Over the last decade, the UK has experienced unprecedented increases in migration associated with the 2004 A8 expansion of the European Union. These migrant workers have been praised by managers in the UK, who have frequently stated that they perceive these workers to have a strong ‘work ethic’ when measured on aspects such as absence from work rates. This article examines this perceived migrant ‘work ethic’ by analysing worker absence data from the UK Quarterly Labour Force Survey for the period 2005-2012. Regression analysis reveals that when A8 migrant workers first arrive in the UK, they record substantially lower absence than native workers, but that these migrant absence levels assimilate within 2-4 years. If employers use this information to make hiring decisions, this may have negative implications for native workers, but, importantly, only in the short run.
Keywords: Absence from work; Work Ethic; Migration; UK Corresponding author: Chris Dawson, School of Management, University of Bath, Claverton Down, Bath BA2 7AY, United Kingdom, E-mail: C.G.Dawson@bath.ac.uk, Telephone: +44 (0) 1225 383147
Introduction On 1 May 2004 the European Union was expanded to include the A8 nations of Central and Eastern Europe. Since their entry into the United Kingdom labour market, this new group of workers has been lauded by employers as having what has been described as a stronger “work ethic” than workers from the UK. Within the academic literature, recent qualitative evidence studying managers’ views of the migrant work ethic, particularly through observations on sickness absence, suggest that the work ethic of migrants was perceived by these managers to be higher than that of native workers (see, for example MacKenzie and Forde, 2009; Matthews and Ruhs, 2007; Tannock, 2013). Despite these voluminous qualitative findings, however, quantitative evidence that substantiates these employer perceptions of a distinctive migrant work ethic is scant.
Using worker absence data from the UK Quarterly Labour Force Survey (QLFS) covering the period 2005 to 2012, this article presents the first quantitative investigation into the migrant work ethic. While the literature recognises that work ethic is a multi-dimensional concept, absence from work has previously been identified by managers within the qualitative research literature as being an important measure of work ethic (MacKenzie and Forde, 2009; Tannock, 2013). Absence from work has also been extensively used as a measure of work effort within the labour economics literature, especially when examining the increased levels of effort (i.e. lower absence) that temporary workers exert in order to increase their chances of being offered a permanent contract (see Bradley et al., 2014; Engellandt and Riphahn, 2005).1 The theoretical framework presented in this article asserts that recent A8 migrants face certain disadvantages in the UK labour market relative to comparable natives, despite their higher levels of human capital as evidenced, for example, with their higher levels of qualifications (Hopkins and Dawson, 2016; Wadsworth, 2015). It is argued that these disadvantages weaken the labour market power of A8 migrants (Vershinina et al., 2011), providing them with an incentive to exert more work effort. Firstly, recent migrants have limited labour market information about the host country; while, on the demand side, UK employers are unaware of the value of migrant characteristics, such as education and other work-related characteristics, if obtained outside of the UK (Clark and Drinkwater, 2008). This latter factor has also been found in workplace studies (Hopkins et al., 2016), and is the result of both the diversity of qualifications across eight different educational systems, and also the lack of information provided to businesses because of the initially low predictions of the number of additional migrants that would enter into the UK (as also found in Hopkins, 2017). Secondly, many recent migrants possess low levels of English language proficiency which will hinder their labour market outcomes, as these migrants are unable to obtain employment that adequately reflects their particular skills.2 In this view, language skills are seen as complementary to job related skills and both are needed in order to match workers with jobs that reflect their skill set (Dustmann et al., 2013; Eckstein and Weiss, 2004). A particular consequence of these disadvantages and information asymmetries is that migrants are unable to signal ex ante, i.e. when applying for a job, their underlying productivity to employers through the traditional channels, such as education (Spence, 1973) and labour market experience. As such, this article argues that recent migrants have an incentive to find new ex post, i.e. after being employed, methods of signalling productivity to employers in order to progress from low skilled, low paying roles and into employment positions thatbetter reflect their skill sets. In this view, migrant workers signal productivity through a stronger work ethic and, within the context of this study, through lower absenteeism. This signalling of effort will be over and above that required to signal underlying productivity when UK employers are fully informed about migrant characteristics. According to the migrant assimilation model pioneered by Chiswick (1978), the employment outcomes of migrants (e.g. their earnings from work) will converge to those of natives as migrants acquire language skills, labour market information, and skills specific to industries in the host nation over the years following arrival. In line with the predictions of the “assimilationist” model, this article also assesses the assimilation of the migrant work ethic. In short, if a longer residency in the UK improves the employment outcomes of migrants, then these migrants will no longer have an incentive to signal productivity through behaviours associated with a stronger work ethic.
Background The current UK context following the A8 EU expansion of 2004 makes the UK a suitable arena for the study of the links between migration and perceived work ethic (Anderson, 2010). The issue of migrant labour has become particularly important in the UK following the A8 expansion of 2004, where eight Central and Eastern European (CEE) nations joined an expanded EU (Ciupijus, 2011). These countries are the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia. Whereas other member states imposed restrictions of up to seven years, Sweden, the Republic of Ireland and the UK were the only three EU member states to allow full access to workers from the A8 nations to work without restriction. The UK government’s decision was influenced by an original predicted figure of increased migration as a result of the expansion of between eight and thirteen thousand (Dustmann et al., 2003), and as such the only requirement for A8 migrants to take work in the UK was to register on the Worker Registration Scheme (WRS). By the time this scheme was closed in April 2011, seven years since the A8 expansion, over a million people had registered. Clark and Drinkwater (2008) show that these changes saw the proportion of the total number of migrants and immigrants to the UK from the A8 countries rise from 4.1% of the total in 2000-2003 to 36.5% of the total in 2004-2007. While there are different reasons to migrate, as examined by Eade et al. (2007) and Hopkins et al. (2016), in-depth qualitative studies have found that a recurring story among A8 migrants is of highly qualified people taking lower skilled roles. These migrants are found in sectors such as hospitality (Alberti, 2014; McDowell et al., 2008) and manufacturing (Hopkins et al., 2016; Tannock, 2013), particularly in roles where interaction with customers is not required (for example, in distribution warehouses or back of house roles in hospitality). A notable feature of these sectors is the use of deskilled work practices, which demotes the importance of English language proficiency. A further recurring theme is that of managers comparing these migrant workers to those from the UK using the term ‘work ethic’ as a differentiator, which we now examine in greater detail.
Theoretical framework and empirical literature One of the key themes that has emerged from workplace studies of migrant workers is a preference among managers for A8 migrant workers over native workers (see e.g. MacKenzie and Forde, 2009; Tannock, 2013). One explanation for this may be the opportunity to pay lower wages to migrant workers. However, in the context of minimum wage legislation that limits this opportunity, managers cite the stronger migrant work ethic as a key reason for choosing migrants over native workers. But what is this work ethic? Managers repeatedly pointed to low absence in particular, with Hopkins’s (2014) study of absence management finding that managers consider A8 migrants to take less sickness absence than their UK colleagues. Hopkins (2014) also examined the influences of factors other than ill health on absence, following Edwards and Scullion’s (1982) view that absence must also be considered as a response to managerial control. Qualitative evidence from workplace studies, such as those of MacKenzie and Forde (2009) and Tannock (2013), reveals that managers propose a link between this work ethic and migration and, as a result, they prefer migrant to native workers. Matthews and Ruhs (2007) suggest that in lower skilled roles employers will actually prefer a ‘good work ethic’ over more recognisable qualifications or skills. This creates complex hierarchies among potential recruits, where “workers are often – and in some cases primarily – distinguished and recruited on the basis of their nationality” (Matthews and Ruhs, 2007: 29). These findings in the UK match with previous research in other countries, for example that of Chiswick (1978) and Waldinger and Lichter (2003), who find a preference for migrant workers among US managers. Waldinger and Lichter (2003: 176) find that managers reported that they preferred Latino migrants as they “liked to work”, while African Americans were reported to be too “Americanized” and thus more likely to demand higher wages and better conditions.
Within the context of this article, it is argued that the stronger migrant work ethic is directly linked to migrant labour market power and, in particular, low levels of English language proficiency and issues around the portability of qualifications. This is despite the higher levels of human capital among this group, as evidenced by higher levels of qualifications (Hopkins and Dawson, 2016; Wadsworth, 2015). Firstly, migrants may endeavour to negate these issues by being more compliant to the demands of employers, which is often termed by these employers as the ‘migrant work ethic’ (see, for example, MacKenzie and Forde, 2009). Secondly, a particular consequence of lower migrant labour market power is that migrants are forced to supply labour in low-paying, low-skilled employment positions, which do not adequately reflect their particular skills (McCollum and Findlay, 2015). More specifically, Clark and Drinkwater (2008) find that recent migrants from the A8 countries have the lowest returns to their skills, and relate this to the issue of English language proficiency. Related to this, Dustmann and Faber (2005) find that language proficiency is lowest among those groups that have the largest disadvantages in the labour market. In conjunction with language proficiency, Friedberg (2000) argues that another reason for this poor ability to obtain higher skilled roles is a lack of portability of skills and qualifications between countries, with managers unaware of the value of these if they are earned outside the host nation (see also Clark and Drinkwater, 2008; Dustmann and Faber, 2005; Dustmann et al., 2013; Eckstein and Weiss, 2004). Although this would also be the case for migrant workers from other EU nations, workplace studies have confirmed the diversity of qualifications across the eight new accession states as contributing to the lack of portability (Hopkins, 2017). This is a contributory factor in the majority of recent migrant workers in the UK taking low-skill jobs (Alberti, 2014), despite their relatively high levels of formal education (Hopkins and Dawson, 2016). Wadsworth (2015) finds that the immigrant workforce in the UK is better qualified than the native workforce – for example, 46 percent of the UK-born workforce left school aged 16 or younger, compared to 8 percent of the A8 migrant workforce.
Within this context, a stronger migrant work ethic, for example through lower work absence, enables migrants to signal their underlying productivity to employers. While signalling usually takes place ex ante (e.g. Spence, 1973), as UK employers are not fully informed of the value of migrant qualifications (Friedberg, 2000; Clark and Drinkwater, 2008) then migrants will have to find an alternative way of demonstrating their skills and commitment ex post, and signalling this through, in particular, lower levels of absence. As a comparison, Bradley et al. (2014) show that those on a probationary contract will demonstrate a superior work ethic through lower absence in an attempt to increase their chances of gaining a permanent contract. This signalling can therefore be seen as an attempt by individuals to overcome asymmetries of information and, for high-productivity migrant workers, to demonstrate to employers that they truly are more productive in order to be reallocated into more highly-skilled roles which adequately reflect their particular skill sets.
While recent A8 migrants are likely to face disadvantages in the UK labour market relative to comparable natives, the pioneering work of Chiswick (1978) showed that, although immigrants earn less than natives when first arriving, there was equality of earnings for immigrants in the US ten to fifteen years following their arrival (although this varied across ethnicities, with Mexican-born immigrants performing less well). The “assimilationist” interpretation of this finding is that, after arrival, migrants will accumulate language skills, labour market information, and other skills specific to industries in the host country. Moreover, employers may have greater information concerning the work-related characteristics of these migrant workers. The accumulation of these skills is expected to increase the labour market power of migrants, leading to better employment prospects and to the assimilation of migrant wages. Empirical evidence consistent with this assimilation is provided by Clark and Lindley (2009) and Dickens and McKnight (2008). As this labour market assimilation process occurs, migrants will no longer have an incentive to signal productivity through additional effort, therefore their reliance on signalling through, for example, lower absence, will lessen.
Data and descriptive statistics Data source and sample In order to study the A8 migrant work ethic, data drawn from the October-December rounds (fourth quarter) of the QLFS for the years 2005-2015 are utilized. The QLFS is particularly rich in information concerning working hours and absence from work and this information can be used to construct absence measures that proxy the work ethic of individuals working in the UK. The QLFS has a rotating panel structure, where each household member of the sample is interviewed for five consecutive quarters/waves. Wave five responses are excluded from the final sample to avoid duplicate observations for respondents that were observed in their first wave in the previous year.3 The sample is also further restricted to employees that are either UK nationals or A8 migrants. In the latter group, only migrants that arrived in the UK in or after 2004 are included, in order to specifically study the wave of migration from the A8 countries after the enlargement of the EU in that year. Workers that are full-time students, those that are under 16 years of age or above the state pension age (64 for men, 59 for women), and those that report over 90 usual weekly hours (to remove extreme and/or invalid information), are also excluded. Finally, as the QLFS allows interviewers to collect information by proxy, i.e. from another related adult in the household, we exclude these responses from our analysis owing to potential measurement error.
In the estimations, four work absence measures are considered as dependent variables: (1) the sickness absence probability, (2) the sickness absence rate, (3) the overall absence probability, and (4) the overall absence rate. To construct these measures, the procedure outlined in Barmby et al. (2004) has been followed. First, an absence rate is calculated for each individual in the sample as follows: let UHi denote the usual hours the employee i works in a week, excluding any overtime work. This is assumed to correspond to the hours the individual is contracted to work. AHi denotes the actual hours the same employee worked in the reference week of the survey, again excluding any overtime. Those respondents who reported working fewer hours than usual during the reference week were asked a follow-up question regarding the reason for this. The exact wording of the QLFS question is as follows:
“What was the main reason that you did fewer hours than usual/were away from work in the week ending Sunday the …..?
Number of hours worked/overtime varies
Maternity or paternity leave
Sick or injured
Attending a training course away from own workplace
Started new job/ changed jobs
Ended job and did not start new one that week
Laid off/short time/work interrupted by bad weather
Laid off/short time/work interrupted by labour dispute at own workplace
Laid off/short time/work interrupted by economic and other causes
Other personal/family reasons
(Source: QLFS questionnaire, 2012)
A dummy variable is then created, which takes the value of 1 if the individual’s response was j = sick or injured (option 6) in the question above, and 0 otherwise. For the case of overall absence, = 1 if the individual’s response was j = sick or injured (option 6), or other personal/family reasons (option 13), or other reasons (option 14), and 0 otherwise.4 By using all the above variables, the sickness or overall absence rate, , for each individual i is constructed as follows:
where 0 ≤ ≤ 1 for each i. This variable measures the proportion of weekly hours lost owing to the reasons mentioned above and is the sickness absence rate or overall absence rate, depending on how is calculated. By using this rate, we can also construct our sickness and overall absence probability measures. These are discrete variables taking the value of 1 if the respective absence rate is positive (and 0 otherwise) and they effectively measure the incidence of at least one hour of absence in the reference week.
In the multiple regression analysis, linear models for the four dependent variables are estimated to investigate whether A8 migrants record more or less absence from work than UK nationals.5 As well as including in the model a dummy variable indicating whether the individual is an A8 migrant, an interaction of this with a variable that measures the number of years an A8 migrant has resided in the UK since migration is also included. The coefficient of the A8 dummy, therefore, measures the absence differential between a UK national and an A8 migrant that arrived in the UK in the same year as the one he/she is observed in the QLFS, while the coefficient of the interaction term measures the rate of absence assimilation as residency in the UK lengthens for A8 migrants.6 In order to account for the heterogeneity in both personal and labour market circumstances between the A8 migrants and UK nationals, a standard barrage of control variables is included in the regression models. Basic demographic variables include gender, age (and its square), education (in years)7, marital status, number of dependent children under 16 years old, and age of the youngest dependent child. Health status, an important variable in all work absence studies (see e.g. Leigh, 1991), is also included and is captured by two dummies indicating (1) whether the respondent suffers from a long-term health problem, and (2) if that problem affects the amount of work for the employee. A series of region of residence and year dummies are also included to control for regional variations in weather conditions and other relevant variations by place and time. Finally, housing tenure and receipt of any state benefits or tax credits are included in the models in order to capture access to the welfare state.8 These are important controls since A8 migrants were not eligible for tax credits before registering with the WRS, while they also could not claim any income-related benefits before having worked continuously for one year (Dustmann et al., 2010: 6). This limited access to the welfare state is, in turn, expected to affect migrant work effort (see also Hansen and Lofstrom, 2011). To account for possible differential effects of benefits receipt on UK nationals and A8 migrants, an interaction term is added in the models.
Labour market heterogeneity is captured through a series of control variables including: usual basic weekly hours worked, paid and unpaid overtime hours, whether the employee works in the public sector, has a second job, a permanent contract, a managerial or supervisory status, whether the employee works at home (or in the same building as his/her home), tenure with current employer, establishment size (see e.g. Barmby and Stephan, 2000), trade union status (see e.g. Allen, 1984) and flexible working arrangements (see e.g. Heywood and Miller, 2015). A series of occupational and industry dummies are also included in the models.9 Finally, since job dissatisfaction is a much studied variable in the work absence literature (see Steers and Rhodes, 1978), it is proxied here by the following variables: (1) a variable that captures dissatisfaction with current working hours (“Fewer hours desired”); (2) a dummy indicating whether the employee is looking for an extra job; and (3) a dummy taking the value of one if the respondent is looking for a new job.
All the above variables are included in the final models, with their corresponding sample means available in Table A1 of the Appendix. A final sample of 113,804 observations is obtained after dropping individuals with missing observations for any of the dependent or independent variables. 112,408 of these (98.8 percent of the total) correspond to UK nationals and 1,396 (1.2 percent) to A8 migrants. The average UK residency of A8 migrants in the final sample is approximately 3.1 years. The full distribution of migrant UK residency is presented in Figure 1.
Descriptive evidence Before presenting the ordinary least squares (OLS) coefficient estimates, the raw differences in work absence between the native and migrant samples are briefly considered. Table 1 presents the relevant sample means. Crucially, A8 migrants are less likely to be absent from work and also record lower levels of absence than UK nationals. T-tests are performed for the difference in means between the groups; they are highly significant, confirming the differences in each case. These differences are not small; in particular, all absence sample means are 70-90 percent higher for UK nationals, providing prima facie evidence in favour of a better work ethic among A8 migrant workers.
Table A1 in the Appendix also shows that A8 migrants have on average 2 more years of education than their UK national counterparts. Migrants are also younger, healthier, more likely to work in a temporary full-time job, less likely to have managerial or supervisory duties in their job, less likely to be unionized, and have on average a shorter tenure with their current employer than UK nationals. Consistent with the literature on the disadvantages faced by A8 migrants in the UK labour market, A8 migrants face a substantial hourly wage penalty relative to natives and their work is heavily concentrated in low-skilled occupations. Around 63 percent of A8 migrants work as plant and machine operatives or in elementary occupations, while the corresponding percentage for UK nationals is only around 16 percent. A8 migrants also exhibit a lower amount of dissatisfaction with current working hours, although they work substantially more hours (basic and paid overtime) than UK nationals.