New thoughts on capital in the twenty-first century
It's very nice to be here tonight.
So I've been working on the history of income and wealth distribution for the past 15 years, and one of the interesting lessons coming from this historical evidence is indeed that, in the long run, there is a tendency for the rate of return of capital to exceed the economy's growth rate, and this tends to lead to high concentration of wealth. Not infinite concentration of wealth, but the higher the gap between r and g, the higher the level of inequality of wealth towards which society tends to converge.
0:48 So this is a key force that I'm going to talk about today, but let me say right away that this is not the only important force in the dynamics of income and wealth distribution, and there are many other forces that play an important role in the long-run dynamics of income and wealth distribution. Also there is a lot of data that still needs to be collected. We know a little bit more today than we used to know, but we still know too little, and certainly there are many different processes — economic, social, political — that need to be studied more. And so I'm going to focus today on this simple force, but that doesn't mean that other important forces do not exist.
So most of the data I'm going to present comes from this database that's available online: the World Top Incomes Database. So this is the largest existing historical database on inequality, and this comes from the effort of over 30 scholars from several dozen countries. So let me show you a couple of facts coming from this database, and then we'll return to r bigger than g.
So fact number one is that there has been a big reversal in the ordering of income inequality between the United States and Europe over the past century. So back in 1900, 1910, income inequality was actually much higher in Europe than in the United States, whereas today, it is a lot higher in the United States. So let me be very clear: The main explanation for this is not r bigger than g. It has more to do with changing supply and demand for skill, the race between education and technology, globalization, probably more unequal access to skills in the U.S., where you have very good, very top universities but where the bottom part of the educational system is not as good, so very unequal access to skills, and also an unprecedented rise of top managerial compensation of the United States, which is difficult to account for just on the basis of education. So there is more going on here, but I'm not going to talk too much about this today, because I want to focus on wealth inequality.
So let me just show you a very simple indicator about the income inequality part. So this is the share of total income going to the top 10 percent. So you can see that one century ago, it was between 45 and 50 percent in Europe and a little bit above 40 percent in the U.S., so there was more inequality in Europe. Then there was a sharp decline during the first half of the 20th century, and in the recent decade, you can see that the U.S. has become more unequal than Europe, and this is the first fact I just talked about. Now, the second fact is more about wealth inequality, and here the central fact is that wealth inequality is always a lot higher than income inequality, and also that wealth inequality, although it has also increased in recent decades, is still less extreme today than what it was a century ago, although the total quantity of wealth relative to income has now recovered from the very large shocks caused by World War I, the Great Depression, World War II.
So let me show you two graphs illustrating fact number two and fact number three. So first, if you look at the level of wealth inequality, this is the share of total wealth going to the top 10 percent of wealth holders, so you can see the same kind of reversal between the U.S. and Europe that we had before for income inequality. So wealth concentration was higher in Europe than in the U.S. a century ago, and now it is the opposite. But you can also show two things: First, the general level of wealth inequality is always higher than income inequality.
So remember, for income inequality, the share going to the top 10 percent was between 30 and 50 percent of total income, whereas for wealth, the share is always between 60 and 90 percent. Okay, so that's fact number one, and that's very important for what follows. Wealth concentration is always a lot higher than income concentration.
Fact number two is that the rise in wealth inequality in recent decades is still not enough to get us back to 1910. So the big difference today, wealth inequality is still very large, with 60, 70 percent of total wealth for the top 10, but the good news is that it's actually better than one century ago, where you had 90 percent in Europe going to the top 10. So today what you have is what I call the middle 40 percent, the people who are not in the top 10 and who are not in the bottom 50, and what you can view as the wealth middle class that owns 20 to 30 percent of total wealth, national wealth, whereas they used to be poor, a century ago, when there was basically no wealth middle class. So this is an important change, and it's interesting to see that wealth inequality has not fully recovered to pre-World War I levels, although the total quantity of wealth has recovered. Okay? So this is the total value of wealth relative to income, and you can see that in particular in Europe, we are almost back to the pre-World War I level.
So there are really two different parts of the story here. One has to do with the total quantity of wealth that we accumulate, and there is nothing bad per se, of course, in accumulating a lot of wealth, and in particular if it is more diffuse and less concentrated. So what we really want to focus on is the long-run evolution of wealth inequality, and what's going to happen in the future. How can we account for the fact that until World War I, wealth inequality was so high and, if anything, was rising to even higher levels, and how can we think about the future?
So let me come to some of the explanations and speculations about the future. Let me first say that probably the best model to explain why wealth is so much more concentrated than income is a dynamic, dynastic model where individuals have a long horizon and accumulate wealth for all sorts of reasons. If people were accumulating wealth only for life cycle reasons, you know, to be able to consume when they are old, then the level of wealth inequality should be more or less in line with the level of income inequality.
But it will be very difficult to explain why you have so much more wealth inequality than income inequality with a pure life cycle model, so you need a story where people also care about wealth accumulation for other reasons. So typically, they want to transmit wealth to the next generation, to their children, or sometimes they want to accumulate wealth because of the prestige, the power that goes with wealth. So there must be other reasons for accumulating wealth than just life cycle to explain what we see in the data. Now, in a large class of dynamic models of wealth accumulation with such dynastic motive for accumulating wealth, you will have all sorts of random, multiplicative shocks. So for instance, some families have a very large number of children, so the wealth will be divided. Some families have fewer children. You also have shocks to rates of return. Some families make huge capital gains. Some made bad investments. So you will always have some mobility in the wealth process. Some people will move up, some people will move down. The important point is that, in any such model, for a given variance of such shocks, the equilibrium level of wealth inequality will be a steeply rising function of r minus g. And intuitively, the reason why the difference between the rate of return to wealth and the growth rate is important is that initial wealth inequalities will be amplified at a faster pace with a bigger r minus g.
So take a simple example, with r equals five percent and g equals one percent, wealth holders only need to reinvest one fifth of their capital income to ensure that their wealth rises as fast as the size of the economy. So this makes it easier to build and perpetuate large fortunes because you can consume four fifths, assuming zero tax, and you can just reinvest one fifth. So of course some families will consume more than that, some will consume less, so there will be some mobility in the distribution, but on average, they only need to reinvest one fifth, so this allows high wealth inequalities to be sustained.
Now, you should not be surprised by the statement that r can be bigger than g forever, because, in fact, this is what happened during most of the history of mankind. And this was in a way very obvious to everybody for a simple reason, which is that growth was close to zero percent during most of the history of mankind. Growth was maybe 0.1, 0.2, 0.3 percent, but very slow growth of population and output per capita, whereas the rate of return on capital of course was not zero percent. It was, for land assets, which was the traditional form of assets in preindustrial societies, it was typically five percent. Any reader of Jane Austen would know that. If you want an annual income of 1,000 pounds, you should have a capital value of 20,000 pounds so that five percent of 20,000 is 1,000. And in a way, this was the very foundation of society, because r bigger than g was what allowed holders of wealth and assets to live off their capital income and to do something else in life than just to care about their own survival.
Now, one important conclusion of my historical research is that modern industrial growth did not change this basic fact as much as one might have expected. Of course, the growth rate following the Industrial Revolution rose, typically from zero to one to two percent, but at the same time, the rate of return to capital also rose so that the gap between the two did not really change. So during the 20th century, you had a very unique combination of events.
First, a very low rate of return due to the 1914 and 1945 war shocks, destruction of wealth, inflation, bankruptcy during the Great Depression, and all of this reduced the private rate of return to wealth to unusually low levels between 1914 and 1945. And then, in the postwar period, you had unusually high growth rate, partly due to the reconstruction. You know, in Germany, in France, in Japan, you had five percent growth rate between 1950 and 1980 largely due to reconstruction, and also due to very large demographic growth, the Baby Boom Cohort effect. Now, apparently that's not going to last for very long, or at least the population growth is supposed to decline in the future, and the best projections we have is that the long-run growth is going to be closer to one to two percent rather than four to five percent.
So if you look at this, these are the best estimates we have of world GDP growth and rate of return on capital, average rates of return on capital, so you can see that during most of the history of mankind, the growth rate was very small, much lower than the rate of return, and then during the 20th century, it is really the population growth, very high in the postwar period, and the reconstruction process that brought growth to a smaller gap with the rate of return. Here I use the United Nations population projections, so of course they are uncertain. It could be that we all start having a lot of children in the future, and the growth rates are going to be higher, but from now on, these are the best projections we have, and this will make global growth decline and the gap between the rate of return go up.
Now, the other unusual event during the 20th century was, as I said, destruction, taxation of capital, so this is the pre-tax rate of return. This is the after-tax rate of return, and after destruction, and this is what brought the average rate of return after tax, after destruction, below the growth rate during a long time period. But without the destruction, without the taxation, this would not have happened. So let me say that the balance between returns on capital and growth depends on many different factors that are very difficult to predict: technology and the development of capital-intensive techniques. So right now, the most capital-intensive sectors in the economy are the real estate sector, housing, the energy sector, but it could be in the future that we have a lot more robots in a number of sectors and that this would be a bigger share of the total capital stock that it is today. Well, we are very far from this, and from now, what's going on in the real estate sector, the energy sector, is much more important for the total capital stock and capital share.
The other important issue is that there are scale effects in portfolio management, together with financial complexity, financial deregulation, that make it easier to get higher rates of return for a large portfolio, and this seems to be particularly strong for billionaires, large capital endowments. Just to give you one example, this comes from the Forbes billionaire rankings over the 1987-2013 period, and you can see the very top wealth holders have been going up at six, seven percent per year in real terms above inflation, whereas average income in the world, average wealth in the world, have increased at only two percent per year. And you find the same for large university endowments — the bigger the initial endowments, the bigger the rate of return.
Now, what could be done? The first thing is that I think we need more financial transparency. We know too little about global wealth dynamics, so we need international transmission of bank information. We need a global registry of financial assets, more coordination on wealth taxation, and even wealth tax with a small tax rate will be a way to produce information so that then we can adapt our policies to whatever we observe. And to some extent, the fight against tax havens and automatic transmission of information is pushing us in this direction. Now, there are other ways to redistribute wealth, which it can be tempting to use. Inflation: it's much easier to print money than to write a tax code, so that's very tempting, but sometimes you don't know what you do with the money. This is a problem. Expropriation is very tempting. Just when you feel some people get too wealthy, you just expropriate them. But this is not a very efficient way to organize a regulation of wealth dynamics. So war is an even less efficient way, so I tend to prefer progressive taxation, but of course, history — (Laughter) — history will invent its own best ways, and it will probably involve a combination of all of these.
Bruno Giussani: Thomas Piketty. Thank you.
Thomas, I want to ask you two or three questions, because it's impressive how you're in command of your data, of course, but basically what you suggest is growing wealth concentration is kind of a natural tendency of capitalism, and if we leave it to its own devices, it may threaten the system itself, so you're suggesting that we need to act to implement policies that redistribute wealth, including the ones we just saw: progressive taxation, etc. In the current political context, how realistic are those? How likely do you think that it is that they will be implemented?
Thomas Piketty: Well, you know, I think if you look back through time, the history of income, wealth and taxation is full of surprise. So I am not terribly impressed by those who know in advance what will or will not happen. I think one century ago, many people would have said that progressive income taxation would never happen and then it happened. And even five years ago, many people would have said that bank secrecy will be with us forever in Switzerland, that Switzerland was too powerful for the rest of the world, and then suddenly it took a few U.S. sanctions against Swiss banks for a big change to happen, and now we are moving toward more financial transparency. So I think it's not that difficult to better coordinate politically. We are going to have a treaty with half of the world GDP around the table with the U.S. and the European Union, so if half of the world GDP is not enough to make progress on financial transparency and minimal tax for multinational corporate profits, what does it take? So I think these are not technical difficulties. I think we can make progress if we have a more pragmatic approach to these questions and we have the proper sanctions on those who benefit from financial opacity.
BG: One of the arguments against your point of view is that economic inequality is not only a feature of capitalism but is actually one of its engines. So we take measures to lower inequality, and at the same time we lower growth, potentially. What do you answer to that?
TP: Yeah, I think inequality is not a problem per se. I think inequality up to a point can actually be useful for innovation and growth. The problem is, it's a question of degree. When inequality gets too extreme, then it becomes useless for growth and it can even become bad because it tends to lead to high perpetuation of inequality over time and low mobility. And for instance, the kind of wealth concentrations that we had in the 19th century and pretty much until World War I in every European country was, I think, not useful for growth. This was destroyed by a combination of tragic events and policy changes, and this did not prevent growth from happening. And also, extreme inequality can be bad for our democratic institutions if it creates very unequal access to political voice, and the influence of private money in U.S. politics, I think, is a matter of concern right now. So we don't want to return to that kind of extreme, pre-World War I inequality. Having a decent share of the national wealth for the middle class is not bad for growth. It is actually useful both for equity and efficiency reasons.
BG: I said at the beginning that your book has been criticized. Some of your data has been criticized. Some of your choice of data sets has been criticized. You have been accused of cherry-picking data to make your case. What do you answer to that?
TP: Well, I answer that I am very happy that this book is stimulating debate. This is part of what it is intended for. Look, the reason why I put all the data online with all of the detailed computation is so that we can have an open and transparent debate about this. So I have responded point by point to every concern. Let me say that if I was to rewrite the book today, I would actually conclude that the rise in wealth inequality, particularly in the United States, has been actually higher than what I report in my book. There is a recent study by Saez and Zucman showing, with new data which I didn't have at the time of the book, that wealth concentration in the U.S. has risen even more than what I report. And there will be other data in the future. Some of it will go in different directions. Look, we put online almost every week new, updated series on the World Top Income Database and we will keep doing so in the future, in particular in emerging countries, and I welcome all of those who want to contribute to this data collection process. In fact, I certainly agree that there is not enough transparency about wealth dynamics, and a good way to have better data would be to have a wealth tax with a small tax rate to begin with so that we can all agree about this important evolution and adapt our policies to whatever we observe. So taxation is a source of knowledge, and that's what we need the most right now.
BG: Thomas Piketty, merci beaucoup.