Wage costs and job relocation since the enlarged European Union. Evidence from Firm Level Data.

Wage costs and job relocation since the enlarged

European Union. Evidence from Firm Level Data.

Author: A.H. Koopman Student number: S1588494 Email: s1588494@student.rug.nl

Tel.: 06 14 03 28 14

University of Groningen

Faculty of Economics and Business;

Supervisor International Business and Management: drs. H.C. Stek

Abstract

The enlargement of the European Union asks for renewed research on job (re)location. Firm level data of manufacturing firms in Europe are used to analyse whether wage cost

differentials between high and low wage location, respectively Western Europe and Central/Eastern Europe, can trigger employment relocation. First I have documented

differences in labour costs and productivity between Western and Central/Eastern European countries. The ratio of labour costs and productivity gives an indication of a country’s labour competitiveness. Labour costs are about 5 times lower in the typical firm in Central/Eastern Europe than labour costs in the high wage countries. But also labour productivity is more than 5 times lower in Central Europe. Despite the fact that this largely neutralizes competitive advantages I found out that since 2002 for Central and Eastern Europe a competitive

1. Introduction

At midnight 9 November 1989 East Germany's Communist rulers gave permission to open the gates along the Berlin Wall after hundreds of people converged on crossing points. After the fall of the Berlin Wall waves of democratisation have swept through Hungary, Poland, East Germany, Bulgaria, Romania and Czechoslovakia replacing many Communist governments

These newly democratised countries then tried to link themselves with the rest of Europe by implementing market and political reforms that would smooth their entry into the European Union. Eventually on 1 May 2004 the ten Central and Eastern European countries Estonia, Latvia, Lithuania, Poland, the Czech Republic, the Republic of Slovakia, Hungary and Slovenia joined the European Union. In the leading up to the enlargement discussion started whether the cheap labour force which was abundantly available in the CEE countries could seriously harm the domestic labour market of the old member states. Among others

policymakers and labour unions from high wages countries in Western Europe argued that low wage import competition from the new member states may result in job losses in the domestic country or that companies in WE could relocate their operations (F. Abramham et al., 2000). In my search for possible job losses I found that according to the Statistics

Netherlands since 2001 more than 600 large companies active in especially the manufacturing industries have relocated business processes abroad. An important contributing factor was that most companies are quite internationally oriented as they are part of multi-national

corporations. The main reason for companies to move activities out of the Netherlands was to save money, particularly saving wage costs. It was, however, difficult to estimate the

consequences of relocation for employment in the home country, some companies lost jobs at home, others even created new jobs at home (G. van Gessel and F. van Berkel, 2007).

The discussion about job losses to Eastern Europe is nowadays not so intensive anymore as it was at the time of the enlargement. The decrease in the interest for this subject reflects itself then also in the number of scientific papers. While research papers dated at the time of the enlargement are abundant recent papers about job relocation within Europe are in contrary rare. That the topic of job relocation has been researched intensively at the time the CEE countries still had to join the EU and the discussion after the actual enlargement quieted down is surprising because the actual enlargement of the EU could have a serious effect on the topic in my opinion. The main difference between the time periods, i.e. before and after the

available whereas the second does not have this data problem. Hence it is important to verify these earlier results, which were in those days still based on expectations.

So why could renewed research give a different outcome when we talk about job relocation and why would the high wage countries in WE start to suffer from cheap labour elsewhere? To answer these questions it is first useful to know more about an important phenomenon about job relocation: globalisation. Civilians but also politicians and labour unions fear job losses in the home countries as globalisation increases and opens borders. As can be seen in table 1 there are arguments in favour of globalisation but also arguments against globalisation. An important contra argument is job destruction, so when we focus on the effects of

globalisation on the labour market we can state that increased globalisation feeds the peoples fear of loosing their jobs as companies relocate their company to lower wage countries (J. Konings, 2001). So it makes sense that, since “The fall of the Berlin Wall symbolized the end of a global ideological division and gave a boost to the latest burst of globalisation itself”(N. Chanda, 2002), the fear of job losses to CEE started to play an important role in WE.

Table 1: pro- and-contra arguments with respect to globalisation

To test if the enlargement does have affect on the labour market within the new EU it is first useful to know whether differences in labour costs have changed and so have become

distinctive enough to be a serious incentive to relocate jobs. In other words do CEE countries really have lower labour costs compared to WE since the enlargement and does this possible difference make it attractive to relocate jobs to these CEE countries? Since countries in CEE are known as low waged and countries in WE as high waged the differences in wage costs are also expected to be large. Hence are according to many in WE the countries in CEE a threat to home labour market. Because there is hardly any empirical evidence that compares labour costs across the CEE countries with those of the EU and little is known about firm level labour productivity in the CEE countries (Koning, 2003) it would be useful to determine this attractiveness of relocating jobs to the CEE. To measure labour costs competitiveness, wages are not the only factor of importance. One should also be aware of the effectiveness of the work by those who are employed. High labour productivity is not guaranteed and is related to many different kind of factors. To sum up these factors is impossible because when you think of any worker you will agree that almost everything can influence this worker his

productivity, it can vary from the level of education and training-policies until for example the culture in a certain country. Hence it is in determining the competitiveness of a country or region important to focus on wages (or more broadly, labour costs) on the one hand, but also for labour productivity on the other hand. To measure this labour competitiveness the

measurement unit labour costs will be used. Unit labour costs is a frequently used

measurement linking productivity and the costs of labour in producing output. After assessing whether the competitiveness from the CEE should be seen as a real threat to WE employment the main question should of course be answered whether WE countries are actual triggered to relocate jobs to the CEE. This is however a much more complex subject matter. International trade effects on labour market outcomes can be studied in different ways. Most papers about job relocation have used sector level data to address the issue and hence do not take into account an important difference. This is that firms can, at firm level, be very heterogeneous as within the same sector both domestic as foreign firms are active (J. Konings et al., 2001), domestic firms include non-multinational enterprises as well as multinational enterprises (MNEs) which in it’s turn can rival foreign-owned firms in terms of productivity levels. Thus, the superior productivity performance of foreign firms may not be a foreign ownership

ideal unit of analyse. It is expected that MNEs have the availability to relocate activities to countries with lower wages in an easier way as they are considered to be ‘footloose’ (Caves 1996). This is confirmed in a paper by Murphy who states that one of the most obvious ways in which employment in WE might be affected by this enlargement is through the

employment re-allocation decisions of MNEs. By operating a portfolio of subsidiaries in diverse national markets MNEs can, in a more easy way than other firms, re-allocate their factors of production across these markets to minimise their production costs (Murphy, 2006) as they are already experienced with the foreign markets. Because of this the MNE has an advantage on other firms when it comes to relocating activities to countries with lower wages hence making the home countries more vulnerable to labour costs in foreign countries (H. Braconier and K. Ekholm, 2000). Another argument to a close empirical examination of MNEs is the fact that more than half of trade is in intermediate goods and the majority of international trade is controlled by MNEs (Riker et al, 1997).

from before 2004. And in my opinion it precisely, as explained later, this EU enlargement that actually could effect business between countries from the CEE and West Europe. By using a comparable research method by Jozef Konings but taking more recent data I try to find evidence whether the enlargement of the EU in Mai 2004 (low wage countries like Poland, Hungary and the Czech Republic joined the EU) triggered more substitution between the workers in the two regions.

I come to the following research question:

Has it become more attractive to relocate jobs from Western European companies to their subsidiaries in Eastern en Central Europe since the enlargement of the EU in 2004 is a fact, and does the enlargement actual triggers this job relocation?

This paper is structured as follows:

The paper starts by giving a brief theoretical overview to clarify the motive why it can be attractive to relocate jobs to low wage countries. From different point of views the subject of job relocation will be discussed. The arguments in favour of a renewed research on possible job losses in WE are discussed in an intensive way, as it is the motive for this research. Finally there is a literature review on competitive labour costs advantages followed by a review of job relocation, which examines the important research studies.

The empirical part is divided in two sections. The first section gives an impression of the attractiveness to relocate labour by measuring the labour competitiveness between WE

countries and the new member states entered the EU in 2004. With the labour competitiveness of a company is meant a comparable indicator of a company’s labour costs. The level of attractiveness is important to determine whether the so-called ‘low wages’ countries from the CEE are a real threat to employment in WE. Despite academic research (see literature review in this paper) no evidence has been found that the labour market in WE is influenced

countries in WE and countries in CEE. So low wages are according his paper clearly not an important argument to relocate jobs. An important minus in referring to his work in recent developments is that his research is based on data from before the enlargement. Hence it would be useful to determine this threat once more and test whether critics should be taken seriously. So if the ratio is still the same over the countries, i.e. the table does not give clear distinctions between WE countries and CEE countries, it would indicate that the critics still use a false argument when they argue that WE loses jobs to CEE as a consequence of differences in labour costs.

By combining the actual average costs of hiring an employee and the average productivity of this employee a ratio of labour competitiveness can be given (Konings, 2005). This ratio can give us an answer whether the criticism towards the low wages countries do have a point when they argue that the countries in the CEE have substantial lower labour costs compared to WE and hence differences labour costs competitiveness could be a true argument of job losses in WE labour market. Groups in WE are afraid many lower skilled jobs will move to the Eastern Europe because of the huge gap in income. Because I use Amadeus as a database I have the advantage that companies are comparable across countries, I namely use the same inclusion criteria across the different countries and Amadeus uses the same type of reporting for all the companies. A large data set of about 10000 manufacturing firms in both WE countries as in CEE countries will be used to define possible differences in labour costs between the both regions. The reasons to choose for only the manufacturing industry are numerous because first this paper is a follow up of an research by Konings who also uses manufacturing industry, as I want to compare results of both papers it is essential that data are identical. This brings us to the second argument where P. Havlik in one of his researches (Havlik, 2005) addresses that branches of industries are not identical when we talk about labour competitiveness. The third and last argument is about the comparability and exchangeability of industrial goods, firms active in the manufacturing industry produce products that are most internationally tradable (B. van Ark et al., 2005) and hence do border issues play a less important role.

is negligible for the period 1995 till 2000, my task is to control for the period 2000 till 2006. As said before this period is interesting, as it includes the enlargement of the EU. It is further interesting how the WE labour market actually responds to possible changes of in the

competitiveness of the labour market in CEE. Because job relocation is difficult to measure (Konings, 2005) a more precise and independent check is needed.

Independently from results in the first section I focus on the actual employment substitution. Independently because regardless of the outcome of the threat level in section one, I can state that it does not undermine the statistical test needed to determine the actually replacement of jobs. The second section will statistically test whether the enlargement of the EU in 2004 affected the number of jobs in WE firms as a response to a more open CEE labour market. Besides measuring actual job substitution I want to analyse how the amount of jobs in parent firms and subsidiary enterprises is associated with changes in subsidiary labour costs. Also this is in some way a follow up of the research by J. Konings in 2005 as it uses a similar method. The reason to choose for the research methods developed by Konings and colleagues is because he uses multinational enterprises as data source and does this on firm level. To get an idea of the effect of low labour costs on job relocation MNEs are, as said before, ideal. Because MNEs are argued to be footloose (Caves, 1996) they are an obvious channel through which employment substitution can take place in an easy way. They are already active in different countries and are able to relocate their factors of production to low wage countries without being confronted with large set up costs (Konings and Murphy, 2006). Hence it is expected that when the incentives to relocate employees to other countries is strong enough MNEs will play a leading role. To test whether the enlargement of the European Union has contributed to job loss in West Europe a distinction is made between subsidiaries either in EU or in CEE or both. A new dataset is needed, this time only from MNEs. Again I use Amadeus since I have the advantage of comparability across firms.

relationship is negative this indicates the employment in the parent and subsidiary firms are substitutes for one another.

2. Theory

This part will first try to shed light on the motive why companies can be triggered to relocate jobs to low wage countries. Secondly the economic implications of the EU enlargement on the European labour market are explained in an intensive way in particular, as it is the motive for this research. Finally there are two literature reviews: one about the topic competitive labour costs advantages and one about the important research studies on job relocation.

2.1 Trigger to relocate jobs

To get an answer on why companies relocate jobs to cheap labour countries can be explained from macro economic view but also from micro economic point of view. When we take a macro economic view we can see that since the fall of the Berlin Wall globalisation is a phenomenon that is unleashed and has become an important effect on the labour market.

Globalization goes together with a remarkable expansion of free trade (R.M. Dunn et al, 2004). Free trade in its turn is an important condition for economic development and it was already in 1776 when Adam Smith developed the basic precept of free trade theory. It suggested that it is more efficient for each country to produce the goods it is more able to produce, due to supply conditions of human resources, natural and physical capital, in comparison to its trade partners. This occurs due to the derived gains from specialization of production. In 1817 the principle of comparative advantage established by David Ricardo however suggests that a country should concentrate on producing goods that have the smallest relative cost of production, and not necessarily on the smallest absolute cost of production. In Ricardo’s formulation, labour is the only production factor. What is unclear in that theory is the effects of free trade on the distribution income, since the theory is based on only one factor of production. According to the traditional trade theory of Heckscher-Ohlin,

From micro economic point of view there is also an explanation why labour is influenced by other countries. We see that the demand for labour comes from the employer and the labour market fore fills the demand. There is normally an inverse relationship between the demand for labour and the wage rate that a business needs to pay for each additional worker

employed. If the wage rate is high, it is more costly to hire extra employees. When wages are lower, labour becomes relatively cheaper than for example using capital equipment and it becomes more profitable for the business to take on more employees. Businesses seek to maximise profits and will therefore search in the long run for the mix of factors of production (labour and capital) that produces the required level of output as efficiently as possible for the lowest possible total cost. By opening up trade a country with low labour costs could increase output and labour demand. Hence the labour demand decisions of companies including MNEs should not only depend on domestic, but also on foreign labour costs (J. Hatzius, 1998). From strategic perspective companies should cross borders as in a globalising world companies need to have competitive advantages. Competitive advantages are capabilities that are difficult to replicate or imitate and are non-tradable (Porter). There are generally spoken two types of capabilities leading to a competitive advantage: Differentiation (A higher level of customer value by performance, quality and brand services) & Cost leadership (A lower cost base) (P. Lasserne, 2003). When choosing the strategy to compete on costs, companies will start to search for methods to save on their costs. An important economic variable considered in the context of globalisation, competitiveness and the production location decisions is the development and the level of labour costs (Brück und Engerer, 2004). Especially for labour intensive production the main factor of the costs is labour. Hence firms concerned with this labour intensive production are clearly dependent on the resource labour. To create a competitive advantage like low labour costs it is in nowadays global business strategy important to have easy access to this globally scarce resource. It sounds logical that when firms heavily depend on labour and once do have this easy access, they relocate their operations to the countries who provide this competitive advantage.

this phenomenon affects labour market in West Europe negatively. Contrary to the policy concerns and popular press literature however pointed out that heavy job losses were not expected.

2.2 Why the enlargement of the EU could affect the WE labour market.

So in which extent could the enlargement effect economic conditions, which in their turn leads in changes in the labour market? I start by stating that the accession of CEE countries clearly adds to the trade liberalisation of the past. Despite the gradual nature of economic integration in Europe, actual joining the EU is likely to have important implications for the economy of the new member state. Advantages of EU membership are that firms located in the EU have an easy access to a large pool of member states and membership also increases the stability and safety of investments. The enlargement of the EU should at least theoretically widen the scope for trade and the free movement of goods and people. International investors are in general negative towards different legal en regulatory environments, discriminatory taxes, expropriation, incomplete information and other risks as foreign currencies (R Dunn et al., 2004) but as the European Union has institutions to regulate them it sounds plausible investment will increase. For example the tariff and nontariff barriers to trade in industrial goods between the accession countries on the one side and the EU on the other are largely removed by the European agreement (there has been a free trade area for industrial goods since 2002), and the EU enlargement will lower the transaction costs of this trade further (Brück et al., 2004). Besides do the accession countries have access to EU Structural Funds intended for basic infrastructure development, human resources development,

competitiveness and enterprise development, rural development, and environmental protection (Kalotay 2006).

When we look at figures we find evidence that before the enlargement most economists expected only limited gains from the actual accession to the EU because CEE countries had gained so much already from integrating with the EU. A research by the Dutch CPB however predicted that the economic benefits for the accession counties tend to be significant.

Especially industrial relocation to these countries is expected as companies reap the gains from trade and exploit comparative advantages of countries (Lejour et al., 2001). This

countries recorded investment growth rates of 5 per cent of more in 2004 and 2006. This investment was supported by a rise in FDI inflows. According to a study from The Executive Committee (TEC) in 2005 there is evidence that especially the smaller companies in WE preferred the CEE as a location for outsourcing. Most of the large companies already invested in the CEE countries before the enlargement. (K. Barysch, 2006). Also UNCTAD mentions about a post development in a similar way. Despite not long after the breakdown of the Berlin Wall Western investors were enthusiastic to invest in the CEE, according to business surveys two thirds of multinational corporations had plans to directly invest in Eastern Europe (J. Donges and J. Wieners, 1994) research by UNCTAD shows that the CEE countries ready to join the EU had not been receiving massive foreign direct investment (FDI) from the older members of the Union as predicted. As can be seen in table 2, the combined inflows during the late 1990s and early 2000s of the 10 new members remained considerably lower than those of such current EU States as France and Germany and, more recently, Ireland and Spain (press release UNCTAD, 2004)

Table 2: FDI inflows of the countries acceding to the EU in 2004 international comparison, 1995-2003

Source: UNCTAD FDI/TNC database

"second wind" of FDI. Several new EU members have drastically lowered their corporate taxes. The report states that “the combination of low corporate tax rates, relatively low wages and access to EU subsidies - enhanced by a favourable investment climate, a highly skilled workforce and free access to the rest of the EU market - makes the accession countries attractive locations for FDI, both from other EU countries and from third countries.” This combination can, in contrary to the disappointing inflows of FDI from shortly after the fall of the Wall, result in a strong increase in FDI. As a resource-seeking motive MNEs can use FDI in search for more efficient factors of production, in this case cheap labour. The relationship between MNEs’ subsidiary jobs and the parent jobs is however not unambiguously a

substitutionary one (H. Broconier and K. Ekholm, 1999). MNEs have different motives to invest abroad. Broadly the literature on Foreign Direct Investment (FDI) speaks about two different types of FDI: Horizontal and vertical FDI. Whether to expect that the expansion of activities in one location substitutes for or is complementary to activities in other locations crucially depends on whether FDI is of the horizontal or vertical type.With horizontal FDI, firms invest abroad in order to avoid the costs that are associated with exporting from the home plant to export markets. Consequently, the MNE compares the costs associated with exporting with the additional costs associated with setting up a new plant abroad.Horizontal FDI therefore has a negative impact on employment at home as home production for exports is being replaced by local production in the host country.According to vertical FDI, firms invest abroad in order to reduce overall production costs, benefiting from factor price

2.3 Labour competitiveness

Before determining whether job relocation really took place after the enlargement it is

important interesting to know what has happened to the competitiveness between WE en CEE with respect to labour costs. An increase in labour costs for WE in comparison with CEE should according the critics result in job losses in WE. When we talk about labour costs in WE, we first can see that there are significant differences in wages between the principal states. The countries with the highest wages are Denmark, Norway, Switzerland,

Liechtenstein, Luxembourg and Germany. At the bottom of the WE wage league are Spain, Greece, Malta and Portugal (FedEE, Pay in Europe 2006). Pay in the EUs’ eastern European member states is generally much lower than in WE, as can been observed in figure 1.

Figure 1: Income per head CEE

But even the labour compensation is not the only factor of importance for labour

between countries. Another paper “Unit labour costs in the new EU member states” by P. Havlik argues that Unit Labour Costs differs between the various branches. For example in the leather products in Hungary and Slovenia or textiles & clothing and wood products in Slovenia there is obviously no comparative Unit Labour Costs advantage any more. However in the electrical and optical, as well as transport equipment industries Unit Labour Costs are very low – especially in the Slovak Republic and in Hungary (Havlik, 2005). Hence it is, when measuring a firm or countries labour competitiveness, important to compare identical branches and not to make the mistake to generalize.

Because of research results by Konings (2005) it is expected that countries with low labour costs will have a lower productivity and visa versa the countries with a high labour costs will be compensated by reaching a higher level of productivity. He used a competitiveness index to make clear that the advantage of low wages is strongly reduced by a low level of

productivity so that no incentive to relocate jobs exists. Konings argues that this relation is that strong that no serious difference in labour costs between WE countries and CEE countries exists. The question however is whether this relation has weakened over time so that

competitive advantages have shifted.

2.4 Literature review about MNEs’ job relocation

When we look at job relocation at the time of the fall of the Berlin Wall the policymakers in WE where concerned of their home labour market as in the countries in the CEE there was a large pool of low paid workers, situated just in the backyard of the WE countries. One of the reasons for these concerns was the possibility that WE could close their doors and replace their activities to CEE. Another reason was the fear of a massive immigration from CEE towards WE. The last but however the most serious threat was that competition from the CEE as a result of low wages would results in the replacement of jobs to the CEE (Konings, 2001). Substantial research has been done on whether globalisation, in the just mentioned case the fall of the Berlin Wall, actually leads to job relocation. The effects of globalisation on the labour markets can be distinguished in three broad sets of studies.

1. The effect of globalisation on wages and employment. A good example of this set of studies is a paper from Feenstra and Hanson (1996). They study the impact of

that outsourcing of certain production processes contributed to an increase in the demand for skilled labour in the U.S..

2. The effect of international trade and technological progress on wage inequality. 3. The effect of globalisation on the relocation of the MNEs’ home employment to

subsidiaries in other countries.

The third set, the research on the effect on MNEs’ job substitution between home and it’s subsidiaries in other countries, has received little attention (J. Konings and A. Murphy 2001). Most of this papers used sector level data to indicate a possible job substitution within the MNEs. Slaughter (2000) has used this sector level data to address the questions of job relocation and comes to the conclusion that MNE transfer tends to have small, imprecisely estimated effects on US relative labour demand. This means that, contrary to Feenstra and Hanson (1996), MNE transfer has not contributed to US skill upgrading.

The first to use firm level data were Brainard and Riker. In two working papers (1997) they examined firm level data of foreign manufacturing subsidiaries owned by U.S. multinationals and analysed the substitution between subsidiaries located in countries with low wages with the labour demand of subsidiaries located in industrialized, high wage, countries. In both papers they conclude that, instead of a negative effect on the labour market of industrialized countries, there is a complementary relationship between the labour market in low wage countries and the labour market in high wage countries. A substitution effect however does seems to exists between subsidiaries with a similar factor endowment, for example both are low wage countries. For example when a developing country lowers its wage it has a negative effect on the labour demand in another developing country.

Hatzius (1998) used in his study a panel of Swedish multinational corporations and concludes that when labour costs in foreign countries increase this has a positive effect on labour

country negatively what is comparable to the results of Brainard and Riker (1997). The main result of the paper of Braconier and Ekholm is that there is a relationship of complementary. In both subsidiaries, high and low waged, labour demand increases when a MNE invest in a low labour location. With a focus on the MNE job relocation as a result of fall of the Berlin Wall, Konings can be considered as the expert on this specific terrain. Konings wrote, sometimes in co-operation with others, multiple papers about the topic (Konings 2005), (Konings 2001), (Konings and Murphy 2001), (Konings and Murphy 2006), (Abraham and Konings 1999). All these papers by Koning have in common that the results show that no significant job relocation takes place between parents and their daughters in the CEE. Labour costs in the WE subsidiaries do however seem to effect its parents amount of jobs, a decrease in labour costs from the daughter attracts jobs from the parent. The most recent year of the data used in these researches is 2002. Interesting to know is whether from 2002 results have changed as a result of the EU enlargement.

3. Statistics and empirics

3.1 The labour competitiveness between firms located in WE vs. firms located in CEE. To clarify whether competitive advantages in labour costs between WE and CEE has changed since the enlargement a same competitiveness index as Konings (see figure 2) will be used however covering a different time period. The type of measurement used in the

Figure 2:

First of all the costs of wages of will be determined. The costs of wages per employed person will be measured on firm level. This will be done by dividing the total wage costs by the amount of personnel. The average costs of wages in a country will be determined by summing up all the average costs of wages of the selected companies and divide the outcome by the amount of firms. Secondly is an important factor the labour productivity per worker. As said before not only the costs of wages are important when searching for a country to relocate production to but also the productivity of the workers is crucial when talking about

competitiveness. The productivity of labour will be measured as output per employed person per year. The average of the country is determined in the same way as wages costs. Both outcomes can be used to show the ratio of wage costs per worker to labour productivity in the selected countries. The wage costs and productivity together give information about the competitiveness of a company. Differences in competitiveness between firms can be

determined in the years 2000 till 2006. Because results are given over the most actual time period directly after the years Konings selected, I will be able to test whether the

competitiveness of labour in the selected countries has changed over the years by comparing my results with the results of Konings.

3.2 The actual job relocation between WE and CEE.

In my search for job losses to subsidiaries in the EEC I will first try to determine the relationship between parent employment and subsidiary employment. First a table a gives information about the evolution of the total subsidiary employment as a fraction of the total MNE employment, i.e., the sum of total subsidiaries and parent employment shown in %. In the table we can analyse whether the employment share of the parent MNE has decreased over the years with regard to subsidiary employment and can give an indication of possible job relocation.

To statistically test the job relocation I compute the correlations between employment growth in parent firms and the employment growth in the subsidiaries. I therefore regress parent employment growth on employment growth in the subsidiaries. A distinction is made between subsidiaries located in WE and subsidiaries located in CEE, but no distinction is made

between diverse subsidiaries located in the same side. Because many MNEs have more than one subsidiary in for example the CEE I take the average of all West European subsidiaries and all CEE subsidiaries and consider in the continuing of the research West European subsidiaries and CEE as two groups.

In the previous part I focused on the relationships between parent and subsidiary employment. Also I want to check whether changes in the subsidiaries wage costs affected the parents’ labour force. The theory behind this is derived from the theoretical framework developed by Konings.This theoretical framework is based on the model of The Heckscher-Ohlin using labour and capital as the two factors of the production process. Depending on of the relative price of labour and of capital an optimum quantity of the input factors is used in the

the parent should lead to an increase in jobs at the parent. Contrary should a decrease in wage costs in the subsidiary result in a lower amount of jobs at the parent. This is however only the case when labour is substitutable between the parent and the subsidiaries. If a fall in wage costs of the subsidiary would result in a rise in jobs at the parent firm we would have a complementary effect between parent and subsidiary. To describe how MNE labour

growth/decline is numerically related to labour costs of the subsidiaries. This time I compute the correlations between employment growth in parent firms and the wage costs in the subsidiaries. I therefore regress parent employment growth on employment growth of wage costs of the subsidiaries. A negative relationship means that negative scores on one variable (parent) tend to be paired with positive scores on the other (subsidiary). This indicates that the MNEs’ parent is not loosing jobs when wage costs are declining in the subsidiary, in contrary the amount of jobs grows. A positive coefficient can be seen as more logical, if the wage costs of the subsidiary decline also the number of jobs of the parent decline or if the wage costs rise also the jobs of parent rise. A limitation is that the changes in wage costs of the MNE in not taken into account, hence I cannot control for the effects of these changes. What I did do to have a better comparison of the wage costs over the years was correcting the wage costs of the subsidiaries by 2.1 % inflation per year.

Fixed effects regression is the main technique used for analysis of panel data. Fixed effects regression lets me use the changes in the variables over time to estimate the effects of the independent variables on my dependent variable. It examines group differences in intercepts, assuming the same slopes and constant variance across groups. Fixed effect models use least square dummy variable (LSDV), within effect, and between effect estimation methods. Thus, ordinary least squares (OLS) regressions with dummies, in fact, are fixed effect models. To control for not observed heterogeneity like quality of management or the level of technology but also to included simultaneously the variables of the subsidiaries in WE and CEE

(Konings, 2001) I use in accordance with Konings this fixed effects technique. I selected the fixed effects method for both regression analyses.

3.3 Data

• Country: The country the firm is located in

• Total employment:Total amount of workers

• Wage costs: Total wage bill of the parent firm

• Total output: the value added

• Year

In the eventual data set the years 2000, 2002, 2004 and 2006 will be taken as reference of the renewed time period. The data should represent the annual company accounts. Out of these annual reports I can get the amount of employees within a firm but also the total salary costs of these employees. By dividing the total wage bill by the amount of employees I have an annually measure of labour costs per employee. The total wage bill will not only include gross salary but also the social security contributions and employer contribution. Labour

productivity can be measured in more ways. Examples are the value adding or sales given in profit and loss account. In my paper labour productivity will be measured in the same way Konings did by taking the profit and loss account and select the value adding (=output) of that specific year (Konings, 2005). I make use of a large data set of about 10000 manufacturing firms located in both WE as CEE. To measure the labour costs and productivity at firm level the database Amadeus will be used. Amadeus is a commercial database collected by “Bureau Van Dijck” which is a quoted software and consulting company on the (Euronext) stock market. To derive to a maximally representative database I selected all firms that where available after making the following selection: active in manufacturing industry, availability of data of at least 2 years where the number of employees, the added value and the costs of employees is given.

subsidiaries and the average values over all CEE subsidiaries. To ensure that the activities between the parent firm and its subsidiaries are comparable I have eliminated subsidiaries that operate in a substantially different environment than the parent firm. The data I retrieve are the total amount of workers within the selected European MNEs but also the amount of workers divided over the parent firm and it’s subsidiaries. In this way the evolution of

employment share of parent firm and its subsidiaries can be measured. Labour costs are again determined by combining wage costs and productivity. As a sample I retrieve firm level data from 350 European MNEs with their parent firm (located in West matched with their

subsidiaries located in the both (West and CEE) regions. This paper only covers MNEs with subsidiaries in the EU, in the CEE or in both. So some MNEs have subsidiaries located only in WE, some only in the CEE an others in both. A parent firm is defined as a holder of at least 50% ownership share in one of more firms located in an other country within the EU. Hence indirect holdings are not included. To prevent missing data only the MNEs will be selected where parent firm report its subsidiaries. In some countries companies are not obliged to report the unconsolidated accounts for both parent firm as its subsidiaries. For example

Belgian firms are required by law to report their subsidiaries, while for other countries like the Netherlands companies can voluntarily choose to report or not (Konings and Murpy, 2006).

The eventual data set will cover the period 2000 till 2005. The year 2006 is excluded, as not enough data is available. The data will represent the annual company accounts. The selected data variables are:

• Total output of the MNE: the weighted sum of value added of the parent firm and its subsidiaries. The weight represent the importance of value added in total value added for the parent versus the subsidiaries.

• Parent employment: Total amount of workers in parent firm

• Daughter employment (west): Total amount of workers in subsidiary (west)

• Daughter employment (CEE): Total amount of workers in subsidiary (CEE)

• Parent wage cost: Total wage bill of the parent firm

• Daughter wage cost (west): Total wage bill of the subsidiary (west)

• Daughter wage cost (CEE): Total wage bill of the subsidiary (CEE)

• Identifier: in Amadeus every firm (incl. subsidiary) contains an ID number

This data is panel data, also called longitudinal data or cross-sectional time series data. With panel data is meant multiple cases (people, firms, countries etc) being observed at two or more time periods (Princeton University, 2008). Panel data sets generally include sequential blocks or cross-sections of data, within each of which resides a time series. My panel data set will include multiple observations of different firms followed for five years and sampled annually. The reason I choose for this type of dataset is because in economics panel data is ideally to study the behaviour of firms and wages of people over time (Yaffee, 2003). Analysing panel data is a method of studying a particular subject where multiple cases are periodically observed over more time periods. With multiple cases you can for example think of people, firms or countries. Time series means periodic observations of a set of variables characterizing these multiple cases over a particular time span. With these repeated

observations I will be able to study the dynamics of change with short time series. Together this panel data makes it possible to do a regression analysis with both a multiple cases and time series. The panel data set contains observations on n individuals (in this case an MNE), each measured at T points in time. In other word, each individual (1 through n subject) includes T observations (1 through t time period). Thus, the total number of observations is

nT. Table 3 illustrates the data arrangement of one of the panel data sets.

Table 3: example panel data set

Firm ID year

Amount jobs Parent

Labour costs /wage costs pp sub WE

Labour costs /wage costs pp sub CEE

1 2000 1000 100 100 1 2001 900 110 90 1 2002 800 120 80 2 2000 2000 100 100 2 2001 1900 110 90 2 2002 2000 120 80 3 2000 2200 100 100 3 2001 2100 110 90

Because my data are structured in a different format like the wide form, I needed to rearrange data first. STATA has the .reshape command to rearrange a data set back and forth between

4. Results

Firm level data of manufacturing firms in Europe are used to analyse whether wage cost differentials between high and low wage location can trigger employment relocation. I describe the results in two sections. To asses the issue of attractiveness I first present the competitiveness index. Second correlation results are reported.

4.1 Results attractiveness

In this section I report the findings of a comparison of wage costs and labour productivity at the firm level in Central Europe with those in Western Europe. According to the paper of Konings labour costs are about 5 times lower in the typical firm in CEE compared to labour costs in the WE countries. But also labour productivity is more than 5 times lower in Central Europe. Because of comparable productivity level Konings suggests that just low wages are not a trigger to cause a massive relocation. The question in this paper is whether a

phenomenon as the EU enlargement could shed a different light on Konings conclusion that low wage countries in CEE are not more competitive than the high wage countries in the West. I reproduce the competiteveness index as can be found in the paper of Konings (Konigs, 2005) but use a time period that succeeds his dataset.

Figure 3 0 20 40 60 80 100 120 BELG IUM BULG AR IA CZE CH RE PUB LIC DEN MA RK ES TON IA FRA NC E GE RM AN Y HU NG ARY NETH ER LAN DS PO LAN D PO RTU GA L Average labour productivity per worker 2000 until 2006 (1000 Euro)

differences do still exists but what is more interesting is that the gap has apparently not declined compared to the earlier results of Konings. This is somewhat surprising as the press report that wages, real estate and taxes are rising fast in CEE, “the newest corner of the European Union just isn't the bargain it used to be” (Kole, 2007). Portugal’s wage costs are somewhere between low and high. This can be explained by the fact that Portugal is located in Southern Europe, a part where average pay is considered to be lower than WE.

Figure 4

Labour costs per worker 1000 Euro

0 10 20 30 40 50 60 Bel gium Bul garia Cze ch R upub lic Den mar k Eston ia Fran ce Ger man y Hun gary Net herla nds Pol and Por tuga l 2000-2006 1995-2000

Figure 5 0 10 20 30 40 50 60 BELG IUM BU LGA RIA CZE CH RE PU BLI C DE NM ARK ES TON IA FRA NC E GE RM AN Y HU NG ARY NE THE RLA ND S PO LAN D PO RTU GA L Average labour cost per worker 2000 until 2006 (1000 Euro)

Figure 6

Labour productivity 1000 Euro

0 20 40 60 80 100 120 Bel gium Bul garia Cze ch R upub lic Den mar k Est onia Fran ce Ger man y Hun gary Net herla nds Pol and Por tuga l 2000-2006 1995-2000

When we compare results of Konings with mine we observe again for almost all the countries an increase of productivity. Probably the increase can also here be partly explained by

Figure 7 0,00 0,10 0,20 0,30 0,40 0,50 0,60 0,70 0,80 BELG IUM BU LGA RIA CZE CH RE PU BLI C DE NM AR K ESTO NIA FR ANC E GE RM ANY HU NG AR Y NETH ERLA ND S PO LAN D PO RTU GA L Index average labour cost relative to average added value 2000 until 2006

Figure 7 combines the wage costs together with the productivity as the ratio of labour costs per worker to labour productivity. This results in the so-called competitive index. The lower the competitive index the more competitive a country is in terms of unit labour costs. As we can note competitiveness is almost the same across countries, i.e. differences are negligible compared to the differences in figures 3 till 6. In some cases WE countries even have a better competitiveness than CEE countries. By denoting the propotional amount of labour costs per worker relative to productivity per worker we can observe that the majority of the differences existing in the first two figures are minimized. The relatively small differences between the countries indicate that the labour competitiveness across the countries is almost the same. Because no distinction has been made between the different branches in the manufacturing industry these graphs just show how the average firm in a manufacturing industry looks alike. Hence there will be for example CEE firms that have a better competitiveness index than WE firms however there will be also CEE firms for which the reverse is true.

Figure 8

Index labour cost relative to added value

0,00 0,10 0,20 0,30 0,40 0,50 0,60 0,70 0,80 Bel gium Bul garia Cze ch R upub lic Den mar k Est onia Fran ce Ger man y Hun gary Net herla nds Pol and Por tuga l 2000-2006 1995-2000

Figure 8 compares the competitiveness index of Konings based on data from 1995 till 2000 with my dataset based on 2000 till 2006. We see that competitiveness has somewhat worsened for most WE countries. When we exclude Portugal, as we could do because it is not a clear high wage country, only Belgium improved its competitiveness. But also two out of four CEE are confronted with a worse competitiveness. We can hence conclude that competitiveness in WE or CEE as a whole has not significantly changed.

Figure 9 0,40 0,45 0,50 0,55 0,60 0,65 0,70 0,75 0,80 1996 2000 2002 2004 2006

BELGIUM BULGARIA CZECH REPUBLIC

DENMARK ESTONIA FRANCE

GERMANY HUNGARY NETHERLANDS

POLAND PORTUGAL

When we take a look at the time period 2000 until 2006 and analyse the trend lines of the various countries we see for Poland and Bulgaria from 2001 till 2004 a sharp fall. All the other countries stay relatively stable. As WE countries and CEE countries are compared individually it is however hard to draw any conclusions on the competitiveness of WE or CEE as a whole. Hence I will sum up the results of the individual countries in the following figure.

Figure 10 1996 ₂₀₀₀ 2002 ₂₀₀₄ 2006 CEE WE 0,00 0,20 0,40 0,60 0,80

Index labour costs relative to added value CEE vs WE

CEE WE

Index labour costs relative to added value CEE vs WE (1000 Euro) 0,00 0,20 0,40 0,60 0,80 1996 2000 2002 2004 2006 CEE WE

2002. In 2004 the growing gap between the labour competitiveness of WE and of CEE comes to a standstill, in the following years the advantage for CEE remains stable so it does not decline anymore. This structural advantage could be a trigger to relocate jobs from WE to CEE. In the table the average of the firms in countries located in WE and in CEE are being compared. I have also compared WE and CEE in a different way: here the average of the outcomes of the countries in WE and CEE are being compared instead of taking firms as individuals. The competitiveness index and suggests that, while the results of my research need perhaps more detailed study, the paper does justify complaints from critics warning about the CEE labour market harming the WE one.

4.2 Results actual job relocation

This section reports the results of employment substitution within MNEs. I analyse the relation between job growth in the parent and the subsidiaries but also whether employment growth in the parent is related to wage changes in the subsidiaries. The panel data of 350 European MNEs with their parent firm and subsidiaries is used as data set.

Table 4: Evolution of employment in WE MNEs 2000-2005 (percent)

(1) Parent firm (2) Subsidiaries (3) West European subsidiaries (4) CEE subsidiaries 2000 54 % 46 % 21 % 25 % 2001 57 % 43 % 20% 23 % 2002 58 % 42 % 19 % 23 % 2003 58 % 42 % 19 % 23 % 2004 56 % 44 % 19 % 25 % 2005 54 % 46 % 19 % 27 %

look at the subsidiaries in columns 3 and 4 we see that especially the subsidiaries from the CEE are accountable for these fluctuations, while the employment shares of the subsidiaries in WE are relatively stable. Hence we have a first indication that from 2004 WE has been loosing jobs to CEE. As said before it is an indication but should be interpreted with care. It is for example a good possibility that the amount of jobs in the parent, the WE subsidiary as well as the CEE subsidiary all increase but because it increases the most in CEE the percentage of the CEE rises while the MNE and WE subsidiary fall. When we compare results with the first part, the so-called attractiveness, we can observe that an improved competitiveness in CEE leads to a percentual increase of CEE jobs.

The second method that I use to analyse the relation between job growth in the parent and the subsidiaries is to determine the correlation between employment in parent firms and

employment in the subsidiaries. Again the panel dataset is used as I regress the amount of parent employees on the amount of the employees in the subsidiaries. If there is a negative relationship between parent employment and subsidiary employment (divided in West Europe and CEE) this means that they can be seen as contrary inputs. A negative relationship means that negative scores on one variable (parent) tend to be paired with positive scores on the other (subsidiary) and visa versa. So an indication whether employment substitution may take place can be obtained by computing the correlations between the number of jobs in the parent firms and the number of jobs in their subsidiaries located in WE and subsidiaries located in CEE.

Table 5: Parent MNE to subsidiary Employment correlations

Parent WE subsidiary -0.051**

(0.025) CEE subsidiary 0,088**

(0,041)

Note: (i) These correlations are computed on the basis of a regression of parent employment growth on subsidiary employment growth, including year dummies. (ii) standard errors in brackets (iii) *** indicates 1% significance level, ** 5% significance and * 10% significance

sampling error attached to them, calculated from the variability of the observations in the sample. From this, a margin of error (confidence interval) is derived. It is this confidence interval that is used to make statements about the likely 'true' value in the population. A confidence interval of twice the standard error is used to state, thus only if the estimated correlation is greater than twice the he standard error can we claim to have found statistically significant evidence of a relationship between the two sets of employment figures. We can see that the correlation between parent employment growth and WE subsidiary employment growth is negative and statistically significant. This indicates that the parent and WE subsidiaries are potentially substitutes for one another. For subsidiaries located in CEE the correlation coefficient is positive and statistically significant. The positive sign suggests that as subsidiary employment increases parent employment also increases, in other words they can be seen as complementary. This result comes not as a surprise as Braconier and Ekholm (2000), Konings (2005) and Konings, J., & Murphy A. (2001) report similar results. All found a substitutionary relationship between employment in MNEs’ parent located in WE and its WE subsidiary but found no evidence of substitution stemming from employment in CEE. The only difference is the complementary effects for CEE subsidiaries, it seems that instead of relocation both parent as subsidiary increase their number of employees. In the mentioned papers no relation, substitutionary nor complementary, has been found. To come back on the earlier result where I observe in CEE a percentual increase of CEE jobs it now seems that this increase is a result of an increase of jobs for both WE as CEE. According to table 4 growth in CEE however seems to be higher.

Finally table 6 shows the parent employment growth and subsidiary wage costs correlations and gives us an first indication whether the Western parent responds to wage changes in the subsidiaries located in WE and CEE.

Table 6: Parent jobs to subsidiary Wage costs correlations

Parent WE subsidiary - 0.003*

(0.034) CEE subsidiary - 0.133*

(0.523)

subsidiary employment growth, including year dummies. (ii) standard errors in brackets (iii) *** indicates 1% significance level, ** 5% significance and * 10% significance

The correlation between parent employment and WE subsidiary wage costs is almost 0 and is not statistically significant. In other words no relation can be observed. For subsidiaries located in CEE the correlation coefficient is clearly negative however not statistically significant. Hence also here no relation can be claimed and suggests that parents do not respond to wage changes in both WE as CEE subsidiaries. This can be explained by lacking factor productivity, not only wages are a competitive factor also productivity is a competitive factor. The analysis is however meant as explorative, further econometric analyses are needed to test the relationship more thoroughly. On basis of these explorative results in combination with the earlier results we can conclude that despite CEE labour market is improving, read more competitive and a percentual higher share of employees, this is not due to wage changes.

Concerning limitations there are the following remarks. In this paper for both data sets, the one with all types of manufacturing firms and the other with only manufacturing MNEs, the various manufacturing sectors are pooled together. It is however likely that substantial differences between the sectors exist in terms of labour competitiveness (Konings, 2005). In some sectors are for example more logistic costs involved. Besides different sectors within the manufacturing industries also differences between the different industries are expected, thus should the nonmanufacturing industries be included. Further research with respect to these sector differences but also to different industries is desirable. Location can also be an

5. Summary and conclusions

This paper has examined some potential implications for Western European labour market at the time of the enlargement of the EU. In the first part of the analysis I checked to what extent low wages are a plausible argument to relocate industrial jobs from Western Europe towards Central and Eastern Europe. Compared to the research paper “Wage costs and industry (re)location in the enlarged European Union” by Konings we see that the wage costs differences between Western Europe and Central and Eastern Europe are years after the enlargement of the European Union still enormous. When I include the factor labour

productivity the competitive advantage of low wages melts away, because combined we see that they largely compensate each other. According to the final competitiveness index we can hence conclude that labour the low wages countries in Central and Eastern Europe are still not necessarily more competitive, i.e. are not necessarily more attractive. The fears related to the incorporation of a substantial pool of low-wage workers inside the European Union are however not completely to ignore as some evidence for the substitution of jobs between parent firms and their foreign subsidiaries has been found. Despite the compensating effect of higher productivity in Western Europe we can observe a trend that attractiveness to relocate jobs to Central and Eastern Europe improves. When both Western Europe and Central and Eastern Europe are compared as a whole, thus instead of comparing the countries separately, we see that since 2002 labour in Central and Eastern Europe has become little by little more competitive.

between the amount of jobs of the parent and to wage change of the subsidiaries no significant results were found. This suggests that jobs in the parent of multinationals do not respond to wage changes in its subsidiaries, i.e. wage cost considerations are not an important factor leading multinational firms to re-allocate jobs among its subsidiaries across Europe, both Western Europe as Central and Eastern Europe.

References

Abraham, F., and Konings J. (1999), Does the Opening of Central and Eastern Europe Kill Jobs in the West? World Economy 22 (4): 585–603.

Ark, van B., Stuivenwold E., Ypma G. (2005), “Unit Labour Costs, Productivity and International Competitiveness”, Research Memorandum GD-80

Braconier, H., and Ekholm K. (1999), Swedish Multinationals and Competition from High- and Low-Wage Locations, Review of International Economics 8 (3): 448–461.

Brainard, S.L., Riker, D.A. (1994), Are U.S. multianations exporting U.S. jobs?, National bureau of economic research

Brück, T., Uhlendorf, A., Wowereis, M. (2004), Lohnkosten im internationalen Vergleich, DIW Wochenbericht 14/2004, April, pp. 161 –169,

Brück, T, Brücker, H, Engerer, H, von Hirschhausen, C, Schrooten, M, Schumacher, D, Thiessen, U, Trabold, H, (2004) The Eastern Enlargement of the European Union: Clear Challenges, Unjustified Fears Economic Bulletin, Vol. 41 Issue 6, p189-198,

Caves, R. (1996), Multinational Enterprise and Economic Analysis, Cambridge, Mass.: University Press.

Donges J.B., Wieners J. (1994), Foreign Investment in the Transformation Process of Eastern Europe, The International Trade Journal, Volume VIII, No. 2

Gessel,van, Berkel, van, G. (2007), One in five large companies relocating activities, Web Magazine Statistics Netherlands

Gorg, H., and Strobl E. (2002), Footloose Multinationals?, Centre for Economic Policy Research, Discussion Paper no. 3402.

Havlik P. (2005), Unit labour costs in the new EU member states, The Vienna Institute for International Economic Studies, Austria

Konings, J. (2001), Loonkosten en Relocatie van Belgische Bedrijven, Leuvense Economische Standpunten, 101, pp.13.

Konings, J., and Murphy A. (2001), Do Multinational Enterprises Substitute Parent Jobs for Foreign Ones? Evidence from Firm Level Panel Data. CEPR Discussion Paper 3402. Centre for Economic Policy Research, London.

Konings J. (2005),Wage costs and industry (re)location in the enlarged European union, Swedish Economic Policy Review, vol. 12, no. 1, pp. 57 – 81

Konings J., Murphy, A.P. (2006), Do Multinational Enterprises Relocate Employment to Low-Wage Regions? Evidence from European Multinationals

Kolotay (2006), The Impact of EU Enlargement on FDI Flows, International Finance Review, Vol. 6, pp. 473-499, 2006

Murphy A.P. (2006), An Assessment of the Implications of EU Enlargement for Foreign Direct Investment and Jobs in Ireland, Quarterly Bulletin

Riker, D.A. , Brainard, S.L. (1994), Multinationals and competition from low wage countries, National bureau of economic research

Miscellaneous reverences

Lasserne, P, (2003) Global strategic Management

CPB Document No 011, September 2001., EU enlargement: Economic implications for countries and industries, A.M. Lejour, R.A. de Mooij and R. Nahuis

Enlargement two years on: Economic success or political failure? Katinka Barysch, chief economist, Centre for European Reform Briefing paper for the Confederation of Danish Industries and the Central Organization of Industrial Employees in Denmark April 2006

World Investment Report 2004, The Shift Towards Services Unctad

Cooper, D, Schindler, P, McGraw-Hill, Business research methods

Yaffee, R, 2003, A Primer for Panel Data Analysis, Information technology services New York University

Data and statistical services, 2008, Princeton University

Nayan Chanda YaleGlobal, 19 November 2002

Kole, W.J., 2007, As wages rise in Eastern Europe, businesses look elsewhere, The Associated Press, Wednesday, April 4, 2007, Herald Tribune

Appendix

Labour productivity (1000 Euro)

0 20 40 60 80 100 120 140 1996 2000 2002 2004 2006 BELGIUM BULGARIA CZECH REPUBLIC DENMARK ESTONIA FRANCE GERMANY HUNGARY NETHERLANDS POLAND PORTUGAL

Labour costs per worker 1000 Euro

1996 2000 2002 2004 2006 CEE WE 0 20 40 60 80 100

Labour productivity per worker CEE vs WE (1000 Euro)

CEE WE 1996 2000 2002 2004 2006 CEE WE 0 10 20 30 40 50

Labour cost per worker CEE vs WE (1000 Euro)

CEE WE

Output Stata: Parent MNE to subsidiary Employment correlations

Fixed-effects (within) regression Number of obs = 1139 Group variable: firmID Number of groups = 267 R-sq: within = 0.0024 Obs per group: min = 1 between = 0.0035 avg = 4.3 overall = 0.0165 max = 6 F(1,871) = 2.08 corr(u_i, Xb) = 0.0539 Prob > F = 0.1495 --- MNE | Coef. Std. Err. t P>|t| [95% Conf. Interval] ---+--- WESUB | .06997 .0485069 1.44 0.150 -.0252341 .1651741 _cons | 407.2888 8.212851 49.59 0.000 391.1695 423.408 ---+--- sigma_u | 592.92002 sigma_e | 121.62442

rho | .95962158 (fraction of variance due to u_i)

--- F test that all u_i=0: F(266, 871) = 87.37 Prob > F = 0.0000

Fixed-effects (within) regression Number of obs = 820 Group variable: firmID Number of groups = 209 R-sq: within = 0.0056 Obs per group: min = 1 between = 0.0286 avg = 3.9 overall = 0.0137 max = 6 F(1,610) = 3.46 corr(u_i, Xb) = -0.1480 Prob > F = 0.0635 --- MNE | Coef. Std. Err. t P>|t| [95% Conf. Interval] ---+--- CEESUB | -.0499986 .0268931 -1.86 0.063 -.1028128 .0028157 _cons | 515.9444 6.197144 83.26 0.000 503.7741 528.1148 ---+--- sigma_u | 730.3997 sigma_e | 113.8187

rho | .97629249 (fraction of variance due to u_i)

--- F test that all u_i=0: F(208, 610) = 135.42 Prob > F = 0.0000

Output Stata: Parent MNE to subsidiary Wage costs correlations

Fixed-effects (within) regression Number of obs = 976 Group variable: firmID Number of groups = 211 R-sq: within = 0.0000 Obs per group: min = 1 between = 0.0005 avg = 4.6 overall = 0.0003 max = 6 F(1,764) = 0.01 corr(u_i, Xb) = -0.0180 Prob > F = 0.9256 --- MNE | Coef. Std. Err. t P>|t| [95% Conf. Interval] ---+--- WE | .0032208 .0344823 0.09 0.926 -.0644705 .070912 _cons | 444.5031 4.50284 98.72 0.000 435.6637 453.3426 ---+--- sigma_u | 640.45302 sigma_e | 130.26877

rho | .96027168 (fraction of variance due to u_i)

--- F test that all u_i=0: F(210, 764) = 94.11 Prob > F = 0.0000

R-sq: within = 0.0002 Obs per group: min = 1 between = 0.0005 avg = 3.9 overall = 0.0017 max = 6 F(1,413) = 0.06 corr(u_i, Xb) = -0.0451 Prob > F = 0.7991 --- MNE | Coef. Std. Err. t P>|t| [95% Conf. Interval] ---+--- CEE | -.1331444 .5228083 -0.25 0.799 -1.160842 .8945527 _cons | 537.027 8.416152 63.81 0.000 520.4832 553.5709 ---+--- sigma_u | 662.74174 sigma_e | 120.79623

rho | .96784676 (fraction of variance due to u_i)