Intro. [Recording date: June 16th 2022.]
Russ Roberts: Today is June 16th 2022, and my guests are Leah Boustan and Ran Abramitzky. Our topic for today is their new book, Streets of Gold: America's Untold Story of Immigrant Success. This is Ran's second appearance on EconTalk. He was here in October of 2018 talking about his book, The Mystery of the Kibbutz. Leah, welcome to EconTalk; and, Ran, welcome back.
Leah Boustan: Great to be here.
Ran Abramitzky: It's great to be back, Russ.
Russ Roberts: So, your book is a rather creative use of a very large dataset, and we'll get to that in a little bit, but I want to start with what I thought was one of the most interesting findings of the book, which is a comparison of the past and the present. So, America has had different waves of immigration. There were times before, particularly 1880 to 1920, when there was a lot of immigration, a lot of issues around immigration, and then in more recent years, the last 30 or 40 years as well. And, we have very different views often of those two periods, and part of that may be because where those immigrants came from was very different. So, first, Leah, tell us what's different in terms of where those immigrants come from, and what have you learned that's the same, which I think may surprise some people.
Leah Boustan: Well, the first thing that's the same about immigration 100 years ago and immigration today is that immigrants make up the same proportion of the population. So, around one in every seven people in the United States today is foreign born and the same proportion was in place in 1900.
But beyond that, there are a number of differences, actually, between immigrants now and then. So, in the past, immigrants overwhelmingly came from European countries, and those that did not come from Europe came from Canada. So, I think we have this perception that immigrants a hundred years ago came from cultures and societies that were closer to the U.S.-born. Today, immigrants come from all over the world, and 75% of immigrants are either from Asia or from Latin America. So, those are newer sending countries.
Another really important difference is that as long as you were coming in from Europe a century ago, you essentially could move to the United States unrestricted. So, you didn't even need a visa or a passport for travel, and you did not need to have an employer sponsoring you or a family member sponsoring you.
These days, there's a much stricter set of rules, and there's a lot of bureaucracy involved with moving to the United States. So, the number of visas that are open is smaller than the number of people who want to come to the United States. Therefore, we have around a quarter of our immigrant population today that's undocumented. So, they either came into the United States with a visa and overstayed their visa or they came into the United States across a land border and came in without papers.
So, those are two really important differences between past and present.
Russ Roberts: The other thing I would just add that I think you can comment on, Ran, that I think is often misunderstood, is that: Well, in the past, America was friendly towards immigrants maybe because Europeans were more likely existing population--as Leah suggested--and there was no problem of assimilation. It wasn't they learned English. And, I think most people--we're talking about America today, by the way, I just want to emphasize, although we may get into Israel where Ran's originally from and where I'm living right now; Israel has some very interesting immigration issues--but the thing I want people to realize is that the idea that immigrants aren't going to assimilate is a very, very old idea.
I lived in St. Louis for 14 years, and St. Louis had a large German population arrive, I think, around in the last part of the 19th century, and the biggest fear there was that those children would never learn English--the children of immigrants--and that Germans were going to stay German. They'd never assimilate. Of course, it turned out not to be true. So, talk about this issue of fear and uneasiness, Ran, when we look at past immigration versus today.
Ran Abramitzky: Yeah. So, it's a very old idea, as you say. You hear today some politicians, they say very bad things about immigrants: They can't assimilate. And we talk about second and third generation: There is no assimilation. You're like, 'Oh, that's shocking. In the past, it must have been much more positive.' But in fact, it was very similar in the past. Senator Henry Cabot Lodge, who was one of the architects of the immigration restrictions that were put in the 1920s, said, 'Immigrants are from the lowest races and lowest classes and they have lower levels of IQ [Intelligence Quotient].'
Over and over again in this book, we see that you kind of hear the exact same concerns, only they are targeted at different immigrant groups.
So, it started with complaints about the Irish, and then later on it was about Italians, and then later on it was those pre-, past-Italians and Germans and English talking about Mexicans and Muslim immigrants. So, you hear very similar concerns, only targeted at different immigrant groups.
Russ Roberts: Leah, do you want to add anything to that?
Leah Boustan: Well, one thing that was fascinating to us and really unexpected is, if you try to quantify attitudes towards immigration rather than just picking out a couple of politicians--that you actually go to the Congressional Record and you find all the speeches about immigration--actually, speeches about immigration were far more negative during that Ellis Island period, where most of the immigrants were from Europe. Overwhelmingly, the way that Congresspeople and Senators spoke about immigration between 1880 and World War II was negative. And, these days, there's a number of positive statements about immigration on the floor of the House and Senate.
Russ Roberts: There's a lot of interesting data analysis in your book, the Congressional Record analysis. You looked at every single speech in the Congressional Record, correct? And used machine learning to look at the ones that were about immigration, and then you had to assess whether they were positive or negative. So, we should just let listeners know: you weren't pouring over a microfilm or leafing through a bunch of--volume after volume of book. You did this rather creative analysis of the digital version of the Congressional Record, correct, Ran?
Ran Abramitzky: Yeah, that's right. We took all the eight million Congressional speeches, and we had a team of research assistants that went through them and decided which one is about immigration and which one is not about immigration and how positive and negative they are. Then we basically were teaching the computers to identify speeches about immigration in the entire Congressional speeches and identified about 200,000 Congressional speeches as a result.
Russ Roberts: I have remarked on here that Allan Meltzer might be the only person in the world to have read every minute of the Federal Reserve meetings; but I wish one of you had actually read all eight million speeches, but in this case, it's not true. The computer did the reading. That's correct, right, Leah?
Leah Boustan: Right. So, you do need to train the computer.
So, we had the army of RAs [Research Assistants] that were reading a number of snippets, thousands of snippets through history, and picking out words and phrases and attitudes that would tell you as a human being, 'Oh, this is a speech that's positive and laudatory towards immigrants, the contributions they make, their family or maybe representing them as victims of persecution that need our help,' or 'These are negative speeches that are talking about immigrants bring crime, they bring drugs.'
So, it's relatively clear to a human being what's a positive speech or what's a negative speech. And then you kind of have to teach the computer what sorts of words and turns of phrases are associated with each. But that way, you can actually classify 200,000 speeches, which is really just not possible with the human eye.
And I was really blown away by those findings. I did not expect that when immigrants were primarily European, that the speeches would be so uniformly negative, and that today with immigrants being much more diverse, there's actually differences of opinion, and there's a number of politicians who are very pro-immigration and then a number who are anti-.
So, my impression or my bias had been when immigrants were from Europe and looked more like the U.S.-born, that attitudes would be more open. And in fact, they were very much not.
So, some of the anti-immigrant politicians that Ran's pointed to, those are not in any way outliers. Those actually represent what was the common sentiment at the time, which was very anti-immigration.
Russ Roberts: Yeah. I would summarize the data as, on average, negative throughout history, and that in recent--and remember we're talking about politicians now, not people in the street, although they often reflect one, the one reflects the other--but in modern times, what was interesting, I thought, was the gap that opened up between Republicans and Democrats. So, that the average political speech--politician speech--in recent times is mildly negative, not much different than the mildly negative attitudes average of the past, but now, there's a partisan difference that was not there in the past. The question I have is: in the older time period, say 1880 to 1920, were there politicians who spoke positively about immigrants?
Ran Abramitzky: You hear it--it's more rare to hear a very positive speech about immigration in the past.
Leah Boustan: Yeah, and there were four attempts to close the border between 1880 and 1920. So, the typical Representative was anti-immigration and consistently so. What was preventing the border from closing in the late 19th century was actually the efforts of the President. So, the President continued to veto border closure legislation such that we needed a super-majority to actually close the border. And eventually, we got a super-majority, and the border closes in 1921. Can you imagine any political issue these days getting a super-majority--
Russ Roberts: Yeah, it's true, very rare.
Leah Boustan: of the House and the Senate unless it's something very ceremonial, like, 'We're going to name a building after this politician who has died.' Otherwise, a real substantive issue where there's a super-majority would be quite rare.
Ran Abramitzky: We do look at Presidents. We have 5,000 Presidential Speeches, and one thing we find is that Presidents tend to be more positive than Congress when they speak about immigration. You can probably guess the one modern exception for that, but overall, Presidents tend to be more positive about immigrants than Congress.
Russ Roberts: Yeah. In general, Presidents tend to be more positive about trade. And of course, immigration is sort of trade. It's at least exchange of certain kinds. It's investment flowing across borders.
But it's interesting to me: I wonder how decades and centuries of gerrymandering have created the unlikelihood that a Member of Congress would be positive. Obviously, a Member of Congress who has a lot of immigrants in his or her district is going to tend to be more positive; or who has members, has constituents, who are sympathetic. I wonder if that was rarer in the past. I don't know. Just a thought.
Leah Boustan: Well, you can also see the positive change in the Senate, too, but it's slower. So, the Senate is not at all gerrymandered, just reflecting the different states; and the polarization in the House happens earlier by party, and then it catches up in the Senate as well. So, I think there's something real going on. It's not just a matter of grouping the voters into clusters that are pro or anti.
And I think that I completely agree with you that that's what differentiates today from the past. There is a steady drumbeat of anti-immigrant sentiment throughout history. What's different now is that there is a group that's pulling away that's pro-immigration and trying to make the case and is supportive. That was not at all present in the past.
Russ Roberts: Let's turn to the financial/economic side of this, which is, actually, I think the most dramatic piece of this comparison in the past and the present. There's a certain view that America in the 19th century, maybe the early 20th century, was the land of opportunity. Immigrants came here because the streets were paved with gold, which is where your title of your book comes from, Streets of Gold. People flocked to American cities, to all kinds of pieces of the American countryside in that pre-1921 immigration period.
Of course, many of us are their children, their grandchildren; and we've done pretty well. So, there's a view that immigration in the past was pretty great. There's a worry that immigration of the present is not like that. It's not nearly as rosy. There's a belief, a widely held belief that mobility in the United States has stagnated, and so that when poor people come here--meaning, not poor people, but people who come here from poor countries with limited skills and often limited English, which means even if they have skills, they might not be able to use them well in the American economy--that they're going to stay poor, they're likely to become members of the welfare state, and that part of the reason we need to restrict immigration is because we're just importing unproductive people into the United States. At worst, and at best, they'll be low-income people who will do a bunch of services, provide a bunch of services that maybe Americans don't want to provide.
And what you argue in the book--and I want you to first summarize the argument and then we'll talk about the evidence for it--what you argue in the book that both those view are wrong: that the past is not as rosy as we think it was and the present is much better than it looks. So, Ran, start us off and tell us, what do you find when we look at the past and the present in terms of mobility?
Ran Abramitzky: Right. So, there is this nostalgic view that in the past things were great and immigrants move from rags to reaches very quickly, whereas today, immigrant groups aren't able to do as well. And so, when we look at the first generation--so the immigrant themselves in the past--we don't find evidence for the rags-to-riches myth. So, we find that both parts of, in fact, of the rags-to-riches story is wrong. They didn't start in rags. So, some immigrant groups started, actually, even when they first came to the United States, they did better than the U.S.-born workers, and they continued to do better throughout their lifetimes; and other immigrants started worse than the native born, and they continue to not catch up even 30 years later.
So, no rags-to-riches in the past just like today. But what we do find is that the children of immigrants--by the second generation, the children of immigrants are doing remarkably well. So, both in the past and today, the children of immigrants are incredibly mobile. They are, in fact, more upwardly mobile than the children of U.S.-born; and even when they start poor, they completely catch up with the children of U.S.-born. And that is true for immigrants, for Mexican immigrants today, and it was true for Danes and Swedes in the past.
So, I guess immigration is a novel. It's not a short story. So, they come. By coming to the United States, they are increasing their income dramatically. Then, they kind of don't fully catch up in the first generation, but their children are doing great, both in the past and today.
Russ Roberts: And, I think for most of us--well, I shouldn't say for most of us--for me, the interesting part is the children part. The fact that immigrants don't come to America and suddenly thrive is not surprising to me. They have all kinds of challenges as an immigrant here in an alien culture in Israel. I have a little taste of that. I haven't gotten a raise, by the way, just for example. I'm stagnant after a year here, but I think that's the less interesting part of the story.
I think the more interesting part of the story is that both in the past and in the present, the children of immigrants do dramatically better than their parents and dramatically better than their American-born counterparts. Leah, why don't you continue the story and talk about that?
Leah Boustan: Well, there's two ways to look at this picture. One is that the children of immigrants on average are being raised in poorer households than children of the U.S.-born. Many immigrants, especially today, are coming in from poor countries, so their parents may never have had the opportunity to go to high school. They're working in low-paid service jobs. So, in that sense, it would look really dramatically like an improvement if the children of immigrants just catch up to the average. And that's what we find if we're looking at all households. Children of immigrants are being raised in poorer households, but yet when they themselves are adults and they're in the labor market, they look very close to the average. And, there's only a few countries where the children of immigrants are below average, but even in those countries--and this is Mexico, Haiti, a few other Caribbean countries--the children have moved up quite a bit from their parents.
So now, a second way of looking at it is to say: 'Let's compare apples to apples.' In a way, we can see the dramatic rise of the children of immigrants in the averages, but let's take two households that are equally poor--households that, let's say, are at the 25th percentile of the income distribution, which would mean both parents working minimum wage jobs full-time. There's nothing special about the 25th. You could take the 30th or the 35th--anything that's below the median and say: 'Let's compare households that look the same in terms of income, and what's different about them is that the parents are either U.S.-born or foreign-born.' And there, you see that the children of the foreign-born parents, by the time they reach adulthood, have surpassed. They've reached a higher income level than the children of the U.S.-born family that was otherwise in similar circumstances.
Russ Roberts: So, they both moved up, often over a generation, American or foreign-born; but the foreign-born move up quickly. I have some thoughts on why that is. Ran, why don't you give us what your argument is?
Ran Abramitzky: Right. So, at first, the first reaction of many people is, 'Of course, immigrants are so wonderful. They are heavy entrepreneurial spirit. They come and they are great.' And part of it may as well be true. But, what we find is a more kind of mundane, if you want, explanation, which has to do with location. So, immigrants tend to move to those areas that offer higher economic mobility for everyone.
In the past, this meant that very few immigrants settled in the U.S. South, which was a place of very little economic mobility for everyone. So, immigrants tend to move to places that offer mobility for everyone. Whereas, the U.S.-born are more rooted in place. If you ae born in Alabama and your parents were born in Alabama and your grandparents were born in Alabama, and you have your friends and network there, moving away from there is leaving home. They don't just do the rational calculation of where the returns to economic activity is the highest, and that's understandable. But immigrants are more footloose. By moving across the Atlantic, they have revealed that they are comfortable to leave home and so they may as well settle in those places that offer mobility to everyone. So, that's kind of one part of explanation.
The other part of the explanation that we find is this story of the Russian scientist who come to the United States and he's a taxi driver. So, somehow, he's much underplaced relative to what his abilities would suggest, but his child is converging more towards the scientist-father rather than towards the taxi-driver-father. So, immigrants are underplaced when they come to the United States and their children reach to the full potential when they grow up.
Russ Roberts: Of course, they also have the disadvantage of usually having imperfect English, the parents, and that's going to handicap them in many occupations--not necessarily, but in many. As a result, their children, who generally don't have that disadvantage, are going to rise more to where they might have been otherwise, Leah, you want to add anything?
Leah Boustan: Yeah. On that last point about underplacement, we compared two different kinds of immigrants: immigrants who arrived themselves as kids, or immigrants who arrived in adulthood. So, in terms of limited English, limited labor market networks, if you arrive as an adult, that's certainly going to describe your experience. An immigrant who arrives as a young teenager, though, might be able to make up some of those disadvantages.
So, exactly we find that it's the children of immigrants who arrive as adults who are experiencing this upward mobility, which suggests that part of the move up from your parent's situation is because your parent, by arriving as an adult, was underplaced in the income distribution. He probably could have earned more if he had native English--if he didn't speak with an accent and so on. But, because he arrived beyond the age where you can really absorb language quickly, he ends up underplaced.
Russ Roberts: Ran has an accent--one of our guests--and of course, he's only made it to Stanford. It's too bad.
Ran Abramitzky: Imagine all I could have done, Russ.
Russ Roberts: Exactly. But the fact is, you arrived here with much worse English than you have now, and you were much younger.
Ran Abramitzky: Oh, absolutely.
Russ Roberts: I happen to know that. Actually, tell listeners when you did get here and a little bit of your story.
Ran Abramitzky: Yeah. So, it's like you said. I write better in English than I speak, and for a long time when I moved to the United States, I felt that I'm not really myself in English. Nobody laughs at my jokes even when I tell them in Hebrew, but when I used to tell them in English, they completely didn't understand what I was saying, and then my personality even didn't come through.
It's a little bit like Amos Oz used to say, 'In Hebrew, I say what I want, and in English I say what I can.' So, that I feel is still true for me today, but in the past, it was much more binding. I came to the United States and I felt that I have to catch up on my English, and that is when I was relatively--I came to do Ph.D. So, think about immigrants who come in lower-skilled occupations, and that's a real challenge.
Russ Roberts: Yeah. I would just add, as would-be Hebrew speaker: part of it is you don't understand the sense of humor of the culture you arrive in, possibly. But the other part is that, even if you are enunciating the words and syllables and letters correctly and not keep using, say, the word for cheese, which is g'vina, with the word for right, which is yamina--which is the kind of mistake that immigrants make all the time, right? They blunder and misplace a letter. But even if you don't misplace a letter, you often put the emphasis on the wrong syllable because Hebrew and English and other languages differ by that.
Or worse, you don't surround well-expected phrases with well-expected context. There are certain words that you expect to hear in a punchline, say, that you are not aware of as a foreigner. And so, you make what you think is a funny joke; it's followed by silence; and you want to say, 'Don't you people understand Hebrew? Isn't it your native language?' But the answer is: 'It is, but you've just butchered it in a way that you can't even be aware of. You got the vocabulary right. You even said it grammatically correctly; but you put the empha'sis on the wrong sylla'ble and you're cooked.' And that kills humor, obviously. Leah, do you want to add anything?
Leah Boustan: Well, I do, and sort of to bring it back to the data, one thing that was incredibly fascinating in the work we did for Streets of Gold is that we uncovered around a thousand oral histories that were recorded. These were audio files of people who had come through Ellis Island now 100 years ago.
Russ Roberts: Wow.
Leah Boustan: So, these were recordings that were done in the 1970s and 1980s when the immigrants were getting to the end of their life, and they were describing why they left Europe, what they found when they first came to the United States, and then how successful they had become or not become later in life. And, what was really interesting about these recordings, to us as as social scientists, is we can actually pick up on all of those subtle elements of language that you really can't get from a survey that's asking someone, 'Are you fluent or not?' or 'Do you speak English well, very well, not so well?'
And, we had research assistants listen to these stories and record for accent. We can record for vocabulary--not so much from listening, but actually from transcripts, where we can see, 'Are you using a simple word like cheese or are you using a fancy word like brie?' So, there is more complex ways and simpler ways of saying the same thing. And then, even: how long and complex are your sentence structures?
So, this was really something that we were able to do for the first time. And we can watch immigrants as they improve in English, and we can see which countries, which arrival periods, which religions, etc., have eventually received higher English ability by the end of life.
One finding that really stuck out was that refugees--people who explained that they were fleeing from persecution--ended up with more complex English by the end of life. We have some theories about it, but it seems like a lot of immigrants that came from Europe for economic reasons thought, 'Maybe I'll go home.'
So, Ran came to the United States for economic reasons--to improve his education. And there was a time where he was thinking, 'Maybe I'll go back to Israel. I'll work at a university in Jerusalem or Tel Aviv.' And he ended up staying at Stanford. But I think that that's a story for many economic migrants--is that, they're always planning on the potential 'I'll go home.' So, that may hold them back in terms of really investing in English ability.
Refugees are different. They think, 'As much as I would like to go home and I may have nostalgia for my childhood place, I feel like I can't go home, and so I really have to go all in on investing in English.'
Russ Roberts: Yeah. I like that. You want to add anything, Ran?
Ran Abramitzky: No. It's good. Leah is perfect.
Russ Roberts: Now, a lot of your analysis--there are different kinds. We talked about the Congressional Record analysis. You just talked about some of the oral history. But, the bulk of the work you did is a very creative culling of data from Ancestry.com; and then merging that, often, with more recent census data and tax data. So, first, tell us a little bit about the Ancestry.com story because it's kind of amusing. Ran, you can start us off on that, and then we'll talk about some of the limitations of the data that might reduce the robustness of some of your findings--and then we'll let people make their own judgment on that. But let's start off with how you did it. Ran, what'd you do with Ancestry.com?
Ran Abramitzky: So, Leah and I were walking down in Huntington Gardens talking about immigration and how to do immigration research when we got to the idea that we can actually--rather, we can go to Ancestry.com, pay $29.99, and get ourself an account; and then you can look up your great-grandparents in the Census records. We can do the same thing, only multiply by millions. We can look at thousands of thousands of immigrants in the Census record.
So, I guess the story that you are alluding to here is that one day, I came back to the office and I had the--the answering machine was blinking, and I listened to the message and it was a lawyer from Ancestry.com who wanted me to call back, and he was like, 'Hey, you seem to have a very large family.'
For a while there, he was using very legalistic language because all they have is their data, and they were afraid that we are after their data. Somebody's downloading their data; and what else can they do with it but selling it?
And, by the end of the conversation, when I explained to him that what we really do is we are social scientists trying to look up immigrants and follow them over time and look at their children and see how they do in the United States and all of that--by the end of the conversation, he was asking me questions, 'Wait, but what about Italian immigrants? Are they doing well?'
And then I knew that we were in good shape.
Russ Roberts: But I was a little confused by that because--so, I know how to look up my grandparents. I've never done it. I have relatives who have, and it's pretty cool. But, I can't look up any other people's grandparents because I don't know their names. So, Leah, how did you find millions of data points? And then, once you did that, how do you link them to taxes or census records? You don't have their Social Security number, the normal things that economists would use even in a privacy-protected dataset.
Leah Boustan: Well, first of all, we don't actually do our work on the Ancestry.com website. We now have a research partnership with Ancestry, and we have access to all of the digitized census records. And, I just want to pause and do a shout-out to the many volunteers who digitized this data, primarily for the Mormon Church. So, it was really an enormous undertaking because, by the time you get to the 1940 Census, you have 100 million people living in the United States, and each one of those records is digitized.
So, now imagine you have access to these files that, by the time you get to 1940, would be a 100 million--as you go further back in time would be somewhat smaller than that--with all of the names, and ages, and places of birth, and street locations, neighborhoods, occupations, etc., of the people in the Census.
We have to make choices about what pieces of information we want to use to try to link over time. We don't have a numerical ID [identification] like a Social Security number. So, we have to think, 'What about a person is so fixed that it won't change by the next census?' And, we selected to use first name, last name, age--of course, you add 10 to age because 10 years have gone by--and either state of birth for the U.S.-born or country of birth for the foreign-born.
And, you know,it's not a cut-and-dry or simple approach. There's a lot of dispute in the literature about which attributes of a person should you use and how should you combine those attributes to try to find the person 10 years later. And there's a variety of algorithms and approaches out there. But,in theory, you can understand how you might be able to find a Leah Boustan 10 years later, born in Massachusetts--that sort of thing in the Census.
Ran Abramitzky: If I can--
Russ Roberts: Yeah, go ahead--
Ran Abramitzky: add one thing to this. So, just think about it this way. You might be worried, 'Oh, you can look me up.' But, after 72 years, all the records become publicly available and you can see the person's name.
And so, for example, if your name is James Smith, we don't really have hope to find you because there are many of them. Even when we look at a James Smith born in a particular year, in a particular country, there are too many of them. But if you look up Ran Abramitzky, there is only one in the United States, so very easy to find in the next census, right?
We do miss out kind of half the population by this procedure because women, they tend to change their last name when they marry. So, we couldn't systematically link them across censuses, although nowadays, there are real good advances in using things other than the census, like the marriage records and the birth certificates to improve improve on that. But, if you just do census-to-census linking, you end up linking people by their first and last name and place of birth.
And then we developed over the years--one of the things we do alongside what we do now is we get more and more sophisticated in our algorithm. So, the way you should think about it is you can be very, very conservative. You can say, 'I want to link you only if I see one Ran Abramitzky in the next census, and I don't see anybody else within five years of the birth date of Ran Abramitzky,' and then for sure you will link the right Ran Abramitzky. But you will have a relatively smaller sample. Or, you can say, 'Well, it's too much to be that restrictive. Why don't I just look for --if there is one person, I'll call it a link.' And then you will have a larger sample, but with higher chance of false positives.
So, the way you should think about it is we developed algorithms that are more or less conservative, and we generate multiple samples that way that are kind of trading off false positives with false negatives. And then what we talk to you about today are all those results that are kind of robust across all these linking algorithms.
And we also--along the way, by the way--we realized that it is a tough thing to--it is very labor intensive to go through and link. So, what we do is we link all the censuses to all the censuses using all the ways we know of how to link and try to make this available to other researchers on our Census Linking Project website.
Russ Roberts: We'll put a link up to that because I'm sure most of our would like to play with that data, but--probably not, but in case you do.
Russ Roberts: So, let me see if I understand this, because it's kind of extraordinary. Let's start with 1940. So, 100 million people in the United States, approximately, in 1940--other than homeless people, people who were sleeping under a bridge that night when the Census came around. In 1940, you have every single living American, approximately--I mean, more or less. It's not a sample?
Ran Abramitzky: That's right. In the past, it was a long form, yeah.
Leah Boustan: I mean, we should.
Russ Roberts: Yeah. Okay. They're both nodding on the Zoom call, for those of you listening at home.
So, I have 100 million people, and for those people, I have an enormous amount of information from the 1940 Census, including income. I have income. I have address. I have age. I have how many children, marital status, etc., etc.--a whole bunch of demographic and financial data. I might have education levels at 1940. Okay? Yes?
So, in 1940, I know a lot about people, but of course, this data is going to go back. We're going to go back and forward. We're going to go back to 1880, ideally, and ahead to 1960 and beyond.
And of course, before 1940, as you remark in the book, there's no income data. So, it's going to be a lot harder to measure people's income in 1910. But basically, you're going to try to create--there's death also, of course, and there's a point where a person isn't going to be born when I go back in the past because they hadn't been born yet. But I'm going to go back in the past. For somebody who is, say, 10 years or older in 1940, I should be able to find them in 1930, and find out what they were doing then. And similarly in 1920.
So, this is a decadal, decade-ish analysis, correct? You don't have annual data for these folks, right? But you have the entire population of the United States. And the reason that's interesting is that until this analysis--which, I'll just plug your 2021 American Economic Review article, which is a more technical article than the book. The book can be read by anybody. You don't have to be an econometrician. You don't have to be an economist. It's very accessible. But, in your formal analysis that the book is based on, what's novel about your analysis? In other words, who cares about Ancestry.com? Why do I need that stuff? I mean, is it because the 1930 and 1920 and 1910 Census isn't available? So, tell me what the Ancestry.com stuff adds.
Leah Boustan: Well, I think people were very excited when Raj Chetty was able to use the tax records to look at social mobility in the modern period. So, he was able to look at kids who were born around 1980 and then follow them to the point where they're around 40 years old in the labor market these days, and because they were a tax dependent of their parent in childhood and then a Social Security number is there for them in childhood, we can see what their parents were earning and then we can see what they themselves were earning. And we can slice and dice that data in a whole variety of ways: look at different racial groups, look at different geographies.
And that's not something that anyone was able to do for the past.
So, it was a real revelation to learn about social mobility today, but without this census data that we're talking about where we can try to link everyone to everyone and see what do their childhood experiences look like and how are they doing as adults, we really can't assess whether social mobility has been declining in the long run. Was there really this age of American exceptionalism in the late 19th century where people could go from rags to riches, immigrant or not. And, how important was immigration in that story? Like, are immigrants particularly socially mobile?
I think we have a lot of myths about that period, some related to immigration and some broader--you know, about the frontier, about what it means to be in a free and open society, and that somehow we have this impression that's been on the decline. But we can't assess that unless we have the historical data.
Ran Abramitzky: Any question that has to do with how a person does over time requires a panel data. You require to link that person over time. Every question that has to do with how the child is doing relative to the parents--so anything that has to do with progress over time requires a linked dataset of sort. Before you link these datasets, what you will have is a snapshot of the economy, say, in one point in time.
And so, for example, just to give you a concrete example from a myth that might have come up that way is: Imagine you have a snapshot of the economy and you ask yourself, 'How do immigrants progress over time?' You might be tempted to compare an immigrant that is here for a few years with an immigrant who is here for many years.
Although the immigrant that is here for a few years is a Mexican immigrant, the one who is 30 years here is a German scientist, and you are not comparing the same person.
So, it's only when you are able to link the same person over time that you can see what happens with the person when he first arrived and then 30 years later, and then what happens to his child. That's, kind of, I think the promise, I guess, of the approach.
Russ Roberts: So, I just want to put in a digression that, like you said, people were very excited when Raj Chetty did that. I was less excited. I'm not as big a fan of that work, I think, as the average economist. I've written a little bit on why I think their analysis is overly pessimistic. We'll put a link up to that. But, I would just mention that what he focuses on--and with his co-authors, I think there's six of them, we should be fair; we call it Chetty, et al--Chetty, et al are looking at often the simple question of whether a child is doing better than the parent when they grow up to the same age. And that's a very difficult thing to measure when there's dramatic changes in both in prices and in quality.
There's really, I think, a challenging question that to take quality into account accurately is hard to do. He has issues around the number of children and the number of people in a household.
And so, he has to make a bunch of assumptions. And when he makes those assumptions, he finds out that in the modern data only half the children are doing as well as their parents. Which means that half are doing worse.
Then when you make different assumptions about those choices you have to make--and what I love about Chetty, et al is that they let you see what those differences are--you get a very different picture.
So, I think a reason much more optimistic that there's much more mobility in the United States, the children--that a significant proportion, I think, it's something in the 1960s or even potentially the low 1970s--do better than their parents, which you could still say that's not good enough, it should be better than that if you want to believe in the American dream. But, I think that listeners and readers of the newspaper who often only get a number--half the American children today are doing as well their parents and half are worse--often, it's important to look at the Appendix. It's always good to keep that in mind, but I still don't--
Ran Abramitzky: But, to be fair for Raj, et al, you're talking about absolute mobility.
Russ Roberts: Correct.
Ran Abramitzky: They also look at relative mobility, which is the same thing that we are looking at now, which is looking at their rank/rank correlation between parents and children, and then they find that the children of the U.S.-born who grew up in the 25th percentile, let's say they reach to about the 45th percentile of the income distribution. And we replicate this finding with the modern data that we do. But what we find is that the children of immigrants growing up in the same households are doing better. So, they kind of get to the, let's say, the 51st percentile of the income distribution.
So, there is a more optimistic story for immigrants relative to the U.S.-born. It's consistently defined.
Russ Roberts: Yeah. It's a great point. When you use that phrase, 'rank/rank,' what you're saying is that if I'm looking at where you are in the distribution relative to your peers, I've controlled implicitly for these issues of inflation and quality--at least they're not relevant in that comparison. It's a great point.
I want to also emphasize the second point, which is fascinating: that poor--excuse me--children growing up in poor immigrant households catch up more, do better, rise to a higher relative place in the income distribution than U.S.-born. And, that could be for reasons that have nothing to do with--as we talked about earlier, that could be for reasons of their parents were unusually low in terms because of language or cultural reasons, or it could just be that different kinds of people come here relative to the native-born in terms of grit, perseverance. Some of it's luck, obviously. Maybe a lot of it's luck.
But, anyway, these are really interesting.
I just want to emphasize that there often are a lot of decisions that have to be made in these kind of analyses that affect what the bottom line is. But, the point you're making here that I think is very important is that, in many ways, your work replicates their findings: that children growing up in poor households, native-born, do better, but as well as we might want or hope, but that immigrants are doing better than that. Leah, do you want to add something?
Leah Boustan: I just want to add that in the end of the day, what's really special is the new data creations. So, we now can argue about how to exactly interpret the mobility findings because we have access to that really tremendous data. Before the work with the tax data, we did have some small longitudinal surveys--
Russ Roberts: Correct--
Leah Boustan: We had surveys that have, like, 10,000 households and follow 10,000 households over time. But really, the box we can open up by looking at the full population is tremendous. Whether we agree on exactly what the findings are, we now have a chance to debate that in the scientific community.
And I think what we're trying to add with bringing in the historical census data is to try to extend our data availability as far back as possible into U.S. history so we can continue to have those debates.
Russ Roberts: So, that's what I want to clarify now, because I think we haven't quite gotten to the punchline there. So, Chetty and others have data from 1940 onward about income. What did you add? You've added data before 1940--you can tell us how far back--that is partly--I mean, in other words, let's take a simple case. I don't want to take Ran Abramitzky because he's a one-of-a-kind. And I don't want to take James Smith, even Joe Smith, who is numerous. Let's take somebody a little bit in-between. So, we'll take--anyway, somebody who's got a relatively unusual name; and we have that person in 1940. We know their name, we know where they lived, etc. Now I want to link--and that person's over 10 years old. So, that person was alive in the 1930 Census. Do we not have them linked before? What's different here?
Ran Abramitzky: Right. So, first of all, I think it's easier to think about linking forward conceptually than linking backward. But, imagine that before what you had was you had the 1930 Census and you could look up and you can see the name and you can see everything, but you can see Russ Roberts and you will see how old you were, and that's fine. Then you might see your wife and your kids and all of that in 1930. Then if I wanted to know what happened 10 years later to you, to your kids, I couldn't do anything about it. And so, I could create--
Russ Roberts: Why?
Ran Abramitzky: Because we don't have Social Security number before, like, in this period that we can link people across time. So, we had to think about, 'Oh, how can we create--despite the fact that we don't have a unique Social Security number--how can we create nevertheless a panel data that we can follow people over time?' And the answer was, 'Oh, here is another Russ Roberts. He was 36. Now, he's 46. He was born in England, still born in England. Oh, he's the only one. Probably that's the Russ Roberts,' and then I have you in 1930 and 1940.
Russ Roberts: What's that have to do with Ancestry.com? How does that come in?
Ran Abramitzky: Yeah. Leah, maybe you want to--
Leah Boustan: Well, if you go look up your own grandparents on Ancestry.com, you can actually do this process for a single family. And I've done that. You know, you can see the most--if you know that they came in 1901, you can say, 'Okay. Well, the first Census I can find them is 1910. Let's see how they're doing nine years after they came from Europe. Oh, they're still living in the Lower East Side in New York, and they're working in the garment industry.' Now, let's look at them in 1920. Maybe they moved out. Maybe they moved to Ohio. Cleveland was a really innovative and expanding city. Maybe they moved there; and now they're working in metal manufacturing, which was a more technical type of job. And so, they've moved geographically. They've also moved up economically from a sort of a basic garment worker that didn't require very much training to a metal machinist. And, now, where are their kids? By 1940, maybe a kid graduated from high school. Maybe they graduated from college.
So, by following your own family over multiple periods, you can really trace out the story of the mobility or lack of mobility.
In the case of my great grandfather, actually, we found him many times and he was always doing the same thing, which we know from the family stories: that he owned a little mom-and-pop shop, and they would just try to find whatever they could to buy cheaply and then sell it for a few extra pennies or nickles just to try to make a profit as a small shop owner. And he never moved to a different place. He never moved to a different occupation. But we can see his kids actually do so. His older kids--he had eight--had a year or two of high school and they were moving into very low-level, kind of clerk-type positions. One of them worked at the race track. One of them was a shoe sales lady. That sort of thing.
Then my seventh and eighth kid in the family--the seventh was my grandfather, and he became a doctor, and his younger brother became a lawyer. The older ones kind of helped the younger ones actually be able to afford college in the sense that they could take some time out of work and go to college.
And that is the story of one family.
But what if you could do that over thousands? Or what if you could do that over millions? That's where we really do need to follow people over time across different census periods.
Now, we can't look at real income growth because we only have the income data at the end of the story--in 1940. But, the occupational moves and the geographic moves do tell us a lot about how someone's income and resources were growing.
Ran Abramitzky: And I feel I should say something about Ancestry to give them the credit they deserve. Before Ancestry.com and FamilySearch and so on, you could be you could go and look up in a basement in microfilm your family; and good luck finding them in 100 million people, and find them again in the 1930 Census. What they did was to digitize those records so that you can type it in the comfort of your computer. You can type the name. And then multiple census records from different periods show up. But even that is an innovation of recent years. So, I guess in part, why hasn't this been done before? Maybe because Leah and I are super great; but in part, it's because 20 years ago, nobody could have done it because we needed Ancestry.com to digitize at least the name so that we can start on this process of looking people up.
Russ Roberts: Well, I think you're definitely super great. But the part that I'd asked at the very beginning is that you didn't always have what name to type in, the way Leah knows her great-great grandparents' name. So, what you're telling me, I think, is you took the 1940 Census and took those names into automated searches in Ancestry.com and then used some algorithm of location, and--made some assumptions.
But the key part of this--now we'll get to the bottom line--the key part of this is that pre-1940, you don't have income. So, when you're talking about mobility of past immigrants and their children, you have to impute income over part of that time. So, how'd you do that?
Leah Boustan: Well, it turns out that if you know someone's occupation, location, and age, you can do a pretty good job at guessing their income. So, you have to be really flexible about this. Like, a professor who is 30 is probably going to earn a lot less than a professor who is 40; but if you know that you're a professor in New Jersey who is 43, you can have a pretty good guess at that person's income. I'm just describing myself.
So, that's what we're using. We're using very flexible relationships between occupation, age, and state, so that you allow everything to interact with everything and build a big statistical model of how those three parts of your profile--your age, your location, and your occupation--relate to income.
We build that model in 1940 because that's when we know income as well. And then we take the predictions that come out of that model and we bring that back in time.
And so, sure, 1940 is not necessarily going to reflect the state of the economy in 1910, but if you use the 1940 predictions in 1940, they work incredibly well. So, essentially, those three elements of a person can get very close to telling us their true income. And then, we're running into trouble with trying to back-cast that back through history, but that's really just--we're looking under the lamppost of what's available.
Russ Roberts: Yeah. You're doing the best you can.
A nice exercise, of course, would be to pretend to do that in 2010, or--yeah, 2010 is good enough--and then back-cast for 1990, 1980, 1970, and 1960. You actually do know the actual incomes for those people. But pretend you don't and see how accurate they are. So, did you do anything like that, Ran?
Ran Abramitzky: It's a bit tough because in the IRS [Internal Revenue Service] data, for example, you don't have the occupations in the same way that allows you to do that. So, it's the information. It's a great idea to do something like that for the present, but I don't know that we can do it with the data that we are working with.
Russ Roberts: Okay. So, let's talk about assimilation. We opened with this a little bit, but I want to go a little bit deeper. The worry, I think, with a lot of people in modern time looking at immigrants is that they won't assimilate. They'll stick with themselves. They won't learn the language. And of course, we've made it in America easier to survive without English either as a measure of compassion or for whatever reason. In the old days, there wasn't anything like that. You had to get along. Again, I think of that as an immigrant in Israel. A lot of signs here are in English. Actually, most signs here in Israel are in English, Arabic, and Hebrew.
Unfortunately, signs are not that helpful in life. They're better than nothing. You do have some idea of what street you're on. That's good, but I think without Google Maps and Google Translate--in a recent episode, I've been somewhat critical with Google. I will give them some props, that I can't imagine what it was like to be an immigrant to a country with an alien language, English-to-Hebrew, without Google Translate. And today, I can take a photograph of a sign in Hebrew if it doesn't[?] in English, and Google Translate will translate it. Poorly, but good enough that I can find out whether I'm poisoning myself, say. So, very high level.
Anyway: the standard story is new immigrants, they don't assimilate, they don't learn English, and we've made it too easy for them. We have too many language things that they don't have to learn English. And yet, you find that assimilation hasn't changed very much. So, how would you--I invite listeners to think about how they might begin to think about assimilation--but, Leah, tell us how you actually try to measure it.
Leah Boustan: Well, I think that this is one of the myths that I really did believe before we went into the data. And, for the same reason that you describe, I thought, 'Well, there was a strong Americanization movement 100 years ago that there was an idea[?]: You send your kids to public school and the school is there to make them Americans and to strip them from their native language--you know, only speak English,' and that these days we have much more acceptance of multiculturalism.
And so, I thought: 'Well, naturally then, immigrants will kind of stick to their own culture longer now than they did then.'
So, we tried to test this; and we needed to come up with measures that you can see today and in the past.
I think today you would be able to come up with a whole variety of measures. What do people eat for dinner? Do they celebrate Thanksgiving? What kinds of clothes do they wear? What kind of music do they listen to?
But we needed to stick to something we could compare to past and present.
So, we look at learning English. We look at who immigrants marry. Do they marry only people from their home country or do they marry people either from other countries or from the United States? The neighborhoods that immigrants live in: Do they live surrounded by other people from the home country or do they live in more integrated environments?
And then, the measure that I think is in some ways the most illuminating, which is: the names that immigrants choose for their own children. And, we like this measure because if you have a couple of kids, you can actually get a snapshot on a person two or three times as they're spending more time in the country. Maybe the first kid you think, 'Maybe I'll go back to Europe. Maybe I'll go back to Mexico.' So, you choose a name that is a very ethnic that would work well back in the home country. And as you spend more time in the United States, you might choose a name that's more American-sounding.
And, on all of these metrics, we see that, first of all, immigrants are assimilating. They are shifting toward the more integrated situation, whether it's neighborhood, whether it's names.
But, secondly, I think most really surprising to me was that the pace of that assimilation is very similar today as it was in the past. And that was really surprising to me.
So, when we say in the book that we're kind of busting some immigration myths, it's not that we think, 'Well, everyone out there, they don't know the real story, and we're the experts kind of coming to teach you what's really going on.' I mean, really, in many cases, these were myths and ideas that we also held and that also we needed to hold up to scrutiny.
Russ Roberts: Ran, do you want to add anything to that?
Ran Abramitzky: Giving a child an American-sounding name, the way we think about it, is a sign of wanting to identify with the U.S. culture. So, it's also financially cost-free. So, it doesn't cost money to name your child.
But of course, we know that being financially cost-free doesn't mean that it's completely cost-free, in the sense that if I called my son 'Russ,' for example, my mom would be on the phone and she would be like, 'Where is Russ coming from? How is this part of your Israeli original culture?'
And so, it is costly for me to not retain my own original identity. But this is exactly what we are feeling, we are capturing, by this name measure, which is: immigrants trading off between retaining their own original cultural identity for the potential benefits of assimilation.
And the fact that we find that there is a remarkable similarity, for example, that immigrants tend to give their children more American-sounding names when they spend more years in the United States, and to a remarkable similar degree in the past and present, was an interesting finding.
And especially the groups that received the most criticism--for example, Italian and the Portuguese and Russians in the past or Mexicans today--are among the groups that name-assimilate the fastest, was another interesting findings that we uncover with the name measure.
Russ Roberts: Of course, some of that can be a simultaneous system, right? It's not just that you feel more American so you give your third or fourth child, if you have one, an American-sounding name. It's that you realize you've kind of handicapped maybe your first child with an ethnic-sounding name, and you realize, for whatever reason, you're going to make it easier for them. And I think you say that, I think, in the book.
Ran Abramitzky: Right. Having, for example, a more foreign-sounding name was associated with less schooling, lower income, higher unemployment, and so on. So, if immigrants look and think that giving their kid an Italian name is something that might generate discrimination at school or at work, they might realize that they want to shift to a more American name; or it could be that they choose to stay in the United States rather than return to their home country. It could be various different things, but all of them indicate that you are willing to reduce your identity marker and get closer to what the more dominant culture is.
Russ Roberts: Of course, if your name is Giuseppe Verde and you come to the United States--if you came to the United States and your parents called you Giuseppe Verde, you might after a while change your name to Joe Green yourself. I assume that ruins your data to some extent, right? People who either legally or through practice--and that's very common, obviously, that people who are given an ethnic name stop using it. I assume that that's a common thing. Leah, do you want to comment on that?
Leah Boustan: Well, a lot of those personal name changes do happen very quickly after arrival. So, the example I was giving of a family that arrives in 1901, the first Census you pick them up is 1910. By 1910, if those name changes have happened, we can continue to follow the family. So, we're not trying to match people back to the shipping records or back to what we--in one particular paper, we tried to match people back to the home country, to Norway, but that's not typical.
So, a lot of those name changes do happen quickly. But you're right that that also--if you do decide to change your name later in life, that is a reason why we might not be able to follow you. That would be identical to us as if you died because you yourself with the old name just cease to exist.
So, we have been continuing to refine our algorithms. Just over the past spring, we've been working on ideas like name umbrellas. So, you said Giuseppe to Joe. You didn't say Giuseppe to Russ. So, are there umbrellas where we can think about common name switching and try to put someone under the umbrella and continue to find them? So, as social scientists, we're continuing to refine our algorithms and we get all sorts of questions and criticisms, and then we try to incorporate them.
Russ Roberts: Cool.
Russ Roberts: Let's close and talk about how this work--which is quite, quite ambitious and has a scope that's really unparalleled in terms of size and scope--how it changed you through the course of it. Now, both of you have alluded to it to some extent. We are all people of a place, right? We were born in a place. I was born in Memphis, Tennessee, and at the age of one left; but there's a piece of me that's a Memphian because my parents grew up there, because my brother and sister live there, and we go back there. Now, I've moved to a totally new country and all of a sudden, I'm a Yerushalmi--I'm a person of Jerusalem. I have the new immigrant's sensibility to some extent, which is a fascinating phenomenon. In doing this kind of work, I assume it makes you think about these things.
We did an episode with Megan McArdle on Roger Scruton's book, Where We Are, which is a really wonderful look at how place and home are associated for some people. Not everyone. Not everyone cares deeply about it, but many people do. I mean, we talked earlier about the person in Alabama who wants to stay there even though there's less economic opportunity. Certainly we have an episode with Chris Arnade, which I thought very powerful set of considerations for people who give up economic mobility to be home.
And, I'm just curious. You're Americans. You're the children of immigrants. Almost all Americans are, to some extent, the children of immigrants; and Ran's an actual immigrant. Talk about how this--Leah, go first and talk about how this work makes you think about yourself differently, if at all.
Leah Boustan: Well, I learned a lot about my family in the process of putting the data together. This is the first time that I myself have been in a dataset, and also my family members have been in a dataset. So, in order to look at the names that immigrant moms give for their kids, we were using data from the California birth certificates. While I'm not an immigrant, I was in the control group as a mom who is U.S.-born, and two of my three kids were born in California. So, I'm in the California birth certificates data.
Then my family is in our census data. Of course, I looked under the microscope at that observation far more than I did at others, and I learned that some of the myths that my family tells about itself are true, and some of them are not true. So, it's kind of like a microcosm of what we saw in the case of America.
So, one that just really blew my mind is that my grandmother--who is still alive; she's 103--she is the daughter of immigrants, and she was given the name that at the time was very ethnic, which was Rose. That was very associated with being Jewish and not very U.S.-born. So, she said: I went to second grade and my second grade teacher said, 'You're never going to make it with the name Rose. You should be Rosalyn. Rosalyn is a name of a movie star. You'd be on the silver screen.'
So, she came home and she told her parents, 'I'm Rosalyn now,' and her parents said, 'Okay.' Which--I have a third grader and a first grader. I was like, 'There's no way that they're going to come home and tell me that their name is something other than it is. That couldn't possibly be true. But in the census data, I found her when she was two, and I found her when she was 12. Indeed, she was Rose when she was two. By the time she was 12, she was Rosalyn.
Russ Roberts: Whoa.
Leah Boustan: So, she herself as this young, strong-willed girl changed her own name in this process of Americanizing herself and her family. So, I was really able to learn more about my own family and actually test my own family's stories against the data, which is very on-brand for me with the skepticism and the empiricism. So, that fit really well.
Russ Roberts: That's very nice. Ran?
Ran Abramitzky: We shouldn't look at the past with nostalgia and that we should take the long-term view. So, people, when they look at the voters and politicians, when they look at immigration, they tend to look at the short-term views--how immigrants do when they first arrive to the United States. And that view tends to undermine immigrants' overall success.
If there is anything our book taught us, taught me, is that you should take the long view--that you should think about immigration in terms of generations, not in terms of years. When you do that, then immigrants are doing remarkably well. If you look at the children of immigrants, they're doing remarkably well. If you look at past groups, like the Norwegians that when they first came they were fishermen and they were not doing as well; but when you look at them with a perspective of 100 years later, they look terrific.
So, somebody in 100 years will be sitting and will say, 'Oh, of course, those great Mexican immigrants, they're doing great, but what about--' you know; and then maybe your own new group that you don't like. So, it fits well with us being economic historians that we should take the long view on immigration, and when you do the American dream, today is just as real as it was 100 years ago for the Ellis Island generation.
Russ Roberts: My guests today have been Ran Abramitzky and Leah Boustan. Their book is Streets of Gold. Ran, Leah, thanks for being part of EconTalk.
Ran Abramitzky: It was really great to be here. Thank you, Russ.
Leah Boustan: Thank you so much.