By Alessandro Caiumi (University of California, Davis)
One out of seven international migrants today is someone who was forced to leave their country of origin because of conflict, persecution, or insecurity. Forcibly displaced migrants, the vast majority of whom are refugees, experience persistent labor market gaps at destination relative to other workers, greatly limiting their integration prospects (Bratsberg, Raaum, and Roed, 2017; Brell, Dustmann, and Preston, 2020; Fasani, Frattini, and Minale, 2022). While policy efforts and program evaluations focus on whether refugees obtain a job, less attention has been paid to who provides that job. When refugees arrive, resettlement policy often restricts where they are allowed to live. In practice, this also restricts which employers they can realistically access. Still, as firms differ widely in productivity, training, management practices, and pay, the first employers a refugee connects with may matter for their long-term success at destination as much as formal integration policies.
This paper provides the first quasi-experimental evidence on the effect of early employer quality on refugees’ labor market outcomes. Using Danish administrative data and a policy that quasi-randomly assigned refugees across locations, we estimate a proxy for employer quality and show that early access to higher-quality employers has long-lasting effects on refugees’ employment probability and earnings, even several years after arrival. This result implies that the type of employers available upon refugees’ first entry is an important determinant of their long-term success at destination, with important policy implications.
A critical entry point
Entry conditions in the labor market matter for all workers (von Wachter, 2020). For workers born and raised in a country, however, the first employer may often be a temporary stop. Natives possess local skills and connections, know the labor market well, including how and where to search for other jobs, face fewer barriers and can move freely. On the other hand, refugees arrive with limited local knowledge, uncertain credential recognition, language barriers, and thin networks. These conditions make the first workplace a much more important point of entry into a new labor market: it carries greater potential for learning, acquiring context-specific human capital, and building professional connections.
To estimate employer quality, we rely on matched employer–employee data and a two-way fixed effects wage model to isolate firm wage premia (Abowd, Kramarz, and Margolis, 1999). Essentially, this methodology estimates which firms systematically pay higher wages than others for similar workers, returning a ranking of employers. While firms differ along many dimensions beyond pay, we follow a large empirical literature in using these wage premia as a summary measure of employer quality, after validating them against other observable measures of firm productivity and desirability (Kline, 2024).
The paper then asks: what is the impact of the quality of refugees’ first accessible employers on their subsequent labor market outcomes? And how can policymakers leverage our insights for a better resettlement of refugees? A key reason for the lack of evidence is that endogenous sorting and selection are pervasive in employer–employee matching, making the causal effect of employers on employees difficult to identify.
Identification: Dispersal policies and co-national networks
Denmark offers an advantageous setting to answer this question. First, the country has accepted a large number of refugees from many origins over the past four decades. Second, the country’s unusually rich administrative data allow us both to estimate firm quality with precision and to track individual refugees over a 15-year period following arrival, providing a longer horizon than most previous studies. Finally, the implementation of a national dispersal policy helps address concerns about worker nonrandom sorting, which has long complicated the estimation of causal effects.
Between 1986 and 1998, newly arrived refugees could not choose their initial location but were resettled across municipalities by Danish authorities. Conditional on basic and observable demographic characteristics such as family size and origin, initial placement was as good as random, and it was not based on economic characteristics or preferences of refugees. Our identification approach combines this policy experiment with the role of co-national networks — as migrants rely heavily on people from the same origin when navigating an unfamiliar labor market — to obtain exogenous variation in refugees’ exposure to the quality of first potential employers. The thought experiment compares refugees who, given the municipality of initial assignment and their co-national networks there, are randomly exposed to different early employers.
Results: Early employer quality affects long-term success
High-quality early employers significantly shape refugee integration. Refugees who are placed in a municipality where their co-national networks are employed by higher-quality firms experience improved employment rates and earnings for up to ten years after arrival (Table 1, Panel A). These effects are economically meaningful: magnitudes are approximately one fourth to one seventh of those documented for other successful, but more costly interventions, such as language training or active labor market policies. According to the estimates, exposure to better employers alone would reduce by 5% the native–refugee earnings gaps ten years after arrival.
Just as important is what does not drive the result. Being placed in an area with higher average firm quality, but without co-national network connections, does not produce similar gains, as the effect appears when refugees have network connections to better employers, not when good firms are simply present but potentially out of reach (Table 1, Panel B). Importantly, these gains do not come from other potentially correlated local characteristics, such as network size and local employment rate. We also show that quality of employers, rather than quality of members of co-national networks, drives our results, and conclude that networks are important friction-reducing mechanisms for migrants.
Indeed, two main mechanisms help explain why firms employing co-nationals are relevant: job referrals and information sharing. Consistent with these channels, we find that refugees are significantly more likely to work at a firm where a co-national is already employed and, consequently, to access higher-quality firms themselves. This channel explains roughly the 60% of the main effect documented for earnings.
Why this matters for policy
Many refugee placement systems are designed to spread arrivals across regions to share fiscal costs and avoid crowding. Our paper suggests that considering labor demand, specifically which employers refugees are likely to reach early on, can be a cost-efficient way to improve integration while still respecting policymakers’ priorities and constraints regarding municipality quotas and the availability of support services. This suggestion does not imply resettling everyone in the most economically active cities; instead, it amounts to using local co-national networks to inform placement in ways that improve access to higher-quality employers.
Building on these insights, the paper extends a data-driven assignment algorithm that optimally matches refugees to municipalities while respecting capacity limits (Bansak et al., 2018; Ahani et al., 2021). The approach combines individual and local characteristics with network measures, including the quality of accessible employers, to predict employment success after arrival. Compared with the random dispersal policy, our optimized assignment increases predicted short-run employment probability by about 46%. This gain arises from leveraging origin–location complementarities to favor access to employers and remains stable across counterfactual policy scenarios (Figure 1).
Taken together, this evidence suggests that integration also depends on firms. For refugees in a new labor market, early employers are especially valuable: a strong first employer can help offset early disadvantages, while a weak one can reinforce them. From a policy perspective, incorporating labor demand into placement decisions for refugees is a promising direction. Algorithms like ours can be integrated into software tools and made available in real time to resettlement authorities, reducing inefficiencies common in manual assignment.
About the Author
Alessandro Caiumi a Ph.D. Candidate in the Department of Economics at the University of California, Davis.
His research interests are in labor and public economics, with a focus on migration. To learn more about his work, visit his website: https://www.alessandrocaiumi.com/
Social Media Handles:
