 Welcome everyone, thanks for coming. First of all, I apologize because the chief economist is not available today, but I'm going to replace him. I'm going to introduce, it's a pleasure for me to introduce Professor Megan McGarvey. She's a researcher and professor at Boston University School of Management and also fellow National Bureau of Economic Research. Publish a lot on leading economic and management journals and topics related to the IIP, but also trade, FDI and so on and so forth. And also high-skilled migration, which is the topic of today's presentation. I have a particular interest in this, in this, in her presentation, and my, the economics and states division has a particular interest in her presentation, because as you might know, we have been working in the last two years on a project of the International Mobility of Inventors, and we have produced a lot of reports and working papers that are available in our website. But we also are interested in serious research, and that's why we brought Megan here, who's gonna, who's gonna talk about high-skilled migration. I think it's better way if she explains a little bit the content of her research. So Megan, the floor is yours. It's a great pleasure to be here. Thank you so much for inviting me. It's a really wonderful opportunity for me to to present my work and to meet with some of you and to hear about your interests. So it's a really special opportunity. The topic of my talk today is Do Return Requirements Increase International Knowledge Diffusion? And this paper is co-authored with my colleague Shulamit Khan at Boston University. The background for this paper really comes from the fact that in the United States and in other countries as well, many, and in some fields most, of the people who receive doctoral degrees in Sciences and Engineering come from outside of the United States to study. So these two charts come from another paper by Bound, Turner, and Walsh, and they show that in the life sciences, a large share of doctoral recipients are non-U.S. students, so they come from other countries, and in engineering actually a substantial majority of those students come from other countries. The primary sending countries being China and India, but you know all the countries of the world are represented in American University's doctoral programs in Science and Engineering. Now the chart on the right comes from a publication by Finn, and he has documented the stay rates of these foreign students in the United States. So after they complete their PhDs, most of these students actually remain in the United States. So the most recent estimates are somewhere around 70, 60, 60, high 60 percent, close to 70 percent, which I think really raises the question for the home countries of, you know, what are the effects of having such a large percentage of their citizens educated in the United States and remain in the United States? So that's a question from the sending countries perspective. I think it also raises a question for the United States, because if you notice this chart shows that the stay rates of foreign students in the United States have really plateaued in recent years, and there may even be a bit of a decline in the stay rates of those students. And so I think the United States really needs to consider what might happen if we started to see substantial declines in the stay rates of foreign students who are educated in the United States. So as I alluded to, a moment ago, in many sending countries have concern about the possibility of emigration of the most talented, which was sometimes known as a brain drain. And, you know, a number of countries have policy initiatives to try to bring back people to those countries who have left and studied abroad. The most common policy lever for bringing people back is to make their graduate fellowship funding conditional on the eventual return to the home country. And we're really going to focus on policy like that in this paper today. So the specific question I'm going to try to answer today is what is the impact of return requirements, so a policy that requires a student to return to his or her home country after finishing their dissertation? What is the impact for science and engineering PhDs on home countries? So you might think that there are positive impacts for the home countries in the sense that, you know, we're bringing these students have been educated abroad and they're bringing sort of truth and light to the home country. But it's possible that there were also negative implications if, by requiring people to return to their home country, they are being forced to leave the place in which they could have been most productive in their careers. On the other hand, from the US perspective, if the US encourages or allows policies that require students to return to their home countries, it's possible that we could be losing out on the contributions of some of the most productive members of the scientific community. So, you know, many of the most productive members of the scientific community, whether measured by Nobel laureates or other measures of the productivity of scientists, many of them have come from abroad to the United States. And I full disclosure, I did not put these two pictures in the presentation because I was going to Switzerland. They just happened to have a Swiss connection that's CERN on the right-hand side. So on the one hand, we may be losing the contributions of some of the most productive scientists. On the other hand, when people leave the United States and go abroad, they may maintain their connections to scientists in the US and this may enable scientists in the United States to tap into knowledge production in foreign countries. So we think that there is a possibility that there could be both positive and negative effects for the United States when people go back to their home countries. And let me just say that, you know, I'm going to be talking a lot about the United States in this presentation because this is where our data come from, but I think that the results that I'm going to be describing today are applicable to any advanced economy. Okay, so there's a prior literature that suggests that these negative effects of brain drain could be important. I won't spend a lot of time talking about these, but just to summarize, a number of papers have suggested that that location is very important for knowledge diffusion in science and technology, okay, that when people are located in a particular place that other inventors or other scientists have privileged access to the innovation created by the innovator. On the other hand, there's a literature that suggests that there might be something called brain circulation rather than brain drain, which is that when people from, for example, India move to the United States and start high-tech startups, that they may then go back to their home countries and source inputs or do outsourcing in their home countries. They may share knowledge about technological innovation that they've acquired in the United States with their home countries and that this can have potential impact, about positive impacts for the home country. There's also some research which suggests that the ties, the social ties between inventors are persistent over time and that when an inventor moves to a new location that that inventor's colleagues in his former location are still aware of the research that the inventors doing in his new location, which might suggest that when people return to their home countries their former colleagues in the U.S. may still be aware of what they're doing, may still be in contact with them and learn about their research. So I think that the prior research has evidence both for and against the idea that location is going to be very important in sharing knowledge between the home country and the scientist. So specifically what this paper does is we compare foreigners, so people who come from outside the United States, who have PhDs in science, technology, engineering, and math from U.S. universities, who are Fulbright Fellows, and these Fulbright Fellows have the requirement that they must go back to their home country when they finish their PhDs if they ever want to come back to the United States and have a work visa. We compare these Fulbright to other students of foreign origin who do not have return requirements, and then we're going to look at a variety of measures related to their research publications in science and engineering journals, and we're going to look at how productive they are, how many papers they author. We're going to look at who they collaborate with, and we're going to look at the citations to their work. We're going to look at whether they're cited by people in the home country. We're going to look at whether they're cited by people in the U.S., and then we're also going to look at who they cite. Now the reason why we're interested in citations is because we think of citations, so when I write an academic paper I list all the other prior papers that have influenced my research of that I'm building upon in my research, and that list of prior kind of like prior art in a patent, or maybe not exactly like prior art in a patent, but you know the citations in an academic paper can be used as an indicator of the influence of one group of scientists on another group of scientists, or one individual scientist on another. So we're going to use those citations as a way of tracking what scientists in one location learn about scientists in another location. So just a little bit of background on the Fulbright program. So the Fulbright program is created in the post-war period, and it's brought a large number of students to study in U.S. programs. It's the major program that funds foreign students to study in the United States. And they receive what's called a J1 visa, and that visa has the condition that they must spend two years in their home countries before they can apply for a permanent or work visa in the United States. We've collected a data set of 249 foreign students who had Fulbright fellowships in the mid-1990s, and we match each Fulbright with another international student from the same program in the same university. Most, much of the time, but not in every case, because we couldn't always find a person with the same advisor in every case, but much of the time what we call the control student who's matched to the Fulbright student has the same dissertation advisor as the Fulbright, but they're always in the same scientific field as the Fulbright. They received their PhD generally in the same year, but sometimes within three years before or after the Fulbright, and then we follow their locations and their research publications and the citations to those publications through 2007. So we're going to track these people over time and look at where they got where they moved and what they wrote. Our data sources come from a variety of places. We started with the list of foreign Fulbright fellows. We then look them up in a database of dissertations and theses in the United States. We then use that database to identify the controls. We Google them. In other words, we look them up on the web. Generally, we found their CVs and use their CVs to track their locations over time. And then we collected publication data from ISI's Web of Science, which is the Thompson Reuters database, and we also got citations from Thompson Reuters. So this is just showing a the distribution of sources for our location data. We got most of our location information from CVs, but also from LinkedIn profiles and web bios. And then there were and some people we identified through the affiliation that they listed in their publications. There are a few other types of sources in there. The regions from which Fulbright's and Controls come are similar in some cases. So for example, for Europe, we have similar sort of balance in the region from which they come, but we have a large number of Latin American Fulbright's. And this just reflects the foreign policy objectives of the United States during the time at which these fellowships were granted. The years in which these people received their PhDs, the median year is 1998. Average is somewhere between 1997 and 98. And this is the distribution of year of PhD is similar for Controls and Fulbright's. And the fields in which they obtain their PhDs are engineering, computer science is the most common field followed by biological and life sciences and agricultural sciences are also a large share of the PhDs in our database. So it's certainly possible that Fulbright's could be different from the controls that we find. But the controls that we identify come from, as I said, the same program, often have the same dissertation. They're working in the same field and they graduated around the same time. So we feel that this is going to give us a fairly close similarity between Fulbright's and Controls in the quality of their training and in their general attitude for research. Although it's certainly possible that there are some differences left over between Fulbright's and Controls in their ability to do research, we think that we've accounted for most of those differences through this matching procedure and that the major difference between Fulbright's and Controls in this sample is that the Fulbright's are required to go back to their home countries while Controls are not. And if you have concerns about whether there is this quality bias in the match between Fulbright's and Controls, I think we can address those when we actually look at the results because I think we can think about them in the context of what we actually see in the data. Okay. So I'm going to talk about three different types of evidence of the impacts of return requirements. So first I'm going to tell you a little bit about what we know about how return requirements affect the productivity of these scientists. Then I'm going to talk about what, how they influence the collaborations between Fulbright's and other scientists. And then I'm going to talk about the citation data. So how does, how do return requirements correlate with the diffusion of knowledge both from the Fulbright's or Controls to other scientists and then from other scientists to the Fulbright's and Controls. So let's start with productivity. The productivity regressions that I'm going to tell you about are Poisson regressions with, with Control variables and the main variable that I'm going to focus on is going to be a dummy variable that's equal to one if the person is a Fulbright and zero if the person is a Control. So the coefficient on this Fulbright variable is going to give us information about the average difference in the productivity of the Fulbright's relative to the Control scientists. And that's going to tell us something about how return requirements, which are important for Fulbright's but non-existent for Controls, how those return requirements influence the productivity of scientists. But we're also going to control for or hold constant the field of study, the year of PhD, the rank of the PhD program that the scientists graduated from, their gender and some information on their, the GDP per capita of their home country and how productive they were as a graduate student, how many articles they wrote as a graduate student. So this table, I mean let's start by focusing on this column of this table. Okay, so this is giving me the percentage difference between the number of publications written by a Fulbright student who must go back to his or her home country and the number of publications written by a Control scientist who doesn't have to go back after holding constant those other variables that I mentioned in the last slide. And it's broken down by the percentiles of the GDP per capita of their home country. So a country at the 25th percentile of GDP per capita is a pretty low income country. A country at the 90th percentile of GDP per capita is very high income country. And what we see here is that for a country at the 25th percentile of GDP per capita, a scientist that is has, that's trained in the United States Okay, but is required to go back home by the conditions of their fellowship is 50% less productive in their subsequent research careers than a scientist who doesn't have a return requirement, but is otherwise similar. Okay, at the 50th percentile the scientists, the Fulbright is 34% less productive than a control who has again all the same training and all the same care you know most of the same characteristics or that we've held constant with those control variables um and uh However, when we look at the 75th percentile and the 90th percentile There these numbers are negative, but they're not statistically significant So that means we cannot reject the hypothesis We cannot statistically reject the hypothesis that there's no difference in the productivity of controls and Fulbright's So controls and Fulbright's look to be about as productive um Even though the Fulbright's are required to go back to their home countries and that's only true for the high income countries Okay, and this is true across several different measures of their publications So first authored publications may give us a better sense of first authored and last authored You know in some in some cases scientists have large collaborations in which there may be 10 people Who have all contributed to an experiment and they all get listed on the article and often The ones who were really most responsible for the article get listed either at the beginning of the author list or at the end of the author list And you know we try to control for that by distinguishing between first authored and last authored publications And we see the same effects. It's the effects are particularly large for last authored publications And that may be because in many fields the person who's the head of the laboratory gets listed as the last author Um, and if we look at publications in high impact journals only we drop all the low impact journals We still see similar results. Okay, so this I think This these results suggest that when people are required to go back to their home countries They produce fewer scientific articles than if they're allowed to remain in the united states But this is only true for people who are required to go back to low income countries Requiring someone to return to switzerland has zero impact on their subsequent productivity Maybe it has a positive effect now our results suggest if you're at the 95 percentile of gqp per capita, there's really zero impact Okay Now let's look at collaboration. So this is a measure of How scientists work with other scientists. So do fulbright's Work mainly with people in the home country. Do they continue to work with people in the united states? And the reason we think this interesting is because we believe that knowledge is shared When people collaborate with each other because they exchange ideas and they influence each other's work We're going to look at the number of articles that have at least one author in the home country And then we will try excluding the author's current location. So Because the author is himself may himself be in the home country We want to look only at collaborations that don't involve the institution of the author himself now. Unfortunately, we can't Identify each author's exact location in the article data that we have but we can rule out Collaborations that are with people at the same institution that the author is working at We'll do the same thing for collaborations with u.s. Authors And we'll look at collaborations between home country authors and u.s. Authors And what we see here this is controlling for their Their productivity in graduate school. So how productive they were as a as a phd student We see a doubling again. These are percentage effects. We see a doubling of the number of Collaborations with people in the home country and this is true even after excluding the author's own institution So it's not just explained by the fact of the author's collaborating with himself You know, we really do see that there are substantially more collaborations in the home country For fulbright's than there are for controls Then we see that There are fewer collaborations with People in the united states although this result is a little bit less robust It's not quite as statistically significant as the other result However, we see More collaborations between the home country and the u.s For fulbright's than we do for controls So these results suggest that the return requirements of the fulbright program while they do appear to reduce the productivity of scientists After we account for that productivity difference. We see much more collaboration Between the home country and the united states When the person is a fulbright and has a return requirement Okay, so this is the first indication that we have That home countries may be gaining access to knowledge produced in the united states through the return of Scientists facing return requirements Okay, that's just summarizing what I just said now. I want to talk about knowledge diffusion Okay, so collaboration as I said is giving us the first indication that there may be knowledge diffusion Created by this return requirement And now we're going to dig a little bit deeper into that using citation data We'll look at two types of Citation data. Okay. Now. I apologize if this is a little confusing But we're going to call one type of citation a forward citation and a forward citation is a citation To a particular scientist's work So if I write a paper and earnest cites my paper Um in his own paper saying that my paper has influenced his paper Then we earn it that we call that a forward citation forward citation is earnest citing me Okay, if alternatively julio writes a paper, uh, which maybe he's doing it right now Uh, and uh, and I cite his paper So my citation is to julio's paper, uh, will be uh considered a backward citation. So forward citations are Are citations that represent knowledge that i'm sending to earnest and backward citations represent Knowledge that julio is sending to me. Okay, so we're going to look at both types of citations here Um So the number of forward citations just to be specific is the number of citations to scientist eyes articles published in year t That are then cited by articles In year capital t And we'll look at citations that come from the home country and citations that come from the united states How do we identify the country or the country of the citing article? We'll use the reprint authors address. So when a scientist submits an article for Publication, there's a corresponding author. Okay, and we use the corresponding author as the one The as a way of identifying the location of the article And this is because we believe that the corresponding author is usually the person who is has the most ownership For the paper now, obviously it could be shared equally among all of the co-authors it could be that all of the co-authors had sort of an equal contribution and equal Authority in in the paper But because authors can come from multiple countries. We had to identify the country Uh of the article in some way and this is we believe a sort of reasonable approximation to the country of the article the citing article We'll restrict our sample to the set of Journals that are listed in the web of science although We didn't find any evidence that foreign publication that people living in foreign countries are That their publications are undercounted when we examined the cvs of those Scientists and compared them with what's listed in the web of science We drop individual citations to their own papers And we again, we're going to use this Poisson model because A Poisson model is most appropriate for count data and we're going to be counting up the the citations made to these papers and we'll cluster the standard errors by scientists because That just Allow accounts for the facts that we have multiple observations on the same scientist over time And we don't want to overestimate the precision of our results by just because we have lots of observations on the same person The backwards citations are going to be very similar only it's going to be Citations to publications in the home country or publications in the u.s So the key variables that we're going that i'm going to talk about and then in the coming slides Include a dummy equal to one if the person is a fulbright And then that dummy Interacted with uh another dummy for whether the person is from a low-income country So that second interaction here Um So actually we'll have two of these we'll have one for fulbrights from low-income home countries And we'll have another variable for fulbrights from high-income home countries And those variables are going to give us Uh an indication of the difference the average difference in citations holding constant other factors between the Fulbright from a high-income country and a control from a high-income country And then the fulbright times low-income home country variable will give us an indication of the The average difference in citations between a fulbright from a low-income country and a control from a low-income home country And again, we think that the major difference between fulbrights and controls Is that the fulbrights have the return requirement and the controls do not Um We'll also look not just at income the gdp per capita of the country But we'll also look at what we call the science base And we'll do that either by looking at the number of articles per capita produced in the home country So the number of scientific articles in the scientist's particular field Uh written by authors in the home country Or we'll look at the number of citations per article that those Scientists receive so the first is a measure This is a measure of just sort of the amount of science produced in the home country per capita and this is uh, we take as a measure of the quality or the importance of that science because A country that receives a lot of citations in a particular field Is a country where a lot of people are paying a lot of other scientists are paying attention to that country's work And then we'll in some regressions I'll show you will also incorporate information on where the scientists are actually located to to try to really nail down The fact that this these results that we're seeing Really are due to the location of the scientist rather than some other aspect of the fulbright program Now it's important to account for the number of potential citations that could have been made and The reason for this is that countries vary in the number of articles that they produce And we want to make sure that we don't just Count over overestimate the number of citations the scientists receive just because they come from a country where a lot of work is going on So we'll control for the number of papers in the country in the field in your tea Because that those are the number of papers that That could cite the articles the scientists articles And we'll also control for the number of papers by the author in your tea because those will be Measure of the number of articles that could be cited Okay And we control for these two variables in logs in their regressions Again, we'll control for field of study The rank of the phd program gender and so on And these are just some summary statistics So one thing that's important to note When i'm for when i show you some of the results that come a little bit later Is that the there's a really small number Of citations coming from the home country on average. Okay. Most most articles are not cited in the home country Most articles are You know Close to almost all Many articles are cited in the united states And then also similarly scientists Oh wait, wait, i'm looking for this one scientists tend To cite their home country to some extent but not as much as they tend to cite the united states And again, this has to do with the scale of the number of potentially Citable papers in the home country relative to the u.s Just a few other things to note um Our scientists publish about 0.8 articles per year um and About exactly half the sample is a fulbright And it's sort of uh We have a little bit more low-income fulbrights than high-income fulbrights. Okay So this first table shows me That if I don't Hold constant The number of publications in the home country or the number of publications by the scientist On average, I see no difference between fulbrights and controls In the tendency of the home country to cite their work. So on average, we don't see Uh any effective return requirements when we don't control for those things however Once we account for the scale Of the home country the size of the home country in terms of the number of publications in the field We estimate a statistically significant and uh pretty large impact of the return requirements on citations from the home country And this is also true when we hold constant the number of publications by the scientists What's most interesting I think is that when we um distinguish between fulbrights from high-income countries And fulbrights from low-income countries That positive and significant effect which is associated with about 200 and more citations Okay, again, remember that we had the average citation rate is 0.05 So the actual increase in the number of citations is is a small increase, but it's a large percentage increase Okay, so this this uh, oh, okay this um this fulbright effect is um only positive and significant For the low-income countries, which means that it's only the people who are required to return to the low-income countries That see more citations from the home country And it's not we don't see an effect for people who are required to return to high-income countries Um this result is robust to looking at the distinguishing rather than by income But by the accord the science base of the home country in terms of the number of articles published Or the number of citations to those articles we see the same results That it's the low income and the low the countries that are have weaker science bases And not the countries that have the strong science bases that seem to really benefit from the return requirements These uh results these in this this increase in citation seems to go up over time And over time meaning number of years since the graduation of the person When we look at citations coming from the united states We see uh no effect on average We only see an effect for fulbrites from low-income countries and there's a negative effect Associated with 36 fewer citation 36 percent fewer citations So if people are required to return to a low-income home country scientists in the united states Make fewer citations to their research suggesting that scientists from the united states may lose May lose May lose access to what they're doing, however when we distinguish but according to the The science base of the home country we don't see statistically significant results So and and i'm not going to go into all of the results from the u.s But it turns out that once i um control for some additional variation in The fulbrites and controls, uh, we don't see a robust result for the united states so There i think the evidence on how the return requirements affect Knowledge sharing with the united states is more inconclusive than the results that we see for scientists going back to Low-income home countries and sharing knowledge with those home countries Um And this just talks about those robustness checks those those additional controls that we've looked into so we found that the The the benefits for the home country for the low-income home countries are robust to controlling for what sector the person works in Using more precisely defined scientific fields Controlling for whether they cite the home country themselves in their work We try dropping journals that are regional journals like the um Australian review of natural sciences So we drop that and we still see these results and we think that agricultural and environmental science is particularly focused on Certain regions of the world and so you might expect expect to see a bigger impact In those fields and we find that we still see the results when we drop agricultural and environmental sciences We also still see this again sharing of knowledge with the home country low-income home country when we Control for publications in high-impact journals when we control for The number of citations coming from other countries So in other words, we are looking at the share of citations that come from the home country And we also still see the result when we exclude the most highly cited articles in our data set when we drop The articles above the 90th percentile of citation Okay, so now this is looking at backwards citations and I realize we don't have a lot more time. So So i'll just try to summarize this With backwards citations. What's interesting is that we see Uh positive and significant effects of return requirements for both low-income and high-income countries Actually, this is a typo. This these two should be switched. This should be this high income This is low income. It doesn't really matter because they're essentially the same What we see here is that when fulbrites go back to their home country They seem to be redirecting their own research towards research that's going on in the home country They're citing more articles by other scientists from the home country And this is true as I said for both types of scientists, which I think is quite interesting And those also go up over time that the more time a person spends in the home country Or the longer it's been since they graduated from their phd program the more citations they make to the home country With when we look at citations to the u.s. On the other hand They continue to cite again. There are no there's these these numbers are not statistically significant That means that we cannot rule out the possibility that scientists who Go back to their home countries continue to have The same amount of access to research produced in the united states as scientists who do not have any requirement to return to their home countries So we see that people go back to their home countries They seem to be sharing knowledge with those with other scientists in the home country But they and they're paying more attention to what's going on in the home country But they still have The same amount of Access or attention paid to research by scientists in the united states And This just looks this just adds the location variables to the regression and here what we see is that Most of this fulbright effect seems to be mainly explained by location And that That the the results that we saw before are just kind of again reflected In the in the actual in the actual location of these people Rather than just by looking at their status as fulbright Okay, so the last thing i'm going to show you because i we're running out of time Is that this is a this is a regression in which we Regress a variable equal to one if the person is located in their home country On a dummy equal to one if they're a fulbright from a high-income country And another dummy equal to one if they're a fulbright from a low-income country And what we see here is that the fulbrites from the low-income countries have a bigger coefficient the four There's a 46 percentage point Increase in the probability of being located in your home country after you finish your phd If you are a fulbright from a low-income home country Whereas for fulbrites from high-income home countries, there's only a 27 percentage point increase And that's because scientists who come from high-income home countries Are more likely to go back home Independent of the return requirement But scientists from low-income home countries Are are less likely to return home than those other And in other words the return requirement has a bigger impact in the probability of returning home For the fulbrites and we think that this is a part of the reason for the results that we've estimated in the In the rest of the paper about the bigger impact for low-income countries because The return requirements change the behavior of people from low-income countries more Than they do the people from high-income countries So to summarize what we see in this in in these results is that for low-income countries or countries with weaker science bases Scientific articles by fulbrites are cited more frequently in the home country And this appears to be because they spend more time in the home country than control scientists But for high-income countries There's really not much difference Between fulbrites and controls in most across most of the measures that we've looked at And so this suggests that return requirements may actually not be necessary for the high-income countries and For all countries return requirements do seem to redirect the research agendas of the scientists towards the home country From the u.s. Perspective returning to a high-income country doesn't impact Diffusion to or from u.s. Scientists and so this this may be consistent with prior research Which has shown that the there are these persistent Relationships between inventors that through which knowledge continues to flow even after the inventor has moved to a new location And overall from kind of a global perspective Return requirements for countries from high for scientists from high-income countries are not associated with any decline in productivity Although return requirements for scientists from low-income countries are associated with a large decline in research output So this suggests that That if we're going to impose return requirements for scientists from low-income countries Perhaps there should be additional policies that are Directed towards supporting them and helping them maintain their Research productivity when after they return to the home country okay, so Thank you very much. And if you have questions or comments my email address is is listed there and I'd love to hear From you. Thank you