The jobs at risk from globalization: the French case

Abstract

This article analyzes the effect of outward foreign direct investment (FDI) on the workforce composition in French firms. We use a detailed employer-employee database constructed with four comprehensive datasets of French manufacturing firms over the period 2002–2007, in order to analyze changes in the workforce composition in terms of skills and tasks. To deal with endogeneity issues, we propose an IV strategy where the level of infrastructure and GDP per capita in the host countries are used as instruments. The fixed effect results show that FDI to low-income countries raises significantly the share of executives and reduces the share of blue-collar workers in company workforces in France. Outward FDI to high-income countries affects negatively the share of workers performing non-routine manual tasks. When controlling for endogeneity, the IV results further show an overall positive effect of offshoring for employees performing interactive and analytical tasks, such as engineers and managers.

Appendices

Appendix 1: The composition of tasks within occupations

We classified the 41 main work activities into five major categories as in ALM (2003). They distinguish between: (i) non-routine analytical tasks are usually performed in technical or managerial occupations. They require cognitive capacity in which responsiveness, creativity, decision making and problem solving are important. (ii) The second major task category includes non-routine interactive tasks. These tasks require physical interaction and adaptability to certain types of situations. (iii) The third category identifies cognitive routine tasks that are usually carried out by administrative or clerical occupations, such as secretaries and accounting officers who perform repetitive tasks using an identified procedure. (iv) Manual routine tasks are performed by production operators such as handlers, machine operators, workers in packaging and transportation. These tasks can be seen as unskilled manual tasks. (v) Finally, the last category lists manual tasks that require specific knowledge and are considered as skilled manual tasks. These tasks are mostly performed by technicians or foremen (Table 11).

Table 11 Occupational tasks

The following table gives examples of different types of tasks performed in four main occupations in the metal industry. Here we see that engineers have a higher score in making decision, whereas mechanics have the lower score. In contrast, skilled assemblers have a higher score in the task of handling and moving objects compared to engineers (Table 12).

Table 12 Examples in the metal industry

Table 13 provides descriptive statistics, showing the mean intensity of each score for the seven major occupations.

Table 13 Tasks’ intensity within occupations

It shows that skilled production workers (composed of skilled blue-collar workers, engineers, technicians and foremen) have an important index of non-routine manual, interactive and analytical tasks. Managers and engineers are the two occupations having the highest score of routine cognitive tasks, whereas skilled and unskilled blue-collar workers perform routine manual tasks more intensively.

Mapping the SOC classification of ONET to the French PCS-ESE The ONET databases provides information on 900 occupations classified according to a classification close to the American SOC classification. In order to build a task index compatible with the French classification of occupations (PCS-ESE), we need to build a mapping table from PCS-ESE to SOC 2010. We have built it thanks to the EurOccupations database.⁴⁷ The EurOccupations project aimed at building a publicly available database containing the most common occupations in multi-country data-collection. The database includes a source list of 1594 distinct occupational titles within the ISCO-08 classification and provides a mapping table between the EurOccupation classification and the ISCO-08 classification, as well as a French translation of these occupations. We then match the 412 PCS-ESE occupational classification for which there is at least a perfect pair with occupations described in the EurOccupation database. Finally, a mapping table from the ISCO-08 to the SOC-2010 classification is used to link the PCS-ESE occupational classification with the SOC-2010. By creating this mapping table, we can use the ONET index to analyze the task content of French occupations.

Appendix 2: Robustness tests

The following section gives results of Eq. 3 by using different measures of offshoring. We define three other measures of FDI.

The first proxy is constructed by using the percentage of votes at general meeting. The aim is to take into account the possibility that a firm uses its blocking minority power.⁴⁸ The underlying idea for using blocking minority as a weighted measure is to assuming increasing control over the foreign subsidiary when the firm has the possibility to veto certain decisions at extraordinary general meetings of shareholders. The investing firm could thus have incentives to actively manage employment inside the subsidiary and to transfer its intangible assets.

The measure is built as follows:

$$\begin{aligned} \rho _{jt} = \sum \limits _k {subsidiary_{kjt} } \end{aligned}$$

(6)

where $\left\{ \begin{array}{ll} subsidiary_{kjt} = 1,&\quad when\, Voting_{kjt} \ge 34\,\% \\ subsidiary_{kjt} = Voting_{kjt},&\quad when\, Voting_{kjt} < 34\,\% \\ \end{array} \right.$

Here, we assume that when a firm owns more than 34 % of the voting stock, then the firm has a blocking minority and can intervene in important decisions concerning its subsidiary. Therefore the firm owns the subsidiary entirely, and the number of FDI projects is equal to 1. If the firm does not hold a blocking minority, then we do not consider that the firm holds the entire subsidiary, but only holds the subsidiary weighted by the voting stock.

The $\rho$ measure controls for the blocking minority, but has the disadvantage of overriding the existing hierarchy when votes are held.

We also control our results by using a second proxy of a firm’s FDI size, by weighting the firm’s number of subsidiaries by firm $j$’s average outward investment stock.⁴⁹ The third measure is constructed as follows:

$$\begin{aligned} \varphi _{jt} = \sum \limits _k {Value_{kjt} } \end{aligned}$$

(7)

where $Value_{kjt}$ is the balance sheet value of firm $j$’s subsidiary $k$ at time $t$. However, information on the gross value of investment purchased in each subsidiary is not available prior to 2004. As it is a stock value, the average over a 4-year period gives a good proxy of a firm’s mean size of foreign investments. We first measure the number of foreign subsidiaries controlled by each firm $j$ over the period $t \in \left[ {2002,2007} \right]$. We weight this number by the firm $j$’s mean outward investment stock amount over the period 2004–2007.

However, the measure $\varphi$ does not account for all multinational companies, since information on subsidiaries’ stock amounts are not referenced for each subsidiary in the LIFI survey.

Finally, we use the number of subsidiaries abroad without weights (we call this $\delta$).

The measure $\kappa$ is preferred for different reasons. First, it allows the effect of a change in the share of voting stock held in a foreign subsidiary to be captured, even if the firm does not change its number of foreign subsidiaries. Several studies have highlighted the impact of control over wages and employment (see Bircan 2011 for a review). Second, it allows control for multiple authority sources in a group. Indeed, if two French subsidiaries inside a domestic group control the same subsidiary abroad, it allows firm holdings with a clear majority to be distinguished.

Tables 14 and 15 report results with the different measures of FDI on the share of executives and blue-collar workers respectively. Columns (1) and (7) report the results with the measure of offshoring $\kappa$. Columns (2) and (8) report the results with the measure of offshoring $\rho$. Columns (3) and (9) report the results with the measure $\varphi$. Columns (4) and (10) report the result with the measure of offshoring $\delta$. Columns (5) and (11) report the results with the measure of offshoring $\kappa$, and include the variables on the wages paid by type of work $i$ relative to the composite wage of work type not in $i$. Whatever the proxy retained, the preceding conclusions are not altered.

Table 14 Robustness test with different measures of offshoring (executives)

Table 15 Robustness test with different measures of offshoring (blue-collar workers)

Columns (6) and (12) report the results with the measure of offshoring $\kappa$, by including the variable of parent company nationality. We add a dummy that takes the value of 1 if a firm is controlled by a foreign parent company, and 0 otherwise. The coefficient associated with this dummy is identified for firms that changed nationality in the sample period. The inclusion of the firm’s nationality does not alter the conclusions. FDI to low-income countries is significantly associated with a higher share of executives employees, and is associated with a lower share of blue-collar workers.⁵⁰

Finally, we report the results when controlling for the FDI strategy at the group level (see Table 16). The results are consistent with those obtained by taking into account the relocation strategy at the firm level. A group’s number of subsidiaries in low-income countries raises the share of executives inside domestic firms. Relocation in high-income countries does not show any significant impact on either types of jobs and tasks. However, contrary to results at the firm level, when the group raises its FDI in high-income countries, blue-collar workers are negatively affected. The reason is certainly due to an overall effect on the rationalization of production processes by eliminating unnecessary domestic capacity at the group level. Indeed, when the group increases FDI to low-income countries it may decide to reduce the employment of blue-collar workers in dedicated subsidiaries, even those that are not directly engaged in outward FDI.

Table 16 International strategies at the group level

Appendix 3: Description of variables

We report a detailed description of groups of occupations, set out in Sects. 5.1, 5.2, 5.3 and 5.4. We also refer to their French classification (PCS-ESE 2003), available at http://www.insee.fr/ (Table 17).

Table 17 Description of the groups of occupations

Appendix 4: Regression on groups of occupations

See Table 18.

Table 18 Changes in the share of occupations by production groups

Appendix 5: Sector decomposition

Table 19 presents the average wage bill share for seven broad occupations per sector. The consumer goods industries are relatively more intensive in administrative workers (managers and employees), while the capital and intermediate goods industries are more intensive in skilled production workers (engineers, skilled blue-collars and technicians). The share of unskilled blue-collar workers and foremen are relatively similar in the three sectors.

Table 19 Decomposition of the wage-bill share by sector

The following tables present the results of the fixed effect model by dividing the sample into three main industry categories. Firms belonging to consumer goods industries are reported in Table 20. Firms belonging to intermediate goods industries are reported in Table 22. Finally, firms belonging to capital goods industries are reported in Table 21. 20.82 % of our sample are in the consumer goods industries, 23.69 % are in capital goods industries and 50.99 % are in the intermediate goods industries.

Table 20 Fixed effect model for the consumer goods industries

Table 21 Fixed effect model for capital goods industries

Table 22 Fixed effect model for intermediate goods industries

Table 20 shows no clear effect of offshoring on employment. Whereas Table 21 shows a positive effect of offshoring to low-income countries on the share of executives, and the share of workers performing intensively cognitive and analytical tasks.

Interestingly, Table 21 shows that the negative bias towards manual workers when offshoring occurs in high-wage countries and is mainly driven by firms in capital goods industries, as described in Sect. 5.2. We also observe a positive association between offshoring in high-income countries and the share of workers performing interactive tasks. This is similar to results obtained in Becker et al. (2013).

Table 22 shows a clear substitution between blue-collar workers and a complementarity between skilled executives and offshored employment in low-income countries. Similar results were obtained for Sweden by Hansson (2005).

Appendix 6: First-stage estimates

Results of the first-stage estimates show that there is a positive correlation between our $\kappa$ measure, reflecting the number of subsidiaries, in high-income countries and the average GDP per capita in the destination country. Conversely, there is a positive correlation between GDP per capita in low-wage countries and the number of subsidiaries in low-wage countries. Overall, we observe that increasing GDP per capita raises the incentive for firms to invest, whatever the type of country (low-income or high-income).

Interestingly, we observe a positive correlation between FDI to high-income countries and the GDP per capita in low-income countries. FDI to low-income countries and more particularly in emerging economies combines strategies of market access and saving labor costs (which are often called mixed investment strategies). On the one hand, production costs are lower than in developed countries, which increasingly drive efficiency-seeking investments. On the other hand, growing demand and strong market potential increase incentives to invest in emerging economies. Mixed investment strategies represent the major share of investment, since roughly 80 % of FDI to low-income countries is directed towards only ten emerging countries with strong growth potential. Large firms in European countries and French CAC 40 companies, in particular, have created more jobs and business in emerging countries than in high-income European countries in recent years. This is because a part of the global demand and growth have shifted to emerging markets (UNCTAD 2013). Therefore, increasing low-income countries’ GDP per capita would raise consumer demand without increasing production costs as much as in developed countries. A firm could gain in efficiency and profitability as a result of increasing low-income countries’ GDP, and would allow it to raise FDI in high-income countries.

In contrast, we observe a negative and significant correlation between high-income countries’ GDP per capita and FDI to low-income countries. These results show that, other things being equal, increasing market potential in high-income countries reduces the incentive for firms to invest in low-income countries. This corroborates the result that market access is a powerful force of attraction for companies (Fontagne and Mayer 2005).

Results for infrastructure variables are intuitive and are in line with the existing, vast literature. They show that the quality of infrastructure is an important source of comparative advantage. Increases in the level of infrastructure in low-income countries raise the incentive for firms to invest in these countries. The same conclusion holds for offshoring to high-income countries.