Export and Firm Performance: Evidence on Productivity and Profitability of Italian Companies

Abstract

Empirical research in international trade has shown that exporting firms display higher productivity than their non-exporting competitors. This paper focuses on the relation between export and profitability. Our evidence on Italian exporting firms shows that exporting activity is not systematically associated to higher firm’s profitability. This is shown both by means of non-parametric methods and, with an approach that is more standard within the empirical trade literature, by regression techniques that try to identify an “export premium”.

Appendix: Thresholds for export status

In the paper we content ourselves with a definition of exporter as that of any firm that in a given year has reported a positive fraction of turnover (however small) from sales abroad. Table 1 reported the percentage of exporting firms without considering any threshold of export intensity, and so did the distributional analysis in Fig. 1.

The purpose of this Appendix is to investigate if the results we obtain in the paper significantly depend on the definition of exporter. Scholars have investigated the issue of export intensity in relation to performance variables, most notably firm’s productivity (see, among the others, Aw et al. 2000; Castellani 2002; Bernard et al. 2003 and Fryges and Wagner 2010 on profitability), however the literature does not, nor the empirical evidence, provide conclusive results. Another relevant dimension in order to try answering our original question, is the time window. A firm that is willing to expand by going international will have to pay a sunk cost of entry, which is also specific to each destination country.²⁵ As such we do expect that once a firm has taken such a decision, and paid the relative entry cost, it will tend to persist in the export market for some years to follow. Thus, a further distinction that might turn useful for understanding who are “exporters” is that between “systematic” exporters, those firms that once have started to export then persist in the export market, and “occasional” exporters, those firms that are sporadically active on the international markets.

We concentrate on those sectors accounting for a relevant share of national exports (as shown in Table 3) and among these, we will choose sectors also in order to provide a representative picture of the whole manufacturing industry in Italy. To this end we consider the sectors of Manufacture of textile (NACE 17), Basic metals (NACE 27), Machinery and equipment n.e.c. (NACE 29) as they satisfy these criteria, and in addition these sectors also have a high number of observations.

Plots in Fig. 2 display a measure of central tendency of some performance variables, namely size as proxied by total sales, productivity (labor and TFP), profitability, and growth, for firms exporting more than a given threshold of their turnover. The aim is to better understand if beyond a certain threshold of export intensity we do observe a given pattern in the variable of interest. In particular, our interest is to verify if the differences between the two groups of firms, exporters and non exporters, arise after some threshold of export intensity, or if, on the contrary, the only discriminant cutoff is the export status, irrespectively of its intensity.

Fig. 2

Row I: median size (as total sales in thousands of euro) at different threshold of export intensity. Row II: real average labor productivity at different threshold of export intensity. Row III: average profitability (as ROS) at different threshold of export intensity. Row IV: average one-year growth rates at different threshold of export intensity

Fig. 3

Row I: empirical density of size (as total sales) for systematic, occasional exporters and non exporters, NACE 17, 27, 29 in 2004. Row II: empirical density of labor productivity. Row III: empirical density of TFP. Row IV: empirical density of ROS. Row V: empirical density of growth rates

Table 9 Fligner–Policello test of stochastic equality on size labor and total factor productivity, returns on sales and growth rates for exporting vs non-exporting firms

Table 10 Labor Productivity regression with dummies for systematic, occasional and never exporting firms

Table 11 TFP (top panel) and ros (bottom panel) regression with dummies for systematic, occasional and never exporting firms

Every dot in the graphs represents the average²⁶ value of a given variable for firms exporting more than x% of their sales. Of course, we are aware that the performance variables that we are investigating are not normally distributed and they do display a very wide support (see for instance Bottazzi et al. 2005, 2007) so that sometimes the mean is a noisy indicator. In principle, one would want to compare at every threshold of export intensity the whole distribution of a variable for exporters and for non exporters in order to get a more precise assessment of the differences between the two groups.

Let us now consider the relation between different thresholds of export intensity and the median size of firms associated to that threshold. Many empirical works show that bigger firms are more likely to be exporters (Serti and Tomasi 2008; Bernard et al. 2007; Castellani 2002), and this is shown also in the plots in the first row of Fig. 2. Such a gap persists also at higher level of export intensity. Also notice that in the machine tool sector (NACE 29) after the clear gap that identifies firms starting to export, we do observe a slight increase of the median size at higher export intensities. The other sectors included in the analysis do not display such a feature.

The relation between labor productivity and export intensity is represented by plots in the second row of Fig. 2. For firms in the textile sector, NACE 17, in 2000 and 2004 the average labor productivity of all non-exporters was around 35 thousands euro (deflated with the sectoral production price index) as marked by the corresponding dots at the extreme left of the plot. A relevant “productivity premium” characterizes exporting firms. Again such difference has more to do with export status than export intensity. The average labor productivity of all firms exporting any positive share of their turnover is, both in 2000 and 2004, around 40 thousands euro, registering an increase of 15% with respect to non-exporting firms. This is very robust evidence, as it holds across all sectors included and is valid for all years.²⁷ Again, one notices that in the sector of machine tool, NACE 29, intensive exporters appear to be more productive than marginal exporters, and that the trend is linear. Results for TFP are much in line with those on labor productivity, in particular this is surprisingly so for the sector of machine tool, NACE 29.

Plots in fourth row of Fig. 2 compare the profitability of non-exporters to that of firms with different intensity of export. As far as profitability is concerned we do register, in most sectors and years, the opposite trend: exporters report lower profitability. The same lack of relationship is also apparent between export and firms’ growth, as displayed by the plots in the fifth and last row of Fig. 2.

Let us now consider the issue of persistency over time in the export status. To this purpose, in the following we define a firm as “systematic” exporter in 2004 if it has exported in 3 or more of the previous 5 years. We define accordingly a firm as “occasional” exporter if it has exported only once or twice in the same time interval, and residually, non exporters are those firms that never exported in these years. Since the database we employ for trade data, COE, contains information on all cross-border transactions performed by Italian firms, we are safe that we are not inducing any selection bias in the analysis by employing these definitions.

Plots in Fig. 3 display the empirical density of total sales, labor and TFP, return on sales and growth rates in 2004 for firms as classified in the categories of “systematic”, “occasional” exporters, and never exporters on the basis of their export activity in the previous five years. From the plots it is apparent that “systematic” exporters are bigger in all the industrial sectors that we consider, Textile, NACE 17, Basic Metal, NACE 27, and Machine tool, NACE 29. Further it is clear that there exists a ranking between the three groups of firms that holds in all sectors: “systematic” exporters are bigger than “occasional” exporters, and the latter are bigger than never exporters. The same ranking is observable also for the distributions of labor and TFP, whether, as in the plots of Fig. 1, the export activity is not apparently associated to better performance in terms of profitability or growth.