The scarring effects of employment bans for asylum seekers
Policy interventions in times of COVID-19 should aim to shield citizens – especially the most vulnerable ones – from the dramatic socio-economic consequences of drastic public health measures (Furman 2020). Its unequal impact on populations is one of the numerous unsettling aspects of this pandemic. Pre-existing disparities in society, indeed, have been starkly highlighted over the last few months as differences in income, employment contracts, ethnicity, and housing conditions are heightening health hazards (Borjas 2020) and economic uncertainty among already weaker groups (Adams-Prassl et al. 2020, Midões 2020, Mongey et al. 2020).
For migrants, the pandemic has increased their vulnerability (World Bank 2020) as well as emphasised the fundamental role they play as key workers in our societies (Fasani and Mazza 2020). Both factors suggest that swift action is needed on migration policy (Foresti 2020).
Some reforms may be easier than others, however. Removing restrictions, for instance, requires far less political effort than designing and implementing new policy measures.
Employment bans in Europe
Recently defined by The Economist as restrictions that “please nobody” (The Economist 2019), bans on asylum applicants from taking up legal employment while their cases for protection are being considered are a persistent and widespread feature of Western countries’ asylum policies (Clemens et al. 2018).
In 2015, at the peak of the European refugee crisis, only four European countries (Greece, Norway, Portugal and Sweden) allowed asylum seekers immediate access to labour markets. Most other countries impose bans of between 2 and 12 months or, in the case of Ireland and Lithuania, even an indefinite restriction. In the US, asylum seekers must wait 6 months from the date of application before they can receive a work permit, with the current administration putting forward proposals to increase such length.
Evidence from past waves of forced migration in Europe (Fasani et al. 2018, Brell et al. 2020) suggests that refugees face significant hurdles to socio-economic integration and underperform in the labour market relative to both comparable natives and non-forced migrants. If initial difficulties in integration are inherent to refugee migration, asylum policy should aim to minimise those hurdles, rather than adding new ones.
In a recent paper (Fasani et al. 2020), we investigate the medium and long-term effects on refugee labour-market outcomes of the temporary employment bans that refugees were exposed to when they arrived in Europe.1 To do so, we use a newly collected dataset covering almost 30 years of employment restrictions, together with individual data from the European Labour Force Survey for refugees entering European countries between 1985 and 2012.
The geographical and temporal variation in employment bans generated by their staggered introduction and removal, coupled with frequent changes at the intensive margin (i.e. in ban length), allows us to adopt a difference-in-differences framework to identify the causal effect of employment bans on refugees’ outcomes.
The medium and long-term impact on refugee outcomes
Restricting the employment of asylum seekers has the obvious effect of keeping them out of the workforce for the duration of the ban. What happens later, once the ban is lifted, is however not clear: does the initial restriction have persistent effects that last when restrictions are lifted?
We find that being banned (i.e. exposed to an employment ban) at entry reduces employment probability in the medium run by 8.9 percentage points, or 15.2%. The negative effect is explained primarily by the lower labour market participation of banned refugees (by 9.2 percentage points) rather than by a higher probability of being unemployed (Figure 1, blue diamonds).
These results suggest that the primary effect of bans is to discourage refugees from actively searching for jobs. The effect is large, being quantitatively equivalent to about a 4-year delay in the integration process of refugees.
Given that our sample excludes refugees who may still be subject to employment restrictions, this effect is in no way a mechanical one. Remarkably, when we conduct a placebo analysis on a sample of non-refugee migrants that closely resemble the refugees, belong to the same arrival cohorts, and reside in the same countries but were not subject to the employment bans, we find zero estimated effects (Figure 1; blue dots).
Figure 1 Effect of employment bans at arrival on future labour market outcomes
Notes: The figure reports estimated coefficients from DID regressions of current labour market outcomes of refugees (blue diamonds) and other migrants (green dots) on an employment ban dummy (which takes value one equal if an employment ban was in place in the destination country at the time of arrival), individual controls and fixed effects. See Fasani et al. (2020) for details on the estimation procedure.
The detrimental effects of employment bans are highly persistent, confirming the evidence from the non-migration literature on scarring effects of entering the labour markets under unfavourable conditions (Schwandt and von Wachter 2019, Rothstein 2020).
As shown in Figure 2, the ban has a 24-percentage-point negative effect on employment probability in the first 2–4 years post entry, which decreases to 19 percentage points after 5–7 years, and 8 percentage points after 8–10 years in the country (Panel A). This latter coefficient, like those for refugees with more than 10 years in the host country, is not statistically different from zero.
Figure 2 Effect of employment bans by years since migration
Notes: The figure graphs the estimates and 95% confidence intervals for the ban effect on employment, participation and unemployment probability by years since arrival. See Fasani et al. (2020) for details on the estimation procedure.
The effect on participation (panel B) follows a very similar pattern, whereas the effect on unemployment (Panel C) is short-lived: it is relatively large and marginally significant for refugees with the shortest duration of residence (2–4 years) but quickly converges to zero for other groups.
Our results are robust to several identification tests. We first rule out the possibility that changes in bans correlate with factors that directly affect refugees’ future integration outcomes, such as economic conditions or other country-specific shocks at the time of entry.
We then provide evidence in favour of the parallel trend assumption, finding no effect of employment bans just before their introduction or right after their removal. Our results are also not explained by refugees sorting into employment bans – by selectively choosing timing and destination of their migration.
Lastly, we implement an instrumental variable strategy that exploits a 2003 EU Directive limiting the maximum employment ban duration in EU countries and obtain two-stage least squares estimates that fully confirm our main findings.
In the second part of the paper we investigate the mechanisms behind our key results. First, the concentration of the detrimental effects of employment restrictions is among less-educated refugees, suggesting that such bans mainly harm migrants whose employability in host countries is already relatively limited.
Banned refugees also experience lower occupational quality (lower likelihood of employment in a high-skilled occupation and higher probability to have a temporary job), report lower proficiency in the host-country language, and have more health issues and greater likelihood of receiving benefits.
These effects are non-linear in ban length. Although exposure to a longer ban (13 months or more) has a slightly larger negative effect than exposure to a shorter one (up to 12 months), the difference is not particularly pronounced. These suggest the most relevant margin at which a ban policy operates is the extensive rather than the intensive.
These are compatible with the fact that initial exclusion from the labour market may severely lessen incentives to make early investment in host-country specific human capital (e.g. language), reduce interactions with natives, and hinder the development of networks, thus slowing down assimilation and increasing isolation.
At the same time, being placed on welfare immediately upon arrival may generate a culture of welfare reliance, leading to lower motivation to engage in the labour market.
Should we lift the bans?
The existence of employment bans for asylum seekers causes a significant economic loss for the host countries. To get a sense of the magnitude of such losses, we quantify the cost of imposing employment bans on asylum seekers who arrived in Europe during the so-called refugee crisis (2015−2016).
Even abstracting from the mechanical effect of employment bans, thereby ignoring asylum seekers’ forced idleness during the first months in the host country, we show that the ban imposed on over 1 million new refugees arrived in those years may have resulted in an overall output loss of €37.6 billion over an 8-year period, equivalent to about €4,100 per banned refugee per year.
Based on this evidence, host-country governments should carefully weigh the (alleged) benefits of such bans against their longer-term costs for both refugees and the host-country economy. Whereas allowing asylum seekers some months to recover physically and psychologically may be beneficial for their long-term integration, forcing them out of the labour market seems an ill-conceived approach. On the contrary, allowing early labour market access is an easily implementable and financially costless policy that effectively accelerates refugee integration.
Unsurprisingly, some European countries have either decided (Belgium) or debated (Germany) lifting the bans in the midst of the pandemic crisis. Although their motivations may be more related to concerns about possible labour shortages in the agriculture sector due to travel restrictions rather than to refugees’ wellbeing, these developments nevertheless constitute a step in the right direction.
Rothstein, Jesse (2020), “The lost generation? Labor market outcomes for post Great Recession entrants”, mimeo, University of California Berkeley.
Schwandt, Hannes, and Till Von Wachter (2019), “Unlucky cohorts: Estimating the long-term effects of entering the labor market in a recession in large cross-sectional data sets”, Journal of Labor Economics 37: 161–98
The Economist (2019), “Britain’s ban on jobs for asylum-seekers pleases nobody”, 19 September.