Immigrants and Employer-provided Training

Alan Barrett & Séamus McGuinness &

Martin O’Brien & Philip O’Connell

Published online: 3 July 2012# Springer Science+Business Media, LLC 2012

Abstract Much has been written about the labour market outcomes for immigrantsin their host countries, particularly with regard to earnings, employment and occupa-tional attainment. However, much less attention has been paid to the question ofwhether immigrants are as likely to receive employer-provided training relative tocomparable natives. As such training should be crucial in determining the labourmarket success of immigrants in the long run it is a critically important question.Using data from a large-scale survey of employees in Ireland, we find that immigrantsare less likely to receive training from employers, with immigrants from the NewMember States of the EU experiencing a particular disadvantage. The immigranttraining disadvantage arises in part from a failure on the part of immigrants to getemployed by training-oriented firms. However, they also experience a training dis-advantage relative to natives within firms where less training is provided.

Keywords Immigrants . Employer-provided training . Ireland

JEL Classification J24 . J61

Introduction

Much has been written about the labour market outcomes of immigrants acrosscountries and over time. Studies often find that immigrants experience labour marketdisadvantages relative to natives soon after arrival in their host countries (Chiswick1978). Such disadvantages are seen through lower wages or higher unemployment

J Labor Res (2013) 34:52–78DOI 10.1007/s12122-012-9148-7

A. Barrett (*) : S. McGuinness : P. O’ConnellEconomic and Social Research Institute, Whitaker Square, Sir John Rogerson’s Quay, Dublin 2,Irelande-mail: Alan.Barrett@esri.ie

M. O’BrienCentral Bank of Ireland (CBI), Dublin 2, Ireland

propensities for immigrants, usually controlling for relevant labour-market character-istics of individuals. Some studies show reductions in the relative disadvantage ofimmigrants as they spend longer in their host countries (Chiswick et al. 2005). Otherstudies show persistence in disadvantages (Borjas 1985).

Other aspects of the experience of immigrants in their host country labour marketshave generally received less attention. In this paper, we aim to contribute to address-ing this by conducting an analysis of the extent to which immigrants receiveemployer-provided training, relative to comparable natives. There are a number ofreasons why the receipt of employer-provided training could be critical for immi-grants and, hence, why any gap in receipt relative to natives could be damaging fortheir labour market prospects. First, in the case of immigrants from countries whereeducational attainment is low or of poor quality, training in the host country may offeran opportunity to compensate in part for any educational disadvantages facingindividual immigrants. Second, even in the case of highly qualified immigrants, skillsacquired in the home country may not be readily transferable to the host country(Friedberg 2000). In such cases, employer-provided training may be important inproviding the necessary location-specific human capital or in allowing the immigrantto adapt their existing skills to the new environment.

With regard to more theoretical frameworks, human capital theory (Becker1964), predicts that employee’s will tend to bear the cost of general training, whileemployers are more likely to fund job specific training on the grounds that they willaccrue a sufficient return over time, in the form of enhanced employee productivity,to cover the costs of their investments. Becker contends that most training is somecombination of both general and specific. On the grounds that immigrant workers arelikely to be perceived as having a lesser tendency to remain within the firm,1

employers may be reluctant to fund job-specific training among immigrants due toincreased uncertainty with respect to fully recouping their training costs. This mayresult in immigrants receiving less training relative to natives within the firm, or beingcorralled into firms with lower levels of training intensity.2 It may also, of course, bethe case that immigrants self-select into low-training intensive firms, or fail to putthemselves forward for training when opportunities arise. Within our data it is notpossible to distinguish the extent to which lower levels of training are driven bystrategic behaviour on the part of firms or self-selection on the part of employees.Nevertheless, given that the wage benefits associated with training are well knownand that self-selection into low-training intensive firms represents a sub-optimalstrategy on the part of the worker, it is probably fair to conclude that any observedimmigrant training disadvantage will be primarily driven by employer-based strategicbehaviour.

A further element in the importance of employer-provided training for immigrantsarises from the literature on return migration. Co et al. (2000) and Barrett andO’Connell (2001a) provide examples of situations in which returned migrants earna wage premium relative to people who have never worked outside of their homecountries. In the case of both papers, the reason for the wage premium is linked to

1 For example, as return migration is a common phenomenon, employers may think that there is asignificant probability that immigrant workers will eventually return home.2 We go on to explicitly test for the relative importance of these two effects.

J Labor Res (2013) 34:52–78 53

human capital formation while in the host country and the rewarding of this humancapital in the home country, following a return. If true, this portrayal of circularmigration holds out the prospect of advantages for immigrants and for their homecountries. However, if immigrants are excluded from employer-provided training,this will clearly have negative implications in terms of human capital formation.

The paper is structured as follows. In “Literature”, we provide a brief overview ofthe literature on the question of which employees are more likely to receiveemployer-provided training. We will also provide details of the limited work onimmigrants and training. In “The Data”, we present details of the data that we useto analyse our core question, namely, are immigrants more or less likely to receiveemployer-provided training. In “Results”, we present our results. As will be seen, wedo find evidence of a lower incidence of training among immigrants, with interestingpatterns emerging across different groups of immigrants. “Conclusions” contains ourconclusions.

Literature

O’Connell and Junblut (2008) in their review of the international literature, argue thatparticipation in training at work is highly stratified. Those with higher skills oreducational attainment are more likely to participate in training, includingemployer-provided training. Larger firms and those that pay higher wages are alsomore likely to train their employees. Part-time workers, those on temporary contractsand older workers are less likely to receive training. O’Connell and Jungblut thusargue that those with the greatest need for training tend to receive less of it,presumably to their long-run disadvantage.

The dominant theoretical framework informing most research on training has beenthe human capital approach. This approach, deriving from Becker (1964), assumesthat individual workers undertake training, and employers invest in training, on thebasis of their estimates of future returns (including employment prospects, wages andproductivity gains). The human capital approach emphasises the distinction between“general” training—of use to both current and future employers—and “specific”training, linked closely to the current job and of use only to the current employer.In this approach it is expected that employers will not pay for general training,because they cannot recoup the cost—other employers would be free to “poach”trained employees and reap the benefits of enhanced productivity. If, as a result of thismarket failure, employees have to pay the full cost of general training—whetherdirectly or through reduced wages—it is likely that there will be under-investment intraining.

However, empirical evidence tends not to support this hypothesis. Theempirical literature has found that the theoretical distinction is difficult tooperationalise and that many employers pay for both general and specific training.O’Connell and Junblut (2008) summarise findings from Germany, Ireland, Sweden,the UK and the US that show that the vast majority of job-related training appears tobe employer paid, at least partially, even though most training appears to be general ortransferrable in nature. Thus, for example, Booth and Bryan (2002) found that 85 %of respondents to the British Household Panel Survey considered their training to be

54 J Labor Res (2013) 34:52–78

general and 89 % reported that it had been financed by their employer. O’Connell andByrne (2012) show that 80 % of all employer-sponsored training received byrespondents to a survey of Irish employees was general in nature, considered to be“of use in getting a job with another employer”.

A growing stream of work challenges the human capital approach, focussing inparticular on its key assumption that labour markets are perfectly competitive. Ifmarkets were perfectly competitive, workers with training would be more likely tomove, and all workers would be treated similarly. One stream of work points to theimportance of institutions—such as unions and labour market regulation(e.g. Acemoglu and Pischke 1999). Booth et al. (2003) show that unions can have apositive or negative impact on training depending on union strategy and the mannerin which unions negotiate on training and wages.

For most workers in advanced industrial societies the employment contractentails an ongoing and relatively long-term relationship that may differ inimportant respects from the competitive labour market assumed in the Beckermodel (Barrett and O’Connell 2001b). This turns our attention to the demand sideof the labour market and to the social organisation of work. The segmented labourmarket approach focuses more on the characteristics of jobs rather than individualsand argues that different labour market sectors impose structural limitations both onthe returns to education and experience and on the career prospects of workers(Doeringer and Piore 1971; Gordon et al. 1982). At their simplest, labour marketsegments can be dichotomised, with the primary market consisting of well paid andsecure employment as opposed to jobs in the secondary market which are poorly paidand are of a precarious nature with few or no prospects for upward mobility. From thisperspective, workers in the secondary labour market are less likely to participate in job-related training, and the returns to such training are lower. Workers in the primarysegment(s) are more likely to receive training, the returns are higher, and such training islikely to be associated with upward mobility, perhaps in an internal labour market.

We now turn to the literature that looks specifically at immigrants and their receiptof training relative to natives. We should firstly ask what theory will predict on theissue but in truth theory can point in either direction. As the human capital modelviews the issue of training in the context of the costs and benefits to both employeesand employers, this allows us to identify some critical issues. For example, ifimmigrants see their stay in the host country as being temporary, they will be lessinclined to invest in training which is specific in nature as the pay-off period will berestricted. Similarly, if employers believe that immigrant workers are more likely toleave, they will be less inclined to provide training. Alternatively, if training allowsimmigrants to acquire location-specific human capital that they are lacking, then thepay-off can be high.

If we are thinking in terms of segmented labour market theories (Doeringer andPiore 1971), then predictions become more clear-cut. Under this theory, secondarylabour markets are characterised by jobs which offer lower wages, fewer fringebenefits and poorer working conditions, including a lower incidence of training.If such secondary markets exist and if immigrants are more heavily concen-trated in them then immigrants will be less likely to be in receipt of training.No studies of immigrants in Ireland have attempted to determine directly if theyare employed in secondary labour markets. However, results from work on the

J Labor Res (2013) 34:52–78 55

earnings of immigrants relative to natives (for example, Barrett and McCarthy 2007)and on the labour market integration of immigrants (Barrett and Duffy 2008) wouldcertainly be consistent with a scenario of some immigrants in Ireland being insecondary labour markets, especially immigrants from the New Member States ofthe EU.

The empirical literature on the question of the relative receipt of employer-provided training across immigrants and natives is thin, but three papers fromAustralia (Miller 1994; Kennedy et al. 1994; Van den Heuvel and Wooden 1997)point firmly in the direction of immigrants being less likely to receive training.This is especially true for immigrant groups with different linguistic back-grounds to the native population. Taking Van den Heuvel and Wooden (1997)as an example, they explore the issue using Australian data from 1993. By estimatinglogit regressions of training receipt, they show how immigrants in Australia fromnon-English speaking countries have lower incidences of training and that the gaprelative to natives is more pronounced among immigrants from these countries whoalso report having weak English language skills. This outcome could emerge, forexample, because a given amount of training could provide better outcomes fornatives if immigrants cannot follow instructions due to a language barrier.3

Hum and Simpson (2003) look at the issue in Canada and also find a lowerincidence of training among immigrants relative to natives. They go on to explorehow training receipt among immigrants differs according to age at arrival in Canadaand show that training decreases with age at arrival. Shields and Wheatley Price(1999a, b)) look at differences by ethnic group, as opposed to immigrants, in the UK.They show that members of ethnic minorities are less likely to receive training, onaverage. One third of the difference can be explained by the characteristics of theindividuals (in the case of men) but the remaining two-thirds cannot be explained bycharacteristics.

The Data

The data used in this study comes from Ireland’s National Employment Survey (NES)from March 2006. The 2006 NES is a workplace survey, covering both the public andprivate sectors, which was conducted by Ireland’s official statistical office, theCentral Statistics Office (CSO).4 The information contained in the NES was collectedfrom a matched employee-employer survey. The employer sample was drawn usingthe CSO Central Business Register (CBR). Selected firms were asked to draw asystematic sample of employees from their payroll. Approximately 8,000 enterprises5

were contacted of which 4,845 responded resulting in employee information on67,766 individuals. After the elimination of employees with information missing onsome variables, part-time students and also the restriction of our sample to those of

3 We are grateful to the Editor for pointing this out.4 While the NES survey was of enterprises with 3 plus employees, the results were calibrated to theQuarterly National Household Survey (QNHS) employment data for employees (excluding agriculture,forestry and fishing), which covers all employees.5 Only employers with more than three employees were surveyed and the data were collected at theenterprise level.

56 J Labor Res (2013) 34:52–78

standard working age, the final sample for this study was just over 50,000employees. When analysing the employee sample, cross-sectional weights wereapplied to ensure that the data was representative of the general population ofemployees in employment. The application of cross-sectional weights ensuresthat the data structure reflects that of the Irish labour market in terms of bothemployment composition and firm size. However, we do adjust the standarderrors within our models to allow for the effects of clustering within firms. Ouranalysis relates to employees in both the public and private sectors given thatthere are no nationality restrictions to public sector employment in Ireland.

The employer questionnaire collected some limited information on firm size andsector that was incorporated into our models. Employees were issued with a separatesurvey within which they were asked to provide information on their age, gender,educational attainment, employment status (part-time or full-time), length of time inpaid employment and also other job-related characteristics (for example, trade unionmembership, supervisory role, tenure with current employer).

Employees were also asked whether or not they had received employer-provided training during the course of the preceding year, 2005. They wereasked a broad question about whether or not they had participated in any“company or company sponsored training courses in 2005”. They were alsoasked a more detailed question about types of training, such as training throughparticipation in quality circles, job rotation etc. If an individual respondedpositively to any of the questions about participation in company training,including the broad question and the more detailed questions, we coded thatperson has having received employer-provided training. This gives us the“broad training” binary dependent variable used in the bulk of the analysis.However, we also examine the extent to which immigrants experience differ-ential rates of access to specific forms of training such as on-the-job, subsidisedself-directed learning, study-circles, conference and workshop attendance etc.

In terms of migration, each individual’s country of origin was coded in a verydetailed way that allowed us to separate out migrants into UK, Pre-accessionEU (other than UK), New Member States of the EU (hereafter referred to asNMS),6 non-EU/English speaking and non-EU/non English speaking. The fact thatthe survey was conducted in March 2006 but that the training questions related to theyear 2005 generated a concern that many immigrants would not have receivedtraining because they were very recent arrivals. In order to account for this, weexcluded people who had entered zero in their response to the question on tenureand so we are only looking at individuals, both natives and migrants, with at least oneyear of tenure with their current employer. This restriction also has the advantage ofremoving any confusion surrounding individuals who changed jobs in early 2006 andwho have answered the questions on training in 2005 with reference to their newemployer. We should note at this point that we do not have information on the year inwhich immigrants arrived in Ireland. For many studies of immigrants this would be asignificant problem but it is less of a problem here. Our focus is on immigrants fromthe NMS and the vast majority of these immigrants arrived in Ireland in recent years,

6 This refers to the 10 countries which joined the EU on 1 May 2004. Citizens of these countries had fullaccess to Ireland’s labour market from the date of accession.

J Labor Res (2013) 34:52–78 57

especially after EU Accession in May 2004. The number of nationals from the NMSgrew from under 25,000 in 2002 to over 120,000 in 2006. Hence, they will be quitehomogeneous in terms of time spent in Ireland.

Results

In Table 1, we present some descriptive statistics for the native and migrant groupingsin our data.7 After natives, immigrants from the UK made up the largest singlegrouping in 2006 with an employment share of 3 %; these were followed by NMSmigrants and those from non-EU non-English speaking countries (2 % respectively).With employment shares of just 1 %, EU 13 and non-EU/English speaking immi-grants represented the smallest groupings. As expected, tenure levels were lowestamong immigrants, with those from the new member states having the lowest of all.The average tenure level of 2.62 years among NMS immigrants confirms the viewthat the vast bulk of these individuals entered Ireland shortly following the accessiondate. Regarding education, relative to natives, immigrants from EU 13 countries weremore likely to be educated to tertiary level while NMS immigrants were more likelyto be educated to post-secondary level. With respect to sectoral location, NMSimmigrants were more heavily concentrated in the manufacturing, and wholesale\retailsectors, while those from EU 13 countries were disproportionately concentratedin manufacturing and financial intermediation.

There are also some marked differences in terms of the distribution of training.With regard to broad training (which captures whether receipt of any form of trainingwas indicated on the questionnaire), the most striking point is the notably lowerincidence among immigrants from the immigrants from the NMS. While 60 % ofnatives received some form of training, only 41 % of accession state nationals were inreceipt. Relative to natives, immigrants from non-EU/English speaking countrieswere more likely to receive on-the-job training, to participate in learning circles\self-directed learning and to attend conferences. The higher incidence of trainingamong the non-EU non-English speaking grouping is most likely a consequence oftheir increased presence within professional occupations. Conversely, MNS immi-grants were much less likely to attend conferences and also appear somewhat at adisadvantage with respect to both on-the-job training and self-directed learning.Variations with regard to the remaining migrant groupings are less obvious. However,there is obviously a need to look at this issue in a multivariate context so that we candistil whether or not these relatively small differences persist when controlling forother characteristics. We know from a series of papers, and the descriptive statisticspresented here, that immigrants in Ireland differ from natives in terms of their levelsof educational attainment (see, for example, Barrett and Duffy 2008). For this reason,we might have expected them to have a somewhat different incidence of training. Inaddition, we also know that immigrants from different regions have very differentlabour market experiences (Barrett and McCarthy 2007). Hence, when undertakingmultivariate analyses, we need to look across different immigrant groups as well aslooking at immigrants as a whole.

7 Fuller descriptions of the variables are given in Appendix Table 15.

58 J Labor Res (2013) 34:52–78

Table 1 Raw probabilities

All Native UK EU13 NMS Non-EUEng Speak

Non-EU/nonEng speaking

Male 0.52 0.52 0.57 0.52 0.62 0.50 0.57

Tenure 10.37 10.83 7.61 4.56 2.62 6.00 5.18

Supervisory role 0.34 0.34 0.39 0.30 0.16 0.38 0.27

Primary education 0.07 0.08 0.06 0.06 0.08 0.02 0.09

Secondary 0.36 0.37 0.31 0.19 0.28 0.20 0.23

Posts-secondaryeducation

0.11 0.10 0.14 0.11 0.27 0.07 0.09

Tertiary education 0.36 0.36 0.36 0.41 0.29 0.52 0.47

Post graduate 0.10 0.10 0.13 0.24 0.08 0.19 0.12

Experience 18.04 18.53 19.89 10.69 7.36 14.52 8.72

Experience squared 462.6 480.0 522.14 190.3 110.32 308.50 142.06

Full-time 0.87 0.87 0.86 0.92 0.95 0.92 0.88

Professional body 0.16 0.16 0.18 0.07 0.03 0.23 0.12

Union 0.33 0.34 0.25 0.12 0.07 0.23 0.19

Public Sector 0.25 0.26 0.19 0.09 0.01 0.22 0.10

firm size 5.09 5.14 4.72 4.66 3.99 5.03 4.62

Broad Training .60 .60 .60 .60 .42 .70 .58

On-the-job training 0.34 0.34 0.34 0.35 0.29 0.42 0.37

Study visit 0.07 0.06 0.06 0.07 0.08 0.06 0.11

Learning circle 0.13 0.12 0.13 0.12 0.10 0.16 0.17

Self-directed 0.11 0.11 0.10 0.11 0.05 0.18 0.12

conference 0.33 0.34 0.34 0.30 0.10 0.48 0.28

Irish Native 0.91 1.00 0.00 0.00 0.00 0.00 0.00

UK 0.03 0.00 1.00 0.00 0.00 0.00 0.00

EU 13 0.01 0.00 0.00 1.00 0.00 0.00 0.00

NMS 0.02 0.00 0.00 0.00 1.00 0.00 0.00

Non-EU/EnglishSpeaking

0.01 0.00 0.00 0.00 0.00 1.00 0.00

Non-EU/Non-EnglishSpeaking

0.02 0.00 0.00 0.00 0.00 0.00 1.00

Manufacturing 0.18 0.17 0.17 0.25 0.28 0.13 0.15

Electricity, Gas & Water 0.05 0.05 0.05 0.04 0.09 0.03 0.03

Construction 0.13 0.14 0.12 0.06 0.15 0.08 0.11

Wholesale & Retail 0.05 0.04 0.06 0.09 0.17 0.08 0.25

Hotels & Restaurants 0.07 0.07 0.08 0.10 0.10 0.07 0.03

Transport, storage &communications

0.06 0.07 0.05 0.06 0.02 0.05 0.01

Financial Intermediation 0.12 0.12 0.17 0.22 0.12 0.18 0.14

Business Services 0.11 0.12 0.06 0.02 0.00 0.10 0.02

Public Admin & Defence 0.08 0.08 0.08 0.07 0.02 0.08 0.04

Education 0.11 0.11 0.12 0.07 0.02 0.17 0.20

Health & Social Work 0.03 0.03 0.03 0.01 0.03 0.03 0.03

J Labor Res (2013) 34:52–78 59

Within the paper we estimate the following specifications:

Ptrain ¼ ai þ bxi þ dImi þ "i ð1Þ

Ptrain ¼ ai þ bxi þ dImi þ fsci þ "i ð2Þ

Ptrain ¼ ai þ bxi þ dImi þ lxi � Imi þ fsci þ "i ð3Þwhere Ptrain is the probability of receiving broad training,8 xi is a vector of personalcharacteristics related to individual human capital and job history, Imi relates toimmigrant status, sc is a vector of controls for occupation and sector9 while Imi*xirepresents a series of interactions between immigrant status and certain individualcharacteristics. In Table 2, we report the results from probit regressions in which thedependent variable is equal to one if the individuals reports that they received someform of training in 2005 and zero if not. The models are estimated on the combinedsample initially before being re-estimated separately for males and females. Beforelooking at the immigrant coefficients, we will briefly consider the other coefficientsto see if the models are producing results that are generally consistent with whatwould be expected.

Looking at Model 1 in Table 2, we can see that the pattern of results is essentiallywhat would be expected based on the existing literature. Tenure and experience havea negative impact on training receipt; this is consistent with a human capital viewwhich would see reduced pay-off periods as lowering the potential value of additionaltraining. Full-time employees, members of professional bodies and union membersare all shown to have a greater likelihood of receiving training. Similarly, employeesin larger companies are also more likely to receive training.

8 This relates to participating in broad training and/or any of five other measures of training.9 Both sector and occupation were controlled for by the inclusion of dummy variables generated using 1digit codes based on the NACE (sector) and UKSOC (occupation) classification systems.

Table 1 (continued)

All Native UK EU13 NMS Non-EUEng Speak

Non-EU/nonEng speaking

Managers & admin 0.09 0.09 0.12 0.08 0.02 0.11 0.03

Professional 0.21 0.21 0.22 0.26 0.06 0.33 0.21

Associate Professional 0.10 0.10 0.11 0.13 0.03 0.14 0.09

Clerical & secretarial 0.21 0.22 0.17 0.25 0.09 0.20 0.11

Craft & related 0.07 0.07 0.05 0.02 0.15 0.02 0.06

Personal & Protective Services 0.06 0.06 0.05 0.04 0.07 0.03 0.12

Sales 0.06 0.06 0.06 0.04 0.06 0.05 0.06

Plant & machine operatives 0.10 0.09 0.11 0.07 0.22 0.05 0.09

Other 0.10 0.09 0.11 0.12 0.31 0.08 0.23

N 50591 46273 1366 633 831 333 1155

60 J Labor Res (2013) 34:52–78

Turning to the marginal effect of the immigrant dummy, we see that it isnegative and significant. In terms of a marginal impact, immigrants are 4percentage points less likely to receive employer-provided training. While thisresult is interesting to a degree, earlier research on immigrants in Ireland hasshown that the labour market outcomes for different immigrant groups varyconsiderably.10 For this reason, it is important to look separately at the differentgroups and we do this in Model 2. Once we do this we see the different outcomesemerging. The marginal effects that are significant across the five immigrant

10 For example, see Barrett and McCarthy (2007) and Barrett et al. (2011) on earnings and Barrett andDuffy (2008) on occupational attainment.

Table 2 Probit regressions broad training: marginal effects

Variables (1) (2) (3) (4) (5)

Model 1 Model 2 Model 3 Males Females

Tenure −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.01*** (0.00)

Supervise 0.18*** (0.01) 0.18*** (0.01) 0.16*** (0.01) 0.16*** (0.01) 0.16*** (0.01)

Secondary 0.08*** (0.01) 0.08*** (0.01) 0.06*** (0.01) 0.06*** (0.01) 0.07*** (0.02)

Post secondary 0.11*** (0.01) 0.12*** (0.01) 0.10*** (0.01) 0.10*** (0.02) 0.10*** (0.02)

Tertiary 0.22*** (0.01) 0.22*** (0.01) 0.16*** (0.01) 0.15*** (0.02) 0.18*** (0.02)

Post graduate 0.23*** (0.01) 0.23*** (0.01) 0.17*** (0.01) 0.16*** (0.02) 0.19*** (0.02)

Experience −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.01*** (0.00) −0.00* (0.00)

Experiencesquared

0.00 (0.00) 0.00* (0.00) 0.00** (0.00) 0.00** (0.00) 0.00 (0.00)

Full time 0.10*** (0.01) 0.10*** (0.01) 0.09*** (0.01) 0.12*** (0.02) 0.08*** (0.01)

Professionalbody

0.19*** (0.01) 0.19*** (0.01) 0.16*** (0.01) 0.15*** (0.01) 0.18*** (0.01)

Male 0.02* (0.01) 0.02* (0.01) 0.03*** (0.01)

Union 0.07*** (0.01) 0.07*** (0.01) 0.07*** (0.01) 0.07*** (0.02) 0.06*** (0.01)

Public 0.00 (0.02) 0.00 (0.02) −0.15*** (0.04) −0.15** (0.07) −0.16*** (0.03)

Firm size 0.05*** (0.00) 0.05*** (0.00) 0.04*** (0.00) 0.04*** (0.01) 0.05*** (0.00)

UK −0.01 (0.01) −0.01 (0.01) −0.01 (0.02) 0.00 (0.02)

EU 13 −0.03 (0.04) −0.03 (0.04) −0.05 (0.06) −0.01 (0.03)

NMS −0.13*** (0.02) −0.09*** (0.02) −0.10*** (0.03) −0.07** (0.03)

Non-EU/Englishspeak

0.03 (0.03) 0.02 (0.03) 0.02 (0.04) 0.03 (0.04)

Non-EU/Non-English speak

−0.05** (0.02) −0.02 (0.02) −0.02 (0.03) −0.02 (0.03)

Immigrants - all −0.04*** (0.01)

Observations 49420 49420 49420 25719 23701

PseudoR-squared

0.141 0.141 0.160 0.136 0.191

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbeside coefficient estimates. Model 3 and both gender based models contain controls for sector andoccupations

J Labor Res (2013) 34:52–78 61

categories are those of the NMS and non-EU/non-English-speaking countries. In addi-tion, the marginal impact for the NMS group, at 13 percentage points, is substan-tially higher than that estimated for immigrants as a whole. In Model 3, we addcontrols for occupation and sector to explore the possibility that our observa-tions arise from a concentration of NMS immigrants in low-training occupationsand sectors. While both the estimated coefficient and marginal impact for NMSimmigrants fall when these controls are added, we are still observing a sig-nificant gap between natives and NMS immigrants in terms of receipt ofemployer-provided training. In the case of immigrants from non-EU/non-Englishspeaking countries, the estimated coefficient is no longer significant in Model 3.When the models are estimated separately for each gender we observe that thetraining penalty is again significant and substantial for both sexes. However, the levelof disadvantage is somewhat higher for male NMS migrants.

As stated earlier, although unlikely, it is possible that immigrant workers may beless disposed to training and, as such, may self-select into jobs (or firms) that require(or offer) less training. As our analysis stands, our models do not account for suchunobserved traits across firms. To overcome this problem, we estimate a fixed effectsmodel including over 4,500 individual firm level dummies, thus ensuring that ourestimates are robust to any unobserved heterogeneity with respect to firm levelbehaviour. We re-estimate model 3 from Table 2 using this format and the results

Table 3 Fixed effects probitregressions broad training:marginal effects

* implies significance at the10 % level; ** implies 5 % and*** implies 1 %; standard errorsare shown beside coefficientestimates. Model 3 and bothgender based models containcontrols for sector and occupa-tions in addition to over 4,500individual firm level dummys

Variables (3)

Model 3

Tenure −0.00*** (0.00)

Supervise 0.18*** (0.01)

Secondary 0.07*** (0.01)

Post secondary 0.8*** (0.01)

Tertiary 0.13*** (0.01)

Post graduate 0.13*** (0.01)

Experience −0.00*** (0.00)

Experience squared 0.00** (0.00)

Full time 0.10*** (0.01)

Professional body 0.16*** (0.01)

Male 0.04*** (0.01)

Union 0.07*** (0.01)

Public −0.06 (0.60)

Firm size 0.10 (0.12)

UK −0.01 (0.01)

EU 13 −0.03 (0.04)

NMS −0.05*** (0.02)

Non-EU/English speak 0.04 (0.03)

Non-EU/Non- English speak −0.01 (0.02)

Observations 49420

62 J Labor Res (2013) 34:52–78

are reported in Table 3.11 Not surprisingly, the controls for both public sector and firmsize are no longer significant; in addition there are some moderate changes to thecoefficients relating to education etc. With respect to immigrant status, while thecoefficient relating to NMS immigrants falls somewhat, it remains both negative andsignificant, confirming that our principal results are not the product of unobservedfirm level heterogeneity.

Before proceeding to other dimensions of the analysis, it is important toconsider selection issues as they arise with respect to the characteristics of theimmigrants. In their work on the relative earnings of immigrants and natives inIreland, Barrett et al. (2011) demonstrate that observed differences in wagesbetween immigrants and natives adjust when account is taken of sample selectionbias. Thus it seems, at least with respect to wages, some immigrant groupings are notrandomly distributed in terms of their personal characteristics and, when account istaken of this, the observed immigrant wage penalty increases significantly. If it provesto be the case that immigrants possess characteristics that result in higher ex antetraining probabilities, then failure to adjust for these influences will lead to someunder-estimation of the immigrant training disadvantage. We need to assess whethersuch selection effects are present here in order to ensure that the results from theprobit models are valid. As will be seen, we only attempt to account for selectionbased on observed characteristics. Selection based on unobserved characteristics,such as motivation, might also be present but our data do not allow us to implementinstrumental variables or related techniques which might account for these unob-served selection effects.

We account for selection effects by estimating matching models whereby theprincipal characteristics of immigrants are initially identified through a probit model.Immigrants are then matched on the basis of their predicted probabilities, or propen-sity scores, with natives holding similar characteristics and the training incidences ofthe two groups are then compared. In terms of the matching technique adopted, weapply Nearest Neighbour with replacement.

With respect to their dominant characteristics, we report the results from the stageone probit models in the Appendix Table 14. Immigrants have much lower levels ofexperience and tenure and are less likely to be educated to secondary level. However,they are more likely be male and to hold post-secondary and third-level qualifica-tions. Immigrants are less likely to be members of professional bodies or be employedin the public sector, they also have a somewhat higher tendency to be employed inlarger firms and they have a lower likelihood of trade-union membership. The resultsvary substantially when the models are estimated by country of origin, with NMS andnon-EU non-speaking immigrants exhibiting the most substantial differences relativeto natives.12 Nevertheless, the dominant distinguishing feature of immigrants fromthe UK, EU 13 and New Member States relates to employment tenure, and as such,the PSM approach will ensure that the training rates of immigrants from these

11 As is the case with dummy variables related to sector etc. which will not vary within firms, the modeldrops at least one firm-level dummy to allow the estimation of other firm-level invariant effects such asfirm-size and sector.12 As indicated by the magnitude of the pseudo R squared statistics.

J Labor Res (2013) 34:52–78 63

groupings will be directly comparable with those of natives with similar levels ofemployment tenure. Such an approach thus helps us to directly control for the morerecent arrival of NMS immigrants.

The estimates generated by the matching process are presented in Table 4. Ingeneral, the pattern of results in Table 4 aligns well with those in Table 2 suggestingthat selection bias is not an important factor in this regard, at least as regardsobservable characteristics. As such, the remainder of the analysis uses probit modelswithout any selection-related adjustments.

Barrett et al. (2011) have shown how the wage disadvantage experienced byimmigrants from the NMS countries in Ireland is not uniform across educationalcategories or across occupations. Instead, no disadvantage is present for immigrantswith low levels of skills, as indicated by occupations and education levels, and is highestfor those with higher levels of skill. Given this, it is of interest to explore whether thedisadvantage in training for immigrants is similarly correlated with education andoccupation. It should be noted that even within this more disaggregated analysis, oursample sizes typically do not fall below 3,000 ensuring that our estimates remain robust.

In Table 5, we present the marginal effects on immigrant dummy variables fromregressions run within educational categories. The models are equivalent toModel 3 of Table 2, although with the education categories dropped. Althoughthe educational distributions indicate that immigrants from each background arerelatively well represented within each category of educational attainment (Table 1),it is possible that some of the estimated standard errors will be affected by small cellsizes and, as such, the results should be treated with a degree of caution. As can beseen, the training disadvantage is quite concentrated when viewed in this way, both interms of nationalities and in terms of educational groups. Of the 25 nationality/educationcells in Table 5, only three contain significant coefficients. Two of these relate to NMSimmigrants. It appears that the training disadvantage is not present at either the lowerand or highest ends of the education distribution but instead is concentrated aroundimmigrants with post-secondary educations and primary degrees.

In Table 6, we present the marginal effects from the analysis conducted withinoccupations. As with the education-base regressions, the training disadvantage isagain seen to be concentrated. Looking specifically at NMS immigrants, we see how

Table 4 PSM estimates (nearestneighbour)

The first stage probit resultsare presented inAppendix Table 14

Immigrants - all −0.04***0.01

UK 0.01

0.01

EU13 −0.000.02

NMS −0.13***0.03

Non-EU/English Speaking 0.06*

0.04

Non-EU/Non-English Speaking −0.030.03

64 J Labor Res (2013) 34:52–78

the disadvantage arises generally towards the upper end of the occupational distribu-tion, although not at the very top. This mirrors the results from the education analysis.However, there is also a clear training disadvantage for this group at the lowest end ofthe occupational ladder. We know from previous research (Barrett and Duffy 2008)that NMS immigrants are often well educated but in low skill jobs. This pattern mayexplain the apparent discrepancy between the Tables 5 and 6, if NMS immigrantswith post-secondary qualifications, for example, are working in elementary occupa-tions. This conclusion is confirmed in Table 7 which demonstrates that, relative to allother groupings, NMS immigrants with post-secondary qualifications are much morelikely to be working in elementary occupations or as plant/machine operatives andhave a much lower incidence of employment within managerial or professionaloccupations.

Table 5 Probit regressions by education level for broad training: marginal effects

Variables (1) (2) (3) (4) (5)

Primary Secondary Post-sec Tertiary Post Grad

Tenure −0.00***(0.00)

−0.01***(0.00)

−0.01***(0.00)

−0.00***(0.00)

−0.00***(0.00)

Supervise 0.15***(0.02)

0.18***(0.01)

0.18***(0.02)

0.13***(0.01)

0.12***(0.02)

Experience −0.00 (0.00) −0.01***(0.00)

−0.00 (0.00) −0.00 (0.00) −0.00* (0.00)

Experience squared 0.00 (0.00) 0.00***(0.00)

0.00 (0.00) −0.00 (0.00) 0.00 (0.00)

Full time 0.09***(0.02)

0.04***(0.01)

0.10***(0.03)

0.12***(0.01)

0.14***(0.03)

Professional body 0.13** (0.06) 0.16***(0.02)

0.16***(0.03)

0.13***(0.01)

0.12***(0.01)

Male 0.04** (0.02) 0.06***(0.01)

0.06***(0.02)

0.00 (0.01) −0.01 (0.01)

Union 0.05** (0.02) 0.06***(0.02)

0.05***(0.02)

0.07***(0.01)

0.06***(0.02)

Public −0.14***(0.04)

−0.20***(0.05)

−0.17***(0.05)

−0.11***(0.03)

−0.04 (0.03)

Firm size 0.04***(0.01)

0.05***(0.00)

0.05***(0.01)

0.03***(0.00)

0.02***(0.01)

UK 0.06 (0.06) −0.02 (0.03) 0.02 (0.04) 0.00 (0.02) −0.03 (0.03)

EU 13 −0.05 (0.08) −0.12* (0.07) 0.04 (0.08) −0.04 (0.04) 0.04 (0.03)

NMS −0.02 (0.06) −0.03 (0.04) −0.13***(0.05)

−0.10***(0.03)

−0.07 (0.05)

Non-EU/English speak 0.13 (0.21) 0.07 (0.06) −0.05 (0.11) 0.00 (0.03) 0.04 (0.04)

Non-EU/Non- Englishspeak

0.05 (0.06) −0.03 (0.05) 0.03 (0.07) −0.02 (0.03) 0.01 (0.04)

Observations 3557 17760 5229 17882 4992

Pseudo R-squared 0.076 0.104 0.096 0.127 0.153

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbelow coefficient estimates. All models contain controls for sector and occupations

J Labor Res (2013) 34:52–78 65

Table 6 Probit regressions by occupation level, dependant variable: “broad” training: Marginal effects

Variables (1) (2) (3) (4) (5)

Managers Professional AssocProfessional

Admin &secretarial

Skilledtrades

Tenure −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.01*** (0.00) −0.00 (0.00)

Supervise 0.31*** (0.02) 0.10*** (0.01) 0.09*** (0.01) 0.20*** (0.02) 0.16*** (0.02)

Experience 0.01*** (0.00) −0.00 (0.00) −0.00 (0.00) −0.00** (0.00) −0.01*** (0.00)

Experiencesquared

−0.00*** (0.00) −0.00 (0.00) −0.00 (0.00) 0.00 (0.00) 0.00*** (0.00)

full time 0.05 (0.05) 0.11*** (0.02) 0.12*** (0.03) 0.08*** (0.02) 0.19*** (0.05)

Professionalbody

0.17*** (0.01) 0.12*** (0.01) 0.13*** (0.01) 0.24*** (0.02) 0.13*** (0.05)

Male −0.03* (0.02) −0.03*** (0.01) 0.03* (0.02) 0.06*** (0.02) 0.04 (0.05)

Union 0.14*** (0.02) 0.07*** (0.01) 0.07*** (0.02) 0.04** (0.02) 0.04 (0.03)

Public −0.07 (0.06) −0.08*** (0.02) −0.12** (0.05) −0.23*** (0.07) −0.19** (0.09)

Firm size 0.04*** (0.00) 0.03*** (0.00) 0.04*** (0.01) 0.05*** (0.00) 0.06*** (0.01)

UK −0.03 (0.04) 0.02 (0.02) −0.03 (0.04) 0.03 (0.04) −0.06 (0.06)

EU 13 0.13*** (0.05) 0.02 (0.03) −0.24** (0.11) 0.05 (0.04) −0.13 (0.13)

NMS −0.13 (0.14) −0.12* (0.07) −0.17 (0.11) −0.08 (0.07) −0.07 (0.06)

Non-EU/English speak

0.10 (0.07) 0.01 (0.04) 0.06 (0.05) 0.08 (0.06) −0.01 (0.17)

Non-EU/Non-English speak

0.08 (0.08) −0.07* (0.04) −0.03 (0.05) 0.12** (0.06) −0.09 (0.08)

4523 10529 4787 10501 3379

Observations 4523 10529 4787 10501 3379

PseudoR-squared

0.1529 0.0914 0.099 0.150 0.078

(6) (7) (8) (9)

Variables Person Services Sales Plant & machine Elementaryx

Tenure −0.01*** (0.00) −0.01*** (0.00) −0.01*** (0.00) −0.00*** (0.00)

Supervise 0.17*** (0.02) 0.15*** (0.02) 0.21*** (0.02) 0.18*** (0.02)

Experience −0.01*** (0.00) 0.00 (0.00) −0.00 (0.00) −0.01** (0.00)

Experiencesquared

0.00*** (0.00) −0.00* (0.00) 0.00 (0.00) 0.00 (0.00)

Full time 0.08*** (0.03) 0.10*** (0.02) 0.04 (0.05) 0.10*** (0.03)

Professionalbody

0.12*** (0.04) 0.24*** (0.05) 0.23*** (0.06) 0.15*** (0.04)

Male 0.08*** (0.03) 0.09*** (0.02) 0.10*** (0.03) 0.03 (0.02)

Union 0.06* (0.03) −0.06 (0.05) 0.04 (0.03) 0.05* (0.03)

public −0.28*** (0.07) −0.08 (0.32) 0.07 (0.07) −0.01 (0.06)

Firm size 0.05*** (0.01) 0.06*** (0.01) 0.08*** (0.01) 0.04*** (0.01)

UK −0.03 (0.07) −0.07 (0.05) 0.01 (0.05) 0.03 (0.04)

EU 13 0.03 (0.11) −0.07 (0.10) −0.17 (0.11) 0.00 (0.10)

NMS −0.02 (0.08) −0.08 (0.08) −0.01 (0.05) −0.09** (0.04)

Non-EU/English speak

−0.09 (0.19) 0.08 (0.13) −0.23** (0.09) 0.04 (0.10)

66 J Labor Res (2013) 34:52–78

We expand the analysis further in Tables 8, 9 and 10. In Table 8, Model 1, weinclude two sets of interactions between (a) the immigrant dummy variablesand firm size and (b) the immigrant dummy variables and whether theimmigrant is a union member. Due to issues relating the estimation ofmarginal effects, a linear probability model estimated with robust standarderrors is preferred over the probit. Before looking at the coefficients on theinteractions, it should be noted that their inclusion has an impact on a numberof the immigrant dummy coefficients. In Table 2 Model 3, only the immigrantsfrom the NMS were found to experience a training disadvantage. However, in Table 8Model 1, we now see that immigrants from the EU-13 also experience the disadvant-age. In the case of both EU 13 and EU-English speaking immigrants we findsignificant interaction effects. Immigrants from both categories who work in largefirms have higher likelihoods of receiving training relative to the base. It is possiblethat this positive effect arises from US and EU-13 nationals working in multinationalsin Ireland. In contrast, there appears to be a penalty to union membership for theseimmigrants. In the case of the NMS immigrants, the inclusion of the interactions hasno effect on the immigrant dummy coefficient and neither of the interaction terms issignificant.

In model 2 and 3 of Table 8, we introduce a union-density variable which is adummy variable equal to one if 50 % of more of the employees in the firm are

Table 6 (continued)

Non-EU/Non-English speak

0.08 (0.06) 0.03 (0.10) −0.13** (0.05) 0.01 (0.05)

Observations 2959 3075 4719 4940

PseudoR-squared

0.109 0.096 0.088 0.071

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbeside coefficient estimates. All models contain controls for sector

Table 7 The occupational distribution of immigrants with post-secondary education

Native UK EU15 NMS Non- EU Non-EU

Eng-spk N End-spk

Managers & senior officials 11.1 13.3 9 2.2 10.8 3.5

Professional 34.5 32.4 32.1 8.1 39.7 29

Associate professional & technical 12.6 15.2 12.4 3.8 15.1 10.7

Administrative & secretarial 20.6 16.3 26 9.8 18.5 11.7

Skilled trades 5.2 3.6 2.1 14.8 1.2 5.3

Personal service 4.1 3.5 4.2 7.3 2.7 10.9

Sales & customer service 4.1 3.8 2.9 6.2 3.1 4.7

Process, plant & machine operatives 3.4 4.5 2.1 19.3 3.5 6.6

Elementary 4.4 7.4 9.2 28.5 5.4 17.6

25784 863 477 533 259 792

J Labor Res (2013) 34:52–78 67

Table 8 Linear probability model broad training with interactions: marginal effects

Variables (1) (2) (3)

Model 1 Model 2 Model 3

Tenure −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00)

Supervise 0.14*** (0.01) 0.14*** (0.01) 0.14*** (0.01)

Secondary 0.06*** (0.01) 0.06*** (0.01) 0.06*** (0.01)

Post secondary 0.10*** (0.01) 0.10*** (0.01) 0.10*** (0.01)

Tertiary 0.16*** (0.01) 0.16*** (0.01) 0.16*** (0.01)

Post graduate 0.17*** (0.01) 0.17*** (0.01) 0.17*** (0.01)

Experience −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00)

Experience squared 0.00** (0.00) 0.00** (0.00) 0.00** (0.00)

Full time 0.08*** (0.01) 0.08*** (0.01) 0.08*** (0.01)

Professional body 0.11*** (0.01) 0.11*** (0.01) 0.11*** (0.01)

Male 0.03*** (0.01) 0.03*** (0.01) 0.03*** (0.01)

Union 0.06*** (0.01) 0.06*** (0.01) 0.06*** (0.01)

Public −0.13*** (0.03) −0.13*** (0.03) −0.13*** (0.03)

Firm size 0.04*** (0.00) 0.04*** (0.00) 0.04*** (0.00)

UK −0.04 (0.03) −0.00 (0.01) −0.03 (0.03)

EU 13 −0.26*** (0.07) −0.02 (0.03) −0.26*** (0.07)

NMS −0.13** (0.06) −0.08*** (0.02) −0.13** (0.06)

Non-EU/English speak −0.08 (0.06) 0.02 (0.02) −0.08 (0.07)

Non-EU/Non- English speak 0.01 (0.05) −0.01 (0.02) −0.01 (0.06)

TU Density 0.00 (0.02) 0.00 (0.02)

INTERACTIONS

UK * Firm size 0.01 (0.01) 0.01 (0.01)

EU 13 * Firm size 0.06*** (0.01) 0.05*** (0.01)

NMS* Firm size 0.01 (0.01) 0.01 (0.01)

Non-EU English * Firm size 0.02** (0.01) 0.02* (0.01)

Non EU N English * Firm size 0.00 (0.01) 0.00 (0.01)

UK * TU member −0.01 (0.03)

EU 13 * TU member −0.15** (0.07)

NMS* TU member 0.04 (0.08)

Non-EU English * TU member −0.11** (0.05)

Non EU N English * TU member −0.17*** (0.05)

UK * TU density 0.00 (0.03)

EU 13 * TU density −0.08 (0.06)

NMS* TU density −0.05 (0.09)

Non-EU English * TU density −0.06 (0.06)

Non EU N English * TU density −0.12** (0.06)

Constant 0.20*** (0.03) 0.20*** (0.03) 0.20*** (0.03)

Observations 49420 49420 49420

R-squared 0.20 0.19 0.20

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbeside coefficient estimates. Model 3 and both gender based models contain controls for sector andoccupations

68 J Labor Res (2013) 34:52–78

union members and zero otherwise. The inclusion of this variable is intended tocapture the possible effect that a strong union presence in a firm might have ontraining provision generally and on immigrants in particular. As can be seen,the inclusion of this variable has little impact. The one exception to this is withregard to the interaction between immigrants from non-EU/non-English coun-tries and the union density variable. This interaction term suggests a large andnegative effect. One possible explanation would be the fact that the healthservice is both heavily unionised and a large employer of nationalities suchas Filipinos (nurses) and Indians (doctors).13

As a further extension of the analysis, we address the following questions: towhat extent are immigrants less likely to be employed in training intensivefirms and, secondly, are immigrants employed in training intensive firms lesslikely to receive employer-provided training than natives? We define a trainingintensive firm as one where 50 % or more employees received training in 2005.We then regress this variable on worker characteristics in order to identify thosefactors that most heavily influence an individual’s likelihood of being employedin a training intensive firm. The results are presented in Table 9. The models are

13 We should note that while these may be non-English speaking countries, the immigrants from them maywell be good English speakers.

Table 9 Probit for employmentin training intensive firm’s:marginal effects

*implies significance at the 10 %level; ** implies 5 % and ***implies 1 %; standard errors areshown beside coefficient esti-mates. Model 3 and both genderbased models contain controlsfor sector and occupations

Variables (1) (2)

Model 1 Model 2

Tenure −0.00*** (0.00) −0.00*** (0.00)

Supervise 0.03*** (0.01) 0.03*** (0.01)

Secondary 0.03** (0.01) 0.03** (0.01)

Post secondary 0.07*** (0.02) 0.08*** (0.02)

Tertiary 0.12*** (0.02) 0.12*** (0.02)

Post graduate 0.14*** (0.02) 0.14*** (0.02)

Experience 0.00 (0.00) −0.00 (0.00)

Experience squared −0.00 (0.00) −0.00 (0.00)

Full time 0.06*** (0.01) 0.06*** (0.01)

Professional body 0.03*** (0.01) 0.03*** (0.01)

Male 0.01 (0.01) 0.00 (0.01)

Union −0.00 (0.01) −0.00 (0.01)

Public −0.12 (0.08) −0.12 (0.08)

Firm size 0.07*** (0.01) 0.07*** (0.01)

UK 0.01 (0.01)

EU 13 −0.01 (0.03)

NMS −0.09*** (0.03)

Non-EU/English speak −0.01 (0.02)

Non-EU/Non- English speak −0.03 (0.02)

Immigrant −0.02** (0.01)

Observations 49420 49420

J Labor Res (2013) 34:52–78 69

well specified and indicate that an individual’s probability of being employed in atraining intensive firm increases with education. The results suggest that trainingintensive firms tend to be larger and have higher concentrations of professional staff.Critically our models indicate that immigrants are indeed less likely to be employedin training intensive organisations with the disadvantage specific to immigrants fromnew member states.

In order to explore the second question, we divide our sample of employeesinto two groups; those working in training intensive firms and the rest. Wethen rerun the probit training regressions within each group. The results areshown in Table 10. A number of striking results emerge. First, there is no trainingdisadvantage for immigrants relative to natives if they are employed in trainingintensive firms. Indeed, in the case of immigrants from non-EU/English speakingcountries there is actually a greater likelihood of training. In contrast, the trainingdisadvantage of immigrants is apparent in the firms that are less intensive trainers. As

Table 10 Probit regressions for employees in training intensive and non-intensive firms, dependantvariable: “broad” training

Variables (1) (2)

Intensive Non-intensive

Tenure −0.01*** (0.00) −0.01*** (0.00)

Supervise 0.44*** (0.03) 0.45*** (0.03)

Secondary 0.17*** (0.04) 0.17*** (0.04)

Post secondary 0.22*** (0.04) 0.24*** (0.05)

Tertiary 0.39*** (0.04) 0.34*** (0.05)

Post graduate 0.42*** (0.05) 0.28*** (0.07)

Experience −0.01*** (0.00) −0.01*** (0.00)

Experience squared 0.00* (0.00) 0.00** (0.00)

Full ime 0.22*** (0.04) 0.12*** (0.03)

Professional body 0.42*** (0.03) 0.50*** (0.05)

Male 0.04 (0.03) 0.15*** (0.03)

Union 0.21*** (0.04) 0.08** (0.04)

Public −0.18*** (0.05) −0.48*** (0.11)

Firm size −0.01 (0.01) 0.14*** (0.01)

UK 0.02 (0.05) −0.12* (0.07)

EU 13 0.11 (0.08) −0.30** (0.15)

NMS −0.11 (0.08) −0.23*** (0.08)

Non-EU/English speak 0.19* (0.11) −0.14 (0.15)

Non-EU/Non- English speak −0.04 (0.07) −0.03 (0.09)

Constant 0.56*** (0.12) −1.19*** (0.18)

Observations 32146 17274

Pseudo R-squared 0.090 0.083

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbeside coefficient estimates. Model 3 and both gender based models contain controls for sector andoccupations

70 J Labor Res (2013) 34:52–78

with earlier tables, we see again the disadvantage experienced by immigrants fromboth the EU-13 and the NMS. However, these groups are now joined by immigrantsfrom the UK.

Finally, we examine the extent to which the immigrant training disadvant-age was more heavily concentrated within specific areas. As discussed earlier,the questionnaire collects information of the following specific forms oftraining:

& On-the-job training& Job rotation, exchanges, secondments or study visits& Participation in learning or quality circles& Self-directed learning at least partially funded by the company& Conferences, workshops, trade fairs and lectures.

While it is not straightforward to classify any of the above in terms of eithergeneral or job-specific, it is probably reasonable to state that the first three categoriesare more specific in nature with the latter two more general. We estimate the modelsinitially for the entire sample then separately for males and females. The results fromTable 11 indicate that NMS immigrants were 13 % less likely to be sent to confer-ences, workshops and trade fairs and were also less likely to have access to both on-the-job training and subsidised self-directed learning. Within the aggregate model, therewas no evidence of any immigrant disadvantage in relation to job rotation/study visits orlearning/quality circles. Interestingly, Non-EU immigrants from non-English speak-ing countries were more likely than natives to have access to job rotation/studyvisits and learning/quality circles. However, the pattern of results change somewhatwithin the gender specific models, with male NMS immigrants incurring penaltiesin all areas of training with the exception of study visits (Table 12). Relative tonatives, female NMS migrants were 4 percentage points less likely to access jobrotation schemes/study visits and 13 percentage points less likely to be sent toconferences/trade fairs (Table 13).

Conclusions

A number of findings have emerged from our analysis of the relative likelihood ofimmigrants and natives receiving employer-provided training. Our baseline analysis(Table 2, Model 3) showed that only immigrants from the EU’s New Member Statessuffered a training disadvantage. These results were robust to the influences ofselection bias. Within this NMS group, the training disadvantage was experi-enced only by those with post secondary education and primary degrees. Theinclusion of interactions terms between the immigrant dummy variables and thevariables union membership, union density and firm size had essentially noimpact on the NMS immigrant coefficient. However, when we divided thesample into employees in training intensive and non-intensive firms, not onlywere immigrants from NMS less likely to be employed in a training intensivefirms, a further training disadvantage was experienced by this group employedin non-training intensive firms. At one level, these results point to a labourmarket disadvantage facing immigrants from the NMS in particular. When

J Labor Res (2013) 34:52–78 71

combined with earlier findings on their lower levels of earnings and poorer occupationalattainment (Barrett and McCarthy 2007; Barrett and Duffy 2008), the findings hereadd to a sense that immigrants from the NMS may find themselves in the sort ofsecondary labour markets that were defined by Doeringer and Piore (1971).

For other immigrants, a more mixed picture emerges with training disadvantagesarising for very specific sub-groups. For example, immigrants from non-EU/Englishspeaking countries working as process, plant and machine operatives were shown tobe 23 percentage points less likely to receive training when compared to natives in thesame occupations. Immigrants from the EU-13 were found to be 26 percentage pointsless likely to receive training relative to natives, once interactions were controlled forwhich picked up the effect of EU-13 immigrant working in large firms. Looking at

Table 11 Probit regressions by training type: marginal effects

Variables (1) (2) (3) (4) (5)

Onjob Study Circle Self Conf

Tenure −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00)Supervise 0.10*** (0.01) 0.03*** (0.00) 0.06*** (0.00) 0.03*** (0.00) 0.16*** (0.01)

Secondary 0.07*** (0.01) 0.01 (0.01) 0.03*** (0.01) 0.01 (0.01) 0.05*** (0.01)

Post secondary 0.11*** (0.01) 0.01** (0.01) 0.04*** (0.01) 0.04*** (0.01) 0.10*** (0.01)

Tertiary 0.12*** (0.01) 0.01** (0.01) 0.04*** (0.01) 0.07*** (0.01) 0.17*** (0.01)

Post graduate 0.11*** (0.02) 0.03*** (0.01) 0.03*** (0.01) 0.08*** (0.01) 0.21*** (0.02)

Experience −0.01*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00 (0.00) 0.00 (0.00)

Experiencesquared

0.00*** (0.00) 0.00*** (0.00) 0.00*** (0.00) −0.00 (0.00) −0.00 (0.00)

Full time 0.04*** (0.01) 0.01 (0.00) 0.02*** (0.00) 0.03*** (0.00) 0.09*** (0.01)

Professionalbody

0.05*** (0.01) 0.03*** (0.00) 0.05*** (0.00) 0.07*** (0.01) 0.19*** (0.01)

Male 0.02** (0.01) 0.00 (0.00) 0.01** (0.00) 0.01*** (0.00) 0.02*** (0.01)

Union 0.06*** (0.01) 0.02*** (0.00) 0.02*** (0.00) 0.00 (0.00) 0.02* (0.01)

Public −0.19*** (0.02) −0.04*** (0.00) −0.07*** (0.01) −0.04*** (0.01) −0.02 (0.02)

Firm size 0.04*** (0.00) 0.01*** (0.00) 0.02*** (0.00) 0.00*** (0.00) 0.01*** (0.00)

UK −0.00 (0.01) −0.01 (0.01) 0.01 (0.01) −0.01** (0.01) −0.01 (0.01)

EU 13 −0.04 (0.03) −0.01 (0.01) −0.02 (0.01) −0.02* (0.01) −0.03 (0.02)

NMS −0.06*** (0.02) 0.01 (0.01) −0.02 (0.01) −0.03*** (0.01) −0.13*** (0.02)Non-EU/English speak

0.03 (0.03) −0.01 (0.01) 0.01 (0.02) 0.03 (0.02) 0.06** (0.03)

Non-EU/Non-English speak

−0.01 (0.02) 0.02*** (0.01) 0.03** (0.01) 0.01 (0.01) −0.04* (0.02)

(0.02) (0.01) (0.01) (0.01) (0.02)

Observations 49420 49420 49420 49420 49420

PseudoR-squared

0.072 0.062 0.066 0.092 0.196

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbelow coefficient estimates. Model 3 and both gender based models contain controls for sector andoccupations

72 J Labor Res (2013) 34:52–78

different forms of training, we found the magnitude of disadvantage was highest withrespect to attendance at conferences, workshops, trade fairs and lectures for NMSmales. This group was also 8 percentage points less likely than natives to receive on-the-job training. There was evidence of exclusion in other forms of training, partic-ularly for males but the marginal effects were somewhat lower. It also appears thatNMS migrants have a lower likelihood of accessing both general and job-specificforms of training.

But on a broader level, the findings also point to a diversity of experience bothwithin the NMS immigrants and across other immigrant groups. Such diversity ofoutcomes is likely to be related in turn to a diversity of processes. In some cases, thelower rate of receipt of employer provided training may result from decisions on thepart of immigrants while in other cases, the decisions of employers may be the keydeterminant. In the case of many immigrants from the NMS, the temporary nature of

Table 12 Probit regressions by training type (males): marginal effects

Variables (1) (2) (3) (4) (5)

Onjob Study Circle Self Conf

Tenure −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00 (0.00)

Supervise 0.09*** (0.01) 0.03*** (0.00) 0.07*** (0.01) 0.03*** (0.00) 0.16*** (0.01)

Secondary 0.06*** (0.01) 0.00 (0.01) 0.02** (0.01) 0.01 (0.01) 0.05*** (0.01)

Post secondary 0.11*** (0.02) 0.01 (0.01) 0.03*** (0.01) 0.04*** (0.01) 0.11*** (0.02)

Tertiary 0.10*** (0.02) 0.01 (0.01) 0.04*** (0.01) 0.06*** (0.01) 0.18*** (0.02)

Post graduate 0.07*** (0.02) 0.02** (0.01) 0.01 (0.01) 0.09*** (0.02) 0.20*** (0.02)

Experience −0.01*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00** (0.00) 0.00 (0.00)

Experiencesquared

0.00*** (0.00) 0.00*** (0.00) 0.00*** (0.00) 0.00 (0.00) −0.00 (0.00)

Full time 0.03 (0.02) 0.01 (0.01) 0.03*** (0.01) 0.04*** (0.01) 0.16*** (0.01)

Professionalbody

0.06*** (0.01) 0.03*** (0.01) 0.05*** (0.01) 0.06*** (0.01) 0.19*** (0.01)

Union 0.07*** (0.01) 0.01** (0.00) 0.02*** (0.01) 0.01 (0.01) −0.00 (0.01)

Public −0.19*** (0.04) −0.04*** (0.01) −0.06*** (0.02) −0.03*** (0.01) 0.00 (0.04)

Firm size 0.04*** (0.00) 0.01*** (0.00) 0.02*** (0.00) 0.01*** (0.00) 0.01*** (0.00)

UK −0.01 (0.02) −0.01 (0.01) 0.01 (0.01) −0.02** (0.01) −0.00 (0.02)

EU 13 −0.04 (0.04) −0.01 (0.01) −0.02 (0.02) −0.01 (0.01) −0.03 (0.03)

NMS −0.08*** (0.02) −0.00 (0.01) −0.04*** (0.01) −0.05*** (0.01) −0.13*** (0.03)Non-EU/English speak

−0.04 (0.04) −0.01 (0.01) −0.03 (0.02) 0.04 (0.02) 0.07* (0.04)

Non-EU/Non-English speak

0.01 (0.03) 0.02 (0.01) 0.03 (0.02) 0.00 (0.01) −0.04* (0.02)

Observations 25719 25719 25719 25719 25719

PseudoR-squared

0.068 0.057 0.059 0.10 0.193

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shownbelow coefficient estimates. Model 3 and both gender based models contain controls for sector andoccupations

J Labor Res (2013) 34:52–78 73

their stay in Ireland14 may have led them to avoid jobs where training was needed, andhence where initial wages may have been lower. While this might have been anincome maximising strategy in the short-run, the failure to acquire training wouldclearly work against any possibility of enjoying a wage premium on returning home,along the lines discussed above (Co et al. 2000; Barrett and O’Connell 2001a).

Clearly, much more needs to be done to gain a deeper understanding of the labourmarket outcomes of immigrants beyond the well-researched areas of earnings andoccupational attainment and the issue of training appears to be a potentially importantavenue of research. As discussed in the literature review above, we know that employer-

14 Central Statistics Office (2009) shows, for example, that of the social security numbers issued toimmigrants from the NMS in 2005, almost a half of these numbers were “inactive” by 2008, where inactivemeans that no taxes were being paid or welfare claimed.

Table 13 Probit regressions by training type (females): marginal effects

Variables (1) (2) (3) (4) (5)

Onjob Study Circle Self Conf

Tenure −0.01*** (0.00) −0.00** (0.00) −0.00* (0.00) −0.00*** (0.00) −0.00*** (0.00)Supervise 0.10*** (0.01) 0.03*** (0.00) 0.05*** (0.01) 0.03*** (0.01) 0.16*** (0.01)

Secondary 0.08*** (0.02) 0.01 (0.01) 0.04*** (0.01) 0.01 (0.01) 0.07*** (0.02)

Post secondary 0.11*** (0.02) 0.03** (0.01) 0.06*** (0.02) 0.05*** (0.02) 0.11*** (0.02)

Tertiary 0.13*** (0.02) 0.02** (0.01) 0.05*** (0.01) 0.07*** (0.01) 0.19*** (0.02)

Post graduate 0.14*** (0.02) 0.04*** (0.01) 0.05*** (0.02) 0.08*** (0.02) 0.24*** (0.02)

Experience −0.01*** (0.00) −0.00*** (0.00) −0.00*** (0.00) −0.00 (0.00) 0.00 (0.00)

Experiencesquared

0.00** (0.00) 0.00** (0.00) 0.00 (0.00) −0.00* (0.00) −0.00** (0.00)

Full time 0.04*** (0.01) 0.01 (0.00) 0.02*** (0.01) 0.03*** (0.00) 0.08*** (0.01)

Professionalbody

0.05*** (0.01) 0.02*** (0.00) 0.05*** (0.01) 0.09*** (0.01) 0.20*** (0.01)

Union 0.04*** (0.01) 0.02*** (0.01) 0.01** (0.01) −0.00 (0.01) 0.04*** (0.01)

Public −0.19*** (0.02) −0.04*** (0.01) −0.08*** (0.01) −0.05*** (0.01) −0.03* (0.02)

Firm size 0.04*** (0.00) 0.01*** (0.00) 0.02*** (0.00) 0.00*** (0.00) 0.01*** (0.00)

UK 0.00 (0.02) 0.00 (0.01) 0.01 (0.01) −0.01 (0.01) −0.02 (0.02)

EU 13 −0.05 (0.03) −0.00 (0.01) −0.02 (0.02) −0.03** (0.01) −0.03 (0.03)

NMS −0.02 (0.03) 0.04** (0.02) 0.02 (0.02) −0.02 (0.02) −0.13*** (0.03)Non-EU/English speak

0.09** (0.04) −0.01 (0.02) 0.06** (0.03) 0.02 (0.02) 0.06 (0.04)

Non-EU/Non-English speak

−0.04 (0.03) 0.03** (0.01) 0.03* (0.02) 0.01 (0.02) −0.03 (0.03)

(0.02) (0.01) (0.02) (0.02) (0.03)

Observations 23701 23701 23701 23701 23701

PseudoR-squared

0.080 0.072 0.078 0.089 0.21

* implies significance at the 10 % level; ** implies 5 % and *** implies 1 %; standard errors are shown belowcoefficient estimates. Model 3 and both gender based models contain controls for sector and occupations

74 J Labor Res (2013) 34:52–78

provided training provision appears to be stratified in a way that magnifies labourmarketdisadvantage as opposed to counteracting it. In many cases, the state intervenes throughstate training agencies to reverse this disadvantage but the intervention is more normallyaimed at the unemployed. The results here indicate that there may be a role for stateagencies in the increased training of certain groups of immigrants and that suchinterventions could assist in the long-term integration of immigrants.

We would like to acknowledge the funding provided by the Irish Research Council for theHumanities and Social Sciences under the Government of Ireland Thematic Research Project GrantsScheme. We would also like to acknowledge helpful comments from two anonymous referees andthe Editor. The usual disclaimer applies. Any views expressed are those of the authors and notnecessarily of the ESRI or CBI

Appendix

Table 14 Stage 1 probits from PSM estimation

All migrants z UK z EU13 z

Tenure −0.084 −21.74 −0.08 −16.6 −0.10 −8.57Supervise 0.005 0.12 0.15 2.4 0.02 0.23

Secondary −0.438 −5.85 0.13 1.02 −0.78 −4Postsec 0.290 3.59 0.57 4 −0.02 −0.11Tertiary 0.042 0.56 0.37 2.77 −0.01 −0.06Post graduate 0.401 4.71 0.72 4.81 0.86 4.31

Experience −0.070 −12.66 0.09 9.2 −0.05 −3.48Experience squared 0.001 10.53 0.00 −5.14 0.00 1.5

Full time 0.281 4.85 −0.14 −1.58 0.56 3.59

Professional body −0.421 −7.77 0.03 0.32 −1.26 −7.71Male 0.340 9.47 0.22 3.6 0.18 2.12

Union −0.308 −6.23 −0.21 −2.77 −0.69 −4.96Public −0.594 −9.43 −0.15 −1.64 −1.02 −6.16Firm size 0.022 2.22 0.00 0.16 0.08 3.3

_cons −1.485 −15.71 −4.31 −24.82 −3.50 −14.35N 49420 46709 45991

Pseudo R-squared 0.098 0.043 0.122

NMS z NUE-eng NEU-N ENG

Tenure −0.30 −14.85 −0.08 −7.25 −0.03 −2.69Supervise −0.33 −3.22 0.10 0.8 0.00 0.03

Secondary −0.92 −6.03 0.46 1.14 −1.29 −9.45Post sec 0.48 3.04 0.66 1.52 −0.65 −4.02Tertiary −0.89 −5.71 1.30 3.26 −0.39 −3.04Post grad −0.74 −3.84 1.65 3.98 −0.41 −2.65Experience −0.13 −9.64 0.02 0.96 −0.19 −12.41Experience squared 0.00 5.02 0.00 −0.5 0.00 3.62

J Labor Res (2013) 34:52–78 75

Table 14 (continued)

Full time 1.44 8.48 0.42 2.02 0.21 1.87

Professional body −1.67 −7.51 −0.01 −0.07 −0.60 −5.41Male 0.47 5.93 0.05 0.43 0.61 8.67

Union −0.72 −4.94 −0.29 −1.86 −0.04 −0.42Public −2.94 −7.01 −0.11 −0.63 −1.13 −7.93Firm size 0.01 0.45 0.03 1.02 0.04 2.03

Constant −2.09 −9.26 −5.96 −13.03 −1.67 −9.89N 46176 45699 46313

Pseudo R-squared 0.27 0.051 0.17

Table 15 Data appendix

Variable Labels and Definitions

Label Definition

Human Capital

Lower secondary Lower secondary qualification (1,0 dummy variable)– took three years of secondary schooling

Upper secondary Upper secondary qualification (1,0 dummy variable)– completed secondary schooling

Post secondary Post secondary qualification (1,0 dummy variable)– completed some form of education after secondarybut not to certificate, diploma or degree level

Cert/diploma Cert/diploma qualification (1,0 dummy variable)– completed a certificate of diploma course

Third-level Third-level qualification (1,0 dummy variable)– completed a degree course

Professional body Member of a professional body (1,0 dummy variable)

Experience Experience (measured in years) – some of time spentin all jobs

Experience squared Experience squared

Union member Member of a trade union (1,0 dummy variable)

Supervise Respondent supervises staff in their job(1,0 dummy variable)

Tenure Length of time with current employer(measured in years)

Firm size Number of employees in the firm (continuous)

Full-time Employed on a full-time basis, (1,0 dummy variable)

Public sector Employed in the public sector, (1,0 dummy variable)

On-the-job training Received on-the-job training, (1,0 dummy variable)

Study visit Participated in study visits, (1,0 dummy variable)

Learning circle Participated in learning circles, (1,0 dummy variable)

Conferences Attended conferences, (1,0 dummy variable)

76 J Labor Res (2013) 34:52–78

References

Acemoglu D, Pischke J-S (1999) Beyond Becker: training in imperfect labour markets. Econ J 109:F112–142Barrett A, Duffy D (2008) Are Ireland’s immigrants integrating into its labour market? Int Migrat

Rev 42(3)Barrett A, McCarthy Y (2007) Immigrants in a booming economy: analysing their earnings and welfare

dependence. Lab Rev Lab Econ Ind Relat 21(4–5)Barrett A, O’Connell PJ (2001a) Is there a wage premium for returning Irish migrants. Econ Soc Rev 32(1):1–21Barrett A, O’Connell PJ (2001b) Does training generally work? The returns to in-company training. Ind Labor

Relat Rev 54(3)Barrett A, McGuinness S, O’Brien M (2011) The immigrant earnings disadvantage across the earnings and

skills distributions: the case of immigrants from the EU’s new member states. Br J Ind Relat (EarlyView, doi: 10.1111/j.1467-8543.2010.00835.x)

Becker GS (1964) Human capital: a theoretical and empirical analysis, with special reference to education.National Bureau of Economic Research, New York

Booth A, Bryan M (2002) Who pays for general training? New Evidence for British Men and Women. IZADiscussion Paper No. 386

Booth A, Francesconi M, Zoega G (2003) Unions, work-related training and wages: evidence for Britishmen. Ind Labor Relat Rev 57(1):68–91

Borjas GJ (1985) Integration, changes in cohort quality and the earnings of immigrants. J Labor Econ 3:463–489Central Statistics Office (2009) Foreign National: PPSN allocations, employment and social welfare

activity, 2008. CSO, DublinChiswick BR (1978) The effect of Americanisation on the earnings of Foreign-born men. J Polit

Econ 86:897–921Chiswick BR, Lee YL, Miller P (2005) Longitudinal analysis of immigrant occupational mobility: a test of

the immigrant integration hypothesis. Int Migrat Rev 39:332–353Co CY, Gang IN, Myeong-Su Y (2000) Returns to returning. J Popul Econ 13:57–79Doeringer PB, Piore MJ (1971) Internal labor markets and manpower. Lexington Books, LexingtonFriedberg R (2000) You can’t take it with you: immigrant assimilation and the portability of human capital.

J Labor Econ 18(2):221–251Gordon D, Edwards R, Reich M (1982) Segmented work, divided workers: the historical transformation of

labour in the United States. Cambridge University Press, CambridgeHum D, Simpson W (2003) Job-related training activity by immigrants to Canada. Canadian Public Policy

29(4):469–490Kennedy S, Drago R, Sloan J, Wooden M (1994) The effect of trade unions on the provision of training:

Australian evidence. Br J Ind Relat 32(4):565–560Miller PW (1994) Gender discrimination in training: an Australian perspective. Br J Ind Relat 32(4):539–564

Table 15 (continued)

Variable Labels and Definitions

Label Definition

Irish Native (1,0 dummy variable)

UK UK born immigrant. (1,0 dummy variable)

NMS New Member State born immigrant.(1,0 dummy variable)

Non-EU/English speaking Non-EU/English speaking immigrant.(1,0 dummy variable)

Non-EU/Non-English speaking Non-EU/Non-English speaking immigrant.(1,0 dummy variable)

J Labor Res (2013) 34:52–78 77

O’Connell P, Byrne D (2012) The determinants and effects of training at work: bringing the workplace backin. Eur Socio Rev 28(3):283–300

O’Connell P, Junblut J-M (2008) What do we know about training at work? In: Mayer KU, Solga H (eds) Skillformation: interdisciplinary and cross-national perspectives. Cambridge University Press, Cambridge, UK

Shields M, Wheatley Price S (1999a) Ethnic differences in the incidence and determinants of employer-funded training in Britain. Scot J Polit Econ 46:523–551

Shields M, Wheatley Price S (1999b) Ethnic differences in British employer-funded on and off-the-jobtraining. Appl Econ Lett 6:421–429

Van den Heuvel A, Wooden M (1997) Participation of non-English-speaking-background immigrants inwork-related training. Ethnic and Racial Studies 20(4):830–848

78 J Labor Res (2013) 34:52–78