Domestic multinationals and foreign-owned firms in Italy: evidence from quantile regression.

• Author: Grasseni, Mara
• Position: Essay

1. Introduction and Literature Background

   Several studies based on firm or plant level data have documented the growing importance of firm heterogeneity. The pioneering works in this area focus on comparison between exporters and non-exporters (Bernard and Jensen, 1999; Clerides et al., 1998; Delgado et al., 2002), documenting that the former tend to outperform the latter. Recently, also foreign direct investment (FDI) has become an important issue in the discussion on heterogeneity. In this regard, the main prediction considered in empirical studies is that productivity differences among firms have a role in explaining the presence of domestic firms, exporting firms and investing firms (Helpman et al., 2004; Head and Ries, 2003). This is consistent with the self-selection hypothesis suggesting that firms engaged in some kind of foreign activity need to have some exante advantages in order to deal with the costs and the complexities of international markets. This idea has clear links with the consolidated literature on multinationals and according to which one reason why firms invest abroad is their desire to exploit firm-specific advantages in host countries (Markusen, 1995; Caves, 1996).

   The paper builds on this analysis and conducts empirical discussion on the less explored issue of heterogeneity across and within multinationals in Italy.

   Only a few studies in the literature explore the role of firm heterogeneity in explaining the relationships between productivity and internationalization strategies, comparing foreign affiliates and domestic firms and distinguishing the latter between non-multinational and multinational enterprises (MNEs) (Castellani and Zanfei, 2006; Temouri et al., 2008; Criscuolo and Martin, 2009; Doms and Jensen, 1998). These studies find essentially that foreign affiliates exhibit higher productivity than domestic non-MNEs, while foreign and domestic MNEs differ only marginally. Thus, in explaining the better productivity performance of MNEs, some of their advantages are more important than foreign ownership advantages per se. Regarding Italy, Castellani and Zanfei (2006) show that foreign-owned firms outperform domestic firms, but the gap disappears when foreign firms and domestic MNEs are compared. Moreover, US foreign firms perform better than affiliates from other countries, and they exhibit productivity levels similar to those of domestic MNEs. In a different work (2007), Castellani and Zanfei find that domestic multinationals with production activities abroad exhibit higher productivity as well as better innovation performance than multinationals with only non-production activities abroad. Overall, MNEs are characterized by better performances than exporters and domestic firms. However, the analyses in that work do not include foreign-owned firms.

   Finally, there are two studies that focus on productivity spillover from foreign investment and which employ the same econometric techniques as used in this paper: Dimelis and Louri (2002) and Kosteas (2008). The former examines data on Greek firms to find a positive effect on the labour productivity of fully or majority owned foreign affiliates. The latter study uses data on Mexican manufacturing plants to distinguish among FDI from North America, Canada, and the rest of the world. Kosteas finds that Canadian-owned plants yield higher spillovers than other foreign-owned plants, suggesting a large amount of heterogeneity among inward FDI. Both papers, however, focus on spillover effects and do not analyze performance differences in great detail: for example, they do not have information on outward investment and concentrate only on productivity.

   This paper makes two contributions to the empirical literature. Firstly, it presents a detailed analysis of the role of multinationality in explaining heterogeneity, focusing in particular on differences between foreign-owned firms and domestic MNEs in Italian manufacturing industries. Much of the analysis in this area has hitherto focused on productivity differences. This study furnishes additional insight into the issue of heterogeneity by investigating not only productivity but also average wages, capital intensity, as well as various measures of non-financial and financial performance, such as return on sales (ROS), return on investment (ROI), and debt leverage. The advantages of these three indicators are that they are easy to calculate and, especially, that their definitions are agreed and well-known: traditionally in the international business literature, the success of a company has been examined using these measures (Tangen, 2003).

   Secondly, the paper highlights the differences in performance among multinationals. To address this issue it distinguishes between foreign-owned firms of different nationalities and domestic MNEs according to the locality of their FDI. The nationality of foreign-owned firms may be crucial for understanding whether there is a performance leader among them. Such leadership, for instance, may be the consequence of an advantage of the home country compared to another. Performance gaps may exist even within domestic multinationals: in fact, the choice of the geographical areas in which affiliates are established may reflect distinct structural characteristics as well as distinct motivations.

   The empirical analysis reported by the paper is performed using Kolmogorov-Smirnov tests of stochastic dominance and by applying Quantile Regression Technique (QR). The non-parametric tests compare the cumulative distribution of the variable for different types of firms and not just the mean. The QR permits evaluation of the differences across and within multinationals at different points of the conditional distribution of the dependent variable. Therefore, if one acknowledges that multinationals are heterogeneous, there are reasons to suspect that the differences across and within firms do not need to be the same across the performance distributions. On the contrary, ordinary least squares method (OLS) assumes that the conditional distribution of the performance variables is homogeneous.

   The rest of the paper is organized as follows: Section 2 discusses the characteristics of the sample and describes the econometric framework. The results are presented in Section 3, and Section 4 concludes.

2. Data and Empirical Model

   2.1 Data and Descriptive Statistics

   Used for the empirical investigation of the issue of heterogeneity across and within multinationals operating in Italy during the 1990s is the "Centro Studi Luca D'Agliano-Reprint" dataset resulting from merging the Reprint dataset of Politecnico of Milan, which contains information on foreign affiliates (FO) and domestic multinationals ([MNE.sup.D]) with the AIDA database of Bureau Van Dijck, which provides balance sheet data and other economic data on Italian firms. (3)

   An useful feature of the dataset is that it stated the nationalities of foreign firms, thus making it possible to distinguish in the empirical study among US, European and Other foreign-owned firms ([FO.sup.US], [FO.sup.Europe] and [FO.sup.Other]). Moreover, regarding domestic multinationals, the dataset provides information on the country of localisation of their foreign subsidiaries, permitting comparative analysis among the different characteristics of domestic multinationals investing only in developed countries, investing only in less developed countries, and investing in both ([MNE.sup.D_DC], [MNE.sup.D_LDC], [MNE.sup.D_Both]). (4)

   Owing to limited information on the localisation of outward investment and given that observations for balance sheet data are missing, the econometric analysis is conducted using an unbalanced panel which includes data on foreign firms and domestic multinationals, with at least one foreign subsidiary, for the years 1995 and 1997.

   The variables used in the empirical analysis are: labour productivity, defined as value added per employee; average wages; capital intensity, defined as total tangible assets over number of employees; return on sales, ROS; return on investment, ROI; and debt leverage, Leverage, defined as total debt over equity.

   An overview of the distribution of firms by firm type, nationalities and localization of FDI and sectors is provided by Table 1, where firms are classified according to the Pavitt classification among traditional, high returns to scale, specialized, and high-tech sectors.

   In 1995, around 40% of domestic MNEs have subsidiaries located only in developed countries. There is the same percentage for outward investment in less developed countries, while only 19.6% of domestic MNEs have higher international involvement by adopting the strategy of investing in developed countries as well as in less developed ones. In 1997 there is an increase in the number of Italian firms choosing to invest abroad, but this rise essentially concerns firms with subsidiaries located in LDC. (5)

   In regard to the ownerships of foreign affiliates, in 1995 about 70% of them have European nationality and about 24% are from the US. In 1997 the foreign firms in Italy increase in number, but the percentage composition does not change. Overall, these stylised facts are not particularly surprising and confirm the great amount of foreign direct investment that takes place among developed countries, and especially among European countries.

   Table 1 also gives an overview of the sectoral distribution of firms. In the sample, the highest percentage of domestic MNEs are active in high return to scale sectors, around 40%, and traditional sectors, around 30%. The former have foreign subsidiaries especially in DC, around 50%, while the latter invest mainly in LDC, around 66%. Foreign-owned firms are more concentrated in high return to scale sectors, around 56%, and specialised sectors, around 22%; and in both sectors most FO have European nationality.

   Table 2 reports mean and standard deviations of the key variables...