing abroad than single-unit entities or firms with a smaller company network. Recent evidence is rather mixed. For example, Bernard and Jensen (2004) and Ilmakunnas and Nurmi (2010) cannot detect a significant positive effect of those firms with multi-plant characteristics on their decision to start exporting. In contrast, the stimulating role of multi-plant characteristics to stop exporting and to divest is empirically confirmed by Ilmakunnas and Nurmi (2010), and also by Bernard and Jensen (2007). The latter show that U.S. based multi-plants display a significantly higher probability of being shut down than single-plant firms. Moreover, Duhaime and Grant (1984) show that business units exhibiting a low interdependency with other units are more likely to be divested than those characterized by strong organizational ties. Drawing on the above discussion, it may be argued that the role of ownership structure is in most cases asymmetric, too. External investors provide capital, specific know-how and access to a larger corporate network. This ownership scenario means that firms with (foreign) corporate shareholders are more embedded in the global community and possess industrial ties to other affiliates, customers or suppliers of this corporate owner. Consequently, firms with those ownership structures are more likely to start exporting and investing abroad. In line with this expectation, Roberts and Tybout (1997) find that firms owned by a corporation display a significantly higher propensity to export than other firms. Roper et al. (2006) detected that in particular foreign-owned Irish manufacturing plants achieve significantly higher export intensities. The result is confirmed by Greenaway et al. (2007) for foreign-owned U.K. manufacturing firms, too. Finally, Ilmakunnas and Nurmi (2010) provide empirical evidence that foreign ownership shortens the duration until a firm will start exporting. The decision of corporate firms to continue exporting or serving foreign markets through foreign affiliates depends mostly upon the degree of control exerted by the headquarters and the existence of defined strategies to serve foreign markets. Shifting resources from one affiliate to another is easier for firms with corporate ownership, since the global network is larger compared to firms with individual shareholders. Firms with (foreign) corporate ownership also have more experience in re-organizing and re-structuring their production processes. The value of knowledge created at foreign affiliates’ location increases with the size of the intra-company network. Knowledge externalities that arise at parent companies with foreign corporate owners can be used by more firms within that network as compared to parent companies with individual shareholders. Hence we postulate that (foreign) corporate ownership has a positive effect on entering foreign markets by exporting and/or FDI, however, it negatively determines the firm’s decision to continue exporting or producing abroad.

3. Data and econometric approach 3.1. Data The company data used in the paper is taken from the AMADEUS (Analyse Major Databases from European Sources) database compiled by Bureau van Dijk (BvD). The data collection is carried out by country-specific information providers (e.g., the credit insurer Coface in France, and the credit rating agency Creditreform in Germany). These providers use published sources (e.g., company reports, stock exchange information, news reports) and they contact companies directly to handle collection orders from companies’ business partners (e.g., banks). In contrast to voluntary surveys for scientific purposes, firms rarely refuse to give information to those providers to avoid a negative entry. The same data is also used for marketing purposes by supplying firm addresses or additional firm information on a fee basis. For this reason, it is obvious that information providers strive to build up a comprehensive firm database containing high-quality up-to-date profiles. The sum of total sales of French companies registered in AMADEUS amounts to 2.503 billion Euro in 2009. In terms of total sales, the coverage ratio of AMADEUS is 73.5 per cent in comparison to the SIRENE register, which covers all French companies (see INSEE, 2012). The sample in this paper comprises French companies from the manufacturing, construction, trade and service industries.3 Only companies with an unconsolidated financial account have been selected for the analysis. This ensures that firms that are part of a larger company group are not counted twice via the consolidated account of the group. AMADEUS provides company related information on financial accounts, ownership structure and affiliated companies. Whereas financial data are available up to ten years, information on shareholders and subsidiaries is static and based on the last available annual report of the company. Hence, dynamics in ownership and (foreign) subsidiaries can only be analysed via the various updates of the database. Having access to five updates allows us to observe the internationalization status of companies in the years 2000, 2002, 2004, 2005 and 2007, although the exact date of a status change is not available. We distinguish between three main modes of internationalization, namely MNEs that are engaged in FDI (DI), exporters (DX) and domestic companies that neither export nor have foreign subsidiaries (D). In order to minimize the time gap for status changes to occur within two years, we compare a company’s mode of internationalization in 2000–2002, 2002–2004

3 The NACE (Nomenclature générale des Activités dans les Communautes Européenes) classification is the statistical industrial code for economic activities in the European Union. We exclude from the analysis industries with the NACE two-digit and four-digit codes 01 and 02 (Agriculture, hunting and forestry), 05 (Fishing), 10–14 (Mining and quarrying), 7415 (Management activities of holding companies), 75 (Public administration and defense, compulsory social security) and 91 (Activities of membership organizations e.g., trade unions) from the analysis. Observations for which variables realize values in the upper and lower 1st percentile of the distribution are eliminated from the dataset in order to control for outliers and coding errors.

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

39

Table 1 Internationalization status. Parent companies

2000–2002

2002–2004

2005–2007

Total

D-D

57,427 (92.1%) 4730 (7.6%) 178 (0.3%) 4936 (22.5%) 16,344 (74.5%) 657 (3.0%) 59 (17.4%) 107 (31.6%) 173 (51.0%) 84,611

71,248 (92.9%) 5384 (7.0) 70 (0.1%) 5118 (21.5%) 18,488 (77.6%) 233 (1.0%) 20 (2.0%) 159 (15.6%) 843 (82.5%) 101,563

113,916 (94.6%) 6433 (5.3%) 75 (0.1%) 6147 (22.5%) 20,900 (76.5%) 261 (1.0%) 58 (4.9%) 86 (7.2%) 1051 (87.9%) 148,927

242,591 (93.5%) 16,547 (6.4%) 323 (0.1%) 16,201 (22.2%) 55,732 (76.3%) 1151 (1.6%) 137 (5.4%) 352 (13.8%) 2067 (80.9%) 335,101

D-DX D-DI DX-D DX-DX DX-DI DI-D DI-DX DI-DI Total

Notes: Changes in the internationalization status can occur between 2000 and 2002, 2002 and 2004, 2005 and 2007. For example, firms in the D-DX group in 2000–2002 were still domestic firms (D) in 2000 which became exporters (DX) by 2002. Total refers to firm-year observations.

and 2005–2007.4 Each company has a unique identification number assigned by Bureau van Dijk. A company enters our sample when its identification number is recorded in at least two consecutive updates. Some companies recorded in one update may not be found in the consecutive update due to attrition. Attrition results from both market exit and M&A activities, with the latter resulting in a different identification number. Since we are not able to identify the source of attrition, our sample is subject to a survivor bias. As the time difference between two consecutive updates is rather small, the survivor bias is likely to be of minor relevance. Empirical evidence of studies with comparable data supports this assumption (e.g., Almus et al., 1999). Table 1 provides an overview of the internationalization status of French companies used in our analysis. The sample contains 335,101 firm – year observations with non-missing values for the internationalization status and other key characteristics. High data requirements imply that the number of exporters is around 20% lower compared to the French customs database (see Bricongne et al., 2012, p. 35). The number of MNEs is relatively small, which might be due the high level of sunk costs that need to be incurred with FDI. For example, the dataset contains 1051 continuing MNEs (DI-DI) in 2005–2007, whereas the number of continuing domestic companies (D-D) is more than a hundred times larger. Nevertheless, domestic companies have the opportunity to change upwards and become exporters (D-DX), or to establish foreign affiliates (D-DI). Exporting firms can change their mode of internationalization up- or downwards, i.e., they can remain as pure exporters (DXDX), become engaged in FDI (DX-DI), or stop exporting (DX-D). Finally, MNEs might either continue their foreign activities (DI-DI), or change downwards with respect to the mode of internationalization (DI-DX, DI-D). On average 10.4% of all sampled French firms change their mode of internationalization within any observed two-year period. The share of changers, however, varies to a large extent across the three company types. In particular, 93.5% of the domestic companies retain their status, compared to only 76.3% of the exporters and 80.4% of the MNEs. Correspondingly, some 6.5% of all domestic companies move upward in terms of their internationalization status in any of the observed twoyear periods, but only 1.6% of the exporters become multinationals. By contrast, 19.2% of MNEs reduce their foreign exposure, with most of them (13.8%) becoming exporters, while 5.4% stop serving international markets altogether. Similarly, among exporters, it is much more likely to observe a downward change (22.2%) than an upward move (1.6%). Related to all firms in the sample we detect that on average 4.9% of firms stopped exporting. That ratio is slightly lower than comparable figures reported by Bricongne et al. (2012) based on the French customs database. In sum, the majority of firms does not change its mode of internationalization, so that foreign investments and divestments are rather rare events in absolute terms. In relative terms, however, divestments occur more often than its neglected relevance in trade statistics would suggest. 3.2. Econometrics and variable description Given our data structure with the internationalization status of firms only being observed in a limited number of years, we refrain from applying an econometric panel data model. Instead, we estimate the transition probabilities for firms between

4 Information stems from the AMADEUS updates no. 88, 113, 136, 146 and 168. Companies that were only observed in e.g. 2000 and 2004, or 2002 and 2005 had to be dropped from the sample. Moreover, given our data structure we implicitly assume that a company only changes its status once within the two year period.

40

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

Table 2 Definition and summary statistics of the explanatory variables. Variable

Definition

Productivity Productivity

Total factor productivity (TFP) (in logs)

2.57 1.00

2.87 1.06

3.36 1.33

Domestic subsidiaries

= 1, if at least one corporate shareholder with ownership share > 10% = 1, if foreign corporate shareholder with ownership share > 10% = 1, if at least one financial shareholder with ownership share > 10% Number of domestic subsidiaries

0.105 0.31 0.006 0.08 0.011 0.10 0.073 0.80

0.292 0.45 0.043 0.20 0.027 0.16 0.218 0.96

0.760 0.43 0.136 0.34 0.128 0.33 2.460 6.29

Financial characteristics Cash flow ratio

Cash flow/tangible fixed assets

Current liability ratio

Current liabilities/total assets

Non-current liability ratio

Non-current liabilities/total assets

1.22 28.4 0.562 0.22 0.125 0.16

1.28 15.2 0.508 0.15 0.079 0.10

1.66 3.3 0.524 0.21 0.101 0.10

Basics Employees

Number of employees

Age

Age of company

Capital intensity

Total assets/employees

Year dummies y2000 (d)

= 1, if company is observed in 2000–2002

y2002 (d)

= 1, if company is observed in 2002–2004

y2005 (d)

= 1, if company is observed in 2005–2007

Multi-unit and ownership structure Corporate shareholder (d) Corporate*foreign shareh. (d) Financial shareholder (d)

Number of observations

D

DX

18.96 490.4 17.23 12.6 23.32 611.0 0.240 0.43 0.296 0.46 0.464 0.50 259,461

DI

54.51 871.2 23.88 17.6 20.61 502.2 0.300 0.46 0.326 0.47 0.374 0.48 73,084

663.65 4149.3 34.08 25.6 28.84 78.7 0.133 0.34 0.412 0.49 0.456 0.50 2556

two consecutive time periods, applying separate multinomial probit models (MNP) conditional on the internationalization status in the respective starting period. For each of the three samples (‘domestic companies’, ‘exporters’ and ‘MNEs’) we pool data from three time periods in order to estimate time-pooled MNP models. In a very similar manner Bernard and Jensen (2007) applied a time-pooled model to estimate marginal effects on the probability of plant closures. The MNP assumes that the decision unit chooses one of the mutually exclusive alternatives. At the end of each period covered the final mode j of internationalization yij for firm i is linked with its latent counterpart yij *, which can be written as follows yij∗ = ˇj xi + εij

with i = 1, . . . N; j = 1, 2, 3

where εij is distributed according to a multivariate normal distribution allowing for correlations across choices j, yij∗ is the

unobserved latent variable, and ˇj are the choice-specific parameter estimates for the vector of firm-specific exogenous variables xi . The parameter estimates can be efficiently obtained by using the so-called GHK (Geweke–Hajivassiliou–Keane) simulator (see Geweke et al., 1994). Table 2 provides a statistical summary of the explanatory variables separated by initial internationalization status.5 In essence, there are three samples of firms with 259,461 domestic firms, 73,084 exporters and 2556 MNEs. All firm-specific factors refer to the parent company and are taken from the starting year, e.g., from 2000 for analyzing company decisions between 2000 and 2002. Following most of the internationalization literature, the productivity measure refers to total factor productivity (TFP). Applying the procedure suggested by Levinsohn and Petrin (2003), consistent estimates of firm-level TFP are obtained from a Cobb–Douglas production function.6 The descriptive statistics in Table 2 confirm the expected productivity ranking with MNEs exhibiting the highest productivity, followed by exporters and domestic companies.7

5

An extended version of Table 2 is provided in the Appendix. Levinsohn et al. (2003) provide a STATA command (levpet) to implement their TFP estimations. The TFP value corresponds to the residual obtained from a firm-specific logarithmized Cobb–Douglas production function. In contrast to labour productivity, which measures turnover per employee, TFP has no obvious scaling or natural base values thereby impeding a direct interpretation. 7 For comparison: the TFP value for exporters in our sample is around the size than the TFP in the sample of Greenaway et al. (2007). However, exporters in our sample are remarkable smaller than in Greenaway’s study and thus, economies of scale might matter to a lower extent. 6

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

41

The ownership structure of the firm is used as a proxy for the underlying strategic interest of its owners. External ownership is defined as direct investment if a financial or non-financial owner holds 10% or more of the firm equity (OECD, 2008). An ownership share of at least 10% indicates an effective voice in the management of the company, implying that the investor is able to decisively influence the corporate strategy. The ownership structure is captured by the three dummy variables corporate shareholder, corporate shareholder × foreign and financial shareholder. The interaction term corporate shareholder × foreign is defined to test for significant differences between domestic and foreign corporate investors. The number of firms with at least one foreign financial shareholder is very low and thus an interaction term is not meaningful. Firms with only individual shareholders serve as a reference group. MNEs exhibit the highest percentage of firms with corporate and financial owners, 76.0% and 12.8% respectively. In contrast, only 29.2% of the exporters and 10.5% of the domestic firms have corporate shareholders. Similarly, a mere 1.1% and 2.7% of the exporters and domestic firms have financial owners, respectively. The multi-unit firm structure is captured by the number of domestic subsidiaries. Descriptive statistics reveal that MNEs have on average 2.46 domestic subsidiaries, whereas only a minority of exporters and domestic companies have established domestic affiliates. Following the investment literature on measuring financial constraints, we employ the cash flow ratio defined as cash flow divided by tangible fixed assets. With a value of 1.66, MNEs exhibit a higher cash flow ratio than exporters and domestic companies, who have ratios of approximately 1.2. In line with Stiebale (2011) we do not detect any difference in the cash flow ratio between exporters and domestic firms. Moreover borrowing constraints constitute a major investment barrier and two further indicators are included to address the firm’s ability to finance foreign investments. The current liability ratio, defined as current liabilities to total assets, represents short-term debt that needs to be re-paid within one year. In contrast, the non-current liability ratio, defined as non-current liabilities to total assets, represents long-term liabilities that a company is not required to repay within the next 12 months.8 The latter is also employed by Stiebale (2011) and Greenaway et al. (2007). A high non-current liability ratio mainly results from investments in tangible and intangible assets. Since factors such as economies of scale, experience, technological differences and the impact of the local environment will affect different companies to varying degrees, our study also accounts for firm’s age to pick up learning effects, firm’s size to consider economies of scale and capital intensity to control for sunk costs of investments. The older and larger a firm, the larger the economies of scale tend to be and the propensity to go abroad. Capital intensity is measured as total assets over the number of employees. MNEs display on average the highest capital intensity. In contrast to Ilmakunnas and Nurmi (2010), French domestic companies exhibit a higher capital intensity than French exporters. Several industry dummies based on the two-digit NACE classification and legal form dummies capture remaining firmspecific heterogeneity. Regional dummies based on the first-level NUTS9 classification control for effects of firm’s local environment and year dummies are included to pick up business cycle and time trends. 4. Results The total sample is split into three subsets according to firm’s initial internationalization status. Separate regressions are estimated for all three subsets in order to empirically test our hypotheses. In each subset firms decide whether to keep their current status or to switch to one of the two other internationalization modes. In Section 4.1 we investigate whether and how domestic companies internationalize (via exporting or FDI). Section 4.2 analyses the role of firm-specific factors on the decision of exporters to stay or to leave the export market, or to become engaged in FDI. Finally, transition probabilities of MNEs either continuing or ceasing their foreign operations are estimated in Section 4.3. 4.1. Internationalization of domestic companies In the first sample domestic companies have the choice to either continue serving the domestic market only, or to internationalize via exporting or FDI. Columns (1) to (2) of Table 3 present the marginal effects, calculated at the means of the independent variables, from a multinomial probit model explaining the transition choices of continuing domestic companies (D-D) (reference group), export starters (D-DX) and firms that become engaged in FDI (D-DI). We clearly find that domestic firms with a higher productivity are more likely to enter international markets. Our estimates suggest that a productivity increase by one standard deviation (equal to an increase of 40%), ceteris paribus, increases the likelihood of becoming an exporter by 0.34 percentage points and of becoming a MNE by 0.006 percentage points. Despite large variation in productivity levels, foreign market entry constitutes a rare event. The predicted probability of domestic companies to start exporting is 5.31%, and less than 0.02% to become engaged in FDI. Thus, evaluated at the sample mean, the transition probability of export starters increases from 5.31% to 5.65% (5.31 + 0.34) when a firm’s productivity increases by one standard deviation. This implies that even a large positive productivity shock only incrementally enhances the transition probability of domestic companies to start exporting.

8 Alternatively, we have constructed these liability ratios by using tangible fixed assets in the denominator (instead of total assets). The main results remain unchanged. Using total assets has the advantage that current assets are also taken into account. 9 The NUTS (Nomenclature des Unités Territoriales Statistiqes) classification is the standard statistical geographic code for the regional sub-division of a country in the European Union.

42

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

Table 3 Foreign market entry by domestic companies. Rare event logit model Base group D-DX

Multinomial probit model Base group D-D

Productivity Total factor productivity Multi-unit and ownership structure Corporate shareholder (d) Corporate × foreign shareholder (d) Financial shareholder (d) Domestic subsidiaries Financials Cash flow ratio Current liability ratio Non-current liability ratio Basics Employees Age Capital intensity Number of observations Pr(change)

D-DX Marginal effect

D-DI Marginal effect

D-DI Marginal effect

0.0034*** (7.36)

0.00006*** (3.52)

0.0013*** (3.59)

0.0213*** (12.77) 0.0295*** (4.95) 0.0133*** (3.08) −0.0003 (0.65)

0.0012*** (5.57) 0.0002 (1.50) 0.0002 (1.45) 0.00002** (2.27)

0.0140*** (5.84) 0.0006 (0.36) 0.0068* (1.73) 0.0011*** (3.49)

−2.61 × 10−6 (0.47) 0.000302 (0.39) −.0456*** (11.58)

7.25 × 10−08 (0.62) −0.0001 (1.54) −.00003 (0.60)

0.00002** (2.00) −0.0029* (1.88) 0.0020** (2.24)

−1.91 × 10−7 (0.15) 0.0001*** (3.58) 0.000 (0.45) 259,461 0.0531

1.83 × 10−08 *** (2.75) 2.79 × 10−06 *** (2.99) −2.03 × 10−07 * (1.95) 259,461 0.0002

2.78 × 10−06 ** (2.04) 0.00004* (1.86) −1.42 × 10−7 *** (2.64) 16,870 0.0062

Notes: Reported are the marginal effects (ME), evaluated at the means, from a multinomial probit regression. D-D is the base group in the multinomial probit model. D-DX is the base group in the rare event logit model. Control dummies are included for the industry, region, sample year and legal type of the companies. The entry (d) indicates that we report the estimated effect of a discrete change of dummy variable from 0 to 1. Pr(change) reports the predicted probability to change from D to DX and DI, respectively. The z-statistics are reported in parentheses. * p < 0.10. ** p < 0.05. *** p < 0.01.

With respect to the ownership structure, the presence of corporate shareholders increases the likelihood of domestic companies to start exporting by 40% (i.e., relating the marginal effect of 2.13 percentage points to the sample mean of 5.31%). The transition probability to become an exporter even doubles for firms with foreign corporate shareholders. In addition, the presence of a financial shareholder increases the likelihood to start exporting by 25%. Corporate ownership is also a significant driver for the foreign investment decision. The transition probability is six times larger for firms with corporate owners compared to domestic companies without corporate shareholders. Foreign ownership has no additional impact for domestic companies to invest abroad. In contrast, the higher the number of domestic subsidiaries the more likely a domestic company will establish a foreign affiliate. Findings for the financial indicators are rather mixed. Only one of the three indicators confirms the prediction that financing constraints matter. Domestic companies with a high non-current liability ratio exhibit a significantly lower likelihood to become exporters. This finding is in line with Stiebale (2011) for his static model. One standard deviation increase implies a 0.73 percentage point higher propensity to export. Since non-current liabilities are binding over a longer period of time, business activities cannot expand very quickly. This may indicate a disciplinary function of high debt levels as suggested by Jensen (1986). He argues that debt often has an important control function in high-leverage firms, since repayment of debts limits free cash flow. Interestingly, the non-current liability ratio does not matter for the decision to invest abroad. This finding can be explained by corresponding investments at home due to strong intra-firm relations with foreign affiliates (see Braunerhjelm and Oxelheim, 2002 for empirical evidence) implying a higher non-current liability ratio. Confirming recent findings of other authors, the age of a company has a positive and significant effect on the probability to internationalize (e.g., Bernard and Jensen, 2004). Whereas the number of employees has no distinctive effect on the export decision, it significantly impacts the establishment of foreign affiliates.10 The latter finding might reflect that a significant

10 Very similar results are obtained for specifications with operating revenue, instead of employees, as to capture general firm size effects. Results are available upon request.

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

43

Table 4 Foreign market entry and exit by exporters. Multinomial probit modelBase group DX-DX Productivity Total factor productivity Multi-unit and ownership structure Corporate shareholder (d) Corporate × foreign shareholder (d) Financial shareholder (d) Domestic subsidiaries Financials Cash flow ratio Current liability ratio Non-current liability ratio Basics Employees Age Capital intensity Number of observations Pr(change)

DX-DIMarginal effect

DX-DMarginal effect

0.0018*** (5.13)

−0.0191*** (9.21)

0.0093*** (10.33) 0.0009 (0.86) 0.0107*** (3.98) 0.0014*** (6.20)

−0.0400*** (10.56) −0.0455*** (5.42) −0.0426*** (4.54) −0.0026 (0.89)

−1.59 × 10−06 (0.20) −0.0024** (2.37) 0.0036*** (4.03)

−0.0001* (1.88) 0.0271*** (4.00) 0.0369*** (3.88)

3.70 × 10−07 (1.57) 0.0001*** (3.67) 1.75 × 10−08 (0.15)

4.56 × 10−06 (1.17) −0.0012*** (10.74) −1.62 × 10−07 (0.08) 73,084

0.0072

0.209

Notes: Reported are the marginal effects (ME), evaluated at the means, from a multinomial probit regression. DX-DX is the base group. Control dummies are included for the industry, region, sample year and legal type of the companies. The entry (d) indicates that we report the estimated effect of a discrete change of dummy variable from 0 to 1. Pr(change) reports the predicted probability to change from DX to DI and D, respectively. The z-statistics are reported in parentheses. * p < 0.10. ** p < 0.05. *** p < 0.01.

amount of domestic resources is needed to make up for the high level of sunk costs associated with foreign investments. Finally, higher capital intensity reduces the probability of a domestic company to become engaged in FDI, although the estimated effect is very small. This result indicates that sunk costs at home are a binding constraint to duplicate production abroad. In a supplementary rare event logit regression (see Tomz et al., 1999 for their implementation in STATA), reported in column (3), we explore the decision between the two internationalization modes in more detail. The dependent variable takes the value 1 if a domestic company decided to become engaged in FDI, and the value 0 if it decided to become an exporter. Larger, older and more productive companies are more likely to enter foreign markets via FDI. Thus, the productivity ranking postulated by Helpman et al. (2004) is clearly confirmed. Firms with corporate and/or financial owners are more likely to internationalize via FDI than by exporting, although no additional effect is found for foreign ownership. Finally, the results confirm that a firm with a higher number of domestic subsidiaries favours internationalizing via FDI. Significant effects with respect to firm’s size, age, and capital intensity suggest that firms starting to invest abroad are a rather selective group as compared to firms merely becoming exporters. 4.2. Foreign market participation of exporters As a next step we analyse the internationalization behaviour of French exporters in a multinomial probit model. They can either remain exporters (DX-DX, reference group), become engaged in FDI (DX-DI) or stop exporting (DX-D). The regression results are presented in Table 4. We find that more productive exporters are significantly more likely to become engaged in FDI and significantly less likely to cease their export operations. Thus, productivity exerts a positive effect on export entry (Section 4.1) as well as on continuing or even expanding export activities. Numerically, the marginal effect of an increase in productivity is even larger on the decision to stop exporting. Furthermore, our results suggest that exporters with corporate and financial shareholders are more likely to become engaged in FDI and less likely to stop exporting. The latter finding corresponds with other empirical studies (e.g., Roberts and Tybout, 1997; Roper et al., 2006). With respect to economic significance, the predicted transition probability to become engaged in FDI differs substantially, from 0.72% for exporters without foreign corporate shareholders to 1.75% for exporters

44

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

with corporate shareholders. Similarly, the predicted probability to exit the export market decreases considerably from 20.9% to 12.4% for exporters with foreign corporate owners. Thus, moving between domestic markets only and exporting is not only affected relatively symmetrically by corporate ownership, but – in both directions – even more so by foreign corporate ownership. Overall, the ownership structure has a sizable influence on companies’ internationalization behaviour. Recalling the results of Table 3, external ownership is a distinctive determinant to start and continue exporting, thereby depicting a symmetric effect on exporting. In line with previous findings for the D-DI decision, a larger network of domestic subsidiaries tends to increase exporter’s likelihood to further expand international activities by establishing foreign subsidiaries. This may be due to learning effects as firms with domestic subsidiaries might be more efficient in establishing foreign affiliates. The probability of multi-unit firms to stop exporting does not significantly differ from single-unit firms. This result contradicts to recent findings of Ilmakunnas and Nurmi (2010) for Finish plants who observe a positive effect on the decision to spare potential export markets. Overall, a multi-unit firm structure is asymmetrically linked to export participation. While current cash flow has no significant effect on exporters’ decision of establishing foreign affiliates (DX-DI), a low cash flow tends to significantly decrease the likelihood of ceasing export activities (DX-D). Quantitatively displaying a different order of magnitude, a higher current liability ratio significantly increases the likelihood to stop exporting and decreases the likelihood to become engaged in FDI. This finding suggests that financing constraints have a symmetric effect. The noncurrent liability ratio has a positive effect on the decision to invest abroad as part of these funds might have been used for corresponding home investments (see Section 4.1). By contrast, a downward change might reflect the binding constraint of poor financial performance, as indicated by a higher level of non-current liabilities. Finally, whereas firm size and capital intensity display no significant effect in the above estimations, older and more mature firms are not only more likely to continue exporting, they are also more inclined to expand their international footprint via FDI than younger firms. Thus, our results clearly confirm a symmetric pattern of firm age for export starters and continuing exporters. 4.3. Divestment decisions by MNEs Our final analysis concentrates on 2556 MNEs that either continue to be engaged in FDI (DI-DI), divest all their foreign affiliates in order to become exporters (DI-DX) or completely retreat from international markets (DI-D). The regression results are presented in Table 5. While our findings in Sections 4.1 and 4.2 suggest that productivity is a strong driver for the internationalization of firms, parent companies’ productivity does not exhibit any significant impact on the decision of MNEs to reduce their international activities. Once the decision to invest abroad is taken, MNEs are obviously prepared to pass through phases of low productivity. High levels of sunk cost could explain why firms go for long-term engagements abroad in order to steadily improve profitability and to recoup initial investment cost. Whereas domestic corporate ownership seems to have no direct effect on MNEs divestment decisions, those MNEs with foreign corporate shareholders display a significantly lower propensity to change their mode of internationalization downward from FDI to exporting only. Thus, foreign corporate owners play a key role for firms’ global market exposure by enhancing the probability of entry and lowering the exit probability. This finding contradicts the assumption that foreign corporate owners induce a higher divestment probability given their possibility to shift resources from one affiliate to another. Moreover, the presence of a financial owner increases the likelihood of a MNE to become a pure exporter by some 7.6 percentage points. In fact, this finding supports the argument that financial investors rather go for short-term returns on investment than for a sustainable and long-term business development of the target firm. The estimated effects of the ownership variables in the DI-D case, however, are all insignificant. The extent to which changes in ownership structure have an impact on the internationalization behaviour of firms cannot be monitored and remains subject to additional research. The existence of domestic subsidiaries increases the likelihood of a downward change for MNEs for both divestment modes. While Bernard and Jensen (2007) detect a significant positive effect of domestic plant closure, we confirm this finding for divesture of foreign affiliates. Specifically, evaluated at the sample mean, the predicted divestment probability increases by 0.2 percentage points for each additional domestic subsidiary. The estimated effects of the financial variables are mainly insignificant. Thus, while we and other authors find mixed empirical results for the entry decision, our data also reveal that MNE’s financial performance at home has little predictive power on complete foreign divestments. The negative sign for current liabilities is unexpected but only weakly significant. Similarly, firm size, measured by the number of employees at the headquarters, does not allow differentiating between continuing and exiting MNEs. Finally, older MNEs are less likely to fully stop their foreign market engagement, although the absolute effect of age on the predicted divestment probability is rather small. Even though most of our analysis is framed on the level of the parent company, when analyzing foreign divestment decisions, foreign affiliates’ characteristics add valuable information, too. Apart from the central role of productivity we extend the model by including some additional and distinctive characteristics of foreign affiliates. With the exception of the foreign location, remaining affiliates’ characteristics suffer from many missing values. For example, labour productivity can only be calculated for 429 of the 8624 foreign subsidiaries. Following Maddala (1997, p. 2002) we include a dummy variable for each characteristic in order to properly account for missing variable information. The results for variables of main interest are depicted in Table 6. Subsidiary’s productivity has the expected negative sign, but remains insignificant

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

45

Table 5 Foreign divestments of MNEs. Multinomial probit modelBase group DI-DI Productivity Total factor productivity (parent company) Multi-unit and ownership structure Corporate shareholder (d) Corporate × foreign shareholder (d) Financial shareholder (d) Domestic subsidiaries Financial characteristics Cash flow ratio Current liability ratio Non-current liability ratio Basics Employees Age Capital intensity Number of observations Pr(change)

DI-DXMarginal effect

DI-DMarginal effect

0.0046 (0.81)

−0.0018 (0.58)

−0.0144 (0.89) −0.0411** (2.26) 0.0760*** (3.21) 0.0020* (1.74)

−0.0043 (0.47) −0.0113 (1.07) 0.0040 (0.36) 0.0020*** (4.30)

3.08 × 10−06 (0.28) −0.0047 (0.17) −0.0221 (0.44)

5.95 × 10−07 (0.11) −0.0304** (1.82) −0.0263 (0.88)

−7.56 × 10−06 (1.32) −0.0003 (1.03) −0.0002* (1.72)

5.20 × 10−08 (0.05) −0.0003** (1.99) 0.0000 (0.27) 2556

0.122

0.038

Notes: Reported are the marginal effects, evaluated at the means, from a multinomial probit regression. DI-DI is the base group. Control dummies are included for the industry, region, sample year and legal type of the companies. The z-statistics are reported in parentheses. Pr(change) reports the predicted probability to change from DI to DX and D, respectively. The entry (d) indicates that we report the estimated effect of a discrete change of dummy variable from 0 to 1. * p < 0.10. ** p < 0.05. *** p < 0.01. Table 6 Foreign divestments of MNEs – subsidiary level. Multinomial probit modelBase group DI-DI Foreign subsidiary: characteristics Labour productivity Employment cost ratio Ownership share Location: Eastern Europe (d) Location: rest of the world (d) Number of observations Pr(change)

DI-DXMarginal effect

DI-DMarginal effect

−0.008 (1.35) 0.001 (0.05) 0.0003 (1.23) −0.011 (0.65) 0.041** (1.97)

−0.009 (1.16) −0.023 (1.01) −0.0001 (0.30) −0.032* (1.76) 0.044** (2.28) 8624

0.108

0.086

Notes: Reported are the marginal effects, evaluated at the means, from a multinomial probit regression. DI-DI is the base group. Variables for parent company (see Table 5) are included. The z-statistics are reported in parentheses. Pr(change) reports the predicted probability to change from DI to DX and D, respectively. The entry (d) indicates that we report the estimated effect of a discrete change of dummy variable from 0 to 1. * p < 0.10. ** p < 0.05.

at conventional levels of significance. The results remain unchanged when estimated for each of the three subsamples (2000, 2002, 2005).11 Therefore the main finding is confirmed: Whereas companies with a higher productivity are more likely to invest abroad, productivity has no significant effect on their decision to completely retreat from foreign markets.

11

Results are available from the authors upon request.

46

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

Furthermore, the estimates suggest that foreign affiliates of French firms located outside Europe have a higher probability to be divested than those located within Europe. 5. Conclusion In this paper we analyse the internationalization behaviour of more than 300,000 French domestic companies, exporters and MNEs between 2000 and 2007. Our data indicate that a rather small but distinctive group of companies increase or decrease their international engagement. The transition probabilities to enter and exit foreign markets via exporting are much higher than for FDI. Yet, symmetries might exist for key company characteristics with respect to the internationalization decisions of firms. Overall, a symmetric pattern of firms’ attributes is mainly found for export starters and continuing exporters, whereas an asymmetric pattern dominates investment and divestment decisions. An exception is the role of firm age. The older the company, the larger is the propensity to start and continue foreign market activities with respect to both exporting and FDI. Relative to the central role of productivity we clearly detect a symmetric pattern for the export decision. High-productivity companies are more likely to start and continue exporting. Similarly, with the ownership structure being key for the internationalization strategy of a company, we find that firms with financial, domestic and/or foreign corporate shareholders are more likely to start and continue serving export markets. The FDI participation deserves a special focus as we notice that key determinants affect FDI and export decisions differently. For example, short-term liabilities reduce the propensity to invest abroad but have no effect on the propensity to start exporting. In line with economic theory and empirical findings, productivity enhances the likelihood of domestic companies and exporters to become engaged in FDI. However, productivity seems to have no significant effect on the divestment decision – neither on parent nor on subsidiary level. In particular, high levels of sunk cost, potential re-entry cost related to FDI and knowledge spillover from foreign affiliates to the parent company might prevent an immediate market exit even if MNEs or their affiliates are underperforming. Multi-unit and ownership characteristics impact the FDI decision asymmetrically: multi-unit firms and those with financial shareholders are more likely to become engaged in FDI. Nevertheless, these MNEs are also more likely to completely divest from abroad. In summary, our empirical findings suggest that entering and exiting foreign markets via exporting or FDI is determined by varying symmetry patterns depending on the specific mode and the influence of various firm characteristics. One potential path for future research could be to analyse incremental and intra-company network changes. MNEs, for example, extend their existing foreign network by an additional affiliate. Moreover, the divestment decision of a domestic subsidiary might be directly linked to the establishment of a foreign subsidiary. Thus, capturing those dynamic network effects could provide valuable information on firm’s continuous internationalization decisions. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jebo.2012.10.016. References Almus, M., Engel, D., Nerlinger, E.A., 1999. Determinanten des Beschaeftigungswachstums junger Unternehmen in den alten und neuen Bundeslaendern: Bestehen Unterschiede hinsichtlich der Technologieorientierung? Zeitschrift für Wirtschafts- und Sozialwissenschaften 119, 561–592. AMADEUS database, release 168, 146, 136, 113 and 88. Bureau van Dijk Electronic Publishing (BvDEP). Amit, R., 1986. Cost leadership strategy and experience curves. Strategic Management Journal 7, 281–292. Benito, G.R., 1997. Divestment of foreign production operations. Applied Economics 29, 1365–1378. Benito, G.R., 2005. Divestment and international business strategy. Journal of Economic Geography 5, 235–251. Bernard, A.B., Jensen, J.B., 1995. Exporters, jobs, and wages in U.S. manufacturing: 1976–1987. Brookings Papers on Economic Activity: Microeconomics, 67–119. Bernard, A.B., Jensen, J.B., 2004. Why some firms export. The Review of Economics and Statistics 86, 561–569. Bernard, A.B., Jensen, J.B., 2007. Firm structure, multinationals, and manufacturing plant deaths. Review of Economics and Statistics 89, 1–11. Bernard, A.B., Wagner, J., 2001. Export entry and exit by German firms. Review of World Economics 137, 105–123. Bernard, A., Eaton, J., Jensen, J.B., Kortum, S., 2003. Plants and productivity in international trade. American Economic Review 93, 1268–1290. Boddewyn, J.J., 1979. Foreign divestment: magnitude and factors. Journal of International Business Studies 10, 21–27. Boddewyn, J.J., 1983. Foreign and domestic divestment and investment decisions: like or unlike? Journal of International Business Studies 14, 23–35. Braunerhjelm, P., Oxelheim, L., 2002. Does foreign direct investment replace home country investment? The effect of European integration on the location of Swedish investment. Journal of Common Market Studies 38, 199–221. Bricongne, J.-C., Fontagné, L., Gaulier, G., Taglioni, D., Vicard, V., 2012. Firms and the global crisis: French exports in the turmoil. Journal of International Economics 87, 134–146. Birkinshaw, J., Hood, N., Young, S., 2005. Subsidiary entrepreneurship, internal and external competitive forces, and subsidiary performance. International Business Review 14 (2), 227–248. Caves, R.E., 1998. Industrial organization and new findings on the mobility and turnover of firms. Journal of Economic Literature 36, 1947–1982. Chaney, T., 2005. Liquidity Constrained Exporters. Mimeo, University of Chicago. Duhaime, I.M., Grant, J.H., 1984. Factors influencing divestment decision-making: evidence from a field study. Strategic Management Journal 5, 301–318. Dunne, T., Roberts, M.J., Samuelson, L., 1988. Patterns of firm entry and exit in U.S. manufacturing industries. The RAND Journal of Economics 19, 495–515. Filatotchev, I., Stephan, J., Jindra, B., 2008. Ownership structure, strategic controls, and export intensity of foreign-invested firms in transition economies. Journal of International Business Studies 39, 1133–1148. Geroski, P.A., 1995. What do we know about entry? International Journal of Industrial Organization 13, 421–440.

D. Engel et al. / Journal of Economic Behavior & Organization 85 (2013) 35–47

47

Geweke, J., Keane, M., Runkle, D., 1994. Alternative computational approaches to inference in the multinomial probit model. The Review of Economics and Statistics 76, 609–632. Greenaway, D., Kneller, R., 2007. Firm heterogeneity, exporting, and foreign direct investment. Economic Journal 117, F134–F161. Greenaway, D., Guariglia, A., Kneller, R., 2007. Financial factors and exporting decisions. Journal of International Economics 73, 377–395. Gupta, A.K., Govindarajan, V., 1994. Organizing for knowledge within MNCs. International Business Review 3, 443–457. Haynes, M., Thompson, S., Wright, M., 2000. The determinants of corporate divestment in the UK. International Journal of Industrial Organization 188, 1201–1222. Haynes, M., Thompson, S., Wright, M., 2003. The determinants of divestment: a panel data analysis. Journal of Economic Behaviour and Organisation 52, 47–166. Helpman, E., Melitz, M.J., Yeaple, S.R., 2004. Export versus FDI with heterogeneous firms. American Economic Review 94, 300–316. Hopenhayn, H., 1992. Entry, exit, and firm dynamics in long run equilibrium. Econometrica 60, 1127–1150. Ilmakunnas, P., Nurmi, S., 2010. Dynamics of export market entry and exit. Scandinavian Journal of Economics 112, 101–126. Institut national de la statistique et des études économiques (INSEE), 2012. http://www.insee.fr/fr/themes/tableau.asp?reg id=0&ref id=NATTEF09225 (March 2012). Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance and takeovers. American Economic Review 76, 323–329. Lancheros, S., Demirel, P., 2012. Does finance play a role in exporting for service firms? Evidence from India. The World Economy 35, 44–60. Levinsohn, J., Petrin, A., 2003. Estimating production functions using inputs to control for unobservables. Review of Economic Studies 70, 317–342. Levinsohn, J., Petrin, A., Poi, B., 2003. Production function estimation in Stata using inputs to control for unobservables. Stata Journal 4, 113–123. Maddala, G., 1977. Econometrics. McGraw-Hill, New York. Manova, K., 2012. Credit constraints, heterogeneous firms and international trade. NBER, Working Paper, 14531. Forthcoming at the Review of Economic Studies. Mayer, T., Méjean, I., Nefussi, B., 2010. The location of domestic and foreign production affiliates by French multinational firms. Journal of Urban Economics 68, 115–128. Melitz, M., 2003. The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71, 1695–1725. OECD, 2008. Benchmark Definition of Foreign Direct Investment, 4th ed. OECD, Paris. Procher, V., 2011. Agglomeration effects and the location of FDI: evidence from French first-time movers. Annals of Regional Science 46, 295–312. Roberts, M.J., Tybout, J.R., 1997. The decision to export in Colombia: an empirical model of entry with sunk costs. American Economic Review 87, 545–556. Roper, S., Love, J.H., Higon, D.A., 2006. The determinants of export performance: evidence for manufacturing plants in Ireland and Northern Ireland. Scottish Journal of Political Economy 53, 586–615. Stiebale, J., 2011. Do financial constraints matter for foreign market entry? A firm-level examination. The World Economy 34, 123–153. Tomz, M., King, G., Zeng, L., 1999. RELOGIT: rare events logistic regression, Version 1.1. Harvard University, Cambridge, MA. The current version of this program, which is available at http://gking.harvard.edu/ Wagner, J., 2003. Unobserved firm heterogeneity and the size-exports nexus: evidence from German panel data. Review of World Economics 139, 161–172. Wagner, J., 2007. Exports and productivity: a survey of the evidence from firm level data. The World Economy 30, 60–82. Wagner, J., 2008. Export entry, export exit and productivity in German manufacturing industries. International Journal of the Economics of Business 15, 169–180. Wernerfeldt, B., 1984. A resource based view of the firm. Strategic Management Journal 5, 171–180. Watts, H.D., Kirkham, J.D., 1999. Plant closures by multi-locational firms: a comparative perspective. Regional Studies 33, 413–424.