Intro. [Recording date: September 7, 2017.]
Russ Roberts: Our topic for today is Gabriel Zucman's recent paper with Emmanuel Saez and Thomas Piketty, "Distributional National Accounts: Methods and Estimates for the United States", which has received a great deal of attention in the popular press and among economists.... So, it's a very provocative and complicated paper. We're going to try to do two things today among a few others, but we're going to try to give listeners an idea of what the paper finds, and then how such a finding is actually constructed using data. So, some of the background of how this kind of empirical work is done. So, let's start with some of the key findings of the paper. In a recent treatment for the average person, not for an economists, which you wrote up--we'll link to that write-up--you suggest there are three key findings. What are they?
Gabriel Zucman: So, what we are trying to do is research to impute income growth for each group of the population with a way that's consistent with macroeconomic growth. And so, there are indeed three big findings. So, the Number 1 finding is that you had some growth in the United States since 1980--that is, the average income per adult has increased about 60% since 1980--
Russ Roberts: Corrected for inflation.
Gabriel Zucman: Corrected for inflation--so it's in real terms. Exactly--
Russ Roberts: And including--I just want to make one--this is really important: This isn't just earnings. This is all forms of income: taxes, transfers, right?
Gabriel Zucman: This is all income as recorded in the National Accounts. So, that is the headline figure that we hear when we talk about GDP [Gross National Product] growth--about the macroeconomic growth rate of the country. So, if you compute macroeconomic growth per adult, what you get is this number of 61% increase in total income since 1980. But the Number 1 finding is that income growth has been very unequal. So, if you look at income growth for the bottom 50% of the distribution, before taxes and transfers--so, before any form of government intervention--you've had zero growth for the bottom 50%. So, for half of the population, they've been completely shut out of economic growth. Their income in real terms has not increased. The second finding we have is that, you know, by contrast, at the top of the distribution there has been a lot of growth. [?] at the top 1%, their income has been multiplied by 3 in real terms since 1980. To give a sense of how heterogeneous growth has been in the United States, so, we can express these numbers in terms of real growth rates per year. So, when I say that average income has increased 60% since 1980, that's equivalent to an annual real growth rate of 1.4%. So, that's the average in the United States. What we find is that, for the vast, vast majority of the population, the growth that they've experienced has been much less than 1.4% per year. In fact, for the bottom 88 percentiles, or for 88% of the population, income has grown less than 1.4% per year. So, only for the top 12% that income has grown at above 1.4%. And among that group, it's really only in the top 1% and top 0.1 and top 0.001% that you see high growth rates of 3, 4, 5% per year. And so, that's important, because some people have a view that what is happening in the United States is that, let's say, the top quintile--so, the top 20%--has pulled off from the rest of the economy. And that's not really what we find. We find that rising inequality in the United States is really very much a story about the top 1%. So, another finding we have is: How do these reserves[research?] change when you take into account taxes and transfers? So, the effect of government intervention. When [?] of government estimates of inequality in the United States is that they don't take properly into account of the role of taxes and transfers; and all the numbers that I've mentioned up until now were about pre-tax- and transfer-income. Now, after taxes and transfers, and when you take into account all forms of taxes--at the Federal level, at the state and local level--and all forms of transfers--whether monetary transfers or in-kind transfers or public goods spending--what we find is that government distribution has made growth slightly more equitable. But only slightly so. That is, if you get back to this bottom 50% of the distribution that has experienced zero growth pre-tax and transfer, after taxes and transfers its income has grown a little bit. It has increased by about 20% since 1980. That is still much, much less than the average growth of 60%, and of course much less than growth at the top of the distribution. So, our conclusion here is that, despite important changes in government transfers and the expansion of certain important programs like Medicare and Medicaid, that has not been enough to significantly lift the income of working class Americans--so the bottom 50%. And, the last finding that we have, which I think is [?] very interesting, is that what we are currently trying to do now is compute similar statistics in other countries. And we've done that in France. And we can compare, let's say, the income growth rates for the bottom 50% in France and in the United States. And the trajectory has been very different. In 1980, the bottom 50% income earners used to be about 10% richer in the United States than France. But, what has happened is that in France, since 1980, bottom 50% incomes have continued to grow at roughly the same rate as macroeconomic growth in France, when the United States has completely stagnated. As a result, now, the bottom 50% in France is significantly richer--has more income--than in the United States. And that is before taxes and transfers. And that, what makes this reason[research?] particularly spectacular--I'm not talking about the generous welfare--French welfare state. That's not what's driving our results. Before tax and transfers. So looking just at market income, the bottom half of the distribution, half of the population now does better in France than in the United States. And that's our last main finding.
Russ Roberts: So, listeners will not be surprised that I'm skeptical of these findings. Right? And--I have to say--I find them hard to believe. So, I'm going to give you a chance to convince me, over the next 50 or so minutes. But let's start with how you constructed these. So, to give listeners a little bit of background: How would you begin to use the National Income Accounts? Let me say it differently. You might--you might, listeners, you might understand--you might hear that capital's share has gone up--or down--compared to labor. Or, 'Men and women have had different wages and growth rates of wages' over a particular time period. But, Gabriel, what you are I think trying to do here is say, 'When GDP grew 3.2% in such-and-such a year, how much of that went to the bottom quintile--the 20% who were the poorest in the United States? How much went to the top 1%?' That's a very hard task. So, how would you begin? How do you begin to make the assumptions that you need to do to apportion those gains to different strata of the income distribution? And let me know if that's the right question.
Gabriel Zucman: Yeah. That's a great question. That's exactly what we're trying to do. What we're trying to do is to bridge the gap between the study of macroeconomic growth, where people use National Accounts data, and the study of inequality, where people use tax data and survey data to study the distribution of income or the distribution of wealth. The problem is that there is a large gap between the total income that you see in the survey and tax data, and total macroeconomic income--total GDP or National Income. And that's a problem, because then it makes it hard to decompose macroeconomic growth by income [?]--to have just growth statistics for each fractile of the income distribution, for each growth of the income distribution that add up to the headline GDP growth number. So, the way that we try to bridge this gap is by combining National Accounts tax data and survey data. So, they all have strengths and weaknesses. But if you combine them--and you can try to approximate the distribution of Total National Income, as we call it in the National Accounts. And, I want to stress that this, what we've done, in my view, is very much a prototype. So, there are uncertainties. Could be improved in many ways, as more data become available, as methods are improved. But I think the objective is worthwhile. That is, it's important to try to not only measure growth but be able to say, 'Here is how growth looks like for people like you.' So, our starting point is: National Income, as recorded in National Accounts. So, what is National Income? That's GDP, Gross Domestic Product, minus capital depreciation--which is not an income for anybody. Plus the net income that the United States receives from abroad--so typically dividends and interest that the United States receives from foreign countries, minus and interest like based in foreign countries. So, it's very close to GDP. And that's our starting point. And then, what we try to do is to allocate this big total to which group of the population. So, some from the income of, are well-captured in tax data. Typically, the dividends and interest income that wealthy individuals earn--tax data of a critical source of information to capture that, because all rich people have to submit a tax return. And, that's how we know how that form of income is distributed. Wage income, as well: Tax income does a very good job there. For other forms of income, you need to look at survey data. Typically, transfers, a lot of transfers, are not taxable and so don't have to be reported on tax returns; and so you need to look at survey data. You need to do [?] CPS[?] [Child Protective Services? More probably in this context: CPS=Current Population Survey--Econlib Ed.] in the United States, the current population Survey. And you have a third category of income that you see, neither see in tax data nor in survey data. Things like, corporate retained earnings, for instance, which have increased a lot since the Great Recession. Then you need to impute them. So, what we do is, in that research is pretty technical. We try to explain, for each income category, how we capture those income categories. Tax data[?], so there is all imputation. We try to be very explicit about the tax [?] imputations[?] that we are making, when there is no readily available source of information. And we try to investigate what happens when we impute income in different ways. And broadly speaking, the research don't change much for the simple reason that the rising inequality in the United States--the trends are so massive that even small deviations--you know, changing imputations from some capital reserve income doesn't affect these big trends, which was already very visible in survey data or [?] data. So, that's a broad methodology.
Russ Roberts: So, let's start with--I am fascinated by the claim that the bottom 50% has had no gains in economic benefit--certainly from market-based economic benefit. So, again, we might want to distinguish between inequality or gains from market-based results. Which could be your investments; It could be your salary, or wages, your fringe benefits. And then there's a second thing you would look at, which would be: Well, what about your taxes? And what about transfers? Government benefits? You might be eligible for and receive unemployment insurance, food stamps, Social Security, etc.? So, to me the most dramatic claim, and the one that I'm most skeptical of, is the idea that the bottom 50% had zero pre-tax and transfer. That is--ignore tax and transfer for a minute. Just look at the outcomes from what people earned and invested and gained, or lost. And that is stagnant. For a 34-year period during which, economic growth, as you say, was quite substantial: 60% per adult, over the--an increase of 60% per adult. It's saying that the bottom 50% got none of that. So that's a very dramatic claim. So, my first question is: What does--and just a clarifying question: Does the additional of the National Income approach or the National Income accounts data affect that very much? Because that bottom 50%--they are not getting a big part of corporate-retained earnings, I assume, or other things that might show up in the National Accounts that aren't in taxes or survey data? America I right in that?
Gabriel Zucman: Yeah. You are absolutely right. I mean, the bottom 50%, it's actually relatively easy to observe their income, because the vast, vast majority of their--all of their income, pre-tax and transfer, it's labor income. It's wages and salaries and second-employment[?] income. And that income is typically very well-recorded by the tax data[?]. And the survey data. So, this trend, that there has been zero growth in pre-tax income for about half of the distribution, you could see it already pretty well in tax and survey data. Where tax and survey data have limitations, important limitations, is not really for wage income or labor income, probably speaking. It's for capital income--
Russ Roberts: right--
Gabriel Zucman: Because, if you look at the flow of macroeconomic capital income--so the capital share that macroeconomists study and they find it's relatively big, about 25, 30% of National Income and they find it's [?] in the United States and in many other countries. About two-thirds of this macroeconomic flow of capital income, you don't see it in tax and survey data. In tax[?] they tell you, only see one third of total capital income because the majority of economic capital income is actually tax-exempt. That includes corporate retained earnings, imputed rents for home owners, a number of taxes like the corporate tax and property taxes. And a big flow of dividends and interest that's paid to pension funds. All this, big flow of capital income, it's part of economic capital income, but you don't see it in tax data. Now, capital income tends to be more concentrated than labor income. Which means that actually with tax data, ordinarily you can't do too good a job at studying the rich. That's kind of--you know, a paradox, because people started using taxes--they started studying the rich. But if you think more about it, there is a [?] two thirds of capital income that all go[?] it goes to the top. So, our imputations are going to play some role for the dynamic of income at the top and for the composition of income at the top. But, for the bottom 50%, they are not playing any significant role.
Russ Roberts: So, let's start with that. Let's just start with the bottom 50%. So, I want to start by saying something that I think sounds like it's impossible. But, I think it's a very important truth. And I just want to get your reaction to it. When you say that the bottom 50% have no gains in income between 1980 and 2014, that does not mean, that does not rule out the fact that, if you went and looked at people in 1980 and you followed them throughout time--in particular, if you took 25-year-olds in 1980 and followed them till they were 60, 59, if I got the years right, in 2014--you are not saying that none of those people had any growth. Because obviously millions of them did. Millions of them are better off in 2014 than they were in 1980. Right?
Gabriel Zucman: That's correct. So, the approach that we have in this research is, let's say, a cross-sectional approach, where this means very simply is that we look at the distribution of income, year after year. We are not trying to photo[?] people over time. Which would be very interesting and important and hard to do. But in this research is very much a first step. We don't follow people over time. Now, is it likely that if you are able to follow people over time you would find a significant amount of growth for the bottom 50%? I'm a bit skeptical. For the following reason. What we tried to do, what we are doing as research is we compute income distributions by age groups, so we can look at the bottom 50% of income earners, age 20-30. 30-40. 40-50. And, when you do that, you see, basically, no growth for the bottom 50% within each age group. And so that suggests that even if, of course, you know, over the life cycle, people's income and wages change, there has been this huge stagnation of wages for working class America at all age levels. Such that, it's unlikely that, even if you were able to follow people over time you would see a lot of income growth. And, you know, more generally, there is no indication that there is a lot of mobility in or out of the bottom 50%. There is no indication that people in the bottom 50% of [?], to move, say, to the top 10% or that they move to the top 1%. The Social Security data that are used to study these questions where you can follow people over time are very clear on that. There is a lot actually of persistence in income over time. So, if you are in the bottom 50% of the distribution at some point, you are still very likely to be in the bottom 50% next year or two years after. But, you know, we need more research on that question. And it's certainly a next step for research.
Russ Roberts: The reason it's important, more generally, is because: If, to take an example, the United States had a lot of low-skill immigrants come into the country between 1980 and 2014--which, there's a decent number of. Obviously it's a bunch. But there's also some high-skilled immigrants who came, which complicates it. But, if you had low-skilled-only immigrants who came, then what you would observe is--you might observe that the average wage level in the United States could go down, but every person who was already here before the immigration are better off. And so, the average growth rate for the bottom 50% in that situation could be very misleading about the state of the economy. Could be just a composition effect.
Gabriel Zucman: Right.
Russ Roberts: So, that's one issue. Now, when people do follow the same people over time, they do find very different levels of growth. And in fact, I mean, it's shocking. When I talk about these numbers, people always assume they can't be true, because we all know so many things about the income distribution. But I don't understand how these numbers can be so different. So, for example, Gerald Auten has done work with colleagues--was in the National Tax Journal--I'm just going to quote these numbers, because they are so striking. He looked at people aged 35-40 in 1987--so he's going to look at a 20-year period, 1987-2007. It doesn't exactly overlap with 1980-2014. But, it's a 20-year piece of your time period. He finds--they find--that, shockingly, the poorest people had the biggest gains in income. The lowest quintile over that 20-year period--that is, you started, you are 35 years old in 1987, 20 years later, the people in the bottom quintile had doubled their income. They had 100% growth. That's the change in median within that quintile. For the next quintile, it's 42%. For the next, it's 27%. So, the three lowest--the median income within the three, the bottom 60%, went up by 27, 42, and 100%. The 4th highest quintile only went up went up 11, and a 5th only went up 5%. So the largest gains went to the poorest people. And that's totally different than the standard finding that people claim about that the average person, the median person, the bottom 50% are making no progress. What do you think is different about their results relative to yours?
Gabriel Zucman: I think there are many differences. So, one thing is: What's the income that you are looking at? Are you looking at pre-tax income? Are you including some of many forms of government transfers? What we are trying to do in our research is to add a really clear distinction between what is income before any form of government intervention--what we call pre-tax income, and income after you include after you include all forms of government taxes and transfers. Another difference is: How do you deal with changes in household size? Changes in marriage rates and divorce rates? A lot of the analysis in inequality, you know, is conducted at the household level, which, you know, can be a problem if household size changes a lot. Which has happened in the United States. So, what we try to do in our research is to address that issue by looking at income per adult, where we split income, the income of married couples, 50-50 between each spouse. So, to have at least consistent unit of observation over time that's not affected by changes in household size. And most importantly, what we've tried to do is to have members that add up to total GDP growth and to total National Income. And so, if you believe that income has grown a lot at the bottom of the distribution, it has to be the case that it has grown a lot less at the top of the distribution than what we estimate. That's contradicted by a huge and overwhelmingly large set of evidence from [tanzlita?] in particular. So, again, I think more work needs to be done to better study changes in income when you follow people over time. But, reading that I do after most recent studies that follow the entire population of working age Americans using Social Security data, even when you do that, when you follow people over time, you see a huge increase in lifetime inequality. And the point about immigration--I mean, it might play a role. I'm not sure, though, that it explains a lot of what is going on. Because, look at Europe. Before the Great Recession, the Crisis of 2008 and 2009, there was actually more immigration in the EU [European Union] than in the United States. That has changed since the Great Recession. But if you look at the 2000-2009 period, that's, that's, that's true. And, 11 developed[?] countries, that includes France, Germany, Scandinavian countries, where immigration was much higher than in the United States. And yet, we don't see the type of stagnation involving 50% incomes that we observe in the United States. Getting back to the case of France, we see that the bottom 50% has been growing at roughly the same percent as the same economy. Looking at Scandinavian countries, is even more spectacular. There is much more inequality--there is much more equality there, in Scandinavian countries as in the United States. And growth has been much more equitably distributed than in the United States. Despite higher immigration as than in the United States. So, I think what's more important to--as an explanatory factor for understanding what is happening in the United States is not low-skill immigration, but it's a number of changes to policy. The decline in the real Federal minimum wage since the late 1960s or 1970s. The decline in the role of unions. More broadly speaking, the decline in the bargaining power of labor. Very unequal access to higher education. All of these things, I think, have contributed to the stagnation of the bottom 50% income in the United States.
Russ Roberts: Okay, we're going to come back to those because those are really interesting and provocative. I disagree with three of them; but I agree with one of them. So, that's pretty good. I just want to make a clarifying point on immigration and then I want to ask you about the family structure issue. So, I agree with you on immigration: I don't think that it's important. I just gave that as an example of how misleading it could be when you look at different snapshots over time because they are not the same people. I think it's very interesting that all--not true any more--but, for a while all of the studies that followed the same people over time showed large gains for the poorest people. That's true in the Panel Study of Income Dynamics [PSID] that people have analyzed, which follows people over time since 1970s.
Gabriel Zucman: That may be true in some studies that use survey data, and PSID, but that's not true in the studies that use the population-wide Social Security data.
Russ Roberts: The recent ones that came out. But the Auten study that I mentioned--we'll put a link up to it--the Auten study that I mentioned is actually using tax returns and it's quite exhaustive; it's an enormous sample of the entire universe of tax returns that showed the largest gains for the bottom. And not just the largest gains--quite large, dramatically larger gains. But the question I want to focus--I want to [?] for a minute before we get to the causation and speculation about what are the explanations for this: Let's talk about households. And you make a very good point, which, because it drives me crazy: There's been a huge change in household structure in the United States over the last 35, 40 years. It goes back to the 1970s when the divorce rate in the United States started to rise very, very rapidly. And it rose very disproportionately by education. So, there was a big increase in divorce for the people with the least education. People with the highest education tended not to divorce. And I think--it's either in your results or in others I've read--among the top 1%, the percent that's married is very high, remains very high. Whereas in the rest of the population it's fallen dramatically--but particularly dramatically among low-skilled, low-education workers. So, what we've seen in the United States since the 1970s is an increase in households that's not due to population growth, but it's due to either divorce or people not marrying at all. And, that that increase in households has not been spread equally across the income distribution--it's disproportionately found in the lower half. Does that--when you apportion family income, household income, equally between husband and wife, do you think you--are you controlling for that?
Gabriel Zucman: We have a way, which is not the only way--and we consider other ways to control for that--which is to always conduct the analysis at the adult individual level. Then the question becomes: How do you split income within married couples? And, there's a lot of research on that, and sharing rules among couples, and very interesting research in this area. But it's hard to have long time series for empirical sharing rules. So, in our benchmark series, what we do is, we are agnostic and we just say: Income is split equally. Now, with the data that we have and that we are going to make online, the micro-files[?] that we are going to make online relatively soon, you can experiment with other sharing rules and you can say, 'Okay, husbands [?] take a greater fraction of the comparison[?] of income, or a smaller fraction; this has changed over time.' But, that's the way that we address this issue. We think that if you want to be consistent, if you want to create series that are consistent with macroeconomic growth, which is typically expressed in terms of per adult income, you need to study the distribution of per-adult income. And so I think that's progress compared to the studies that look at household incomes and sometimes use no equivalence scales to individualize[?] income, because we searched studies: You can't be consistent with macroeconomic growth. Your country[?] decompose macroeconomic growth across social groups.
Russ Roberts: I want to add that, when I raised the data findings of the paper by Auten, et al, in the National Tax Journal, you correctly asked the question, made the observation it depends on what kind of income you are looking at. They were, I think, looking at after-tax, after-transfer; so they find large effects. They may not have found those same effects if they had narrowed it down to labor income. But, it does also challenge your conclusion that taxes and transfers don't have a big effect. And Auten has that other paper with Splinter from 2016 where they find large effects in reductions of--they find much smaller growth in inequality when you include government taxes and transfers. What do you think explains that difference relative to yours? Have you looked at it?
Gabriel Zucman: Yes, yes; we've looked at these carefully. And there are a number of differences. One big difference is--we tried, and we do distribute 100% of National Income, and they don't. Another difference is, when you [?] pre-tax and transfer income, whereas they look at income concepts which are more mixed. So it's really in the area of data[?] that you mentioned, it would be income including some forms of government transfer, but not excluding taxes, for instance. Now, we know, taxes for the bottom 50%, people don't know that but they've increased quite a lot--
Russ Roberts: Social Security--payroll taxes--
Gabriel Zucman: because of payroll taxes. Exactly. They've increased a lot. The overall tax system in the United States, if you compute average tax rates by income group, taking into account all taxes at all levels of government, you find that the top 1% average tax rate is a bit higher than the average macroeconomic tax rate in the United States, which is 30%. You find that the bottom 50% average tax rate is a bit below 30%. But the difference is very small. That is, you know, all together, the tax system in the United States is barely progressive. It's close to a flat tax where everybody almost pays 30% of their income. And that's a big change compared to the 1960s and 1970s where the top 1% average tax rate was significantly higher and bigger than the average tax rate; and the tax rate for the bottom 50% was significantly lower, below the average tax rate. And so, you know, it's very important to be consistent. You can't just look at transfers but forget about taxes, and vice versa.
Russ Roberts: Yep; I agree.
Gabriel Zucman: So, that explains a lot of the differences.
Russ Roberts: Let's talk about the possible explanations. The last thing I want to add is that consumption data also, to me, casts some doubt on the claim that the bottom 50% have made no change in their economic wellbeing since 1980 or a very small change. The bottom 20% has much greater access, and of course, a fortiori, the bottom 50% has a much greater access to a huge range of consumer goods. Cellphones, washing machines, air conditioning, cars; houses have gotten larger, the median house is larger, it's not just houses at the highest end. So, to me, that's also a challenge to the finding that there's been no gain. I was alive in 1980--I think I'm older than you are, than you were then--were you born in 1980? But I was an adult; so I remember what 1980 looked like. And the world has changed a lot. It's a lot richer; there's a lot more stuff. And it's not just going to the top 1% or the top 10%. It's all over the place. It's in every Walmart--I'm not saying it's particularly important or good. But it's hard to understand the claim that there's no gain for that group. This is not a very sophisticated argument. It's a sniff test.
Gabriel Zucman: Well, I hear this argument, yeah. The thing is that, when you look at the macroeconomic data--so, let's forget about distributions for a second--is it the case that there's been a lot of growth in the United States since 1980? In the macro data, the answer frankly is, 'Not really.' Average income per adult, in real terms, has increased only 1.4% per year since 1980. That's what the macro data tell you. Now, maybe macroeconomic statistics understate the actual growth rate of the economy maybe because--
Russ Roberts: the price index--
Gabriel Zucman: of a problem with the price index, the price indices--yes--and they overstate inflation and they don't properly take into account new products. So, the price index that we use, and it is also a difference with some of the studies that you mentioned, is, I think, what's probably the best price index, the one that takes into account the substitution bias and other problems with the CPI [Consumer Price Index]. It's not the CPI. It's the National Income Price Deflator--it's the deflator that's used, the price index that's used to compute real macroeconomic growth. And it shows less inflation than the CPI. So, we already take into account one standard important criticism that's been addressed to this literature[?], which is, 'Oh, no; you understate real income growth because you have overstated inflation.' Now, maybe even the GDP Deflator or the National Income Deflator might overstate inflation. That's possible. What we do is that we take the National Accounts data as given; that's really one[?] limitation of what we do. We know that they have problems. But we think it's valuable to say: Let's introduce distributional measures in this macro data, and then if the macro data change then our results also will be changed.
Russ Roberts: Yeah, no, I think that's exactly where I do think you are using a better--it's such a complicated question, right? Because ideally what you want is a basket of goods that the bottom 50% buys. Which is, of course--the macro, the GDP Deflator has business purchases and there's a big interesting technical question about how computers have changed the price over time and how that's affected productivity; and whether we are measuring it correctly for individuals versus the economy as a whole, and not something to get into right now. But the--I mean, it's obviously a really hard problem; and you did the best you could. Obviously. I think you did. There's nothing particularly easy or definitive about that decision. You have to make a decision; and that's probably the best one, and it's consistent with the GDP accounts, so it's a good choice.
Gabriel Zucman: And, I didn't answer your question on consumption growth, which I think is very important. So, first of all, consumption--the data we have to study consumption are frankly of limited quality, unfortunately. They miss a lot of consumption. They [?] capture well the consumption of the wealthy and so on. But the second thing that's important is that consumption and income can evolve differently, if saving rates change.
Russ Roberts: True.
Gabriel Zucman: And certainly that's part of what has happened in the United States, which is that household debt has skyrocketed before the Financial Crisis; the saving rates of the middle class and the working class has collapsed. What we find is that in the 10 years before the Great Recession, the saving rate of the bottom 90% of the wealth distribution--when you rank people by wealth--was actually negative. Okay? So, they were consuming more than their income. And that's a big part of, you know, the macroeconomic story. And also can help reconcile the trends in consumption--maybe there's been some increase in area[?] of consumption with the trends in income.
Russ Roberts: Yeah, it's an interesting point. Obviously, in the run-up to 2007, 2008, a lot of people saw their home--the values of their homes or what they thought was the value of their home--rising. And many people borrowed from that future, what they thought was a future capital gain, in order to increase their consumption now. And then it turns out that capital gain never materialized. And, it's definitely the case that there was some consumption that was financed out of thin air that wasn't real. Financed out of creditors who never got their money back. For sure. Who consumed less, presumably, than they expected.
Russ Roberts: Let's turn to the causes. So, the first--you mention four, which are all interesting: the decline in unionization in the United States; the stagnation or actual fall in the real value of the minimum wage; the loss of bargaining power of labor; and access to education. So, I don't understand those arguments in general. I do understand the last one. Let me raise the issues I don't understand about the first three. So, let's start with unionization. Unionization in the United States in the private sector, pretty much fallen steadily since 1945. In the aftermath of WWII, manufacturing as a percentage of total employment fell steadily; unionization fell steadily both in percentage and absolute terms. And so, it's hard to understand why, when we look at--let me say it differently. When we look at all the data on these issues of stagnation of the middle class--again, I'm moving away from inequality right now; just this question of how the average person is doing and whether they benefit from growth--it's hard to understand why 1973, say, or in your case with your data, 1980, it looks so bleak. Because that unionization was falling all along. And those gains in the measured data are very different in the post-War, 1945-1973 or 1974 period than they are in the later period. Suggesting maybe it's not unionization. What are your thoughts on that?
Gabriel Zucman: I think it's possible that the decline in unionization affects income distribution. We saw some lag because it takes time to change, you know, wage contracts. There is some stickiness in labor agreements. What I find pretty striking is when I look at the cross-section of countries today, and also, you know, the panel[?] of countries, at the global level, there is a pretty strong correlation between unionization and income inequality. So, you look at the countries that, where the middle class is doing well these days. Germany, Scandinavian countries: These are economies where unions have power. You know, unions are in corporate[?]. A big function of the workforce is unionized, much more than in the United States. So, these are economies that, just like the United States, have been subject to the same trends of globalization, open international trade; they do import a lot of stuff[?] from China; they do export a lot; they are very integrated in global markets. They have faced the same trends in terms of technological trend, technological change. And yet, the income distribution, and the growth rate of income for the middle class and the working class, has been very different. And it's hard to say what fraction of this owes to unions and what fraction of this owes to minimum wages, education, and other factors. But all together, I think the reasons we have suggest that, broadly speaking, policies matter a lot. And policies affect the pre-tax and transfer distribution of income. And, they matter a lot. Because, if they didn't matter, we should see the same orders or same trends in all the world's developed countries as in the United States. And that's not at all what we see.
Russ Roberts: Well, I guess the question is: What other trends are going on beside policy changes? The demographic changes we talked about, to the extent that they are hard to control for, could be very different in the United States than elsewhere. I think worldwide, in general, there are similar trends. You are right--they could be different in some countries relative to others--the magnitudes may not be the same. But the growth of command of the top 1%, the income share as measured to going to the top 1%, has grown very dramatically in a lot of industrialized countries. And, I just think about--well, anyway--I'm not--I'm open to the possibility it's unionization. I don't think it's as you say. It could be just one factor. But there are other factors that are not the same across countries. Globalization is one that is similar across a lot of countries. But do you have any idea of what unionization is in Germany? In the United States, I think it's now under 10% in the private sector.
Gabriel Zucman: I don't want to say, you know, [?] know off the top of my head, but I do know that unions in Germany, they play a much bigger role in corporations than in the United States. So, typically, unions have several board members. They are in the corporate world. So, they are involved in decision-making at the--within corporations. And, yeah, that's a big difference with the United States. And it's likely to be part of the reason why the German working class is doing working than the United States. You know, just the buying power that it has is why it's stronger. In the United States, corporations increasingly since the 1980s have been viewed as having for sole purposes the sole purposes of maximizing shareholder value. So, that's something that's, that's [?] very strongly in the United States--what a CEO [Chief Executive Officer] should do is maximize shareholder value. Now, ask this question to people in Germany, to people in Scandinavia, 'Is the role of a CEO to maximize shareholder value?' And they have very different answers. You know, most of them would say, 'No. There are several stakeholders. There are shareholders. There are workers. Customers. Local governments. And we have a duty towards all of these stakeholders, not only shareholders.' And so, of course, if this, such a different state of mind, I think that explains, you know, part of the difference between the United States and other countries. And I think if you look carefully at the data, I'd like to push back a little bit on this idea that the top 1% has surged everywhere. I think the rise in the top 1% income share among rich countries is actually a uniquely American phenomenon. It is not something that you see in Continental Europe. It is not something that you see in Canada. It is something that you see a little bit in the United Kingdom, but the magnitude in the United Kingdom of the rise is much less than the United States. The United States is the main people[?] the outlier among rich countries. The top 1% used to have 10% of total pre-tax income in 1980. Now, around 20% of the total pre-tax income. Whereas, for the bottom 50%, it's exactly the opposite. The bottom 50% used to have 20% of total pre-tax income in 1980. Now, it's[?] around about 12% of total pre-tax income. There is no other developed country in the world where such a phenomenon has happened since 1980.
Russ Roberts: Well, you know, there's two possible--we were talking mainly about the bottom 50% before, but now we're on the top. The United States has--again, remembering that we're not following the same people. We're looking at a point in time. I always use the example of an athlete. In 1980, Larry Byrd and Magic Johnson were the best basketball players, and they made a lot more than the average person in the stands who was following them. In 2014, the best basketball players, I would say--you could argue who they are, but let's say--LeBron James and Stephen Curry. They make much more than--the ratio of their income to a fan in the stands is much higher. Because basketball is a lot more popular around the world; globalization has occurred; the gains to being the best have increased. And one could argue that, 'Yes, in the United States it's much easier to make a lot more money today than it was to make a lot more money in 1980.' And that's not a bad thing. There are examples of it that are bad things--people on Wall Street, I think, have benefited in a grotesque way from the willingness of the U.S. government to treat them with special favors; and I think that's a distortion that we should be upset about. Whether we should be upset about LeBron James--we might want to tax him more. You're right. We might agree that that's a good thing. We might not. But if we just looking at the underlying effectiveness of the economy and the role of market forces versus, say, government policy to set salaries, I don't find anything surprising or disturbing about the fact that the United States is in many ways a more entrepreneurial economy than outside the United States--the opportunity of access to venture capital is unparalleled. So, there are a lot of reasons why Sergey Brin and Larry Page and Mark Zuckerberg and LeBron James and all these folks--entertainers would also be included--do a lot better today than they did a long time ago. And I don't find that disturbing. I find parts of it disturbing. What are your thoughts?
Gabriel Zucman: Well, I think--and I know--basketball, particularly, is not a good example, because I'm sorry to say but people in Europe and outside of the United States, most of them don't really care about American basketball. And so it has always been and still is something that is of interest only to the United States.
Russ Roberts: I think they care in China, actually. Which is kind of a big country.
Gabriel Zucman: Maybe. But, to me, what drives, what has been driving top incomes and in particular for top corporate executives is there is much less globalization than the big changes that have happened in terms of tax policy in the United States. So, think about it. Globalization happens everywhere. But, it's only in the United States that you've seen CEO pay skyrocket. What has happened in the United States--the United States was the country in the 1960s with the highest marginal income tax rate in the world--90%. It's hard to have more than 90%.
Russ Roberts: You can have it.
Gabriel Zucman: You can.
Russ Roberts: There's not much room. It's not a good strategy.
Gabriel Zucman: You can have 100%. But it was 90%. It's moved from 90% to something in the mid-20% in 1986, just until the Tax Reform Act of 1986. It's a dramatic development. Where, recently[?] in the 1960s, you face a 90% of marginal income tax rate. There is absolutely no incentive to try to earn $50 million dollars in income, when out of any extra dollar that you earn, 90 cents are going to go to the IRS [Internal Revenue Service]. There's just no such incentive to do so. Of course, when you face a top marginal income tax rate of 20% or 30%, now it becomes valuable to try to earn very high incomes. And then the question becomes: 'Okay. Is it now, top earners have incentives to earn more? Is it good? Is it, you know, translating into a lot of growth for the United States as a whole? Or is it mostly at the expense of other stakeholders?' And here, there is no perfect evidence. But, at a high level what I find compelling is that you don't see that macroeconomic growth has been spectacularly high in the United States since the 1980s. Again, actually, it has been pretty low. You don't see that. But you see the income of top earners has boomed. So, one reading of this evidence is that, thanks to these lower tax rates, high earners have been earning more. But it's not because they have been producing so much more. It's because what they've earned is other crops[?] of the population which have not earned, actually, that income. So, it's at the expense of other stakeholders. And typically, corporate executives are better able now to extract very high salaries from corporate bonds in the United States. That's at the expense of shareholders. That's at the expense of, maybe, other workers in the firms. That, I think, explains not--certainly not everything--but part of the increase in the very highest incomes in the United States.
Russ Roberts: Well, it's interesting you mentioned that Europeans don't care much about basketball. I thought you were going to say that basketball players don't make a very large portion of the 1%. Even athletes don't, and even entertainers don't. And some might even say it's not so much Wall Street, even. There's a big debate, and you're going to help me understand it better because I think you know the data much better than I do as to what the real source of those income gains are. When you mention CEOs--a lot of people have argued--I don't know if it's true, but a lot of people have argued that American CEOs make more because they tend to have more responsibility. Their companies are dramatically larger than companies outside the United States. The part of this--so, I'm interested. So, CEOs are part of the 1%, for sure. So, I'm interested in what your thoughts are on sort of the decomposition of where that income comes from in the top 1%. Because I know you've looked at it.
Gabriel Zucman: In terms of who are the top earners, a lot of them are indeed corporate executives in various industries. So, finance is an important component; it is far from all of it. In lots of industries--in finance, in the health care industry, manufacturing. So, across the board--in the pharmaceutical industry--you've seen the pay of the top executives grow automatically faster than average worker pay. That's part of what's happening. But, the data we have now, getting back to our distributional national accounts, shows that most of the--now, the majority of the income of top 1% earners is not labor income. Is not wages and salaries and stock options and bond indices[?]. It's actually capital income. And that's a relatively new development. In the 1980s, 1990s, the rise of U.S. income inequality was essentially driven by an increase in labor income inequality--the upsurge of top corporate executive pay. Since 2000, it's been very different: labor income inequality actually has not increased, might even have declined internally[?]. All of the rise of the top 1% income share since 2000 owes to an increase in capital income, in the dividend income, corporate profits, interest that high-income earners get. And as important, because, of course, the forces that shape the distribution of labor income and the distribution of wealth and capital income are quite different. And so if you want to understand rising inequality in recent years in the United States, you need to ask yourself, 'Okay. Is coming from capital. So, what's the reason for that?' So, one potential explanation is that these high-labor incomes of the 1980s, 1990s, have been saved at a pretty high rate, and so these high earners have been accumulating quite a lot of wealth. That wealth itself, it generates some return; and so capital income, which in turn is flow of capital income, is being saved at high rates. So, wealth further accumulates and capital income concentrations further increases. And, I think that this is what is happening in the United States at the moment. Not everything corresponds to that. But that was not very important in the 1980s and 1990s. Now it's becoming very important. Capital income at the top is more important than labor income.
Russ Roberts: Yeah, no; that's very interesting.
Russ Roberts: The point I wanted to make--I lost my train of thought a minute ago, but the point I wanted to add is that: I just don't agree that saying that you want to maximize shareholder value means therefore that you don't care about stakeholders. Outside the profit and loss. Obviously, if you treat your workers badly, you are not going to have very good returns for your shareholders. I guess the question, though, is whether there is some room to, or cultural constraints, on either CEO pay or what you have to pay workers that would be more likely in those places you are talking about. I think that's the question.
Gabriel Zucman: Yes. And I think when you have worker preemptives[?] in corporate boards[?], that sets limits to CEO pay--and that--there's no research limits in the United States. In the United States it is possible to get salaries of dozens or hundreds of millions of dollars. In a way that's for--you know, that seems impossible in Spain[?] or in countries or in most German corporations. And there are cultural reasons. There are reasons that the Internet[?], the role of unions, incorporate both. And there are tax reasons as well. And in the United States, again, if you manage, let's say, to earn income that's taxable as capital gains, whereas the top marginal income tax rate on that at the Federal level, 23.8%, that's, that's, that's not a lot. And so, when it's 23.8--yes, it's really worth it to try to earn a hundred million. In most other countries it's significantly more than that.
Russ Roberts: You are presuming--I want to--we can end with is, because I want to come back to your point you made earlier--we won't quite end on it if you have another few minutes. But the point you made about bargaining power. Usually market forces tend to result in higher salaries when taxes go up, and lower salaries when taxes go down. So, you are talking the desire people have to make more money. What your desire is, is that a market doesn't matter: you are stuck with, usually the market rate. So, when your tax rate goes--say, your tax rate goes up. If you can still get the job, usually that's going to increase the market wage you are going to earn. But, if you don't believe wages are set by market forces--which maybe they aren't; but for CEOs, they might not be. There might be some ability of Boards to act capriciously if sufficiently large. But, going back to the average person: I've never understood this claim that bargaining power matters. There's--I'm not in a bazaar. I'm not in a street market or a farmer's market where I can haggle over prices. There's a going rate for, usually, a certain type of skill. So, my bargaining power--maybe it means something differently than what I hear when you use that phrase.
Gabriel Zucman: Well, I think it--this is a question that is connected to the growing evidence about the rise of market power in the United States, and the fact that it seems we are increasingly so, and not on a perfectly competitive market. And, neither on the labor market nor on the product market. There is growing evidence of rising market, of rising concentration, maybe of rising monopsony power in the labor markets. And as soon as you on[?] it in the Econ 101, perfectly competitive, perfect-information markets, you know, pay can be different than the Marginal Product of Labor. It can be higher; it can be lower--in ways that are determined a lot by intra-firm bargaining, and so by directive[?] power of unions. And by policy. And, that, I think is--you know, explains, again, part of the divergence between wage inequality in the United States and wage inequality in Europe--in countries where salaries tend to be fixed by rigid salary scales. And much less by pure market forces. Which don't necessarily give you a wage equal to a Marginal Product, depending on the type of competition that you have.
Russ Roberts: Yeah; well it's interesting, though, at the--you are in the Bay area in California, where I spend my summers. And I run into and chat with a lot of engineers at Google and Facebook. And, Google is not using their monopoly power to the extent they have it to pay low salaries and treat their workers badly. They are very pleasant places to work. Maybe be even more pleasant if they had less market power. I don't know. I think--I actually think they are making some monopoly power and they are using it--they have some and they are using it partly to reduce their turnover rate. And the, their workers, the ones I talk to, feel that they are relatively well-treated. And they do make a relatively large amount of money. I think get free lunch, too.
Russ Roberts: Let's close with something we can agree on, with a little more eagerness, perhaps, which is the education system as the last factor that you mention. I think we probably agree that the people at the bottom half, and certainly the bottom 20%, have bad access, inadequate access to skills, and certainly to even formal education. What do think we can do to do a better job there?
Gabriel Zucman: That's a complicated question. But it's one of things that I find most striking and shocking about the U.S. economy--is, how well your probability to attend college is predicted by the rank of your parents in the income distribution. That is, it's almost a perfect 1-to-1 correlation. If your parents are in the top 1%, you have 100% chance to attend college. If your parents are in the bottom 1%, it's not like you have zero probability you have to attend college, but it's close to zero. It's not 10%. So, think about it. This is incredibly unfair. And a huge problem for the United States. Now, what are the ways to address that issue? It's a complicated question. So, I personally think that tuition fees, you know, tend to prevent access to 4-year colleges for young Americans from a working class background. And so that having more free public higher education could help having more equal access to higher education. Now, if that's one thing that might help, it's far from the only thing that's important--
Russ Roberts: The important thing is to graduate. And graduation rates are even--
Gabriel Zucman: [?] Absolutely. [?] I also think that lots of problems of, at the high school level, the elementary school level. And so, these are [?] very important questions. Maybe we can talk about these in another show.