good evening well i should say perhaps a good late morning to our guest david card who is connected from berkeley um this is our third alan krueger lecture alan krueger has been at the seventh edition of the festival delivering wonderful lectures lecture that i think transferred to us two unforgettable characteristics of his way of conceiving the profession of economist one was the joy of research really the passion the enthusiasm that he devoted to addressing always very very relevant issues and the second is the fact that always policy must be based on empirical evidence so the idea that you should always carry out evidence-based economic policies so we always want to remember these two features of alan krueger contribution outstanding contribution to the profession in this lecture we had a lecture by hillary coins on safety nets on minimum guaranteed income schemes this was two years ago and last year we had another lecture by george hangrist on education and on measuring the returns to education so tonight we have another co-author of alan krieger who has been working not only on education but on other very relevant fields for policy making in particular there is a work of david and alan on the impact of a minimum wage rise in new jersey that has been really a pioneering study and one of the most widely quoted cited and discussed paper in economics it is a very interesting studies because it applies a new method establish causal effects of policies in particular of the rise of a minimum wage the idea there is that you not only should look at what happens in this case in new jersey in the labor market of new jersey when you increase the minimum wage but you should also take into account of all other relevant factors that may confound or alter the impact of a minimum wage and therefore you should take as a comparison in other state in this case pennsylvania where the minimum wage rise was not implemented but went through the same type of shocks of changes that could have been affecting the new jersey so the idea is that not only you should take the difference between the situation before and after the minimum wage hike but also the difference between new jersey and pennsylvania this double difference approach has been really a pioneering component of studies in labor economics and not only labor economics in development economics in many other fields and so not only the result of the studies were particularly interesting and widely debated because they found that the minimum wage hike in new jersey was not destroying jobs as many would have fought before and many and according to many prediction of economic theory uh but uh but also because of the new methodology that we study introduced so i think that david card is is a person who really is very important for us for this uh uh third alan krueger lecture alan david has been working on many other issues very relevant for economic policy making he has been working very much for instance on the issue of migration and there also his studies are extremely and widely quoted and has been very very influential on policy making without taking all the stance of what what is nice of his way of thinking about the role of an economist is one that provide information to policy makers and then the policy maker have to make a decision based on the trade-off that economic research suggests migration as i said is the issues that he has been addressing are always very important and at the same time very controversial migration was one income support schemes for you know people being in in poverty is is another one tonight is david is going to talk about affirmative action for racial and gender diversity affirmative action is a case where you tend to alter selection procedure for some position because you want to give more representation to some groups that seem to be underrepresented in this selection we have many examples also in italian case for instance gender quotas in board of corporations and the application of the method of analysis of david carr to two cases in the u.s will be for us extremely interesting so we are very glad that uh david you accepted our invitation and without further ado i don't want to take any more of your of your time the floor is yours for this third alan krieger lecture thank you so much um it's a great honor for me to be able to give this lecture i have actually given a lecture live in trento some years ago and i enjoyed it very much and i know as you mentioned that alan had given a talk there and he i have to say he loved he loved the trento set up he loved the idea of being able to communicate with a broad audience and get across some of his ideas and also to publicize economics and make people comfortable with some of the ideas that the newer ideas that allen and others were coming out with um it's also nice for me because i was alan's colleague for many years um for about 12 years the way he came as an assistant professor when i was already on the faculty at princeton and we worked on many different topics together and actually the last major project we worked on together was about affirmative action so this is i think a good good choice for for me i'm going to share my screen now and let's see if i can make this work so uh let me just go on quickly because i i want to try and cover a bunch of material and leave lots of time for questions um so as was mentioned affirmative action is a process that is introduced in cases where there's a selection process in the background that uh there are concerns about so examples would be uh hiring processes admission to selective university which is extremely important in controversial in the united states and honors an award so who gets to be for instance the national academy of science in the united states or other prestigious organizations similar in europe and lots of times when these processes people notice that the people being selected are much different than the underlying population and that's a concern and so it's widely proposed to have some kind of a a boost for the underrepresented groups and i'm going to use that term boost uh it's a kind of an unusual term even in english um so i apologize if it doesn't translate very well but it what by what i mean a boost is an increase coming about through an explicit change in the policy for one of the groups so examples of this is already mentioned the female shares on corporate boards there's a 2000 law in italy the gulf of moscow law california recently adopted a similar law actually admission policies at selective universities also hiring goals so for instance um my own university at berkeley in california we have kind of soft targets for the gender mix of um who we're going to try and hire as faculty with the goal of trying to increase some particular fraction of of underrepresented minorities so african american and hispanic usually and um and women especially in fields like economics where there's relatively few women um i'm going to show you some evidence through the talk that the use of these policies is on the rise an example here the national academy a very prestigious organization in the united states it's been around for more than 200 years it admits about a hundred people a year these days as a as an academy fellow in 2010 so only 10 years ago only 20 of the new fellows were were female this year they broke we almost broke the 50 rate for the first time so they got to 49 and so they've really been changing that policy and interestingly that change has occurred even under the trump administration so it hasn't been the case that um uh this particular type of policy has been was rolled back uh substantially uh in the last few years nevertheless these policies are extremely controversial uh and so that's one of the reasons why i thought it would be helpful to give a lecture and and was mentioned this was really uh characteristic of alan that um alan always felt that even in a very controversial area an economist could help by trying to set some of the facts straight and help people to make informed decisions about how a process works or how a new proposal works and so that's going to be my goal in this talk really in honor of alan's ideas so i'm going to talk about how affirmative action policies work um some of the consequences of those policies and i'm going to be focusing on two specific settings that i happen to know about one is a recent federal law case in the united states and this was a case where harvard university was litigated in other words there was a legal case against harvard for their admissions policy and some part of the paul of the litigation you may have heard about was uh to do with whether asian students were admitted to harvard at um proportional rates or high enough rates but the main part of the litigation was to an attempt to dismantle um the affirmative action policy at harvard um and so i'll talk about that and and i was the expert witness for harvard university in that case and so i was intimately involved in preparing all the exhibits and presenting the argument the statistical arguments for harvard at the court and the second context i'm going to be using is going to look at honors for academic achievement for people later in their careers and i'm going to be focusing on three particular types of honors the national academy of science honors i mentioned before another organization known as the american academy of arts and sciences and the third actually the most important for some purposes is the econometric society and uh those of you who are really economics nerds will know the econometric society is a pretty old long-established society it was established in the 1930s in europe by a bunch of european economists including norwegians and british people and so on and this since the 1930s has had honorific fellows and so we'll be looking at who gets into that and this is a joint work with uh stefano delavinya who's a native son from northern italy he's uh he was born in como nagori ira berry who's at the basque university and patricia funk who's in the italian university in switzerland um so why do organizations need affirmative action well um usually the the need or the concern about affirmative action or the argument for it comes about because of concern about the lack of diversity in the outcomes of the selection process and so in for instance in the harvard case they have a very explicit policy that says harvard university as a matter of policy needs to have a critical mass of african-american students so that other students at harvard can see the perspectives and understand some of the problems and understand some of the issues that really affect americans and can then be better equipped to deal with american life similarly the honorific societies have in the last few years especially started to realize that there would be benefits to the societies of having a broader perspective having more for instance women in the in these societies possibly also as some kind of a symbolic gesture to young women who are thinking about going into science or wondering whether they should go into science to make people feel that actually it's perfectly legitimate and and um natural for women to be scientists as well as men and so there's a lot of different values um but the main one is to try and i think to try and change the balance of who's getting in so i'm going to talk about somewhat of a kind of a almost mathematical but very simple mathematical model of how affirmative action works and it turns out that's actually very helpful uh this isn't going to be a you know complicated lecture in calculus or anything like that but it's going i think i'm going to try and convince you that there is some value in understanding how affirmative action actually works the simplest model of a candidate selection process for instance going to college or who gets to be going into a college or who gets to be awarded an honor would be to rank all this candidates by some qualifications and then choose those with qualifications above some cutoff so if a qualifications are cue that's the and those are variables that are distributed across the population in some way and then everybody that's above a certain rank would get in and sometimes there's an explicit rule like there's only going to be a 100 people admitted so the top 100 people you would find the cutoff that define the top 100 and that's how you go now that what that does is we have a technical problem i hope it's going to be solved as we can't hear david let's see if our technicians okay have i come back yes i think so i don't know because this is the first time actually in this festival that we have a problem of this type okay i'm going to turn off my video so it uses a little bit as bandwidth no i think there is no problem i mean we can see you and everything was working well we just had a small problem but now please you can share the screen and and we can keep on okay so i'll let me go back just 10 or 15 seconds and restart i apologize for that um so i'm going to be thinking about how criterion works and i'm thinking of a of one possible model of the selection process where there's a cutoff for criterion and in the case where there's a very sharp criterion that everyone above that criteria gets in and everybody below it does not get in we have a uh a rule that looks like this graphically on the x-axis is qualifications so candidates are qualified in different degrees so people may have for instance in the harvard case they would have different levels of academic achievement and different other characteristics like whether they're good in sports and so on and they would be ranked and there would be a cut off here that would determine who gets in and those above that cut-off get in and on the y-axis is the probability of selection so that jumps from zero to one as we pass through the cutoff so that's a very sharp straightforward rule now in reality that's not how it really works and the reason is twofold one is that most of the time we don't fully observe all the qualifications uh and so inevitably there's some noise in the process so somebody's looking at two candidates to be selected for a process or to be hired oftentimes it's kind of hard to see the difference between them and at the end of the day there may be something that happens and there's quite a few studies and economics of the variation that caused by things like what time of day it is or what you had for lunch or whether your sports team won the football match last night that determined to some extent the gatekeepers decisions so i'm going to be thinking about a slightly more complicated model where there's a combination of qualifications and noise or discretion in some cases so sometimes you could have even a biased judge or a biased hiring manager that could potentially tip the balance one way or the other in that case there's going to be an s curve and i'm going to spend a little bit of time talking about this s curve in the probability so let me just show what that looks like so instead of the blue line here which is the sharp curve when there's a little bit of noise you get an s curve and i've shown two s curves here one with a small amount of noise which is almost a step curve and one with a larger amount of noise which looks more like an s and you can see what happens with an s curve on the next diagram so here i'm showing what you get if there is some noise in the system is you get three types of candidates you get people to the left who have no chance of selection so they basically are so far below any kind of cutoff that they're never going to be selected and that's pretty important for instance in the harvard case uh more than 60 or 70 percent of candidates who apply for harvard really in all honesty have no chance of getting in there's a few candidates to the right the sure things i'm going to call them those are the candidates who would get in almost for sure so they're so outstanding they're so stellar now that's the size of that group can actually be surprisingly small uh depending on how uh elite is the selection process you're thinking about so if you're thinking about the national academy of science where only 100 scientists in the whole united states get selected each year there are virtually no sure things so you could you could have a the curve you would never see anybody who with with a probability uh 100 percent would be selected comes from college admissions in the united states and i have to say i did not grow up in the united states i grew up in canada where things work differently but every student in the united states knows about this concept of being kind of in between the group that has no chance of getting in and the group that has a 100 chance of getting in and the the americans refer to that as the bubble i don't know why the term bubble is used but the people on the bubble are the people that have some chance of getting in and sometimes people think specifically about someone whose probability of getting in is 50 50. so they're kind of halfway along the s curve where there there's going to be just a very arbitrary selection so if that's the kind of process for selection you can still have different cutoffs but what it's going to do is it's going to change the s-curve for this groups and so i'm going to think about a situation where we have a group a of people that are the ordinary group so that could be for instance the men b who are the group that we're trying to improve their chances of getting in and so we're going to give them a different selection criteria and so what you're going to get is two s curves one for group one that's the green group for instance that could be the men being selected for national academy of science or the white students being selected for admission to harvard and the blue line that's the s curve for the other group now again notice that there are a very interesting pattern here because to the left there's still a large group of people who aren't going to get in not going to get selected but it's a little bit smaller for the privileged group the b group the surescene group is also present it's a little bit bigger for the group that's shifted to the right and then there's a again a complete set of bubble students and what happens or or candidates and what happens is that those that group of people on the bubble is shifted a little bit to the right now if you think about how does the affirmative action policy benefit and who does it benefit you can see that by thinking about a candidate with a given level of qualifications so that's some point on the x-axis so for instance think of the point here at the very beginning of the s-curve those candidates have very very low probability almost zero if they're in the ordinary group for instance the white students group they have a little bit higher probability if they're in the advantage groups so the minority group for instance in the harvard case but notice that the gain in their probability isn't very big similarly over here at the far right or the right you can see at the end of the s curve the same pattern there's a little bit higher chance but it can't be raised very much because actually you can only go up to 100 so if you're pretty highly qualified very close to the end of the s you can only go up to 100 so the gain is small the group in the middle of the bubble that's the group that has the big gain and you can see for people right in the middle of the bubble that there could be quite a significant benefit and this is kind of the paradox of affirmative action policies many many people are not benefiting at all so this broad group of people that are no chance is still present they don't get any of benefit from the definitive action policy similarly the most highly qualified group gets no benefit it's only the people in the bubble range and in the bubble range there's quite a heterogeneity of treatment there could be a benefit that's big or small okay now that's a model is it really true is that how it really works well i had a lot of experience with this because in the harvard case we had to analyze how affirmative action worked in the admissions process and what we did was we developed a model for admissions that included all the characteristics of the students that the admissions committee would see so sat scores that's the score that everyone knows about if they're applying to american school athletic achievements uh other kinds of volunteer work you had done and so on and we were able to estimate models with and without affirmative action and see what happened and what we found is summarized here so this is two of the advantage groups on the left is african-american students and the right is hispanic students and we rank them by their strength of their portfolio so that's exactly the same as the qualification type of graph i was showing before and you can see for the bottom five deciles or even six deciles there's virtually no benefit for the affirmative action program at harvard for african-americans in those deciles exactly as the graph we had before you can see there's a pretty strong benefit for as we go a little bit further and these are the people kind of in the bubble range and at the tenth that's how there's a little bit smaller effect than for the ninth which is consistent with them them being closer to the top of the s so there's a little bit less benefit for hispanic students there's a for hispanic students there's a similar pattern not quite as start uh stark and you can see that um there's a kind of an increase but all there's a benefit for affirmative action for hispanic students but only for the top 20 or so students 25 of students at most okay so how much affirmative action do you need well it depends on what criteria you have but if your criteria is related to the fraction of people in the selected pool that we'll actually get in then what you need to do is you need to consider two things first how big is the underlying share of group b this is the group we're worried about i'm going to call that s and secondly what are the probabilities of selection for group a and group b and those i'm going to think of as two numbers pa and pb and it's a very simple exercise in math to show that the fraction of people who get selected that are in group b is affected by s and pb so what you need to have a high fraction of say african american students at harvard is you need to have a relatively high s and you need to have a relatively high pb and of course for something like african americans that's only about 10 of all students so s is only 0.1 so you know that you can only have if if the pa and pb are equal you could only have 10 of students at harvard would be african-american even in the absence of any difference between white and black students and of course if the underlying qualifications of african-american students are lower then you're going to have an even lower share and you can think about that mathematically in terms of two parameters or two constants that you could play with one would be what's the relative rate of preference or probability in the absence of affirmative action so i'm going to call that p0b that's the probability that say black students would get into harvard in the absence of affirmative action relative to the same probability for white students pa and so delta is the ratio of probabilities in the absence of affirmative action and you might think that's less than one might be like 80 percent or 20 25 percent or some number like that now what affirmative action does is boosts the probability from that uh p0b and so the boost factor i'm going to call beta and as a result of the combination of some differences in qualification of the two groups and the affirmative action policy you really have two parameters one is where would you be without affirmative action that's the delta and then how much do you change that that's the beta and you can effectively increase the beta by lowering the qualification bar for the group of interest the group b and the optimal degree of that depends on how much you want to push the selection process to achieve a certain fraction in the ultimate pool now i have the next question i want to address is i want to ask affirmative action is very controversial and a lot of people have thought about an alternative to affirmative action they say well why can't we do something that's a little that seems a little fair for instance in the admissions to university case what if instead of benefiting people who were african-american we tried to give a preference for low-income people we know that african-americans on average have lower income than white americans and so if we try and uh privilege lower income that's that's kind of a race neutral policy that might be politically more saleable there's a very important mathematical result that's been proved by a number of different economists and that is that any kind of indirect affirmative action of this type where i i don't focus on race per se but i focus on something correlated will necessarily lead to lower qualifications in the selected group holding constant the fraction of people that you get of the target group in that pool so if you want to get 15 african americans at harvard the most efficient way to do that is to target directly uh african americans by giving them a lower boost and again we had a we had a chance to empirically test this in the harvard case because um we could take the admissions process that they use at harvard and we could evaluate what would happen if instead of using um affirmative action directly you gave a boost or a change in the award that the admission system gives for lower income families and so what we did was we experimented with a range of possible positive benefits that the admission system could give to lower income families and we showed that um you know if you do that now this is a somewhat complicated graph this is directly from the trial so it's a complicated uh graph but very colorful though the left-hand panel the left-hand bar shows the actual composition of the class at harvard in terms of the racial composition so 40 percent white 24 asian 14 hispanic 14 african-american and 8 other the next panel shows what would happen if we took away all the racial preferences and you could see that the fraction of whites would go up the fraction of asians would go up the fraction of black and hispanic would fall quite substantially so instead of representing 28 those two groups together they would only represent around 15 or about half as many as they currently have and then these additional bars show what would happen if you implied bigger and bigger boosts to this target income metric and said okay i want to give a higher and higher and higher boost to income you can see as we go raising them the multiple of the boost you can see that what happens is we gradually get more minority students but most of that effect actually goes to hispanic students they're much lower income than the african american students who apply to harvard and so you get some increase in the boost if you want to get to something like 28 of the student pool at harvard being either african-american or hispanic you have to go to kind of the middle here the one that's highlighted and that's about a in the range of these boosts that we considered that was about a four times boost now what did that do we evaluated that by looking at the four criteria that harvard used and they rank students in four dimensions on a one to four scale one is exceptional really basically and no one we've ever met is a one a two is really very very strong and so they really want we're concerned to have one or two so at least two rating on these four dimensions academics extracurricular activities so that's things like clubs you belong to things you do to help your local hospital personal rating that would be things like are you helping out with your family are you somebody that everybody looks to in a leadership role and athletics of course that's very important in american universities because they want to have a strong intramural sports and you can see that across all four dimensions if you were to try and replace direct affirmative action with a boost to ses or boost to socioeconomic status you would actually lose on all the dimensions the biggest loss is actually in athletics which harvard puts a very high weight on okay so the last thing i want to do is i want to take a very brief but somewhat deeper look at the a selection of honors for economists and so the reason to do this is because we can find very long historical data so what we did was we constructed lists of all the economists who had published papers in any journal any of about 40 journals over the last century and we constructed for each economist in the evolving curriculum vita a cv that showed how many papers they had published at that point how many citations they had accumulated at that point and were able to do this all the way back to the 1920s and we were able to assign genders to everyone based on their names so um except in italy and andrea is female so we we also had a luckily we had stefano on the team so we could straighten out the andreas so we can assign names and we can then develop models of who gets selected and we do this for the econometric society which started selecting fellows in the 30s the american academy of arts and sciences which started selecting fellows in the 40s and the national academy of science which only decided that economists were scientists in 1968 so they've been selecting lots of other scientists but not economists and uh the interesting thing about all three of these societies is there's a nomination process which involves some individual input from members as well as a committee and then there's a vote by existing fellows so the preferences that you're seeing here are not the preferences of some committee or some hiring manager these are the preferences of the members of these committees the people who themselves have become members of these organizations people themselves are fellows so think of it as the the most distinguished people in the field as selected in the past are selecting new people and what we see is very interesting and here i'll start by showing the data for the econometric society and we're looking at the impact on the fraction of women who become new fellows and the red line is kind of a very important reference that's the fraction of all the existing economists out there in the world as far as we can tell so that's around 25 000 people who have published a paper and are still active in their careers and in the 1970s only five percent of those were female and as you can see from the red line that has been steadily rising and is now around 20 so we now have around 20 percent of the active publishing economists are female the green line is a measure of the fraction of highly qualified economists who are female so there what we did was we said let's take all the economists in in our set that we've been following and let's get all of those that have published at least three papers in one of the five top journals in the field and so the economists who are listening will know what what journals i mean but in the interest of time i won't go through that that list but there are five very prestigious journals the same five for the last hundred years um and you can see that in that fraction the fraction of women was also but even much lower was about half as big but two and a half percent back in the 70s and it's still about half as big uh today so it's around um just under 10 percent today out of 20. so the women are rising in the field they're also rising as a share of those who are highly highly successful with at least these three pubs but it's still quite a low rate and finally the purple line that's the fraction of women who are selected as new fellows and you can see that that was following the trend for the the green line very similarly up to around 2008 or 2009 and then it starts to take off and so we go from a situation where uh only about uh five or six percent of the new fellows are female to a situation where it's getting close to 20 at this point so it's about the fraction of all economists who are female much higher than the fraction who are highly qualified and that's evidence of this positive affirmative action now interestingly we if we fit more complicated models very complicated models in fact we see sort of three eras we see prior to 1980 that um women were probably just very strongly discriminated against so there were many examples of very famous uh prominent female scholars who never were made fellows of the economic society in the 1960s and 70s then in the 80 to 2000 period 2010 period we see kind of approximate equality and then from 2010 onward we see a kind of a change and interestingly we see the same kind of pattern in the other two societies so this is the fraction of women who are new fellows in the american academy of arts and sciences the econometrics society and the national academy sciences and you can see that all three of those were very dismal back until around maybe around 2000 and then they gradually start to increase nowadays the fraction of uh new fellows for the american academy of arts and sciences is over 40 percent so they're doing a pretty strong affirmative action policy to try and bring in eligible women but the and the economic society and the national academy of sciences are lower but still substantially above where they were so now my final graph is going to show you what's been going on in other fields economics isn't the only field that this has happened in so this is a total of five fields psychology that's the green line social science which is sociology that's the purple line physics and math which are often compared to economics kind of a math heavy uh side of the fields and you can see that um in in all cases there's been some very significant change this the change occurred about a decade or 15 years earlier in psychology psychology went from almost no female candidates to it's now been kind of hovering around 50 percent for quite some time sociology which some of many of you probably know a little bit about that field is fairly high fraction women uh in the base of people who are scholars in the field and now is up to around 30 percent of all fellows but even into the 1990s it was pretty low and even math and physics are now making some strides so let me stop with a slight of conclusions affirmative action policies appear to be on the rise we've just seen some evidence of that for instance the these honorific societies have really dialed up preferences for women direct affirmative action is more efficient than policies that favor correlated attributes that was a kind of a mathematical theorem that the harvard case proves very clearly so i think we have very strong evidence that that's true affirmative action policies mainly benefit highly qualified people in the target group so the vast majority of people in many cases especially cases where the selection probability is relatively low like harvard for instance only five or six percent of all people get in or to become an honorary fellow of the national academy the probabilities are well under a tenth of a percent per year so most people don't benefit the people who do benefit are at the very top of the field and so what you get kind of paradoxically is that you get a widening of inequality within the target group so now there's a very small set of women or a very small set of african-american students who get the big prize but the vast majority get nothing the size of the aaa preference that an organization will need depends on the gap between the desired outcome for instance um do you want to have 50 percent of all women in the fellow the new fellows rate and the share of the group in the highly qualified pool so if the if the if the target group the group b that i've been talking about is a relatively small fraction of the target pool uh then you're going to have to have fairly significant affirmative action to get a desired outcome like 50 50 in the new uh selected pool and in the harvard admissions case and then on honors for economists in the recent decades that has implied very significant boosts on the order of doubling or even tripling the probability that you get in uh because of the your target status so that's i'm going to stop with those conclusions and and i'll be glad to take some questions or comments from the audience so thank you so much for this uh wonderful lecture uh you uh told us at the outset that you had been before at the festival of economics in presence and that's one of the reasons why i didn't take some of the time available to us to remind all of us all the honors you achieved for your achievement in in research and you know while being a member of the surefing group um so it was not needed because you had been before here um your results are extremely interesting and we hope that you will be back to trento in presence when we will all be not only the people the nice thing is that we are back to having people in the room uh unlike last year and there are many young people here so that's wonderful to see and we hope to have you with us back here soon what you said is very relevant actually also for decisions that have to be made in italy very soon because the national plans for reform that is being presented by the italian government for the recovery fund involve the fact that they want to get more hirings of young and women in the public sector to carry out these plans and given that rules for hirings in the public sector do not allow to have quotas or preferential rules for men for women or for young people based on age or gender they plan to use some indirect rule like you know education or other things and what you indicate is that this indirect method may not be that efficient so what you said is extremely relevant also from this standpoint and something very uh you know important nowadays in our countries but i'm sure that there are questions from the floor and so we can take a few so if someone i please ask you to come here keep the mask and talk from this speak from this microphone over here also you can also ask questions in italian because there is simultaneous translation so okay no no keep the mask on please okay thank you for the for the lecture first of all and also i just wanted to ask you whether you were aware of any evidence of persistent effects of these policies in the longer term um because muscle many times these policies are justified as uh having longer-term effects on on the shares uh because of maybe role models the increase of number enrolled role models and things like that i was wondering even though i know these policies are very recent i i was wondering whether you you were aware of anything uh any work on longer term effects thank you well again yeah thank you thank you that that's an extremely good question um i think in fact in the case of affirmative action for university admissions um the supreme court of the united states had at one time said that they really wanted the affirmative action policies to be uh short duration you know to not go on forever um so because as you as you pointed out um these policies are um fairly recent for instance you can see um in the national academy case or the economic society case we're talking about the last 10 years we have to kind of rely on slightly indirect evidence and and i'm sure many of the people in the room are aware of studies of peer group effects that are very widespread in economics and one class of those peer effects are studies of things like if you have a female teacher or you have a female professor or a female mentor are you more likely to succeed or are you more likely to pursue a certain class of policies if you're a female student and there's a related work on african-americans if you have a if you're an african-american student in a university and you have a professor your math professor is african-american does that influence you so there's a fair amount of work like that and i would say although the results are somewhat mixed i think they tend to point toward a positive role model effect so you could expect some changes in outcomes i think people are most interested in the us context in um what they call stem so science technology engineering and math and so that's actually economics is a stem subject in the united states so that's like physics math engineering all those kind of fields the fields that the national academy of science actually represents and in that's historically very underrepresented for women and has we've been making some very significant progress in many of the stem fields but not all actually economics has been really pretty far behind so i think that there's some evidence of role model effects it's i'm not aware of any really great evidence um of really quick and large effects i think the effects are often modest and maybe somewhat slow to build up so i think we shouldn't we should be hoping that there are such effects but we have to be a little bit cautious i'm over here um thank you for your lecture i'm sorry i just have to look at the picture while i ask the question uh in the in the uh female shares of new fellows graph for the um society of the academic society uh you show with the the green line that um just a certain part of the newly selected women is uh is highly qualified in some way so um you show that we have about 20 of the members which are women and just about half is uh some woman with at least three top five publications uh so some some could argue that uh to to make room for the women we decided to take some women which are not actually high qualified but you didn't show the the same uh the same graph for men so i was wondering if all men selected are actually high qualified or it's almost the same so just the part is highly qualified while the other part is of course very very good academy members but not exactly the the best as as the um definition of highly qualified says right that's a very good question thanks uh thanks so much for i'm glad you brought that up um so actually you know almost everybody that gets to be uh say an econometric society fellow has an extraordinary record um so they would have multiple journal publications but not always the econometric society 10 also rewards people who for instance were very influential in running a major institution a statistical agency in some cases or say a university where lots of there was a strong economics department for a very long time and the society also has a program of trying to increase diversity um for representation from around the world so in asian countries and south america and in even african countries and what you see is that um there to the extent that there are people with with limited cvs less publications they would tend to be men from those settings so some other type of affirmative action has pushed them in now almost inevitably they are men but they are probably the beneficiaries of a different kind of preference but if you're just going with straight academics a typical person who would be admitted to the contract society would have let's say five to ten publications in a top five journal and the the female candidates would tend to have one to two fewer uh the way we in our estimation models we we codify it in terms of um different types of journals and the most important journal for getting into the condimentary society is the journal of the society itself which is called econometrica and so being female is kind of the equivalent of having 1 or 1.2 extra econometricus so as as i was trying to say with that s-curve diagram you're really benefiting quite highly qualified women there's not any women getting in who are um you know not with a pretty outstanding cv but there is a gap a noticeable gap we can take another last question yes please as always come over here yes uh thank you for the question apologies this is not my field at all so sorry apologies this is not my field at all so uh i apologize if this is a naive question but i'm wondering um you know rather than actually having policies towards affirmative action should we actually treat the root cause because it seems to me that we are really treating the symptoms here so rather than you know having disadvantaged communities rather than having the quotas and everything else what about focusing on really improving their scs thank you i think thanks for that i think that's um that's actually um that's a very that's an alan krueger kind of comment um and i think that everyone you know certainly believes that um we should try and encourage um you know talented women to pursue science and and math and economics profession type fields we should try and get um you know talented african-american students to um do well in in high school and and be prepared to enter into um university i think the reality is though that the costs of that and the benefit are so high and the set of programs we have that seem to work are somewhat limited so if you're thinking about where will we be 5 or 10 or 15 years from now um one of the strongest cases for affirmative action i think would be for instance um in the harvard case if you said i need to have a bunch of lawyers who were educated at ivy league schools and go to the best law schools who are african-american and can make arguments for uh maintaining civil rights law and expanding civil rights law and keeping voting rights and so on and the only way that you can have that group of people on the ground 10 years from now is going to be to make a choice today about who gets into harvard or yale or princeton this year so i think that's the that's the kind of the choice that we have to think about a little bit obviously you know my personal view would be that it would be very important to try and spend more money and and do more research and figure out what works and try lots of different things and alan kruger spent a huge amount of his career emphasizing uh changes in the education system that could potentially benefit especially al and i did quite a bit of work on minority uh education programs so i think we i think we that's a separate thing though in my view because the reality is with the students we have today they are going to be the leaders tomorrow and if we don't have african-americans or we don't have women tomorrow or five years from now we're going to then delay things much longer so that that's the argument that people would make indeed but that's a very good question and obviously one of the deepest ones some of the justification being provided for affirmative action is that affirmative action create the condition to overcome also this type of regulation and but whenever we we we have this type of presentation i always have a kind of envy with respect to the data that often are available in the other side of the atlantic but we will get there at some stage and we will be able to carry out the same type of analysis and to guide economic policy as you did so well uh there so thank you again david for being with us and again we hope to have you with us when the festival will be back to normality with all the people and the speakers in the same room so thank you thank you again thanks so much you