The Economics of Investing in Children – James J. Heckman

welcome to the 10th anniversary celebration of the Center for trial family policy here at Duke University I’m Ken dodge center director I’d like to thank you for coming today and being a part of the festivities the skies have opened up the skies are indeed Carolina blue we welcome all our friends from Chapel Hill thank you for coming absolutely I would like to before we get going I’d like to acknowledge a couple of the people who are responsible for today first quickly but very importantly I want to thank Barbara and Eric aleyko for making today happen they have been with the center for a long time Barbara was the very first person that I hired even before I got here and it’s been a wonderful collaboration I also want to acknowledge a couple of people who might be here not sure but have been part of the inspiration for the center Joel Fleischman this is Fleischman Commons we are sitting in now and Joel Fleischman it was the inspiration through Terry Sanford for the Sanford Institute of Public Policy and I’m really indebted to him and and we all are I also wish to acknowledge and thank the directors of the Sanford Institute which is now the Sanford School of Public Policy which we are sitting Phil Cook was director 10 years ago and then Bruce gentle s’en and now Bruce kind of home all three have been terrific leaders and I’m very grateful to them there are many other people that could be acknowledged and I have been trying to acknowledge the number of people all day long those that I have acknowledged the least are those from within our own staff you know who you are thank you very much before you get to meet Jim Heckman I get the honor of introducing Bill chafe who will introduce our speaker and I wanted to do this William H chafe is the very Alice Mary Baldwin professor of history at Duke bill Jeff’s life has been marked by his sense that the university should contribute solutions to society’s challenges I first heard of Bill chafe when I was a graduate student here at Duke in the 70s I didn’t get to meet him them but heard about him what a legacy and in his own prize-winning research he has addressed historical challenges of race and gender equality in the American South I first got to meet Bill chafe in 1989 when our paths crossed that Stanford on the west coast and now I’m very lucky to call him my friend bill was the Dean of the College of Arts and Sciences at Duke University in 1999 who had a vision for the Center for Child and Family Policy that led to its creation he had an unwavering faith that the Duke University community of faculty members staff and students could contribute solutions to problems facing children families and educators if only we put our minds to the task and devoted resources where necessary I’m very grateful to Dean Shea for his vision faith and friendship and I am just thrilled to be able to introduce him to you thanks again I want to make a couple of remarks but before that Phil Costanza wants to finish some remarks that he was in the process of making Thank You Phil before I do this professor Heckman I just want to say a couple of words it is true that I think all of us have opportunities at different points in our lives to try to make some difference than the institution’s we work for I tried to do that as dean but there was nothing that ever reached the point of having a greater impact than the chance that came to me in 1999 to appoint Ken dodge as

director of the Child and Family Policy Center at Duke it was an extraordinary opportunity I knew Ken from Stanford would work together the Center for B Advanced Study in the behavioral sciences I knew this would make a critical difference in our University and of the many many moments of my deanship that were times of exaltation this was one of the best and I just want to say how important it is to have been here with you all these years and all you’ve done to make this a much better place now I have the great honor of introducing our speaker today James Heckman is the Henry Schulz Distinguished Service professor of economics at the University of Chicago where you served since the 1970s in 2000 he shared the Riksbank prize in Economic Sciences in memory of Alfred Nobel with Danny and McFadden he directs the economics research center in the center of a social program evaluation at the Harris school for Public Policy and he also teaches at the University of Dublin which is a wonderful thing to be able to do as an alternative to being in Chicago I think it’s marvelous and he’s also as a research fellow at the American Bar Foundation his work has been devoted to marrying scientific methods with economic policy evaluation and he is focused especially on models of how to measure problems and possibilities created by heterogeneity diversity on people’s lives love the life experience their economic experience this merging of scientific evaluation and social policy is most revenant and some of the pioneering research he has done on outcomes of people who have obtained the high school equivalency diploma which is meted May – impact as well as for his incredible work on the importance and salience e of early childhood education programs on the futures future lives of individuals his research has given important policymakers important new insights into how they should proceed with the best kind of scientifically verifiable information professor Hickman has published over 200 articles in several books his most recent books include inequality in America what role for human capital policy written with Alan Krueger and evaluating human capital policy and law in employment lessons from Latin America in the Caribbean he is deeply invested himself already in his two days here at Duke he spent three hours yesterday with barber friends talking about the eastern childhood initiative and we know that was gonna be very productive and when I came in here he was spending an hour I think with Governor James Hunt explaining the nuances of all the things that he was pretty he’s going to be presenting here today he has received numerous awards for his work including the John Bates Clark award of the American Economic Association in 1983 the 2005 and 2007 Dennis Agner award for applied econometrics from the Journal of econometrics the 2007 Theodore W Schultz award from the American agricultural economics Association and the 2005 Jacob mincer award for lifetime achievement in labor economics clearly we have someone who has an enormous amount to tell us and we look forward to hearing his remarks pres Ahmed thank you very much for those kind words and I’m really very pleased to be here to speak on the 10th anniversary I didn’t realize the full significance that that you were fully ten years old I guess I should have read the letter more closely but I didn’t but it’s it’s a it’s a good occasion and it’s one that I found extremely useful it’s a chance to meet old friends like Joe Hudson Duncan Thomas and many others in this room I don’t want to single just them out funny lad and others as well as new collaborations that we’ve had here with the with the group I guess across the way the ones the boys are the girls in blue from North Carolina Franklin Porter Graham Center and it is a very interesting two days that I spent here and I hope I’m invited back because we have a lot of interest yeah I can and I spent some time in Chicago last December and I look forward to further interactions with him on this very same kind of question I’m talking about now see if I can get it going so what I’m going to talk about today is the economics of investing in children and I actually gave a private tutorial to your former governor a few minutes ago and I actually told him I he asked me if there’s anything that I would have said differently had he had I been in his shoes and I said yes there was and that he had actually forgotten half of what’s this talk is all about which is non-cognitive skills and so I want to try to shape what a case is and sort of why this these these topics that were discussing in in the seminars today and discussions are so important and to think something differently about policy towards inequality and disadvantage and some of what I’m going to say repeats or maybe restates some of what we heard in

a very interesting panel earlier today but I think the repetition may be of some use especially for those who weren’t here so I want to make a statement which i think is so well known that it’s almost a bromide if people know that this is true American society the graphs that were handed out by Ron Haskins certainly demonstrated this American society is becoming polarized and another dimension that he didn’t feature was essentially less productive the u.s. productivity rate especially labor productivity has been in decline for some time it picked up it’s picked up over some episodes you know in the late 90s early 2000s but generally speaking labor productivity is a serious issue and so let me just give you a few facts I think some of you have a sheet where some of these facts are there but let me just lay out the sheets I have I am an academic economist I’m not a former governor how you would never be elected governor of anything I promise you but but I and so I’m I’m less politically sensitive maybe than others in the room but I still think it’s useful to kind of look at the structure of these policies of these problems and what American society has experienced and what I think is a refined way a new way to think about public policy so the polarization is something that’s real a lot of books are written about it articles people have noted this one dimension of this polarization is that at the same time we have a greater percentage of all of our children graduating attending and graduating college and ever in our past so the college graduation rate is actually picking up in the last few years at the same time if you properly measure it there’s a greater percentage of people dropping out of secondary school and therefore what we have is a polarization on the American Society now the official statistics actually mask that we have something called a GD which has already mentioned by that by Dean chafe and the the GD graduates about 12% of all of our high school graduates about 12% some states is as high as 20% GDS are actually performing at the rate of high school dropouts so I think it’s the only accurate calculation as GDS our dropouts and if you count GDS as dropouts we have a growing dropout population not just stagnant growing and this point this fact was alluded to earlier I’ll just repeat it but it’s a study that has retracted a lot of attention but if you look at the recruiting records and you look at American youth who apply to get in the military about 75 percent are ineligible either because of criminal records obesity or low cognitive capacities they can’t pass a test and so this means then by this measure now this isn’t a full inventory of American society but it is an inventory that we really are finding that a lot of American youth are not living up to a minimal level of standard we also have measured from the US adult literacy surveys at about 20 percent of the workforce has a very low literacy rate so low that if you actually gave them a vial of pills and said you know take one a day or something claim that slightly complicated they wouldn’t fully understand it so it’s a very low level of workplace productivity and we’ve seen from a number of studies work by DeLong and Katz and golden and others that there’s been a slowdown and the growth of skills and a workforce and so we know that this has itself real consequences for workforce productivity and the performance of the American economy in the future now all these problems are familiar a lot of people discuss these problems and I’m not telling you anything you haven’t heard or maybe a few facts you haven’t but the main thrust is there but the way that we analyze these problems is is in isolation of each other these problems are typically treated in a piecemeal fashion so there are a lot of lobbies really intellectual lobbies groups of people who will blame one institution or another or look for one sort of source of solution over another so a lot of people will blame the public schools a lot of people talk about high tuition costs so a lot of people we know tuition costs have gone up and most economists all the economists in the room would certainly recognize that when you raise the price of a good you typically buy less of it and that’s true for college as well as other goods but if you ask where the gaps in college going have appeared and I’ll show you some statistics on that what you’ll find is in fact that these factors are not playing the major role and so we really have to more than just take a list of these items we have to really prioritize and understand that the public schools I don’t think of the main problem although they certainly can be made better rising tuition costs are certainly a contributing factor but they’re not the main problem and I think it’s important for us to understand what these competing instant proposals have been and to try to put them in a firm ground so today I want to lay out a framework that allows us to sort of integrate on bodies of knowledge in economics and psychology and epidemiology and neuroscience that puts us in a common framework for thinking about these problems so this is what I want to do today and I want to argue that there is

a strong economic case all of these cases are provisional any empirical serious empirical person will never say they know the truth they’ll just say I know it as of today this is the ballot my best call and I have to say I’ve tried my best and and you lay out the case for that so this is no different from any of those cases but I think the arguments are strong and I’ll try to show you what they are so this is the argument let me make a few and this isn’t the handout if you happen to have it this this argument so what I want to argue is that many major economic and social problems like crime teenage pregnancy even obesity high school dropout rates adverse health conditions can be traced to low levels of skill and ability skill and ability are really important factors now that by itself doesn’t say too much but I want to try to argue it that that these factors are very important I don’t want to minimize any of the other incentive effects so if we don’t help police on the street we don’t hit it’s crime is going to occur so none of this is saying but when we have in a standardized incentive environment these early factors play a very important role but now this is the part I was quarreling with your former governor because when we think about ability we need to recognize the multiple aspects of ability so much of our discussion is focusing on the question of cognitive ability so we know for example in the No Child Left Behind emphasis and the Bush administration and in much of the way we think about whether our schools are succeeding or not even the way we think about the achievement gap is typically in terms of a cognitive test and we know what the cognitive abilities are very important there are lots and lots of studies that show cognitive abilities are important but we’ve also learned something else and this is something that doesn’t receive as much attention as it should and it’s something that actually is extraordinarily important if we really are to think about what good policies will be so if we look for example at other dimensions of human performance we would look at social and emotional skills physical and mental health we would look at things like perseverance attention motivation and self-confidence I was telling the governor the story about you know you little kids you know your little train that could a little engine that could you know I think I can I think I can we read those to our kids we know they’re important Thomas Edison made the remark that genius was right 1% inspiration and 99% perspiration that’s a statement about the importance of non-cognitive skills you’ll find this in Aesop’s fables Tortoise and the hare and so forth so that all around and yet in our public policy we typically are focusing only on these cognitive skills as measured by a test now the two things wrong with that first of all the chief minh on a test is partly perform related to non cognitive skills so you’re a Willa T your desire to actually succeed your motivation and your non cognitive traits play a very important role even on the getting the very test score but to the extent that the test score is just capturing kind of some aspect of the human behavior I want to try to convince you it’s missing a big part of the story and the reason why that’s interesting is because we can do we can first of all measure these skills sometimes in public policy discourse these are called soft skills skills it can’t be measured but there’s a lot of work that’s been done recently I’m trying to make those soft skills hard there’s a group of economists and psychologists that have come together not the behavior and not the behavioral economists working with a cognitive psychologist so much as other economists working with personality psychologists and the like and showing the importance of these skills and also showing that they can be manipulated and they can be manipulated they can be they can be altered in favorable directions and ways that I don’t think we even fully understood ten years ago and I want to show you that some of the case for early childhood that we think of as being mostly focused as teaching little kids you know maybe baby Einsteins are creating kind of more verbal skills at age three so they’ll be smart at seven but a large part not all but a substantial part of what these early programs are doing is basically creating non cognitive skills social and emotional skills and in fact I want to show you a few slides on the Perry preschool program showing much of its effect is mediated through its consequence on non cognitive traits these non so the understanding and how soft skills are really quite hard in the sense they have predictive power and that the evidence is not just correlational evidence but some causal evidence actually helps us understand much better and that the that we need to understand that that another component of ability is cognitive and is cognitive ability as we thought about for a long time but also non cognitive ability so that’s part of the argument another part of the argument is some that longitudinal data sets have given us a lot of information about and that is the ability gaps between the haves and the have-nots the advantage and the disadvantage open up and that’s they open up very early in life and these are gaps that actually are associated with

with both cognitive abilities and non cognitive abilities now that’s important and I’m gonna show you some evidence on that but we also want to argue that these abilities are produced one of the common themes in public policy discussions is somehow that these traits cognitive and non-cognitive are more or less genetically given their god-given gifts and well we’ve come to understand as how the environment does affect the expression of these gifts how in fact even things that are genetically determined are moderated by the effects of the environment on those and the reason why that’s so important is that and we look at family environments and this is already alluded to earlier about Senator Moynihan our guest at that time assistant secretary Moynihan’s fears about the family the black family in particular materialized for the whole society that in fact we do have serious issues with what the quality is of environments of early children so these environments matter and this is a source of great concern because the environments in some sense are getting worked much worse and so more than genetics is at work the family plays a powerful role and yet family conditions are not doing so well in many dimensions if you depends on how you measure family the quality of family life and I’ll try to try to discuss that now how do I know that it’s not just genetics well the reason why it’s not just Dianetics is we have a lot of experimental interventions some very distinguished interventions here Francis Campbell and a group in the abecedarian project at the Franklin Porter Graham Center as well as work at peri and other studies as well have shown by experimental manipulation that early interventions can make a difference at least in the study of the group that I’ve studied the most the pairing children the lion’s share of those treatment effects that come out of the program the lion’s share of the high rate of return to that are being mediated by increases in non cognitive skills and so what we’ve learned then is that these traits which we used to think of as being fixed in stone I mean William James the famous psychologist once said that plastic that the that that personality traits of an individual were more or less set like plaster early in life and would never change that’s simply wrong it’s also equally wrong to say the personality traits are completely malleable so you are what you need to be in every situation that’s the kind of Woody Allen selig version of the human personality so the truth is somewhere in between but it’s important when we look at this policy that we understand these policies that we understand the power of these non-cognitive traits so society intervenes early enough it can raise both cognitive and social-emotional abilities and even health I’m not going to spend a lot of time on health and a lot of evidence shows how they reduce inequality by promoting schooling reducing crime reducing teenage pregnancy and fostering workforce productivity and INRI analyses and analyses of the benefit cost ratios or we actually go and test and kick the tires on these we find that the rates of return on these programs especially the peri program where the the data are out the longest the kids are forty years old now we can actually get rates of return that are above the return historically on equity so we basically in terms of investing in people early on versus investing in the stock market say between 1945 and 2008 you’d get a higher return in persons rather than in the stocks so what we also need to understand and this is the question of prioritization in some of the discussions I’ve heard the day they’re well-meaning discussions but I’m an economist after all and I need and in a time of tight budgets we have to be really hard nosed we really have to understand that if we have a limited budget we need to spend our money where it has the highest rate of return I wish this principle had actually been applied more in the last year in the stimulus package because you know there have been actually Orszag computed the rate of return to pothole replacement it’s there is a positive return but it’s a lot less than they return to the Perry preschool program and if you go down the line and a number of intervention programs for example reduce pupil-teacher ratios these are things that received so much attention David Card and Alan Krueger did a series of studies that showed that if you have smaller classroom size if you pay higher teacher salaries you will find that this has an effect in terms of increased earnings for the individuals however when you do a cost-benefit analysis and say what if you hire the teachers to get the reduced classroom size what is the cost of actually doing this you get a negative rate of return for those same interventions so again we have to prioritize Joe and I Joe Hudson I did some a lot of work on the job training programs some 20 years ago the ones that President Clinton killed finally and basically the rate of return there was either zero in some cases negative so when we go down this list and we think about this large body of programs that are out there they’re all viewed as competing programs what we find and I’ll show you a little bit of evidence that things like adult literacy convict rehabilitation and and some a lot of the programs that were

funded last year under the ARA was I have much lower rates of return than the kind of programs we’re talking about today but it’s that kind of discipline needs to be made not just saying we want to do good things and we can qualitatively say that we want to do all these good things we want to actually quantitatively assess how far we can go how far we should go now the reason why these early programs have an effect is that we know that life cycle skill formation is dynamic in nature there really is a self reproducing a self productive notion your skill begets skill motivation beget motivation and they cross foster each other and we know and this is something that I’ll show you a little bit of evidence on that if we wait too late in the life cycle it’s not that it’s impossible to remediate it’s just much more costly so it’s always a question of so I was being I guess you were attributing to me the notion you should invest only in the early years that’s insane I mean obviously you don’t want to invest only in the first three years and then walk the kid up in a closet and then send him off to the work place I understand but no nice then but there is a sense in which the prioritization would actually be is toward of moving more investments towards the early years than we currently do especially for disadvantaged children because really I’m talking a lot about disadvantaged children and I will contradict one of your points to later on which is about targeting but I well I still will accommodate it so what we’ve come to understand though and I want you to show you is that the longer he waits the more costly it is and for some of these interventions we really haven’t found exist accessible ways so if you get a kid who’s got you know he’s an 18 year old illiterate say with a prison record so far the track record is pretty poor there may be a few exceptions about average effects for groups of individuals most of the interventions have been very ineffective but now that’s not to say we can’t do better and we shouldn’t try to do better that’s nothing as saying that at all but it is saying that what we know now suggests that’s not a very useful way so what we need to do is rethink our public policy to understand the understan to incorporate a life cycle of skill and health formation and the importance of the early years so that’s what I want to talk about today so let me give you some evidence on this and I am a victim of being of liking to many slides I am an academic after all I’m not a governor so I I like my slides well let me show you a few facts and not to mine too governor I have a great respect for your former governor he’s great but here’s the here’s a piece of fact information that got me onto this subject in the first place this is some work I did some years ago now nine years ago published with the owner Rubenstein and Jingjing C with the University of Chicago and what we were doing was you’re looking at G DS and G DS are the same groups of people I mentioned who are now about 12% of all high school graduates 12 to 14% and I was really curious about the G DS because I learned about them actually in the course of looking at job training programs a lot of g DS a lot of job job core used to mainly produce g DS and so it turned out that the g DS are earning at the rate of high school dropouts and that’s a fact if you look at the g DS they’re earning at the rate of high school dropouts but there’s a test and the test was this GED test that certified that they’re the equivalents of high school graduates so what we did is we looked at this is for white males or white females it’s true for all the other demographic groups so I just choose to not to bombard you so if you look at this is just the baseline distributions this is the distribution of the test scores for white males and white females my purpose here isn’t to make a comparison across the groups it’s to compare within each group where the GED is why so if I say where did the graph so the test scores of the gd’s lie relative to the ordinary high school graduates who complete seat time they do equally well the G DS are as smart at least as measured by this test the test does its job because the test is looking basically for people to who are successful in in in in passing a test but now if we ask what do we get in terms of earnings the rate of return on the GED is about the same as the rate of being a high school dropout so the G DS are as smart as high school graduates they complete who complete education by a class time but they lack non-cognitive skills how do we know that because we look at a variety of dimensions of their lives later on the G DS and not only the people who drop out of high school there are also the people who drop out of the military if they get into the military there are also the people who drop out of marriages that they get into marriages they drop out of a variety of workplace situations almost every environment they lack social skills and in some recent work we actually quantified those social skills and found that we can literally find that the G d’s have the same level of cognitive skills but by the new measures of non-cognitive skills they’re very low and as dimensions so that’s another that’s an evidence that that here here was the first piece of evidence that I found what was interesting is that the G d’s the military got onto this a long

time ago and in fact the Marines refused to take G DS because they’re too violent and undisciplined you know you have to be vite you have to be disciplined in your violence if you’re in the Marine Corps and they and the G DS lack that they just literally couldn’t they couldn’t finish the D couldn’t do anything so let me show you some evidence that I have some papers with Joris D crude and Sergio itsuo and we looked at some of the same factors that motivated a lot of attention and still receive a lot of attention the effects say of cognition Herrnstein the famous psychologist years ago looked at this relationship between crime Herrnstein and Wilson between crime and IQ and he’s noticed this very strong relationship a negative relationship that that the higher the IQ the lower the level of crime and this this graph reproduces that relationship what you find as you go from the bottom of the distribution which is low skills in terms of cognition to the top what you find is the people at the at the top are much less likely to be in jail at age 30 so that’s the cognitive if you ask the question what about non cognitive skills you get in this sense of asking if you where is the movement from the bottom of the distribution to the top you actually find a similar even more dramatic drop-off so you find that higher levels of non-cognitive skills I would even argue this is a causal relationship but certainly correlational er associated with much lower levels of crime and we can do the same thing for a lot of other dimensions of performance here’s the relationship between teenage pregnancy and cognition you’ve seen very high levels of teenage pregnancy for the low ability girls and then you go to the top of the distribution very low level that’s in terms of high cognitive skills in non cognitive skills the two distributions almost overlap if you ask as we’ve been talking about looking at the dimension of four-year college graduates we can ask again holding constant one factor moving the other as you go from the bottom of the distribution which does on the left side of each figure to the top you find how much higher levels if who goes to college so if you ask somebody who has a very low level of cognitive and non-cognitive skills say measured at age 9 and age 10 who’s going to actually graduate 4-year college you’re gonna get very high level of predictability when and and even after you include things like family income tuition and all that it’s really much these are these are powerful factors so I will I will so let me give you one factor though that’s very important many people are concerned about the achievement gap it’s something that’s disturbing about American society what we found this is what work I did with Steve Cameron again some 10 years ago or so is that once we control for ability measured at the high school going age it’s a fact that us minorities are more likely to attend college more likely despite having lower family incomes and beside other measures of adversity they’re more likely and so what we find that is that the factors that are leading to this are not high levels of tuition it’s not necessarily family income we’re not talking about attending Duke we’re talking about attending any college so this is community colleges and so forth and so let me just give you an idea I’ll just show you one graph here this is let me show you just the bottom which is these are all three relevant but I don’t want to beat you to death with too many statistics so let me see if I can use the mouse can this yeah this shows up that’s good so if you look at this bottom two figures here what you can actually see is that the gap this is taken around 2000 the gap between whites and minorities in terms of college going college entry now this is all levels of college it’s 2-year and 4-year schools was twelve percent favoring whites and if you look at the Hispanic gap was even bigger 14 percentage points but if you adjust for those abilities cognitive and non-cognitive at age 17 you can actually reverse the gap you’re saying once you standardize the ability you’re finding that that gap is actually minus 16 so in that says blacks are more likely to be going to college and Hispanics are more likely to be going to college that’s partly a matter or firm ative action and other issues but it’s still the case and the other factors play them and so what we found was interesting that the gaps and the abilities that plays such an important role are actually these these factors and that these factors open up at early ages these gaps in the haves and they have nots open up so we found that the the school’s the schools received so much attention about the quality of the school affecting the the scores and so forth we found that schooling even though it varies a lot in the United States high quality of low quality schools the schooling quality plays very little role in accounting for these gaps and that most of the measures of teacher quality that we least use I’m not saying the teachers don’t matter but the conventional measures pupil-teacher ratio teacher salaries have a very minor effect so let me just show you one graph and this is a graph taken from a study that I’m doing with Greg Duncan and Brooks Gunn and some other people this is basic let me just put the whole graph out here because what it shows is the

cognitive score this is just a measure of cognition by classified by the mothers educational level so the top graph would be associated with college-educated women the bottom line would actually be associated with the high school dropout mothers big gaps at age but the most important thing about this graph is that the gaps that are there at 18 are already pretty much there at age 5 and even to a considerable degree at age 3 long before these kids even enter school ok so that’s an important factor and there are hundreds of graphs like this I mean in this book that was mentioned with Krueger Pedro Cornero and I actually spent a lot of time going through this and there’s a lot of independent verification by many other scholars so this this finding is not unique to me and if we do similar kind of accounting in terms of social and emotional skills this is just one index and it’s not the same age range this is now between age 4 and age 12 but still the shows a similar pattern this is a problem of behavioral problems so there’s a behavioral problem index and if you look at the kids from the lowest income classes they now a score high they have a lot of behavioral problems those at the bottom are those bottom of this graph of those from not more advantaged families and you’ll notice again those gaps at age 12 are already there at age 4 are mostly there at age 4 there’s some role but those gaps are there those early years are really important and there’s also some similar findings although not the parallelism there’s some work by case liubot ski and pass and showing if in a measure of health disparities if you go from exactly individuals who go from here’s a scale where one to five the higher the score the worse the health there are real disparities that open up and they continue opening up across age disparities across income level and then by income level they’re actually widening with age so these gaps emerge but now when you look at graphs like that there are immediate questions that come to mind if Charles Murray were here you would immediately say of course because that’s a genetic predisposition so the evidence leaves open a lot of questions and this is the kind of research that we’re doing and a lot of people in this room are doing to try to understand what produces this so the first explanation is genetics you see so smart kids are born to smart parents so you’re just seeing a reproduction of a genetic process or is it due to family environments is it really due to the fact the families are doing better or more than just environments really active roles by families not just that the kid mother’s smart and that kind of spreads of an environment but also that they take the kid to the they read to the kid and sent me well there’s a lot of research that’s looked into that and it’s still going on some of the people in the room are actively contributing to that research and I probably shouldn’t mention names I will by attribution here the on some of these graphs but this is the kind of question so the evidence is that from the intervention studies which are extremely valuable that we can actually manipulate family environments and we can look at environments not only for humans we can look at it for right rats and we look at it from monkeys and I’ll show you some evidence on both because with monkeys we can do a lot of things we can’t do with people and in many ways we can actually get a firmer understanding through experimental manipulations that are basically illegal for human beings but but nonetheless the so then the question then becomes what do we know so let me let me just give you one dark fact which is I think it’s actually quite dark in this this is kind of reinforces the point so we talked about the Moynihan issue so family environments will matter let me just tell you why that should be a great source of concern because really what we’re talking about today is family policy people like to talk about early childhood because they don’t want to get anywhere near the issue about families but if you look for example at the single biggest growth sector in American families what you’ll see is that the percent of kids under 18 living with one parent and who where they’ve been actually never seen the the mother’s never been married and the father is really absent in a fundamental sense that’s been the biggest growth that’s the growth trend up there so if you look at the overall statistics for the u.s population it’s around twenty eight twenty nine percent but the biggest single growth sector has been in this never married category so the family is substantially different than what it was even 40 years ago the the the divorced do the percentage of kids raised in divorced families that’s been fairly constant actually over the last few years and why is that important and this is work of so young moon it happens to be here in the in the audience and then in fact we have measurements from a lot of data sources I’m only giving you a small snapshot of a large literature each one of these has like an iceberg the tip of an iceberg there are lots of things there but for example if you look by family type single mother broken intact one of the crudest categories we can come up with if we look at measures of cognitive stimulation what we see is this figures on the right the ones the blue figure

what we find is the intact families are spending great deal more time with cognitive stimulation reading to the child promoting learning and so forth and this has been very well documented and the patterns continue over the lifecycle of the child so what we’ve also come to understand then is that a major source of inequality in the American Society is the quality of the American environment so now let me just show you some figures it may not be pleasant news but I’ll just show you again from C Jung Woon’s work and this is basically looking at a graph of how resources are distributed differentially across different ethnic groups so looking at Hispanics and we’re looking at African American populations now the way to read these graphs would be the following I’m putting the Hispanic and black distributions in the in the white distribution if things were equally distributed you get a straight line across that would be roughly at point 10% these are deciles of the distribution what this shows is that blacks and Hispanics amilies our way over-represented in the low end of the material goods distribution and if we look at other measures for example and so you say well that’s family income but look at measures of cognitive stimulation again a massive over-representation and so what we’re fine for emotional support so developmental factors that we know to be important and even when we make adjustments for things like family income in a long run scale and we look at things like like mothers education we still find substantial gaps and that’s important because I want to show you that these factors are playing a very important role in explaining child outcomes so this early deprivation is there and it’s the real measure of poverty see this this is something that you merge from these studies when we think about poverty the traditional role I saw the picture of Lyndon Johnson as I walked in you think of the war on poverty forty years ago and Mollie Orshansky and this whole literature about how to measure poverty well if you want to measure poverty for young children it’s not going to be just family income going to be these resources like emotional support parenting because those are the factors that matter the most and that’s actually the correct way to do so and if we do that we get a much richer notion of what poverty really is all about it’s not just family income there can be some very rich families that apparently give poor emotional support my favorite example is Paris Hilton actually yeah as an example but I mean there are other other examples as well so let me talk about another line of work though which we’ve come to understand this is where the neuroscience comes in I think it’s interesting I don’t want to go into this because there’s always a danger of flirtation with areas you don’t really know and so forth but actually I’ve gotten some experience in this the genetic argument is always there on the table it’s been on the table and always will be on the table and hundred years ago in the era of social Darwinism that was the argument right then we really did have reproduction people didn’t really understand genetics like we do now but there was this argument that there was people born dumb and there were these social classes and they reflected a certain wisdom and and the aristocracy was smart and the the poor we’re not what we’ve come to understand in the last 30 years is from genetics now and from and from hard biology is how experience gets embodied in the biology of the organism and we understand how poverty actually operates how these low levels of family skills actually operate and what it does is it actually affects the way the gene affects you so we know a lot more about genetics and we did 100 years ago we had discovered DNA we have all these processes one of the most interesting features about Janine’s of course is the genes themselves do nothing they have to be expressed and the expression of the DNA or the lack of expression can lead to powerful differences in them dividuals so two individuals can have exactly the same genetic traits but they literally will have a differentially methylated gene how the gene expresses itself and how that gene is making the protein that makes the behavior that we think is associated with genetic behavior so let me just give you some a little bit of evidence on this because it’s a deeper understanding of how the early years play a role here I’ll bring in some rats just because I can’t resist it but it’s it’s actually quite an advance in some monkeys so the point is is that experience gets under the skin in a modifying DNA so let me show you I don’t know if these graphs are that helpful they’re a little too small I didn’t notice that till this morning but here it is what this is is it shows the DNA it’s it’s actually the patterns of the methylation of the DNA in identical twins these are people who have literally the same DNA they literally share the same genetic material now as I said the genetic material is expressing itself though only through this methylation so methylation can neither expert can suppress or accentuate the the expression of the genetic material so if you look at these three year-olds they there’s a twin so that’s one twin is on the left the other twins on the right if in fact the the expressed DNA was the same for the in these two groups the coloration pattern should be identical

and what you all I really want you to see and I’m not trying to obscure this they got a little too small but it’s a paper published in the Proceedings of the National Academy and you can see that they are really different the coloration patterns are there you get 250 year old twins identical twins they’re dramatically different and so the methylation literally the experience getting under the skin is affecting the behavior and it’s affecting the things that generate the behaviors what we’ve come to understand also is that early environments play a very powerful role so although the genetic arguments on the table the genetic argument is not all powerful so there was a finding I won’t show you that finding but there was a finding that Wilson actually featured many years ago and that is there’s a certain gene or a certain short allele of the gene the name isn’t so important but that gene was much highly correlated with criminal behavior so you went to the State Penitentiary you’d find an excess abundance of this gene compared to others now what we found recently what I didn’t find it but what was found recently was in fact that the expression of that gene only took power only had predictive power if that genetic configuration was put into an adverse environment so if you had kids who were born and raised in ordinary middle-class environments having that gene was not at all predictive of criminal behavior on the other hand criminal and aggressive behavior so the environments matter let me just show you one one example and this isn’t a medical example actually so and this is turning the corners around but it’s showing the gene-environment interactions it is so important so the recent paper in the Proceedings of the National Academy by Denice and he shows that maltreated children are much more likely in adult lives to have inflammation so a process that leads you makes much more likely to you know yeah it’s it’s a it’s a it’s a condition that makes you much more susceptible to disease and a number of diseases and so there was a serious difference between the maltreat and the non maltreated but then what he found what it was mediated through a genetic mediator so it wasn’t all male treated this doesn’t endorse maltreatment bother but it does suggest though then effectively that when we came to understand these early genetic associations that were the stuff of pop science 20 years ago that a lot of that just missed the point that it’s a gene environment interaction there’s a rich field of epigenetic interaction this is a study done by francesc Champaign who’s a psychologist at the University of at Columbia University she worked with some scalp leading scholars and in Montreal but she’s a leading scholar and on write she’s looking at rats and there turns out there’s a maternal pattern which is licking and grooming so if a rat mother licks and grooms rat pups that is good mothering for a rat she doesn’t you know low level of licking and grooming that that leads to it’s been known they’ve been associations well what she did what what’s Aransas champagne did was actually look at the methylation of the gene this is kind of an index of the differential methylation and so what she finds is that at Birth the the kid the mud the offspring of looking low licking and grooming dams and high licking and grooming dams are basically the same that’s if you look at the figure under birth and then when you go out into life what you find is the low licking and grooming mothers have a much differential methylation much higher methylation suppression of sort of gene products and so forth and life cycle and the low licking and high licking and grooming mothers much less and these persists throughout the lifetime so this is a sense in which adverse environments so we have fragments of evidence I don’t want a predictor we should go from the rat studies to passing a bill in Congress tomorrow but it does suggest very strongly and I’ll show you something even more I’ll skip over this but in the monkey studies work by Swami it shows even more dramatically so this is like looking at licking and grooming and it’s looking at behavior really it’s not the full genetic material it’s only a part of the mouse’s brain but there’s a recent dramatic study by Sumi and his co-authors and I’ve actually participated in that study to an extent and some of the modeling and so what happened is if we look at putting monkeys into adverse environments these are monkeys that are reared pure reared environments versus mother reared environments you can literally experimentally assign parenting strategies and then you can look at what the consequences are for the methylation of the gene and see it into adulthood and so what happened is the peer-reared monkeys think of that as monkey poverty are ones who exhibited expression of genes that involve more inflammation all of these negative aspects in terms of adult health so in terms of adult health attributes serious limitations surrogate mothers if you adopt them in instead of having this just rear by themselves surrogate mothers can partially reverse it and the ordinary mother you don’t get this reversal at all so I’ll just show you I don’t know if you can see this graph very well I guess you can’t I’m

sorry I’ve made these shrunk more than I thought in in the last few days they shrank in my computer but the only thing you really need to see is you have three conditions where the mother is rearing it that’s the top the surrogate and the peer reared and the different this is a heat map and it’s showing differential methylation so instead of just going through the fact is you see very different methylation patterns and these are full scans on the genome this isn’t just one gene it’s not just one part of the brain it’s not just the rats hippocampus this is the whole gene and this recent study by me and Swami and zip show that basically early peer rearing is affecting close to 22 percent of the whole genetic material it’s not even known yet about what the full ramifications of that are but child poverty plays a huge role okay so what do we know about this we also know from again from the from the literature and psychology but are critical and sensitive periods I know a lot of people have talked about this before and I got to take a graph from Charles Nelson and his co-author Thompson Russell Thompson and looking at the looking at what critical and sensitive periods are these are periods for development of different traits we know for example in the acquisition of the second language but learning a second language before age 11 is generally more efficient you literally learn what the different part of your brain later on so typically it’s very hard to acquire a second language about an accent after age say 10 to 13 we know in a lot of stages of development there are Syria there are critical and sensitive periods where if we don’t intervene it makes a big difference so let me let me let me just refer you to that literature it’s a very important literature but it’s not fully developed so in the sense that right now I want to take it back to the human studies and take it back to the studies that are a little less biologically motivated so the real question and this is always the important question is always from the point of view of economic analysis we want to understand how easy is it to remediate the effects of early disadvantage now why do I say that it’s it’s not that early factors matter early factors do matter we know that but the question is always going to be well from a purely economic standpoint it’s always better to postpone right if I can do the same thing at age 18 its earn the same money at 18 then at age 1 I can actually spend more money at 18 because I put the money in a bank so economically it’s always a better thing to wait everything else the same but the real question is how costly is it to delay and that gets us back to the question of critical and sensitive periods and what do we know about the optimal time for interventions so let me let me show you some some evidence this is now going to be excursion into some of the early childhood literature I mentioned already of the peri program let me show you some evidence the peri program was a program that’s probably the best studied of all these programs not the best program and maybe in ten years so now I’m not necessarily the best study but but right now it’s the kids are 40 years of age people randomly assigned into treatments of control and treatment conditions and they were followed for 40 years all of these kids were disadvantaged forget American children they were all in one town west of Detroit very poor town very very deprived area generally a few neighborhoods that are good but this was not a good area in terms of affluence this was all in the period 1962 to 1967 the treatment in this program was fairly modest it basically consists of two and a half hours of classroom instruction and the instruction is interesting because what was done was you weren’t reading you weren’t teaching alphabet to these kids although there are alphabets I’ve been to the classroom was pretty much the same and you can see the alphabet all around the room it’s there but what was actually done is what they call the plan do review sequence so kids came in each day they defined tasks they finished the task they stayed with the task remember the perseverance the self-control and then they reviewed the task they also worked with each other and there was some some training aspect of social skills there was also a family that parents were actually visited once a week by the 90 minutes each week by the teacher and various aspects about this training of the teacher a training of the school what the school was doing were instructed okay so what’s amazing about this is that this program these people have been followed for 40 years and they’re all actually turning 50 so they plan to follow them age 50 now this is the important part taking me back to the beginning of this talk Perry did not raise IQ did not what it did do is it raised the non cognitive skills and it was exactly what that wasn’t even what they were looking for in the 1960s the whole emphasis at that period was raising IQ so in the late 60s some of you or remember this Westinghouse report that was put out by by on the Head Start program and they argument was a failure of a program couldn’t possibly work the reason is is that treatment and control groups they had a big surge on their IQ and then they faded out so Head Start couldn’t work so that was the was the argument well just like Head

Start Perry did not raise the IQ he had an initial surge I’ll show you the ground the treatment group in the Perry program they were all enrolled at age three these kids were subnormal IQ they were chosen to be subnormal IQ they were all disadvantaged kids so they were chosen they have low IQs the kids had an average of 80 into the program during the program just like head start big surge in the IQ and then it dwindled down to the average of 85 the control group basically started at the same level didn’t experience the same surge but by age 10 the treatment and control groups are basically the same no effect on IQ so if you evaluate American policy by IQ effects this is a miserable failure and yet Perry had a huge effect I’m just going to show you the results for males this is a reanalysis that we did so we can actually decompose so all of these effects I don’t know if you can see them I can read them off the very first one is in achievement test it’s California achievement test at age 14 the second one is employment at nineteen monthly income a 27 if you look at people participation on welfare arrest which is a huge dimension so in this this reanalysis of the Perry preschool program we looked at the cost-benefit calculations we found very high rates of return a lot of the analysis group is here right now so we can decompose the effect they say how much of this had to do with the boost extra boost in cognition that was due to due to the Perry program well for males there wasn’t this is a slight difference there but it’s not statistically significant so a slight effect on the achievement test but not much if you ask how much of the treatment effect this is just the percentage of the total treatment effect all of these treatment effects are statistically significant and the rate of return is statistically significant so I’m not giving you just numbers at random and we do adjust for multiple hypothesis testing so these are significant tests these are things that test for multiple testing procedures so this gets rid of cherry-picking of Selective results but for example if you look down here at this bottom slide associated with reduced crime at age 40 total cost associated the crime came because of the boost and non-cognitive traits we have direct measurements of those traits we can then ask how much of these direct measurements translate into explaining the achievement effect so this is just a mediating analysis and so forth and we look at other measures of non-cognitive traits that there’s still some additional component as well so even the rest of it is not fully explained we’re not explaining all the treatment effect but still 70% isn’t bad of a pure treatment effect and even in the worst cases on welfare participation we’re getting twenty twenty five percent but the measured cognitive and non-cognitive factors cognitive factors are playing a fairly minimal role that’s the rest by the way okay so what am I saying I’m taking too long maybe but what am I saying well what I want to try to emphasize and I think the center here is focusing on I talked to a lot of people I know that that ken is talking a lot about this and adolescent interventions but there are others as well is the developmental focus and so what we want to try to do in our work is to move beyond this kind of level of just looking at treatment effects I already did a little bit already and decomposing it but we want to think about the basic principles that underlie what these programs do so that the danger with these programs is we do a experimental evaluation say see this program worked and we say okay here’s another program and this program works we look at a bunch of treatment effects typically we don’t even look at the rates of return so we don’t even compare them rates of return and we don’t understand why one program worked better than another or what was it that was really going on so what we’re trying to do is understand how these programs and how families together these programs are just supplements to the family these are nothing more than supplemental parenting programs compensating for adversity in American family life so we want to understand through what channels do these operate that both the family affects the non-experimental family studies and the treatment effects personality cognition and health and so to do this I realize now I’m running afoul of Hawking’s rule Hawking said do you lose you lose half your audience for each equation now technically this isn’t an equation so your cells should be here but this is so I hope you don’t I’m near the end anyway so but what this is showing is just a basic comment so this is trying to codify Thomas Edison and so forth it’s basically saying that outcomes that’s what Y stands for they can be a bunch of things it could be an achievement test score it could be crime it could be how well individual did on a test it could be how well a person did in the workplace how well the person fixed a gun in the military there are a number of things you can think of je is a large number of attributes and the basic theory of course in economics is one of

comparative advantage and it’s one where individual outcomes are determined by capacities and by the incentives to use the capacities so again if we look at crime we’re not going to say just criminally prone people you know some Thetas are there and there and people who commit the same amount of crime I suspect we’d all be happy to steal a billion dollars if we could do it without ever being caught so that’s the incentive environment so so really that we’re none of us are completely immoral but the point is is that is that we want so I want to understand and the way to organize this evidence is to think of both the family studies in the experimental studies as affecting theta as its measured in different control different environments and those controlling for environments is a huge literature but the second ingredient is this dynamics of skill formation so what it’s saying is these capacities these Thetas are self productive so that’s where skill begets skill it’s where investment plays a role and where parental environments even independent of environments play a role and peer effects and other larger groups the distinction the last two arguments is and so isn’t so important so the central question though so so here’s the idea capacities matter capacities affect a lot of decisions how we decide to specialize in life you know comparative advantage is not just something international trade it’s our life for me we all know we all tried probably to be great musicians great athletes great this and we all settled into our mediocre current status as a result of finding out our comparative advantage was there we’re not going to be Einsteins but we are going to have we are going to be relatively good so that’s where that’s what that first equation is saying and the second equation is saying that we can improve those capacities and that’s the new part because there’s a huge literature I could give you specialized literature’s there’s a literature and criminology that basically talks about types that are pure early on in life and you’re like rocket ships born into you’re over your whole life and yet I would argue no that second argument and the third argument of those functions play a very important role environments can mediate what initial endowments are but we need to understand that I’ll show you a little bit of evidence on that and in shut up so the so what are the central questions the central question is for Public Policy what’s the most efficient way to intervene so these capacities matter I’ve tried to show you some of the evidence on that there’s a large and growing body of evidence it’s cognitive non cognitive and health components but on top of that there’s also a sense that these matter and that they can be manipulated so we can now start understanding of how families and the peri program and the abecedarian program compete on the same footing we can literally show okay in a rich family environment is improving social and emotional skills by this amount so some preliminary calculations at Flavio Cunha and I made a few years ago basically said that Perry if you took our technology function was basically like moving a disadvantage Kent from the bottom decile of the endowment of the family investment distribution to about the 70th percentile so you’re really moving it up now not changing the parental characteristics just doing raw investment just doing that so you can you can codify that and then you can understand which dimensions and are which capacities and where and the lifecycle investments are optimally placed so this kind of framework is a guide gabriela Conte’s here and we’re we’re integrating this with health and many other dimensions but if you look for example at this the goal is cross fertilizing so maybe I should wind down with making a few remarks to simply say this scheme Ana has so see people falling asleep already because it’s the equations are there they haven’t left yeah that’s nice but but let me let me let me just conclude so the idea is to really try to understand the development will focus on capacities and do we we can quantify these and so let me let me just summarize what are two polar cases in the economics literature and the public policy literature now I won’t put up any equations or I’ll flash by them the first equation that really dominates the first view of human skill and I think this was the Lyndon Johnson view of skill this was an optimistic view of skill was that we can invest more or less at any age and we can make good so remember the war on poverty wasn’t just headstart it wasn’t just it was like 40 and 50 year old steel workers who were disadvantaged it was illiterate living in Appalachia and the whole spectrum of people and so that was an idea that economists would call perfect substitution at some point you can compensate might be more costly later on but you can do it the opposite case is the case which we can you couldn’t think of as the left shoe right shoe technology economists call it Leon T of Technology it’s basically saying that youyou need a left shoe and you need to write you both to walk if you just have one it’s much harder to walk so so the

idea would be that if in the early years you receive no no investment in the later years no investment that you make later on can compensate so if you don’t get the minimal investment early on later on you can’t do anything that’s a very bleak view you can actually estimate you can quantify that I won’t show you the estimates but we have a series of estimates of how much you can remediate early worth late and the literature it’s not it’s certainly not perfect substitute it is not we do find that the early investment has a much higher rate of return and the cost of remediating is high later on especially for cognitive traits it turns out the substitutability across the lifecycle for non cognitive traits is typically it doesn’t change that much you can remediate early and late for non cognitive traits but for cognitive traits gets to be much harder but so where do we land in this spectrum it’s much closer to this left shoe right shoe technology than it is to the perfect substitution and what does that technology say it says you can’t remediated on four early disadvantage if you have a low level of skill later on don’t have the base it’s very hard to remediate later on but it also says and this is your point that if you don’t do later on you’re not gonna get any effect from the early on so it really is saying you need complementary investment over the whole lifecycle the technical term is complementarity so we can measure that so let me skip past all this that’s what I said in words and let me just let me just so I’m gonna skip all this and and conclude actually I should conclude because I’m over my time so if you we can out of this technology we actually had there literally this is a freehand graph and this still is a freehand graph but it’s actually rooted in some data that we have it’s an implication of the technology with Cunha and basically what we can say is suppose you imagine the hypothetical experiment I invest a dollar in the life of a child that each day in the life of the child $1 if I do that give everybody at least something now I ask having done that say what’s a $1 each year I now ask where should the next dollar go where is the marginal return the highest and the marginal return turns out to be highest for the earliest years why because of the self productivity I raced through that argument because the skill begets skill what happens is an early investment makes it much easier for later invested so when when the governor and the panelists we’re talking about trying to route this into education this has a direct tangible output you motivate children early on and you get them directed and you get them highly motivated and you can avoid a lot of the special burdens that are placed on the school system special education and a lot of the difficulties that are associated with special education and so immediately you get a higher but that translates into a higher return why because you save those costs and you make the child more investable later in the life of the child and so that’s the notion that you really get a dynamic so it’s a multiplier if you will not a Keynesian multiplier but a lifecycle multiplier where you’re actually showing that the investment because it’s raising the productivity of future investments is making it easier to invest in the child less costly because you don’t have to remediate and producing a very high rate of return and so you find this pattern which is the early years have a very high rate of return the later years very low rate of return not zero though but nonetheless lower now this is a this is what economists are calling out of equilibrium relationship so what the optimal policy would be to invest so that the rate of return is the same at all ages but if you look at the data likes Young’s data and other data where you actually find what the investment patterns are advantage kids in the earliest years you see the greatest deprivation so the pattern of optimal investment is input for the most advantage kids they’re getting a lot of maybe they’re getting smothered you can refer to other Woody Allen movies besides Ealing about mothering and so forth I don’t know if you want to hear some public policy discussions but there are there’s some practical issues which I want to conclude so let me just talk about this because I think it relates to some of the discussion of the panel there are practical issues this is an area that’s in flux I’m obviously excited by them very excited because I see a symbiosis of economics psychology neuroscience epidemiology a lot of interesting trends a lot of work still to be done I think we we have a lot of open questions whom should we target and what is the best program and who should provide the programs who should pay for them and what would be the compliance so let me just talk very briefly all the studies that have been done to date have shown very high returns that disadvantaged people and it raises the question of what the proper measure of disadvantage is I don’t think it’s poverty like we traditionally measure it I think it has much more to do with measures of disadvantage in the early family life related to parenting and the resources given to the child quality of parenting that plays a very key role there was an experiment on a natural experiment if you will in an Indian Reservation in Connecticut about 15 or 20 years ago and

what happened was that an Indian tribe opened a casino this tribe had been the subject of many many studies by sociologists and anthropologists and this tribe was destitute very poor suddenly you get they were unable to open a casino and they became very rich and meanwhile the anthropologist sociologists have been studying child welfare and so they could actually take before-and-after measures the casino was hoping it well they found improvement overall in child measures but then they started looking more closely and they had been measuring child parenting practices where they found the improvements were in those families where there seemed to be a definite improvement or change in the parenting style that came from the increased wealth so the structure then was me so it wasn’t just the money because the parents all got shares equally it was really way parenting was doing so that’s at least a hint so I think that’s why so and cultural groups and have been have been more successful even in the face of adversity and poverty and we know that within any any ethnic cultural group you find very sharp differences that very poor people can do extremely well and raising the children simply the quality of the parenting how well the mother sticks so what’s the program I’m not going to advocate one particular program sorry the abecedarian group but these are all promising I mean there are a lot of very interesting programs they they’re being evaluated now into the adult years that plays a huge role in understanding what rates of return are we know that what i would argue from the peri evidence and I would argue will show up in other cases that programs that are essentially building character and motivation not just focusing on that cognition not just focusing on verbal skills are ones they’re gonna last but we need to sort through this list much more finely there’s a danger here people saying okay this program worked the one thing we do know is that there are a lot of other possibilities out there so we need a really an experimental approach here we really need to be thinking yes we need to evaluate what was try we need to know what the mechanisms work but we also need to think more broadly we just shouldn’t think about replicating one program that’s had some measure of success and thinking that’s going to work now the question is and this is a big issue and this is something that I don’t think was discussed adequately in the panel today so I’ll get my my two minutes in and that is who should provide the programs there’s been a general feeling that these are governmental II surprised programs but that’s not true that should not be true for many reasons it shouldn’t be true first of all one of the biggest sources of opposition is the question about these early programs intruding into early family life I mean there’s a reason why programs are not adopted a lot of these programs because when you’re taking zero to three year old kids and you’re taking kids away from their parents and you’re saying now I’m gonna let somebody from another group start raising my kid or inculcating different values you’re really challenging a basic premise of American and most societies so I think you need to design programs that are understand that you really need to respect that for so you need to respect cultural diversity and I think you need to think about bringing in a lot of actors in this program precisely not only because you would expect in a way that maybe a single government you know the problem with a lot of government programs is is forced towards uniformity as a sense of treating everybody alike but here’s a case where different ethnic and religious groups have different values they should I mean I mean there’s I can give you scenarios that are quite bizarre where you know you might train people in certain ways that you may not like and others may not like but nonetheless I think the structure should be that we bring in people not only because it will respect the sanctity of the family and not the cultural diversity is somebody would mentioned a discussion in Holland you and sunny you were there in Holland but there was also an issue in Holland that they again the Dutch were very anxious to have a universal pre-k program right and and to make it fair and not to adapt and a very culturally sensitive way to the North African population so kids came in they were actually teaching in Dutch some of these programs not really respecting a lot so there’s that and then there’s the other component which is funds at this point in time we really do have shortages of government funding and we need to try to bring in support then the question is should they be universal or not and so my viewpoint is fine make them universal but have a sliding fee schedule and so if sums there’s always this fear of stigmatization okay so make them universal everybody can come to Head Start or enriched headstart but then there’ll be a tax by income that could be arranged privately yes or the IRS or something so I think what you really want to do is I don’t think the universality issue is such a deep issue and I do think with the time of limited resources as an economist I cannot in good conscience say we should spend money and what we can obviously call deadweight loss namely taking the kids from affluent parts of Durham or the suburbs of Chicago and giving them free childcare when they’re already have a lot of resources and if anything they may over over a parent their children and then there’s the question of issues

of compliance and and this really gets back to the Family Values question and that is a lot of successful programs change the values that runs counter to the values of parents and we should just respect put it out on the table there may be serious challenges this is something that many people don’t like to to admit but I think that’s the serious tension that’s out there I gave a talk on this Edinboro a couple of years ago and I remember I mentioned a program called Big Brother Big Sister which is an adolescent intervention program and the newspaper headline the next day says Chicago economists advocates Big Brother coming into the house fall but that’s the fear that’s the fear that’s out there and there really is this is a very delicate issue you are dealing with the family these this is not just a template you have to you have to package these in a way so let me summarize a lot of current social problems I think have the roots and abilities there are other factors as well so I mentioned crime you know I’m not saying that the only reason why there’s an education deficit is because of abilities but I’m saying if I had to put a first order accounting that would be the first order ability deficits open up early they produce inequality and they also reduced productivity and a lot of evidence shows that these interventions play there were critical and sensitive periods and this is associated with some very storm neurological basis animal and human studies which I reviewed and what we found is that early investment is generally more productive and it’s because of what we’ve come to understand is the biology and psychology the the evolution of the brain and the evolution of how skills are created these critical insensitive periods are are documented and increasingly getting documented so if we construct an optimal portfolio it’s generally weighted towards the early years and because the early investments create a base for later investments and so this leads to the question of should we invest in early versus late well the advantage versus disadvantaged I think we want to look at the disadvantaged primarily getting the subsidy anyway but in anything we have to be careful and know that the knowledge base needs to be expanded this is kind of a trite saying that every academic talk ends with but in this area is particularly strong because I worry about the rush I think when the leaders in early childhood get together and meet when they’re honest with each other they say we don’t really know what the best policy is we know that we know hints of certain things that’s why it’s important not to say here’s the treatment effect but why did the treatment effect work and what what did we learn from it I think what we learned from it was social and emotional skills I think what we’re gonna learn from it in the future and we really haven’t looked at it enough is health health early health is gonna play a big role but a lot of the early intervention programs didn’t even think about health they didn’t even sew the Prairie study you can’t even get the body mass index at the very people at age 40 that was how unimportant the antha the body measurements were so that’s the point so what we’ve learned then is that yes remediation can work if we don’t do investment early on but it actually is much less effective and it’s costly we should work at improving it but it nonetheless is costly and so I think that Social Policy really should be redirected so if you think of how we spend our budgets so you’re talking about you know spending money on the on the over 60 set as opposed to the under 5 set well I would certainly say the disadvantaged under five sets should receive a lot more attention but it should receive attention and accountable we guided by these principles but with knowledge that we don’t really have a best program yet and that I very much worried about people rushing out and saying well I like this or I read this particular graphic account of my program or I read the website of this program or that there’s a danger in that because there is a lot to be known still but I think the contours are there and I think that’s the exciting part of it so there’s some of this that some of this is this website I didn’t create myself but if you want to you can download some of this material and it actually does have a lot of documents associated with it with the technical scientific papers that support it so thank you very much for your attention and I so I guess I get to be moderator so I can stop huts from asking any questions ha ha no seriously I’d be happy to take questions I much prefer I went on too long I realized but yes what’s that well no I mean there’s certainly evidence that suggests that cognitive traits are more highly valued I mean the environment is much less of a blue-collar manual labour environment so there’s definitely some strong evidence that suggests that technology changes the price of skills or been surprisingly hard to actually isolate a huge change it’s just related to return to cognitive ability and people are working on it but there seems to be some indication Murnane and Willett and Levy have actually done some work on that so

there’s some evidence that technology is moving towards a more cognitively and that’s common sensical you look at the use of computers and look at the use of the technology that’s using higher skills yes yes I race through that so actually the estimates of the technology are really showing the following that a higher level of non cognitive skill actually makes it easier for investment in cognitive skills so if you actually would so this is common sensical putting it forgetting the language if you just put in the in the in the common sense is saying if the kids are motivated to learn and you make learning investments in them they’re gonna learn more that’s what I just said in a common sense term and and yes there’s evidence for that interestingly though and the some of the estimates with Flavio there wasn’t the reverse effect that the cognitive components didn’t seem to help that much in promoting non cognitive we don’t know what the estimates are for health I mean that’s to be determined we were still estimating those technologies but so that’s the sense in which higher level of cognitive non-cognitive skills promotes the base for learning and then there’s a self productivity effect so literally there’s a cross effect that higher cognitive skills would promote the growth of higher levels of non-cognitive skills directly but they’re really strong effect is through this an enhancement of investment oh yes the eternal question about public policy the eternal question I I would say this though we have some again it’s a rough guide I mean you could formalize this there are lot of interesting papers that do just that but the short and intuitive answer is basically I think now if you look at the way we spend our budgets we’re spending a lot of money’s fell on job training programs youth so the question about say where do we get the additional resources to do this if we simply applied cost-benefit analysis to programs currently in place you know there was a cost-benefit analysis done on the Job Corps the Job Corps raised the earnings of participants by three dollars a year for the standard error of a $500 they had no had no effect and so we could spend the Job Corps money on other things by the way Job Corps is producing the gd’s I mentioned earlier but even even now I bet it’s not that good so you can redistribute the budget you should really sensibly understand that this this curve that I was showing you actually I think is real this is a curve which indicates how much priority should be where you should relatively spend more of your money relative to less so in that sense I would say that for guidance of where the policy comes on but when you get to the specific issue should it be this particular program or that I’m very hesitant to say that precisely because I think it’s we don’t really know I mean I think to me was a surprise that Perry was operating mainly through non cognitive traits and what I found amazingly this is something a recent discovery this is I was meeting with the people there’s a program that was written up in the New York Times a few months ago called tools of the mind they was teaching self control to children it turned out the tools of the mind there’s a famous Russian psychologist named by Akatsuki I don’t know if I’m pronouncing his name correctly but basically that was the inspiration for tools of the mind then I started realizing that tools in the mind had been implemented in some form or another in the peri program even though they didn’t use that language in the 60s they had implemented a version of tools of the mind a crude version so I actually sent the peri protocols to the tools of the mind people there was a Eureka moment they actually had an evaluation of their program about 40 years out but it’s those kinds of cuts oh so in other words instead of thinking of just vocabulary there are whole groups of people saying we must put in phonics or must do that I would say that that the missing gap and a lot of the thinking is understanding so I did talk to your former governor at length he gave him part of these slides and I said you know you already have to think about non cognitive traits and at least as he left this platform he seemed to be convinced we’ll see if he’s removed the next time you hear me may not have heard a thing but we’ll see but but the fact of the matter is that that’s the condos are the kinds of measures but to say it more precisely I worry about it and so and and to engage the parents so it seems like any successful program is going to engage the parents because the one thing we know about education is that successful schooling programs generally have parents helping so you know the Coleman report showed that most of the variability across schools much of the variability way was not caused by the schooling quality that Coleman measured it was actually caused by the parents that the parental environment so family factors not schooling factors matter so that’s that’s not going to help you in the immediate answer maybe some other people can Elizabeth you can you can give some advice you can I’m more than happy to listen to that but but I think but their worries always that policy because you never make

policies for full information and we never have and we never will but I on the other hand I think we don’t have to do repeat some of the mistakes of the war on poverty I don’t think we can think we can just do a scattershot program across all stages the lifecycle and think that’s going to work I think even within childhood we want to target much more of the early years than we do now and then within that a much greater emphasis on these social-emotional skills okay now yes every instinct I have makes me think that you’re right and the families paying a key role and in the production of cognitive and non-cognitive skills but but the time series evidence you know there’s been a quite a decline in crime and there’s there’s been quite a decline in teenage child bearing over the last ten years and at the same time there’s been a rise in the number of single families right there may have been a change in the quality of marriages with the sorting from changing divorce laws so I do you have any comment on the on the time time series evidence itself wouldn’t suggest I’m worried about that and actually a sealing that’s looking at exactly that question asking how much because you see that when you look at the change in the family life you have to recognize that there are multiple dimensions so families are now richer and they’re actually more educated and they where that’s all a good plus but at the same time families are worse in sense of these dimensions so that should contribute to negative outcomes but in the particular measure of crime I think we have to look at measures above and beyond just this question of cognitive and non-cognitive skills so I’m not saying that’s the full story and it clearly runs contrary to the kind of time-series evidence what you’re saying so in other words if I were to say does this story explain the RAI the decline in the crime rate in the last 15 years I’d say certainly abortion doesn’t but I also say that these non-cognitive skills and in the family environment seemed not to either so in that sense it looks like an improvement on the other hand there are a lot of other measures right we’ve heard some statistics if you look for example what seems to be a growing underclass so maybe these kids aren’t committing the same level of crime but you are finding higher levels so it’s kind of this group this is not just true in the US but the UK a greater fraction of kids who are born who are basically neither employed and are going to school in the teenage years that’s been increasing so maybe the severity of the violence but by many mentions you know the skinheads and the this to class society business actually is as a real business but I completely agree with what you’re saying and you can’t just take the time series it was run against if I were just to say take what I’ve said today and that’s gonna explain the decline in the in the crime rate no American families have probably gotten worse I could give you kind of a cheap story and say well see education levels went up and if I just took a look at one thing at a time and just look at your crime I’d say yeah well the education levels have gone up maybe that more than compensates for the adverse family for crime maybe it’s just that particular trait I don’t know the answer I don’t know the sharp answer but it’s a good question what’s your answer why is it no seriously what what’s the reason why the crime rate has declined that’s a huge mystery right teenage childbearing is it declining at the same to climbing to that sort of interesting Russ no no a lot of the but yeah no but we know that contraception sorting and so for there certainly seems to be but there are other dimensions by which you can say we’re getting to a more deviant society right and that we’re getting a growing underclass and more high-school dropouts and so forth so that’s a good that’s a really great question and I wish I had a better answer for it but I mean there was a book recently written I forgotten the name of this guy he was looking at the determinants of the decline in crime and you know basically said we still don’t have a good explanation for that and there isn’t one maybe you think of one anybody have one I’d be happy to surrender the phone and microphone and listen yes oh yes no look you’re talking about classroom productivity right away you’re talking a big burden on schools is things like special students categories you’re talking about remediation you’re talking about aggression you’re talking about a whole series of chains even in the public schools even the first few use outside of public schools to see one of the peri studies so one of the things studies we use was he looked at teacher ratings about aggressive behavior in the first and second grade you actually found that was declining dramatically between the treatment and control groups and for males and females both it was a real thing the amazing thing about these peri results is it was a very small sample you’re talking a sample sizes of 60 for each gender group 30 treatment 30 control you’re still finding strong significant results even after adjusting for pre testing so so I would say they do show up now the full benefit won’t show up for 40 years and so there was a simulation done recently the Brookings

Institution sahil and Dickens had a paper where they actually did the calculation and they were showing of course that the big problem with early childhood investments tended to be the money was spent early on and that and the big returns came later on and that’s true but if you think about a longer term steady state then you’re gonna get very high returns but yes there is so you will get some returns but the full returns will come later