share interactive transcript request transcript/captions live captions
download
|
MyPlaylist
DICK MILLER: My name is Dick Miller. I teach in the philosophy department and direct the program on Ethics in Public Life. Our speaker today is Claudia Goldin of the Harvard economics department. For a decade she has illuminated the leading questions about inequality. In this process she has known few limits. She has worked on gender inequality-- the topic of her lecture today. She has offered powerful explanations of the steep rise in income inequality that troubles people in the United States. She studied educational inequality and differences among the life trajectories of men and women going to American universities.
In this process she has also respected known limits of disciplinarity. She is an eminent economist and an eminent historian, and so far as I can see, she records those two disciplines as a seamless combination in her work. For these reasons alone she would be a perfect speaker for the program in Ethics in Public Life in the Cornell philosophy and the Cornell economics department to co-sponsor as part of the sesquicentennial celebrations of the College of Arts and Sciences. But, in fact, she is a super perfect speaker for that purpose because she received her BA from the College of Arts and Sciences here in 1967.
To introduce Claudia Goldin, Fran Blau of the Cornell department of economics will tell us more about her.
[APPLAUSE]
FRAN BLAU: Well, I'm absolutely delighted to welcome Claudia-- actually welcome Claudia back to Cornell. And, as Dick said, this lecture is part of our sesquicentennial celebration, and it is perfect to have Claudia as a graduate of Cornell. But also, one of the many things that I am proud of about Cornell is that, when we opened the doors 150 years ago it was as a co-ed institution. So the topic of this talk is quite relevant to Cornell.
Claudia Goldin is the Henry Lee professor of economics at Harvard and the director of the Development of the American Economy program at the National Bureau of Economic Research. After receiving her BA at Cornell, she got her PhD in economics from the University of Chicago where, among other luminaries, she studied with Robert Fogel, a Nobel laureate, in economics and also a graduate of Cornell.
Before joining the Harvard faculty in 1990-- oh, and let me mention, too, that among the many, many achievements that she's had, Claudia Goldin became the first tenured woman in the economics department at Harvard when she joined the faculty there. Before going to Harvard, she was on the faculty at the University of Wisconsin, Princeton, and the University of Pennsylvania.
At Harvard, in addition to her stellar scholarship, she is an award-winning teacher, and widely known as an outstanding mentor for graduate students. Claudia has received numerous awards. And since I know you want to hear her speak, I won't go through all of them, but let me hit a few high points.
She is the recipient of the prestigious Jacob Mincer Award for lifetime contribution to the field of labor economics from the Society of Labor Economists, and she also received the Carolyn Shaw Bell Award from the Committee on the Status of Women in the Economics Profession of the American Economic Association, which recognizes excellence in research and mentoring.
She just recently stepped down as president of the American Economic Association, and is a former president of the Economic History Association, as well as a former Vice President of the American Economic Association. She's been elected a member of the National Academy of Sciences and as a fellow of the Econometric Society, the American Academy of Arts and Sciences, the Society of Labor Economists, and a number of others.
As Dick said, Professor Goldin is an economic historian and a labor economist, and she has profoundly shaped both those areas of economics. One of the hallmarks of her research is that she doesn't hesitate to ask really big, really difficult questions. And then, miraculously-- it seems to some of us-- she answers them convincingly, with all the care, rigor, and precision that most economists bring to much smaller inquiries.
A second hallmark is that she marshals absolutely every weapon in the social scientist arsenal to do this. She collects data from musty, forgotten historical archives; she fields her own surveys; she uses large, publicly-available data sets. In gaining insight she's willing to and seeks out firsthand contemporary reports from historical individuals, but also estimates sophisticated econometric models on large data sets.
One theme that runs through her work is that she seeks to understand the present through the lens of the past as she probes the long-run historical roots of contemporary trends. And she's examined a host of topics, and I thank Dick for mentioning a number of them. Let me just break out a couple to give you more of a flavor.
I would say that Claudia is perhaps best known for her historical work on women in the US economy. Her 1990 book, Understanding the Gender Gap provides a sweeping analysis of women's economic status from the colonial period to today, and it's absolutely foundational for understanding gender issues. And this work is complemented by much other research, both historical and contemporary issues, relating to gender and could anchor topically an insightful study of the growth of women's college-going that sheds light on why women have not only caught up to men in higher education, but actually surpassed that.
As Dick mentioned, she's also widely known and respected for her contribution to our understanding of the dramatic growth in income inequality that has occurred in the United States, and actually in other countries as well in recent decades. Her recent book with Lawrence Katz on the Race Between Education and Technology has garnered considerable attention both within and beyond the academy. This book takes a long view of the relationship between education and technology in influencing income inequality, and sheds really profound light on the fundamental sources of our current situation.
So you're in for a treat today. Professor Goldin is not only a deep thinker, but she's a great speaker. As usual, she takes on a big question. While women have made dramatic gains in the labor market, full equality continues to be elusive. What is necessary to achieve full equality between men and women in the labor market or, as she puts it, write the last chapter in gender convergence?
[APPLAUSE]
CLAUDIA GOLDIN: Thanks very much, Fran. Just to make me a little bit more real, something that Fran left out-- because we know each other well, but she doesn't know me maybe that well-- I'm a dog trainer.
[CHUCKLES]
So that makes it a lot more down to earth. I think I'll put this over here. OK. So, I'm going to talk about what I call a grand gender convergence. It's the last chapter. And the converging economic roles of men and women are among the grandest advances of the last century, for sure. And each part of this gender convergence is what I call a figurative chapter-- there is one chapter right there-- in the history of gender roles.
But if, in fact, this is a history and it has chapters, it's got to have a last chapter. Right? So the question is-- so, the last chapter of this metaphorical book-- for it to be the last chapter, has to contain some degree, if not absolute, gender equality in the economy. So the question is, how will the last chapter be written?
So, this is a bit of a detective story. And the answer that I'm going to provide is that the last chapter can be written. And how it must be written may come as a surprise to you because, in fact, as I was doing this it came as a surprise to me. Always remember that if you go into a project knowing the answer, why are you doing it?
And what we do in research really is detective work. That's what makes it so exciting. That's why I became an academic. I became an academic because when I was at Cornell I saw the likes of Walter LaFeber and Fred Kahn who were two of my mentors here. And I said, I want to be them, because they do great detective work.
So the solution that I will tell you about does not have to involve government intervention, and it does not have to improve women's bargaining skills and their desire to compete. It doesn't have to improve their ability at math because, in fact, they're really good at math. And, in fact, men do not have to become more responsible in the home, although that wouldn't hurt.
But the answer is that it must involve changes in the labor market. In particular, how jobs are structured and remunerated to enhance what I'm going to call temporal flexibility. That's sort of the allocation of time over time. OK? And it must involve a decrease in the cost to firms of substituting the hours of one worker for another. This is not something that's going to be legislated away.
The gender gap in pay-- and when I say that, I mean per unit time-- would vanish if firms did not have an incentive to pay an employee who's working 80 hours a week more than two times what an employee who's working 40 hours a week earns. Similarly, with regard to working particular hours.
So such change has actually begun in various sectors such as in technology jobs, in science, and in parts of the health care sector. But change has been much slower in the corporate, in the financial, and in the legal worlds. Beneficial change has decreased costs of having workers substitute for each other, has increased teams of substitutes, and has enhanced the handing-off of clients. And I'll give you lots of different examples of these.
Does the last chapter has to have a lower cost of temporal flexibility? And that is an amenity than in general is valued more by women right now than by men. It must have what I call greater linearity of total earnings with respect to ours. OK? By linearity I simply mean that old algebraic notion that if you-- here it's if you work twice as many hours and you earn twice as much, that's linear. If you work twice as many hours and you earn three times as much, then that is nonlinear.
But I've gotten a bit ahead of myself. And I must first very briefly summarize, so that we're all on the same page, the previous figurative chapters in this book and the Grand Gender Convergence before I get to the last.
So, there has, over the last many decades, been a great narrowing between the human capital attributes of men and women. So, for example, labor force participation rates between men and women have greatly converged. Not equal, but greatly converged. Time out of the labor force has decreased for most women. Years of education of women have actually surpassed those of men. We see that certainly in college enrollment and college graduation rates. And college majors are far more equal between men and women, and in many professions women are actually in the majority.
Well, what about earnings differences, on which my talk will largely focus? Well, we also see that particularly in the 1980s and the 1990s there was great convergence, but not equality. Because of the grand convergence in human capital differences, what we called the residual gender earnings gap-- that's the gap narrowed in various ways, in various econometric ways in terms of people's human capital differences. The residual gender gap is now about equal to the raw difference, particularly for college graduates.
And what that means is that most differences in the attributes have actually been squeezed out. There was a time, some time ago-- several decades ago-- when many of us studying this subject looked at the numbers and we said, oh, there are lots of differences in education, in college majors. What are women majoring in? They're majoring in lots of fun stuff, but it's consumption not investment. You know? They stay out of the labor force longer. There are lots of differences. If we can squeeze these out, we will have squeezed out all of the differences in terms of earnings. That isn't the case.
So we need help. And so when I need help, I ask the great detective of all times. OK? And, by the way, I do not mean those silly shows on television. I mean the real thing. What I mean by the real thing is the books, you know?
[LAUGHTER]
And the stories. So the question that I pose to the great detective Holmes is, what will it take to eliminate gender differences in earnings? What must be in the last chapter for it to be the last? And Holmes says-- he says, ah. It's a capital mistake to theorize before one has data. Insensibly, one begins to twist facts to suit theories instead of theories to suit facts. Thanks very much. OK.
So therefore, I have to first provide data clues to the detective-- really good data clues. Then I have to step back from these data clues once I have them to say, what type of general framework or theory is consistent with these? And then I have to see if the evidence is consistent with the theory. Then I have to go back to the evidence. And that's the way Holmes always works.
So I'm going to demonstrate that the gender pay gap, given time worked, greatly increases with age or labor market experience. And it's mainly as women have children. I will also show that the gender pay gap differs enormously by occupation and by the sector that that occupation is in-- health, science, technology, finance, corporate, and so on. Although my evidence is for the US, the main facts and the findings can be extended to many other nations' labor force.
So first I'm going to consider pay gaps among college graduates born from 1958 to 1978. And the data for each of these cohorts and the cohort birth dates are given next to the lines. The data for each of these cohorts are coefficients on the interaction between female and age group from lots of cross-section regressions using the American Community Survey and the various US population censuses for 1970 to around 2010.
The points are connected to produce the relationship between age and the gender pay gap for what we call synthetic cohorts, because these aren't actual cohorts. I've connected the dots. So, first of all, the horizontal axis is age. The vertical axis is the log female over male earnings. Now, because most of you do not have the exponentiation chip in your brain, I have done that for you. And you can see that the ratio of the earnings of females to males is-- it begins with 1. I don't have the 1 there. 0.9, 0.82, 0.74. And that is merely taking the log and making it in to the ratio. OK? And each of the lines, as I said, is for a different birth cohort.
So note first that there is a decreasing pay gap across cohorts. Each of the lines is on top of the other so that the younger ones are highest. Most important to my story though, is that there is a greatly-increasing gap within cohorts for all of them. Now, this is just for a group of somewhat younger cohorts. If I increased the number of cohorts, the gap widens even further for the older cohorts and then actually turns up. Exactly why that's the case is another story.
Thus, the first clue is that the gender pay gap greatly widens with age. A second group of clues comes from estimating gender earnings gaps by occupation. But before I present this, I need to note and convince you that the relationship between the gender pay gap and occupations is mainly within occupation and not across occupation. And I'm doing this for the approximately 473-digit occupations in the census. These are a lot of occupations. Some of them are highly-aggregated, some not, but it's still a lot of occupations.
Now, for all women and for men, about 85% of the total gender pay gap is due to the gap within occupations, and only about 15% to the distribution of occupations by sex. For college graduates-- a little bit different-- about 70% of the gender pay gap is due to that within occupations and, somewhat larger, 30% to the distribution of occupations by sex.
Now, this may come as a surprise to many who think women are occupationally segregated into occupations that are very low paying. That is not wrong, but it's just that even if you gave men and women the same occupations, the difference within even these large number of occupations is so great that you would not wipe out that much. If we did this same calculation for 1980, you would get somewhat larger numbers in terms of across occupations rather than within occupations. So it was truer in the past than it is today.
So, now that we know how important the within-occupations gender pay gap is, we can actually focus on it. So what I'm going to do is I'm going to populate this empty figure with coefficients on female for the 473-digit occupations in the American Community Survey for about 2010, graphed against the average male income for each occupation. That is the horizontal axis is now earnings in the occupation. Every point here that I'm going to put down is an occupation. And, once again, this is in logs, just to give you a sense of what this is.
So 60,000-- the log of 60,000 is 11, so that line is being drawn at 11. And the very higher income at 12 would be 160,000. So it runs from about 25,000 to 120,000. OK. And the coefficients come from a log annual earnings regression for males and females who, at 25 to 64 years old, working full-time, full-year. That means 35 or more hours a week and 40 or more weeks per year. I include age in this regression. I'm trying to include as much as I can using the census. Age, [INAUDIBLE], race, education. Log of hours, log of weeks, occupation dummies.
Whenever I do this, it sounds like when people on the food channel-- and they give you a recipe. They say, and I add this and I add this, and you're trying to copy it down and you can't. You know? And then the cake comes out terrible. OK. But here you don't have to worry because we're not baking.
[CHUCKLES]
OK. And I'm giving this to you for all workers. If I did it for college graduates or young workers only, the results are pretty similar. So this is what it looks like. Ah. For all occupations, the result looks like, of course, a bad case of the measles or some other disease. But once I classify the occupations into various groups-- business, health, science, technology-- the categories-- it looks a lot more interesting.
In addition, I am going to look only at the occupations above the 60,000 mark. The categorization really can only be done for those in this group, many of whom are term professional service workers. And many of you here are professionals. So, you may not have known it, but you are a professional service workers, and others of you will probably become professional service workers. So this is all about you. OK? And this is it.
Note the business occupations are the red squares. So this is the same type of graph that I gave you before where the horizontal axis is the log of income in the occupation. Here I'm using male income. And the vertical axis is the gender gap. So the lower that number is, the greater the gap is.
So the red squares are much lower than the tech science ones, which are green and orange, and which, if we didn't have so many lights in the room, you might be able to see are green and orange. Do we need all these lights? Nobody is in charge of the lights. OK. Then it doesn't matter. Can anyone see that these are green and orange triangles? You can. Great. OK.
I once gave this talk and someone came up to me and said, which were the green ones and which were the orange ones? [LAUGHS] It doesn't matter. They're triangles and they're pretty similar. OK.
Now-- we can leave the lights the way they are, and then people can-- then I can see my notes much better, actually. Well, you can see this a lot clearer here. This is just the tech graph, and the science. So these are only the triangles.
And you can see that the tech occupations-- and these are the engineers, the information systems people, the programmers, all of those techie people. And remember, this isn't tech in tech industries. Don't think Google. Tech is all over. Tech is in the room. Right? You're tech. Yeah. He's tech. Tech is everywhere. There's more tech in manufacturing than there is in the techie stuff that you think is the tech sector.
So, these occupations have among the lowest gender gaps in-- remember from the previous one, and I'll show it to you again-- these are really the low ones. They may not have a ton of women, but they have the lowest gender gaps. And with the exception of the one outlier, is that dude in the pilot's uniform. And he's in a pilot's uniform because-- yeah. He's a pilot. OK? And that's a whole other story why gender gaps are very, very large in aviation. OK? A whole other story.
So gender gaps in health, which are the blue diamonds, are really spread out. And they depend, to some degree, on the extent of self-ownership. I'll go through one very interesting occupation, which is pharmacy, towards the end. Examples of the occupational names-- I know people look at this and they think, what are these, anyhow? And so I just put in some names over here so that you can just see where some of the occupations are. You can't really put in a lot of names with-- it's about 95 occupations there.
Another piece of evidence, therefore, to provide the detective, is the large difference in the residual gender pay gap across occupations. So the point here is that some occupations have really small residual gender gaps and some have, like, wacko large residual gender gaps. So what accounts for this difference?
So a really interesting hint as to what must be in the last chapter comes from freeing up this thing that I'm calling the coefficient on hours by occupation. So the regression that produced all these numbers-- it was a regression that produced all these numbers, because these are the interactions of occupation and female. That occupation did not allow hours-- remember, this is all full-time, year-round workers, so it didn't allow hours above 34 a week to affect earnings differently by occupation. It constrained that to be the same.
So when I allow hours to affect earnings differently by occupation, I see that the business occupations actually have the greatest coefficient on log hours and the tech occupations the smallest. Thus, the business occupations have the greatest elasticities of earnings with respect to weekly hours, and the tech and the science occupations the smallest. What that means is exactly what I said before. The business occupations and law, finance-- they have far more of this nonlinearity. You work more hours, you work crazy hours, and you get not a "crazy amount" more, you get "crazy amount squared" more. You get a lot more. OK?
So what does it look like? It looks like this. So on the horizontal axis is my computation of this elasticity of annual earnings with respect to weekly hours. On the vertical axis, once again, is the gender gap-- this coefficient on female times occupation. So I graphed the elasticities here against the gender pay gap from that original regression, and there is a clear negative association. Occupations with more nonlinear earnings with respect to hours are associated with larger gender gaps.
Now, this estimation results in a lower bound to this elasticity so, in fact, it's pretty amazing that many of these red squares are above 1. In fact, probably if I could do this where I didn't get this bias, they'd be all above 1, and the others would be sort of closer to the 1 range. So let me summarize the data clues. So this is my presentation to Holmes.
Number one, earnings gaps rise with age to some point. Number two, earnings gaps differ greatly by occupation and by sector. Number three, hours matter differently by occupation and sector. And the way of measuring hours here-- I'm measuring number of hours. I'm not measuring the fact that someone had to be there at 11 o'clock on Friday. OK? That would probably give me even stronger results.
We have three findings, so we're ready for some theorizing-- some really simple theorizing to help us. So the framework that I'm going to provide explores what happens when employees cannot hand off clients and patients and customers and even students in a costless fashion. The framework fits into a model that many labor economists use, called the model of compensating differentials, and it provides this simple framework, provides the foundations for the costs of providing an amenity. In this case, it's temporal flexibility.
So consider just about the simplest production process of all, the one that I've graphed over here. So consider this production process. This is completely linear here. On the horizontal axis is this lambda, which is time, number of hours, and on the vertical axis is someone's production.
So this person, if they work lambda max hours, then produces lambda max times that k, that slope. So that's their productivity. So consider a production process such that a formally-trained employee-- let's say the person has a JD-- they're a lawyer. They can go around practicing law-- and is in position 1, this position. And let's say it's a lawyer in some moderately-sized law firm. And the person has a value k1 per unit time, which is lambda. OK?
Now, if the individual is around for less than lambda star, the person's value declines. There's a penalty. And it declines to this k1 times 1 minus delta 1. That's like a tax. It's a penalty. And that means that if the person works at lambda max, the person can make a lot. But if the person works under lambda 1 star, the person's value drops down.
And that gives the person the following choice over here. That's a very stark choice. You work lots of hours, you make a lot. You work suddenly less, you're on this cliff and you fall down. Very stark way of putting it.
But let's say there's another position. And let's call that corporate counsel. And that exists-- remember, this lawyer could go to large law firms, small law firms, government, corporate counsel. And corporate counsel exists with a lower value per unit time. OK? So you can see that the corporate counsel, even if she worked tons of hours, would still make less than if she were at the law firm.
But there's also a lower penalty to not being around, giving the person the following choice. So notice that you make k2 if you work as corporate counsel per unit time, which is less than k1, but the penalty is less as well. Uh-huh.
Let's say there is another occupation that this person can work in. Let's say a lawyer in government-- let's call that the reservation occupation. And it has a linear output with respect to hours, giving us the green line. And under these conditions the individual is faced then with a nonlinear relationship between earnings and hours. One of these occupations is perfectly linear, but the person is choosing among them, and therefore the person gets-- uh-huh-- in total a nonlinear relationship between earnings and hours where the nonlinearity thus depends on the size of the penalty.
So the implications of this are the following. The framework suggests that nonlinearity arises when it is costly for employees to be away, when it is difficult to hand off clients, and when interdependent teams must coordinate schedules as in finance occupations, consultants, and lawyers. You know the problem-- if you have an 11 o'clock meeting with 10 other people, the problems of getting everyone together.
Linearity, on the other hand, arises when employees can substitute for each other in a somewhat costless fashion, when there are many independent team members, when information systems lower the costs of handing off clients and patients as has occurred in health and in pharmacy. So the more nonlinear, the lower the relative earnings of women, particularly those with children. And fewer of these women will eventually be able to achieve family and career. The more linear the occupations, the higher the relative earnings of women, and more will remain in the labor force, and more therefore will be able to achieve family and career.
So given the theory, what occupational characteristics should be related to these residual gender gaps? What is consistent with the theory that will help answer the question that I began with? And I'm going to approach this in two ways. The first one involves finding relevant characteristics on all of the three-digit industries that I've been looking at. And there are almost 100 of them in these graphs.
The source of this data is something called O Net, and it's the modern version of the Dictionary of Occupational Titles put out by the US Department of Labor. And the US Department of Labor has been putting this out, I think, since the 1940s. There may be one that preceded that. But O Net is the Dictionary of Occupational Titles on steroids. OK? And it's very interesting. O Net gives hundreds-- probably thousands of occupational characteristics, and it's done in a very involved way that would bore you to tears, so I won't even start.
Many of these characteristics concern things we're just not interested in in this particular example, such as strength and cognitive abilities. But many are interesting. And I selected characteristics from O Net head on things that would be relevant to what the model is telling me. They concern employee time pressure, whether employees need to be around, whether they need to have face time.
And I must say, I don't understand face time. When many of my students who take jobs at Goldman come back to see me and I ask them what they're doing, they tell me and I say, that sounds like face time. And they say yes, and I ask them what it is, and they can't tell me. It's like being there. So, also the need to have client contact. And also whether individuals work independently or on specific projects.
So I take these characteristics-- there are five of them here-- and I normalize them so that they all have zero mean and a standard deviation equal to 1, and then I add them up. And what I discover is that the tech and the science occupations score far lower than business and law. In fact, they're about one standard deviation below. Which means, according to O Net characteristics, that there is less need in the tech occupations to be around, less time pressure. Techies work more on specific projects. They work more independently relative to those in the business world and law.
Health is a mixed bag, and I'll talk more about that. Not surprisingly, therefore, there is an association between the average of the five O Net scores and the horizontal axis here, is the O Net scores that have been normalized-- so, that has a mean of 0 and standard deviation of 1. So there is an association between the average of these five O Net scores and the gender pay gap from the other work. And I took the one here for college graduates.
The second method that I'm going to use to understand differences in the gender pay gap is to explore individual occupations taken from the two tails of this distribution that I know a lot about. And it's going to be-- one of them will be the large gender gap group, and one is going to be a small gender gap group.
So, from the large gender gap group I'm going to pick business and finance, and also I'm going to pick law. And in each case, the data that I'm going to use are very good data. They are either longitudinal or retrospective, and they contain enormously rich information on productive characteristics. So these types of data are rare. So, what I was giving you before is information from data that doesn't have these sort of rich set of co-variates.
So the data from business comes from administrative records and a survey of the University of Chicago Booth School MBAs who graduated from 1990 to 2006, and it's data that I collected jointly with Marianne Bertrand, who did most of the collection-- she's at Booth-- and with Larry Katz.
And in that work we found that the gender pay gap greatly increases with time since MBA. This information is similar to what I was presenting before on how the gender pay gap increases. So on the horizontal axis is years since they received their MBA. On the vertical axis, once again, is the gender pay gap. So the smaller here would be a smaller gap. This is in log terms once again.
So what we found, as you can see, is that the gender pay gap greatly increases with time since MBA so that it is around 56 log points, at about 10 years to 15 years beyond the MBA. And that is a ratio of male to female earnings of 1.75. So that's big. And it's 45 log points, or the ratio is 1.57, correcting for what we call MBA performance. And that is that men take-- at Booth in particular-- far more finance courses. So there is a bunch of stuff wrapped up in MBA performance. And that would be the green line.
But the largest factors explaining the gender pay gap are hours and time out even though the difference in hours worked between men and women is not that large, and even though women working in this sector work very long hours. And it's also the case that time out for women is not that extensive.
So we've found that MBA women, when they have children-- and in the longer work we look at when all these things happen. You know? And so we have exactly when they have their first kid, for example, and their second and their third. And so we found that when MBA women have children they shift into lower-paying positions. Or, for some small group leave the labor force entirely to gain temporal flexibility. The finance and the corporate sectors heavily penalize lower hours, and flexible hours are quite rare.
So the material on lawyers-- I was fortunate to get access to the University of Michigan Law School Alumni Research data set, which contains extremely rich information on hours and earnings. The relationship between the gender earnings gap and time since JD is very similar to this picture here, so I'm not going to go through that. But instead I'm going to ask whether hourly earnings are nonlinear with respect to hours worked.
So here I've graphed annual earnings and constant dollars in four weekly hours bins 15 years after these individuals got their JD, and they're men and women who left University of Michigan Law School from 1982 to 1991. And earnings here are clearly nonlinear with regard to weekly hours. Those working more hours per week earn more on an hourly basis.
And this exercise stands up to many controls-- for years off, for years part-time, and also within each sex. The fraction female, not surprisingly, is much higher in the lower hour groups, and the fraction of women with children is also much higher for the lower hour groups. And that is given here.
Now, it's also the case that hourly fees are reported by many in this survey, and the hourly fee for those who reported also rises with weekly hours. So it's not surprising that lawyers know their hourly fee, because I recall that they even know their fee per 15 minutes.
[CHUCKLES]
So, earnings with respect to hours are strongly nonlinear among JDs. Let me move to my last example, and that is an example of a very high-income occupation. But, unlike in the previous cases, it has a very small gender pay gap and almost no penalty for low hours. And, as you can see here, this is for men. Pharmacists are a very high-income occupation. They're about eighth or ninth for men. For women they're about fourth or fifth. That's really high. OK?
But let me ask you a question. Is there anyone here who has never been to a pharmacy? OK. Very interesting. That must mean that you've all been to a pharmacy. Is there anyone here who knows their pharmacists by first name? And-- thank you. I do, too-- and insist, when you go to the pharmacy, that that person fill your prescription? OK. It's OK to know your pharmacist by first name. That's very kind. OK.
In fact, almost no one cares because we don't have our prescriptions compounded. There is standardization of medications. The information that your pharmacist sees is the same information that any pharmacist could see and interpret. In fact, you may think when you go to Walgreens that the people there are only seeing the prescriptions that you filled at Walgreens. That isn't true. They can see-- because they have the insurer data-- they can see the prescriptions you have filled everywhere for your entire lifetime, and the ones filled by your kids who were acting instead of you. You know? So anyone who falsifies who you are, they will see that as well. OK? So the point is that it's very easy for pharmacists to hand off clients.
Now, pharmacy has undergone enormous changes in the past many decades-- the past half-century. Ownership and the fraction working in independent practice, which is the blue line-- they have plummeted. I still go to an independent pharmacist in Cambridge. I bet many of you do not because they're hard to come by.
The fraction female, which is the red line, has risen quite a bit. And the ratio of female to male earnings, which is the green line, has increased a lot. So most pharmacists today are employees. They work for large firms and in hospitals. The spread of vast information systems; as I said, the standardization of drugs has enhanced their ability to hand off clients.
And the result is that short and irregular hours are not penalized. Pay is almost perfectly linear in hours. Those who work lower hours-- say, because of family-- are paid less, but it's linearly. Those who work more because they are managers are paid more, linearly. And part-time work is extensive, particularly for women, and there is almost no part-time wage penalty in pharmacy. And you can see that in the red circle. The coefficient on part-time from a regression of log hourly pay using an extensive data set that I got on pharmacists. And you can also see-- if you doubt that data set-- I also do it with CPS MORG data, which gives hourly wages.
So let me conclude. We now know what must be in the last chapter for it to be the last. It must contain considerable economic change. It can't just be a Band-Aid with firms offering flexibility at bargain prices. I'm always amazed that people applaud various firms that stand up, have the CEO be paraded at conferences, and the CEO says, oh, we are so family-friendly. And I always say, give me your data and I'll tell you if you really are. OK? And they don't.
[LAUGHTER]
And the worst offenders are the big accounting firms. And that's Deloitte and Ernst and Young. I met the CEO of Ernst and Young-- and if he's in the audience, I don't care-- and I asked him about their policies, which actually were put in place in the 1990s, so I couldn't really do an experiment with them. And I said, you know, no one at Ernst and Young would ever give me data on how women and men progress and time taken out and how these policies have really affected the earnings and promotions.
And he looked at me-- I kid you not-- he laughed. He said-- and I don't really like when people laugh at what I say, but still-- he laughed. And he said, you really think we have those data? And I didn't want to remind him what his firm was.
[LAUGHTER]
OK. So temporal flexibility must become less expensive, with more linearity of earnings with respect to hours. So a restructuring of jobs has happened organically in many occupations, particularly those in health care. Pharmacists, physicians in particular, hospitals and doctors have learned that they convinced their patients that there is a group that's going to take care of you, and it's going to be better than just your one.
I had a minor-league operation on my eye-- a procedure on my eye, and a couple of weeks later I got a note from my ophthalmologist congratulating me. And it was a form note-- congratulating me that I did such a good job taking care of my eyes, so I didn't have to see anyone. But if I did, that he has 20 people working with him, and isn't that great. And then at the bottom it essentially told me that he would never have to see me again. So they have this great way of handing off.
Optometrists, veterinarians-- certain physician specialties have lower hours, fewer on-call hours, and namely planned procedures. Many of the tech and the science occupations have built-in flexibility because the projects are often independently done, and the spread of information systems has led to change in many other sectors.
The previous metaphorical chapters of this grand gender convergence have concerned the relative increase in the human capital attributes of women-- education and job experience. But the last chapter is not going to be about changing productive attributes. The last chapter is going to concern the utilization or remuneration of those productive attributes, and it's going to be about how firms respond to changes in technology and the changing preferences in employees. Just like in a model of compensating differentials, we have technology and we have preferences interacting.
So-- but the most important thing is that the last chapter is not just about women. This isn't only a woman's problem, and it isn't a zero sum game. The changes that I've described can better just about everyone's life. And, thanks very much.
[APPLAUSE]
DICK MILLER: [INAUDIBLE]
CLAUDIA GOLDIN: Yeah. I can do one. OK. So, if there are any questions-- and what I would really appreciate are suggestions.
DICK MILLER: Would you stand up though and speak loudly so we can all hear?
CLAUDIA GOLDIN: Yes. Certainly so that I can hear. Yes.
AUDIENCE: So I wonder if there might be some sacrifice involved. I wonder if there are some cases in which there are inherent and perhaps unavoidable nonlinear relationships between productivity and hours worked by a single person and what we would do about that?
CLAUDIA GOLDIN: Right. Yeah. There's absolutely no question that I do not want the President of the United States-- we've had some who believed that this was a part-time job, and that there was linearity. But there just isn't. OK? And you're absolutely right. There are always going to be cases like that.
But we have long lives. And there are going to be times when we can put in-- we can go to those jobs. We could switch either within the firm or across the firms. But you're absolutely right. There are some things that are just not part-time. OK? And there's some things in which-- you know, I want my president to be on-call. You know? OK.
AUDIENCE: A very nice talk. Thank you very much. I was interested in the MBA data, because that was the one where you found the women were working-- so the pharmacy data was great. Right? That fits your story absolutely perfectly and you get it all in there. But the MBA data-- even though the women were working lots and lots of hours, you found that the difference in hours did not explain their--
CLAUDIA GOLDIN: No, it does. It explains a tremendous amount.
AUDIENCE: But there were still a lot of variance, wasn't there? Or did I miss that slide? Or misunderstand it?
CLAUDIA GOLDIN: No. That-- here we don't have good detail on ours because they were retrospective. For the lawyers we do. So for that we're just saying that the penalty for working lower hours is very great. And, of course, numbers of hours isn't the best indicator of which hours. So you may work 62 hours, but you have to be there during particular times.
You know, stock traders don't worked such long hours, but they have to be there from the start to the finish. So hours does go a pretty long distance. But what I was saying was that they're a difference in hours. It's not as if women-- female MBAs were working 25-hour weeks. OK?
AUDIENCE: [INAUDIBLE]
CLAUDIA GOLDIN: No, no, no. I said that they were not-- I said, the guys may be working in the 60s, they're working more in the 50s. So, they're not working very, very low hours. Sorry. Yeah.
AUDIENCE: So I different interpretation is that there is-- two different ideas. One is that, maybe a much more fine-grained look at occupation interactive with organization explains this. And so that corporate council where it's not the same job as a government lawyer, it's not the same job as a corporate lawyer, versus a [INAUDIBLE] or a small firm lawyer. And so that fine-grained occupational segregation still goes a pretty long way-- occupational/organizational segregation goes a long way to explain it. So that's one idea. And the--
CLAUDIA GOLDIN: Yeah. But the person has the choice. They come-- there's always going to be a division of occupation so fine that each one of us has a separate occupation. And that's going to be a problem. So it is a fine dance in terms of theory and empirics to find the right level of aggregation. But here, saying that you're coming to this with a degree that allows you to take any of these positions and then allows the individual to select among them.
AUDIENCE: Fair. But a second, I think, thing that's common in sociology is to think about this as a process of de-skilling, and that pharmacy-- pharmacists are an example of a profession that is being de-skilled.
CLAUDIA GOLDIN: OK. Then-- not exactly. In fact, the number of years that you have to go to school to become a pharmacist has increased over time. So pharmacists without a PharmD are not doing very well now. So, you know, it may appear to those of us to swallow little pills, that our pharmacist is doing nothing more than counting them. And, in fact, techs usually do that.
But they have to deal with far greater complexity in the drugs-- the number of drugs, the type of chemical compounds, the injectables, cancer treatments-- has grown enormously. And they're our front line. They're the person that we go to-- they're the only people open 24 hours, actually. You know, we depend on our pharmacists a lot. I think sometimes when I give this it's like, rah-rah, pharmacists, but-- yeah.
So, I understand that it looks like it's de-skilling. If it's so much de-skilling, why are they the eighth-highest paid occupation for men?
AUDIENCE: But maybe they won't be for long.
CLAUDIA GOLDIN: They have been for quite some time. Yeah.
AUDIENCE: [INAUDIBLE]
CLAUDIA GOLDIN: What's going to bring it down-- you know, there are regulations. And, you know, we have regulations about what physicians do as well. And if there are changes in laws concerning what physician assistants and nurse practitioners can do state-by-state, and dental hygienists, we'll also see changes there. It's not just pharmacists. And I don't think anyone here is saying that physicians are being de-skilled.
AUDIENCE: Just the last idea was the idea to remove a winner-take-all instead of [INAUDIBLE].
CLAUDIA GOLDIN: Certainly on this campus.
AUDIENCE: [INAUDIBLE]
CLAUDIA GOLDIN: I meant Bob [INAUDIBLE]. I didn't mean that there was--
AUDIENCE: [INAUDIBLE]
CLAUDIA GOLDIN: So, what was the other question?
AUDIENCE: Well, just that that is some of the-- is that a different way of explaining-- it's not occupational organizational segregation, but it's a winner-take-all counterpart.
CLAUDIA GOLDIN: That's a different way of-- a different type of model that I can push into somewhere. Yeah. Yeah.
AUDIENCE: This is really a continuation, I think, of the first question. You mentioned that you want the president of the United States to be on call. That seems right. But the logic of the job or, in any case, the [INAUDIBLE]. To push in that direction, a lot of other occupations, I would say, perhaps because of our reputational importance to institutions of higher education, breaking your [INAUDIBLE] to get tenure seems to be built into being a college professor at an elite institution. I guess I could see how face time and stable relations with clients and working with a fixed team is important for investment banking and corporate law.
I guess the only case in which this kind of being on-call and there attaches to a fairly low-income profession would be clergy men or women. I guess clergy has to be on-call. But basically there seems to be a high-income thing. You see the worry if one were to bow to the logic of the job when it is the logic of the job. Then it does seem that there's a new version of the glass ceiling. That women are in the fancy occupations-- OK. But because of the nonlinearity, they don't do at all as well in the fancy [INAUDIBLE].
CLAUDIA GOLDIN: Right. What I'm questioning, though, is-- it's not that there's a-- I think we're speaking the same language. You're using the word "logic" and I'm using the word "productive cost." And the question is whether these costs are actually high or whether they're not. And I can give you a lot of examples in which-- firms make trade-offs and individuals make trade-offs.
And so a firm making a trade-off-- and a firm might say, well, I would like to work my workers, my employees, very, very hard and I'm going to get more because I believe that this handing-off of clients to say, my clients really want this person around, so I'm going to demand that they be there sort of 24/7. And then they discover, ooh, that really costs a lot of money. I'm going to sort of change things.
So I got a note from Bank of America that-- once again, it was this very congratulatory note. Congratulations. Joe was your personal banker, but you are now very lucky because there's a team of seven personal bankers who are equal to Joe-- equal or better than Joe-- and they're around all the time to help you. So don't be offended if Joe isn't there, because any one of them can help you just as well.
So that was a company that realized that it was costing them too much to have workers that were on call. And that's really the way things are going to change. If firms say, this is costing me too much-- and why is it costing them too much? Because the individual says, I really care about my time. OK? And the more individuals who say that-- men, in particular-- the better off.
Now, someone who was my student long ago wrote a book called Lean In. And as much as I adore her and everything she writes is interesting, I wish to hell she had written a book called Lean Out for men. Because the more man lean out, the more these costs are going to be more apparent to these firms and, quote, the "logic," which I call the costs, are going to be higher. Yeah?
AUDIENCE: This is somewhat of a follow-up. It sounds like you've identified something [INAUDIBLE] equilibrium where people are more interchangeable and [INAUDIBLE] hours [INAUDIBLE] firm. So what do you think are the constraints that are keeping us from getting to that equilibrium?
DICK MILLER: I think everything that I just said. You know? If, in fact, it is a true cost, then-- if, in fact-- and one of our students wrote a paper-- I don't want to say who it is, because Larry said to me when I tell them, wow, this is such a great finding, he said, if you knew what went into that sausage, you would not take that seriously.
[CHUCKLES]
But it was the following finding, that he was using a group-- I don't know how he got this data. But he was using a group of accountants who were doing taxes for firms, and they then dropped dead. OK? And then someone took over. And he discovered that there was a-- so that's a very interesting experiment. You know? You don't have to kill them off. They just drop dead. And then you see how much more the firms-- or less-- the firms paid. Because that's exactly the cost-- the immediate cost-- of handing off. OK?
So there may be real costs of this. Or it may be that using information technology does a tremendous amount to reduce these costs. There are some cases in which there-- I think of these as sort of two cases. One in which technology simply changes, as in the case of pharmacy, and one in which there is a change in preferences.
So this, once again, is the compensating differential model. You either have a change in cost or a change in preferences. OK? So the change in preferences is the case of pediatrics. So pediatricians, not surprisingly, like children. OK? And because they like children, they actually like their own children a lot, and they value their time with their own children.
So, one-by-one and en masse, they changed the face of pediatrics so that 35%-- the last numbers I looked at, which were for 2006-- 35% of all female pediatricians worked under 35 hours a week. And for men it was about 10%. And this was all increasing over time. That has to have happened because places like Kaiser and others changed their way of dealing with their doctors, and they have teams of doctors that can deal with these kids.
Another interesting case is-- I was looking at data on sub-specialties in medicine, and I noticed that women have generally eschewed the surgical specialties, with the exception of OBGYN. And, for one thing, it's a much longer residency, fellowship period. OK? And another thing is sort of-- the claim is, the culture of the orthopedic surgeons and the surgeon macho culture.
But I noticed that women were going disproportionately into a type of surgery, which is rectal surgery. You know, I know pediatricians love children, but it wasn't clear to me what was going on. And they were also going into other fields. And then I realized that they were fields in which there were planned procedures that-- in small places, if you're the only surgeon doing this, you also do colonoscopies. And colonoscopies are now required, you know, x-number of years of everyone over some amount. That's a lot of colonoscopies being done. These are planned procedures.
So there are lots of different things that are moving us along. Some by large numbers of people and some almost organically by technology, and some sort of in-between by firms deciding, well, it's just too costly to do this. I'm going to sort of explore over here in terms of new technologies. I don't have a magic wand for this. Yeah?
AUDIENCE: But I guess some of this could have the outcome of segregation. So women could crowd in areas where this is more feasible, and then it might-- well, it has to happen in pharmacy in terms of percent female, not in terms of lowering pay.
CLAUDIA GOLDIN: Well, it certainly has stabilized. The percent female-- and I'll give you an example of where it has gone-- there are some examples. But percent female in pharmacy schools is about 60%. So, it's been that way for a long time. So it's about 60%.
But you're absolutely right. You know, is this the new mommy track? But in many of these occupations, if you can create linearity-- I mean, women will earn less for a while, but not on a time-adjusted basis. And, of course, we have long lifetimes. So keeping your foot in the door-- you know, rather than leaving the labor force, keeping your foot in the door is really the best way to have a long and productive time in the labor force. Yeah?
AUDIENCE: It seems to me that it's going to be really hard to get the best, most talented people not to make themselves replaceable, in a way. So, in other words, people who are most expert in their field or most specialized aren't going to want to be able to be not on-call or able to sort of clock out. And yeah, I sort of worry about the occupational segregation at the top. How are you going to get women to be more likely to go into academic neurosurgery, for example? Because that's not a very easy thing to clock in and clock out of.
CLAUDIA GOLDIN: Well, it depends. I'm not quite certain if it's that hard to clock in and out of. It depends upon how the procedures are planned and how the procedures are organized. Take anesthesiology. Do you think of anesthesiology as a high-status occupation? Is that one of the-- No? Yes?
AUDIENCE: Well, I mean, it depends. Relative to some jobs, yes. In medicine not as much. I mean, compared to surgeons, no.
CLAUDIA GOLDIN: OK. Take the anesthesiologist away, and that patient just died. [LAUGHTER]
AUDIENCE: Well, that's what I mean. It depends. I mean, if you're asking a bunch of people in med school-- I mean, it sort of depends.
CLAUDIA GOLDIN: So-- OK. So we have a group of surgeons who are the highest, and then we have this whole other group. So you're saying that using this sort of metric denies women the ability to be surgeons.
AUDIENCE: I'm saying it denies women the ability to be sort of the most expert specialist person that is needed when there's a [INAUDIBLE].
CLAUDIA GOLDIN: No, no, no. This surgeon is the only expert specialist person?
AUDIENCE: Well, maybe in an area. Or, I mean--
CLAUDIA GOLDIN: Yeah, but that doesn't mean that-- you know, for the individual to be able to-- first of all, there are-- I don't know how many thousands of surgeons there are in the United States today, but my guess is that in any great research hospital there are a number of surgeons who look at each other and say, you know, we're all in the same band of great surgeons.
AUDIENCE: But they don't all do the same thing.
CLAUDIA GOLDIN: They don't all do the same thing?
AUDIENCE: If you have a very specialized problem.
CLAUDIA GOLDIN: You know, I absolutely agree with you that there are going-- it's just like the president of the United States. But let me talk about anesthesiology, because anesthesiology really is up there. And anesthesiologists-- I don't know how many of you have had surgery in the past couple of years or have any recollection if you did, but you probably did not-- you saw your surgeon probably for about 15 minutes before you had surgery. Weeks before, because you had to figure out whether you actually needed surgery. You had a heart-to-heart conversation with your surgeon.
You saw that person who's actually keeping you alive for, what, 10 minutes before you went into surgery. Because that is a team. And the surgeon has a group of anesthesiologists-- maybe three, four, five, whatever the group is, and one who is free. So within that group-- now, that group was chosen because that surgeon believed that that is the group he or she trusts. OK? So that's a team of pretty good substitutes. They're all excellent, but they're pretty good substitutes.
Just saying that-- do you have to be absolutely unique so that you're one in a billion to be great and high-status? I don't think so. OK? I agree with you. There are some people-- the husband of my primary care physician is the greatest surgeon on the parotid gland. And my guess is that if I had a problem with my parotid gland, he's the only one I could go to. So, you're right. There are some people like that. But I think that that's sort of rare, and I don't think that's as big an issue as you might think it is. Yeah. Are you a surgeon?
[CHUCKLES]
Yeah?
AUDIENCE: I'm just wondering, don't you think there's going to be push-back from clients and patients that maybe want that personal [INAUDIBLE]?
CLAUDIA GOLDIN: Sure. And that's exactly right. You know, don't get me wrong. I am not standing up here and telling you what should be. I'm telling you what would happen if it did happen. OK? I'm not a normative economist. I'm a positive economist. OK? And the degree to which there is push-back is precisely why these things exist.
There are tons of people who think that their tax preparer is the only person who could prepare their taxes. OK? And I'm on record for saying I can do their taxes just as well, because I use the same program. OK?
[LAUGHTER]
And I know the code just as well as their tax preparer does, because this doesn't really take that much. But their tax preparer has sort of convinced them that their tax preparer sort of knows you well, just like their clergyman has convinced them that he or she is that conduit from God to them. OK? And if that's the case, then they're irreplaceable. QED.
[LAUGHTER]
Yeah?
AUDIENCE: A number of us have been asking you what [INAUDIBLE] to be spontaneous processes in the labor market. Why don't you just turn this around to a more constructive question. What role do you think there is in ending objectionable gender inequalities now for government? And also for public opinion-- public opinion if you tried to stir up in this room by getting us not to think so well of the nature of auditing firms. How can we supplement spontaneity, or is this one of these cases in which maybe that was a good idea in the past, but there's nothing to be done now?
CLAUDIA GOLDIN: Well, I think I defer to people who study laws concerning leave policies and paternity leave and the changing of social norms in that regard. And I see the millennial generation-- and, if you go out to places like Silicon Valley, you see a generation of techie people with lots of skills and a tremendous desire, if it's a nice day, to go bike riding at 2 o'clock and to come back and do something.
The idea there is that it's work-life balance, not career-family balance. And the more people who see that, the better. I'm probably not a good example of this.
[LAUGHTER]
Yeah?
AUDIENCE: So, in your talk it seems like if you're a woman, science and technology is really where you want to be. So why do you think there is still so few women in science and technology jobs?
CLAUDIA GOLDIN: Well, that's-- first of all, in science jobs, there are a lot. I mean, biology as a major-- all the different biological sciences are way over 50% female in universities. And engineering and CS is increasing. Economics is not, and it's something I'm working on, and I would like to know myself. OK?
So there are lots of people who would talk about the culture of computer science and techie jobs, but I'm not-- I don't know. What do you think?
AUDIENCE: [INAUDIBLE]
CLAUDIA GOLDIN: Well, it'd be nice to ask people who came to college-- women who came to college and thought that they were going to be engineers and then-- see, one very interesting thing about physicians-- and it may have to do with the techie jobs, too, is that women tend to go in to sub-specialties where they actually deal with patients.
So, there's a curiosity in medicine which is that radiology would seem to be exactly the field that women would like-- exactly the field. OK? And yet, they have not been-- and in the beginning they did go into radiology. In fact, Jacob Mintz's wife was a radiologist. And it's really been going nowhere-- women in radiology.
And the reason is that you don't have patients when you're a radiologist. You're not dealing person-to-person. And in many of the techie jobs you're also not dealing-- so there are a lot of women in the tech sector. There are a lot of women in the tech sector, and they're in marketing and they're in management, and they're dealing with people. And then there are the coders, and they're off somewhere on the spectrum. And women tend to want to be more dealing with people. And you see it very clearly in the selection of specialties in medicine. But it's a great question, and it's something that we're going to have to think very hard about.
DICK MILLER: Let's thank Claudia.
[APPLAUSE]
Harvard economist Claudia Goldin '67 laid out ways to end the gender disparity in wages between women and men Oct. 23, 2014, in a Sesquicentennial talk sponsored by the Ethics and Public Life Program and the Department of Economics.
