Occupational Recognition and Immigrant Labor Market Outcomes Herbert Brücker, Albrecht Glitz, Adrian Lerche, and Agnese Romiti

Occupational Recognition and Immigrant LaborMarket Outcomes

Herbert Brücker Albrecht Glitz

University of Bamberg Humboldt University Berlin

Institute for Employment Research Universitat Pompeu Fabra & Barcelona GSE

Adrian Lerche Agnese Romiti

Universitat Pompeu Fabra Institute for Employment Research

January 2016

AbstractIn this paper, we analyze how the formal recognition of immigrants’ foreign occupationalcredentials a�ects their subsequent labor market outcomes. Our empirical analysis isbased on a novel German data set that links respondents’ survey information to theiradministrative records, allowing us to observe immigrants at monthly intervals before,during and after their application for occupational recognition. To address the inherentissue of self-selection into the application process, we focus on a sample of immigrantswho eventually all receive full recognition, and exploit variation in the timing of therecognition process. As a robustness check, we further employ a pooled synthetic con-trol group approach, matching workers who obtain full recognition to control groups ofworkers with similar pre-recognition labor market trajectories. Our empirical findingsshow substantial employment and wage gains from occupational recognition, both in theshort and in the long run. After four years, immigrants who obtained full recognitionare around 45 percentage points more likely to be employed and earn around 40 percenthigher hourly wages than immigrants without recognition. We show that the increasein employment is driven almost exclusively by a higher propensity to work in regulatedoccupations. Relating our findings to the relative earnings position of immigrants inGermany, we finally show that occupational recognition leads to a substantially fasterconvergence of immigrants’ earnings to those of their native counterparts.

Keywords: Occupational Recognition, Immigrants, Labor Markets

JEL Classification: J15, J24, J44, J61

1

1 Introduction

It is a well documented fact in most developed economies that immigrants perform sig-

nificantly worse in the labor market than their native counterparts (see, e.g., Dustmann

and Frattini, 2013). In many cases, the main reason appears to be a lack of human cap-

ital, which pushes immigrants into low paying and precarious jobs and prohibits them

from moving into more desirable segments of the labor market. What is more, even if

immigrants accumulated valuable skills in their country of origin prior to migration, the

transferability of these skills to the host country economy is often problematic, partly

because of insu�cient language skills (Chiswick and Miller, 2003) but partly also because

legal restrictions make it di�cult for immigrants to work in certain occupations. While

the safeguarding of quality standards is the main rationale for such restrictions, their

strict enforcement often leads to an under-utilization of immigrants’ skills, as reflected in

their often observed occupational downgrading in the host countries (see, for example,

Friedberg, 2001, for Israel, Mattoo et al., 2008, for the US, and Dustmann et al., 2013, for

the UK).1 Gaining recognition for foreign occupational credentials is a way to potentially

overcome this ine�ciency, and even though it is a costly process, the economic benefits

are likely to be su�ciently large to make it worth its while. Yet, so far, there exists little

empirical evidence on the issue.

In this paper, we estimate the causal impact of occupational recognition on immi-

grants’ labor market outcomes. To get their foreign credentials recognized in Germany,

immigrants are required to go through a formal process, at the end of which, if success-

ful, the responsible authorities certify the equivalence between the immigrants’ foreign

qualification and its German counterpart. From a labor market perspective, occupational

recognition has two main e�ects. First, it reduces uncertainty about the skill set of im-

migrant workers, which allows employers to better screen in the hiring process, leading

to higher quality matches between workers and firms. Second, a successful recognition

gives the immigrants access to segments of the labor market that they could previously

not enter. These regulated segments are typically characterized by high wages, both be-

cause of high returns to skills and because of monopoly rents from occupational licensing1A related manifestation of the low transferability of human capital are the remarkably low returns

to foreign education and experience observed in many destination countries (see Dustmann and Glitz,2011, for a comprehensive overview of this literature).

2

(compare Kleiner and Krueger, 2010, 2013). Both e�ects suggest a positive impact of

occupational recognition on immigrants’ employment outcomes and wages, part of which

is driven by immigrants moving into regulated occupations.

Identifying the causal impact of occupational recognition is not straightforward due

to self-selection on the part of the immigrants. Presumably, those immigrants who obtain

occupational recognition would also perform comparatively well in the labor market if

they had not received it, conditional on other observable characteristics. This is because

having obtained recognition reflects a specific set of skills that is likely to be generally

valued in the labor market, both in the regulated and unregulated segment. In addi-

tion, immigrants who decide to go through the costly application process are likely to

di�er from those who do not in terms of unobservable characteristics such as ambition

and motivation, factors that on their own would be associated with better labor market

outcomes. We deal with these issues by exploiting a novel German data set that links

detailed survey information on the exact timing of the application process for recognition

with comprehensive social security data on the respondents’ entire work histories in Ger-

many. Our starting point is an estimation sample that consists exclusively of immigrants

who eventually all obtain full recognition for their foreign credentials. To identify the

impact of recognition, we take advantage of the di�erential timing at which applicants

receive their final decisions, following an approach similar to Bratsberg et al. (2002) and

Arai and Thoursie (2009). Our estimates are thus derived from comparisons between

labor market outcomes of individuals with full recognition and individuals who applied

but have not yet received their final decision. We estimate both average e�ects of occu-

pational recognition and dynamic impacts which reflect the evolution of the employment

and wage e�ects over time. As a robustness check for our dynamic estimation, we also

use a pooled version of the synthetic control method proposed by Abadie et al. (2010),

which leads to similar estimates of the dynamic e�ects of occupational recognition.

Our findings show substantial positive e�ects of occupational recognition on employ-

ment and wages. On average, immigrants in our sample who obtained full recognition are

26.7 percentage points more likely to be employed and earn 13.2 percent higher hourly

wages than comparable immigrants who are still in the application phase. In addition,

successful immigrants are 19.0 percentage points more likely to work in a regulated oc-

3

cupation while the e�ect on employment in unregulated occupations is much lower in

magnitude and statistically insignificant. These results are robust to a variety of alter-

native sample selection rules. Turning to the dynamic process underlying these average

e�ects, our estimates show that the probability of being employed relative to the control

group increases rapidly with the receipt of occupational recognition, reaching 30.1 per-

centage points within the first twelve months. In the following years, the employment

gap continues to widen, though at a lower pace, reaching a value of around 45 percent-

age points after three years. The wage gains from occupational recognition take a little

longer to materialize but start increasing steadily from the second year after recognition

onwards, reaching 40 percent four years after recognition. Finally, the relative shift into

the regulated segment of the labor market starts directly after recognition and continues

at a steady rate, reaching a relative gap of 45 percentage points three years after ob-

taining recognition. In contrast, the impact on the propensity of working in unregulated

occupations remains close to zero throughout the observed time period.

Our paper relates to the literature on the economic assimilation of immigrants (see,

e.g., Borjas, 1995, or Lubotsky, 2007) in that it studies a specific mechanism through

which immigrants may be held back in the host country’s labor market. In compari-

son to this extensive literature, the evidence on the impact of occupational recognition

on immigrant labor market outcomes is scarce. One important aspect of occupational

recognition is that it helps overcome employment barriers to licensed occupations. In this

context, Kleiner and Krueger (2010) document that 20 to 30 percent of US jobs require

licensure or certification, making it likely that these types of labor market regulations

a�ect immigrants as well. Peterson et al. (2014) provide corresponding evidence by look-

ing at the migration decision of physicians in the US and find that states with stricter

licensing requirements for immigrants receive fewer foreign physicians. Regarding the

causal impact of occupational recognition on labor market outcomes, Kugler and Sauer

(2005) use an instrumental variable strategy which exploits the fact that Soviet trained

physicians who immigrated to Israel in the early 1990s were exogenously assigned to dif-

ferent re-training tracks that di�erentially a�ected the probability of eventually obtaining

a medical license. Their results show substantial monetary returns from obtaining a med-

ical license of the order of 200% of monthly earnings within 3 to 4 years after arrival in

4

Israel. Contrary to what one might expect, the di�erence between OLS and IV estimates

suggest a strong negative selection of Soviet physicians into the acquisition of medical

licenses, with the former estimates being about half of the latter in magnitude. Tani

(2015) looks at recognition processes in Australia and estimates the e�ects of the assess-

ment of o�cial post-secondary foreign degrees on wage outcomes using the Longitudinal

Survey of Immigrants to Australia. To deal with the endogeneity problem inherent to

the assessment process, he proposes an instrument that captures the higher incentives to

apply for recognition for immigrants with degrees from less known foreign universities.

The empirical findings suggest that a positive assessment directly after entry to Australia

leads to 40% higher wages within the first year and further increases in subsequent years.2

Even though the qualitative results of these studies are similar to ours, there are a

number of important di�erences. First, we look at a broader set of qualifications, includ-

ing post-secondary education as well as occupational training. Second, we use rich panel

data and apply an event-study type di�erence-in-di�erences strategy exploiting temporal

variation in recognition decisions, which under plausible assumptions gives rise to causal

estimates. In addition, we use a pooled synthetic control method as a second estimation

approach to corroborate our findings. Third, we extend the empirical analysis beyond

wages, studying employment and occupational mobility as additional outcome variables

to get a more complete picture of the labor market e�ects of occupational recognition.

Fourth, having complete work histories of immigrants, we estimate dynamic e�ects on a

monthly basis, thus identifying both short- and long-run e�ects of occupational recogni-

tion. Finally, the access to social security earnings data of natives allows us to interpret

our results in the context of immigrants’ earnings assimilation, providing a further piece

of evidence for an informed policy approach regarding the occupational recognition of

foreign credentials.

The paper is structured as follows. In the next section, we present the theoretical

background against which our empirical results can be interpreted. In Section 3, we then

sketch the institutional setting in which the occupational recognition process takes place2Complementing the literature on the role of occupational recognition on labor market outcomes,

Chapman and Iredale (1993) find that men who successfully apply for recognition in Australia earn 15to 30 percent higher wages than those who do not obtain recognition, while Houle and Yssaad (2010)analyze descriptive statistics for Canada and find that immigrants with higher education levels are morelikely to successfully apply for recognition.

5

in Germany. In Section 4, we present our empirical model and identification strategy,

which is followed by a description of our data set and summary statistics in Section 5.

Section 6 is dedicated to the discussion of our main results, which we then link to the

earnings assimilation of immigrants in section 7. Section 8 concludes.

2 Theoretical Background

The implications of recognizing foreign credentials can be illustrated most easily in a

framework of a two-sector economy, where final output is produced by two intermediate

inputs. The first input is produced by a high-productivity sector, h, where production

technologies require that workers have to exceed a minimum productivity threshold level,

yú. The second input is produced by a low-productivity sector l, where a productivity

threshold level does not exist. Product market regulation requires that firms can hire

only workers with approved credentials in sector h, which ensures that actual labor pro-

ductivity exceeds the minimum threshold level, i.e. that y > yú for all workers hired

by sector h. Before hiring workers, firms invest in sector-specific capital, ki, which in

turn determines production technologies and the subsequent skill requests. We assume

that kh > kl. Accordingly, workers decide whether to invest in the recognition of for-

eign credentials, which involves a certain cost, c. Instead of a two sector-economy one

might also consider the case of a labor market which is stratified by productivity and

skill requirements of workers into two segments.

Since firms cannot observe the true productivity of their workers, they form expec-

tations on their productivity based on the signals they receive. More specifically, the

expected average productivity level of workers who have achieved the recognition of their

foreign credentials is above that of workers who have not, i.e. E(y|r) > E(y|n), where

the index r denotes a worker who has an approved credential and n a worker who has

not.3 Similarly, workers form expectations about the returns of an investment in the

recognition of foreign credentials depending on the probability of becoming employed in3The way in which di�erences in expected productivities of workers with and without o�cal recog-

nition translate into di�erences in labor market outcomes may further be amplified by latent racialdiscrimination on the part of the employers. Experimental evidence by Dietz et al. (2009), for instance,shows that the credentials of black immigrants in Canada are being significantly less favorably evaluatedif the evaluators harbor latent racial biases, but that this negative bias disappears once the immigrants’foreign credentials have been o�cially accredited.

6

the respective sectors and the wages paid there.

The details of the implications of recognizing foreign credentials can be derived for-

mally from a search- and matching framework (e.g. Pissarides, 1990). The predictions

of such a framework in the steady-state are however straightforward so that they can

be outlined verbally: wages of workers whose foreign credentials are recognized and who

are employed in the high-productivity sector are higher than those of workers whose

credentials are approved as well but are employed in the low-productivity sector, and

the wages of the latter group of workers are higher than those of workers who do not

possess approved credentials, i.e. wrh > wr

l > wnl . Note that the higher capital endow-

ments in the high-productivity sector assure that wages of workers with approved foreign

credentials are higher compared to those of the same type of workers employed in the

low-productivity sector.

Assume moreover for convenience that the job destruction rate, the replacement rate

and the bargaining power of both types of worker are similar. Then it follows that steady-

state unemployment rates of workers with recognized foreign credentials are below those

of workers without approved foreign credentials, i.e. us > uh. This can be traced back

to the fact that workers whose foreign credentials have been recognized can also seek for

jobs in the high-productivity sector.

These considerations can be easily extended to the case where workers, who have

already applied for recognition but did not yet receive a formal decision, also send a

signal to the labor market. In this case, wages of workers who have applied for recognition

are higher compared to those who have not, but are below those who have successfully

applied. The steady-state unemployed rate of this type of workers is also below that of

workers who have not applied, but above the unemployment rate of those whose degrees

have already been approved. From an empirical point of view, however, it is di�cult

to systematically test these predictions since the true signalling e�ect from applying for

recognition is likely to be confounded by unobserved labor market shocks that incentivize

people to do so.

7

3 Institutional Setting

In regulated professions, occupational licensing is a precondition for access to the pro-

fession as well as for using the job title. Entering or practising regulated professions is

governed by legal or administrative provisions which require proof of specific professional

qualifications. Examples of regulated professions are medical doctors, psychotherapists,

nurses, lawyers, teachers, nursery school teachers and engineers. Only people who have

obtained the required professional qualifications are entitled to work in these professions

and use the professional title. In Germany, the main professions and occupations are

regulated at the national level. The individual federal states (Länder) are responsible

only for the regulation procedure of the following occupations: teachers, nursery school

teachers, youth social workers, engineers and architects. In contrast to regulated occupa-

tions, recognition is not a precondition for exercising unregulated occupations. Persons

may apply for work directly and work in these occupations without a state license and

without obtaining recognition for foreign vocational qualification. However, in case of

some unregulated occupations, the person can voluntarily decide to apply for having

the qualification assessed for equivalence with a German qualification. The notice of

equivalence received is an o�cial and legally secure document confirming equivalence of

the foreign qualification with the relevant German reference qualification. Examples of

unregulated occupations where the application is possible but not compulsory are the

so-called training occupations (e.g. o�ce management clerks, mechanics or electricians)

or advanced training occupations (e.g. master craftsman level qualifications in one of

the craft trades or in industry or agriculture, certified advisors, certified senior clerks,

specialist commercial clerks or business economists).4

In order to apply for and obtain recognition of the foreign qualifications, it is not

necessary to hold German citizenship or a residence permit for Germany. There is also

no need to be resident in Germany at the time of the application, so that they can be

submitted from abroad.

On 1 April 2012, a new law regulating the procedure of recognition was introduced,4All training occupations, i.e. occupations for which training takes place within the dual system, are

unregulated in Germany. In contrast, recognition is compulsory in order to work as a self-employed insome craft trades occupations where a license is required.

8

the Federal Recognition Act (Anerkennungsgesetz).5 Then aim of the law was to sim-

plify and standardize the procedure for the recognition of professional and vocational

qualifications obtained abroad governed by the Federal Law, in addition to opening up

such procedures to groups not covered by the previous legislation. Prior to the enact-

ment of the Recognition Act, the recognition of European professional and vocational

qualifications was regulated at the European level. In particular the legislation in place

was the EU Directive 2005/36/EC on the recognition of professional qualifications, which

came into force on October, 20th in 2005 and was introduced in Germany in 2007. For

individuals from third countries outside the EU, the EAA and Switzerland there was no

common o�cial procedure regulating the recognition process before the Regulation Act.

Therefore the country of origin played a decisive role in the result of the application, as

a legal basis was only available for EU-citizens and Ethnic Germans (Aussiedler). Ad-

ditional shortcomings in place before the introduction of the Recognition Act were that

no deadline was binding to process the application; therefore the procedure could be

quite lengthy.6 With the introduction of the Recognition Act, the duration of the pro-

cedure was regulated. The responsible authority begins the assessment procedure when

all documents have been received in completed form. A decision has to be made within

3 months of receipt of all documents. The decision deadline may be extended on a single

occasion in di�cult cases. The Recognition Act does not apply in the following areas:

recognition of occupations which are regulated at the federal state level (such as teachers,

nursery school teachers, youth social workers, engineers and architects), the recognition

of higher education qualifications which do not lead to a regulated occupation (e.g. math-

ematicians, chemists, economists), the academic recognition of studies and examinations

completed abroad within the context of admission to higher education, and the recog-

nition of school leaving certificates.7 The main authority responsible for the assessment

of the recognition are the Chambers, which as organized typically at the Länder level.

In some cases it is possible to apply to other local authorities. In our sample, 36 per-5Our sample was largely covered by the legislation in place before the introduction of the Federal

Recognition Act; only 7 percent of the sample applied thereafter.6According to our sample, before the introduction of the law the average duration between application

and result received is 4.96 months (with a standard deviation of 9.07 months) as opposed to 2.7 thereafter,with a standard deviation of 1.92.

7In our data set, we are not able to identify those applying for the recognition of school leavingcertificates, because the question only relates to post-secondary school certificates.

9

cent of the respondents apply to one of the Chambers, the rest to other local authorities

(Zeugnisanerkennungsstelle, Bezirksregierung).

After the assessment, in addition to denial, there are two outcomes possible: partial or

full recognition. The partial recognition is only possible in case of unregulated occupations

and it certifies if there are substantial di�erences between the foreign and the German

reference qualification. The assessment notice issued by the responsible authority includes

a detailed description of the existing vocational skills and of the knowledge which is

missing. This enables a new application to be made at a later point if appropriate.

In order to apply for recognition, the following documents are required: tabular sum-

mary in German of the training courses completed including previous occupational activ-

ity if relevant, proof of identity, proof of vocational qualification, proof of relevant occu-

pational experience, evidence of other qualifications (e.g. continuing vocational training

courses), a declaration of having not previously submitted an application, evidence of

intention to work in Germany (does not apply to nationals of the EU/EEA/Switzerland

and persons residing in the EU/EEA/Switzerland). All documents should be submitted

in German. Translations must be made by interpreters or translators who are publicly

authorized or certified. The procedure is subject to a charge. The fees for submitting an

application range from 100 to 600 Euros, depending on the occupation and the federal

state in which the application is submitted. The costs of fees and other costs for, e.g.

translations and certifications of documents must in all cases be borne by the applicants

themselves. In some circumstances and on an individual case basis, these fees may be

paid by other bodies (e.g. German Social Security Code II and III).8

Among immigrants who hold a foreign certificate and could therefore, in principle,

apply for occupational recognition, only 35 percent actually do so, according to our survey

data. Excluding people with a certificate who state that they do not need recognition, the

main reasons put forward for not applying are the belief of having no chance of obtaining

recognition (24 percent of the sample), the complexity of the system (23 percent), and

problems with the provision of the necessary documentation (7 percent). Monetary costs,8Prior to submitting an application, unemployed applicants or applicants registered as job seekers

should seek clarification from their local Employment O�ce or Job Centre whether the labour adminis-tration authorities will meet the costs of the procedure. The labour administration authorities will onlyprovide such support if it is of the view that recognition of a foreign training qualification is necessaryin order for the holder to be integrated into the labour market. Adaptation measures such as continuingtraining courses or examination preparations may also be funded in these cases.

10

in contrast, seem to constitute only a minor obstacle to applying (4 percent).

4 Empirical Framework

In our data set, we observe individuals receiving occupational recognition at di�erent

points in time. We use this variation to compare labor market outcomes of individuals

with recognition to those of individuals who have not yet received recognition at the

same point in time. We focus on a group of individuals who eventually all receive full

occupational recognition to minimize possible selection e�ects.

Following Bratsberg et al. (2002) and Arai and Thoursie (2009), we start with the

following fixed e�ects regression to obtain an overall estimate of the impact of recognition:

yit = —1CertRecogit + —2Appliedit + ⁄i + ⁄p + ⁄t + Áit, (1)

where yit denotes a specific labor market outcome of individual i in period t. In particu-

lar, we examine the impact of occupational recognition on an immigrant’s probability of

being employed, the log hourly wage rate, and the probability of working in a regulated

occupation and working in an unregulated occupation. The first two outcomes provide

general insights into the e�ects of occupational recognition on immigrants’ labor market

performance, and are important when viewed in the context of the rather poor employ-

ment and wage outcomes of immigrants documented in much of the migration literature

(for Germany, see, for example, Algan et al., 2010). The latter outcomes are more spe-

cific to our setup and provide an insight into the mechanism through which occupational

recognition a�ects labor market outcomes. We are interested in the di�erential movement

into regulated and unregulated occupations and therefore, provide results for both cases.

The main regressor of interest, CertRecogit, is a dummy variable taking the value

one if individual i has a foreign qualification that was recognized before or in period t,

while Appliedit is a dummy variable taking the value one if individual i has applied for

recognition before or in period t. We are interested in identifying —1, the causal e�ect of

occupational recognition on labor market outcomes, which would be straightforward if

participation in the recognition process was uncorrelated with the unobservable determi-

nants of these outcomes. However, applying for recognition is a necessary condition for

obtaining recognition and starting the application process is likely to be correlated with

11

changes in individual unobservables such as the preference for working or the incentive

to earn money. To deal with this fundamental issue, we include the dummy for having

applied, Appliedit, as another regressor in our specification. Conditional on applying,

the time until the decision depends on factors outside the control of the individual, such

as the workload of caseworkers in the recognition o�ce or their e�ectiveness in handling

applications. Thus, the date of decision is exogenous conditional on having applied,

which allows us to identify the causal e�ect of occupational recognition on labor market

outcomes.

To control for other confounding factors that may be driving individual labor market

outcomes, we additionally include a number of fixed e�ects in our empirical specifica-

tion. First, we include individual fixed e�ects (⁄i) that control for all time-invariant

characteristics of each immigrant. Their inclusion accounts for much of the personal

characteristics associated with better labor market outcomes and the selection into the

occupational recognition process. Furthermore, we add time (month ◊ year) fixed ef-

fects (⁄t) to account for general changes in labor market conditions, for example due to

seasonal variation or business cycle fluctuations. Finally, we include a full set of months

since migration fixed e�ects (⁄p) which capture the dynamic evolution of immigrants’

labor market outcomes as a result of their ongoing assimilation into the host country’s

economy.9

While specification (1) provides a useful summary measure of the average impact of

occupational recognition on employment and wage outcomes in our sample of immigrants,

it conceals valuable information about the dynamic process with which the e�ects of

recognition evolve over time. As an extension, we therefore introduce individual dummy

variables for the months around the recognition decision in an event-study framework,

which allows us to distinguish between short- and long-term labor market e�ects. More

specifically, we use the following regression model:

yit =48ÿ

q=≠12”t≠qCertRecogMthi,t≠q + ”t≠49CertRecogi,t≠49

+ —2Appliedit + ⁄i + ⁄p + ⁄t + Áit,

(2)

where the dummy variables CertRecogMthi,t≠q, which equal one if individual i’s qualifi-9For some robustness checks on the importance of these fixed e�ects, we refer to the Appendix.

12

cation was recognized in period t ≠ q, now capture the e�ect of occupational recognition

in specific months around the recognition date. We create these dummy variables start-

ing twelve months before the recognition date and ending 48 months thereafter. All

dummy variables are equal to one only in the relevant time period and zero otherwise.

For example, CertRecogMthi,t≠10 is equal to one when the successful recognition was ten

months before period t, so that the corresponding estimate ”t≠10 measures the e�ect of

recognition ten months after it was obtained. CertRecogi,t≠49 is a dummy variable for

individuals having a foreign qualification that was recognized before or in period t ≠ 49.

Thus, this variable picks up the long-run average e�ect of recognition on labor market

outcomes during all months more than four years after the recognition date. This setup

leads to a well defined reference period that includes all periods that are at least twelve

months before the recognition date. In these months, none of the just explained dummy

variables is equal to one and therefore, all these dummy variables measure the e�ect of

occupational recognition relative to the e�ect in this reference period.

Our dynamic setup and, in particular, the inclusion of separate dummy variables for

the twelve months prior to recognition also allow us to check our exogeneity assumption. If

the coe�cients on the occupational recognition dummies are to be interpreted as causal

e�ects on immigrants’ labor market outcomes, one should not observe any significant

estimates in months prior to the actual recognition date.

For both estimation models, we add a small set of time-varying control variables. We

include age squared in the spirit of Mincerian wage equations.10 We also use a proxy

of German language proficiency to capture to some extent di�erences in the assimilation

process of immigrants. Given the extensive use of fixed e�ects, we do not expect a large

influence of these control variables on our point estimates but they should help increase

their precision. We cluster our standard errors at the individual level, thus allowing the

error terms to be heteroscedastic and correlated over time for a given individual.

Finally, to check the robustness of our results, we use the same estimation model but

vary the sample of included individuals. Our baseline sample includes only individuals

who arrived in Germany at an age of at least 18 and who received full occupational

recognition at some point during the sample period. We also exclude individuals who,10We already implicitly control for linear age e�ects by including individual and time fixed e�ects.

13

at any point in their lives, were self-employed or worked in the government. In some

specifications, we further restrict the sample to individuals who are at least 25 years old

or migrated to Germany only once. By imposing these restrictions, we can avoid some of

the ambiguities regarding an immigrant’s employment status which arise in the register

data since we do not observe if a person is in school, migrated temporarily to another

country, or works in employment not covered by social security.

5 Data

The basis of our empirical analysis is a novel survey data set, the IAB-SOEP Migration

Sample11, which has been linked to the German social security data of the IEB (the

so-called Integrierte Erwerbsbiografie). The latter comprise the universe of workers in

Germany covered by the social security system.12 The IAB-SOEP Migration Sample is

restricted to individuals with a migration background, hence including both first and

second generation immigrants. More information on the survey and on the way it was

linked to the social security records is provided in the appendix.

The linked IAB-SOEP Migration Sample is particularly suited for our analysis for

two reasons. First, the survey component contains a detailed set of information on

occupational certificates obtained both before migration and after arrival in Germany. In

addition, there is a full module devoted to the process of requesting recognition of foreign

credentials, including information about the month and year when the application process

was initiated and the month and year when a final decision (denial, partial recognition,

full recognition) was obtained. Second, the social security component of the data allows

us to observe an immigrant’s entire work history after arrival in Germany. Linking the

information about the precise timing of the recognition process to the spell structure of

the registry data, we can observe each individual’s labor market outcomes before, during,

and after the application process at monthly intervals.

All our outcome variables are taken from the registry data. Employment represents an

indicator for having worked at least one day within a given month. Wages represent log

gross hourly wages of the longest working spell within the month and are right-censored11The survey has a panel structure and was started in 2013. In our analysis, we use both the first and

most recent second wave.12Civil servants, self-employed and military personnel are thus excluded.

14

at the social security contribution ceiling.13 In addition to wage and employment, we also

use as additional outcomes indicator variables for working in a regulated or unregulated

occupation. In order to be entitled to work in a regulated occupation (such as doctor,

pharmacist), the formal recognition of foreign credentials is required. To classify occupa-

tions as regulated or unregulated, we proceed as follows. Each 8-digit occupation in the

German system is o�cially classified as regulated or unregulated. We follow the approach

implemented by Vicari (2014) in which each 3-digit occupation is assigned an index that

measures the extent to which formal recognition is required using the full IEB registry of

2012. Since occupations in the registry data are aggregated to the 3-digit level, we first

transform the 8-digit classification into a 3-digit classification, then we link the index to

the data. The index represents the share of 8-digit subcategories in each 3-digit category

that require the recognition of foreign credentials, thus ranging from zero (no subcate-

gories requiring recognition) to one (all subcategories requiring recognition), where each

8-digit occupation is weighted by its relative size among the working population. We use

this continuous index to construct an indicator for regulated occupations, whose cut-o�

value is determined by the median in the occupation distribution of the IAB-SOEP Mi-

gration survey, weighting each occupation by the number of observations that are not

associated with the recognition process or before application (0.014). This indicator is

set equal to zero either for values below the cut-o� or for individuals not working. Table

1 reports the ten 3-digit occupations with the highest (Panel A) and lowest (Panel B)

share of regulated occupations.14 For each occupation, we report the value of the regu-

lation index, the fraction of the working population, the 3-5 years rate of wage growth,

and the average firm tenure. The Table suggests that the occupations above the median

value of the index have higher wages, with the average di�erence ranging between 7.5

and 10.6 percent (and between 21.6 and 24.4 percent on average if we compare the first

10 occupation with the highest regulation intensity with the first 10 occupations with

the lowest regulation intensity). Also in terms of the rate of wage growth, occupations

with higher intensity of regulation have a much steeper wage profile: those working in13Note that in the context of this study, right-censoring is a relatively minor issue since immigrants in

Germany tend to earn wages well below the censoring limit.14The ranking of occupations and the descriptives are very similar if the cut-o� value is obtained after

weighting occupations according to the overall population in each occupation as taken from the 2 percentof the IEB registry data.

15

the 10 occupations with the highest regulation intensity have a wage growth that is 42

percentage points higher than the wage growth of those working in the 10 occupations

with the lowest regulation intensity. The same trend applies whether we consider the full

sample or only natives. Firm tenure, in contrast, is on average higher for occupations

with lower regulation intensity.

As mentioned above, we restrict our analysis to the sample of immigrants who even-

tually receive full recognition. Out of this group, we further select all individuals who

migrated to Germany at 18 years or older and for whom we have valid information about

the recognition of foreign credentials. We consider as immigrants those who are foreign-

born, thus improving upon previous studies using German registry data, which, due to

data limitations, define migration status based on nationality. Our final estimation sam-

ple consists of 107 individuals. Table 2 shows a number of descriptive statistics for our

estimation sample (column (1)), as well as those immigrants in the survey who only got

partial recognition (column (2)), were denied recognition (column (3)), or did not apply

at all (column (4)). Focusing first on the full recognition sample, we see that there are

47% male individuals who are on average 43.5 years old at the end of our panel in Decem-

ber 2013. The schooling level of these immigrants is relatively high with almost 11 years

of education (excluding tertiary education). The table also provides information about

the typical recognition process. On average, immigrants enter Germany for the first time

when they are 30.5 years old. After that, they take about 1.2 years before making an

o�cial recognition request. One of the reasons for this delay is the demanding recogni-

tion process since various documents and their translations and authentications must be

delivered to the authorities. Finally, after on average four month, successful immigrants

get to know the result of their application. However, as indicated by the large standard

deviation of 6 years, there is a wide range of waiting times, leading to variation that is

important to identify our parameters of interest.

Table 2 also provides information about each group’s labor market outcomes, both

during the first year after migration to Germany and for all available time periods. In

general, there is large variation in the employment rate between the first year and subse-

quent periods, particularly for those who receive full recognition whose employment rate

increases from 18.3% to 63.9%. Hourly wages, even if by less, also increase, for the full

16

recognition group from 8.9 Euros in the first year to 9.9 Euros in later years, but note

that the latter figures average across periods with and without occupational recognition.

When looking across immigrant groups, we see a high level of heterogeneity. Those immi-

grants who obtain full recognition are positively selected in terms of schooling and initial

wages relative to all other groups. They also tend to be younger when making their

request than those immigrants whose application is eventually denied. Overall, the sub-

stantial di�erences in observable characteristics between the di�erent immigrant groups

justify the decision to focus on the sample with full recognition in our empirical analysis.

The last information refers to the share of people working in a regulated occupation

by di�erent groups and over time. Working in a regulated occupation refers to working

in an occupation whose regulation index is above the median value of the occupation

distribution, where the median is weighted by the number of immigrants that before the

application worked in each occupation, as opposed to working in an occupation whose

index is below the threshold or not working. The biggest increase between the first year

and later on is reported by those receiving full recognition, followed by those receiving

partial recognition. Among the former group, only 11 percent work in a regulated occu-

pation at arrival, whereas the average is 31 percentage points higher. These shares refer

to movement either from non-employment to employment in a regulated occupation, or

from working in an unregulated occupation to a regulated one. However, only half of

this increase in probability is driven by moving into employment. This is clear when we

look at the change in employment in unregulated occupations; the increase between the

first year and later on is almost half as much as the increase in the employment rate in

regulated occupations. For those whose application is denied, the increase over time in

the probability of working in a regulated or unregulated occupation is almost the same,

and overall much lower than for those who receive full recognition.

6 Main Results

In this section, we present our estimates of equations (1) and (2). We first show the

average impact of recognition on employment, wages and the propensity to work in a

regulated occupation and unregulated occupation, and present a number of robustness

checks based on di�erent estimation samples. We then graphically show the results from

17

our dynamic specification.

6.1 Average Impacts of Recognition

Table 3 reports the results based on equation (1). The estimates in column (1) show

that obtaining full occupational recognition increases an immigrant’s probability of being

employed by 26.7 percentage point. This is a large e�ect, suggesting that occupational

recognition helps immigrants find and maintain jobs, both because it serves as a signal of

high human capital levels and because it gives them access to labor market segments that

were previously unavailable to them. The point estimate for having applied is negative

and statistically not significant, suggesting that there is no positive signaling e�ect from

having applied for occupational recognition itself. However, since this variable is likely

to also pick up any unobserved time-varying shocks that induce immigrants to start the

application process, its interpretation as the causal e�ect of applying is problematic. Col-

umn (2) shows the corresponding results for the log hourly wage rate. Full recognition

increases average wages by 13.2 percent, which suggests that recognition enables immi-

grants to better utilize and demand higher rewards for their human capital in the host

country’s labor market. Column (3) shows that the probability of working in a regulated

occupation (identified as being employed in an occupation that is above the migrant

weighted median value of regulation) relative to working in an unregulated occupation

or not working at all increases by 19.0 percentage points after recognition. If recogni-

tion did not influence the distribution of migrants’ occupations, we would expect similar

results for the regression in column (4) on the probability of working in an unregulated

occupation relative to working in a regulated occupation or not working at all since the

chosen cuto� value divides the migrant population in half in terms of their occupational

distribution in regulated and unregulated occupations. However, we find an insignificant

average e�ect of 6.7 percentage points in this regression suggesting that immigrants move

disproportionately into regulated occupations after successful recognition.

6.2 Robustness Checks

Table 4 shows a number of robustness checks in which we vary the underlying estimation

sample. Column (8) restates the baseline results of Table 3 as the most restrictive sample.

18

This is our preferred sample. In column (1), we impose the restriction of migration after

the age of 18 with little e�ect on the estimates. In column (2), we exclude individuals

who we do not observe for more than three consecutive years in the registry data. The

estimated e�ects of full recognition on employment, wages and the probability of working

in a regulated or unregulated occupation remain very similar. In column (3), we impose

the additional restriction with respect to column (1) that the immigrants declared to live

in Germany since first arrival. We do this to avoid any biases resulting from di�erential

sample attrition, for example due to return migration. Again, this restriction leaves our

point estimates of interest relatively unchanged. In column (4), we restrict the sample to

those at least 25 years of age to avoid the influence from education periods, again with

no important consequences for our results. Columns (5) and (6) then impose further

restrictions with respect to the sample underlying column (4) with little e�ect. Column

(7) introduces the restriction that the occupational recognition had to take place after the

arrival of the immigrants in Germany. Note, that applying for recognition from abroad

was possible, giving rise to an additional source of selection if only those individuals moved

to Germany who already obtained recognition while abroad and who had already secured

an attractive job or faced particularly good job prospects. Again, we see little changes

in our point estimates, which also holds for our most restrictive specification reported in

column (8) which combines all previous sample restrictions and is estimated on a reduced

sample of only 76 immigrants. Overall, Table 4 shows that our results are very robust to

di�erent sample selection rules, with average employment e�ects of around 25 percentage

points, wage e�ects of around 13 percent, impacts on the probability of working in a

regulated occupation of around 12 percentage points and impacts on the probability of

working in an unregulated occupation of an insignificant 5 percentage points.15

15Table 6 in the Appendix shows how our estimate of the impact of recognition on the di�erent labormarket outcomes varies with the set of control variables included in the specification. Essentially, aftercontrolling for time since migration and individual fixed e�ects, the further inclusion of year fixed e�ects,age squared and German language proficiency has little impact on our point estimates, suggesting thatby focusing on a sample of individuals who eventually all receive full recognition, we capture a large partof unobserved heterogeneity across immigrant workers, as long as we control for the time that has passedsince their arrival in Germany and time-invariant heterogeneity.

19

6.3 Dynamic E�ects

We now turn our attention to the results from the dynamic specification given in equa-

tion (2). The estimation sample corresponds to our main results of Table 3. For better

readability, we summarize the estimates of the period-specific e�ects ”t≠q graphically,

together with the corresponding confidence intervals. Figure 1 displays the e�ects of oc-

cupational recognition on the employment (left panel) and wage (right panel) outcomes

of immigrant workers in the twelve months before and 48 months after recognition. In

the months after recognition, the di�erence in the probability of being employed relative

to those whose application is still under consideration increases rapidly, reaching 30.1

percentage points after twelve months. After that, the employment gap continues to

grow albeit at a slower rate, peaking at 47.6 percentage points 38 months after recog-

nition. This pattern suggests that occupational recognition increases the labor market

opportunities of immigrants relatively quickly following the positive decision, and that

their employability keeps improving even in the long run, possibly due to faster rates of

human capital accumulation in the higher quality jobs immigrants are now able to access.

Reassuringly, there is no discernible di�erence in employment rates between those who

obtain recognition within the following year and those who do not, as indicated by the

insignificant set of parameter estimates prior to the recognition date.

The corresponding dynamic pattern for log hourly wages displayed in the right panel of

Figure 1 shows that it takes around one year before the recognition of foreign credentials

translates into positive wage e�ects. However, from then onwards, the wage di�erential

relative to those without occupational recognition keeps increasing, leveling o� in the very

long run at a log di�erence of around 0.336 or 40 percent. The reason for the delayed onset

of significant wage gains from occupational recognition could be due to employers’ initial

skepticism regarding the equivalence between foreign and native credentials, which only

with time is overcome and, in conjunction with faster rates of human capital accumulation

in the higher quality jobs now accessible to the immigrants, reflected in higher wage rates.

Again, there is no evidence of a significant wage gap in the months prior to the recognition

date, lending credibility to the claim that the subsequent positive wage e�ects are indeed

causally related to the occupational recognition.

Figure 2 decomposes the observed positive employment e�ects and documents how the

20

propensity to work in a regulated (left panel) and unregulated (right panel) occupation

evolve over time, both before and after the recognition date. For the probability of

working in a regulated occupation, initially, there is a noticeable and steady increase in the

share of workers with successful recognition who work in a regulated occupation reaching

20.8 percentage points after 12 months. Subsequently, the probability is increasing further

at a lower rate, reaching 36.8 percentage points after 4 years. Again there is some time

delay until all advantages from recognition materialize, likely because of di�culties of

locating a suitable job in the regulated market segment for some migrants. In contrast,

the probability of working in an unregulated occupation is completely flat throughout the

time before and after recognition with some small but insignificant increase directly after

recognition. Taken together, these two dynamic regressions show that migrants move

into more regulated occupations after recognition, which are the jobs with higher wages

and faster wage growth.

6.4 Synthetic Control Estimates

As a robustness check for our dynamic estimation, we apply a pooled version of the

synthetic control method proposed by Abadie et al. (2010). In contrast to our original

approach, each immigrant who receives recognition (the treatment) is here matched to a

set of other immigrants who never applied for recognition but whose labor market out-

comes in the period prior to recognition are similar to that of the treated immigrant. We

obtain a synthetic control group for each treated immigrant and then average the dynamic

treatment e�ects in each pre- and post-treatment month across all treated individuals in

the sample in those months. The thick black lines in Figure 3 show the resulting dy-

namic impacts of occupational recognition on employment (left panel) and hourly wages

(right panel) between 12 months before and 48 months after recognition.16 The overall

patterns are very similar to those obtained in our fixed e�ects regression approach, with

substantial and relatively quick increases in both employment and hourly wages in the

months immediately after recognition, and continuing divergence at a slower pace in the

long run. At around 20 percentage points, the estimated long-run employment e�ect16Note that we use hourly wages rather than log hourly wages since otherwise it would be di�cult to

find potential control individuals who were earning positive wages in precisely the same periods as thetreated individuals.

21

in the synthetic control approach is smaller than in our regression approach, but note

that the e�ects in the latter are measured relative to current applicants for recognition,

and the point estimates for this group of workers suggest a negative e�ect of applying

for recognition on employment, wages, and the probability of working in an unregulated

occupation, and a positive e�ect on the probability of working in a regulated occupation

(compare Table 3).

To assess the statistical significance of the synthetic control-based dynamic e�ects,

we perform 30 placebo estimations in which, for each iteration, we randomly pick for

each treated immigrant an untreated immigrant from his or her donor pool, assign the

same hypothetical recognition date as for the treated immigrant, find a suitable synthetic

control group for this placebo immigrant, and then aggregate all dynamic impact esti-

mates across all placebo immigrants. As illustrated by the thin gray lines in Figure 3, our

estimated employment and wage e�ects of occupational recognition are unusually large

relative to the distribution of dynamic placebo e�ects, suggesting that they actually pick

up real employment and wage e�ects.

Figure 4 displays the corresponding dynamic e�ects for the probability of working in a

regulated (left panel) and unregulated occupation (right panel). As already documented

in Figure 2, the observed increase in the probability of being employed is exclusively

driven by a strong increase in the probability of working in a regulated occupation, while

there is no clear impact of occupational recognition on the probability of working in

an unregulated occupation, at least for the first 12 post-recognition, after which the

probability of working in an unregulated occupation starts to decline relative to the

synthetic control groups.

7 Occupational Recognition and Immigrant Earnings As-similation

Our results so far have shown a significant positive long-run e�ect of occupational recog-

nition on immigrants’ employment and wage outcomes. In this section, we put these gains

into perspective by relating them to standard earnings assimilation profiles of immigrants

in Germany. For this purpose, we merge a 2% random sample of native Germans to our

IAB-SOEP Migration Sample, and jointly estimate the following immigrant and native

22

earnings profiles:

Immigrants: logwlt = „ÕiXlt + –i · agelt + — · ysmlt + “ · ysrlt + ”Cl + ◊ifit + Ált

Natives: logwlt = „ÕnXlt + –n · agelt + ◊n · fit + Ált,

(3)

where wlt are the monthly earnings of individual l at time t, Xlt is a vector of socioeco-

nomic characteristics (educational attainment, gender, federal state of residence), agelt

is a quadratic or quartic, depending on specification, of the individual’s age, ysmlt is

either a quadratic or a quartic in the number of years the immigrant has resided in Ger-

many, ysrlt is a quadratic or quartic in the number of years passed since the result of the

recognition process was obtained (set to zero for all immigrants who never applied for

recognition), Cl is a vector of dummy variables indicating an immigrant’s arrival cohort

(1970-1994, 1995-2005, 2005-2013), and fit is a vector of year fixed e�ects. Since aging,

cohort and period e�ects are perfectly collinear, we impose the standard assumption that

period e�ects are the same for immigrants and natives (◊i = ◊n), as suggested by Bor-

jas (1995). We estimate this model using all available monthly native and immigrant

observations, clustering our standard errors at the individual level. The immigrants in

the sample belong to four distinct groups: immigrants who never applied for recognition,

immigrants who applied but were denied recognition, immigrants who applied and gained

partial recognition, and immigrants who applied and gained full recognition. We drop

immigrants who applied for recognition but whose decision is pending at the time of the

survey. We allow the age, years since migration and years since recognition profiles to

vary between each of these four groups.

Rather than presenting the full regression results (which can be found in Table 5 in

the Appendix), we use our coe�cient estimates from the two-equation regression model in

(3) to predict native and immigrant earnings profiles, allowing for a quartic in age, years

since migration and years since recognition (see column (4) of Table 5). We simulate

earnings profiles for immigrants who enter Germany at the age of 25 and compare them

to the corresponding earnings profile of natives of the same age. We compute each profile

for the mean values of all socioeconomic characteristics in the sample, thus accounting for

observable di�erences in educational attainment, gender, federal state of residence and

time period between the di�erent immigrant groups and natives. The intercepts of the

23

four immigrant groups reflect the weighted means of their cohort e�ects.17

For clarity, the left panel of Figure 5 only depicts the earnings profiles of native

Germans, immigrant non-applicants, and immigrants who eventually received full recog-

nition, suppressing the corresponding profiles for immigrants whose application was de-

nied and immigrants who only received partial recognition, which together make up only

a small fraction of the overall sample. The right panel depicts the di�erences in log

earnings relative to natives, together with 95% confidence intervals of these di�erences.

Immigrants who never apply for recognition (who make up 79.5% of the immigrant sam-

ple) initially face an earnings gap relative to native Germans of 41.7% (0.539 log points)

which steadily declines over time, leveling o� at around 11.4% after 15 years of residence

in Germany. For immigrants who eventually obtain full recognition (12.1% of the immi-

grant sample), we see initially, prior to obtaining full recognition, a relatively fast rate

of earnings convergence which, however, decelerates noticeably after the first few years

in the country. After full recognition, which we assume to occur after three years of

residence in Germany (the mean duration between arrival and recognition in our assim-

ilation sample), we see a pronounced renewed acceleration of the speed of convergence

(dashed line), suggesting a catch-up and eventual overtaking of native earnings after 10

years in the country. However, as evident from the right panel in Figure 5, due to the

small sample size, we lack precision in the estimates for the immigrant group with full

recognition, so that from 5 years since migration onwards, their earnings gap relative to

natives is no longer statistically significant. Notwithstanding, our findings suggest that

occupational recognition has a significant e�ect on the speed of immigrants’ economic

assimilation in Germany.

8 Conclusion

In this paper, we analyze how the formal recognition of immigrants’ foreign occupational

credentials a�ects their subsequent labor market outcomes. We use a new linked survey-

social security data set that explicitly asks participants about details in the timing of their

recognition process and includes comprehensive information about their work histories17Similarly to Bratsberg et al. (2006), we allow the returns to education (and gender) to vary between

natives and immigrants, but not between di�erent immigrant groups. We further assume that the regione�ects are the same for immigrants and natives.

24

in Germany. In our panel data setup, we control for individual fixed e�ects and exploit

the variation in the timing between application beginning and recognition outcome to

identify the causal e�ects of occupational recognition.

Overall, the evidence from our dynamic specification suggests large and long-lasting

positive e�ects of occupational recognition on labor market outcomes. Recognizing immi-

grants’ foreign credentials may thus be a highly e�ective way of tapping into their human

capital and fostering their integration into the host country’s economy. The results also

suggest that part of the often substantial employment and wage gaps between natives

and immigrants may be due to the lack of formal recognition of the latter’s occupational

training. The large positive wage e�ects and the eventual full convergence to native

earnings furthermore indicate that foreign credentials, once declared equivalent to native

ones, are indeed valued in the German labor market, mitigating fears of a watering-down

of occupational standards.

25

References

Abadie, Alberto, Alexis Diamond, and Jens Hainmueller, “Synthetic Control Methods

for Comparative Case Studies: Estimating the E�ect of California’s Tobacco Control

Program,” Journal of the American Statistical Association, 2010, 105 (490), 493–505.

Algan, Yann, Christian Dustmann, Albrecht Glitz, and Alan Manning, “The Economic

Situation of First- and Second-Generation Immigrants in France, Germany, and the

UK,” The Economic Journal, 2010, 120 (542), F4–F30.

Arai, Mahmood and Peter Skogman Thoursie, “Renouncing Personal Names: An Em-

pirical Examination of Surname Change and Earnings,” Journal of Labor Economics,

2009, 27 (1), 127–147.

Borjas, George J., “Assimilation and Changes in Cohort Quality Revisited: What Hap-

pened to Immigrant Earnings in the 1980s?,” Journal of Labor Economics, 1995, 13

(2), pp. 201–245.

Bratsberg, Bernt, Erling Barth, and Oddbj’orn Raaum, “Local Unemployment and the

Relative Wages of Immigrants: Evidence from the Current Population Surveys,” The

Review of Economics and Statistics, 2006, 88 (2), 243–263.

, Jr. F. Ragan James, and Zafar M. Nasir, “The E�ect of Naturalization on Wage

Growth: A Panel Study of Young Male Immigrants,” Journal of Labor Economics,

2002, 20 (3), 568–597.

Chapman, Bruce J. and Robyn R. Iredale, “Immigrant Qualifications: Recognition and

Relative Wage Outcomes,” International Migration Review, 1993, 27 (2), pp. 359–387.

Chiswick, Barry R. and Paul W. Miller, “The complementarity of language and other

human capital: immigrant earnings in Canada,” Economics of Education Review, 2003,

22 (5), 469 – 480.

Dietz, Joerg, Victoria M. Esses, Chetan Joshi, and Caroline Bennett-AbuAyyash, “The

Evaluation of Immigrants’ Credentials: The Roles of Accreditation, Immigrant Race,

and Evaluator Biases,” CLSRN Working Paper No. 18, 2009.

26

Dustmann, Christian and Albrecht Glitz, “Chapter 4 - Migration and Education,” in

Stephen Machin Eric A. Hanushek and Ludger Woessmann, eds., Handbook of The

Economics of Education, Vol. 4 of Handbook of the Economics of Education, Elsevier,

2011, pp. 327 – 439.

and Tommaso Frattini, “Immigration: the European Experience,” in David Card and

Steve Raphael, eds., Immigration, Poverty, and Socioeconomic Inequality, Russell Sage

Foundation, 2013, chapter 13.

, , and Ian P. Preston, “The E�ect of Immigration along the Distribution of Wages,”

The Review of Economic Studies, 2013, 80 (1), 145–173.

Friedberg, Rachel M., “The Impact of Mass Migration on the Israeli Labor Market,”

Quarterly Journal of Economics, 2001, 116 (4), 1373–1408.

Houle, René and Lahouaria Yssaad, “Recognition of newcomers’ foreign credentials and

work experience,” Perspectives - Statistics Canada, 2010, pp. pp. 18–33.

Kleiner, Morris M. and Alan B. Krueger, “The Prevalence and E�ects of Occupational

Licensing,” British Journal of Industrial Relations, 2010, 48 (4), 676–687.

and , “Analyzing the Extent and Influence of Occupational Licensing on the Labor

Market,” Journal of Labor Economics, 2013, 31 (2), S173–202.

Kugler, Adriana D. and Robert M. Sauer, “Doctors without Borders? Relicensing Re-

quirements and Negative Selection in the Market for Physicians,” Journal of Labor

Economics, 2005, 23 (3), 437–466.

Lubotsky, Darren, “Chutes or Ladders? A Longitudinal Analysis of Immigrant Earn-

ings,” Journal of Political Economy, 2007, 115 (5), pp. 820–867.

Mattoo, Aaditya, Ileana Cristina Neagu, and Caglar Ozden, “Brain Waste? Educated

Immigrants in the US Labor Market,” Journal of Development Economics, 2008, 87

(2), 255–269.

Peterson, Brenton D., Sonal S. Pandya, and David Leblang, “Doctors with borders:

occupational licensing as an implicit barrier to high skill migration,” Public Choice,

2014, 160 (1-2), 45–63.

27

Pissarides, Christopher A., Equilibrium Unemployment Theory, Oxford: Blackwell, 1990.

Tani, Massimiliano, “Local Signals and the Returns to Foreign Education,” Technical

Report, IZA Discussion Paper Series No 9597 2015.

Vicari, Basha, “Degree of Standardised Certification in Occupations,” Bundesagentur für

Arbeit, 2014, 4.

28

Table 1: Regulated and Unregulated Occupations - Migrant Workers

Index of Fraction of Mean 3-5 Yrs Rate of Wage 3-5 Yrs Rate of Wage Firm TenureRegulation Working Pop. % Wage Growth % Growth. Natives % (in Years)

Panel A. First 10 Occupations with High Regulation IntensityOccupations in veterinary medicine 1.000 0.513 17.192 107.358 108.699 4.544Occupations in human medicine and dentistry 1.000 0.023 12.185 60.987 55.975 4.705Teachers in schools of general education 0.991 0.551 15.044 66.148 65.677 5.736Occupations in police, jurisdiction and the penal institution 0.875 0.042 9.498 12.026 12.090 5.848Occupations in nursing, emergency medical services and obstetrics 0.760 2.924 9.744 64.571 65.365 6.212Occupations in technical research and development 0.753 1.493 14.504 21.597 21.590 6.216Occupations in construction scheduling, and architecture 0.708 0.635 14.113 25.973 25.537 5.669Occupations in geriatric care 0.628 0.152 7.892 21.124 21.331 5.557Occupations in education and social work 0.445 3.236 9.895 140.868 143.383 5.395Ship’s o�cers and masters 0.442 0.048 12.055 16.770 16.161 4.447

First 10 occupations 0.760 0.962 12.212 53.742 53.581 5.433First 10 occupations (weighted average) 11.510 82.226 83.316 5.761Above the median index 0.199 0.837 10.459 39.289 40.051 5.717Above the median index (weighted average) 9.859 48.043 48.814 5.610

Panel B. Last 10 Occupations with Low Regulation IntensityOccupations in editorial work and journalism 0.000 0.198 15.514 65.586 67.392 5.776Service occupations in passenger tra�c 0.000 0.003 13.657 11.569 11.692 5.748Sales occupations in retail trade 0.000 0.009 7.852 22.401 24.297 5.788Occupations in human resources management and personnel service 0.000 0.069 7.466 47.142 47.762 5.317Occupations in event organisation and management 0.000 0.016 8.082 36.390 37.312 6.558Sales occupations (retail) 0.000 0.080 11.732 63.060 68.937 6.639Occupations in the humanities 0.000 0.134 10.696 152.695 183.189 3.112Managing directors and executive board members 0.000 0.004 6.348 3.809 2.965 5.431Sales occupations (retail) selling foodstu�s 0.000 0.003 10.422 7.950 8.087 6.825Artisans working with metal 0.000 0.004 8.646 2.259 2.259 6.609

Last 10 occupations 0 0.052 9.869 23.252 23.627 5.894Last 10 occupations (weighted average) 8.997 30.097 30.511 6.278Below the median index 0.002 0.577 9.487 28.538 29.329 5.622Below the median index (weighted average) 8.651 31.345 31.364 5.129

Note: Data source: IEB data. The distinction between regulated and unregulated occupation is made with respect the median value of the regulation index across all occupation. Panel A refers to the first 10occupations with the highest vale of the index among the occupations with a value above the median. Panel B refers to the first 10 occupations with the lowest vale of the index among the occupations with avalue below the median. The median value of the weighted index is 0.014. The index is provided by (Vicari, 2014), and it is weighted according to the population in each occupation in the full IEB registryin 2012. In addition, we use as further weight the immigrant population in the survey data in each occupation. Only the spells before the application for recognition are considered. All descriptive values arecomputed using a 2 percent sample of the full IEB registry, and refer to the years 1975-2013. Wages refer to real hourly wages of the longest working spell within a year.

29

Table 2: Descriptive Statistics by Recognition Outcome

Full Recognition Partial Recognition Denied Recognition Non-Applicant

Panel A. ImmigrantsMale % 46.7 53.8 30.6 43.5

(50.1) (50.8) (46.7) (49.7)Years of schooling 10.7 10.4 9.9 10.6

(1.8) (1.5) (1.4) (1.7)Age in December 2013 43.5 41.3 45.0 44.2

(9.4) (8.7) (7.9) (11.0)Age at first migration 30.5 28.6 32.4 32.1

(7.6) (7.1) (8.2) (9.6)Age at request of recognition 31.7 31.2 35.5 ≠

(7.7) (9.2) (7.9)Time request to result (months) 4.0 7.9 5.8 ≠

(6.0) (15.3) (10.4)

Panel B. Observations - First Year In GermanyEmployed % 18.3 9.5 12.2 21.2

(38.7) (29.3) (32.7) (40.9)Employed in regulated occupation % 11.3 5.3 4.7 10.7

(31.7) (22.4) (21.2) (30.9)Employed in unregulated occupation % 5.5 0.0 3.6 9.0

(22.9) (0.0) (18.6) (28.6)Real hourly wage 8.9 5.5 5.7 8.2

(4.7) (2.4) (3.1) (5.8)

Panel C. Observations - Average Over TimeEmployed % 66.2 51.0 53.6 56.7

(47.3) (50.0) (49.9) (49.5)Employed in regulated occupation % 42.5 37.1 25.9 26.4

(49.4) (48.3) (43.8) (44.1)Employed in unregulated occupation % 22.1 8.2 23.4 27.7

(41.5) (27.4) (42.4) (44.7)Real hourly wage 9.9 8.9 5.8 8.7

(5.8) (5.2) (2.9) (5.4)

Individuals 107 26 36 322

Note: Statistics depicted are means with standard deviations in parentheses. Statistics are based on individuals in upper panel andon monthly observations in the lower two panels. Employed % compares time periods of employment to times of employment andnon-employment. Employed regulated occupation % and employed unregulated occupation % look at the employment in specificoccupations and theoretically add up to the figure of Employed %. Since occupational information is missing for some employmentspells both statistics combined are lower in our calculated statistics. Because occupation information is given on an aggregate levelwe assign the degree of regulation according to a regulation index provided by the IAB. We classify occupations as regulated ifthey are above the median of the recognition index among the migrant working observations that are una�ected by the recognitionprocess. Real hourly wage uses daily wage information assuming that full-time employment is 8 hours and part-time employmentis 5 hours or 2 hours.

30

Table 3: Occupational Recognition and Average Labor Market Outcomes

Regulated UnregulatedEmployment Wages Occupations Occupations

(1) (2) (3) (4)

Applied for recognition -0.049 -0.058 0.050 -0.090(0.073) (0.133) (0.089) (0.069)

Received full recognition 0.267*** 0.132** 0.190** 0.067(0.083) (0.061) (0.092) (0.049)

Individuals 76 69 76 76Observations 11,544 6,740 10,909 10,909

Note: Data source: IAB-SOEP Migration Sample linked to IEB data. The dependent variable is an indicatorfor working in column (1), log hourly real wages in column (2), an indicator for working in a regulated occupa-tion relative to working in an unregulated occupation or being non-employed in column (3), and an indicator forworking in an unregulated occupation relative to working in a regulated occupation or being non-employed incolumn (4). Additional controls are individual fixed e�ects, time fixed e�ects, time since migration fixed e�ects,age squared, and German language proficiency. The sample only comprises immigrants who eventually receivefull recognition, who migrated to Germany at the age of at least 18, received their recognition outcome after mi-grating to Germany, stayed in Germany after arrival and do not have 3 years of missing employment information.Observations are only included when migrants are at least 25 years old. For the wage regression the sample onlyincludes observations from full-time employment.Standard errors in parentheses are clustered at the individual level: * p<0.10, ** p<0.05, *** p<0.01

31

Table 4: Occupational Recognition and Average Labor Market Outcomes - Robustness Checks

Mig after 18yr (1)+w/o 3yr gaps (1)+stay (1)+age >25 (4)+stay (5)+w/o 3yr gap (5)+recog after mig (6)+(7) combined(1) (2) (3) (4) (5) (6) (7) (8)

EmploymentApplied for recognition -0.096 -0.043 -0.105 -0.097 -0.103 -0.049 -0.103 -0.049

(0.059) (0.062) (0.063) (0.066) (0.069) (0.074) (0.069) (0.073)Received full recognition 0.267*** 0.290*** 0.236*** 0.241*** 0.226*** 0.263*** 0.227*** 0.267***

(0.061) (0.070) (0.063) (0.070) (0.073) (0.082) (0.073) (0.083)

Individuals 107 85 98 107 98 81 92 76Observations 16,794 12,875 15,376 16,004 14,800 11,812 14,466 11,544WagesApplied for recognition -0.031 -0.076 -0.031 -0.015 -0.014 -0.056 -0.014 -0.058

(0.119) (0.131) (0.119) (0.121) (0.120) (0.133) (0.120) (0.133)Received full recognition 0.136** 0.127** 0.130** 0.149** 0.143** 0.137** 0.138** 0.132**

(0.054) (0.057) (0.055) (0.060) (0.060) (0.062) (0.059) (0.061)

Individuals 90 76 84 89 84 72 81 69Observations 8,042 7,267 7,799 7,796 7,570 6,883 7,427 6,740Regulated OccupationsApplied for recognition -0.080 -0.014 -0.053 -0.032 -0.008 0.057 -0.011 0.050

(0.070) (0.083) (0.070) (0.076) (0.075) (0.088) (0.076) (0.089)Received full recognition 0.212*** 0.231*** 0.222*** 0.170** 0.184** 0.201** 0.173** 0.190**

(0.071) (0.081) (0.073) (0.079) (0.081) (0.091) (0.081) (0.092)

Individuals 107 85 98 107 98 81 92 76Observations 15,976 12,191 14,593 15,211 14,040 11,152 13,736 10,909Unregulated OccupationsApplied for recognition -0.017 -0.037 -0.053 -0.053 -0.080 -0.098 -0.077 -0.090

(0.052) (0.066) (0.050) (0.058) (0.056) (0.068) (0.057) (0.069)Received full recognition 0.053 0.057 0.012 0.062 0.035 0.051 0.047 0.067

(0.041) (0.047) (0.041) (0.045) (0.044) (0.049) (0.044) (0.049)

Individuals 107 85 98 107 98 81 92 76Observations 15,976 12,191 14,593 15,211 14,040 11,152 13,736 10,909

Note: Standard errors in parentheses are clustered at the individual level. Additional controls are individual fixed e�ects, time fixed e�ects, time since migration fixede�ects, age squared, and German language proficiency.* p<0.1, ** p<0.05, *** p<0.01

32

Figure 1: Dynamic E�ect of Recognition on Employment and Log Hourly Wages

Note: The estimations are analogous to the static employment and wage regressions in Table 3, controlling for an indicatorfor having applied, individual fixed e�ects, time fixed e�ects, time since migration fixed e�ects, age squared, and Germanlanguage proficiency, and using the sample of all individuals who received full recognition. For wages, the sample is restrictedto full-time employees. 95% confidence intervals displayed using clustered standard errors at the individual level.

Figure 2: Dynamic E�ect of Recognition on Employment in Regulated and UnregulatedOccupations

Note: The estimations are analogous to the static occupation regressions in Table 3, controlling for an indicator for havingapplied, individual fixed e�ects, time fixed e�ects, time since migration fixed e�ects, age squared, and German languageproficiency, and using the sample of all individuals who received full recognition. Occupations with a recognition indexabove the median among the migrant working observations are considered as regulated occupations, occupations with aregulation index below the median are considered as unregulated occupations. 95% confidence intervals displayed usingclustered standard errors at individual level.

33

Figure 3: Dynamic E�ect of Recognition on Employment and Hourly Wages - SyntheticControl Approach

−.2

−.1

0.1

.2.3

Empl

oym

ent

−12 0 12 24 36 48Months around Recognition

Effect of Recognition over Time on Employment

−2−1

01

23

Hou

rly W

age

−12 0 12 24 36 48Months around Recognition

Effect of Recognition over Time on Hourly Wage

Note: Estimates derived using synthetic control approach. Estimation sample as in column (8) of Table 4. For each treatedimmigrant, we find a set of immigrants who never applied for recognition and who can jointly serve as a suitable controlgroup based on their outcomes in the periods just prior to the treated immigrant’s recognition date. The donor pool foreach treated unit is restricted to those individuals who are observed in exactly the same periods as the treated unit. Thedisplayed estimates along the thick black lines are the average di�erentials in employment (left panel) and hourly wages(right panel) in each pre- and post-treatment period between all treated units and their synthetic control groups. The thingray lines depict 30 placebo estimations, in which we iteratively apply the synthetic control method to randomly pickednon-treated immigrants in each treated immigrant’s donor pool.

Figure 4: Dynamic E�ect of Recognition on Working in Regulated Occupations - Syn-thetic Control Approach

−.2

0.2

.4.6

Regu

late

d O

ccup

atio

n

−12 0 12 24 36 48Months around Recognition

Effect of Recognition over Time on Regulated Occupation

−.2

−.1

0.1

.2Un

regu

late

d O

ccup

atio

n

−12 0 12 24 36 48Months around Recognition

Effect of Recognition over Time on Unregulated Occupation

Note: Estimates derived using synthetic control approach. Estimation sample as in column (8) of Table 4. For each treatedimmigrant, we find a set of immigrants who never applied for recognition and who can jointly serve as a suitable controlgroup based on their outcomes in the periods just prior to the treated immigrant’s recognition date. The donor pool foreach treated unit is restricted to those individuals who are observed in exactly the same periods as the treated unit. Thedisplayed estimates along the thick black lines are the average di�erentials in the probability of working in a regulatedoccupation (left panel) and an unregulated occupation (right panel) in each pre- and post-treatment period between alltreated units and their synthetic control groups. Occupations with a recognition index above the median among the migrantworking observations are considered as regulated occupations, occupations with a regulation index below the median areconsidered as unregulated occupations. The thin gray lines depict 30 placebo estimations, in which we iteratively apply thesynthetic control method to randomly picked non-treated immigrants in each treated immigrant’s donor pool.

34

Figure 5: E�ect of Recognition on Immigrant Assimilation Profiles

6.5

77.

58

25 30 35 40 45Age

Native Germans Non−ApplicantsBefore Full Recognition After Full Recognition

Log

Mon

thly

Ear

ning

s

−1.5

−1−.

50

.5

25 30 35 40 45Age

Non−Applicants Full Recognition

Diff

eren

ce in

Log

Mon

thly

Ear

ning

s

Note: The displayed simulations of earnings profiles are based on parameter estimates reported in column (4) of Table 5.Immigrants are assumed to enter Germany at the age of 25, with the comparison being relative to natives of the sameage. We compute each profile for the mean values of all socioeconomic characteristics in the sample, thus accounting forobservable di�erences in educational attainment, gender, federal state of residence and time period between the di�erentimmigrant groups and natives. The intercepts of the di�erent immigrant groups reflect their weighted mean cohort e�ects.The left panel shows the actual predicted earnings profiles, the right panel the profiles of the predicted log earnings di�erencesrelative to natives, together with 95% confidence intervals.

35

9 Appendix

Linked Survey-Social Security Data

The IAB-SOEP Migration Sample is a new longitudinal survey of people with migration

background in Germany, jointly carried out by the Institute for Employment Research

(IAB) and the German Socio Economic Panel (GSOEP). Both first and second generation

immigrants are part of the survey, with the first generation representing around 75 percent

of the sample. The first two waves were conducted in 2013 and 2014, with the former

comprising around 5,000 individuals, and the latter adding another 3,800 individuals.

Using a personal identifier, survey respondents are linked to the social security data

(IEB). Due to data protection, respondents are required to give their prior consent for

the record linkage by signing a document. The overall approval rate amounts to about

50 percent.

Assimilation Regression Results

36

Table 5: Assimilation Regressions

Quadratic Quartic

(1) (2) (3 (4)

Never -0.905 (0.511) -0.901 (0.510) -2.175 (6.492) -2.198 (6.491)Denied -2.499 (2.427) -2.099 (2.582) 13.831 (23.554) 19.375 (30.138)Partial 0.492 (1.447) 0.490 (1.433) 7.796 (28.077) 5.784 (26.524)Full 0.144 (1.305) 0.236 (1.299) -14.540 (15.962) -12.000 (16.103)Never ◊ YSM 0.035*** (0.009) 0.035*** (0.009) 0.010 (0.028) 0.010 (0.028)Denied ◊ YSM 0.142 (0.074) 0.132 (0.073) 0.078 (0.355) -0.242 (0.424)Partial ◊ YSM 0.071 (0.040) 0.051 (0.063) 0.216* (0.109) 0.230 (0.141)Full ◊ YSM 0.051** (0.018) 0.001 (0.041) 0.068 (0.070) 0.040 (0.074)Never ◊ YSM2/10 -0.006 (0.003) -0.006 (0.003) 0.036 (0.036) 0.036 (0.036)Denied ◊ YSM2/10 -0.074 (0.049) -0.059 (0.044) -0.176 (0.911) 0.668 (1.068)Partial ◊ YSM2/10 -0.002 (0.012) 0.021 (0.033) -0.236 (0.131) -0.320* (0.161)Full ◊ YSM2/10 -0.007 (0.005) 0.012 (0.022) -0.041 (0.076) -0.103 (0.090)Age 0.066*** (0.003) 0.066*** (0.003) 1.247*** (0.074) 1.247*** (0.074)Age2/10 -0.008*** (0.000) -0.008*** (0.000) -0.461*** (0.027) -0.461*** (0.027)Never ◊ Age 0.030 (0.026) 0.030 (0.026) 0.147 (0.658) 0.150 (0.657)Denied ◊ Age 0.105 (0.126) 0.086 (0.131) -0.894 (2.307) -1.403 (2.960)Partial ◊ Age -0.040 (0.072) -0.041 (0.071) -0.951 (2.909) -0.655 (2.745)Full ◊ Age -0.018 (0.063) -0.017 (0.063) 1.300 (1.616) 1.022 (1.645)Never ◊ Age2/10 -0.005 (0.003) -0.005 (0.003) -0.044 (0.243) -0.045 (0.243)Denied ◊ Age2/10 -0.017 (0.015) -0.014 (0.016) 0.125 (0.820) 0.306 (1.047)Partial ◊ Age2/10 0.003 (0.008) 0.003 (0.008) 0.397 (1.100) 0.240 (1.037)Full ◊ Age2/10 0.000 (0.007) -0.000 (0.007) -0.428 (0.595) -0.315 (0.610)Medium Edu 0.373*** (0.014) 0.373*** (0.014) 0.374*** (0.014) 0.374*** (0.014)High Edu 0.752*** (0.019) 0.752*** (0.019) 0.753*** (0.019) 0.753*** (0.019)Immigrant ◊ Medium Edu -0.214*** (0.062) -0.223*** (0.063) -0.228*** (0.062) -0.234*** (0.063)Immigrant ◊ High Edu 0.028 (0.087) 0.015 (0.088) 0.022 (0.087) 0.011 (0.088)Female -0.507*** (0.011) -0.507*** (0.011) -0.509*** (0.011) -0.509*** (0.011)Immigrant ◊ Female -0.119* (0.060) -0.115 (0.060) -0.118* (0.059) -0.116 (0.060)Cohort 1970-1994 0.249* (0.108) 0.243* (0.107) 0.261* (0.107) 0.255* (0.107)Cohort 1995-2004 0.119 (0.076) 0.123 (0.076) 0.117 (0.075) 0.117 (0.075)Denied ◊ YSR -0.003 (0.024) -0.033 (0.037)Partial ◊ YSR 0.013 (0.037) 0.040 (0.058)Full ◊ YSR 0.045 (0.029) 0.073* (0.033)Denied ◊ YSR2/10 -0.022 (0.024) 0.059 (0.036)Partial ◊ YSR2/10 -0.029 (0.033) -0.043 (0.117)Full ◊ YSR2/10 -0.018 (0.020) -0.006 (0.017)Never ◊ YSM3/100 -0.023 (0.016) -0.023 (0.016)Denied ◊ YSM3/100 0.387 (0.865) -0.487 (1.005)Partial ◊ YSM3/100 0.130* (0.065) 0.252 (0.141)Full ◊ YSM3/100 0.019 (0.030) 0.084 (0.043)Never ◊ YSM4/1000 0.004 (0.002) 0.004 (0.002)Denied ◊ YSM4/1000 -0.202 (0.274) 0.106 (0.317)Partial ◊ YSM4/1000 -0.023* (0.011) -0.061 (0.046)Full ◊ YSM4/1000 -0.003 (0.004) -0.015 (0.008)Age3/1000 0.749*** (0.043) 0.749*** (0.043)Age4/100000 -0.449*** (0.025) -0.449*** (0.025)Never ◊ Age3/1000 0.058 (0.389) 0.060 (0.389)Denied ◊ Age3/1000 0.111 (1.258) -0.166 (1.593)Partial ◊ Age3/1000 -0.726 (1.801) -0.381 (1.700)Full ◊ Age3/1000 0.596 (0.948) 0.399 (0.975)Never ◊ Age4/100000 -0.032 (0.229) -0.033 (0.229)Denied ◊ Age4/100000 -0.245 (0.706) -0.089 (0.884)Partial◊ Age4/100000 0.484 (1.084) 0.213 (1.029)Full ◊ Age4/100000 -0.302 (0.551) -0.177 (0.569)Denied ◊ YSR3/100 0.040 (0.028)Partial ◊ YSR3/100 -0.074 (0.112)Full ◊ YSR3/100 -0.037* (0.017)Denied ◊ YSR4/1000 -0.074** (0.024)Partial ◊ YSR4/1000 0.045 (0.059)Full ◊ YSR4/1000 0.009 (0.006)Constant 5.361*** (0.063) 5.361*** (0.063) -5.767*** (0.731) -5.767*** (0.731)

R-squared 0.27 0.28 0.28 0.28Observations 1,864,925 1,864,925 1,864,925 1,864,925

Note: The dependent variable are log monthly earnings, conditional on working at least one day in a given month. Theomitted categories are males, low educational attainment, immigrant cohort 2005-2013, period 1975. Standard errors areclustered at the individual level. The sample comprises monthly observations of 12,234 individuals.

37

Robustness Checks for Static Regression Model

38

Table 6: Impact of Control Variables

(1) (2) (3) (4) (5) (6)EmploymentApplied for recognition -0.041 -0.052 -0.021 -0.051 - -0.049

(0.088) (0.092) (0.078) (0.074) (0.073)Received full recognition 0.536*** 0.318*** 0.257*** 0.267*** 0.231*** 0.267***

(0.085) (0.114) (0.090) (0.083) (0.076) (0.083)

Individuals 76 76 76 76 76 76Observations 11,544 11,544 11,544 11,544 11,544 11,544EmploymentApplied for recognition -0.077 -0.027 -0.137 -0.063 - -0.058

(0.188) (0.183) (0.138) (0.142) (0.133)Received full recognition 0.561*** 0.443** 0.175** 0.133** 0.090 0.132**

(0.199) (0.201) (0.077) (0.065) (0.101) (0.061)

Individuals 69 69 69 69 69 69Observations 6,740 6,740 6,740 6,740 6,740 6,740Regulated OccupationsApplied for recognition 0.102 0.082 0.067 0.049 - 0.050

(0.088) (0.098) (0.094) (0.089) (0.089)Received full recognition 0.313*** 0.198* 0.190* 0.190** 0.226*** 0.190**

(0.092) (0.111) (0.097) (0.092) (0.075) (0.092)

Individuals 76 76 76 76 76 76Observations 10,909 10,909 10,909 10,909 10,909 10,909Unegulated OccupationsApplied for recognition -0.124*** -0.128*** -0.082 -0.092 - -0.090

(0.045) (0.046) (0.067) (0.069) (0.069)Received full recognition 0.218*** 0.120** 0.061 0.068 0.002 0.067

(0.049) (0.047) (0.050) (0.049) (0.068) (0.049)

Individuals 76 76 76 76 76 76Observations 10,909 10,909 10,909 10,909 10,909 10,909Time since migration fixed e�ects Yes Yes Yes Yes YesIndividual fixed e�ects Yes Yes Yes YesTime fixed e�ects Yes Yes YesControls Yes Yes

Note: Standard errors in parentheses clustered at individual level. Sample selection according to main specifi-cations in Table 3.* p<0.1, ** p<0.05, *** p<0.01

39