Jan Stuhler is Associate Professor at Universidad Carlos III de Madrid, and is affiliated with both the Swedish Institute for Social Research in Stockholm, as well as the Centre for Research and Analysis of Migration in London.
Welcome everybody good afternoon welcome to this very nice seminar by Yan stoer I’m going to introduce Yan in a second I’m an economics I’m visiting for one year and from georetown University so let me say a few words about y y is becoming one of the superstars in the economics literature of migration
And he also works on intergenerational Mobility you received a phds from University College London and his advisors were two of the most prolific authors in the economics of migration so Yan has worked with Christ basman and Sh when we will uh for those of you who will take my class on
My would see these are um very important figures inage Yan is affiliated with PR CPR and IA and less than 10 years from his PhD he has published for very well so very nice po oriented paper and the title is the initial of migrants in the German label Market evidence
Over yeah let me start presenting we have a bit of a technical issue here there’s some some feedback we’re not sure where it’s coming from I’m kind of searching my pockets if I have anything but I don’t you give me a second yeah I think this one is fine I don’t
Think the microphone is on I mean now it’s fine maybe maybe it’s stop it okay yeah and otherwise of course we can try one of the other ones okay let’s let’s try let’s try with this one okay yeah thanks a lot for for the uh invitation thanks for having me here let’s right
With okay just very close it’s already on okay great okay um yeah thanks for having me I’m very happy to present this work here on labor market integration of migrants in Germany um so I’m generally working on as you said on migration topics related to migration but this is actually the
First project where really focus a lot on the integration experience for migrants themselves okay so hopefully some of this is going to be new to you uh it certainly was some of this was new to me uh so this is kind of worked trying to get really into this topic
Trying to understand better what’s going on and because this was sort of the first work I did on this topic uh on this particular topic um I really sort of first wanted to get an overview of what’s happening okay so this paper is really very much a paper trying to get
An overview over it’s in the title over the last 50 years trying to see how well did migrants integrate into lab Germany how has changed over time and so on okay so very much an overview article so a bit of background Germany has actually by now become the second most important destination for
Migrants work right um so right after the UN us of course most migrants still go to the United States um but after that Germany so we have 30 million foreign born now as population share that’s 177% if you include second generation migrants then that’s up to 27% and because there’s of course a
Different demographic structure of migrants if you look at those at Young um young people Germany then we’re actually reaching a migrant share of about 40% okay so it’s like really large sh of the population now either forign born or has forign born parents and so this is clearly a very
Important issue and you know anyone who’s reading the news not only in Germany but also many other countries migration and integration of migrants is of course a very important isue in Germany I think it’s fair to say most immigration episodes took the country by surprise so we kind of I talk a little
Bit about we started with sort of the so-called guest worker period where we really actively recruited workers from other countries but then a lot of the other migration wave that happened after that they have more as surprise and they were often accompanied by kind of controversial debates rather than
Positive uh narratives we talked about that a little bit earlier that you know for example if you compare the public debate for example around migration in the US or in Germany so it’s a bit different and so one thing that’s maybe interesting about Germany is that part
Of this debate was always a bit negative German and part of it was was negative because people didn’t expect migrants to arrive so the country in some sense was was not so there was a reluctance to actually acknowledge that Germany has become an what in German call iron B immigration
Country for a long time I think now kind of we understood it right now it’s we we understand that Germany has become an important destination but for a long time that underst that understanding was missing so what I want to talk about I want to give you really like a big
Picture overview of what has happened in terms of integration outcomes in Germany over the last 50 years so first I’m going to show you some very general pattern and then I’m going to ask two specific question the first is how predictable are the integration outcomes of different groups right so if you have
A new group of migrants coming in is it kind of predictable what’s going to happen to what extent is it predictable based on on the knowledge that we have to arrival and another question I’m going to ask is did integration outcomes improve over time right is is this something that has systematically
Changed um and if it has changed maybe do we do do we now know why that is cas if we have time sufficient time so probably not but maybe uh we will talk briefly about there was quite dramatic collapse of employment among some immigrant groups in the 1990s so I want
To show you this briefly and talk briefly about why that might be the case and if we have enough time at the very end we can talk a little bit about the recent Refugee CS okay if you read the news nowadays of course it’s a lot about it’s about ukrainians who came in
Recently it’s about those who came in around 2015 and so we also have some evidence so the evidence I’m going to show you today it’s based on What’s called the German micros sensus it’s like a 1% sample of the population that’s been taken very regularly and so if you ask
How does it differ what I show you today compared to all the other work that’s already out there there is actually a lot of work that um it’s primarily the difference is the data source so a lot of the work out there uses survey data or for example uses administrative data
And so these have different Advantage this advantages I think the advantage of our data source that it’s it’s very representative because really random sample of population in contrast to for example the administrative data which is just those who are working and it’s being taken every couple of years so you
Have sort of a very upto-date picture and a representative picture of what’s happening over time which is in for example in contrast to survey data such as the G um where you always have to wait for new samples to be taken uh of of of new imig varieties there’s one
Other paper which is has a bit of a similar perspective like ours and that is um by uh Strang holds and co-authors and they are very focused on the gender dimension of integration okay so they very carefully compare men and wom and sort of what’s the gender differences
Today I’m completely going to abstract from that I’m going to very much focus on man uh the idea being is that if you’re interested in um how well people are doing on the labor market it’s a bit easier to interpret what you see for men than for wom because for women you also
Have strong culture difference across groups uh to what extent they’re attached to okay so that is very interesting and it’s discussed in this paper but today we sort of focus on more on working so we in particular we’re going to focus on first generation uh men that who have forign
Nationalities um we’re going to consider 36 different cohorts that I’m going to show you in a second by where they came from and when they arrived and first I want to show you a bit of evidence on what happens to their employment you know what was their com rage Natives and
I want to show you what were the income uh over the uh over the time that this so what are these different groups so I think it’s kind of useful for those of you from Germany who will know a lot about it U others maybe not so much so I
Think this is one way kind of of categorizing these different periods of immigration Germany and of course most important one in the beginning was the so-called recruitment period where we tried to get what we call guest workers from other countries okay so the German economy was booming there were all these
Factories manufacturing Mining and they needed workers and so we very actively went to other countries as also some other European countries did to actively recruit workers and those workers came well they came particular from Turkey Yugoslavia but also from Italy for example Spain many other un then there was a what we
Call here consolidation period so this was a period where there was a lot of family reunification so where many of these guest workers got their family in um and we continue to have relatively High migration then we had another wave of immigration in the early 90s and that
Was of course because of the fall of the Iron Curtain so now we had lots of Eastern Europeans coming into Germany um and then we had sort of this increased East West integration um and so we had a few other spikes integration and immigration more recently of course in particular around 2015
Um so for this Refugee arrivals around 2015 from Syria and other countries and of course most recently after the Russian Ukrainian War arrival ofra um I’m not going to bore you sort of how exactly we do this there is a slide uh explaining how we do this but it is actually pretty straightforward
Yeah yeah exactly so the question here is exactly so why do we have these good question like why do we have these different periods it’s related to sample size so we for some groups we had many many people coming in for example from turkey and so we can go into a bit more
Detail we can distinguish you know those who arrived in ‘ 67 and those who arrived later while for other groups we have to look at a bit broader um we have to look for example here from Northwestern Europe at broader intervals because just sample size as much okay so it’s a
Tradeoff between having large enough groups and trying to go in a bit group great so I want to show you kind of how these integration profiles look like uh here’s an example for the rivals of of these guest worker these so-called guest workers uh for example Italy here
Italians are here this would be this line and we plot this by years since migration uh so since when uh are these uh individuals in Germany in this case the very first years are missing because the data doesn’t uh doesn’t start in the 60s it starts a little bit later um but
You see the general pattern here if you compare this to natives which is this thin uh black line for these group we see that the employment rates were very similar so it kind of becomes a bit lower uh if you go sort of to 20 25th years since migration and that’s related
To age because also the natives they grow older and then you have some early retirements and employment rates dip a little bit but generally for these groups employment rates are all very high if you look at some other groups so these are for example immigrants from Turkey from Yugoslavia so also guest
So-called guest workers so their employment profiles happen to look similar in the early period but then you see that something happened uh and so some of these groups H they have this dramatic drop in their employment um to the point that the employment rates towards the end of this period is now uh
20 or 25 percentage points below the employment rate of um of similarly aged names okay so this is something that’s one of these sort of case studies I want to mention later on again it’s like what has happened there why suddenly are these groups doing so badly um but first
Sort of to be uh show you all the evidence let me show you these profiles uh for all these different arrival periods so these were the ones we were looking at this what kind of the first the recruitment period this is the consolidation period so again you see that for some groups
They have very high employment rates but there are quite a few groups that have employment rates that are systematically below the employment rates of Natives and the gaps are often quite large right so the gaps are often 10 percentage points 15 percentage points 20 percentage points so large gaps in in
Employment and by the way if we would look at something like welfare dependency this would be mirrored in dependency in welfare so for these groups also the dependency in welfare is much higher than the dependency of of M workers these are the 1988 995 arrivals
So I think here you see nicely kind of these typical integration profiles that we find in many countries when migrants come in of course they first have very low employment rates compared to natives uh but then they catch up they increasingly find employment and so you get these concave profiles um or
Convergence profiles uh where immigrants increasingly catch up with problem of course what we see here is that in the Geral context is that for many groups we have these large gaps remaining okay so yes they converge to some extent but they end up at a point where the employment rates are still
Much much lower than the employment rate of sim and this is by the way very different to the US context where in the US uh the employment rate of migrants is often higher than the employment rate of mes so this is that’s quite specific to Germany or maybe more generally to
Europe that we have this fairly low inlo rates in more market for completion these are the sort of the most recent calls of course we can’t follow them since so long but you kind of see the same patterns right that for certain Source countries you have very low employment rates and they also
Don’t seem to fully uh fully okay so I think none of this what I showed you here none of this is completely new I think we kind of you know if you open the newspaper of course some of that is is going to be described in the
Newspaper um I think what is new is like really to have this very comprehensive overview for many different groups many different arrival periods and I think one thing that is a bit new is is uh this chmatic grop in employment year for these early exp and that’s why we want
To look at that in a bit more detail later on yeah yeah the literature finds for the United States um and you said that the employment rates in Germany for first generation migrants are lower than in the US um do you have a sense of what drives these differences between the two
Locations yeah so um I think there have of course different hypothesis so this paper is very descriptive I’m not going to put my money on a particular uh particular hypothesis but some ideas are very clear one thing idea is that it might have something to do
With the welfare state that we have a much more generous welfare State many countries compared to the US so the in some sense in the US you’re being forced to find work otherwise as migrant will have greater problems and maybe this is not so much the um not not true to the
Same extent in Germany or other so it’s kind of it’s a of course a well-known problem there’s sort of this trade-off between when when we design welfare systems there’s this trade-off between trying to help individuals who are in difficult situation so you want to provide welfare particular of course to
The poorest and those who struggle on the other hand you’re worried about the disincentive effects you’re worried about that these kind of benefits disincentivize work and so this is a kind of obvious hypothesis that this might be Ro the second I think with this bit more specific potential explanation
Is that I think some of these groups were very unlucky in which industry they ended up so for example many of these guest workers they ended up in manufacturing they ended up then Mining and initially that was great because these are industries that are paying relatively well and they did relatively
Well but then in the early 9s it’s kind of this story taking away some of what we’re going to see later on in the early 90s being in those Industries was very bad um and so some of what we see here is basically immigrants being in the
Wrong industry at the wrong time um and an interesting question is why didn’t they recover right so it’s very clear what happened to them as in terms of like a shock so that many Turkish workers lost their job in the early 90s because for examp the mining industry
Did very badly the question of course the followup question is why didn’t they find out right and we’ll come back to that later on yeah as a follow up to this question there is evidence for the US that there is quite a lot of Mobility internal mobility of migrants so
Migrants tend to react quite a lot to economic shops yeah is there any evidence on that for Germany so we know that of course generally Mobility is lower in Germany in terms of spatial Mobility that’s one of the things that we know that us workers are more mobile they’re maybe more actively seeking
Opportunities and that’s indeed one question is of course why did migrants stay in these communities in these areas with these heavy industries that were depressed and there still depressed in cases um yeah I know the answer one hypothesis could of course be that this has some networks you want to stay in
Your community some of these communities are very tused yeah great right exactly and that was the original idea right for those uh who maybe don’t know this kind of historical context of his guest work period the idea really when they designed these bilateral agreements with sending countries was
Okay we’re going to get some of your workers they come for a couple of years and then they’re going to go home and that’s of course not what happened right those workers stayed um so and I think that’s part of the explanation of what we for example this period this
Consolidation period this was very much a period where the guest workers did not go home but instead they got their families in so this kind of when people realized okay now this guest worker thing is going to work out very differently than we when we intended and
I think this is also the explanation for these type of patterns that we see here is so some of these guest workers in the early 90s they started losing their jobs um and one response of course could have been to say okay why don’t I go back to my home country right um
And we show in in a companion paper well my co-author shows in a companion paper that for example for the Spanish and the Italian migrants this is what happened they were also affected by these closing of Industries in the early 90s but then many of them went back
Home and that’s why they don’t show up as low employment years example if you look at Italians You Don’t See any decline in the P Rak it’s because many went those while for the Turkish they were affected by job loss and they didn’t work they didn’t they
Didn’t go home they stayed in Germany uh partly perhaps because economic conditions were not great in turkey at that time and that we must have played a role yeah yeah you asked about skills or about skills you said no if there was a m mismatch in terms of skills yeah I
Think in particular for I think for the group of Turkish migrants one hypothesis we were really interested in is that this might have something to do with language so the idea is that so first of all it’s much harder for maybe for for immigrants from Turkey to learn the the
German language um and they worked in these industries where language was not that important right if you work in a factory uh if you work in manufacturing heavy Industries you work on certain machines and if you understand how the machine works that’s maybe sufficient so
We think as a hypothesis and we are not the only ones there also other papers that the Turkish Community particular they were in big trouble because when they lost their job in these industries where language that was not so important they should have found jobs in other Industries but the lack of language
Skills may be also coupled with the fact that they were very clustered in their communities which is of course not fostering learning the German language that this must have really help them okay but in this paper we don’t really have direct evidence of that that’s just a plausible
Interpretation okay so I showed you what happened to employment let me summarize what we see so often these employment profiles are concave low employment at arrival but then increasing employment over time so it’s kind very standard we see that in other countries but what’s specific here in Germany is that for
Many of these groups they keep having substantially lower employment rates than the natives so if you wonder what is sort of the uh average Gap so if you average across all migrants the Gap in employment rates to Native workers after one decade is 10 percentage points uh so
That is quite sizable but of course there’s a lot of hogen groups so for those groups who have large gaps at arrival the gaps never close saw that here right that we see uh for these groups here I mean we only have them for 20 years but this doesn’t
Look as as if these gaps are going to close anytime soon and one thing that’s maybe was not so obvious from those graphs um but I can show you in another graph uh that we have over here so this is kind of looking at these averages uh either unconditional so really just
Comparing migrants and natives or conditional where conditional means conditional on education okay so the conditional graph here we look at individuals who have the same education made as in migrants and so these are kind of these average gaps that I was talking about the Blue Line some of that
Oh touch screen some of that of course is explained U by uh difference in education between migrants and natives um and so I wanted to show you this year here you have average employment profiles relative to natives so so the gaps relative to natives for different immigrant groups that either come from
The from EU countries or immigrant groups that have a high share of refugees or low share of refugees okay it’s kind of like just an interesting way of splitting up the data and so what you see is that and again this is not something you know completely new we we
We see similar patterns in other papers is that immigrant groups that come in as refugees they of course do worse early on at arrival okay so they they don’t migrate into a job they often have to flee flee their country very rapidly and so they don’t they don’t do so well in
The first years maybe not surprisingly but if you look at the very long run so after 15 20 years I think what’s interesting is they catch up with the other groups okay so it takes them much longer it takes refugees much longer to integrate in the labor market
But they might end up at the same place as as non-refugees okay on on on average of course it depends on many other things but the the integration process for them takes simply takes longer and finally I think one interesting observation that for some courts we saw that the gaps actually
Woring again over time right so they some groups did relatively well actually full conversions to Natives and then something happened which is maybe also kind of interesting warning for policy makers integration is not necessarily one-way spee so even groups that look like they’re doing well on the labor
Market there can still be much more vulnerable to economic shocks than natives okay that’s one of the patterns we see we see over here okay so we could do the same for income but I don’t want to lose uh too much time so I’m just going to show you
One graph for income so these are these earliest arrival corts um the guest workers and so it’s the same type of graph but now the y- axis is a real personal income monthly income and so again the black line are the natives and then you have the different imigrant groups
And well there’s kind of these outliers these are the Northwest Europeans uh they do better than the natives but one thing that I found striking is how similar all the other imrs are right if you if you remember the for employment there was a lot of variation right if
You look at these plots they often look like that as you know some groups do better than others um but for income there’s very little variation basically almost all immigrant groups have substantially lower incomes uh natives what I think is also striking is that there’s no convergence so typically we would expect
That immigrants at least to some extent catch up overtime Natives and in employment we see that right we see that for most groups that they do catch up to some extent but in income we instead see Divergence the gaps are large why is that well part of the story
Here is of course that the natives have higher education and more highly educated individuals tend to have higher earning BS okay which perhaps good news for everyone sitting here okay so it’s kind of the idea that if you have let’s say master degree or some some higher
Education you uh you will tend to have higher income Grows Right exactly why exactly what could explain this I mean there are many mechanisms one is more U compositional it’s difference in education but then conditional we see these Divergence even conditional education so let me show you
This graph the blue line is just comparing Natives and immigrants that have similar uh age the orange line is controlling for education so now we are looking at migrants and natives who have a comparable education and we still see this Divergence so you know even if you
Compare a native and an immigrant who let’s say both have a a high school degree the incomes will tend to diver diverge over time okay so discrimination could be one uh one story I think what’s what’s probably part of the story is that also right at the beginning migrants tend to sort into
The worst jobs okay and tend to sort into worst jobs not in terms of current pay but also in terms of career um and yeah so we don’t go deeper in this paper but this will be interesting to to study for okay this I already
Said so this was kind of just giving you like a very broad overview what happened to the first generation let me show you very briefly what happened to the second generation okay so we can also to some extent study second generation migrants in this data and I want to show you this
Plot where we compare the employment gaps first generation and the employment gaps in for second generation migrants so second generation migrant here are defined as uh children who were born in Germany but they had at least one of their parents came from one of these Orin
GRS and so what you see is so kind of this diagonal line here this is of course um this is the line that shows uh what would happen if there would be no improvements across Generations but most of these dots are to the left of the line which means that on average of
Course so for some groups we had for example here for some of these Turkish groups we have these massive employment gaps and their children now they their children now they they do much better right compared compar however of course almost all of these dots are still below the zero line
So that means for all of these groups we still see gaps to to similarly AG so that means these gaps shrink across Generations but they only shrink uh to some extent if you want to have sort of the number 25% okay so the gaps that we had in the first generation on average
Shink by about 25% in this I think another interesting a littleit more sepal Point here is that the gaps also become more uniform across the Rival years with an orig so what I mean by that look for example at the Turkish migrants their employment outcomes here are vastly different depending uh by
Arrival Year and that partly reflects that they came under very different conditions right the guest workers they really mated into an employment contract while some of the later arrivals they came maybe as family reunification uh so they had very different situation but the children they do all relatively
Similar independent of when when when their parents and of course the last thing I think that kind of already saw is there are strong inter populations in the sense that yeah for those groups where we had large gaps in the first generation we will also tend to have
Large gaps in this generation okay so it’s it’s there like a relatively clear pattern so the Gap shrink but kind of sort of the ordering between different groups remains relatively distinct yeah so my question again relates to the experience of the us we were talking about it there’s this very nice book by
Leah bustan and ran binski which is called streets of gold and um uh in this book they show that um both now and during the first global ization wave first generation migrants to the United States did not assimilate completely did not integrate completely but the um full catchup took place during the second
Generation um and actually often times the second generation of migrants um more than catch up they do even better than Americans coming from the same socio economic background so here clearly the question SS are very different and my impression is that in the European context the experience of
Second generation migrants matters a lot for the narrative the political narrative about migration um in the United States now withstanding all the debates that there are the there is a an overall positive narrative about the idea of migration so to what extent do you think this second
Generation matters of course we saw that there are big differences in first generation migrants as well but by the second generation maybe there are also higher expectations yeah exactly and I think one key difference of course that the US experienced migration since much longer time right AC us many European countries
Is still a relatively new thing that means we don’t have so many clearly visible second third for Generation migrants so that means we don’t have so many of these success stories right we do have some very very nice success stories and um I think Germany is like
The those with German I think like biontech examples like the primary example right like the migrants from um coming in sort of having this wonderful Innovation and very successful companies um but maybe we don’t have so many of those stories as in us simply because there are not so many second
Third fourth generation migrants around but as you said even in the first is very different right the the key difference with the S is also that or one key difference is that we have these huge employment gaps which of course are mirrored in Weare dependency so if you
Look at what is the share of people in each group who are depending on welfare for some immigrant groups it’s very high and again this is very different yes with employment rund so I think it’s multiple things it’s not having this you know more visibility of these successful second third generation migrants but
Also already in the first generation things that very different but it probably shapes the narrative in in some way yeah great okay so yeah so there was kind of a bit first second generation migrant comparison of course one could go a lot deeper but you know today I want to give
You more like an overview so let’s move on so the next question I want to ask is okay how predictable are all these differences in integration right we saw that different groups are doing very differently and so one question you might ask is if like a new group arrives
Is it kind of can we kind of guess how well that particular group is is doing and I think the short answer is that yes to some extent it’s it’s quite predictable if you if I say predictable you you might ask okay predictive based on what so the first thing we show is
That things interesting is that if you look at individual outcomes like how well they’re doing employment income and so on of course the most natural thing to do would be to relate individual outcomes to individual level characteristics right you might ask okay what’s the education of a person what’s
Kind of age other demographic information of that person but one thing we do in the paper is to show that actually cohort level characteristics are at least or more predictive than individual levels okay so if I pick a particular migrant and I want to know
How well that migrant is going to do um yes I can ask what is that migrants education but I can also ask what is the average education of the group that this migrant belongs to like the origin and rival group and it turns out that these average characteristics are often more
Predictive how well an individual is doing than the individual’s own characteristics yeah yeah yeah right exactly yeah yeah yeah yeah yeah yeah yeah yeah different waves being very different exactly I I get you want exactly yeah so here I mean what I have on slid and what I’m telling you is all
About these average qus we are kind of completely abstracting from the variation within qu so you’re it’s a good question to ask about it um and you’re right I mean what we find here that also these qu level averages that they’re so predictive it might be suggestive that maybe there’s not that
Much variation within each of these cors and one reason is indeed that for some of these waves I think they were fairly homogene genius in terms of their background characteristics for example the guest workers they would come often from certain regions that were not the most highly educated um and so there
Might be reasons why certain cohorts come with certain characteristics on average yeah yeah yeah I know I got it yeah yeah yeah yeah exactly yeah exactly we could maybe look at qu groups that had more variation within the group and then where less variation and then see whether these results change I agree
Yeah now that’s that’s a good idea uh thanks um yeah so what we show in a nutshell then is here if you know very basic things about a cohort such as the average education or the share of refugees in their Cort you can explain about 75% of the VAR that are show okay
Like you have these different profiles for different groups and employment and income if you want to look at that in terms of a regression but you don’t have to uh so those of you who like to look at regression results what we will be interested in here are these R squares
So this is a measure of how much of the variation uh in the outcome of interest for example the employment gaps or the income gaps how much of that can be explained with these cor characteristics and the answers if we we take a couple of those cor characteristics we explain
About 75 or even 80% of the gaps okay so this is to say uh sorry this was initial gaps this is time years after arrival so what we’re saying here is that if you know some very basic things about immigrant arrivals about those cours it’s fairly easy to predict how well they’re doing
On historically of course if you now take this and you kind of try to forecast the future that’s always risky I’m going to do it at the end of today’s talk and then in five years you’re going to tell me that everything I said was completely wrong because the forecasts went completely
Off but at least historically it’s it’s very systematic which groups are doing well or not and like this very simple information you can you can take okay um second question uh let me first ask how are you doing on time before I okay then we’re good so the second
Question you might ask is okay have integration outcomes systematically improved or not over time okay so I think I want to show you just sort of this one graph here uh these are the same employment gaps that we saw earlier now measured 10 years after arrival okay so previously I showed you these
Pictures where we looked at the entire life cycle for the migrant now let’s focus on how well are different immigrants doing 10 years after the arve to Germany and we’re going to look at um look at that over over time sorry the title of the xaxis not not ideal um and
So at the same groups as before and so the blue line and these solid dots these are what we call unconditional observations so this is really just comparing immigrants and natives of the same age the uh oh the hollow dots have a Miss Point um the hollow dots and the stash
Line This is a conditional comparison where we compare immigrants and natives of the same education okay so we’re kind of trying to control a bit for the composition of different CS okay so of course if a court comes in of migrants with very low education sure that affects their employment outcomes and we
Can kind of control for that for these compositional changes of so unconditionally if you just look at the raw gaps things have worsened a lot right that’s kind of this time Trend the solid line we see that in employment gaps they have increased by about Li the table here here by about
Five percentage Points each decade okay so they went basically from zero to 20 25% gaps that we see more of course part of the reason why that is case is well the guest workers did relatively well at least initially um not surprisingly because they basically migrated into a p right so
Maybe a bit mechanical that did it well but some of these and then we sort of see this uh large uh things really worsening kind of in the early 90s late 90s okay so then we started having groups that did really badly on the labor market and that is still the case
For for many many many more rests if you control for education so this is now the dash line this becomes pretty flat actually which is to say that yes integration outcomes have worsened a lot but most of that is just composition it’s in some sense not surprising because many of the groups
Who came in more recently have relatively low education uh they have high Refugee shares and we know that cohorts with less education and higher Refuge shares will will will tend to be the worse okay so our summary here is yes integration outcomes have worsened a lot but it’s mostly because the
Composition of migrants has has changed a lot okay so what I want to spend sort of the last five six minutes on is sort of um kind of this case studies the one I want to just show you very briefly which we call you know we try to pick a
Dramatic name uh to get the readers interested we called it kind of this 1990s employment collapse and so you saw that a little bit earlier uh when we talk for example about some of these Turkish cohorts uh sorry I’m jumping a lot um that seem to suddenly do a lot
Worse um let’s look at that again but let’s this time at the x-axis I’m not going to have years since migration but let let me put on the xaxis just the year of observation yeah so if we we do that then this is the picture we get so
These are the actual Gap uh between uh these Turkish arrivals that came in the 50s and 60s um and similarly H natives so we see that the Gap was relatively small oh initially this um and then something happened right and so what we do in this
Graph is we we talk about this AG component like the first time we saw this we thought okay maybe it’s just H you know maybe these migrants now in the 90s they maybe reaching 50s 60s maybe some of them go into early retirement maybe it’s just that they go to
Retirement earlier than the natives but we control for that that’s this red line and it’s not that so it’s really it’s it’s not an age effect it’s a Time effect something has happened in the early 90s that was really tellable uh for these mic okay and so I’m not going
To go into details now but what we we do in the paper we do different things and one takeaway is that yeah it really had a lot to do with industry structure uh they were unlucky to be in these industries that did really badly in the early 90s and for some reason even
Though the same happened to Italian and to Spanish migrants um for some reason for Turkish migrants they basically remained unimed okay so they did not find back into the labor market after being exposed to these uh industry specific shocks and so you see you know that’s why call it
Dramatic because it is really 25 percentage points right this is really basically a complete large chunk of those cohs can unemploy and never found in the space of a couple of years okay the last thing I want to do is let’s talk about recent arrivals um so maybe we can use what we
Learned about the past and think about a bit more about recent groups right just kind of think about what what we might expect and so what we do here is and I use the word forecast I said that’s always a bit risky um let’s look at what
Might happen to the kind of large number of refugees who arrived in around 2015 and let’s look at maybe what we should expect for Ukrainian refugees right so there’s some details about the data we use and the assumptions we use and so on but let me just show you what
We get so on the left side we have the 2015 arrivals uh after the R Spring from Syria from other countries and so for those groups we also have some information on how well they actually did because they have been now in the country for a couple of years and that’s
The green line right so with the green line is the gap in employment rates between uh these Refugee cohorts and Native Germans of similar age and so what you see is again it’s kind of this typical concave profile right so of course they had very low employment rates in the beginning but
They increasingly do bad better over time so we see convergence but if we take what we learned in the past you know how these profiles usually look like for people with similar education and or refugees with similar education then what we would expect to happen um is this orange
Line and so what you see is that and that’s of course one reason why I’m showing it to you is that yeah what’s happening is quite in line with what you would expect from the past this again it’s kind of this point that to some extent it’s quite predictable the
Patterns now I said the Orange Line you might ask well actually two Orange Lines so what’s the difference between the two Orange Lines well one thing that’s interesting to keep in mind is that in around 2015 the labor market conditions in Germany were actually really favorable we had record low rates of
Unemployment um and if you take that into account then this shifts a little bit of what you would have expected so here the dash line with these short dashes this is taking into account that we had really favorable condition on the labor market around 2015 and so if you
If that’s your comparison point then the sort of 2015 cohort they’re actually doing a little bit worse than we would have expected based on the past okay so this is to say uh yes kind of what we see lines up with what we saw in the past maybe they’re doing a little bit
Worse than what we saw for compar GR groups in the past you might ask why did they worse I don’t know in this paper we we don’t study that one idea is of course maybe it’s just the sheer number of people came right if you have like
One million people coming in at the same time that’s of course a lot more stressed potentially a lot more difficult to find employment than if if smaller numbers what about the ukrainians so here we do the same exercise for the ukrainians there we don’t have so much
Actual data yet on how well they’re doing although it’s if you read the newspapers you will find some reports and this time we actually for once we do it both for men and women and the reason is of course that yeah for ukrainians most who came in were actually women
Right they were not men so doing the exercise only for men for them particularly strange and so if you look at what what you would expect uh to happen compared to natives it’s actually fairly favorable they we would based on what we saw in the past we would expect
Those groups to do well why is that well those groups have quite higher education uh so the educational attainment of Ukrainian refugees who came in is much higher than the some other Refugee um that we have I said this is the forecast might not turn out and actually if you do read
The news you will see that there are some reports that actually the employment rates are quite low for Ukrainian immigrants at the moment in Germany okay so there’s a bit of a puzzle because we would expect them to do relatively well um but from these early reports that’s that’s not yet
Happened okay I want to conclude um so I wanted to try to give you sort of a little bit this overview of integration outcomes in Germany over the last few years we saw sort of like a few basic patterns yeah we see convergence employment but for many groups we have
Large gaps remaining income gaps actually widening over the life cycle we saw this dramatic collapse of employment for some groups in the 1990s so that means as sort of integration is not a one-way speech as a policy maker you should remain potentially you should remain on guard s to say about what
Happens to your migrant groups even if for the moment things look look okay labor market gaps close only partially in the second generation and a lot of what we saw there is highly predictable if you know very basic things about different port characteristics uh you can predict
Relatively well how this ports are doing and we use that to uh to show you some of these forecasts so what do we sort of I don’t want to just just leave you with results I want to leave you with a puzzle so what’s the puzzle so the puzzle after
We’ve seen all of that for us was that in the social science in economics I think in the last decades we’ve learned a lot about what health s is we know a lot about how different integration policies work out for example in Germany now we put a lot more
Emphasis on language training than we did 30 years ago we have a much more permissive citizenship policy because we realize that if you offer migrants a way towards citizenship that this can be really beneficial for their labor market outcomes and so you know there all these good intentions and all this sort of
Evidence that should move policy in the right direction but then if you look at sort of the big picture we we don’t see any improvements right the sort of there’s no systematic improvements in G outcomes it’s still relatively poor in particular in terms of employment um and so this puzzle I want
To you know sorry for leaving sort of with a bit of a puzzle but I think that’s an interesting puzzle of what is going on what can we do what what has been changed okay thanks a lot for thank you very much Yan and I will
Open it up to questions if you can tell us your name where you’re from and uh your program and then uh go ahead yes hello okay well um hello I’m Harrison I’m a first year M student and I have a first question then a conditional one
After that so where where was your data taken from was it the federal republic or did you also get it from the Democratic Republic during the war oh yeah I should have said that this is West Germany uh partly because we started in the 560s and then we just
Kept on going with West Germany to keep it comparable yeah all right now the conditional question um so given the fact that you didn’t take um data from East German IM groups which often came from around the around the warar paack how do you think that could confound your your predictions with groups
Following 1991 and 1992 do you think this will end up having an effect on how groups integrate based on you know where you you took your data from from the West considering the economic conditions are quite different yeah so I think it means that of course we are missing part of the
Picture so there were certain groups in the East for example the Vietnamese uh who were prominent and so we yeah we don’t have anything to say on them um I don’t think it messes up the pictures that I showed you because those immigrants that we had in the west they
Didn’t overlap much with the Immigrant groups that we had in the East okay so for example the West we had many Turkish migrants uh there was not so much the case in the East so there were different groups and generally there were of course much fewer fewer foreigners in
The East so yeah we missing that part of the picture but I think it doesn’t change the what we saw today yeah hi thank you for presenting your findings my name is masimo I’m a first year maiaia student here in bolognia and I guess my question is a little bit more
About the puzzle that you left us with at the end um and then maybe thinking about this in my question directly is do we think that integration policies need to take in more than just the economic well-being of of migrants when they arrive in the country and I think you’ve
Touched on this a little bit by talking about the the need to learn language and to to then be integrated into that society and my mind goes into thinking about maybe even um seeing your findings and more about like where do these people live within the new um host country that
They’re living in for example I think of this idea of the us having more of an assimilation policy that you leave your culture for the US’s new one rather than maybe integrating into one where you still keep your identity as a migrant but then now living in let’s say Germany
With the Turkish example but it’ be really interesting to see about where because you mentioned that Turkish migrants tend to live within societies that are mostly Turkish and I’m thinking about that also in like a Swedish context where they brought in a lot of refugees from let’s say you know
Kurdistan who like you said then refugees then don’t have the same economic benefits but then they also lived in very segregated communities um and then we could see the social tensions that are happening now with the Kurdish population in Sweden so I’m just thinking about how do you design these
Integration policies to not only just make sure that people have a paying job yeah so exactly how should we design integration policy so I I don’t know the answer but let me give you some ideas um you mentioned language I think that’s something that people realized is very
Important right so that’s one of the big differences to what Germany’s doing nowadays to what they’ve done in the past is that now we put a lot of emphasis in language training because we realize that this is probably going to be important in the long run in many d
Dimensions for migrants um and that of course reflects that we kind of understood that migrants will tend to stay for long for long right when the guest workers came in the idea was that they’re going to stay for a couple of years so as long as they speak enough
German just to do their particular job you know why do they need to learn Germans so to say um and now sort of you if you realize that migrants going to stay for their entire life then of course it’s much more worth to really invest first a couple of years into
Language and so on and so that is happening and I think that’s part of the puzzle why don’t we see the benefits yet one potential answer is that these policy changes are just too recent I think another potential answer is that there’s a problem of take up so there
Are some papers that we site in our work saying the problem is maybe not so much what’s on offer in terms of integration policy the problem is that many of these offerings are not being taken up very well um so there are lots of good intentions but maybe a lot of those
Policies are not reaching uh reaching part um regarding segregation and different communities yeah I think we don’t have anything new to say on that in this project but it’s clearly a topic of interest in the literature right there people working on trying to understand what’s the effect of living sort of in
Your community from the host country they have course many potential benefits um you know you might they might help you in integrating in terms of learning what you have to do and finding jobs and so on so we also for example there are some studies showing that depending on where you send refugees
They might have very different outcomes so for example if you send them to some rural areas where they don’t have many networks maybe that’s actually a bad thing right so it can really go both ways uh but on the other hand we also worried about some of these clusters
Where really people don’t have interact much outside of their their Community um and yeah I think there’s a lot of interest in what to do about it of course it’s difficult what to do about it because if you you don’t want to really force people to live somewhere
Else um but that’s indeed for example I know there’s some uh policy experiments going on in Denmark Denmark has a very aggressive uh immigration policy now um and they are I think really thinking about these type of policies like to really break up neighborhoods trying to get sort of uh break the segregation
Patterns uh but of course it’s a very aggressive kind of type of policies if you force people to to live somewhere else yeah yeah then hi thank you so much Dar I’m a first year my student I actually think I saw a Sim lecture by you a couple years
Ago in the UK so thanks so much great um my question relates to you said in the 1990s there was this economic shock and that different migration cohorts were affected differently so that especially the Turkish cor had had difficulty finding um jobs afterwards you said that
Those was part hardly explained by um a lot of the Italian Spanish or some of the Italian Spanish Community migrating back to the countries but the other part that’s not explained by that I was wondering what your prediction is or if you found anything the outcomes why
There was a difference I think there could be interesting why there was difference in return migration why the Turkish didn’t no no no why not the difference of return but for the people the Spanish and the the Italian C State why it was different for this court in
Relations to the it the Turkish Court why it was different yeah um yeah I think the the one reason of course was that at that time Spain Italy did relatively uh better so you know maybe going back from Germany to Spain and Italy was not such a big change in
Economic opportunities while turkey did relatively badly at that time so the Turkish migrants maybe had an incentive to stay in Germany uh even though even if they lost their job um yeah and within the courts that stay you see difference there um in in how people
Manage to find jobs um yeah in the sense that we yeah we see that those who stayed the Italians and the Spanish for example they did much better than the Turkish but part of that it’s hard to say part of that might be the selection that it’s a different selection of
People who who stay for the Turkish as compared to the Italians and Spanish okay so so we do for example we we see that the as a cohort as a group the Italians and Spanish do much better but maybe that’s just because those who were unemployed went back home okay so there’s
A what would be really useful is to have individual level data to really follow people across country to be able to say something more direct on that I think with this data we we that’s just what we have thank you so much thanks and there was was a questioning the
Yeah hi I’m Michael I’m a m student um I was curious if you considered differences in access to the labor market in terms of changing laws over time and just permissibility of migrants versus refugees versus migrants who came in the 70s versus migrants in the 90s the legal system and ability to work
Within Germany because obviously as an American and Europe you already can tell that it’s difficult for us to get a job depending on the VC you have and the laws that have changed over time so yeah exactly yeah I know that’s clearly going to play a large role right like how how
Easy is are you actually allowed to enter the labor market in the first place and of course refugees who arrived they were often not uh allowed to enter the labor market until they were um had their cases being heard um yeah I think that is and I think that’s the problem that
Now we have actually very good evidence not in this paper but other people are working on this question like what’s the effect of giving better legal rights to refugees for example and some of those recent papers have very good research designs they have really evidence causal evidence that’s I think very reliable
And they sometimes find very large effects so maybe not surprisingly if you give uh first if you give immigrants better access to the labor market and also maybe put less red tape on what they allowed to do or not you will see apparently much better outcomes and I think in Germany policy
Makers know that they to some extent some of the reforms we’ve seen we are trying to to make labor to give better labor market access to to immigrants because we know that this might might play a big role the problem with that is of course there’s sort of this trade-off
There is this concern that if you give too easy access to the labor market this might make it also more attractive for more people to come in I think that’s one reason why we don’t see more permissive policies in lots of questions because there’s sort of this trade-off
On the one hand you know that for the existing migrants who already there this might really improve the situation but then policy makers are worried is that this is going to bring in more migrants um and that’s I think where we’re kind of stuck in this discussion
Yeah I’m going to ask a question um what about uh undocumented migration my sense is that in Germany there is not as much undocumented mulations in the United States and of course undocumented migration um raises some issues not everybody likes the idea that people come and work without documents but at
The same time and it provides easier access to the labor market for those who still don’t have documents um so the fact that undocumented migrants can work in the at least the gray market so to say yeah and that exactly which in the US is a important consideration it’s like the
Group of undocumented migrants very large yes and but then often they are still employed and exactly yeah as you said I think it’s less of an issue in in Germany for example in the early 90s there was a period where there was also a lot of undocumented migrants partly
Because there was so much pressure from Eastern European countries and at that time it was still difficult to go from an Isam European country to to Germany um in the data we using they in principle should be contained in there because this is survey data so in principle they should ask everyone in
Contrast to administrative data they should ask everyone documented and undocumented of course there’s still some concern maybe the undocumented migrants are less likely to answer these type of service um so there’s a bit of a concern that we might be undercounting some of the undocumented migrants
Um we can’t split it up we don’t have a question in the data saying I’m an undocumented migrant um but yeah but it could um to the extent that you capture them it could explain some of the differences in employment rates because some migrants in Germany who still don’t have
Documents they just cannot access the market in the United States they would be able to still work exactly and that extended they’re captured in the statistics you would see that they work exactly so s exactly so I see a point you’re saying Germany like some groups are really more sharply excluded from
The labor market because there are fewer options yeah ex yeah I think that plays a role in particular in the first years I mean of course after you saw some of these patterns after 10 15 20 years and there these very persistent gaps I think that’s not about legal restrictions
Anymore that’s about uh just conditional on being allowed to work it’s not not not working out well yeah be that some specific Board of immigrants did not uh integrate in in Germany and then you make this comparison with the United States but I’m thinking that the comparison is not
100% correct in the sense that us is a huge country country and if you end up arriving in LA and then if if there is a downturn in the economy in La you can easily move to another to the Atlantic part of the state so it’s kind of easy
If you end up in Germany and it’s downt in specific local labor market in Germany then crossing the border to France is a little bit more complicated yeah so is it this kind of C cross EU countries coordination immigration yeah a source of lack of integration Germany’s too small There’s
No Escape for labor market shocks no exactly I think this could play a role I think it might also play a role that it’s it was really industrywide downturns it was not just um you know something happened in a particular city in Germany and then you just go to the
Next city and do the same job nor it was really entire Industries doing really badly and that actually we see even in the US even though the US is so large you know those papers for example on trade shocks in the US there are this evidence on that certain areas like the
Rust Belt and in the US as it’s called they also had these industrywide shocks and also there we see that workers in those areas they did badly so it was apparently not so easy for those workers to escape those shocks even though and that’s a puzzle right it’s
Like because it’s a large country I mean you would think they just go somewhere else and then you recover um but it did not happen apparently and so it did not happen for these Turkish migrants either in Germany so these are uh this is a this
Is a synthetic P right so and related to this point you’re not able to get uh entry and exit over time so I mean exactly related to this I mean this is a curiosity again to which extent do you think I’m a big fan of um synthetic
Panel I use it in several papers but you know one of the crucial part when when you do when you use a cohort fix effect then you’re somehow able to look at variations here if you look a long a Trends in particular in a migration setting in which Naturally Speaking you
Have the U enlargement in between so migrants could move and then you have the two financial crisis actually three because also the it one but but you’re I think a crucial point that here could be problematic is the fact that you’re missing when people are going out and in
Because these are repeated screenshots of the of the situation so uh yeah I mean this is no I agree exactly this is an important limitation interation of the estimat so so people going in is not a problem because we always know the arrival here so you know if new people
Are coming in we are not accidentally mixing them up with the old people because we can split it up by arrival what is a problem is people going out uh so basically we are following these as you said synthetic cohorts and of course the composition changes over time because some of those initial
Arrivals uh might have left the country and we don’t know who exactly because it’s not panel data and so you’re exactly right so this affects interpretation um one thing we did is I didn’t show it today but in the appendic we have these figures for each cohort
Showing you just the size of the cord so for example if the size of the cord stays relatively constant then you know that not so many people left okay so then you know that it’s not a but of course for some cords indeed quite a lot of people left for example like the
Earlier guest workers from Turkey from from Italy or Spain and then so this affects interpretation of the estimates I don’t think it makes them necessarily less uh interesting it just means that for example if we see for example employment outcomes improving this could be either because the existing workers
Are finding employment at a higher rate or it could be that those who didn’t find employment maybe went back to their home country these are different interpretations but from policy perspective just knowing that employment outcomes improved is still kind of interesting even though different mechanisms yeah catching up yeah and I was
Wondering is it catching up or is it that you know they’re becoming more skilled and they’re integrated in other ways so on I completely agree it’s a matter of interpretation in terms of policy should I make them integrate more should make them not leave so yeah exactly it starts mattering exactly so
It’s so it’s uh but yeah exactly that’s why for the in the one case where we really started digging to really say something on that is the case of this this employment collapse in the 1990s right and there we really we my co-author has another paper where he
Uses panel data to exactly look into that question and that’s why for those cohorts we kind of know what happened because we had panel data um but you’re right I mean you want to go into more details there depending what’s your question you have in mind yeah
Good so I’m gonna ask a final question and it is aren’t you a little bit too task on yourself in the sense that you’re comparing these migrants um at most controlling for education but you know there are many other characteristics of individuals that affect their success economic and
Non-economic success so uh um for example for second generation migrants the income of parents and I wonder if you control for that if you’re gonna find again this Divergence exactly yeah I think that’s a matter of perspective I mean you’re exactly right like many studies they really do conditional analysis you try
To find migrants and natives who have very similar characteristics maybe similar family background in terms of income and then you see there’s still differences and that’s very interesting for example if you’re interested in questions such as discrimination you want to compare sort of apples to apples and then you want to control for
As much things in some sense as possible here that’s not the perspective that we have so the perspective we have is more the one from a policy maker um you know if certain migrant group does badly um because they have very low education for example that’s part of what we want to
Capture we don’t want to sort of explain that way um yeah that’s a short answer that’s a long thank thank you very much thank you for all the questions and comments and um thanks for all the questions including the very critical ones very good yeah