foreign-owned firms around the world: a comparative analysis of wages and employment at the...

Contents lists available at SciVerse ScienceDirect

European Economic Review

European Economic Review 60 (2013) 170–188

0014-29

http://d

$ The

results

Nurnbe

may be

under s

contact

program

5259 (Hn Corr

E-m

journal homepage: www.elsevier.com/locate/eer

Foreign-owned firms around the world: A comparativeanalysis of wages and employment at the micro-level$

Alexander Hijzen a, Pedro S. Martins b, Thorsten Schank c, Richard Upward d,n

a OECD, France and IZA, Germanyb Queen Mary University of London, United Kingdom and CEG-IST, Portugalc University of Mainz and IZA, Germanyd School of Economics, University of Nottingham, Nottingham NG7 2RD, United Kingdom

a r t i c l e i n f o

Article history:

Received 13 October 2011

Accepted 4 February 2013Available online 16 February 2013

JEL classification:

F16

F23

J31

Keywords:

Foreign direct investment

Foreign wage premia

Multinational enterprises

21/$ - see front matter & 2013 Elsevier B.V

x.doi.org/10.1016/j.euroecorev.2013.02.001

data used in this paper are confidential bu

in this paper. The data for Germany are av

rg, Germany. The data for the UK can be acce

accessed within the facilities of several rese

pecific rules agreed with the Statistics dep

ing the Ministry of Labour and Employmen

s used to estimate all the results in the pa

ijzen et al., 2010).

esponding author. Tel.: þ44 115 951 4735.

ail address: [email protected].

a b s t r a c t

This paper provides the first microeconomic cross-country analysis of the effects of

foreign ownership on wages, employment and worker turnover rates. Using firm-level

and linked worker-firm data, we apply a standardised methodology for three developed

(Germany, Portugal, UK) and two emerging economies (Brazil, Indonesia). We find that

wage effects are larger in developing countries, and that for each country the largest

effect on wages comes from workers who move from domestic to foreign firms.

Employment growth after foreign takeover is concentrated in high-skill jobs. In contrast

to widespread fears, there is no evidence that wage gains come at the expense of greater

job insecurity; separation rates actually fall slightly after takeover. We conclude that the

positive effect of foreign ownership on wages is not primarily driven by its impact on

incumbent wages, but by its impact on the creation of high-wage jobs.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Multinational enterprises have become key drivers of the world economy. The share of FDI in world GDP has more thantripled since the 1980s. The share of FDI to and from non-OECD countries has also increased substantially, particularlysince the 1990s. And in many developing countries FDI now represents the main source of external finance. Policy makershave generally welcomed these trends, emphasizing the potential benefits that FDI may bring to the host country. Sincemultinational enterprises (MNEs) are thought to have superior management or production techniques (or some other sortof ‘‘firm-specific asset’’) to be able to effectively compete with local firms in foreign markets, this operational advantagemay also benefit workers who are employed in the foreign affiliates of MNEs, or may spillover to the wider economy in

. All rights reserved.

t not exclusive, and may be used for non-commercial research by others, including replication of the

ailable by visiting the research data centre of the German Federal Employment Agency at the IAB,

ssed via the Secure Data Service at the UK Data Archive, University of Essex, UK. The data for Portugal

arch centres located in Portugal, including the Faculty of Economics at the University of Porto, and

artment of the Ministry of the Economy and Employment. The data for Brazil may be accessed by

t. The data for Indonesia can be purchased from Badan Pusat Statistik (BPS), Indonesia. The Stata

per are available from the authors on request. This article is a revised version of IZA working paper

uk (R. Upward).

www.elsevier.com/locate/eer

www.elsevier.com/locate/eer

http://dx.doi.org/10.1016/j.euroecorev.2013.02.001



http://crossmark.dyndns.org/dialog/?doi=10.1016/j.euroecorev.2013.02.001&domain=pdf



mailto:[email protected]


A. Hijzen et al. / European Economic Review 60 (2013) 170–188 171

which the foreign affiliate operates. However, the increased importance of MNEs in the world economy and in emergingmarkets in particular has also raised social concerns.

The way one evaluates the social impact of MNEs depends crucially on the normative standard that is used (OECD,2008). Based on a home-country standard — which involves comparing working conditions in the host country with thosein the home country — MNEs that have exploited international differences in labour costs by relocating some of theiractivities abroad have sometimes been accused of practising unfair competition. It is argued that foreign workers are notgiven their ‘just’ reward and, as a result, workers in the home country have to withstand unfair competition based onexcessively low pay. While such an argument may have a place in the debate on the social impact of outward investmentat home, it is potentially counterproductive in the context of the debate on the social impact of inward investment in thehost country. Alternatively, one may evaluate the social behaviour of MNEs by comparing working conditions against a setof universal standards such as those enshrined in the ILO’s core labour standards. Since in many low-income countrieslabour standards are not effectively enforced, human right activists have demanded that accountability mechanisms beput in place to ensure that core labour standards are respected throughout the operations of MNEs. While this is aninteresting issue to look at, systematic information on compliance levels of MNEs to core labour standards is lacking.

This paper assesses the way MNEs treat their workers in their foreign operations using a local standard that involvescomparing working conditions in the foreign affiliates of MNEs with those in comparable domestic firms. The differencemay be interpreted as the social impact of MNEs in the host country. This allows one to simultaneously analyse thepotential positive benefits emphasised by policy makers as well as social concerns over the tendency of MNEs to use theirbargaining power to force workers into sub-standard working conditions. Moreover, Porter and Kramer (2006) argue thatthe actual social impact also represents the appropriate benchmark to evaluate the corporate social responsibility of MNEsrather than, more narrowly, the extent to which corporate reputations for responsible business conduct are harnessed andstakeholder expectations satisfied.1 This notion is also reflected in major multilateral initiatives to promote responsiblebusiness conduct such as the ILO’s Tripartite Declaration of Principles Concerning Multinational Enterprises and SocialPolicy (ILO, 2000) and the OECD’s Guidelines for Multinational Enterprises (OECD, 2011). They both recommend MNEs toobserve standards of employment no worse than those observed by similar employers in the host country. This paper thusseeks to address the question whether it is better to work for a multinational or a national firm. The paper does notconsider the indirect social impact that MNEs may have on their environment, although this may be important as well.2

A large literature already exists that analyses differences in pay between foreign and domestic firms, so-called ‘foreignwage premia’. The literature has largely focused on differences between foreign and domestic firms rather thanmultinational and national firms as it is typically not possible to identify domestic MNEs with the available data. Thesame approach will be used in the present paper. Most of the previous evidence is based on firm-level data. Until recently,there was a consensus that foreign firms pay higher wages than their domestic counterparts and that foreign wage premiatend to be higher in less developed countries (e.g. Girma and Gorg, 2007; Lipsey and Sjoholm, 2006; Moran, 2006).However, with the increasing availability of linked employer–employee data (LEED) this consensus has been challenged(e.g. Martins, 2004; Heyman et al., 2007; Andrews et al., 2010). These authors all find that after controlling for worker andfirm characteristics, wage effects become much smaller, and in some cases disappear altogether or even become negative.This seems to be because foreign-owned firms select on worker quality: workers in foreign-owned firms would haveearned more even if they had worked for domestically-owned firms.

However, the implications of these recent studies for the conventional wisdom are not well understood. In part, this isbecause the results are qualitatively mixed. Why do some studies find small positive effects, and other insignificant oreven negative effects? Does this reflect differences in the econometric methodology (and particularly the use of differentcontrols), differences in country characteristics or differences in the nature of FDI? Another reason why the implicationsare difficult to gauge is that these recent studies are all limited to European countries, while the effects are generallybelieved to be much more important in developing countries (e.g. Moran, 2006). As a result, it is an open question what theeffect of controlling for firm and worker selection would be for the estimation of foreign wage premia in developingcountries.

This paper analyses the role of foreign ownership for wages, worker turnover and employment by focusing on changesin ownership status as a result of cross-border acquisitions or worker movements. In order to overcome the problem ofselection bias that is associated with the non-random nature of firm acquisitions and worker movements the study makesuse of propensity score matching in combination with difference-in-differences methods.3 In doing so, the paper makesthree key contributions. First, we replicate the consensus in the empirical literature by applying a standardisedmethodology to firm-level data for three developed (Germany, Portugal and the UK) and two emerging economies (Brazil

1 According to Porter and Kramer (2006, 2011) Corporate Social Responsibility (CSR) involves maximising ‘shared value’, i.e. benefits that accrue to

both business and society and, in the present context, outcomes that raise both labour practices and firm profitability. Shared value may come about not

just through the implementation of cost-increasing CSR policies that lead to higher product prices (the ‘‘demand-side’’ of CSR), but also through the

integration of CSR into management strategies that raise both labour practices and long-term productivity (the ‘‘supply-side’’ of CSR). The extension of

CSR to the supply-side expands the scope for CSR beyond the ‘‘willingness to pay’’ by consumers and investors for better labour practices. However, it

also makes it harder to distinguish the impact of CSR from the social impact of day-to-day business.2 See Chapter 5 of OECD (2008) for a recent overview.3 This method will yield estimates of causal effects if the unobservable determinants of firm acquisitions or worker mobility are fixed over time.

A. Hijzen et al. / European Economic Review 60 (2013) 170–188172

and Indonesia). Consistent with the conventional wisdom, the results indicate that foreign-owned firms offer substantiallyhigher average wages than domestic firms and that this difference is particularly important in emerging economies.Second, we further show that large wage effects tend to be associated with significant scale effects in terms of employmentand no, or small, reductions in worker turnover. This suggests that the bulk of the wage difference is likely to be associatedwith increases in hiring of relatively skilled or at least highly-paid workers. It also provides some evidence in favour oftheoretical models that suggest MNEs are more likely to pay efficiency wages. Third, we provide internationallycomparable evidence on the role of foreign ownership for average wages that controls for worker selection usingworker-level data for Brazil, Germany, Portugal and the United Kingdom.4 The results indicate that the positive wageeffects of foreign takeovers are substantially smaller when controlling for changes in the composition of the workforce,although they tend to remain positive. Moreover, the wage effects associated with worker movements from domestic toforeign-owned firms are potentially important, particularly in emerging economies.

The remainder of the paper is structured as follows. Section 2 discusses under what circumstances foreign-owned firmsmay have incentives to offer different wages and better working conditions to similar workers doing similar jobs indomestic firms. Section 3 discusses the empirical literature on the effects of foreign ownership on wages and workingconditions. Section 4 describes the various data sources used and their comparability across countries. Section 5 presentsthe econometric methodology and Section 6 presents the results. The final section provides some concluding remarks.

2. Theory

In a perfectly competitive setting, firms are price-takers in output and input markets and all workers are paid theirmarginal product. Consequently, one would expect foreign-owned firms to offer similar pay and working conditions toindividuals with similar characteristics doing similar jobs. Of course, when looking at the individual components of thepackage of pay and working conditions one may still observe differences between foreign-owned and domestic firms in aperfectly competitive setting. To the extent that working for foreign-owned firms implies certain advantages ordisadvantages, these may need to be compensated through the reward package that is being offered. For example, it issometimes suggested that jobs are less secure in foreign-owned firms because they are ‘‘footloose’’ or have more elasticlabour demand (e.g. Gorg and Strobl, 2003). This could provide a rationale for foreign-owned firms to offer higher wagesthan their local competitors to compensate for lower job security. However, this does not affect the net value of workingfor a foreign-owned firm. In non-competitive labour markets, differences in pay and working conditions between foreign-owned and domestic firms may occur even for individuals with similar characteristics doing a similar job. There may beseveral reasons for this. Most of these reasons suggest that foreign firms may provide better wages and workingconditions, but this does not necessarily have to be the case.

First, search frictions reduce the degree to which arbitrage takes place across firms due to differences in labourproductivity for identical workers. Search frictions may be particularly important for multinational enterprises as they arelikely to have less developed local networks. Search frictions are likely to induce a link between firm productivity andworking conditions.5 Since multinational firms tend to be more productive (Helpman et al., 2004), this means that workingconditions are likely to be better than in domestic firms, but not necessarily different from local firms with the same levelof productivity.6 However, to the extent that search frictions tend to be more important among MNEs because of theirlimited local networks, they may need to offer better wages and working conditions even than their local counterparts inorder to attract suitably qualified workers.

Second, efficiency-wage arguments may explain why the foreign affiliates of MNEs provide better wages and workingconditions than their domestic counterparts. For example, firms that derive their productivity advantage from firm-specific knowledge may wish to provide better working conditions in the hope that this would reduce worker turnover andminimise the risk of their productivity advantage spilling over to competing firms (Fosfuri et al., 2001; Glass and Saggi,2002). Alternatively, MNEs may be inclined to use better pay and working conditions to motivate the workforce tocompensate for the greater importance of monitoring costs or to overcome problems in managing industrial relations in acontext of different legal and cultural traditions. A third efficiency-wage argument is suggested by Egger and Kreickemeier(2013), who develop a model where (more productive) foreign-owned firms pay higher wages because of the fair wagepreferences of workers and the role of global rent sharing. They argue that, in such a context, MNEs operating betweensimilar countries pay higher wages because they are more productive, while MNEs that operate across countries that differin terms of their level of development will pay higher wages in the less advanced country even than domestic firms with

4 Indonesia is excluded from this part of the analysis for reasons of data availability.5 In standard search models, wages are determined through Nash bargaining which implies that high-performance firms are likely to pay higher

wages and provide better working conditions.6 The cost-saving that can be achieved by a MNE thanks to its monopsony power is likely to fall with the availability of comparable jobs outside the

MNE which may be closely related to the level of economic development. Decreuse and Maarek (2008) refer in this context to a technology-rent effect,

which allows MNEs to derive monopsony power from their technological advance, and a competition-wage effect, that follows from the competition

between firms for labour services.


similar levels of productivity. In sum, the foreign affiliates of MNEs may have stronger incentives to provide efficiencywages than their domestic counterparts because of its potential beneficial effects for worker retention and worker effort.7

The extent to which foreign-owned firms offer better pay and working conditions than their local counterparts forsimilar workers doing a similar job may vary across the country of ownership and the countries in which they operate. Wehypothesise that the gap in working conditions between foreign-owned and domestic firms will be greater in developingcountries because the technological difference is likely to be greatest, and the availability of comparable alternative jobopportunities lowest. To the extent that MNEs pay higher wages because of compensating differentials, such as higher jobinsecurity, we would expect foreign wage premiums to be positively associated with worker separation. However, whenMNEs pay higher wages because of efficiency wage arguments, foreign wage premia and worker separation should benegatively correlated. Moreover, as emphasised by Egger and Kreickemeier (2013), wage premia are likely to be larger indeveloping countries. To the extent that it takes time to acquire firm-specific knowledge, the incentive to offer betterworking conditions should also increase with job tenure, particularly among skilled workers.8

3. Empirical literature

3.1. Firm-level evidence

There is a large empirical literature on foreign wage premia. Until recently, there was a consensus that foreign firms payhigher wages than their domestic counterparts, particularly in developing countries. In an early study for Mexico, the USand Venezuela, Aitken et al. (1996) show that average wages in foreign-owned establishments are about 30% higher thanin domestic establishments. These wage differences persist after controlling for size, geographic location, skill mix andcapital intensity in Mexico and Venezuela, but not in the United States. This suggests that foreign-owned firms pay higherwages than their local competitors in developing countries. However, this does not necessarily mean that foreignownership improves employment conditions when a domestic firm is taken over by a foreign firm. Foreign firms mayselect the best domestic firms on the basis of characteristics that are not controlled for in the regression analysis, but areassociated with higher average wages. One such variable is the quality of the labour force. In order to address thispossibility, subsequent studies have analysed the extent to which foreign wage premia persist after controlling forobservable measures of worker quality.9

Most recent studies focus on cross-border takeovers to analyse the causal effect of a change in ownership status on workeroutcomes by making use of firm-level panel data. The main advantage of panel data is that it allows one to control for firm-selection, i.e. the possibility that foreign-owned establishments differ from domestic establishments because foreign investorsselect their targets on the basis of unobservable time-invariant characteristics rather than ownership status per se. Girma andGorg (2007) find, for the UK, that foreign takeovers of domestic firms tend to increase wages, but also that their effect depends onthe industry of target firms and the nationality of acquirers. For Indonesia, Lipsey and Sjoholm (2006) find that even aftercontrolling for firm-fixed effects, foreign takeovers raise production workers’ wages by 17% and non-production workers’ wagesby 33%. More generally, these studies show that controlling for fixed effects reduces the estimated foreign wage premiumwithout, however, challenging the basic finding that foreign-owned firms pay higher wages than domestic firms.

3.2. Evidence from linked employer–employee data

The results from firm-level studies may be somewhat misleading because they do not control for worker selection, i.e.the possibility that ownership changes are associated with changes in the composition of the workforce. To the extent thatunskilled workers tend to leave after takeovers and skilled workers join, this would bias the estimated foreign wagepremium upwards. Using linked employer–employee data, it is possible to control for changes in the composition of theworkforce by focusing on the wage effects for individual workers who stay in the same firm. Those data also allow one tolook at the role of ownership for workers who change jobs between domestic and foreign firms. This is interesting becauseit allows an analysis of differences in pay conditions between foreign and domestic firms for new workers. As productivitydifferences may have more important implications for workers at the moment of hiring than for stayers (Beaudry andDiNardo, 1991; Pissarides, 2009), one may expect the role of ownership to be more important for this category of workers.10

7 The social impact of MNEs on pay and working conditions does not necessarily have to be positive. For example, MNEs may use their bargaining

power to force workers to accept sub-standard employment conditions or to negotiate exemptions from labour provisions from governments.8 In this version of the paper, we do not look at the effects of foreign ownership by skill group. However, results by skill group are presented in the

working paper version of this article (Hijzen et al., 2010).9 For example, Lipsey and Sjoholm (2004) ask whether foreign wage premia may simply reflect differences in worker composition between foreign

and domestic firms. In order to address this possibility, they use an establishment-level dataset for Indonesia with detailed information on the

composition of workers across educational categories. They find that, while differences in average labour quality account for a significant part of the raw

foreign wage premium, the premia remain large, i.e. wages in foreign-owned establishments are 12% higher for production workers and 20% for non-

production workers. Morrissey and Te Velde (2003) present similar findings for five Sub-Saharan African countries.10 In addition, the analysis of worker movements takes account of both foreign-owned firms that were previously domestic, but have been acquired

by a foreign owner, and those that are established through ‘‘Greenfield’’ investment.


An increasing number of studies have made use of linked employer–employee data to analyse the role of foreignownership for individual wages. The results challenge the conventional wisdom by suggesting that foreign takeovers indeveloped countries have, at best, a small positive effect on individual wages and that this effect could even be negative.For example, Martins (2004) shows for Portugal that the foreign wage premium disappears after controlling for workerselection and may even reduce individual wages for workers in foreign firms relative to their counterparts in domesticfirms. Heyman et al. (2007) present similar findings for Sweden which indicate that foreign takeovers reduce individualwages relative to their counterparts in domestic firms, while Andrews et al. (2010) for Germany, Malchow-Moller et al.(forthcoming) for Denmark and Huttunen (2007) for Finland find small positive effects (1%–3%). Relatively few studiesexploit worker mobility to analyse the role of foreign ownership. Two exceptions are Andrews et al. (2010) and Balsvik(2011), who show that workers moving from a domestic to a foreign firm experience a 6% increase in wages in Germanyand 8% in Norway.11 These findings may indicate that the short-term effects of foreign ownership may be more importantfor new hires in foreign firms than workers who stay in firms that change ownership.12

Overall, the recent evidence based on worker-level data provides a somewhat mixed message with respect to theimpact of foreign ownership on wages. While most studies indicate that foreign ownership has a positive impact on wages,a number of studies indicate small negative effects. It is not clear what drives these differences in estimated wage premiaacross studies. They may reflect differences in country characteristics or the nature of FDI, as well as differences inmethodology. Moreover, the effect of controlling for changes in the composition of the workforce on the estimation offoreign wage premia in developing countries (where such premia are believed to be much larger) remains an openquestion. In order to better understand the implications of these new findings, we provide the first comparable cross-country evidence for a number of developed and developing countries.

4. Data sources and cross-country comparability

4.1. Data sources

The data used for this paper are drawn from the national administrative systems of five countries.

4.1.1. Brazil

The main data source is RAIS (Annual Social Information Report), an annual census of all firms and their employees.It contains detailed information for each employee (wages, hours worked, education, age, tenure, gender, etc) and eachfirm (industry, region, size, establishment type, etc.).13 With the establishment identification number it is possible tofollow all establishments that file the RAIS survey. Moreover, with the worker’s national insurance number, it is possible tofollow all workers that remain in the formal sector and to match the workers’ characteristics with those of theestablishment. Therefore, we can create a panel that matches workers to their establishments and follow each of themover time. The establishment identifier is used to link RAIS with external sources with information on cross-bordermergers and acquisitions and foreign ownership (Thomson Financial and ORBIS).

4.1.2. Germany

Two data sources are used. First, the Institut fur Arbeitsmarkt- und Berufsforschung (IAB) Establishment Panel. The IABEstablishment Panel is an annual survey of approximately 16,000 establishments located in both Western and Eastern Germany,covering 1% of all establishments and 7% of all employment in Germany. The IAB Panel includes information on employment,bargaining arrangements, total sales, exports, investment, wage bill, location, industry, profit level and nationality of ownership.Ownership is defined as either Western German, Eastern German, foreign, or public. Complete information on establishmentownership is available for all establishments only in 2000 and 2004. Second, we use the employment statistics register of theGerman Federal Office of Labour (Beschaftigtenstatistik). The Beschaftigtenstatistik covers all employees or trainees registered by thesocial insurance system. Reported daily gross wages are censored at the social security contribution ceiling. The register coversabout 80% of employees in Western Germany and about 85% in Eastern Germany. Information on employees includes basicdemographics, start and end dates of employment spells, occupation and industry, earnings, qualifications, and an establishmentidentification number. By using the establishment identification number, we can associate each worker with an establishment inthe panel. The analysis is restricted to the years 2000 and 2004 and to Western Germany.

4.1.3. Portugal

The analysis draws on a large matched employer–employee data set, Quadros de Pessoal (Personnel Records),a compulsory survey conducted annually by the Ministry of the Economy and Employment of all firms located in Portugal

11 Martins (2011) finds similar results for Portugal.12 Table 1 in Hijzen et al. (2010) provides a schematic summary of the literature.13 RAIS is an administrative report filed by all tax registered Brazilian establishments. Since the information may be used for investigations of labour

legislation compliance, firms that do not comply with it do not file in RAIS. Thus, this data set can be considered a census of the formal Brazilian labour

market (state-owned enterprises, public administration and non-profit organisations are also required to file the report). Firms that do not provide

accurate information will be committing an offence sanctioned by law, a threat that is likely to lead to very high standards of data quality.


with at least one employee. Quadros de Pessoal provides detailed information on all workers in each firm including gender,education, tenure, wages and hours worked, and their firms including foreign ownership, sales, industry andregion.14 There is also a unique identifier for each firm and individual that allows one to follow firms and individualsthrough time.

4.1.4. United Kingdom

The analysis makes use of the Annual Respondent’s Database (ARD) for the firm-level analysis, and the Annual Survey ofHours and Earnings (ASHE) linked to the Business Structure Database (BSD) for the worker-level analysis. The ARD is acensus of large businesses and a statutory survey of smaller businesses conducted annually by the UK Office for NationalStatistics (ONS). The BSD is a series of annual snapshots from the Inter-Departmental Business Register, a live register ofUK businesses maintained by the ONS, containing information on employment, sales, foreign ownership, industry andregions as well as a unique enterprise reference number. The ASHE is a 1% random sample of employees with detailedinformation on pay and hours of work and can be linked to the BSD using the enterprise reference number.

4.1.5. Indonesia

The analysis makes use of the Indonesian Census of Manufacturing (Survei Manufaktur), which is conducted annually bythe National Statistical Office (BPS). The data include all manufacturing establishments with more than 20 employees.Linked worker-firm data are not available, and therefore Indonesia can only be included in the firm-level analysis.

4.2. Cross-country comparability

Table 1 summarises information on the various data sources in order to provide a sense of their comparability acrosscountries. The datasets differ along three key dimensions: the sampling methodology, the unit of observation andcoverage. The sampling methodology may in turn differ in terms of the sampling frame and the sampling method. One candistinguish two types of sampling frames (Vilhuber, 2008): worker-based sampling frames and firm/establishment-basedsampling frames. The German data present an example of an establishment-based sample, where we have a sample ofestablishments for which all (or most) workers can be identified. The UK data is a worker-based sample in the sense thatfor a fraction of the workforce all employers can be identified. In addition to the sampling frame, the data sources may alsodiffer in terms of the sampling method. The German data are an example of a non-random size-weighted sample. The UKASHE data, by contrast, include a 1% random sample of all employees.

Differences in the sampling frame affect the extent to which specific issues can be appropriately analysed in differentcountries, and as a result, affect the scope for cross-country analysis. Worker-based samples are appropriate for theanalysis of worker outcomes, but may be ill-suited for the analysis of firm outcomes. For example, worker-based samplescan be usefully employed to analyse the effects of globalisation (including foreign ownership) for individual workers, butare less adequate for the analysis of worker composition for firm outcomes (worker turnover or skill composition). Themain drawback of firm-based samples is that it reduces the ability to analyse worker movements across establishments.This is due to the fact that in firm-based samples the probability that a worker switches from a firm in the sample toanother firm in the sample depends on the sample size of firms. It is thus possible that the German sample is notappropriate for the analysis of foreign ownership based on worker movements due to the relatively small number ofworker movements between establishments in the sample.

In addition to sample-based data, some data sources include the entire universe of (formal sector) firms and workers.The data for Brazil and Portugal are examples of this. Such data are ideal in the sense that they are not subject to any apriori limitations due to the characteristics of the sampling methodology. Unfortunately, data such as these are onlyavailable for a limited number of countries.

The cross-country comparability of the analysis is also impaired due to differences in the unit of observation. Theanalysis for the UK is conducted at the level of the firm (enterprise), whereas the analysis for Germany is necessarilyconducted at the level of the establishment. While the analysis for Brazil and Portugal can, in principle, be conducted ateither level, we use data at the firm-level. These differences create differences in firm-size statistics and also affectmeasures of worker turnover.15

Finally, the various samples also differ in terms of their coverage across sectors and over time. The data for Brazil,Portugal, and Germany include all market sectors. The data for the UK exclude the banking sector. The data for Indonesiaonly span the manufacturing sector. The data for Portugal and the United Kingdom broadly span the same time period,whereas the years covered in the German data is more limited.

14 Particular emphasis is placed on the reliability of the information, as it is used by the Ministry of the Economy and Employment for the purpose of

checking the employer’s compliance with labour law.15 The definition of foreign ownership however is unambiguous, whether or not the sample is firm- or establishment-level.

Table 1National data sources.

Country Data sources Unit of

observation

Sample selection Sectoral

coverage

Time coverage

Germany Institut fur Arbeitsmarkt- und

Berufsforschung (IAB) Establishment Panel

and the employment statistics register of

the German Federal Office of Labour

(Beschaftigtenstatistik)

Establishment All establishments with employees subject

to social security. Large establishments are

oversampled. The sample comprises about

1% of establishments and 7% of employees

All sectors 2000 and 2004

Portugal Quadros de Pessoal or ‘Personnel Records’

(Ministry of Employment)

Firm All firms with at least one employee All sectors 1997–2004

except 2001

UK Annual Respondent’s Database (ARD) for

the firm-level analysis. Business Structure

Database (BSD) and Annual Survey of

Hours and Earnings (ASHE) for worker-

level analysis

Firm The ‘‘selected sample’’ of the ARD is a

census of firms with 250 or more

employees, and a sample of smaller firms.

The BSD includes all enterprises whose

establishments are subject to VAT or social

security. The ASHE is a 1% random sample

of employees

All sectors

except banks

1997–2005

Brazil RAIS, Global Mergers and Acquisitions

Database (Thomson Financial Securities)

and Orbis (Bureau van Dijk)

Firm All firms with at least one employee All sectors 1994–2005

Indonesia Survei Manufaktur, the Indonesian Census

of Manufacturing (Statistical Office, BPS)

Establishment The census surveys all registered

manufacturing establishments with more

than 20 employees.

Manufacturing 1997–2005

except 2001


4.3. Descriptive statistics

In this subsection, we briefly discuss a number of descriptive statistics with the aim of both illustrating potentialproblems related to the cross-country comparability of the results, and to preview the econometric analysis in the nextsection. The definition of all the variables used in the analysis can be found in Table A1 in Appendix A. Note that,throughout, foreign ownership is defined on the basis of majority ownership.

Table 2 reports summary statistics on log wages at the individual and firm/establishment-level by ownership statusand country.16 The table shows that, as one would expect, wages in foreign-owned firms are on average higher than thosein domestic firms. Of course, such comparisons do not reveal whether wage differences are due to differences in thecomposition of the workforce, the characteristics of foreign and domestic firms or ownership status per se. Moreover,average differences might hide differences in the distribution of wages across foreign and domestic firms. For example,foreign-owned firms may provide larger wage differentials for skilled than for unskilled workers.

As we explain in Section 5, the econometric analysis will focus on the wages of workers whose employer changesownership status, or workers who move between employers of different ownership status, relative to a control group.Table 3 therefore reports the number of ownership changes in the sample for each type of comparison and for each country.

The number of firms that change ownership status as a result of cross-border takeovers is substantial except inGermany17 and Brazil in the case of domestic takeovers of foreign firms. Takeovers as a fraction of domestic firms ishighest in Indonesia and the UK, and very low in Brazil. The number of domestic takeovers of foreign firms is less than 10in Brazil, and we therefore drop this case from the analysis for this country.

At the worker level, we observe a much larger number of observations because each takeover affects the entireworkforce in the target firm. The proportion of workers experiencing takeover activity is usually larger than the proportionof firms, reflecting the fact that takeover activity is greater in larger firms. Again, the UK has the highest rate of workersaffected by takeover activity. Worker mobility from domestic to foreign firms is also important in Portugal and the UK,affecting over 1% of workers per year, while mobility from foreign to domestic firms affects over 5% of workers in foreign-owned firms. The much smaller fraction in Germany reflects the nature of the sample: the probability of observing aworker moving from one establishment to another which are both in the sample is very low. In Brazil, the low mobilityrate reflects the smaller fraction of foreign-owned firms.

16 Worker-level information is not available for Indonesia.17 Note that the rate for Germany refers to the change between 2000 and 2004, so the yearly rate is much lower than reported in Table 3.

Table 2Summary statistics of log wage by ownership status (in national currency).

Firm/establishment-level Worker-level

N Mean log w S.D. log w N Mean log w S.D. log w

Germany

All firms 3474 4.351 0.345 397,584 4.625 0.299

Foreign-owned firms 290 4.587 0.323 87,697 4.701 0.281

Domestic firms 3184 4.330 0.339 309,887 4.604 0.300

Portugal

All firms 146,843 1.464 0.447 6,928,076 1.537 0.583


Domestic firms 139,762 1.436 0.422 5,943,245 1.499 0.567

United Kingdom

All firms 78,850 2.921 0.588 441,159 2.199 0.525


Domestic firms 69,630 2.877 0.587 362,515 2.162 0.513

Brazil

All firms 156,524 1.964 0.660 12,775,660 2.484 0.947

Foreign-owned firms 2165 3.282 0.666 2,536,778 3.021 0.849

Domestic firms 154,359 1.945 0.641 10,238,882 2.351 0.922

Indonesia

All firms 74,723 8.408 0.998 – – –

Foreign-owned firms 9632 9.018 0.924 – – –

Domestic firms 65,091 8.318 0.977 – – –

Table 3Foreign ownership changes of workers and firms.

Germanya Portugal United Kingdom Brazil Indonesia

Foreign takeovers of domestic 36 535 967 114 793

firms (number of firms) 2.25% 0.46% 2.03% 0.08% 2.07%

Domestic takeovers of foreign 19 256 351 3 463

firms (number of firms) 13.87% 4.54% 5.33% 0.16% 7.08%

Foreign takeovers of domestic 11,976 36,222 5,715 144,601 –

firms (number of workers) 7.54% 0.75% 1.94% 1.77%

Domestic takeovers of foreign 3754 18,195 1736 1650 –

firms (number of workers) 9.41% 2.37% 2.96% 0.08%

Workers moving from 341 50,529 4556 31,601 –

domestic to foreign firms 0.21% 1.04% 1.55% 0.39%

Workers moving from 313 39,517 2980 19,891 –

foreign to domestic firms 0.78% 5.16% 5.09% 1.00%

Table reports number of firms and workers who change ownership status between t�1 and t during the sample period. Percentage is as a proportion of

observations in original state at t�1, e.g. 535 Portuguese firms changed from domestic to foreign which is 0.46% of all establishments which were

domestic in t�1.a German data refer to changes between 2000 and 2004.


5. Methodology

Our methodology utilises both firm-level18 and worker-level information. We focus on three key measures: wages,employment and worker turnover, and we seek to answer the following questions. First, what is the simple difference ineach measure between foreign and domestic firms, and how much of this difference is explained by observablecharacteristics of firms? Second, how much of this difference is explained by firm selection? Finally, how much of thisdifference is explained by worker selection?

To deal with firm and worker selection, we consider a variety of ‘‘treatments’’ based on changes in firms’ ownershipstatus. At the firm-level, there are two treatments: foreign takeovers of domestic firms and domestic takeovers of foreign-owned firms. The natural counterfactual in these cases are firms who do not change ownership status. We then examinewhat happens to average firm-level wages as a result of these changes in ownership. At the worker-level, we use

18 Strictly speaking, this is firm- or establishment-level, depending on country.


information on the identity of workers in foreign and domestic firms to shed more light on the mechanisms by whichforeign ownership increases average wages. First, we test whether the change in firm-level wages is a result of changingworker characteristics by comparing workers who remain in the same firm after the change in ownership status(‘‘stayers’’). Second, we test whether those workers who join foreign-owned firms experience greater wage increases thanthose workers who move between firms of the same ownership type (‘‘movers’’). Each treatment will be evaluated at apoint in time less than 12 months after the change in ownership status occurs (t¼0), one to two years after the change inownership status (t¼1), and two to three years after the change in ownership status at (t¼2).19

Let Fjt ¼ 1 indicate foreign ownership of firm j at time t. Define a firm-level foreign takeover treatment indicator TjF

TFj ¼

0 if Fj,t ¼ �1 ¼ 0 and Fj,t ¼ 0 ¼ 0

1 if Fj,t ¼ �1 ¼ 0 and Fj,t ¼ 0 ¼ 1

(: ð1Þ

Similarly, TjD

indicates domestic takeovers of foreign firms.There are also worker-level treatment indicators which account for worker composition effects. For example, the

indicator for stayers whose firm changes ownership is defined as

TSFi ¼

0 if FJði,t ¼ �1Þ ¼ 0 and FJði,t ¼ 0Þ ¼ 0 and Jði,t¼�1Þ ¼ Jði,t¼ 0Þ

1 if FJði,t ¼ �1Þ ¼ 0 and FJði,t ¼ 0Þ ¼ 1 and Jði,t¼�1Þ ¼ Jði,t¼ 0Þ

(, ð2Þ

where the function j¼ Jði,tÞ maps worker i at time t to a particular firm j. Similarly, TiSD

indicates workers who remain in afirm which changes from foreign to domestic.

Finally, the analysis of worker movements involves comparing workers who switch to another firm with the sameownership status with workers who switch to another firm with different ownership status. For example

TMFi ¼

0 if FJði,t ¼ �1Þ ¼ 0 and FJði,t ¼ 0Þ ¼ 0 and Jði,t¼�1ÞaJði,t¼ 0Þ

1 if FJði,t ¼ �1Þ ¼ 0 and FJði,t ¼ 0Þ ¼ 1 and Jði,t¼�1ÞaJði,t¼ 0Þ

(: ð3Þ

Similarly, TiMD

indicates movers from foreign to domestic firms.We use the method of propensity-score matching (PSM) in combination with difference-in-differences. PSM involves

constructing treated and control groups ex post with similar observable characteristics. Like OLS, PSM provides estimatesof the causal impact if selection into the treatment is on the basis of the observed covariates used in the propensity model.The mean difference in outcomes between the treated and untreated gives the average treatment effect on the treated.

The propensity score is estimated with a Probit model which specifies the probability of changing ownership status as afunction of characteristics observed before the treatment occurs. When making firm-level comparisons we use only firm-level characteristics; for worker-level comparisons we use both firm- and worker-characteristics. This ensures thattreatment and control group firms do not systematically differ in terms of the composition of workers in those firms. Thefirm-level characteristics used are industry and region dummies, log employment, the change in log employment20 andthe log average wage. The worker-level characteristics used are log individual wage, a gender dummy, age, age squaredand tenure. All these variables are measured at t¼�1, the year before Fjt may change. Treated individuals are matched totheir untreated counterparts using one-to-one nearest-neighbour matching which attributes a weight of one to the nearestuntreated neighbour of each treated observation and zero to others. The estimation of the propensity score and thematching procedure are conducted separately for each year, broad industry (manufacturing or services) and skill group(unskilled, semi-skilled and skilled).21 Various robustness checks which use different matching methods are carried out toensure that the particular matching method used does not affect the results.22

Propensity-score matching is complemented with the difference-in-differences (DiD) estimator, following Heckmanet al. (1997). The DiD estimator controls for any pre-existing constant differences in the outcome variable before thechange in ownership, even if those differences are caused by unobservable attributes. The actual regressions are estimatedwith (firm or worker) fixed effects, which represent a generalisation of DiD. In order to avoid conflating treatment andcomposition effects related to the appearance pattern of individuals, each cohort should be balanced. Therefore the sampleis restricted to firms/individuals that are present in each year of relative time period t¼�1 to t¼2. The firm-level DiDestimate for foreign takeovers comes from

yjt ¼ ajþXs ¼ 2

s ¼ 0

dsDstþ

Xs ¼ 2

s ¼ 0

gsðDst � T

Fj Þþejt , ð4Þ

where aj is a fixed effect for firm j and Dts

is a relative time indicator which takes the value 1 if t¼s and zero otherwise.The coefficients ds capture the effect of relative time for both treatment and control firms, while the coefficients gs capture

19 Note that because we have annual panel data, the precise date of change of ownership is not known.20 Including the change in log employment between t¼�2 and t¼�1 ensures that we match firms with similar trends in employment growth.21 In the case of Germany, where we have many fewer observations changing ownership status, these variables are treated as additional matching

variables rather than splitting the sample.22 The results of these robustness checks are reported in Table B1 in Appendix B.


the additional treatment effect on those firms with TFj ¼ 1. Similar models for yit are estimated for the worker-level

information.For one country in our sample (Germany) we only observe ownership status at two points in time (2000 and 2004), and

so we cannot estimate separate treatment effects for t¼0, 1, 2. We therefore also estimate (for all countries) a simplified‘‘before and after’’ version of (4)

yjt ¼ ajþdaDat þgðD

at � T

Fj Þþejt , ð5Þ

where Dat ¼ 1 if t4 ¼ 0 (‘‘after’’) and zero otherwise. In this model g captures the average post-takeover change in yjt.

One drawback of Eqs. (4) and (5) is that they only account for the effects of ownership changes up to three years afterthe event. To the extent that the positive effects of ownership changes take time to materialise, the DiD results may notcapture the full effect of foreign ownership. For this reason, it is worth complementing the DiD results with simple levelcomparisons between foreign and domestic firms that control for observable characteristics. While these comparisons mayprovide upward biased estimates of foreign–domestic differences in employment conditions, they are also more likely tocapture the long-term effects of foreign direct investment on employment conditions in foreign-owned firms. Theseresults may thus be interpreted as giving an upper bound on the long-term effects of foreign ownership on employmentconditions.

6. Results

In this section the empirical evidence on the effects of foreign ownership will be presented. Section 6.1 focuses on theeffects of cross-border takeovers using firm-level data, while Section 6.2 reports the estimated effects of cross-bordertakeovers and worker movements between domestic and foreign firms using linked worker-firm data.

Before discussing the results in detail, we should note that the interpretation of our results depends to a large extent onwhether we have successfully matched ‘‘treated’’ observations (firms or workers which change ownership) with ‘‘control’’observations. We therefore conduct a battery of tests to examine the effect of the propensity score matching on theobservable characteristics of the control and treatment groups, shown in Table B2. This table reports the number ofobservations in each group, the mean standardised bias as in Rosenbaum and Rubin (1985), the result of individual t-testson matching covariates and the standardised difference in matching covariates as in Imbens (2009). In almost all cases thematching procedure greatly reduces the mean bias and the number of covariates which are significantly different betweenthe treatment and control groups.23 In Table B1 we also examine whether our findings are robust to the choice of matchingmethod used. There are only minor differences in the results across the three different methods used, and none of thesedifferences alter our main findings.

6.1. The effect of foreign ownership at the firm-level

6.1.1. Average wages

Table 4 presents firm-level evidence using recent data for Germany, Portugal, the UK, Brazil, and Indonesia. Thebasic level comparisons show that differences in average wages between foreign-owned and local firms are large inall of the five countries. The raw differences indicate that foreign-owned firms pay considerably more on average thanlocal firms, with pay differences varying from 29% in Germany to 44% in the UK, 79% in Portugal, 116% in Indonesia, and280% in Brazil.24 Controlling for observable firm characteristics (log employment, industry and region) reduces the averagewage differences between foreign-owned and domestic firms, but they still remain sizable. The wage gap ranges from 11%in Germany to 35% in the UK, 40% in Indonesia, 42% in Portugal and 150% in Brazil.

However, these level comparisons are misleading if one is interested in the causal impact of foreign ownership. Foreign-owned firms also tend to be more productive, more capital-intensive and tend to use more skilled labour. All thesefactors are positively associated with size and higher wages. In order to disentangle the role of foreign-ownership fromother factors that affect employment and wages, the analysis below focuses on changes in ownership as a result of cross-border takeovers. We therefore also report estimates based on the change in wages following changes in foreignownership status as a result of cross-border takeovers. This controls for any pre-existing differences between differenttypes of firms.

Foreign takeovers of domestic firms do increase average wages, although the size of the effect varies considerablyacross countries. The effects range from 2% in Germany to 5% in the UK, 8% in Portugal, 16% in Brazil, and 21% inIndonesia, and furthermore the effect is insignificantly different from zero in Germany and the UK. These results, thefirst to use a consistent methodology across countries, confirm previous studies, and demonstrate clearly that thewage premium is much larger in the two less developed economies (see the discussion of the empirical literature inSection 3).

23 Ideally, we would like to ensure that the treatment and control groups are balanced across every covariate, but this was impractical due to the

requirement that we use the same matching specification across all countries.24 Log-point estimates are converted to percentages using expðbÞ�1.

Table 4The effects of cross-border takeovers on average wages: firm-level evidence.

Germany Portugal United Kingdom Brazil Indonesia

(a) Level comparisonsa

Without controls 0.255nnn 0.585nnn 0.366nnn 1.336nnn 0.771nnn

(0.020) (0.014) (0.010) (0.038) (0.010)

With controls 0.106nnn 0.354nnn 0.297nnn 0.937nnn 0.334nnn

(0.017) (0.011) (0.010) (0.039) (0.011)

(b) Foreign takeovers of domestic firmsb

Average effect 0.020 0.078nnn 0.048 0.147nn 0.189nnn

(0.015) (0.027) (0.025) (0.064) (0.046)

Effect at t¼0 – 0.066nn 0.032 0.148nn 0.175nnn

(0.030) (0.027) (0.069) (0.044)

Effect at t¼1 – 0.110nnn 0.049 0.126n 0.206nn

(0.033) (0.028) (0.067) (0.084)

Effect at t¼2 – 0.057n 0.064 0.167nn 0.221nn

(0.034) (0.034) (0.075) (0.090)

(c) Domestic takeovers of foreign firmsb

Average effect 0.001 �0.022 �0.015 – �0.110

(0.029) (0.036) (0.057) (0.068)

Effect at t¼0 – 0.000 �0.048 – �0.119

(0.038) (0.064) (0.072)

Effect at t¼1 – �0.063 0.012 – �0.097

(0.049) (0.055) (0.093)

Effect at t¼2 – �0.002 �0.009 – �0.058

(0.047) (0.062) (0.108)

nsignificant at 10%, nn significant at 5%, nnn significant at 1%, standard errors clustered at the firm-level.a Estimated using OLS. Controls include log employment, industry and region dummies.b Estimated using difference-in-difference propensity-score matching; see Eqs. (4) and (5). The propensity score is estimated using a Probit model which

includes firm-level characteristics measured at t¼�1: log employment, change in log employment, log average wage, industry and region dummies.


There is some evidence that the effects of foreign takeovers on average wages become larger over time. The effects offoreign takeovers after two years tend to be larger than their direct effect immediately after takeovers in the UK, Brazil andIndonesia, although the picture is less clear for Portugal, where the effect is largest after two years. The increase in theeffect may reflect the time it takes to transfer technology from parent to affiliate and for employees to accumulate humancapital. However, it may also reflect the impact of foreign takeovers on the composition of the workforce or the possibilitythat wage contracts do not change instantaneously.

While foreign takeovers of domestic firms tend to lead to wage increases, there is little evidence of wage falls fromdomestic takeovers of foreign firms, shown in the bottom panel of Table 4. Although almost all coefficients are estimatedto be negative, none are statistically significant at conventional levels. The largest estimated effect is for Indonesia (a fall ofabout 10% with a standard error of 7%), but even this is much smaller than the positive effect of foreign takeover. Thisasymmetry supports the hypothesis that foreign takeovers are accompanied by the transfer of modern production andmanagement practices from the parent to the foreign affiliate.25

6.1.2. Employment

The first panel of Table 5 shows clearly that foreign-owned firms employ many more workers than domestic firms, withemployment differences in all countries being greater than 100%. Even after controlling for industry and region, foreign-owned firms are still much larger, ranging from 85% in Portugal to over 500% in Brazil.

However, we get quite a different picture when we consider changes in employment which occur after foreigntakeover. Foreign takeovers also tend to raise employment in some of the countries analysed, but not in all. Foreigntakeovers of domestic firms raise employment by 25% in Indonesia and 24% in Portugal. The estimated coefficients forBrazil are not precisely estimated due to the small number of foreign takeovers observed (see Table 3), while for the UKand Germany foreign takeovers are associated with falls in employment, although these effects are also not preciselyestimated. To the extent that the large expansion in employment and production associated with cross-border takeovers inthose countries affects the composition of the workforce, this may have important implications for the estimated foreign-wage premia discussed above. More specifically, to the extent that the increase in employment following foreign takeoversin those countries is associated with an increase in the average skill intensity, firm-level estimates of foreign wage premiamay be biased upward. Even though the scale effects of foreign takeovers appear to be less important in the case ofGermany and the UK, this does not necessarily imply that foreign takeovers do not result in important changes in the

25 The data do not allow us to look at the impact of domestic mergers and acquisitions in all countries.

Table 5The effects of cross-border takeovers on employment: firm-level evidence.


Panel I: total employment


Without controls 1.632nnn 0.834nnn 0.890nnn 2.675nnn 1.244nnn

(0.102) (0.032) (0.031) (0.107) (0.014)

With controls 1.155nnn 0.613nnn 0.872nnn 1.827nnn 1.065nnn

(0.089) (0.032) (0.031) (0.110) (0.014)


Average effect �0.079 0.213nnn�0.030 0.017 0.220nnn

(0.092) (0.066) (0.029) (0.116) (0.027)

Effect at t¼0 – 0.219nnn�0.025 �0.085 0.213nnn

(0.062) (0.020) (0.104) (0.027)

Effect at t¼1 – 0.215nnn�0.035 0.069 0.245nnn

(0.070) (0.030) (0.122) (0.046)

Effect at t¼2 – 0.205nnn�0.031 0.068 0.247nnn

(0.079) (0.041) (0.150) (0.060)


Average effect �0.000 �0.131 �0.055 – �0.011

(0.126) (0.117) (0.068) (0.032)

Effect at t¼0 – �0.063 0.036 – �0.012

(0.119) (0.043) (0.031)

Effect at t¼1 – �0.095 �0.105 – �0.037

(0.129) (0.085) (0.054)

Effect at t¼2 – �0.235 �0.097 – 0.035

(0.143) (0.070)

Panel II: employment by skill-group

Foreign takeovers of domestic firmsb

Average effect �0.083 0.207nnn – �0.246n –

(Low skill)c (0.101) (0.080) (0.140)

Average effect 0.111 �0.142 – �0.027 –

(Medium skill)c (0.134) (0.081) (0.150)

Average effect 0.056 0.211nnn – 0.069 –

(High skill)c (0.079) (0.073) (0.090)

nsignificant at 10%, nn significant at 5%, nnn significant at 1%, standard errors clustered at the firm-level.a Estimated using OLS. Controls include log average wage, industry and region dummies.b Estimated using difference-in-difference propensity-score matching; see Eqs. (4) and (5). The propensity score is estimated using a Probit model which

includes firm-level characteristics measured at t¼�1: log employment, change in log employment, log average wage, industry and region dummies.c Skill based on highest educational qualification; not available from UK data.


composition of the workforce. However, the potential bias due to the role of composition effects is likely to be somewhatsmaller.

To further examine the effect of foreign takeovers on worker composition, the final panel of Table 5 examinesemployment change by skill group for the three countries where worker-level information on skill is available (Germany,Portugal and Brazil). In Brazil there appears to be a clear pro-skill bias in the employment growth which occurs aftertakeover, in particular with large falls in employment of the bottom skill group. In Germany there are also falls inemployment of the bottom skill group, but in Portugal there is a distinct U-shape, with expansion in both low- and high-skill employment. Thus, we cannot say unambiguously that foreign takeovers lead to skill-biased employment growth, notleast because in the case of Germany and Brazil the effects are rather imprecisely estimated.

6.1.3. Worker turnover

As noted in Section 3, one motivation for the higher wages paid by multinationals may be to reduce worker turnover tolower the risk of productivity advantages spilling-over to rival firms (Fosfuri et al., 2001; Glass and Saggi, 2002). On theother hand, other authors have argued that multinationals may offer less secure employment because they can more easilyshift production across locations (Fabbri et al., 2003; Gorg and Strobl, 2003). An important empirical issue therefore iswhether foreign-owned firms offer more or less secure employment than domestic firms. In Table 6 we estimate the effectof takeover on worker turnover for Germany, Portugal and Brazil, where worker turnover is defined as the separation rateat the firm-level.26

26 Worker-level data are not available for Indonesia, so separation rates cannot be calculated. For the UK the sample is a 1% sample of workers,

preventing a firm-level analysis of separation rates for all but the largest firms.

Table 6The effects of foreign takeovers of domestic firms on worker separation rates.

Germany Portugal UK Brazil

Average effect �0.042 �0.026 – �0.062nn

(0.035) (0.024) (0.031)

Effect at t¼0 – �0.027 – �0.064nn

(0.023) (0.029)

Effect at t¼1 – �0.020 – �0.058

(0.027) (0.039)

Effect at t¼2 – �0.029 – �0.062nn

(0.032) (0.035)

nsignificant at 10%, nn significant at 5%, nnn significant at 1%, standard errors clustered at the firm level.

Estimated using difference-in-difference propensity-score matching; see Eqs. (4) and (5). The propensity score is estimated using a Probit model which

includes firm-level characteristics measured at t¼�1: log employment, change in log employment, log average wage, industry and region dummies.

Dependent variable is the number of workers who leave the firm or establishment between t�1 and t as a proportion of total employment at t�1.

Table 7The effects of cross-border takeovers of domestic firms on individual wages of workers who do not change firm.



Without controls 0.092nnn 0.263nnn 0.197nnn 0.701nnn

(0.001) (0.031) (0.004) (0.086)

With controls 0.040nnn 0.132nnn 0.117nnn 0.257nnn

(0.001) (0.014) (0.003) (0.034)


Average effect 0.030n 0.034 0.011 0.057n

(0.016) (0.031) (0.008) (0.033)

Effect at t¼0 – 0.010 0.015n 0.069nn

(0.020) (0.008) (0.033)

Effect at t¼1 – 0.052 0.011 0.050

(0.040) (0.010) (0.039)

Effect at t¼2 – 0.039 0.008 0.052

(0.040) (0.011) (0.033)


Average effect 0.011 �0.025 0.008 –

(0.011) (0.024) (0.031)

Effect at t¼0 – �0.096n�0.015 –

(0.053) (0.035)

Effect at t¼1 – �0.028 0.036 –

(0.022) (0.035)

Effect at t¼2 – 0.049 0.004 –

(0.036) (0.041)

nsignificant at 10%, nn significant at 5%, nnn significant at 1%, standard errors clustered at the firm-level.a Estimated using OLS. Controls include log employment, industry and region dummies, sex, age, skill-level and tenure.b Estimated using difference-in-difference propensity-score matching; see Eqs. (4) and (5). The comparison is restricted to those workers who remain in the same

firm before and after the change in ownership; see Eq. (2). The propensity score is estimated using a Probit model which includes firm- and worker-level

characteristics measured at t¼�1: log employment, change in log employment, industry and region dummies, log wage, sex, age, skill-level and tenure.


For all three countries we find that separation rates decline after takeover, although the effect is only significant inBrazil. The decline in separation rates varies from 2.6% (Portugal) to 4.1% (Germany) to 6% (Brazil). The relatively limitedeffect of foreign takeovers on separations suggests that the changes in the composition of the workforce that tend to beassociated with foreign takeovers are largely driven by new hires and separations are relatively unimportant. Thesefindings are consistent with existing evidence for Germany from Andrews et al. (2012), who find no significant effect offoreign ownership on job security, and with theory which suggests that higher wages and lower separation rates reflectforeign-owned firms’ desire to reduce spillovers.

6.2. The effect of foreign ownership at the worker level

6.2.1. Cross-border takeovers

Table 7 presents new evidence of the effects of cross-border mergers and acquisitions on individual wages using linkedemployer–employee data for Brazil, Germany, Portugal and the United Kingdom. These results control for the role of workercomposition by focusing on individuals who stay in the same firm during the window of observation. Foreign takeovers of

Table 8The effects of worker mobility on individual wages.a


(a) From domestic to foreign firms

Average effect 0.062nnn 0.092nnn 0.068nnn 0.148nnn

(0.017) (0.007) (0.022) (0.016)

Effect at t¼0 – 0.089nnn 0.055nnn 0.157nnn

(0.007) (0.023) (0.016)


(0.008) (0.024) (0.017)


(0.008) (0.026) (0.017)

(b) From foreign to domestic firms

Average effect �0.029 0.018n�0.017 0.002

(0.016) (0.009) (0.040) (0.026)

Effect at t¼0 – 0.006 �0.020 0.042

(0.010) (0.041) (0.027)

Effect at t¼1 – 0.022nn�0.010 �0.007

(0.010) (0.043) (0.028)

Effect at t¼2 – 0.025nn�0.021 �0.028

(0.010) (0.044) (0.028)

nsignificant at 10%, nn significant at 5%, nnn significant at 1%.a Estimated using difference-in-difference propensity-score matching; see Eqs. (4) and (5). The treatment group in panel (a) are those workers who move

from a domestic firm to a foreign-owned firm; the control group are those workers who move between domestic firms. The treatment group in panel (b)

are those workers who move from a foreign-owned firm to a domestic firm; the control group are those workers who move between foreign-owned

firms. See Eq. (3). The propensity score is estimated using a Probit model which includes the same firm- and worker-level characteristics as used in

Table 7, measured at t¼�1.


domestic firms have only very small effects on the wages of workers who stay in the same firm, relative to similar workers whostay in domestic firms that are not taken over. More specifically, estimates vary from 1% in the UK to about 6% in Brazil.27

These estimates show that the wage effect of foreign takeovers is, in most cases, largely explained by workforcecomposition effects. By comparing the firm-level results with the worker-level results on workers who remain in the firm,we can see that for Portugal, the UK and Brazil the estimated wage effect is much smaller. Only in the case of Germany isthe wage effect approximately the same. The most likely explanation is that foreign takeovers of domestic firms tend to beassociated with significant increases in average skill-intensity (recall that employment effects were always positive forhigh-skill workers, as shown in the bottom panel of Table 5).

The time-profile of the effects of foreign takeovers on individual wages may differ across countries, although theprecision of our estimates is rather low.28 The effects of domestic takeovers of foreign firms on individual wages are smalland generally insignificantly different from zero.

6.2.2. Worker mobility

The effects of foreign ownership on the labour market are not only confined to those workers who remain within the samefirm. As we have seen, firm-level wage effects are typically larger than worker-level effects, and employment growth is typicallygreater for high-skilled employment. We are therefore also interested in identifying the effect of foreign ownership for individualswho join and who leave foreign-owned firms using the treatment indicator defined in (3). Note that the control group in this caseare workers who move between firms which have the same ownership status. The results are reported in Table 8.

We find significant wage gains for workers who move from domestic to foreign firms, but zero or even negative effects ofmoving from foreign to domestic firms. It seems plausible that wage gains reflect mostly voluntary worker movements (quits),whereas wage losses mostly reflect involuntary worker movements (layoffs). This is consistent with the notion that foreign-owned firms provide better wages, working conditions and career prospects than comparable domestic firms. The foreign-wagepremia associated with worker movements from domestic to foreign firms are also economically important and more so forworkers in less developed countries: in the order of 6% to 10% for the three developed countries and 16% in Brazil. For Portugal, UKand Brazil wage gains appear to be quite flat across t¼0,y,t¼2, suggesting that movement leads to a one-off gain in wages ratherthan an increase in the returns to tenure.

27 Using the same data but a somewhat different methodology Andrews et al. (2010) also find that foreign takeovers raise individual wages by 3% in

Germany. The results for Portugal differ somewhat from earlier results in Martins (2004) but are in the same range as those reported in Almeida (2007).

The differences with Martins (2004), whose study is the most similar to the present one in terms of methodology and set-up can be attributed to the fact

that the present analysis controls for lagged wages whereas Martins (2004) did not. The time period and sectoral coverage are also different.28 Although we have many more treated individual workers than firms, we use standard errors clustered at the firm-level.


The results on worker movements have a number of interesting implications. First, to the extent that substantial wagedifferences between similar workers within firms may be hard to sustain one would expect the positive effects of foreignownership to new hires to trickle down to the rest of the workforce with time. Second, the results on worker mobility suggest thatthe difference between the firm-level and worker-level estimates of the wage effects of foreign takeovers may not just be due tocompositional changes in the workforce associated with such takeovers, but also capture the greater importance of foreignownership for the wages of new hires as opposed to incumbent workers. As a result, the difference between the firm and worker-level estimates cannot be entirely attributed to composition bias, but part of it may reflect a genuine foreign ownership effect thatis not captured by the worker analysis of incumbents. Third, the worker mobility results provide some evidence that the humancapital that is accumulated in foreign firms can be transferred to domestic firms by workers who move from foreign to domesticfirms. Comparing the magnitude of wage gains associated with worker movements from domestic to foreign firms with the wagelosses associated with worker movements from foreign to domestic firms gives an idea of the extent to which worker mobilitymay be a potentially important channel for wage spillovers. To the extent that wage gains are not completely offset bycorresponding wage losses, workers may be able to carry with them some of the knowledge that they have accumulated inforeign firms. The results indicate that wage gains are considerably larger than wage losses in each of the four countries analysed.Thus, worker mobility could be an important channel for wage spillovers.29

6.2.3. Level comparisons between foreign- and domestically-owned firms

While identifying the impact of foreign ownership may be the most appropriate way to obtain unbiased estimates of thecausal effect of foreign ownership, it also has a number of limitations as it requires restricting the scope of the analysis to theshort-term effects of foreign ownership, and does not allow one to analyse the role of greenfield investments. This is a pity, sincethe effects of foreign ownership are often considered to become more important over time, and the effects of greenfieldinvestment may be quite different from those of cross-border takeovers. This subsection therefore concludes by discussing theresults from simple level comparisons for individual workers in domestic and foreign-owned firms that are presented in the toppanel of Table 7. These results indicate that level effects are fairly large even after controlling for observable characteristics. Thewage gap varies from 4% in Germany, 12% in the UK, 14% in Portugal and 29% in Brazil. While these estimates are likely to bebiased upward due to the roles of worker and firm selection, they are still interesting as they provide a useful upper bound on theaverage effect of foreign ownership by also capturing the effects of greenfield investments and the longer-term effects of foreignownership, which may be more positive than the short-term effects of cross-border takeovers alone.

7. Concluding remarks

Until recently, the empirical literature on the role of foreign ownership on wages was characterised by a consensus thatforeign-owned firms pay higher wages than domestic firms and that foreign-wage premia are particularly important in emergingeconomies. This consensus was, to a large extent, based on evidence that identifies the role of foreign ownership on averagewages by focusing on cross-border takeovers using longitudinal firm-level data. This paper provides consistent and systematiccross-country evidence which helps to sharpen and qualify the consensus in the literature.

First, this paper applies a standardised methodology to firm-level data for three developed countries and two large emergingeconomies. The results confirm that foreign-owned firms offer higher average wages than their domestic counterparts in all fivecountries. Moreover, consistent with the conventional wisdom, foreign-wage premia appear to be particularly important inemerging economies. Foreign-wage premia in the three developed countries are consistently below 10% (between 2% and 8%) andforeign-wage premia in the two emerging economies in the range of 15% to 20%. While there are enormous differences in scalebetween foreign and domestic firms, particularly in the two developing countries, there is a less systematic relationship betweenemployment growth and the foreign takeover of domestic firms. In some countries takeovers are associated with quite largeincreases in employment (Portugal, Indonesia) while in others (Germany, the UK and Brazil) there are no such scale effects. Wefind that employment effects of foreign takeovers are always positive for high-skill employment and usually negative for low-skillemployment.

Second, this paper shows that the conventional wisdom overstates the true wage premium, but also that it remainsqualitatively valid after controlling for composition bias. Using linked worker-firm data suggests that foreign wage premia aremuch smaller than previously believed: foreign-wage premia range from 1% to 6% in all the four countries analysed and, in somecases, the effect becomes insignificantly different from zero. Foreign-wage premia are still more important in emergingeconomies, although one should be careful in drawing strong conclusions as the worker-level analysis only involves one emergingeconomy (Brazil). The difference between the firm and worker level estimates of foreign wage premia indicates importantchanges in the composition of the workforce, which is largely consistent with the employment effects by skill-group. Thus, byfocusing on incumbent workers, the worker-level analysis excludes a key channel, perhaps even the main one, through whichforeign takeovers may affect workers.

Third, this paper shows that foreign-wage premia associated with worker movements from domestic to foreign firms may beeconomically important and that such movements may be more beneficial for workers in less developed countries (in the order of

29 However, the analysis is limited to the private returns to worker mobility. In order to analyse wage externalities, one would also have to analyse

the wage effects of worker mobility on incumbent workers.


6% to 10% for the three developed countries and over 16% in Brazil). The results on worker mobility suggest that the differencebetween the firm and worker-level estimates of foreign wage premia may not just be due to compositional changesin the workforce, but also capture the greater importance of foreign ownership for the wages of new hires as opposed toincumbent workers. Moreover, to the extent that substantial wage differences between similar workers within firms may be hardto sustain one would expect the positive effects of foreign ownership to new hires to trickle down to the rest of the workforcewith time.

The positive wage and employment effects of foreign takeovers do not appear to be compensated by a reduction of jobsecurity. Although some authors have argued that multinationals are inherently ‘‘footloose’’, or that jobs in multinationalsmay be less secure, our results indicate consistently small but negative effects on separation rates, consistent with theorieswhich stress the role of higher wages in preventing worker movement and associated spillover effects.

Acknowledgements

The authors would like to thank Andrew Clark, Stefano Scarpetta, Jordan Siegel and Paul Swaim, participants of the2008 CAED conference (Budapest), a research workshop of the Research Institute of Industrial Economics (Stockholm), andthe World Bank/IZA conference on Employment and Development (Rabat) and seminar participants at the HarvardBusiness School, the Paris School of Economics, Nihon University and the Cabinet Office of the Government of Japan, forhelpful comments and suggestions. The opinions expressed in this paper are those of the authors and do not necessarilyreflect those of the OECD or its member states. All remaining errors are our own.

Upward acknowledges financial support through the Leverhulme Trust (Grant Number F/00 114/AM).

Appendix A. Variable definitions

Table A1Variable definitions.


Foreign

ownership

More than 50% of assets

owned by a foreign entity

(establishment)



(firm)


owned by a foreign

entity (firm)



(firm)

More than 50% of

assets owned by a

foreign entity

(establishment)

Employment Log total number of

employees

Log total number of

employees

Log total number of

employees

Log total number of

employees

Log total number of

employees

Average

wage

Log of the average

individual wage

Log total wage bill divided

by employment

Log total wage bill

divided by employment

Log total wage bill divided

by employment

Log total wage bill

divided by

employment

Individual

wage

Log daily wage, censored at

the social security ceiling

Log hourly wage Log gross hourly wage Log hourly wage Not available

Worker

turnover

The number of worker

separations between t and

t�1 over total employment

at t�1

The number of worker



at t�1

Not available The number of worker



at t�1

Not available

Industry 15 categories Two-digit SIC codes One digit SIC92 codes

(9)

One-digit SIC codes (9) SIC codes

15–37 (23)

Region Bundeslander (11 regions) Regions (5) UK Government Office

Region (10)

States (27) Provinces (34)

Sex Indicator variable Indicator variable Indicator variable Indicator variable Not available

Age Age Age Age Age Not available

Skill Dummy for high, semi- and

low-skilled based on

highest educational

qualification

Based on education groups Dummy for high, semi-

and low-skilled based

on SOC2000 1-digit

categories

Based on education groups Production and

non-production

Tenure Number of years in current

establishment

Number of years in current

firm

Dummy for 4 1 year in

current position in firm

Number of years in current

firm

Not available


Appendix B. Matching and robustness results

Table B1Robustness tests.

Germany Portugal UK Brazil Indonesia

(a) Foreign takeovers of domestic firms

Baseline average effecta 0.020 0.078nnn 0.048n 0.147nn 0.189nnn

(0.015) (0.027) (0.025) (0.064) (0.046)

Matching with replacement 0.017 0.081nnn 0.045n 0.094n 0.163nnn

(0.015) (0.028) (0.026) (0.051) (0.046)

Caliper matching (max distance 0.05) 0.020 0.078nnn 0.046n 0.107nn 0.190nnn

(0.015) (0.027) (0.025) (0.052) (0.047)

(b) Domestic takeovers of foreign firms

Baseline average effecta 0.001 �0.022 �0.015 – �0.110

(0.029) (0.036) (0.057) (0.068)

Matching with replacement 0.007 �0.007 �0.015 – �0.185nn

(0.028) (0.039) (0.053) (0.073)

Caliper matching (max distance 0.05) 0.009 �0.022 �0.012 – �0.113

(0.029) (0.037) (0.051) (0.070)

(c) Stayers, foreign takeovers of domestic firms

Baseline average effectb 0.030n 0.034 0.002 0.057n –

(0.016) (0.031) (0.008) (0.033)

Matching with replacement 0.039n 0.024 0.002 0.039 –

(0.021) (0.031) (0.008) (0.028)

Caliper matching (max distance 0.05) 0.022nn 0.026 0.002 0.069n –

(0.011) (0.030) (0.008) (0.036)

(d) Stayers, domestic takeovers of foreign firms

Baseline average effectb 0.011 �0.025 0.031 – –

(0.011) (0.024) (0.026)

Matching with replacement 0.003 �0.016 0.031 – –

(0.022) (0.026) (0.026)

Caliper matching (max distance 0.05) �0.001 0.034 0.031 – –

(0.011) (0.026) (0.026)

(e) Movers from domestic to foreign firms

Baseline average effectc 0.062nnn 0.092nnn 0.068nnn 0.148nnn –

(0.017) (0.007) (0.022) (0.016)

Matching with replacement 0.064nnn 0.089nnn 0.063nnn 0.132nnn –

(0.017) (0.008) (0.023) (0.021)

Caliper matching (max distance 0.05) 0.062nnn 0.118nnn 0.068nnn 0.193nnn –

(0.017) (0.007) (0.022) (0.019)

(f) Movers from foreign to domestic firms

Baseline average effectc�0.029n 0.018n

�0.017 0.002 –

(0.016) (0.009) (0.040) (0.026)

Matching with replacement �0.035nn�0.005 �0.040 0.020 –

(0.017) (0.009) (0.041) (0.026)

Caliper matching (max distance 0.05) �0.029n 0.015 �0.031 0.014 –

(0.016) (0.009) (0.039) (0.031)

nsignificant at 10%, nn significant at 5%, nnn significant at 1%, standard errors clustered at the firm level.a The baseline results come from Table 4, panels (b) and (c).b The baseline results come from Table 7, panels (b) and (c).c The baseline results come from Table 8, panels (a) and (b).

Table B2Outcomes from matching procedures.

Treatment group Control group Germany Portugal United Kingdom Brazil Indonesia

Unmatch Match Unmatch Match Unmatch Match Unmatch Match Unmatch Match

Firms which switch

from domestic to foreign

Firms which stay domestic N (treated)a 37 36 185 155 850 524 105 87 793 702

N (untreated)a 1583 36 42,017 155 41,471 524 99,523 87 66,629 702

Mean biasb 20.15 12.77 22.6 8.5 21.5 8.5 53.6 15.9 12.9 4.9

t-testsc 3/21 0/21 46/114 2/114 65/280 1/280 39/114 1/114 65/190 2/190

Imbens diff.d – 0/21 – 21/114 – 12/280 44/114 – 0/190

Firms which switch

from foreign to domestic

Firms which stay foreign N (treated) 21 19 72 56 251 117 – 463 283

N (untreated) 153 19 1450 56 3432 117 10897 283

Mean bias 21.69 10.26 20.9 11.2 16.1 13.9 14.7 4.9

t-tests 2/16 0/16 12/57 1/57 17/120 0/120 33/126 0/126

Imbens diff. – 0/16 – 17/57 – 10/120 – 2/126

Workers who stay in a firm which

switches from domestic to foreign

Workers who stay in the same domestic firm N (treated) 11,976 11,976 11,910 9,348 1566 923 123,338 95,384 –

N (untreated) 139,857 11,976 952,076 9,348 56649 923 4,888,635 95,384

Mean bias 24.26 7.33 27.2 5.6 26.7 12.4 46.9 8.6

t-tests 27/29 23/29 295/386 83/386 119/335 11/335 516/622 209/622

Imbens diff. – 2/29 – 21/368 – 27/312 – 28/575

Workers who stay in a firm which

switches from foreign to domestic

Workers who stay in the same foreign firm N (treated) 3754 3611 4,640 3,947 104 61 – –

N (untreated) 34,975 3611 123,454 3,947 9222 61

Mean bias 20.79 6.13 26.5 18.1 27.8 13.5

t-tests 22/24 12/24 63/81 43/81 20/62 0/62

Imbens diff. – 0/24 – 30/78 – 11/58

Workers who move from

domestic to foreign firms

Workers who move between domestic firms N (treated) 341 341 17,936 10,808 722 500 7570 4181 –

N (untreated) 5657 341 183,813 10,808 6721 500 32,632 4181

Mean bias 35.8 8.61 14.6 7.8 12.7 5.6 29.4 15.1

t-tests 22/30 3/30 398/677 109/677 105/217 3/217 367/731 73/731

Imbens diff. – 0/30 – 34/659 – 1/212 – 75/683

Workers who move from

foreign to domestic firms

Workers who move between foreign firms N (treated) 313 313 12,742 9,148 443 273 5817 1714 –

N (untreated) 5657 313 183,813 9,148 6721 273 32,632 1714

Mean bias 34.46 7.51 18.6 4.9 12.1 7.4 36.3 15.0

t-tests 21/30 4/30 373/558 31/558 68/210 5/210 271/622 21/622

Imbens diff. – 0/30 – 4/540 – 1/205 – 90/576

a Number of observations (N) in the treatment and control groups before and after matching.b Mean standardised bias before and after matching, see Rosenbaum and Rubin (1985).c Number of t-tests where the p-value on the difference in the mean is less than 0.05.d Following Imbens (2009), we report the number of cases in which the standardised difference in means is greater than 0.25 standard deviations.

A.

Hijzen

eta

l./

Eu

rop

ean

Eco

no

mic

Rev

iew6

0(2

01

3)

17

0–

18

81

87


References

Aitken, B., Harrison, A., Lipsey, R., 1996. Wages and foreign ownership: a comparative study of Mexico, Venezuela and the United States. Journal ofInternational Economics 40 (3–4), 345–371.

Almeida, R., 2007. The labor market effects of foreign owned firms. Journal of International Economics 72 (1), 75–96.Andrews, M., Bellmann, L., Schank, T., Upward, R., 2010. The takeover and selection effects of foreign ownership in Germany: an analysis using linked

worker-firm data. Review of World Economics/Weltwirtschaftliches Archiv 145 (2), 293–317.Andrews, M., Bellmann, L., Schank, T., Upward, R., 2012. Foreign-owned plants and job security. Review of World Economics 148 (1), 89–117.Balsvik, R., 2011. Is mobility of labour a channel for spillovers from multinational to local domestic firms? Review of Economics and Statistics 93 (1),

285–297.Beaudry, P., DiNardo, J., 1991. The effect of implicit contracts on the movement of wages over the business cycle: evidence from micro data. Journal of

Political Economy 99 (4), 665–688.Decreuse, B., Maarek, P., 2008. Foreign Direct investment and the labour share in developing countries. GTRQAM Document de Travail, No. 2008-17.Egger, H., Kreickemeier, U., 2013. Why foreign ownership may be good for you. International Economic 54, 693–716.Fabbri, F., Haskel, J., Slaughter, J., 2003. Does nationality of ownership matter for labor demands? Journal of the European Economic Association 1 (2),

698–707.Fosfuri, A., Motta, M., Ronde, T., 2001. Foreign direct investment and spillovers through workers’ mobility. Journal of International Economics 53 (1),

205–222.Girma, S., Gorg, H., 2007. Evaluating the foreign ownership wage premium using a difference-in-differences matching approach. Journal of International

Economics 72 (1), 97–112.Glass, A., Saggi, K., 2002. Multinational firms and technology transfer. Scandinavian Journal of Economics 104 (4), 495–513.Gorg, H., Strobl, E., 2003. Footloose multinationals. The Manchester School 71 (1), 1–19.Heckman, J., Ichimura, H., Todd, P., 1997. Matching as an econometric evaluation estimator: evidence for evaluating a job training programme. The

Review of Economics and Statistics 64, 605–654.Helpman, E., Melitz, M., Yeaple, S., 2004. Export versus FDI with heterogeneous firms. American Economic Review 94 (1), 300–316.Heyman, F., Sjoholm, F., Gustavsson Tingvall, P., 2007. Is there really a foreign ownership wage premium? Evidence from matched employer–employee

data. Journal of International Economics 73 (2), 355–376.Hijzen, A., Martins, P., Schank, T., Upward, R., 2010. Do foreign-owned firms provide better working conditions than their domestic counterparts? A

comparative analysis’’, IZA working paper 5259.Huttunen, K., 2007. The effect of foreign acquisition on employment and wages: evidence from Finnish establishments. The Review of Economics and

Statistics 89 (3), 497–509.ILO, 2000. Tripartite declaration of principles concerning multinational enterprises and social policy, Geneva: International Labour Office.Imbens, G., 2009. Estimation of treatment effects under unconfoundedness: Part 1. In: Imbens, G., Wooldridge, J. (Eds.), New Developments in

Econometrics: Lecture Notes, CEMMAP.Lipsey, R., Sjoholm, F., 2004. Foreign direct investment, education and wages in Indonesian manufacturing. Journal of Development Economics 73 (1),

415–422.Lipsey, R., Sjoholm, F., 2006. Foreign firms and Indonesian manufacturing wages: an analysis with panel data. Economic Development and Cultural

Change 55 (1), 201–221.Malchow-Moller, N., Markusen, J., Schjerning, B. Foreign firms, domestic wages. Scandinavian Journal of Economics, forthcoming.Martins, P., 2004. Do foreign firms really pay higher wages? Evidence from different estimators, IZA Discussion Paper 1388.Martins, P., 2011. Paying more to hire the best? Foreign firms, wages and worker mobility. Economic Inquiry 49 (2), 349–363.Moran, T., 2006. Harnessing Foreign Direct INvestment for Development: Policies for Developed and Developing Countries. Center for Global

Development, Washington, DC.Morrissey, O., Te Velde, D., 2003. Do workers in Africa get a wage premium if employed in firms owned by foreigners? Journal of African Economies

12 (1), 41–73.OECD, 2008. Do multinational promote better pay and working conditions? In: OECD Employment Outlook 2008, Paris: OECD (Chapter 5).OECD, 2011. Guidelines for multinational enterprises. Available online at /http://www.oecd.org/S.Pissarides, C., 2009. The unemployment volatility puzzle: is wage stickiness the answer? Econometrica 77 (5), 1339–1369.Porter, M., Kramer, M., 2006. Strategy and society: the link between competitive advantage and corporate social responsibility. Harvard Business Review

84 (12), 78–92.Porter, M., Kramer, M., 2011. Creating shared value. Harvard Business Review 89 (1), 1–17.Rosenbaum, P., Rubin, D., 1985. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. The

American Statistician 39 (1), 33–38.Vilhuber, L., 2008. Adjusting imperfect data: overview and case studies. In: Lazear, E. and Shaw, K. (Eds.), The structure of wages: an international

comparison, Chicago: University of Chicago Press.

http://www.oecd.org/

foreign-owned firms around the world: a comparative analysis of wages and employment at the...

Documents