Introduction
The ‘refugee crisis’ of 2014 and 2015 acted as a catalyst for the reformation of integration policies in most European countries. In Denmark and many other countries, previous integration and active labour market policies (AMLPs) were deemed mostly unsuccessful in securing employment for newly arrived refugees (Galgóczi, Reference Galgóczi and Galgóczi2021), because migrants in general, and refugees in particular, have lower employment rates than the general population (Brochmann and Hagelund, Reference Brochmann and Hagelund2011; Calmfors and Gassen, Reference Calmfors, Gassen, Calmfors and Gassen2019; Hernes et al., Reference Hernes, Arendt, Joona and Tronstad2019). The majority of new policy initiatives have been supply-side policies in the form of activation, training programmes and economic incentives targeting the refugees (Galgóczi, Reference Galgóczi and Galgóczi2021).
Conceptually, it is possible to distinguish among three general approaches to integration of refugees into the labour market: the supply-side approach (with a focus on the unemployed), the matching approach (with a focus on employment services) and the demand-side approach (with a focus on employers). Each carries implicit definitions of the problem, explanations for labour market outcomes – and lack thereof – and each entails a specific set of policy solutions (Bredgaard and Thomsen, Reference Bredgaard and Thomsen2018). In this article, we begin by focusing on the demand-side approach and the matching approach, as these have been somewhat neglected in existing research on labour market integration of refugees (cf. the literature review). The role of employers is generally underexplored, but a growing strand of research focuses on the role of employers in relation to employment of other disadvantaged groups and regarding employer participation in active labour market policies (e.g. Bredgaard, Reference Bredgaard2018; Bredgaard and Salado-Rasmussen, Reference Bredgaard and Salado-Rasmussen2021; van Berkel, Reference van Berkel2021). We have, however, not identified any studies focusing on employment of refugees that also emphasises the role of workplaces and employers, despite the fact that refugees are a disadvantaged group of particular interest with regard to employer practices. They often face a number of barriers that hamper their employment prospects, such as lack of familiarity with their new host country, including skills in that country’s language, social and health-related problems, inadequate skills and lack of work experience, et cetera. Furthermore, they lack a network that can assist them in getting jobs, and they may face discrimination in the recruitment process (cf. the literature review).
As shall become evident from review of the literature, much of the existing research on labour market participation of migrant and refugees focuses on supply-side explanations, which overshadow the demand-side and matching perspectives (cf. the literature review).
In this article, we therefore turn our attention to the employers and investigate the following research question:
Which workplace factors and employer attitudes and preconceptions are associated with employment of refugees?
To explore this research question, we use a national and representative survey among Danish workplaces with three or more employees. We run a series of logistic regression analyses that explore these associations.
In the regression analyses, we include workplace characteristics that represent demand-side explanations for employment of refugees, as well as matching explanations represented by public employment services contacts with employers. Here, it is important to note that we do not claim to find pure causal relationships between the covariates and the employment of people with a refugee background. Rather, we wish to explore the question of which factors are related to employment of refugees and to test hypotheses regarding the influence of demand-side and matching perspectives, while controlling for other factors. We take this approach because, in our view, the relevance of the role of employers in relation to employment of refugees and other disadvantaged groups has been somewhat neglected in existing research (cf. the literature review). The consequences of this knowledge gap are multiple. From a practical point of view, it may entail that the employment rates of refugees are lower than they could potentially be if employers could be engaged to a stronger degree in ALMPs. In other words, unrealised potential may exist both in relation to employment rates and employer engagement in ALMPs. The implications of stronger engagement could be that more people could be lifted out of poverty while contributing to the financial sustainability of the welfare state by payment of income taxes.
We contribute to two related bodies of literature through our study. Overall, we contribute to literature on conditions for demand-side and matching ALMPs based on an employer perspective, as well as the literature on employer engagement and social responsibility. Going into more detail, first, we contribute to the literature on labour market integration of refugees by finding associations between workplace characteristics and employment of refugees. Knowing these associations can help employment services target their placement efforts towards specific types of workplaces with the highest likelihood of success. Second, we contribute to the ‘matching literature’ through our findings that contacts of employers by employment services are associated with employment of refugees, thereby providing preliminary support for the assumptions of the matching perspective. Third, we contribute to the literature on employer engagement by demonstrating that employer notions of social responsibility are associated with employment of refugees. Fourth, also in relation to the demand-side focus and employer engagement, our main contribution to the literature is the finding that the likelihood of employing refugees increases if the employer exhibits socially responsible attitudes and socially responsible behaviour towards other disadvantaged groups. We also find that attitudes and some preconceptions about refugees are related to employment of refugees.
In relation to the matching perspective, we find that workplaces that use public employment services to recruit labour are more likely to have refugee employees. The same is the case for workplaces that have been contacted by public employment services with the purpose of encouraging the employer to hire a refugee or getting a refugee into workplace training.
The article is structured as follows. First, we review the literature on labour market outcomes for refugees and migrants in general. We structure the review in line with the three conceptual (theoretical) approaches. We furthermore include the literature on employer engagement and employer attitudes. This allows us to identify relevant variables representing the demand-side approach and the matching approach for the regression analyses. Subsequently, we describe the data and the methods used for the analyses. Then we present the results of our analyses and discuss the findings.
Conceptual approach, review of the literature and selection of covariates
In this section, we review the literature on factors influencing employment outcomes and introduce the three conceptual approaches. Of course, the three approaches are not mutually exclusive. Each perspective has value, and in most countries, the distinct policy mix will incorporate elements of each perspective.
We structure the review according to the three conceptual approaches, including insights from labour market research that focuses on the disadvantaged unemployed, in general, as well as research on labour market integration of refugees and migrants, specifically. We are well aware that there are vast differences between refugees and migrants more generally, but we include both groups below, since they may face some of the same obstacles in the recruitment process – for instance, discrimination, and not being (fully) proficient in the language of the host country. Labour migrants, for instance, are, however, generally better off in the recruitment process, due to skills and qualifications in demand, few experiences of trauma and a lower degree of work-inhibiting health problems et cetera. We furthermore touch upon the literature on employer engagement in relation to the demand side.
The supply-side – individual factors influencing employment
In the supply-side understanding, lower labour market performance is ascribed to characteristics of the refugees themselves (i.e. individual factors influencing employment). Supply-side policy solutions involve a plethora of well-known initiatives, such as training and education programmes, active labour market programmes, workplace training and language acquisition education, in addition to lowering welfare benefits (Bredgaard and Thomsen, Reference Bredgaard and Thomsen2018).
A vast body of research has explored supply-side factors that influence employment outcomes. In particular, this is the case with the literature that evaluates ALMPs. For reviews on the effectiveness of ALMPs for migrants see – for instance, Butschek and Walter (Reference Butschek and Walter2014) and Heinesen et al. (Reference Heinesen, Husted and Rosholm2013). Given that this article is not concerned with exploring supply-side explanations for labour market outcomes, this body of research will only be briefly outlined, and it will solely focus on the employment-related barriers that migrants and refugees face.
A typical explanation for the lower employment rate of refugees and migrants is that because refugees are not fully proficient in the language of their new host country, they face difficulties in obtaining employment (e.g. Lundborg and Skedinger, Reference Lundborg and Skedinger2016; Bredgaard and Thomsen, Reference Bredgaard and Thomsen2018). An additional barrier is related to refugees’ qualification levels for employment. Many have limited schooling from their country of origin and therefore lack the qualifications in demand among employers (Brücker et al., Reference Brücker, Liebau, Romiti and Vallizadeh2014; Schultz-Nielsen and Skaksen, Reference Schultz-Nielsen and Skaksen2017). Even if they have a degree from their home country, it may not be considered valid in the new host country. In line with this, many immigrants who succeed in gaining employment often work in positions below their level of qualification – this is also the case for refugees with more than basic level schooling (Schultz-Nielsen, Reference Schultz-Nielsen2020). Some studies also stress that refugees lack economic incentives or ‘motivation’ to obtain employment. Danish research examining the effects of reducing welfare benefits has found positive employment effects for refugee men but not for refugee women (Arendt, Reference Arendt2020). Welfare benefit reductions, however, come at a cost in the form of lower disposable income, increases in property crime, and lower language test scores and educational outcomes among children whose parents were subject to benefit reduction (Andersen et al., Reference Andersen, Dustmann and Landersø2019).
Poor health is also often considered a significant barrier for obtaining employment. Scandinavian research finds that refugees have poorer health than both the native population and other minority groups, which is a significant obstacle for employment (Borsch et al., Reference Borsch, de Montgomery, Gauffin, Eide, Heikkilä and Jervelund2018).
The matching perspective – the role of employment services
The starting point of the matching perspective is that the lower employment rates of refugees (and other unemployed people, as well) is caused by information asymmetries on both the supply-side (the refugees) and the demand-side (Larsen and Vesan, Reference Larsen and Vesan2012; Bredgaard and Thomsen, Reference Bredgaard and Thomsen2018). Employers are not well-informed about the possibility of recruiting people with a refugee background, and refugees do not know where or how to apply for vacant job openings, and they are missing formal and informal person contacts who might inform them of job openings. In essence, matching perspective explanations for the lower employment rates of refugees highlight the fact that refugees do not have access to formal and informal recruitment networks in their new host country (Bredgaard and Thomsen, Reference Bredgaard and Thomsen2018). Refugees are at a disadvantage in this regard, as a large share of positions are filled through networks and personal contacts, and refugees often need to rely on the public employment services to assist them in the job search process, despite the fact that these organisations are rarely used for recruitment (Granovetter, Reference Granovetter1974; Larsen and Vesan, Reference Larsen and Vesan2012; Håkansson and Nilsson, Reference Håkansson and Nilsson2019). The role of the public employment services (PES) is to facilitate contact between refugees and potential employers, thus acting like a job broker. In other words, the employment service needs to ‘match’ refugee jobseekers with employers. The problem, however, is the employment service may lack credible information about the qualifications of refugees and about job openings as well. Nevertheless, the employment service needs to try to break down these information asymmetries – for instance, by proactively contacting employers and trying to get the unemployed into work-place training programmes. In the Danish case, this takes the form of an active labour market policy in the form of unpaid internships where the trainee performs work for the employer for a limited time. This allows the employers to assess a potential employee for a limited time before making a decision about recruitment.
The above represents the ideal in the matching perspective. In practice, employment services may contribute to ensuing institutional racism, and the caseworkers’ high levels of discretion may result in creaming and parking of clients (Greer et al., Reference Greer, Schulte and Symon2018) with caseworkers focusing their efforts and clients with a high chance of success while parking other clients. The investigation of the caveats of the matching approach is, however, beyond the scope of the article.
Based on the existing literature, we believe contacts to the employers can help facilitate employment of refugees, and we arrive at three hypotheses. The likelihood of having a refugee employee is higher if:
-
1) The employer uses the PES as a recruitment channel.
-
2) The employer has been contacted by the PES with the purpose of hiring a refugee jobseeker.
-
3) The employer has been contacted by the PES with the purpose of getting refugees into workplace training.
In the regression models, we therefore include three covariates to investigate each of these hypotheses.
The demand-side perspective – workplace factors influencing employment outcomes
As a starting point, the demand-side perspective attributes lower employment rates among refugees to the fact that some employers discriminate against refugees and minorities in the recruitment process, either directly or indirectly or have inadequate incentives to employ refugees (Bredgaard and Thomsen, Reference Bredgaard and Thomsen2018; Ravn and Bredgaard, Reference Ravn and Bredgaard2021). Extensive research, in the form of field experiments, has found that religious and ethnic minorities face discrimination in the recruitment process, which puts them at a disadvantage in the competition for jobs (e.g. Midtbøen and Rogstad, Reference Midtbøen and Rogstad2012; Zschirnt and Ruedin, Reference Zschirnt and Ruedin2016). For instance, in a cross-national field experiment, Di Stasio et al. (Reference Di Stasio, Lancee, Veit and Yemane2019) found a penalty for signalling Muslim faith in a job application, and in a meta-analysis of ninety-seven studies Quillian et al. (Reference Quillian, Heath, Pager, Midtbøen, Fleischmann and Hexel2019) found extensive evidence of racial discrimination. In the Danish case, Dahl and Krog (Reference Dahl and Krog2018) finds evidence of discrimination against job applicants with a Middle-Eastern sound name, and in a vignette experiment, Ravn and Bredgaard (Reference Ravn and Bredgaard2021) find that refugees are rated less positively by employers than a control where no refugee background is mentioned. Demand-side policy solutions include strengthened economic incentives for employers to hire people with a refugee background (e.g. economic bonuses or wage subsidies). An additional demand-side strategy would be to enact quotas, as some European countries have regarding people with disabilities, making it mandatory for larger companies to employee refugees (Bunt et al., Reference Bunt, van Kessel, Hoekstra, Czabanowska, Brayne, Baron-Cohen and Roman-Urrestarazu2020).
In addition to possible discriminatory practices among employers, there are other workplace factors that influence employment of refugees. We focus on these factors in the following section and in the article more generally.
The size of the workplace is one factor that may influence the likelihood of having refugee employees. If the composition of employees reflects the composition of the general population, the likelihood of finding (at least) one refugee employee will be higher in larger workplaces/firms. In addition, a study by Banerjee, Reitz and Oreopoulos (Reference Banerjee, Reitz and Oreopoulos2018) argued that larger firms may discriminate to a lesser extent than small and medium-sized firms. Due to this, we hypothesise that the larger the workplace, the higher the likelihood of having at least one refugee employee, and we therefore include a covariate measuring the number of employees in our regression model.
Azlor et al. (Reference Azlor, Damm and Schultz-Nielsen2018) find that local demand for labour is a decisive factor for employment of immigrants. To account for this, we include a variable in the models that measures whether the workplace has recruited at least one new employee over the course of the previous year. We expect that workplaces that have hired new employees will be more likely to have a refugee employees.
Also related to the literature on discrimination, Villadsen and Wulff (Reference Villadsen and Wulff2018) investigate whether public sector employers discriminate less against ethnic minorities than their private sector counterparts. They find evidence that discrimination occurs in both sectors. In addition, immigrants are, to a larger extent than native Danes, employed in the public sector (Ejrnæs, Reference Ejrnæs2012). We therefore include a variable measuring whether the workplace is in the public or private sector.
The Danish labour market, like other Nordic labour markets, is characterised by demand for highly skilled labour and by a high degree of regulation, with rather high minimum wages secured through collective agreements in most workplaces (Brochmann and Hagelund, Reference Brochmann and Hagelund2011). Furthermore, Denmark has a long-standing tradition of social partnership between unions and employer associations (Rasmussen and Høgedahl, Reference Rasmussen and Høgedahl2021), and corporatist policy-making, which entails willingness from both employers and unions to help combat societal problems (Binderkrantz and Christiansen, Reference Binderkrantz and Christiansen2015). The latest example of corporatist policy-making in relation to the integration of refugees came in 2016 when the government, unions and employer associations reformed the national integration programme.
In Denmark, being a member of an employer association entails adherence to collective agreements. Non-members are not obliged to follow the collective agreements or to pay the minimum wages secured by such agreements.
We hypothesise that membership of an employer association may influence the likelihood of employment of refugees in two directions. It may positively be related to having refugee employees, given that having a collective agreement in place signals that an employer is in favour of decent wages, working conditions and promotion of corporate social responsibility. However, being a member and thus having a collective agreement in place that dictates rather high minimum wages might also be negatively related to employment of refugees, since collective agreement entails rather high minimum wages. As we saw earlier in the review, refugees and immigrants are often low-skilled, have little education and are likely to be employed in jobs below their skill-level. Not having a collective agreement in place means that an employer is permitted to pay low(er) wages, which might entail a higher likelihood of having refugee employees. For these reasons, we include a variable measuring membership in an employer association.
A study by Lundborg and Skedinger (Reference Lundborg and Skedinger2016) finds that prior experience matters for current practice in relation to employment of refugees. Some have had positive experience resulting in their continued willingness to employ refugees, whereas others have had negative experiences, meaning that they have been discouraged from hiring refugees. We therefore include a variable measuring whether or not employers have previously had a refugee employee.
Existing corporate social responsibility practice might also play a role in relation to employment of refugees (e.g. Bredgaard, Reference Bredgaard2018; Bredgaard and Salado-Rasmussen, Reference Bredgaard and Salado-Rasmussen2021). For instance, if the workplace employs other disadvantaged groups, the likelihood of employing refugees might also increase. We include a variable measuring the employment of people coming from the three other disadvantaged groups (a long-term unemployed person, a person in a wage-subsidy job and a person in a flex-job). Long-term unemployment needs no further explanation. Wage-subsidy jobs are a special, limited form of employment representing an ALMP for those unemployed people in Denmark who face difficulties in gaining employment. In wage-subsidy schemes, employers hire a jobseeker for a limited time and are reimbursed for some of the wage expenses for the person in question. The logic behind this is that employers get to assess the work of an employee before they make a decision about longer-term employment on ordinary terms. The flex-job scheme is similarly a special type of employment where eligibility is limited to people with severe and permanently reduced work ability (i.e. disabled; Ravn and Bredgaard, Reference Ravn, Bredgaard, Klindt, Rasmussen and Jørgensen2020). The variable we use in the analyses to assess the influence of these practices ranges from 0 to 3, where ‘0’ indicates that the workplace does not employ any of the three groups and ‘3’ indicates that they have employees from all three groups.
The above measures constitute ‘objective’ factors or workplace characteristics, meaning they are more or less observable characteristics that employment services can use to target their efforts if they know which of the factors is associated with employment of refugees. The following factors are not directly observable, but they are nevertheless important for increasing our knowledge about which employer factors are relevant for employment of refugees.
A growing body of literature has turned attention to the role of employers (i.e. the demand-side) in relation to employment of disadvantaged groups and participation in active labour market policies (Ingold and Stuart, Reference Ingold and Stuart2015; Hemphill and Kulik, Reference Hemphill and Kulik2016; Bredgaard, Reference Bredgaard2018; Frøyland et al., Reference Frøyland, Andreassen and Innvær2018; Hyggen and Vedeler, Reference Hyggen and Vedeler2021; van Berkel, Reference van Berkel2021).
Lundborg and Skedinger (Reference Lundborg and Skedinger2016) and Bredgaard (Reference Bredgaard2018) find that employers generally exhibit positive attitudes toward the notion that employers should be socially inclusive. However, these positive attitudes are often not reflected in their behaviour (for instance, by having disabled or refugee employees). A discrepancy between behaviour and attitudes is thus often found.
Some employers are motivated to be socially inclusive by a sense of social responsibility (Snape, Reference Snape1998; Coleman et al., Reference Coleman, McGinigal, Thomas, Fu and Hingley2014; Bredgaard, Reference Bredgaard2018). In the model, we therefore include a variable that measures whether or not the respondent believes that employers have a social obligation to employ refugees.
In contrast, some existing literature finds that some employers are motivated to be socially inclusive by self-interest (Snape, Reference Snape1998; Coleman et al., Reference Coleman, McGinigal, Thomas, Fu and Hingley2014) – for instance, to avoid labour shortages or to be able to pay low wages. We therefore include two additional variables. One measures whether or not employers believe that employment of refugees can help avoid labour shortages, and the other measures whether or not employers believe the wages specified by collective agreements are too high to make employment of refugees worthwhile.
Furthermore, preconceptions are at the heart of the literature examining discriminatory recruitment practices. In this regard, the terms statistical discrimination and stereotypical discrimination are important (Midtbøen, Reference Midtbøen2016; Dahl, Reference Dahl2019). It refers to the fact that discrimination might occur because an employer believes that a particular refugee jobseeker might have undesirable attributes because of the employer’s negative perceptions of refugees in general (e.g. that refugees will be unmotivated to work more often than the general population, that they lack relevant work experience, that they suffer from health-related problems, etc.). Like the general population, employers form stereotypes and preconceptions about refugees through a variety of factors such as media and political discourse, family, friends and co-worker attitudes as well as prior experience in some instances. As demonstrated in the review, the typical supply-side barriers for refugees were a lack of relevant work experience and qualifications (education), health-related and social problems, and in some instances, a lack of incentives (motivation) to obtain employment. We therefore include four variables in our model measuring each of these preconceptions about refugees.
Case, data and methods
As described above, the role of employers in relation to employment of refugees is a relevant area of study, as they face a number of supply-side, demand-side and matching barriers in relation to obtaining employment. Denmark is a particularly interesting country to study in this regard, as it has a low degree of regulation concerning dismissal and employer recruitment decisions. This supports a flexible labour market with many job changes among employees and in which employers can recruit and dismiss whom they want (Bredgaard et al., Reference Bredgaard, Larsen and Madsen2006). Furthermore, Denmark has high minimum wages in place through collective agreements, as well as demand for skilled labour (Brochmann and Hagelund, Reference Brochmann and Hagelund2011), which puts refugees at a disadvantage.
Furthermore, Denmark has worse integration outcomes in terms of the share of refugees in employment and education five and seven years after being granted residency, as compared with its Scandinavian neighbours (Hernes et al., Reference Hernes, Arendt, Joona and Tronstad2019). As such, Denmark can be considered a ‘hard case’ in terms of labour market integration of refugees.
Our study is based on a national and representative survey among Danish workplaces with at least three employees. A net sample of 5,000 public and private workplaces was drawn from the CVR-register, which includes all Danish businesses and firms. The cut-off point of at least three employees was chosen to increase the likelihood of obtaining a sample that could, in theory, include employees with a refugee background by excluding very small businesses and the self-employed.
The survey was conducted in June 2019. The target group was managers responsible for recruitment or the employers themselves. Invitations to participate in the study were sent by email, and potential respondents were given the complete survey online through a link, or they could opt to receive a phone call and be guided through the questions by an interviewer. To increase the response rate, two email reminders were sent to the potential respondents who had not yet answered the survey. The final response rate was 42.9 per cent (1,977 completed responses). Analyses of non-response and dropout rate showed no clear patterns.
Missing values on variables have not been imputed. Instead, observations with missing values have been dropped from our analyses, reducing the number of valid observations by roughly 300.
In order for us to examine which workplace factors are associated with employment of refugees, we performed a series of logistic regression analyses with the dependent variable measuring if the workplace currently employs at least one person with a refugee background.
The term person with a refugee background is used in the survey (and in the analyses of the paper) and needs some elaboration in terms of validity and reliability. The participants in the survey were presented with the following definition of a refugee prior to answering questions about refugee employees:
By refugees we mean people who have left their home country due to war or persecution and have been granted asylum and are living in Denmark. The group does not include people from, for instance, other European countries who have come to Denmark to work. A person who has been family reunified to a refugee is someone who has come to Denmark because a close relative is a refugee living in Denmark.
We furthermore stated that respondents should think of both refugees and people who had been granted family reunification to a refugee when answering questions about people with a refugee background. As such, our questions measures both legal status refugees as well as people who have been family reunified to a refugee.
Related to this, it should also be discussed for how long people would be regarded as having a refugee background. For instance, would an adult who was a refugee as a child be regarded as someone with a refugee background by the respondents? The short answer is no, since we believe that the definition above states only people with a temporary residence permit (due to asylum or family reunification) should be included. In the Danish case, they can have this status for a rather long time. To be eligible for permanent residency one has to have resided legally in Denmark for eight years and fulfil other rather strict requirements.
It could also be argued that employers and HR personnel might not know if some of their employees have a refugee background. This might primarily be an issue at large workplaces where employers are not familiar with every single employee and thus uncertain if they have a refugee employee.
To accommodate this uncertainty, we included a ‘don’t know’ option to the question. Only three per cent of the respondents answered ‘don’t know’ which we interpret as evidence that the vast majority knows whether or not they have employees with a refugee background. The ‘don’t know’ answers have been dropped from the analyses in the process outlined earlier in this section.
Going back to the regression analyses, odds ratios and coefficients in logistic regression analyses can be difficult to interpret. We therefore present the average marginal effects of the variables in the regression models.
In interpreting our results, it is important to state that we do not claim to find isolated causal effects for any of our variables. Instead, our claim is more modest. We find associations when controlling for factors, which can support (or can contradict) hypotheses found in the existing literature. Nevertheless, we may approximate causality in some instances, as we have included a large number of covariates known to affect workplace practices.
Results
Before presenting the results of the regression model (Table 1), it is important to state that 13.4 per cent of the employers in the survey have a refugee employee at the workplace. Thus, the absolute likelihood of having refugee employees is rather low.
Table 1 Logistic regression. Having a refugee employee. Marginal effects. Delta method standard errors in parentheses
*** = significant at a 0.001 level, ** = significant at a 0.01 level and * = significant at a 0.05 level.
In Model 1, we add the covariates that measure the ‘objective’ workplace characteristics (recruitment of new employees during the last year, sector, size of the workplace and membership in an employer association). All variables in the model are statistically significant, except for public/private sector. We interpret the marginal effect of 0.06 for the variable measuring recruitment of new employees as follows: when controlling for other variables in the model, having recruited a new employee over the last year, which reflects a workplace in demand of labour, increases the likelihood of having at least one refugee employee by six percentage points, compared with not having recruited new employees. Furthermore, the number of employees at the workplace also is statistically significant. The marginal effect of 28 percentage points for workplaces with seventy-five or more employees is rather large. Logically, this makes sense, as large workplaces have a higher capacity to on-board and train a new employee with a refugee background. For large workplaces, having one employee with a refugee background constitutes a low share of the total number of employees. As noted above, the results in Model 1 are based on observable factors. Employment service workers might increase their chances of success if they target their efforts towards larger workplaces and workplaces that have recently recruited new employees (i.e. workplaces that are in demand of labour) and workplaces that are members of an employer association.
In Model 2, we add the three covariates that are based on the ‘matching’ perspective. All of the variables are significant, and they remain significant throughout the models. Workplaces that use the PES for recruitment purposes (compared with workplaces that never use the PES) are more likely to employ refugees. If a workplace has been contacted by the PES, either with the purpose of employing a refugee or getting a refugee in workplace training, this increases the likelihood of having at least one refugee employee. It would appear that proactive outreach to employers by the PES can help better the labour market integration of refugees.
In the third model, we add the covariates measuring existing and prior social responsibility practice. These variables are significant and stay statistically significant throughout all permutations. The first variable measures the number of (other) disadvantaged groups that are employed at the workplace. If the workplace has employees coming from two (or more) other disadvantaged groups, the likelihood of having a refugee employee increases. The same is the case if the workplace has previously (also) had refugee employees. Social responsibility practice thus also seems to matter when it comes to employment of refugees.
In the fourth and final model, we add the covariates that measure employer attitudes and preconceptions. Here, we find that employers who believe they have a social obligation to employ refugees are five percentage points more likely to do so, as compared with employers who do not believe they have such an obligation. This holds true when controlling for all covariates in model 4. The variable measuring employer preconception that refugees are (or are not) motivated to work becomes significant. The statistical effect is surprisingly large. Employers who believe (compared to those who do not believe) that refugees are motivated to work are eleven percentage points more likely to employ refugees. In relation to the other newly added variables, we believe it is an important finding that none of these is significant. Controlling for other workplace factors and employer attitudes and preconceptions, it is not important whether or not employers believe that employment of refugees can help avoid shortages of labour, if they believe refugees have inadequate qualifications or work experience or if they believe that refugees have health-related or social problems. Nor does it matter whether or not employers believe the wages in the collective agreements are too high to make employment of refugees worthwhile.
Dwelling on the other variables that remain statistically significant, we see that the size of the workplace, the variables from the ‘matching’ perspective and the variables measuring social responsibility practice remain statistically significant, even when taking employer attitudes and preconceptions into account. Thus, these factors seem to matter for employment of refugees.
The associations that are ‘controlled away’ are furthermore interesting to dwell on. Here, when including the other covariates, we find no statistical effect of having recruited new employees over the last year or membership in an employer association. Nevertheless, for the daily practices of employment services, staff can still target these types of workplaces, as they are significant in Model 1, which includes only observable characteristics.
Conclusion and discussion
In this article, we have explored which workplace factors among Danish employers are associated with employment of refugees. We have furthermore investigated hypotheses derived from the literature on the ‘matching perspective’ and the role of employers (the demand-side) in relation to employment of refugees. In addition, we have explored which employer preconceptions about refugees are related to employment of refugees.
In relation to the matching perspective, our most important finding is that being contacted by the PES is associated with employment of refugees. This offers hope that the placement efforts of the PES are not in vain and that a proactive approach towards employers can contribute to the labour market integration of refugees. Existing research has found that workplace inventions are effective in relation to getting disadvantaged groups into employment (e.g. Card et al., Reference Card, Kluve and Weber2018). However, the isolated effect of employers being contacted and asked if they would like to employ a refugee is, to our knowledge, not yet explored. Theoretically, insofar as an effect exists, this effect could be caused by employers not being aware of the possible of recruiting refugees; or how to recruit them. Further research could be conducted to explore this. In addition, if employers use the PES as a recruitment channel, the likelihood of employing refugees is higher. Our findings also suggest that the PES could potentially target their efforts towards larger workplaces and employers who already employ other disadvantaged groups.
In relation to the demand side perspective, we surprisingly found that having recruited new employees over the last year, sector, collective agreement coverage and membership in an employer association were not associated with employment of refugees after controlling for the more unobservable covariates. However, taking the results from Model 1 at face value, employment services might improve the placement efforts if they target larger workplaces, workplaces that have recruited new employees and workplaces that are members of an employer association.
Our most important findings in relation to the demand side literature are that, in relation to employment of refugees, socially responsible attitudes, and some employer preconceptions, matter. We found a positive relationship between positive attitudes (a belief that employers have a social responsibility to employ refugees) and employment of refugees. A belief in social responsibility increases the likelihood of having refugee employees. Existing social responsibility behaviour in the form of employment of other disadvantaged groups is also positively related to employment of refugees. In other words, social responsibility matters for employment of refugees. Theoretically, attitudes and prior behaviour may be mutually reinforcing.
Furthermore, having previously employed (other) refugees strongly increases the likely employment of refugees currently. To the existing literature, we add and underpin that prior experience (with refugees and other disadvantaged groups) to some extent shapes current practice. Our results suggest that getting the first refugee employee is the hardest obstacle to overcome. This could indicate that if the PES were to persuade employers to recruit one refugee, it could be easier to persuade them to on-board additional refugees in the future (also confer our findings in relation to the matching perspective). Additionally, the more socially responsible employers are, the more prone to use the PES as a recruitment channel. In relation to our finding that contacts by the PES are associated with employment of refugees, a preliminary policy recommendation could be that the PES should act even more proactively and reach out to more workplaces.
Previous research has also found that employer notions of social responsibility, wage concerns and prior experience are important in relation to employment of other disadvantaged groups (see, for instance, Coleman et al., Reference Coleman, McGinigal, Thomas, Fu and Hingley2014; Lundborg and Skedinger, Reference Lundborg and Skedinger2016; Bredgaard, Reference Bredgaard2018). We may add that this is also the case in relation recruitment of refugee employees. We may also add to the literature that some employer preconceptions are of importance. This is the case for the positive preconception that refugees are very motivated to work. In fact, it is one of the variables most strongly associated with employment of refugees. We do, however, also find that most of the variables measuring employer preconceptions are not statistically significant. This could either be because these variables do not affect employment of refugees, or this may be attributable to social desirability bias (Kreuter et al., Reference Kreuter, Presser and Tourangeau2008), which means that employers who are prejudiced against refugees will not signal this in normal survey questions. The causal effects of preconceptions and prejudice against refugees could be explored in future research – for instance, through survey experiments.
In sum, our empirical and theoretical contributions consist of finding that many factors and concerns are at play (attitudes towards social responsibility, prior experience, existing social responsibility practice, size of the workplace, using the PES as a recruitment channel and being contacted by the PES). The social reality is complex, and employment of refugees is no exception.
