Capital in the Twenty-First Century

Problems of Synthetic Indices

Before turning to a country-by-country examination of the historical evolution of inequality in order to answer the questions posed above, several methodological issues remain to be discussed. In particular, Tables 7.1–3 include indications of the Gini coefficients of the various distributions considered. The Gini coefficient—named for the Italian statistician Corrado Gini (1884–1965)—is one of the more commonly used synthetic indices of inequality, frequently found in official reports and public debate. By construction, it ranges from 0 to 1: it is equal to 0 in case of complete equality and to 1 when inequality is absolute, that is, when a very tiny group owns all available resources.

In practice, the Gini coefficient varies from roughly 0.2 to 0.4 in the distributions of labor income observed in actual societies, from 0.6 to 0.9 for observed distributions of capital ownership, and from 0.3 to 0.5 for total income inequality. In Scandinavia in the 1970s and 1980s, the Gini coefficient of the labor income distribution was 0.19, not far from absolute equality. Conversely, the wealth distribution in Belle Époque Europe exhibited a Gini coefficient of 0.85, not far from absolute inequality.22

These coefficients—and there are others, such as the Theil index—are sometimes useful, but they raise many problems. They claim to summarize in a single numerical index all that a distribution can tell us about inequality—the inequality between the bottom and the middle of the hierarchy as well as between the middle and the top or between the top and the very top. This is very simple and appealing at first glance but inevitably somewhat misleading. Indeed, it is impossible to summarize a multidimensional reality with a unidimensional index without unduly simplifying matters and mixing up things that should not be treated together. The social reality and economic and political significance of inequality are very different at different levels of the distribution, and it is important to analyze these separately. In addition, Gini coefficients and other synthetic indices tend to confuse inequality in regard to labor with inequality in regard to capital, even though the economic mechanisms at work, as well as the normative justifications of inequality, are very different in the two cases. For all these reasons, it seemed to me far better to analyze inequalities in terms of distribution tables indicating the shares of various deciles and centiles in total income and total wealth rather than using synthetic indices such as the Gini coefficient.

Distribution tables are also valuable because they force everyone to take note of the income and wealth levels of the various social groups that make up the existing hierarchy. These levels are expressed in cash terms (or as a percentage of average income and wealth levels in the country concerned) rather than by way of artificial statistical measures that can be difficult to interpret. Distribution tables allow us to have a more concrete and visceral understanding of social inequality, as well as an appreciation of the data available to study these issues and the limits of those data. By contrast, statistical indices such as the Gini coefficient give an abstract and sterile view of inequality, which makes it difficult for people to grasp their position in the contemporary hierarchy (always a useful exercise, particularly when one belongs to the upper centiles of the distribution and tends to forget it, as is often the case with economists). Indices often obscure the fact that there are anomalies or inconsistencies in the underlying data, or that data from other countries or other periods are not directly comparable (because, for example, the tops of the distribution have been truncated or because income from capital is omitted for some countries but not others). Working with distribution tables forces us to be more consistent and transparent.

The Chaste Veil of Official Publications

For similar reasons, caution is in order when using indices such as the interdecile ratios often cited in official reports on inequality from the OECD or national statistical agencies. The most frequently used interdecile ratio is the P90/P10, that is, the ratio between the ninetieth percentile of the income distribution and the tenth percentile.23 For example, if one needs to earn more than 5,000 euros a month to belong to the top 10 percent of the income distribution and less than 1,000 euros a month to belong to the bottom 10 percent, then the P90/P10 ratio is 5.

Such indices can be useful. It is always valuable to have more information about the complete shape of the distribution in question. One should bear in mind, however, that by construction these ratios totally ignore the evolution of the distribution beyond the ninetieth percentile. Concretely, no matter what the P90/P10 ratio may be, the top decile of the income or wealth distribution may have 20 percent of the total (as in the case of Scandinavian incomes in the 1970s and 1980s) or 50 percent (as in the case of US incomes in the 2010s) or 90 percent (as in the case of European wealth in the Belle Époque). We will not learn any of this by consulting the publications of the international organizations or national statistical agencies who compile these statistics, however, because they usually focus on indices that deliberately ignore the top end of the distribution and give no indication of income or wealth beyond the ninetieth percentile.

This practice is generally justified on the grounds that the available data are “imperfect.” This is true, but the difficulties can be overcome by using adequate sources, as the historical data collected (with limited means) in the World Top Incomes Database (WTID) show. This work has begun, slowly, to change the way things are done. Indeed, the methodological decision to ignore the top end is hardly neutral: the official reports of national and international agencies are supposed to inform public debate about the distribution of income and wealth, but in practice they often give an artificially rosy picture of inequality. It is as if an official government report on inequalities in France in 1789 deliberately ignored everything above the ninetieth percentile—a group 5 to 10 times larger than the entire aristocracy of the day—on the grounds that it was too complex to say anything about. Such a chaste approach is all the more regrettable in that it inevitably feeds the wildest fantasies and tends to discredit official statistics and statisticians rather than calm social tensions.

Conversely, interdecile ratios are sometimes quite high for largely artificial reasons. Take the distribution of capital ownership, for example: the bottom 50 percent of the distribution generally own next to nothing. Depending on how small fortunes are measured—for example, whether or not durable goods and debts are counted—one can come up with apparently quite different evaluations of exactly where the tenth percentile of the wealth hierarchy lies: for the same underlying social reality, one might put it at 100 euros, 1,000 euros, or even 10,000 euros, which in the end isn’t all that different but can lead to very different interdecile ratios, depending on the country and the period, even though the bottom half of the wealth distribution owns less than 5 percent of total wealth. The same is only slightly less true of the labor income distribution: depending on how one chooses to treat replacement incomes and pay for short periods of work (for example, depending on whether one uses the average weekly, monthly, annual, or decadal income) one can come up with highly variable P10 thresholds (and therefore interdecile ratios), even though the bottom 50 percent of the labor income distribution actually draws a fairly stable share of the total income from labor.24

This is perhaps one of the main reasons why it is preferable to study distributions as I have presented them in Tables 7.1–3, that is, by emphasizing the shares of income and wealth claimed by different groups, particularly the bottom half and the top decile in each society, rather than the threshold levels defining given percentiles. The shares give a much more stable picture of reality than the interdecile ratios.

Back to “Social Tables” and Political Arithmetic

These, then, are my reasons for believing that the distribution tables I have been examining in this chapter are the best tool for studying the distribution of wealth, far better than synthetic indices and interdecile ratios.

In addition, I believe that my approach is more consistent with national accounting methods. Now that national accounts for most countries enable us to measure national income and wealth every year (and therefore average income and wealth, since demographic sources provide easy access to population figures), the next step is naturally to break down these total income and wealth figures by decile and centile. Many reports have recommended that national accounts be improved and “humanized” in this way, but little progress has been made to date.25 A useful step in this direction would be a breakdown indicating the poorest 50 percent, the middle 40 percent, and the richest 10 percent. In particular, such an approach would allow any observer to see just how much the growth of domestic output and national income is or is not reflected in the income actually received by these different social groups. For instance, only by knowing the share going to the top decile can we determine the extent to which a disproportionate share of growth has been captured by the top end of the distribution. Neither a Gini coefficient nor an interdecile ratio permits such a clear and precise response to this question.

I will add, finally, that the distribution tables whose use I am recommending are in some ways fairly similar to the “social tables” that were in vogue in the eighteenth and early nineteenth centuries. First developed in Britain and France in the late seventeenth century, these social tables were widely used, improved, and commented on in France during the Enlightenment: for example, in the celebrated article on “political arithmetic” in Diderot’s Encyclopedia. From the earliest versions established by Gregory King in 1688 to the more elaborate examples compiled by Expilly and Isnard on the eve of the French Revolution or by Peuchet, Colqhoun, and Blodget during the Napoleonic era, social tables always aimed to provide a comprehensive vision of the social structure: they indicated the number of nobles, bourgeois, gentlemen, artisans, farmers, and so on along with their estimated income (and sometimes wealth); the same authors also compiled the earliest estimates of national income and wealth. There is, however, one essential difference between these tables and mine: the old social tables used the social categories of their time and did not seek to ascertain the distribution of wealth or income by deciles and centiles.26

Nevertheless, social tables sought to portray the flesh-and-blood aspects of inequality by emphasizing the shares of national wealth held by different social groups (and, in particular, the various strata of the elite), and in this respect there are clear affinities with the approach I have taken here. At the same time, social tables are remote in spirit from the sterile, atemporal statistical measures of inequality such as those employed by Gini and Pareto, which were all too commonly used in the twentieth century and tend to naturalize the distribution of wealth. The way one tries to measure inequality is never neutral.

{EIGHT}

Two Worlds

I have now precisely defined the notions needed for what follows, and I have introduced the orders of magnitude attained in practice by inequality with respect to labor and capital in various societies. The time has now come to look at the historical evolution of inequality around the world. How and why has the structure of inequality changed since the nineteenth century? The shocks of the period 1914–1945 played an essential role in the compression of inequality, and this compression was in no way a harmonious or spontaneous occurrence. The increase in inequality since 1970 has not been the same everywhere, which again suggests that institutional and political factors played a key role.

A Simple Case: The Reduction of Inequality in France in the Twentieth Century

I will begin by examining at some length the case of France, which is particularly well documented (thanks to a rich lode of readily available historical sources). It is also relatively simple and straightforward (as far as it is possible for a history of inequality to be straightforward) and, above all, broadly representative of changes observed in several other European countries. By “European” I mean “continental European,” because in some respects the British case is intermediate between the European and the US cases. To a large extent the continental European pattern is also representative of what happened in Japan. After France I will turn to the United States, and finally I will extend the analysis to the entire set of developed and emerging economies for which adequate historical data exist.

Figure 8.1 depicts the upper decile’s share of both national income and wages over time. Three facts stand out.

First, income inequality has greatly diminished in France since the Belle Époque: the upper decile’s share of national income decreased from 45–50 percent on the eve of World War I to 30–35 percent today.

FIGURE 8.1. Income inequality in France, 1910–2010

Inequality of total income (labor and capital) has dropped in France during the twentieth century, while wage inequality has remained the same.

Sources and series: see piketty.pse.ens.fr/capital21c.

This drop of 15 percentage points of national income is considerable. It represents a decrease of about one-third in the share of each year’s output going to the wealthiest 10 percent of the population and an increase of about a third in the share going to the other 90 percent. Note, too, that this is roughly equivalent to three-quarters of what the bottom half of the population received in the Belle Époque and more than half of what it receives today.1 Recall, moreover, that in this part of the book, I am examining inequality of primary incomes (that is, before taxes and transfers). In Part Four, I will show how taxes and transfers reduced inequality even more. To be clear, the fact that inequality decreased does not mean that we are living today in an egalitarian society. It mainly reflects the fact that the society of the Belle Époque was extremely inegalitarian—indeed, one of the most inegalitarian societies of all time. The form that this inequality took and the way it came about would not, I think, be readily accepted today.

Second, the significant compression of income inequality over the course of the twentieth century was due entirely to diminished top incomes from capital. If we ignore income from capital and concentrate on wage inequality, we find that the distribution remained quite stable over the long run. In the first decade of the twentieth century as in the second decade of the twenty-first, the upper decile of the wage hierarchy received about 25 percent of total wages. The sources also indicate long-term stability of wage inequality at the bottom end of the distribution. For example, the least well paid 50 percent always received 25–30 percent of total wages (so that the average pay of a member of this group was 50–60 percent of the average wage overall), with no clear long-term trend.2 The wage level has obviously changed a great deal over the past century, and the composition and skills of the workforce have been totally transformed, but the wage hierarchy has remained more or less the same. If top incomes from capital had not decreased, income inequality would not have diminished in the twentieth century.

FIGURE 8.2. The fall of rentiers in France, 1910–2010

The fall in the top percentile share (the top 1 percent highest incomes) in France between 1914 and 1945 is due to the fall of top capital incomes.

Sources and series: see piketty.pse.ens.fr/capital21c.

This fact stands out even more boldly when we climb the rungs of the social ladder. Look, in particular, at the evolution of the top centile (Figure 8.2).3 Compared with the peak inequality of the Belle Époque, the top centile’s share of income literally collapsed in France over the course of the twentieth century, dropping from more than 20 percent of national income in 1900–1910 to 8 or 9 percent in 2000–2010. This represents a decrease of more than half in one century, indeed nearly two-thirds if we look at the bottom of the curve in the early 1980s, when the top centile’s share of national income was barely 7 percent.

Again, this collapse was due solely to the decrease of very high incomes from capital (or, crudely put, the fall of the rentier). If we look only at wages, we find that the upper centile’s share remains almost totally stable over the long run at around 6 or 7 percent of total wages. On the eve of World War I, income inequality (as measured by the share of the upper centile) was nearly three times greater than wage inequality. Today it is a nearly a third higher and largely identical with wage inequality, to the point where one might imagine—incorrectly—that top incomes from capital have virtually disappeared (see Figure 8.2).

To sum up: the reduction of inequality in France during the twentieth century is largely explained by the fall of the rentier and the collapse of very high incomes from capital. No generalized structural process of inequality compression (and particularly wage inequality compression) seems to have operated over the long run, contrary to the optimistic predictions of Kuznets’s theory.

Herein lies a fundamental lesson about the historical dynamics of the distribution of wealth, no doubt the most important lesson the twentieth century has to teach. This is all the more true when we recognize that the factual picture is more or less the same in all developed countries, with minor variations.

The History of Inequality: A Chaotic Political History

The third important fact to emerge from Figures 8.1 and 8.2 is that the history of inequality has not been a long, tranquil river. There have been many twists and turns and certainly no irrepressible, regular tendency toward a “natural” equilibrium. In France and elsewhere, the history of inequality has always been chaotic and political, influenced by convulsive social changes and driven not only by economic factors but by countless social, political, military, and cultural phenomena as well. Socioeconomic inequalities—disparities of income and wealth between social groups—are always both causes and effects of other developments in other spheres. All these dimensions of analysis are inextricably intertwined. Hence the history of the distribution of wealth is one way of interpreting a country’s history more generally.

In the case of France, it is striking to see the extent to which the compression of income inequality is concentrated in one highly distinctive period: 1914–1945. The shares of both the upper decile and upper centile in total income reached their nadir in the aftermath of World War II and seem never to have recovered from the extremely violent shocks of the war years (see Figures 8.1 and 8.2). To a large extent, it was the chaos of war, with its attendant economic and political shocks, that reduced inequality in the twentieth century. There was no gradual, consensual, conflict-free evolution toward greater equality. In the twentieth century it was war, and not harmonious democratic or economic rationality, that erased the past and enabled society to begin anew with a clean slate.

What were these shocks? I discussed them in Part Two: destruction caused by two world wars, bankruptcies caused by the Great Depression, and above all new public policies enacted in this period (from rent control to nationalizations and the inflation-induced euthanasia of the rentier class that lived on government debt). All of these things led to a sharp drop in the capital/income ratio between 1914 and 1945 and a significant decrease in the share of income from capital in national income. But capital is far more concentrated than labor, so income from capital is substantially overrepresented in the upper decile of the income hierarchy (even more so in the upper centile). Hence there is nothing surprising about the fact that the shocks endured by capital, especially private capital, in the period 1914–1945 diminished the share of the upper decile (and upper centile), ultimately leading to a significant compression of income inequality.

France first imposed a tax on income in 1914 (the Senate had blocked this reform since the 1890s, and it was not finally adopted until July 15, 1914, a few weeks before war was declared, in an extremely tense climate). For that reason, we unfortunately have no detailed annual data on the structure of income before that date. In the first decade of the twentieth century, numerous estimates were made of the distribution of income in anticipation of the imposition of a general income tax, in order to predict how much revenue such a tax might bring in. We therefore have a rough idea of how concentrated income was in the Belle Époque. But these estimates are not sufficient to give us historical perspective on the shock of World War I (for that, the income tax would have to have been adopted several decades earlier).4 Fortunately, data on estate taxes, which have been levied since 1791, allow us to study the evolution of the wealth distribution throughout the nineteenth and twentieth centuries, and we are therefore able to confirm the central role played by the shocks of 1914–1945. For these data indicate that on the eve of World War I, nothing presaged a spontaneous reduction of the concentration of capital ownership—on the contrary. From the same source we also know that income from capital accounted for the lion’s share of the upper centile’s income in the period 1900–1910.

FIGURE 8.3. The composition of top incomes in France in 1932

Labor income becomes less and less important as one goes up within the top decile of total income. Notes: (i) “P90–95” includes individuals between percentiles 90 to 95, “P95–99” includes the next 4 percent, “P99–99.5” the next 0.5 percent, etc.; (ii) Labor income: wages, bonuses, pensions. Capital income: dividends, interest, rent. Mixed income: self-employment income.

Sources and series: see piketty.pse.ens.fr/capital21c.

From a “Society of Rentiers” to a “Society of Managers”

In 1932, despite the economic crisis, income from capital still represented the main source of income for the top 0.5 percent of the distribution (see Figure 8.3).5 But when we look at the composition of the top income group today, we find that a profound change has occurred. To be sure, today as in the past, income from labor gradually disappears as one moves higher in the income hierarchy, and income from capital becomes more and more predominant in the top centiles and thousandths of the distribution: this structural feature has not changed. There is one crucial difference, however: today one has to climb much higher in the social hierarchy before income from capital outweighs income from labor. Currently, income from capital exceeds income from labor only in the top 0.1 percent of the income distribution (see Figure 8.4). In 1932, this social group was 5 times larger; in the Belle Époque it was 10 times larger.

FIGURE 8.4. The composition of top incomes in France in 2005

Capital income becomes dominant at the level of the top 0.1 percent in France in 2005, as opposed to the top 0.5 percent in 1932.

Sources and series: see piketty.pse.ens.fr/capital21c.

Make no mistake: this is a significant change. The top centile occupies a very prominent place in any society. It structures the economic and political landscape. This is much less true of the top thousandth.6 Although this is a matter of degree, it is nevertheless important: there are moments when the quantitative becomes qualitative. This change also explains why the share of income going to the upper centile today is barely higher than the upper centile’s share of total wages: income from capital assumes decisive importance only in the top thousandth or top ten-thousandth. Its influence in the top centile as a whole is relatively insignificant.

To a large extent, we have gone from a society of rentiers to a society of managers, that is, from a society in which the top centile is dominated by rentiers (people who own enough capital to live on the annual income from their wealth) to a society in which the top of the income hierarchy, including to upper centile, consists mainly of highly paid individuals who live on income from labor. One might also say, more correctly (if less positively), that we have gone from a society of superrentiers to a less extreme form of rentier society, with a better balance between success through work and success through capital. It is important, however, to be clear that this major upheaval came about, in France at any rate, without any expansion of the wage hierarchy (which has been globally stable for a long time: the universe of individuals who are paid for their labor has never been as homogeneous as many people think); it was due entirely to the decrease in high incomes from capital.

To sum up: what happened in France is that rentiers (or at any rate nine-tenths of them) fell behind managers; managers did not race ahead of rentiers. We need to understand the reasons for this long-term change, which are not obvious at first glance, since I showed in Part Two that the capital/income ratio has lately returned to Belle Époque levels. The collapse of the rentier between 1914 and 1945 is the obvious part of the story. Exactly why rentiers have not come back is the more complex and in some ways more important and interesting part. Among the structural factors that may have limited the concentration of wealth since World War II and to this day have helped prevent the resurrection of a society of rentiers as extreme as that which existed on the eve of World War I, we can obviously cite the creation of highly progressive taxes on income and inheritances (which for the most part did not exist prior to 1920). But other factors may also have played a significant and potentially equally important role.

The Different Worlds of the Top Decile

But first, let me dwell a moment on the very diverse social groups that make up the top decile of the income hierarchy. The boundaries between the various subgroups have changed over time: income from capital used to predominate in the top centile but today predominates only in the top thousandth. More than that, the coexistence of several worlds within the top decile can help us to understand the often chaotic short- and medium-term evolutions we see in the data. Income statements required by the new tax laws have proved to be a rich historical source, despite their many imperfections. With their help, it is possible to precisely describe and analyze the diversity at the top of the income distribution and its evolution over time. It is particularly striking to note that in all the countries for which we have this type of data, in all periods, the composition of the top income group can be characterized by intersecting curves like those shown in Figures 8.3 and 8.4 for France in 1932 and 2005, respectively: the share of income from labor always decreases rapidly as one moves progressively higher in the top decile, and the share of income from capital always rises sharply.

In the poorer half of the top decile, we are truly in the world of managers: 80–90 percent of income comes from compensation for labor.7 Moving up to the next 4 percent, the share of income from labor decreases slightly but remains clearly dominant at 70–80 percent of total income in the interwar period as well as today (see Figures 8.3 and 8.4). In this large “9 percent” group (that is, the upper decile exclusive of the top centile), we find mainly individuals living primarily on income from labor, including both private sector managers and engineers and senior officials and teachers from the public sector. Here, pay is usually 2 to 3 times the average wage for society as a whole: if average wages are 2,000 euros a month, in other words, this group earns 4,000–6,000 a month.

Obviously, the types of jobs and levels of skill required at this level have changed considerably over time: in the interwar years, high school teachers and even late-career grade school teachers belonged to “the 9 percent,” whereas today one has to be a college professor or researcher or, better yet, a senior government official to make the grade.8 In the past, a foreman or skilled technician came close to making it into this group. Today one has to be at least a middle manager and increasingly a top manager with a degree from a prestigious university or business school. The same is true lower down the pay scale: once upon a time, the least well paid workers (typically paid about half the average wage, or 1,000 euros a month if the average is 2,000) were farm laborers and domestic servants. At a later point, these were replaced by less skilled industrial workers, many of whom were women in the textile and food processing industries. This group still exists today, but the lowest paid workers are now in the service sector, employed as waiters and waitresses in restaurants or as shop clerks (again, many of these are women). Thus the labor market was totally transformed over the past century, but the structure of wage inequality across the market barely changed over the long run, with “the 9 percent” just below the top and the 50 percent at the bottom still drawing about the same shares of income from labor over a very considerable period of time.

Within “the 9 percent” we also find doctors, lawyers, merchants, restaurateurs, and other self-employed entrepreneurs. Their number grows as we move closer to “the 1 percent,” as is shown by the curve indicating the share of “mixed incomes” (that is, incomes of nonwage workers, which includes both compensation for labor and income from business capital, which I have shown separately in Figures 8.3 and 8.4). Mixed incomes account for 20–30 percent of total income in the neighborhood of the top centile threshold, but this percentage decreases as we move higher into the top centile, where pure capital income (rent, interest, and dividends) clearly predominates. To make it into “the 9 percent” or even rise into the lower strata of “the 1 percent,” which means attaining an income 4–5 times higher than the average (that is, 8,000–10,000 euros a month in a society where the average income is 2,000), choosing to become a doctor, lawyer, or successful restaurateur may therefore be a good strategy, and it is almost as common (actually about half as common) as the choice to become a top manager in a large firm.9 But to reach the stratosphere of “the 1 percent” and enjoy an income several tens of times greater than average (hundreds of thousands if not millions of euros per year), such a strategy is unlikely to be enough. A person who owns substantial amounts of assets is more likely to reach the top of the income hierarchy.10

It is interesting that it was only in the immediate postwar years (1919–1920 in France and then again 1945–1946) that this hierarchy was reversed: mixed incomes very briefly surpassed income from capital in the upper levels of the top centile. This apparently reflects rapid accumulation of new fortunes in connection with postwar reconstruction.11

To sum up: the top decile always encompasses two very different worlds: “the 9 percent,” in which income from labor clearly predominates, and “the 1 percent,” in which income from capital becomes progressively more important (more or less rapidly and massively, depending on the period). The transition between the two groups is always gradual, and the frontiers are of course porous, but the differences are nevertheless clear and systematic.

For example, while income from capital is obviously not altogether absent from the income of “the 9 percent,” it is usually not the main source of income but simply a supplement. A manager earning 4,000 euros a month may also own an apartment that she rents for 1,000 euros a month (or lives in, thus avoiding paying a rent of 1,000 euros a month, which comes to the same thing financially). Her total income is then 5,000 euros a month, 80 percent of which is income from labor and 20 percent from capital. Indeed, an 80–20 split between labor and capital is reasonably representative of the structure of income among “the 9 percent”; this was true between the two world wars and remains true today. A part of this group’s income from capital may also come from savings accounts, life insurance contracts, and financial investments, but real estate generally predominates.12

Conversely, within “the 1 percent,” it is labor income that gradually becomes supplementary, while capital increasingly becomes the main source of income. Another interesting pattern is the following: if we break income from capital down into rent on land and structures on the one hand and dividends and interest from mobile capital on the other, we find that the very large share of income from capital in the upper decile is due largely to the latter (especially dividends). For example, in France, the share of income from capital in 1932 as well as 2005 is 20 percent at the level of “the 9 percent” but increases to 60 percent in the top 0.01 percent. In both cases, this sharp increase is explained entirely by income from financial assets (almost all of it in the form of dividends). The share of rent stagnates at around 10 percent of total income and even tends to diminish in the top centile. This pattern reflects the fact that large fortunes consist primarily of financial assets (mainly stocks and shares in partnerships).

The Limits of Income Tax Returns

Despite all these interesting patterns, I must stress the limits of the fiscal sources used in this chapter. Figures 8.3 and 8.4 are based solely on income from capital reported in tax returns. Actual capital income is therefore underestimated, owing both to tax evasion (it is easier to hide investment income than wages, for example, by using foreign bank accounts in countries that do not cooperate with the country in which the taxpayer resides) and to the existence of various tax exemptions that allow whole categories of capital income to legally avoid the income tax (which in France and elsewhere was originally intended to include all types of income). Since income from capital is overrepresented in the top decile, this underdeclaration of capital income also implies that the shares of the upper decile and centile indicated on Figures 8.1 and 8.2, which are based solely on income tax returns, are underestimated (for France and other countries). These shares are in any case approximate. They are interesting (like all economic and social statistics) mainly as indicators of orders of magnitude and should be taken as low estimates of the actual level of inequality.

In the French case, we can compare self-declared income on tax returns with other sources (such as national accounts and sources that give a more direct measure of the distribution of wealth) to estimate how much we need to adjust our results to compensate for the underdeclaration of capital income. It turns out that we need to add several percentage points to capital income’s share of national income (perhaps as many as 5 percentage points if we choose a high estimate of tax evasion, but more realistically 2 to 3 percentage points). This is not a negligible amount. Put differently, the share of the top decile in national income, which according to Figure 8.1 fell from 45–50 percent in 1900–1910 to 30–35 percent in 2000–2010, was no doubt closer to 50 percent (or even slightly higher) in the Belle Époque and is currently slightly more than 35 percent.13 Nevertheless, this correction does not significantly affect the overall evolution of income inequality. Even if opportunities for legal tax avoidance and illegal tax evasion have increased in recent years (thanks in particular to the emergence of tax havens about which I will say more later on), we must remember that income from mobile capital was already significantly underreported in the early twentieth century and during the interwar years. All signs are that the copies of dividend and interest coupons requested by the governments of that time were no more effective than today’s bilateral agreements as a means of ensuring compliance with applicable tax laws.

To a first approximation, therefore, we may assume that accounting for tax avoidance and evasion would increase the levels of inequality derived from tax returns by similar proportions in different periods and would therefore not substantially modify the time trends and evolutions I have identified.

Note, however, that we have not yet attempted to apply such corrections in a systematic and consistent way in different countries. This is an important limitation of the World Top Incomes Database. One consequence is that our series underestimate—probably slightly—the increase of inequality that can be observed in most countries after 1970, and in particular the role of income from capital. In fact, income tax returns are becoming increasingly less accurate sources for studying capital income, and it is indispensable to make use of other, complementary sources as well. These may be either macroeconomic sources (of the kind used in Part Two to study the dynamics of the capital/income ratio and capital-labor split) or microeconomic sources (with which it is possible to study the distribution of wealth directly, and of which I will make use in subsequent chapters).

Furthermore, different capital taxation laws may bias international comparisons. Broadly speaking, rents, interest, and dividends are treated fairly similarly in different countries.14 By contrast, there are significant variations in the treatment of capital gains. For instance, capital gains are not fully or consistently reported in French tax data (and I have simply excluded them altogether), while they have always been fairly well accounted for in US tax data. This can make a major difference, because capital gains, especially those realized from the sale of stocks, constitute a form of capital income that is highly concentrated in the very top income groups (in some cases even more than dividends). For example, if Figures 8.3 and 8.4 included capital gains, the share of income from capital in the top ten-thousandth would not be 60 percent but something closer to 70 or 80 percent (depending on the year).15 So as not to bias comparisons, I will present the results for the United States both with and without capital gains.

The other important limitation of income tax returns is that they contain no information about the origin of the capital whose income is being reported. We can see the income produced by capital owned by the taxpayer at a particular moment in time, but we have no idea whether that capital was inherited or accumulated by the taxpayer during his or her lifetime with income derived from labor (or from other capital). In other words, an identical level of inequality with respect to income from capital can in fact reflect very different situations, and we would never learn anything about these differences if we restricted ourselves to tax return data. Generally speaking, very high incomes from capital usually correspond to fortunes so large that it is hard to imagine that they could have been amassed with savings from labor income alone (even in the case of a very high-level manager or executive). There is every reason to believe that inheritance plays a major role. As we will see in later chapters, however, the relative importance of inheritance and saving has evolved considerably over time, and this is a subject that deserves further study. Once again, I will need to make use of sources bearing directly on the question of inheritance.

The Chaos of the Interwar Years

Consider the evolution of income inequality in France over the last century. Between 1914 and 1945, the share of the top centile of the income hierarchy fell almost constantly, dropping gradually from 20 percent in 1914 to just 7 percent in 1945 (Figure 8.2). This steady decline reflects the long and virtually uninterrupted series of shocks sustained by capital (and income from capital) during this time. By contrast, the share of the top decile of the income hierarchy decreased much less steadily. It apparently fell during World War I, but this was followed by an unsteady recovery in the 1920s and then a very sharp, and at first sight surprising, rise between 1929 and 1935, followed by a steep decline in 1936–1938 and a collapse during World War II.16 In the end, the top decile’s share of national income, which was more than 45 percent in 1914, fell to less than 30 percent in 1944–1945.

If we consider the entire period 1914–1945, the two declines are perfectly consistent: the share of the upper decile decreased by nearly 18 points, according to my estimates, and the upper centile by nearly 14 points.17 In other words, “the 1 percent” by itself accounts for roughly three-quarters of the decrease in inequality between 1914 and 1945, while “the 9 percent” explains roughly one-quarter. This is hardly surprising in view of the extreme concentration of capital in the hands of “the 1 percent,” who in addition often held riskier assets.

By contrast, the differences observed during this period are at first sight more surprising: Why did the share of the upper decile rise sharply after the crash of 1929 and continue at least until 1935, while the share of the top centile fell, especially between 1929 and 1932?

In fact, when we look at the data more closely, year by year, each of these variations has a perfectly good explanation. It is enlightening to revisit the chaotic interwar period, when social tensions ran very high. To understand what happened, we must recognize that “the 9 percent” and “the 1 percent” lived on very different income streams. Most of the income of “the 1 percent” came in the form of income from capital, especially interest and dividends paid by the firms whose stocks and bonds made up the assets of this group. That is why the top centile’s share plummeted during the Depression, as the economy collapsed, profits fell, and firm after firm went bankrupt.

By contrast, “the 9 percent” included many managers, who were the great beneficiaries of the Depression, at least when compared with other social groups. They suffered much less from unemployment than the employees who worked under them. In particular, they never experienced the extremely high rates of partial or total unemployment endured by industrial workers. They were also much less affected by the decline in company profits than those who stood above them in the income hierarchy. Within “the 9 percent,” midlevel civil servants and teachers fared particularly well. They had only recently been the beneficiaries of civil service raises granted in the period 1927–1931. (Recall that government workers, particularly those at the top of the pay scale, had suffered greatly during World War I and had been hit hard by the inflation of the early 1920s.) These midlevel employees were immune, too, from the risk of unemployment, so that the public sector’s wage bill remained constant in nominal terms until 1933 (and decreased only slightly in 1934–1935, when Prime Minister Pierre Laval sought to cut civil service pay). Meanwhile, private sector wages decreased by more than 50 percent between 1929 and 1935. The severe deflation France suffered in this period (prices fell by 25 percent between 1929 and 1935, as both trade and production collapsed) played a key role in the process: individuals lucky enough to hold on to their jobs and their nominal compensation—typically civil servants—enjoyed increased purchasing power in the midst of the Depression as falling prices raised their real wages. Furthermore, such capital income as “the 9 percent” enjoyed—typically in the form of rents, which were extremely rigid in nominal terms—also increased on account of the deflation, so that the real value of this income stream rose significantly, while the dividends paid to “the 1 percent” evaporated.

For all these reasons, the share of national income going to “the 9 percent” increased quite significantly in France between 1929 and 1935, much more than the share of “the 1 percent” decreased, so that the share of the upper decile as a whole increased by more than 5 percent of national income (see Figures 8.1 and 8.2). The process was completely turned around, however, when the Popular Front came to power: workers’ wages increased sharply as a result of the Matignon Accords, and the franc was devalued in September 1936, resulting in inflation and a decrease of the shares of both “the 9 percent” and the top decile in 1936–1938.18

The foregoing discussion demonstrates the usefulness of breaking income down by centiles and income source. If we had tried to analyze the interwar dynamic by using a synthetic index such as the Gini coefficient, it would have been impossible to understand what was going on. We would not have been able to distinguish between income from labor and income from capital or between short-term and long-term changes. In the French case, what makes the period 1914–1945 so complex is the fact that although the general trend is fairly clear (a sharp drop in the share of national income going to the top decile, induced by a collapse of the top centile’s share), many smaller counter-movements were superimposed on this overall pattern in the 1920s and 1930s. We find similar complexity in other countries in the interwar period, with characteristic features associated with the history of each particular country. For example, deflation ended in the United States in 1933, when President Roosevelt came to power, so that the reversal that occurred in France in 1936 came earlier in America, in 1933. In every country the history of inequality is political—and chaotic.

The Clash of Temporalities

Broadly speaking, it is important when studying the dynamics of the income and wealth distributions to distinguish among several different time scales. In this book I am primarily interested in long-term evolutions, fundamental trends that in many cases cannot be appreciated on time scales of less than thirty to forty years or even longer, as shown, for example, by the structural increase in the capital/income ratio in Europe since World War II, a process that has been going on for nearly seventy years now yet would have been difficult to detect just ten or twenty years ago owing to the superimposition of various other developments (as well as the absence of usable data). But this focus on the long period must not be allowed to obscure the fact that shorter-term trends also exist. To be sure, these are often counterbalanced in the end, but for the people who live through them they often appear, quite legitimately, to be the most significant realities of the age. Indeed, how could it be otherwise, when these “short-term” movements can continue for ten to fifteen years or even longer, which is quite long when measured on the scale of a human lifetime.

The history of inequality in France and elsewhere is replete with these short- and medium-term movements—and not just in the particularly chaotic interwar years. Let me briefly recount the major episodes in the case of France. During both world wars, the wage hierarchy was compressed, but in the aftermath of each war, wage inequalities reasserted themselves (in the 1920s and then again in the late 1940s and on into the 1950s and 1960s). These were movements of considerable magnitude: the share of total wages going to the top 10 percent decreased by about 5 points during each conflict but recovered afterward by the same amount (see Figure 8.1).19 Wage spreads were reduced in the public as well as the private sector. In each war the scenario was the same: in wartime, economic activity decreases, inflation increases, and real wages and purchasing power begin to fall. Wages at the bottom of the wage scale generally rise, however, and are somewhat more generously protected from inflation than those at the top. This can induce significant changes in the wage distribution if inflation is high. Why are low and medium wages better indexed to inflation than higher wages? Because workers share certain perceptions of social justice and norms of fairness, an effort is made to prevent the purchasing power of the least well-off from dropping too sharply, while their better-off comrades are asked to postpone their demands until the war is over. This phenomenon clearly played a role in setting wage scales in the public sector, and it was probably the same, at least to a certain extent, in the private sector. The fact that large numbers of young and relatively unskilled workers were mobilized for service (or held in prisoner-of-war camps) may also have improved the relative position of low- and medium-wage workers on the labor market.

In any case, the compression of wage inequality was reversed in both postwar periods, and it is therefore tempting to forget that it ever occurred. Nevertheless, for workers who lived through these periods, the changes in the wage distribution made a deep impression. In particular, the issue of restoring the wage hierarchy in both the public and private sectors was one of the most important political, social, and economic issues of the postwar years.

Turning now to the history of inequality in France between 1945 and 2010, we find three distinct phases: income inequality rose sharply between 1945 and 1967 (with the share going to the top decile increasing from less than 30 to 36 or 37 percent). It then decreased considerably between 1968 and 1983 (with the share of the top decile dropping back to 30 percent). Finally, inequality increased steadily after 1983, so that the top decile’s share climbed to about 33 percent in the period 2000–2010 (see Figure 8.1). We find roughly similar changes of wage inequality at the level of the top centile (see Figures 8.3 and 8.3). Once again, these various increases and decreases more or less balance out, so it is tempting to ignore them and concentrate on the relative stability over the long run, 1945–2010. Indeed, if one were interested solely in very long-term evolutions, the outstanding change in France during the twentieth century would be the significant compression of wage inequality between 1914 and 1945, followed by relative stability afterward. Each way of looking at the matter is legitimate and important in its own right, and to my mind it is essential to keep all of these different time scales in mind: the long term is important, but so are the short and the medium term. I touched on this point previously in my examination of the evolution of the capital/income ratio and the capital-labor split in Part Two (see in particular Chapter 6).

It is interesting to note that the capital-labor split tends to move in the same direction as inequality in income from labor, so that the two reinforce each other in the short to medium term but not necessarily in the long run. For example, each of the two world wars saw a decrease in capital’s share of national income (and of the capital/income ratio) as well as a compression of wage inequality. Generally speaking, inequality tends to evolve “procyclically” (that is, it moves in the same direction as the economic cycle, in contrast to “countercyclical” changes). In economic booms, the share of profits in national income tends to increase, and pay at the top end of the scale (including incentives and bonuses) often increases more than wages toward the bottom and middle. Conversely, during economic slowdowns or recessions (of which war can be seen as an extreme form), various noneconomic factors, especially political ones, ensure that these movements do not depend solely on the economic cycle.

The substantial increase in French inequality between 1945 and 1967 was the result of sharp increases in both capital’s share of national income and wage inequality in a context of rapid economic growth. The political climate undoubtedly played a role: the country was entirely focused on reconstruction, and decreasing inequality was not a priority, especially since it was common knowledge that inequality had decreased enormously during the war. In the 1950s and 1960s, managers, engineers, and other skilled personnel saw their pay increase more rapidly than the pay of workers at the bottom and middle of the wage hierarchy, and at first no one seemed to care. A national minimum wage was created in 1950 but was seldom increased thereafter and fell farther and farther behind the average wage.

Things changed suddenly in 1968. The events of May 1968 had roots in student grievances and cultural and social issues that had little to do with the question of wages (although many people had tired of the inegalitarian productivist growth model of the 1950s and 1960s, and this no doubt played a role in the crisis). But the most immediate political result of the movement was its effect on wages: to end the crisis, Charles de Gaulle’s government signed the Grenelle Accords, which provided, among other things, for a 20 percent increase in the minimum wage. In 1970, the minimum wage was officially (if partially) indexed to the mean wage, and governments from 1968 to 1983 felt obliged to “boost” the minimum significantly almost every year in a seething social and political climate. The purchasing power of the minimum wage accordingly increased by more than 130 percent between 1968 and 1983, while the mean wage increased by only about 50 percent, resulting in a very significant compression of wage inequalities. The break with the previous period was sharp and substantial: the purchasing power of the minimum wage had increased barely 25 percent between 1950 and 1968, while the average wage had more than doubled.20 Driven by the sharp rise of low wages, the total wage bill rose markedly more rapidly than output between 1968 and 1983, and this explains the sharp decrease in capital’s share of national income that I pointed out in Part Two, as well as the very substantial compression of income inequality.

These movements reversed in 1982–1983. The new Socialist government elected in May 1981 surely would have preferred to continue the earlier trend, but it was not a simple matter to arrange for the minimum wage to increase twice as fast as the average wage (especially when the average wage itself was increasing faster than output). In 1982–1983, therefore, the government decided to “turn toward austerity”: wages were frozen, and the policy of annual boosts to the minimum wage was definitively abandoned. The results were soon apparent: the share of profits in national income skyrocketed during the remainder of the 1980s, while wage inequalities once again increased, and income inequalities even more so (see Figures 8.1 and 8.2). The break was as sharp as that of 1968, but in the other direction.

The Increase of Inequality in France since the 1980s

How should we characterize the phase of increasing inequality that began in France in 1982–1983? It is tempting to see it in a long-run perspective as a microphenomenon, a simple reversal of the previous trend, especially since by 1990 or so the share of profits in national income had returned to the level achieved on the eve of May 1968.21 This would be a mistake, however, for several reasons. First, as I showed in Part Two, the profit share in 1966–1967 was historically high, a consequence of the restoration of capital’s share that began at the end of World War II. If we include, as we should, rent as well as profit in income from capital, we find that capital’s share of national income actually continued to grow in the 1990s and 2000s. A correct understanding of this long-run phenomenon requires that it be placed in the context of the long-term evolution of the capital/income ratio, which by 2010 had returned to virtually the same level it had achieved in France on the eve of World War I. It is impossible to fully appreciate the implications of this restoration of the prosperity of capital simply by looking at the evolution of the upper decile’s share of income, in part because income from capital is understated, so that we tend to slightly underestimate the increase in top incomes, and in part because the real issue is the renewed importance of inherited wealth, a long-term process that has only begun to reveal its true effects and can be correctly analyzed only by directly studying the changing role and importance of inherited wealth as such.

But that is not all. A stunning new phenomenon emerged in France in the 1990s: the very top salaries, and especially the pay packages awarded to the top executives of the largest companies and financial firms, reached astonishing heights—somewhat less astonishing in France, for the time being, than in the United States, but still, it would be wrong to neglect this new development. The share of wages going to the top centile, which was less than 6 percent in the 1980s and 1990s, began to increase in the late 1990s and reached 7.5–8 percent of the total by the early 2010s. Thus there was an increase of nearly 30 percent in a little over a decade, which is far from negligible. If we move even higher up the salary and bonus scale to look at the top 0.1 or 0.01 percent, we find even greater increases, with hikes in purchasing power greater than 50 percent in ten years.22 In a context of very low growth and virtual stagnation of purchasing power for the vast majority of workers, raises of this magnitude for top earners have not failed to attract attention. Furthermore, the phenomenon was radically new, and in order to interpret it correctly, we must view it in international perspective.

FIGURE 8.5. Income inequality in the United States, 1910–2010

The top decile income share rose from less than 35 percent of total income in the 1970s to almost 50 percent in the 2000s–2010s.

Sources and series: see piketty.pse.ens.fr/capital21c.

A More Complex Case: The Transformation of Inequality in the United States

Indeed, let me turn now to the US case, which stands out precisely because it was there that a subclass of “supermanagers” first emerged over the past several decades. I have done everything possible to ensure that the data series for the United States are as comparable as possible with the French series. In particular, Figures 8.5 and 8.6 represent the same data for the United States as Figures 8.1 and 8.2 for France: the goal is to compare, in the first figure of each pair, the evolution of the shares of income going to the top decile and top centile of the wage hierarchy and to compare, in the second figure, the wage hierarchies themselves. I should add that the United States first instituted a federal income tax in 1913, concluding a long battle with the Supreme Court.23 The data derived from US income tax returns are on the whole quite comparable to the French data, though somewhat less detailed. In particular, total income can be gleaned from US statements from 1913 on, but we do not have separate information on income from labor until 1927, so the series dealing with the wage distribution in the United States before 1927 are somewhat less reliable.24

FIGURE 8.6. Decomposition of the top decile, United States, 1910–2010

The rise of the top decile income share since the 1970s is mostly due to the top percentile.

Sources and series: see piketty.pse.ens.fr/capital21c.

When we compare the French and US trajectories, a number of similarities stand out, but so do certain important differences. I shall begin by examining the overall evolution of the share of income going to the top decile (Figure 8.6). The most striking fact is that the United States has become noticeably more inegalitarian than France (and Europe as a whole) from the turn of the twentieth century until now, even though the United States was more egalitarian at the beginning of this period. What makes the US case complex is that the end of the process did not simply mark a return to the situation that had existed at the beginning: US inequality in 2010 is quantitatively as extreme as in old Europe in the first decade of the twentieth century, but the structure of that inequality is rather clearly different.

I will proceed systematically. First, European income inequality was significantly greater than US income inequality at the turn of the twentieth century. In 1900–1910, according to the data at our disposal, the top decile of the income hierarchy received a little more than 40 percent of total national income in the United States, compared with 45–50 percent in France (and very likely somewhat more in Britain). This reflects two differences. First, the capital/income ratio was higher in Europe, and so was capital’s share of national income. Second, inequality of ownership of capital was somewhat less extreme in the New World. Clearly, this does not mean that American society in 1900–1910 embodied the mythical ideal of an egalitarian society of pioneers. In fact, American society was already highly inegalitarian, much more than Europe today, for example. One has only to reread Henry James or note that the dreadful Hockney who sailed in luxury on Titanic in 1912 existed in real life and not just in the imagination of James Cameron to convince oneself that a society of rentiers existed not only in Paris and London but also in turn-of-the-century Boston, New York, and Philadelphia. Nevertheless, capital (and therefore the income derived from it) was distributed somewhat less unequally in the United States than in France or Britain. Concretely, US rentiers were fewer in number and not as rich (compared to the average US standard of living) as their European counterparts. I will need to explain why this was so.

Income inequality increased quite sharply in the United States during the 1920s, however, peaking on the eve of the 1929 crash with more than 50 percent of national income going to the top decile—a level slightly higher than in Europe at the same time, as a result of the substantial shocks to which European capital had already been subjected since 1914. Nevertheless, US inequality was not the same as European inequality: note the already crucial importance of capital gains in top US incomes during the heady stock market ascent of the 1920s (see Figure 8.5).

During the Great Depression, which hit the United States particularly hard, and again during World War II, when the nation was fully mobilized behind the war effort (and the effort to end the economic crisis), income inequality was substantially compressed, a compression comparable in some respects to what we observe in Europe in the same period. Indeed, as we saw in Part Two, the shocks to US capital were far from negligible: although there was no physical destruction due to war, the Great Depression was a major shock and was followed by substantial tax shocks imposed by the federal government in the 1930s and 1940s. If we look at the period 1910–1950 as a whole, however, we find that the compression of inequality was noticeably smaller in the United States than in France (and, more generally, Europe). To sum up: inequality in the United States started from a lower peak on the eve of World War I but at its low point after World War II stood above inequality in Europe. Europe in 1914–1945 witnessed the suicide of rentier society, but nothing of the sort occurred in the United States.

The Explosion of US Inequality after 1980

Inequality reached its lowest ebb in the United States between 1950 and 1980: the top decile of the income hierarchy claimed 30 to 35 percent of US national income, or roughly the same level as in France today. This is what Paul Krugman nostalgically refers to as “the America we love”—the America of his childhood.25 In the 1960s, the period of the TV series Mad Men and General de Gaulle, the United States was in fact a more egalitarian society than France (where the upper decile’s share had increased dramatically to well above 35 percent), at least for those US citizens whose skin was white.

Since 1980, however, income inequality has exploded in the United States. The upper decile’s share increased from 30–35 percent of national income in the 1970s to 45–50 percent in the 2000s—an increase of 15 points of national income (see Figure 8.5). The shape of the curve is rather impressively steep, and it is natural to wonder how long such a rapid increase can continue: if change continues at the same pace, for example, the upper decile will be raking in 60 percent of national income by 2030.

It is worth taking a moment to clarify several points about this evolution. First, recall that the series represented in Figure 8.5, like all the series in the WTID, take account only of income declared in tax returns and in particular do not correct for any possible understatement of capital income for legal or extralegal reasons. Given the widening gap between the total capital income (especially dividends and interest) included in US national accounts and the amount declared in income tax returns, and given, too, the rapid development of tax havens (flows to which are, in all likelihood, mostly not even included in national accounts), it is likely that Figure 8.5 underestimates the amount by which the upper decile’s share actually increased. By comparing various available sources, it is possible to estimate that the upper decile’s share slightly exceeded 50 percent of US national income on the eve of the financial crisis of 2008 and then again in the early 2010s.26

Note, moreover, that stock market euphoria and capital gains can account for only part of the structural increase in the top decile’s share over the past thirty or forty years. To be sure, capital gains in the United States reached unprecedented heights during the Internet bubble in 2000 and again in 2007: in both cases, capital gains alone accounted for about five additional points of national income for the upper decile, which is an enormous amount. The previous record, set in 1928 on the eve of the 1929 stock market crash, was roughly 3 points of national income. But such levels cannot be sustained for very long, as the large annual variations evident in Figure 8.5 show. The incessant short-term fluctuations of the stock market add considerable volatility to the evolution of the upper decile’s share (and certainly contribute to the volatility of the US economy as a whole) but do not contribute much to the structural increase of inequality. If we simply ignore capital gains (which is not a satisfactory method either, given the importance of this type of remuneration in the United States), we still find almost as great an increase in the top decile’s share, which rose from around 32 percent in the 1970s to more than 46 percent in 2010, or fourteen points of national income (see Figure 8.5). Capital gains oscillated around one or two points of additional national income for the top decile in the 1970s and around two to three points between 2000 and 2010 (excluding exceptionally good and bad years). The structural increase is therefore on the order of one point: this is not nothing, but then again it is not much compared with the fourteen-point increase of the top decile’s share exclusive of capital gains.27

Looking at evolutions without capital gains also allows us to identify the structural character of the increase of inequality in the United States more clearly. In fact, from the late 1970s to 2010, the increase in the upper decile’s share (exclusive of capital gains) appears to have been relatively steady and constant: it passed 35 percent in the 1980s, then 40 percent in the 1990s, and finally 45 percent in the 2000s (see Figure 8.5).28 Much more striking is the fact that the level attained in 2010 (with more than 46 percent of national income, exclusive of capital gains, going to the top decile) is already significantly higher than the level attained in 2007, on the eve of the financial crisis. Early data for 2011–2012 suggest that the increase is still continuing.

This is a crucial point: the facts show quite clearly that the financial crisis as such cannot be counted on to put an end to the structural increase of inequality in the United States. To be sure, in the immediate aftermath of a stock market crash, inequality always grows more slowly, just as it always grows more rapidly in a boom. The years 2008–2009, following the collapse of Lehman Brothers, like the years 2001–2002, after the bursting of the first Internet bubble, were not great times for taking profits on the stock market. Indeed, capital gains plummeted in those years. But these short-term movements did not alter the long-run trend, which is governed by other forces whose logic I must now try to clarify.

To proceed further, it will be useful to break the top decile of the income hierarchy down into three groups: the richest 1 percent, the next 4 percent, and the bottom 5 percent (see Figure 8.6). The bulk of the growth of inequality came from “the 1 percent,” whose share of national income rose from 9 percent in the 1970s to about 20 percent in 2000–2010 (with substantial year-to-year variation due to capital gains)—an increase of 11 points. To be sure, “the 5 percent” (whose annual income ranged from $108,000 to $150,000 per household in 2010) as well as “the 4 percent” (whose income ranged from $150,000 to $352,000) also experienced substantial increases: the share of the former in US national income rose from 11 to 12 percent (or one point), and that of the latter rose from 13 to 16 percent (three points).29 By definition, that means that since 1980, these social groups have experienced income growth substantially higher than the average growth of the US economy, which is not negligible.

Among the members of these upper income groups are US academic economists, many of whom believe that the economy of the United States is working fairly well and, in particular, that it rewards talent and merit accurately and precisely. This is a very comprehensible human reaction.30 But the truth is that the social groups above them did even better: of the 15 additional points of national income going to the top decile, around 11 points, or nearly three-quarters of the total, went to “the 1 percent” (those making more than $352,000 a year in 2010), of which roughly half went to “the 0.1 percent” (those making more than $1.5 million a year).31

Did the Increase of Inequality Cause the Financial Crisis?

As I have just shown, the financial crisis as such seems not to have had an impact on the structural increase of inequality. What about the reverse causality? Is it possible that the increase of inequality in the United States helped to trigger the financial crisis of 2008? Given the fact that the share of the upper decile in US national income peaked twice in the past century, once in 1928 (on the eve of the crash of 1929) and again in 2007 (on the eve of the crash of 2008), the question is difficult to avoid.

In my view, there is absolutely no doubt that the increase of inequality in the United States contributed to the nation’s financial instability. The reason is simple: one consequence of increasing inequality was virtual stagnation of the purchasing power of the lower and middle classes in the United States, which inevitably made it more likely that modest households would take on debt, especially since unscrupulous banks and financial intermediaries, freed from regulation and eager to earn good yields on the enormous savings injected into the system by the well-to-do, offered credit on increasingly generous terms.32

In support of this thesis, it is important to note the considerable transfer of US national income—on the order of 15 points—from the poorest 90 percent to the richest 10 percent since 1980. Specifically, if we consider the total growth of the US economy in the thirty years prior to the crisis, that is, from 1977 to 2007, we find that the richest 10 percent appropriated three-quarters of the growth. The richest 1 percent alone absorbed nearly 60 percent of the total increase of US national income in this period. Hence for the bottom 90 percent, the rate of income growth was less than 0.5 percent per year.33 These figures are incontestable, and they are striking: whatever one thinks about the fundamental legitimacy of income inequality, the numbers deserve close scrutiny.34 It is hard to imagine an economy and society that can continue functioning indefinitely with such extreme divergence between social groups.

Quite obviously, if the increase in inequality had been accompanied by exceptionally strong growth of the US economy, things would look quite different. Unfortunately, this was not the case: the economy grew rather more slowly than in previous decades, so that the increase in inequality led to virtual stagnation of low and medium incomes.

Note, too, that this internal transfer between social groups (on the order of fifteen points of US national income) is nearly four times larger than the impressive trade deficit the United States ran in the 2000s (on the order of four points of national income). The comparison is interesting because the enormous trade deficit, which has its counterpart in Chinese, Japanese, and German trade surpluses, has often been described as one of the key contributors to the “global imbalances” that destabilized the US and global financial system in the years leading up to the crisis of 2008. That is quite possible, but it is important to be aware of the fact that the United States’ internal imbalances are four times larger than its global imbalances. This suggests that the place to look for the solutions of certain problems may be more within the United States than in China or other countries.

That said, it would be altogether too much to claim that the increase of inequality in the United States was the sole or even primary cause of the financial crisis of 2008 or, more generally, of the chronic instability of the global financial system. To my mind, a potentially more important cause of instability is the structural increase of the capital/income ratio (especially in Europe), coupled with an enormous increase in aggregate international asset positions.35

The Rise of Supersalaries

Let me return now to the causes of rising inequality in the United States. The increase was largely the result of an unprecedented increase in wage inequality and in particular the emergence of extremely high remunerations at the summit of the wage hierarchy, particularly among top managers of large firms (see Figures 8.7 and 8.8).

Broadly speaking, wage inequality in the United States changed in major ways over the past century: the wage hierarchy expanded in the 1920s, was relatively stable in the 1930s, and then experienced severe compression during World War II. The phase of “severe compression” has been abundantly studied. An important role was played by the National War Labor Board, the government agency that had to approve all wage increases in the United States from 1941 to 1945 and generally approved raises only for the lowest paid workers. In particular, managers’ salaries were systematically frozen in nominal terms and even at the end of the war were raised only moderately.36 During the 1950s, wage inequality in the United States stabilized at a relatively low level, lower than in France, for example: the share of income going to the upper decile was about 25 percent, and the share of the upper centile was 5 or 6 percent. Then, from the mid-1970s on, the top 10 percent and, even more, the top 1 percent began to claim a share of labor income that grew more rapidly than the average wage. All told, the upper decile’s share rose from 25 to 35 percent, and this increase of ten points explains approximately two-thirds of the increase in the upper decile’s share of total national income (see Figures 8.7 and 8.8).

FIGURE 8.7. High incomes and high wages in the United States, 1910–2010

The rise of income inequality since the 1970s is largely due to the rise of wage inequality.

Sources and series: see piketty.pse.ens.fr/capital21c.

Several points call for additional comment. First, this unprecedented increase in wage inequality does not appear to have been compensated by increased wage mobility over the course of a person’s career.37 This is a significant point, in that greater mobility is often mentioned as a reason to believe that increasing inequality is not that important. In fact, if each individual were to enjoy a very high income for part of his or her life (for example, if each individual spent a year in the upper centile of the income hierarchy), then an increase in the level characterized as “very high pay” would not necessarily imply that inequality with respect to labor—measured over a lifetime—had truly increased. The familiar mobility argument is powerful, so powerful that it is often impossible to verify. But in the US case, government data allow us to measure the evolution of wage inequality with mobility taken into account: we can compute average wages at the individual level over long periods of time (ten, twenty, or thirty years). And what we find is that the increase in wage inequality is identical in all cases, no matter what reference period we choose.38 In other words, workers at McDonald’s or in Detroit’s auto plants do not spend a year of their lives as top managers of large US firms, any more than professors at the University of Chicago or middle managers from California do. One may have felt this intuitively, but it is always better to measure systematically wherever possible.

FIGURE 8.8. The transformation of the top 1 percent in the United States

The rise in the top 1 percent highest incomes since the 1970s is largely due to the rise in the top 1 percent highest wages.

Sources and series: see piketty.pse.ens.fr/capital21c.

Cohabitation in the Upper Centile

Furthermore, the fact that the unprecedented increase of wage inequality explains most of the increase in US income inequality does not mean that income from capital played no role. It is important to dispel the notion that capital income has vanished from the summit of the US social hierarchy.

In fact, a very substantial and growing inequality of capital income since 1980 accounts for about one-third of the increase in income inequality in the United States—a far from negligible amount. Indeed, in the United States, as in France and Europe, today as in the past, income from capital always becomes more important as one climbs the rungs of the income hierarchy. Temporal and spatial differences are differences of degree: though large, the general principle remains. As Edward Wolff and Ajit Zacharias have pointed out, the upper centile always consists of several different social groups, some with very high incomes from capital and others with very high incomes from labor; the latter do not supplant the former.39

FIGURE 8.9. The composition of top incomes in the United States in 1929

Labor income becomes less and less important as one moves up within the top income decile.

Sources and series: see piketty.pse.ens.fr/capital21c.

In the US case, as in France but to an even greater degree, the difference today is that one has to climb much further up the income hierarchy before income from capital takes the upper hand. In 1929, income from capital (essentially dividends and capital gains) was the primary resource for the top 1 percent of the income hierarchy (see Figure 8.9). In 2007, one has to climb to the 0.1 percent level before this is true (see Figure 8.10). Again, I should make it clear that this has to do with the inclusion of capital gains in income from capital: without capital gains, salaries would be the main source of income up to the 0.01 percent level of the income hierarchy.40

FIGURE 8.10. The composition of top incomes in the United States, 2007

Capital income becomes dominant at the level of top 0.1 percent in 2007, as opposed to the top 1 percent in 1929.

Sources and series: see piketty.pse.ens.fr/capital21c.

The final and perhaps most important point in need of clarification is that the increase in very high incomes and very high salaries primarily reflects the advent of “supermanagers,” that is, top executives of large firms who have managed to obtain extremely high, historically unprecedented compensation packages for their labor. If we look only at the five highest paid executives in each company listed on the stock exchange (which are generally the only compensation packages that must be made public in annual corporate reports), we come to the paradoxical conclusion that there are not enough top corporate managers to explain the increase in very high US incomes, and it therefore becomes difficult to explain the evolutions we observe in incomes stated on federal income tax returns.41 But the fact is that in many large US firms, there are far more than five executives whose pay places them in the top 1 percent (above $352,000 in 2010) or even the top 0.1 percent (above $1.5 million).

Recent research, based on matching declared income on tax returns with corporate compensation records, allows me to state that the vast majority (60 to 70 percent, depending on what definitions one chooses) of the top 0.1 percent of the income hierarchy in 2000–2010 consists of top managers. By comparison, athletes, actors, and artists of all kinds make up less than 5 percent of this group.42 In this sense, the new US inequality has much more to do with the advent of “supermanagers” than with that of “superstars.”43

It is also interesting to note that the financial professions (including both managers of banks and other financial institutions and traders operating on the financial markets) are about twice as common in the very high income groups as in the economy overall (roughly 20 percent of top 0.1 percent, whereas finance accounts for less than 10 percent of GDP). Nevertheless, 80 percent of the top income groups are not in finance, and the increase in the proportion of high-earning Americans is explained primarily by the skyrocketing pay packages of top managers of large firms in the nonfinancial as well as financial sectors.

Finally, note that in accordance with US tax laws as well as economic logic, I have included in wages all bonuses and other incentives paid to top managers, as well as the value of any stock options (a form of remuneration that has played an important role in the increase of wage inequality depicted in Figures 8.9 and 8.10).44 The very high volatility of incentives, bonuses, and option prices explains why top incomes fluctuated so much in the period 2000–2010.

{NINE}

Inequality of Labor Income

Now that I have introduced the evolution of income and wages in France and the United States since the beginning of the twentieth century, I will examine the changes I have observed and consider how representative they are of long-term changes in other developed and emerging economies.

I will begin by examining in this chapter the dynamics of labor income inequality. What caused the explosion of wage inequalities and the rise of the supermanager in the United States after 1980? More generally, what accounts for the diverse historical evolutions we see in various countries?

In subsequent chapters I will look into the evolution of the capital ownership distribution: How and why has the concentration of wealth decreased everywhere, but especially in Europe, since the turn of the twentieth century? The emergence of a “patrimonial middle class” is a crucial issue for this study, because it largely explains why income inequality decreased during the first half of the twentieth century and why we in the developed countries have gone from a society of rentiers to a society of managers (or, in the less optimistic version, from a society of superrentiers to a somewhat less extreme form of rentier society).

Wage Inequality: A Race between Education and Technology?

Why is inequality of income from labor, and especially wage inequality, greater in some societies and periods than others? The most widely accepted theory is that of a race between education and technology. To be blunt, this theory does not explain everything. In particular, it does not offer a satisfactory explanation of the rise of the supermanager or of wage inequality in the United States after 1980. The theory does, however, suggest interesting and important clues for explaining certain historical evolutions. I will therefore begin by discussing it.

The theory rests on two hypotheses. First, a worker’s wage is equal to his marginal productivity, that is, his individual contribution to the output of the firm or office for which he works. Second, the worker’s productivity depends above all on his skill and on supply and demand for that skill in a given society. For example, in a society in which very few people are qualified engineers (so that the “supply” of engineers is low) and the prevailing technology requires many engineers (so that “demand” is high), then it is highly likely that this combination of low supply and high demand will result in very high pay for engineers (relative to other workers) and therefore significant wage inequality between highly paid engineers and other workers.

This theory is in some respects limited and naïve. (In practice, a worker’s productivity is not an immutable, objective quantity inscribed on his forehead, and the relative power of different social groups often plays a central role in determining what each worker is paid.) Nevertheless, as simple or even simplistic as the theory may be, it has the virtue of emphasizing two social and economic forces that do indeed play a fundamental role in determining wage inequality, even in more sophisticated theories: the supply and demand of skills. In practice, the supply of skills depends on, among other things, the state of the educational system: how many people have access to this or that track, how good is the training, how much classroom teaching is supplemented by appropriate professional experience, and so on. The demand for skills depends on, among other things, the state of the technologies available to produce the goods and services that society consumes. No matter what other forces may be involved, it seems clear that these two factors—the state of the training system on the one hand, the state of technology on the other—play a crucial role. At a minimum, they influence the relative power of different social groups.

These two factors themselves depend on many other forces. The educational system is shaped by public policy, criteria of selection for different tracks, the way it is financed, the cost of study for students and their families, and the availability of continuing education. Technological progress depends on the pace of innovation and the rapidity of implementation. It generally increases the demand for new skills and creates new occupations. This leads to the idea of a race between education and technology: if the supply of skills does not increase at the same pace as the needs of technology, then groups whose training is not sufficiently advanced will earn less and be relegated to devalued lines of work, and inequality with respect to labor will increase. In order to avoid this, the educational system must increase its supply of new types of training and its output of new skills at a sufficiently rapid pace. If equality is to decrease, moreover, the supply of new skills must increase even more rapidly, especially for the least well educated.

Consider, for example, wage inequalities in France. As I have shown, the wage hierarchy was fairly stable over a long period of time. The average wage increased enormously over the course of the twentieth century, but the gap between the best and worst paid deciles remained the same. Why was this the case, despite the massive democratization of the educational system during the same period? The most natural explanation is that all skill levels progressed at roughly the same pace, so that the inequalities in the wage scale were simply translated upward. The bottom group, which had once only finished grade school, moved up a notch on the educational ladder, first completing junior high school, then going on to a high school diploma. But the group that had previously made do with a high school diploma now went on to college or even graduate school. In other words, the democratization of the educational system did not eliminate educational inequality and therefore did not reduce wage inequality. If educational democratization had not taken place, however, and if the children of those who had only finished grade school a century ago (three-quarters of each generation at that time) had remained at that level, inequalities with respect to labor, and especially wage inequalities, would surely have increased substantially.

Now consider the US case. Two economists, Claudia Goldin and Lawrence Katz, systematically compared the following two evolutions in the period 1890–2005: on the one hand the wage gap between workers who graduated from college and those who had only a high school diploma, and on the other the rate of growth of the number of college degrees. For Goldin and Katz, the conclusion is stark: the two curves move in opposite directions. In particular, the wage gap, which decreased fairly regularly until the 1970s, suddenly begins to widen in the 1980s, at precisely the moment when for the first time the number of college graduates stops growing, or at any rate grows much more slowly than before.1 Goldin and Katz have no doubt that increased wage inequality in the United States is due to a failure to invest sufficiently in higher education. More precisely, too many people failed to receive the necessary training, in part because families could not afford the high cost of tuition. In order to reverse this trend, they conclude, the United States should invest heavily in education so that as many people as possible can attend college.

The lessons of French and US experience thus point in the same direction. In the long run, the best way to reduce inequalities with respect to labor as well as to increase the average productivity of the labor force and the overall growth of the economy is surely to invest in education. If the purchasing power of wages increased fivefold in a century, it was because the improved skills of the workforce, coupled with technological progress, increased output per head fivefold. Over the long run, education and technology are the decisive determinants of wage levels.

By the same token, if the United States (or France) invested more heavily in high-quality professional training and advanced educational opportunities and allowed broader segments of the population to have access to them, this would surely be the most effective way of increasing wages at the low to medium end of the scale and decreasing the upper decile’s share of both wages and total income. All signs are that the Scandinavian countries, where wage inequality is more moderate than elsewhere, owe this result in large part to the fact that their educational system is relatively egalitarian and inclusive.2 The question of how to pay for education, and in particular how to pay for higher education, is everywhere one of the key issues of the twenty-first century. Unfortunately, the data available for addressing issues of educational cost and access in the United States and France are extremely limited. Both countries attach a great deal of importance to the central role of schools and vocational training in fostering social mobility, yet theoretical discussion of educational issues and of meritocracy is often out of touch with reality, and in particular with the fact that the most prestigious schools tend to favor students from privileged social backgrounds. I will come back to this point in Chapter 13.

The Limits of the Theoretical Model: The Role of Institutions

Education and technology definitely play a crucial role in the long run. This theoretical model, based on the idea that a worker’s wage is always perfectly determined by her marginal productivity and thus primarily by skill, is nevertheless limited in a number of ways. Leave aside the fact that it is not always enough to invest in training: existing technology is sometimes unable to make use of the available supply of skills. Leave aside, too, the fact that this theoretical model, at least in its most simplistic form, embodies a far too instrumental and utilitarian view of training. The main purpose of the health sector is not to provide other sectors with workers in good health. By the same token, the main purpose of the educational sector is not to prepare students to take up an occupation in some other sector of the economy. In all human societies, health and education have an intrinsic value: the ability to enjoy years of good health, like the ability to acquire knowledge and culture, is one of the fundamental purposes of civilization.3 We are free to imagine an ideal society in which all other tasks are almost totally automated and each individual has as much freedom as possible to pursue the goods of education, culture, and health for the benefit of herself and others. Everyone would be by turns teacher or student, writer or reader, actor or spectator, doctor or patient. As noted in Chapter 2, we are to some extent already on this path: a characteristic feature of modern growth is the considerable share of both output and employment devoted to education, culture, and medicine.

While awaiting the ideal society of the future, let us try to gain a better understanding of wage inequality today. In this narrower context, the main problem with the theory of marginal productivity is quite simply that it fails to explain the diversity of the wage distributions we observe in different countries at different times. In order to understand the dynamics of wage inequality, we must introduce other factors, such as the institutions and rules that govern the operation of the labor market in each society. To an even greater extent than other markets, the labor market is not a mathematical abstraction whose workings are entirely determined by natural and immutable mechanisms and implacable technological forces: it is a social construct based on specific rules and compromises.

In the previous chapter I noted several important episodes of compression and expansion of wage hierarchies that are very difficult to explain solely in terms of the supply of and demand for various skills. For example, the compression of wage inequalities that occurred in both France and the United States during World Wars I and II was the result of negotiations over wage scales in both the public and private sectors, in which specific institutions such as the National War Labor Board (created expressly for the purpose) played a central role. I also called attention to the importance of changes in the minimum wage for explaining the evolution of wage inequalities in France since 1950, with three clearly identified subperiods: 1950–1968, during which the minimum wage was rarely adjusted and the wage hierarchy expanded; 1968–1983, during which the minimum wage rose very rapidly and wage inequalities decreased sharply; and finally 1983–2012, during which the minimum wage increased relatively slowly and the wage hierarchy tended to expand.4 At the beginning of 2013, the minimum wage in France stood at 9.43 euros per hour.

FIGURE 9.1. Minimum wage in France and the United States, 1950–2013

Expressed in 2013 purchasing power, the hourly minimum wage rose from $3.80 to $7.30 between 1950 and 2013 in the United States, and from €2.10 to €9.40 in France.

Sources and series: see piketty.pse.ens.fr/capital21c.

In the United States, a federal minimum wage was introduced in 1933, nearly twenty years earlier than in France.5 As in France, changes in the minimum wage played an important role in the evolution of wage inequalities in the United States. It is striking to learn that in terms of purchasing power, the minimum wage reached its maximum level nearly half a century ago, in 1969, at $1.60 an hour (or $10.10 in 2013 dollars, taking account of inflation between 1968 and 2013), at a time when the unemployment rate was below 4 percent. From 1980 to 1990, under the presidents Ronald Reagan and George H. W. Bush, the federal minimum wage remained stuck at $3.35, which led to a significant decrease in purchasing power when inflation is factored in. It then rose to $5.25 under Bill Clinton in the 1990s and was frozen at that level under George W. Bush before being increased several times by Barack Obama after 2008. At the beginning of 2013 it stood at $7.25 an hour, or barely 6 euros, which is a third below the French minimum wage, the opposite of the situation that obtained in the early 1980s (see Figure 9.1).6 President Obama, in his State of the Union address in February 2013, announced his intention to raise the minimum wage to about $9 an hour for the period 2013–2016.7

Inequalities at the bottom of the US wage distribution have closely followed the evolution of the minimum wage: the gap between the bottom 10 percent of the wage distribution and the overall average wage widened significantly in the 1980s, then narrowed in the 1990s, and finally increased again in the 2000s. Nevertheless, inequalities at the top of the distribution—for example, the share of total wages going to the top 10 percent—increased steadily throughout this period. Clearly, the minimum wage has an impact at the bottom of the distribution but much less influence at the top, where other forces are at work.

Wage Scales and the Minimum Wage

There is no doubt that the minimum wage plays an essential role in the formation and evolution of wage inequalities, as the French and US experiences show. Each country has its own history in this regard and its own peculiar chronology. There is nothing surprising about that: labor market regulations depend on each society’s perceptions and norms of social justice and are intimately related to each country’s social, political, and cultural history. The United States used the minimum wage to increase lower-end wages in the 1950s and 1960s but abandoned this tool in the 1970s. In France, it was exactly the opposite: the minimum wage was frozen in the 1950s and 1960s but was used much more often in the 1970s. Figure 9.1 illustrates this striking contrast.

It would be easy to multiply examples from other countries. Britain introduced a minimum wage in 1999, at a level between the United States and France: in 2013 it was £6.19 (or about 8.05 euros).8 Germany and Sweden have chosen to do without minimum wages at the national level, leaving it to trade unions to negotiate not only minimums but also complete wage schedules with employers in each branch of industry. In practice, the minimum wage in both countries was about 10 euros an hour in 2013 in many branches (and therefore higher than in countries with a national minimum wage). But minimum pay can be markedly lower in sectors that are relatively unregulated or underunionized. In order to set a common floor, Germany is contemplating the introduction of a minimum wage in 2013–2014. This is not the place to write a detailed history of minimum wages and wage schedules around the world or to discuss their impact on wage inequality. My goal here is more modest: simply to indicate briefly what general principles can be used to analyze the institutions that regulate wage setting everywhere.

What is in fact the justification for minimum wages and rigid wage schedules? First, it is not always easy to measure the marginal productivity of a particular worker. In the public sector, this is obvious, but it is also clear in the private sector: in an organization employing dozens or even thousands of workers, it is no simple task to judge each individual worker’s contribution to overall output. To be sure, one can estimate marginal productivity, at least for jobs that can be replicated, that is, performed in the same way by any number of employees. For an assembly-line worker or McDonald’s server, management can calculate how much additional revenue an additional worker or server would generate. Such an estimate would be approximate, however, yielding a range of productivities rather than an absolute number. In view of this uncertainty, how should the wage be set? There are many reasons to think that granting management absolute power to set the wage of each employee on a monthly or (why not?) daily basis would not only introduce an element of arbitrariness and injustice but would also be inefficient for the firm.

In particular, it may be efficient for the firm to ensure that wages remain relatively stable and do not vary constantly with fluctuations in sales. The owners and managers of the firm usually earn much more and are significantly wealthier than their workers and can therefore more easily absorb short-term shocks to their income. Under such circumstances, it can be in everyone’s interest to provide a kind of “wage insurance” as part of the employment contract, in the sense that the worker’s monthly wage is guaranteed (which does not preclude the use of bonuses and other incentives). The payment of a monthly rather than a daily wage was a revolutionary innovation that gradually took hold in all the developed countries during the twentieth century. This innovation was inscribed in law and became a feature of wage negotiations between workers and employers. The daily wage, which had been the norm in the nineteenth century, gradually disappeared. This was a crucial step in the constitution of the working class: workers now enjoyed a legal status and received a stable, predictable remuneration for their work. This clearly distinguished them from day laborers and piece workers—the typical employees of the eighteenth and nineteenth centuries.9

This justification of setting wages in advance obviously has its limits. The other classic argument in favor of minimum wages and fixed wage schedules is the problem of “specific investments.” Concretely, the particular functions and tasks that a firm needs to be performed often require workers to make specific investments in the firm, in the sense that these investments are of no (or limited) value to other firms: for instance, workers might need to learn specific work methods, organizational methods, or skills linked to the firm’s production process. If wages can be set unilaterally and changed at any moment by the firm, so that workers do not know in advance how much they will be paid, then it is highly likely that they will not invest as much in the firm as they should. It may therefore be in everyone’s interest to set pay scales in advance. The same “specific investments” argument can also apply to other decisions by the firm, and it is the main reason for limiting the power of stockholders (who are seen as having too short-term an outlook in some cases) in favor of a power-sharing arrangement with a broader group of “stakeholders” (including the firm’s workers), as in the “Rhenish model” of capitalism discussed earlier, in Part Two. This is probably the most important argument in favor of fixed wage scales.

More generally, insofar as employers have more bargaining power than workers and the conditions of “pure and perfect” competition that one finds in the simplest economic models fail to be satisfied, it may be reasonable to limit the power of employers by imposing strict rules on wages. For example, if a small group of employers occupies a monopsony position in a local labor market (meaning that they are virtually the only source of employment, perhaps because of the limited mobility of the local labor force), they will probably try to exploit their advantage by lowering wages as much as possible, possibly even below the marginal productivity of the workers. Under such conditions, imposing a minimum wage may be not only just but also efficient, in the sense that the increase in wages may move the economy closer to the competitive equilibrium and increase the level of employment. This theoretical model, based on imperfect competition, is the clearest justification for the existence of a minimum wage: the goal is to make sure that no employer can exploit his competitive advantage beyond a certain limit.

Again, everything obviously depends on the level of the minimum wage. The limit cannot be set in the abstract, independent of the country’s general skill level and average productivity. Various studies carried out in the United States between 1980 and 2000, most notably by the economists David Card and Alan Krueger, showed that the US minimum wage had fallen to a level so low in that period that it could be raised without loss of employment, indeed at times with an increase in employment, as in the monopsony model.10 On the basis of these studies, it seems likely that the increase in the minimum wage of nearly 25 percent (from $7.25 to $9 an hour) currently envisaged by the Obama administration will have little or no effect on the number of jobs. Obviously, raising the minimum wage cannot continue indefinitely: as the minimum wage increases, the negative effects on the level of employment eventually win out. If the minimum wage were doubled or tripled, it would be surprising if the negative impact were not dominant. It is more difficult to justify a significant increase in the minimum wage in a country like France, where it is relatively high (compared with the average wage and marginal productivity), than in the United States. To increase the purchasing power of low-paid workers in France, it is better to use other tools, such as training to improve skills or tax reform (these two remedies are complementary, moreover). Nevertheless, the minimum wage should not be frozen. Wage increases cannot exceed productivity increases indefinitely, but it is just as unhealthy to restrain (most) wage increases to below the rate of productivity increase. Different labor market institutions and policies play different roles, and each must be used in an appropriate manner.

To sum up: the best way to increase wages and reduce wage inequalities in the long run is to invest in education and skills. Over the long run, minimum wages and wage schedules cannot multiply wages by factors of five or ten: to achieve that level of progress, education and technology are the decisive forces. Nevertheless, the rules of the labor market play a crucial role in wage setting during periods of time determined by the relative progress of education and technology. In practice, those periods can be fairly long, in part because it is hard to gauge individual marginal productivities with any certainty, and in part because of the problem of specific investments and imperfect competition.

How to Explain the Explosion of Inequality in the United States?

The most striking failure of the theory of marginal productivity and the race between education and technology is no doubt its inability to adequately explain the explosion of very high incomes from labor observed in the United States since 1980. According to this theory, one should be able to explain this change as the result of skill-biased technological change. Some US economists buy this argument, which holds that top labor incomes have risen much more rapidly than average wages simply because unique skills and new technology have made these workers much more productive than the average. There is a certain tautological quality to this explanation (after all, one can “explain” any distortion of the wage hierarchy as the result of some supposed technological change). It also has other major weaknesses, which to my mind make it a rather unconvincing argument.

First, as shown in the previous chapter, the increase in wage inequality in the United States is due mainly to increased pay at the very top end of the distribution: the top 1 percent and even more the top 0.1 percent. If we look at the entire top decile, we find that “the 9 percent” have progressed more rapidly than the average worker but not nearly at the same rate as “the 1 percent.” Concretely, those making between $100,000 and $200,000 a year have seen their pay increase only slightly more rapidly than the average, whereas those making more than $500,000 a year have seen their remuneration literally explode (and those above $1 million a year have risen even more rapidly).11 This very sharp discontinuity at the top income levels is a problem for the theory of marginal productivity: when we look at the changes in the skill levels of different groups in the income distribution, it is hard to see any discontinuity between “the 9 percent” and “the 1 percent,” regardless of what criteria we use: years of education, selectivity of educational institution, or professional experience. One would expect a theory based on “objective” measures of skill and productivity to show relatively uniform pay increases within the top decile, or at any rate increases within different subgroups much closer to one another than the widely divergent increases we observe in practice.

Make no mistake: I am not denying the decisive importance of the investments in higher education and training that Katz and Goldin have identified. Policies to encourage broader access to universities are indispensable and crucial in the long run, in the United States and elsewhere. As desirable as such policies are, however, they seem to have had limited impact on the explosion of the topmost incomes observed in the United States since 1980.

In short, two distinct phenomena have been at work in recent decades. First, the wage gap between college graduates and those who go no further than high school has increased, as Goldin and Katz showed. In addition, the top 1 percent (and even more the top 0.1 percent) have seen their remuneration take off. This is a very specific phenomenon, which occurs within the group of college graduates and in many cases separates individuals who have pursued their studies at elite universities for many years. Quantitatively, the second phenomenon is more important than the first. In particular, as shown in the previous chapter, the overperformance of the top centile explains most (nearly three-quarters) of the increase in the top decile’s share of US national income since 1970.12 It is therefore important to find an adequate explanation of this phenomenon, and at first sight the educational factor does not seem to be the right one to focus on.

The Rise of the Supermanager: An Anglo-Saxon Phenomenon

The second difficulty—and no doubt the major problem confronting the marginal productivity theory—is that the explosion of very high salaries occurred in some developed countries but not others. This suggests that institutional differences between countries rather than general and a priori universal causes such as technological change played a central role.

I begin with the English-speaking countries. Broadly speaking, the rise of the supermanager is largely an Anglo-Saxon phenomenon. Since 1980 the share of the upper centile in national income has risen significantly in the United States, Great Britain, Canada, and Australia (see Figure 9.2). Unfortunately, we do not have separate series for wage inequality and total income inequality for all countries as we do for France and the United States. But in most cases we do have data concerning the composition of income in relation to total income, from which we can infer that in all of these countries the explosion of top incomes explains most (generally at least two-thirds) of the increase in the top centile’s share of national income; the rest is explained by robust income from capital. In all the English-speaking countries, the primary reason for increased income inequality in recent decades is the rise of the supermanager in both the financial and nonfinancial sectors.

FIGURE 9.2. Income inequality in Anglo-Saxon countries, 1910–2010

The share of top percentile in total income rose since the 1970s in all Anglo-Saxon countries, but with different magnitudes.

Sources and series: see piketty.pse.ens.fr/capital21c.

This family resemblance should not be allowed to obscure the fact that the magnitude of the phenomenon varies widely from country to country, however. Figure 9.2 is quite clear on this point. In the 1970s, the upper centile’s share of national income was quite similar across countries. It ranged from 6 to 8 percent in the four English-speaking countries considered, and the United States did not stand out as exceptional: indeed, Canada was slightly higher, at 9 percent, whereas Australia came in last, with just 5 percent of national income going to the top centile in the late 1970s and early 1980s. Thirty years later, in the early 2010s, the situation is totally different. The upper centile’s share is nearly 20 percent in the United States, compared with 14–15 percent in Britain and Canada and barely 9–10 percent in Australia (see Figure 9.2).13 To a first approximation, we can say that the upper centile’s share in the United States increased roughly twice as much as in Britain and Canada and about three times as much as in Australia and New Zealand.14 If the rise of the supermanager were a purely technological phenomenon, it would be difficult to understand why such large differences exist between otherwise quite similar countries.

FIGURE 9.3. Income inequality in Continental Europe and Japan, 1910–2010

As compared to Anglo-Saxon countries, the share of top percentile barely increased since the 1970s in Continental Europe and Japan.

Sources and series: see piketty.pse.ens.fr/capital21c.

Let me turn now to the rest of the wealthy world, namely, continental Europe and Japan. The key fact is that the upper centile’s share of national income in these countries has increased much less than in the English-speaking countries since 1980. The comparison between Figures 9.2 and 9.3 is particularly striking. To be sure, the upper centile’s share increased significantly everywhere. In Japan the evolution was virtually the same as in France: the top centile’s share of national income was barely 7 percent in the 1980s but is 9 percent or perhaps even slightly higher today. In Sweden, the top centile’s share was a little more than 4 percent in the early 1980s (the lowest level recorded in the World Top Incomes Database for any country in any period) but reached 7 percent in the early 2010s.15 In Germany, the top centile’s share rose from about 9 percent to nearly 11 percent of national income between the early 1980s and the early 2010s (see Figure 9.3).

If we look at other European countries, we observe similar evolutions, with the top centile’s share increasing by two or three points of national income over the past thirty years in both northern and southern Europe. In Denmark and other Nordic countries, top incomes claim a smaller share of the total, but the increase is similar: the top centile received a little more than 5 percent of Danish national income in the 1980s but got close to 7 percent in 2000–2010. In Italy and Spain, the orders of magnitude are very close to those observed in France, with the top centile’s share rising from 7 to 9 percent of national income in the same period, again an increase of two points of national income (see Figure 9.4). In this respect, continental Europe is indeed an almost perfect “union.” Britain, of course, stands apart, being much closer to the pattern of the United States than that of Europe.16

FIGURE 9.4. Income inequality in Northern and Southern Europe, 1910–2010

As compared to Anglo-Saxon countries, the top percentile income share barely increased in Northern and Southern Europe since the 1970s.

Sources and series: see piketty.pse.ens.fr/capital21c.

Make no mistake: these increases on the order of two to three points of national income in Japan and the countries of continental Europe mean that income inequality rose quite significantly. The top 1 percent of earners saw pay increases noticeably more rapid than the average: the upper centile’s share increased by about 30 percent, and even more in countries where it started out lower. This was quite striking to contemporary observers, who read in the daily paper or heard on the radio about stupendous raises for “supermanagers.” It was particularly striking in the period 1990–2010, when average income stagnated, or at least rose much more slowly than in the past.

FIGURE 9.5. The top decile income share in Anglo-Saxon countries, 1910–2010

The share of the top 0.1 percent highest incomes in total income rose sharply since the 1970s in all Anglo-Saxon countries, but with varying magnitudes.

Sources and series: see piketty.pse.ens.fr/capital21c.

Furthermore, the higher one climbs in the income hierarchy, the more spectacular the raises. Even if the number of individuals benefiting from such salary increases is fairly limited, they are nevertheless quite visible, and this visibility naturally raises the question of what justifies such high levels of compensation. Consider the share of the top thousandth—the best remunerated 0.1 percent—in the national income of the English-speaking countries on the one hand (Figure 9.5) and continental Europe and Japan on the other (Figure 9.6). The differences are obvious: the top thousandth in the United States increased their share from 2 to nearly 10 percent over the past several decades—an unprecedented rise.17 But there has been a remarkable increase of top incomes everywhere. In France and Japan, the top thousandth’s share rose from barely 1.5 percent of national income in the early 1980s to nearly 2.5 percent in the early 2010s—close to double. In Sweden, the same share rose from less than 1 percent to more than 2 percent in the same period.

To make clear what this represents in concrete terms, remember that a 2 percent share of national income for 0.1 percent of the population means that the average individual in this group enjoys an income 20 times higher than the national average (or 600,000 euros a year if the average income is 30,000 per adult). A share of 10 percent means that each individual enjoys an income 100 times the national average (or 3 million euros a year if the average is 30,000).18 Recall, too, that the top 0.1 percent is by definition a group of 50,000 people in a country with a population of 50 million adults (like France in the early 2010s). This is a very small minority (“the 1 percent” is of course 10 times larger), yet it occupies a significant place in the social and political landscape.19 The central fact is that in all the wealthy countries, including continental Europe and Japan, the top thousandth enjoyed spectacular increases in purchasing power in 1990–2010, while the average person’s purchasing power stagnated.

FIGURE 9.6. The top decile income share in Continental Europe and Japan, 1910–2010

As compared to Anglo-Saxon countries, the top 0.1 percent income share barely increased in Continental Europe and Japan.

Sources and series: see piketty.pse.ens.fr/capital21c.

From a macroeconomic point of view, however, the explosion of very high incomes has thus far been of limited importance in continental Europe and Japan: the rise has been impressive, to be sure, but too few people have been affected to have had an impact as powerful as in the United States. The transfer of income to “the 1 percent” involves only two to three points of national income in continental Europe and Japan compared with 10 to 15 points in the United States—5 to 7 times greater.20

The simplest way to express these regional differences is no doubt the following: in the United States, income inequality in 2000–2010 regained the record levels observed in 1910–1920 (although the composition of income was now different, with a larger role played by high incomes from labor and a smaller role by high incomes from capital). In Britain and Canada, things moved in the same direction. In continental Europe and Japan, income inequality today remains far lower than it was at the beginning of the twentieth century and in fact has not changed much since 1945, if we take a long-run view. The comparison of Figures 9.2 and 9.3 is particularly clear on this point.

Obviously, this does not mean that the European and Japanese evolutions of the past few decades should be neglected. On the contrary: their trajectory resembles that of the United States in some respects, with a delay of one or two decades, and one need not wait until the phenomenon assumes the macroeconomic significance observed in the United States to worry about it.

Nevertheless, the fact remains that the evolution in continental Europe and Japan is thus far much less serious than in the United States (and, to a lesser extent, in the other Anglo-Saxon countries). This may tell us something about the forces at work. The divergence between the various regions of the wealthy world is all the more striking because technological change has been the same more or less everywhere: in particular, the revolution in information technology has affected Japan, Germany, France, Sweden, and Denmark as much as the United States, Britain, and Canada. Similarly, economic growth—or, more precisely, growth in output per capita, which is to say, productivity growth—has been quite similar throughout the wealthy countries, with differences of a few tenths of a percentage point.21 In view of these facts, this quite large divergence in the way the income distribution has evolved in the various wealthy countries demands an explanation, which the theory of marginal productivity and of the race between technology and education does not seem capable of providing.

Europe: More Inegalitarian Than the New World in 1900–1910

Note, moreover, that the United States, contrary to what many people think today, was not always more inegalitarian than Europe—far from it. Income inequality was actually quite high in Europe at the beginning of the twentieth century. This is confirmed by all the indices and historical sources. In particular, the top centile’s share of national income exceeded 20 percent in all the countries of Europe in 1900–1910 (see Figures 9.2–4). This was true not only of Britain, France, and Germany but also of Sweden and Denmark (proof that the Nordic countries have not always been models of equality—far from it), and more generally of all European countries for which we have estimates from this period.22

The similar levels of income concentration in all European countries during the Belle Époque obviously demand an explanation. Since top incomes in this period consisted almost entirely of income from capital,23 the explanation must be sought primarily in the realm of concentration of capital. Why was capital so concentrated in Europe in the period 1900–1910?

It is interesting to note that, compared with Europe, inequality was lower not only in the United States and Canada (where the top centile’s share of national income was roughly 16–18 percent at the beginning of the twentieth century) but especially in Australia and New Zealand (11–12 percent). Thus it was the New World, and especially the newest and most recently settled parts of the New World, that appear to have been less inegalitarian than Old Europe in the Belle Époque.

It is also interesting to note that Japan, despite its social and cultural differences from Europe, seems to have had the same high level of inequality at the beginning of the twentieth century, without about 20 percent of national income going to the top centile. The available data do not allow me to make all the comparisons I would like to make, but all signs are that in terms of both income structure and income inequality, Japan was indeed part of the same “old world” as Europe. It is also striking to note the similar evolution of Japan and Europe over the course of the twentieth century (Figure 9.3).

I will return later to the reasons for the very high concentration of capital in the Belle Époque and to the transformations that took place in various countries over the course of the twentieth century (namely, a reduction of concentration). I will show in particular that the greater inequality of wealth that we see in Europe and Japan is fairly naturally explained by the low demographic growth rate we find in the Old World, which resulted almost automatically in a greater accumulation and concentration of capital.

At this stage, I want simply to stress the magnitude of the changes that have altered the relative standing of countries and continents. The clearest way to make this point is probably to look at the evolution of the top decile’s share of national income. Figure 9.7 shows this for the United States and four European countries (Britain, France, Germany, and Sweden) since the turn of the twentieth century. I have indicated decennial averages in order to focus attention on long-term trends.24

FIGURE 9.7. The top decile income share in Europe and the United States, 1900–2010

In the 1950s–1970s, the top decile income share was about 30–35 percent of total income in Europe as in the United States.

Sources and series: see piketty.pse.ens.fr/capital21c.

What we find is that on the eve of World War I, the top decile’s share was 45–50 percent of national income in all the European countries, compared with a little more than 40 percent in the United States. By the end of World War II, the United States had become slightly more inegalitarian than Europe: the top decile’s share decreased on both continents owing to the shocks of 1914–1945, but the fall was more precipitous in Europe (and Japan). The explanation for this is that the shocks to capital were much larger. Between 1950 and 1970, the upper decile’s share was fairly stable and fairly similar in the United States and Europe, around 30–35 percent of national income. The strong divergence that began in 1970–1980 led to the following situation in 2000–2010: the top decile’s share of US national income reached 45–50 percent, or roughly the same level as Europe in 1900–1910. In Europe, we see wide variation, from the most inegalitarian case (Britain, with a top decile share of 40 percent) to the most egalitarian (Sweden, less than 30 percent), with France and Germany in between (around 35 percent).

FIGURE 9.8. Income inequality in Europe versus the United States, 1900–2010

The top decile income share was higher in Europe than in the United States in 1900–1910; it is a lot higher in the United States in 2000–2010.

Sources and series: see piketty.pse.ens.fr/capital21c.

If we calculate (somewhat abusively) an average for Europe based on these four countries, we can make a very clear international comparison: the United States was less inegalitarian than Europe in 1900–1910, slightly more inegalitarian in 1950–1960, and much more inegalitarian in 2000–2010 (see Figure 9.8).25

Apart from this long-term picture, there are of course multiple national histories as well as constant short- and medium-term fluctuations linked to social and political developments in each country, as I showed in Chapter 8 and analyzed in some detail in the French and US cases. Space will not permit me to do the same for every country here.26

In passing, however, it is worth mentioning that the period between the two world wars seems to have been particularly tumultuous and chaotic almost everywhere, though the chronology of events varied from country to country. In Germany, the hyperinflation of the 1920s followed hard on the heels of military defeat. The Nazis came to power a short while later, after the worldwide depression had plunged the country back into crisis. Interestingly, the top centile’s share of German national income increased rapidly between 1933 and 1938, totally out of phase with other countries: this reflects the revival of industrial profits (boosted by demand for armaments), as well as a general reestablishment of income hierarchies in the Nazi era. Note, too, that the share of the top centile—and, even more, the top thousandth—in Germany has been noticeably higher since 1950 than in most other continental European countries (including, in particular, France) as well as Japan, even though the overall level of inequality in Germany is not very different. This can be explained in various ways, among which it is difficult to say that one is better than another. (I will come back to this point.)

In addition, there are serious lacunae in German tax records, owing in large part to the country’s turbulent history in the twentieth century, so that it is difficult to be sure about certain developments or to make sharp comparisons with other countries. Prussia, Saxony, and most other German states imposed an income tax relatively early, between 1880 and 1890, but there were no national laws or tax records until after World War I. There were frequent breaks in the statistical record during the 1920s, and then the records for 1938 to 1950 are missing altogether, so it is impossible to study how the income distribution evolved during World War II and its immediate aftermath.

This distinguishes Germany from other countries deeply involved in the conflict, especially Japan and France, whose tax administrations continued to compile statistics during the war years without interruption, as if nothing were amiss. If Germany was anything like these two countries, it is likely that the top centile’s share of national income reached a nadir in 1945 (the year in which German capital and income from capital were reduced to virtually nothing) before beginning to rise sharply again in 1946–1947. Yet when German tax records return in 1950, they show the income hierarchy already beginning to resemble its appearance in 1938. In the absence of complete sources, it is difficult to say more. The German case is further complicated by the fact that the country’s boundaries changed several times during the twentieth century, most recently with the reunification of 1990–1991, in addition to which full tax data are published only every three years (rather than annually as in most other countries).

Inequalities in Emerging Economies: Lower Than in the United States?

Let me turn now to the poor and emerging economies. The historical sources we need in order to study the long-run dynamics of the wealth distribution there are unfortunately harder to come by than in the rich countries. There are, however, a number of poor and emerging economies for which it is possible to find long series of tax data useful for making (rough) comparisons with our results for the more developed economies. Shortly after Britain introduced a progressive income tax at home, it decided to do the same in a number of its colonies. Thus an income tax fairly similar to that introduced in Britain in 1909 was adopted in South Africa in 1913 and in India (including present-day Pakistan) in 1922. Similarly, the Netherlands imposed an income tax on its Indonesian colony in 1920. Several South American countries introduced an income tax between the two world wars: Argentina, for example, did so in 1932. For these four countries—South Africa, India, Indonesia, and Argentina—we have tax data going back, respectively, to 1913, 1922, 1920, and 1932 and continuing (with gaps) to the present. The data are similar to what we have for the rich countries and can be employed using similar methods, in particular national income estimates for each country going back to the turn of the twentieth century.

My estimates are indicated in Figure 9.9. Several points deserve to be emphasized. First, the most striking result is probably that the upper centile’s share of national income in poor and emerging economies is roughly the same as in the rich economies. During the most inegalitarian phases, especially 1910–1950, the top centile took around 20 percent of national income in all four countries: 15–18 percent in India and 22–25 percent in South Africa, Indonesia, and Argentina. During more egalitarian phases (essentially 1950–1980), the top centile’s share fell to between 6 and 12 percent (barely 5–6 percent in India, 8–9 percent in Indonesia and Argentina, and 11–12 percent in South Africa). Thereafter, in the 1980s, the top centile’s share rebounded, and today it stands at about 15 percent of national income (12–13 percent in India and Indonesia and 16–18 percent in South Africa and Argentina).

Figure 9.9 also shows two countries for which the available tax records allow us only to study how things have changed since the mid-1980s: China and Colombia.27 In China, the top centile’s share of national income rose rapidly over the past several decades but starting from a fairly low (almost Scandinavian) level in the mid-1980s: less than 5 percent of national income went to the top centile at that time, according to the available sources. This is not very surprising for a Communist country with a very compressed wage schedule and virtual absence of private capital. Chinese inequality increased very rapidly following the liberalization of the economy in the 1980s and accelerated growth in the period 1990–2000, but according to my estimates, the upper centile’s share in 2000–2010 was 10–11 percent, less than in India or Indonesia (12–14 percent, roughly the same as Britain and Canada) and much lower than in South Africa or Argentina (16–18 percent, approximately the same as the United States).

FIGURE 9.9. Income inequality in emerging countries, 1910–2010

Measured by the top percentile income share, income inequality rose in emerging countries since the 1980s, but ranks below the US level in 2000–2010.

Sources and series: see piketty.pse.ens.fr/capital21c.

Colombia on the other hand is one of the most inegalitarian societies in the WTID: the top centile’s share stood at about 20 percent of national income throughout the period 1990–2010, with no clear trend (see Figure 9.9). This level of inequality is even higher than that attained by the United States in 2000–2010, at least if capital gains are excluded; if they are included, the United States was slightly ahead of Colombia over the past decade.

It is important, however, to be aware of the significant limitations of the data available for measuring the dynamics of the income distribution in poor and emerging countries and for comparing them with the rich countries. The orders of magnitude indicated here are the best I was able to come up with given the available sources, but the truth is that our knowledge remains meager. We have tax data for the entire twentieth century for only a few emerging economies, and there are many gaps and breaks in the data, often in the period 1950–1970, the era of independence (in Indonesia, for example). Work is going forward to update the WTID with historical data from many other countries, especially from among the former British and French colonies, in Indochina and Africa, but data from the colonial era are often difficult to relate to contemporary tax records.28

Where tax records do exist, their interest is often reduced by the fact that the income tax in less developed countries generally applies to only a small minority of the population, so that one can estimate the upper centile’s share of total income but not the upper decile’s. Where the data allow, as in South Africa for certain subperiods, one finds that the highest observed levels for the top decile are on the order of 50–55 percent of national income—a level comparable to or slightly higher than the highest levels of inequality observed in the wealthy countries, in Europe in 1900–1910 and in the United States in 2000–2010.

I have also noticed a certain deterioration of the tax data after 1990. This is due in part to the arrival of computerized records, which in many cases led the tax authorities to interrupt the publication of detailed statistics, which in earlier periods they needed for their own purposes. This sometimes means, paradoxically, that sources have deteriorated since the advent of the information age (we find the same thing happening in the rich countries).29 Above all, the deterioration of the sources seems to be related to a certain disaffection with the progressive income tax in general on the part of certain governments and international organizations.30 A case in point is India, which ceased publishing detailed income tax data in the early 2000s, even though such data had been published without interruption since 1922. As a result, it is harder to study the evolution of top incomes in India since 2000 than over the course of the twentieth century.31

This lack of information and democratic transparency is all the more regrettable in that the question of the distribution of wealth and of the fruits of growth is at least as urgent in the poor and emerging economies as in the rich ones. Note, too, that the very high official growth figures for developing countries (especially India and China) over the past few decades are based almost exclusively on production statistics. If one tries to measure income growth by using household survey data, it is often quite difficult to identify the reported rates of macroeconomic growth: Indian and Chinese incomes are certainly increasing rapidly, but not as rapidly as one would infer from official growth statistics. This paradox—sometimes referred to as the “black hole” of growth—is obviously problematic. It may be due to the overestimation of growth of output (there are many bureaucratic incentives for doing so), or perhaps the underestimation of income growth (household surveys have their own flaws), or most likely both. In particular, the missing income may be explained by the possibility that a disproportionate share of the growth in output has gone to the most highly remunerated individuals, whose incomes are not always captured in the tax data.

In the case of India, it is possible to estimate (using tax return data) that the increase in the upper centile’s share of national income explains between one-quarter and one-third of the “black hole” of growth between 1990 and 2000.32 Given the deterioration of the tax data since 2000, it is impossible to do a proper social decomposition of recent growth. In the case of China, official tax records are even more rudimentary than in India. In the current state of research, the estimates in Figure 9.9 are the most reliable we have.33 It is nevertheless urgent that both countries publish more complete data—and other countries should do so as well. If and when better data become available, we may discover that inequality in India and China has increased more rapidly than we imagined.

In any case, the important point is that whatever flaws the tax authorities in poor and emerging countries may exhibit, the tax data reveal much higher—and more realistic—top income levels than do household surveys. For example, tax returns show that the top centile’s share of national income in Colombia in 2000–2010 was more than 20 percent (and almost 20 percent in Argentina). Actual inequality may be even greater. But the fact that the highest incomes declared in household surveys in these same countries are generally only 4 to 5 times as high as the average income (suggesting that no one is really rich)—so that, if we were to trust the household survey, the top centile’s share would be less than 5 percent—suggests that the survey data are not very credible. Clearly, household surveys, which are often the only source used by international organizations (in particular the World Bank) and governments for gauging inequality, give a biased and misleadingly complacent view of the distribution of wealth. As long as these official estimates of inequality fail to combine survey data with other data systematically gleaned from tax records and other government sources, it will be impossible to apportion macroeconomic growth properly among various social groups or among the centiles and deciles of the income hierarchy. This is true, moreover, of wealthy countries as well as poor and emerging ones.

The Illusion of Marginal Productivity

Let me now return to the explosion of wage inequality in the United States (and to a lesser extent Britain and Canada) after 1970. As noted, the theory of marginal productivity and of the race between technology and education is not very convincing: the explosion of compensation has been highly concentrated in the top centile (or even the top thousandth) of the wage distribution and has affected some countries while sparing others (Japan and continental Europe are thus far much less affected than the United States), even though one would expect technological change to have altered the whole top end of the skill distribution in a more continuous way and to have worked its effects in all countries at a similar level of development. The fact that income inequality in the United States in 2000–2010 attained a level higher than that observed in the poor and emerging countries at various times in the past—for example, higher than in India or South Africa in 1920–1930, 1960–1970, and 2000–2010—also casts doubt on any explanation based solely on objective inequalities of productivity. Is it really the case that inequality of individual skills and productivities is greater in the United States today than in the half-illiterate India of the recent past (or even today) or in apartheid (or postapartheid) South Africa? If that were the case, it would be bad news for US educational institutions, which surely need to be improved and made more accessible but probably do not deserve such extravagant blame.

To my mind, the most convincing explanation for the explosion of the very top US incomes is the following. As noted, the vast majority of top earners are senior managers of large firms. It is rather naïve to seek an objective basis for their high salaries in individual “productivity.” When a job is replicable, as in the case of an assembly-line worker or fast-food server, we can give an approximate estimate of the “marginal product” that would be realized by adding one additional worker or waiter (albeit with a considerable margin of error in our estimate). But when an individual’s job functions are unique, or nearly so, then the margin of error is much greater. Indeed, once we introduce the hypothesis of imperfect information into standard economic models (eminently justifiable in this context), the very notion of “individual marginal productivity” becomes hard to define. In fact, it becomes something close to a pure ideological construct on the basis of which a justification for higher status can be elaborated.

To put this discussion in more concrete terms, imagine a large multinational corporation employing 100,000 people and with gross annual revenue of 10 billion euros, or 100,000 euros per worker. Suppose that half of this revenue figure represents purchases of goods and services by the firm (this is a typical figure for the economy as a whole), so that the value added by the firm—the value available to pay the labor and capital that it directly employs—is 5 billion euros, or 50,000 euros per worker. To set the pay of the firm’s CFO (or his deputies, or of the director of marketing and her staff, etc.), one would in principle want to estimate his marginal productivity, that is, his contribution to the firm’s value-added of 5 billion euros: is it 100,000, 500,000, or 5 million euros per year? A precise, objective answer to this question is clearly impossible. To be sure, one could in theory experiment by trying out several CFOs, each for several years, in order to determine what impact the choice has on the firm’s total revenue of 10 billion euros. Obviously, such an estimate would be highly approximate, with a margin of error much greater than the maximum salary one would think of paying, even in a totally stable economic environment.34 And the whole idea of experimentation looks even more hopeless when one remembers that the environment is in fact changing constantly, as is the nature of the firm and the exact definition of each job.

In view of these informational and cognitive difficulties, how are such remunerations determined in practice? They are generally set by hierarchical superiors, and at the very highest levels salaries are set by the executives themselves or by corporate compensation committees whose members usually earn comparable salaries (such as senior executives of other large corporations). In some companies, stockholders are asked to vote on compensation for senior executives at annual meetings, but the number of posts subject to such approval is small, and not all senior managers are covered. Since it is impossible to give a precise estimate of each manager’s contribution to the firm’s output, it is inevitable that this process yields decisions that are largely arbitrary and dependent on hierarchical relationships and on the relative bargaining power of the individuals involved. It is only reasonable to assume that people in a position to set their own salaries have a natural incentive to treat themselves generously, or at the very least to be rather optimistic in gauging their marginal productivity. To behave in this way is only human, especially since the necessary information is, in objective terms, highly imperfect. It may be excessive to accuse senior executives of having their “hands in the till,” but the metaphor is probably more apt than Adam Smith’s metaphor of the market’s “invisible hand.” In practice, the invisible hand does not exist, any more than “pure and perfect” competition does, and the market is always embodied in specific institutions such as corporate hierarchies and compensation committees.

This does not mean that senior executives and compensation committees can set whatever salaries they please and always choose the highest possible figure. “Corporate governance” is subject to certain institutions and rules specific to each country. The rules are generally ambiguous and flawed, but there are certain checks and balances. Each society also imposes certain social norms, which affect the views of senior managers and stockholders (or their proxies, who are often institutional investors such as financial corporations and pension funds) as well as of the larger society. These social norms reflect beliefs about the contributions that different individuals make to the firm’s output and to economic growth in general. Since uncertainty about these issues is great, it is hardly surprising that perceptions vary from country to country and period to period and are influenced by each country’s specific history. The important point is that it is very difficult for any individual firm to go against the prevailing social norms of the country in which it operates.

Without a theory of this kind, it seems to me quite difficult to explain the very large differences of executive pay that we observe between on the one hand the United States (and to a lesser extent in other English-speaking countries) and on the other continental Europe and Japan. Simply put, wage inequalities increased rapidly in the United States and Britain because US and British corporations became much more tolerant of extremely generous pay packages after 1970. Social norms evolved in a similar direction in European and Japanese firms, but the change came later (in the 1980s or 1990s) and has thus far not gone as far as in the United States. Executive compensation of several million euros a year is still more shocking today in Sweden, Germany, France, Japan, and Italy than in the United States or Britain. It has not always been this way—far from it: recall that in the 1950s and 1960s the United States was more egalitarian than France, especially in regard to the wage hierarchy. But it has been this way since 1980, and all signs are that this change in senior management compensation has played a key role in the evolution of wage inequalities around the world.

The Takeoff of the Supermanagers: A Powerful Force for Divergence

This approach to executive compensation in terms of social norms and acceptability seems rather plausible a priori, but in fact it only shifts the difficulty to another level. The problem is now to explain where these social norms come from and how they evolve, which is obviously a question for sociology, psychology, cultural and political history, and the study of beliefs and perceptions at least as much as for economics per se. The problem of inequality is a problem for the social sciences in general, not for just one of its disciplines. In the case in point, I noted earlier that the “conservative revolution” that gripped the United States and Great Britain in the 1970s and 1980s, and that led to, among other things, greater tolerance of very high executive pay, was probably due in part to a feeling that these countries were being overtaken by others (even though the postwar period of high growth in Europe and Japan was in reality an almost mechanical consequence of the shocks of the period 1914–1945). Obviously, however, other factors also played an important role.

To be clear, I am not claiming that all wage inequality is determined by social norms of fair remuneration. As noted, the theory of marginal productivity and of the race between technology and education offers a plausible explanation of the long-run evolution of the wage distribution, at least up to a certain level of pay and within a certain degree of precision. Technology and skills set limits within which most wages must be fixed. But to the extent that certain job functions, especially in the upper management of large firms, become more difficult to replicate, the margin of error in estimating the productivity of any given job becomes larger. The explanatory power of the skills-technology logic then diminishes, and that of social norms increases. Only a small minority of employees are affected, a few percent at most and probably less than 1 percent, depending on the country and period.

But the key fact, which was by no means evident a priori, is that the top centile’s share of total wages can vary considerably by country and period, as the disparate evolutions in the wealthy countries after 1980 demonstrate. The explosion of supermanager salaries should of course be seen in relation to firm size and to the growing diversity of functions within the firm. But the objectively complex problem of governance of large organizations is not the only issue. It is also possible that the explosion of top incomes can be explained as a form of “meritocratic extremism,” by which I mean the apparent need of modern societies, and especially US society, to designate certain individuals as “winners” and to reward them all the more generously if they seem to have been selected on the basis of their intrinsic merits rather than birth or background. (I will come back to this point.)

In any case, the extremely generous rewards meted out to top managers can be a powerful force for divergence of the wealth distribution: if the best paid individuals set their own salaries, (at least to some extent), the result may be greater and greater inequality. It is very difficult to say in advance where such a process might end. Consider again the case of the CFO of a large firm with gross revenue of 10 billion euros a year. It is hard to imagine that the corporate compensation committee would suddenly decide that the CFO’s marginal productivity is 1 billion or even 100 million euros (if only because it would then be difficult to find enough money to pay the rest of the management team). By contrast, some people might think that a pay package of 1 million, 10 million, or even 50 million euros a year would be justified (uncertainty about individual marginal productivity being so large that no obvious limit is apparent). It is perfectly possible to imagine that the top centile’s share of total wages could reach 15–20 percent in the United States, or 25–30 percent, or even higher.

The most convincing proof of the failure of corporate governance and of the absence of a rational productivity justification for extremely high executive pay is that when we collect data about individual firms (which we can do for publicly owned corporations in all the rich countries), it is very difficult to explain the observed variations in terms of firm performance. If we look at various performance indicators, such as sales growth, profits, and so on, we can break down the observed variance as a sum of other variances: variance due to causes external to the firm (such as the general state of the economy, raw material price shocks, variations in the exchange rate, average performance of other firms in the same sector, etc.) plus other “nonexternal” variances. Only the latter can be significantly affected by the decisions of the firm’s managers. If executive pay were determined by marginal productivity, one would expect its variance to have little to do with external variances and to depend solely or primarily on nonexternal variances. In fact, we observe just the opposite: it is when sales and profits increase for external reasons that executive pay rises most rapidly. This is particularly clear in the case of US corporations: Bertrand and Mullainhatan refer to this phenomenon as “pay for luck.”35

I return to this question and generalize this approach in Part Four (see Chapter 14). The propensity to “pay for luck” varies widely with country and period, and notably as a function of changes in tax laws, especially the top marginal income tax rate, which seems to serve either as a protective barrier (when it is high) or an incentive to mischief (when it is low)—at least up to a certain point. Of course changes in tax laws are themselves linked to changes in social norms pertaining to inequality, but once set in motion they proceed according to a logic of their own. Specifically, the very large decrease in the top marginal income tax rate in the English-speaking countries after 1980 (despite the fact that Britain and the United States had pioneered nearly confiscatory taxes on incomes deemed to be indecent in earlier decades) seems to have totally transformed the way top executive pay is set, since top executives now had much stronger incentives than in the past to seek large raises. I also analyze the way this amplifying mechanism can give rise to another force for divergence that is more political in nature: the decrease in the top marginal income tax rate led to an explosion of very high incomes, which then increased the political influence of the beneficiaries of the change in the tax laws, who had an interest in keeping top tax rates low or even decreasing them further and who could use their windfall to finance political parties, pressure groups, and think tanks.

{TEN}

Inequality of Capital Ownership

Let me turn now to the question of inequality of wealth and its historical evolution. The question is important, all the more so because the reduction of this type of inequality, and of the income derived from it, was the only reason why total income inequality diminished during the first half of the twentieth century. As noted, inequality of income from labor did not decrease in a structural sense between 1900–1910 and 1950–1960 in either France or the United States (contrary to the optimistic predictions of Kuznets’s theory, which was based on the idea of a gradual and mechanical shift of labor from worse paid to better paid types of work), and the sharp drop in total income inequality was due essentially to the collapse of high incomes from capital. All the information at our disposal indicates that the same is true for all the other developed countries.1 It is therefore essential to understand how and why this historic compression of inequality of wealth came about.

The question is all the more important because capital ownership is apparently becoming increasingly concentrated once again today, as the capital/income ratio rises and growth slows. The possibility of a widening wealth gap raises many questions as to its long-term consequences. In some respects it is even more worrisome than the widening income gap between supermanagers and others, which to date remains a geographically limited phenomenon.

Hyperconcentrated Wealth: Europe and America

As noted in Chapter 7, the distribution of wealth—and therefore of income from capital—is always much more concentrated than the distribution of income from labor. In all known societies, at all times, the least wealthy half of the population own virtually nothing (generally little more than 5 percent of total wealth); the top decile of the wealth hierarchy own a clear majority of what there is to own (generally more than 60 percent of total wealth and sometimes as much as 90 percent); and the remainder of the population (by construction, the 40 percent in the middle) own from 5 to 35 percent of all wealth.2 I also noted the emergence of a “patrimonial middle class,” that is, an intermediate group who are distinctly wealthier than the poorer half of the population and own between a quarter and a third of national wealth. The emergence of this middle class is no doubt the most important structural transformation to affect the wealth distribution over the long run.

Why did this transformation occur? To answer this question, one must first take a closer look at the chronology. When and how did inequality of wealth begin to decline? To be candid, because the necessary sources (mainly probate records) are unfortunately not always available, I have thus far not been able to study the historical evolution of wealth inequality in as many countries as I examined in the case of income inequality. We have fairly complete historical estimates for four countries: France, Britain, the United States, and Sweden. The lessons of these four histories are fairly clear and consistent, however, so that we can say something about the similarities and differences between the European and US trajectories.3 Furthermore, the wealth data have one enormous advantage over the income data: they allow us in some cases to go much farther back in time. Let me now examine one by one the four countries I have studied in detail.

France: An Observatory of Private Wealth

France is a particularly interesting case, because it is the only country for which we have a truly homogeneous historical source that allows us to study the distribution of wealth continuously from the late eighteenth century to the present. In 1791, shortly after the fiscal privileges of the nobility were abolished, a tax on estates and gifts was established, together with a wealth registry. These were astonishing innovations at the time, notable for their universal scope. The new estate tax was universal in three ways: first, it applied to all types of property: farmland, other urban and rural real estate, cash, public and private bonds, other kinds of financial assets such as shares of stock or partnerships, furniture, valuables, and so on; second, it applied to all owners of wealth, whether noble or common; and third, it applied to fortunes of all sizes, large or small. Moreover, the purpose of this fundamental reform was not only to fill the coffers of the new regime but also to enable the government to record all transfers of wealth, whether by bequest (at the owner’s death) or gift (during the owner’s lifetime), in order to guarantee to all the full exercise of their property rights. In official language, the estate and gift tax has always—from 1791 until now—been classified as one of a number of droits d’enregistrement (recording fees), and more specifically droits de mutation (transfer fees), which included both charges assessed on “free-will transfers,” or transfers of title to property made without financial consideration, by bequest or gift, and “transfers for consideration” (that is, transfers made in exchange for cash or other valuable tokens). The purpose of the law was thus to allow every property owner, large or small, to record his title and thus to enjoy his property rights in full security, including the right to appeal to the public authorities in case of difficulty. Thus a fairly complete system of property records was established in the late 1790s and early 1800s, including a cadastre for real estate that still exists today.

In Part Four I say more about the history of estate taxes in different countries. At this stage, taxes are of interest primarily as a historical source. In most other countries, it was not until the end of the nineteenth century or beginning of the twentieth that estate and gift taxes comparable to France’s were established. In Britain, the reform of 1894 unified previous taxes on the conveyance of real estate, financial assets, and personal estate, but homogeneous probate statistics covering all types of property go back only to 1919–1920. In the United States, the federal tax on estates and gifts was not created until 1916 and covered only a tiny minority of the population. (Although taxes covering broader segments of the population do exist in some states, these are highly heterogeneous.) Hence it is very difficult to study the evolution of wealth inequalities in these two countries before World War I. To be sure, there are many probate documents and estate inventories, mostly of private origin, dealing with particular subsets of the population and types of property, but there is no obvious way to use these records to draw general conclusions.

This is unfortunate, because World War I was a major shock to wealth and its distribution. One of the primary reasons for studying the French case is precisely that it will allow us to place this crucial turning point in a longer historical perspective. From 1791 to 1901, the estate and gift tax was strictly proportional: it varied with degree of kinship but was the same regardless of the amount transferred and was usually quite low (generally 1–2 percent). The tax was made slightly progressive in 1901 after a lengthy parliamentary battle. The government, which had begun publishing detailed statistics on the annual flow of bequests and donations as far back as the 1820s, began compiling a variety of statistics by size of estate in 1901, and from then until the 1950s, these became increasingly sophisticated (with cross-tabulations by age, size of estate, type of property, etc.). After 1970, digital files containing representative samples from estate and gift tax filings in a specific year became available, so that the data set can be extended to 2000–2010. In addition to the rich sources produced directly by the tax authorities over the past two centuries, I have also collected, together with Postel-Vinay and Rosenthal, tens of thousands of individual declarations (which have been very carefully preserved in national and departmental archives since the early nineteenth century) for the purpose of constructing large samples covering each decade from 1800–1810 to 2000–2010. All in all, French probate records offer an exceptionally rich and detailed view of two centuries of wealth accumulation and distribution.4

The Metamorphoses of a Patrimonial Society

Figure 10.1 presents the main results I obtained for the evolution of the wealth distribution from 1810 to 2010.5 The first conclusion is that prior to the shocks of 1914–1945, there was no visible trend toward reduced inequality of capital ownership. Indeed, there was a slight tendency for capital concentration to rise throughout the nineteenth century (starting from an already very high level) and even an acceleration of the inegalitarian spiral in the period 1880–1913. The top decile of the wealth hierarchy already owned between 80 and 85 percent of all wealth at the beginning of the nineteenth century; by the turn of the twentieth, it owned nearly 90 percent. The top centile alone owned 45–50 percent of the nation’s wealth in 1800–1810; its share surpassed 50 percent in 1850–1860 and reached 60 percent in 1900–1910.6

Looking at these data with the historical distance we enjoy today, we cannot help being struck by the impressive concentration of wealth in France during the Belle Époque, notwithstanding the reassuring rhetoric of the Third Republic’s economic and political elites. In Paris, which was home to little more than one-twentieth of the population in 1900–1910 but claimed one-quarter of the wealth, the concentration of wealth was greater still and seems to have increased without limit during the decades leading up to World War I. In the capital, where in the nineteenth century two-thirds of the population died without any wealth to leave to the next generation (compared with half of the population in the rest of the country) but where the largest fortunes were also concentrated, the top centile’s share was about 55 percent at the beginning of the century, rose to 60 percent in 1880–1890, and then to 70 percent on the eve of World War I (see Figure 10.2). Looking at this curve, it is natural to ask how high the concentration of wealth might have gone had there been no war.

FIGURE 10.1. Wealth inequality in France, 1810–2010

The top decile (the top 10 percent highest wealth holders) owns 80–90 percent of total wealth in 1810–1910, and 60–65 percent today.

Sources and series: see piketty.pse.ens.fr/capital21c.

The probate records also allow us to see that throughout the nineteenth century, wealth was almost as unequally distributed within each age cohort as in the nation as a whole. Note that the estimates indicated in Figures 10.1–2 (and subsequent figures) reflect inequality of wealth in the (living) adult population at each charted date: we start with wealth at the time of death but reweight each observation as a function of the number of living individuals in each age cohort as of the date in question. In practice, this does not make much difference: the concentration of wealth among the living is barely a few points higher than inequality of wealth at death, and the temporal evolution is nearly identical in each case.7

FIGURE 10.2. Wealth inequality in Paris versus France, 1810–2010

The top percentile (the top 1 percent wealth holders) owns 70 percent of aggregate wealth in Paris on the eve of World War I.

Sources and series: see piketty.pse.ens.fr/capital21c.

How concentrated was wealth in France during the eighteenth century up to the eve of the Revolution? Without a source comparable to the probate records created by the revolutionary assemblies (for the Ancien Régime we have only heterogeneous and incomplete sets of private data, as for Britain and the United States until the late nineteenth century), it is unfortunately impossible to make precise comparisons. Yet all signs are that inequality of private wealth decreased slightly between 1780 and 1810 owing to redistribution of agricultural land and cancellation of public debt during the Revolution, together with other shocks to aristocratic fortunes. It is possible that the top decile’s share attained or even slightly exceeded 90 percent of total wealth on the eve of 1789 and that the upper centile’s share attained or exceeded 60 percent. Conversely, the “émigré billion” (the billion francs paid to the nobility in compensation for land confiscated during the Revolution) and the return of the nobility to the forefront of the political scene contributed to the reconstitution of some old fortunes during the period of limited-suffrage monarchy (1815–1848). In fact, our probate data reveal that the percentage of aristocratic names in the top centile of the Parisian wealth hierarchy increased gradually from barely 15 percent in 1800–1810 to nearly 30 percent in 1840–1850 before embarking on an inexorable decline from 1850–1860 on, falling to less than 10 percent by 1890–1900.8

The magnitude of the changes initiated by the French Revolution should not be overstated, however. Beyond the probable decrease of inequality of wealth between 1780 and 1810, followed by a gradual increase between 1810 and 1910, and especially after 1870, the most significant fact is that inequality of capital ownership remained relatively stable at an extremely high level throughout the eighteenth and nineteenth centuries. During this period, the top decile consistently owned 80 to 90 percent of total wealth and the top centile 50 to 60 percent. As I showed in Part Two, the structure of capital was totally transformed between the eighteenth century and the beginning of the twentieth century (landed capital was almost entirely replaced by industrial and financial capital and real estate), but total wealth, measured in years of national income, remained relatively stable. In particular, the French Revolution had relatively little effect on the capital/income ratio. As just shown, the Revolution also had relatively little effect on the distribution of wealth. In 1810–1820, the epoch of Père Goriot, Rastignac, and Mademoiselle Victorine, wealth was probably slightly less unequally distributed than during the Ancien Régime, but the difference was really rather minimal: both before and after the Revolution, France was a patrimonial society characterized by a hyperconcentration of capital, in which inheritance and marriage played a key role and inheriting or marrying a large fortune could procure a level of comfort not obtainable through work or study. In the Belle Époque, wealth was even more concentrated than when Vautrin lectured Rastignac. At bottom, however, France remained the same society, with the same basic structure of inequality, from the Ancien Régime to the Third Republic, despite the vast economic and political changes that took place in the interim.

Probate records also enable us to observe that the decrease in the upper decile’s share of national wealth in the twentieth century benefited the middle 40 percent of the population exclusively, while the share of the poorest 50 percent hardly increased at all (it remained less than 5 percent of total wealth). Throughout the nineteenth and twentieth centuries, the bottom half of the population had virtually zero net wealth. In particular, we find that at the time of death, individuals in the poorest half of the wealth distribution owned no real estate or financial assets that could be passed on to heirs, and what little wealth they had went entirely to expenses linked to death and to paying off debts (in which case the heirs generally chose to renounce their inheritance). The proportion of individuals in this situation at the time of death exceeded two-thirds in Paris throughout the nineteenth century and until the eve of World War I, and there was no downward trend. Père Goriot belonged to this vast group, dying as he did abandoned by his daughters and in abject poverty: his landlady, Madame Vauquer, dunned Rastignac for what the old man owed her, and he also had to pay the cost of burial, which exceeded the value of the deceased’s meager personal effects. Roughly half of all French people in the nineteenth century died in similar circumstances, without any wealth to convey to heirs, or with only negative net wealth, and this proportion barely budged in the twentieth century.9

Inequality of Capital in Belle Époque Europe

The available data for other European countries, though imperfect, unambiguously demonstrate that extreme concentration of wealth in the eighteenth and nineteenth centuries and until the eve of World War I was a European and not just a French phenomenon.

In Britain, we have detailed probate data from 1910–1920 on, and these records have been exhaustively studied by many investigators (most notably Atkinson and Harrison). If we complete these statistics with estimates from recent years as well as the more robust but less homogeneous estimates that Peter Linder has made for the period 1810–1870 (based on samples of estate inventories), we find that the overall evolution was very similar to the French case, although the level of inequality was always somewhat greater in Britain. The top decile’s share of total wealth was on the order of 85 percent from 1810 to 1870 and surpassed 90 percent in 1900–1910; the uppermost centile’s share rose from 55–60 percent in 1810–1870 to nearly 70 percent in 1910–1920 (see Figure 10.3). The British sources are imperfect, especially for the nineteenth century, but the orders of magnitude are quite clear: wealth in Britain was extremely concentrated in the nineteenth century and showed no tendency to decrease before 1914. From a French perspective, the most striking fact is that inequality of capital ownership was only slightly greater in Britain than in France during the Belle Époque, even though Third Republic elites at the time liked to portray France as an egalitarian country compared with its monarchical neighbor across the Channel. In fact, the formal nature of the political regime clearly had very little influence on the distribution of wealth in the two countries.

FIGURE 10.3. Wealth inequality in Britain, 1810–2010

The top decile owns 80–90 percent of total wealth in 1810–1910, and 70 percent today.

Sources and series: see piketty.pse.ens.fr/capital21c

In Sweden, where the very rich data available from 1910, of which Ohlsonn, Roine, and Waldenstrom have recently made use, and for which we also have estimates for the period 1810–1870 (by Lee Soltow in particular), we find a trajectory very similar to what we observed in France and Britain (see Figure 10.4). Indeed, the Swedish wealth data confirm what we already know from income statements: Sweden was not the structurally egalitarian country that we sometimes imagine. To be sure, the concentration of wealth in Sweden in 1970–1980 attained the lowest level of inequality observed in any of our historical series (with barely 50 percent of total wealth owned by the top decile and slightly more than 15 percent by the top centile). This is still a fairly high level of inequality, however, and, what is more, inequality in Sweden has increased significantly since 1980–1990 (and in 2010 was just slightly lower than in France). It is worth stressing, moreover, that Swedish wealth was as concentrated as French and British wealth in 1900–1910. In the Belle Époque, wealth was highly concentrated in all European countries. It is essential to understand why this was so, and why things changed so much over the course of the twentieth century.

FIGURE 10.4. Wealth inequality in Sweden, 1810–2010

The top 10 percent holds 80–90 percent of total wealth in 1810–1910 and 55–60 percent today.

Sources and series: see piketty.pse.ens.fr/capital21c.

Note, moreover, that we also find the same extremely high concentration of wealth—with 80 to 90 percent of capital owned by the top decile and 50–60 percent by the top centile—in most societies prior to the nineteenth century, and in particular in traditional agrarian societies in the modern era, as well as in the Middle Ages and antiquity. The available sources are not sufficiently robust to permit precise comparisons or study temporal evolutions, but the orders of magnitude obtained for the shares of the top decile and centile in total wealth (and especially in total farmland) are generally close to what we find in France, Britain, and Sweden in the nineteenth century and Belle Époque.10

The Emergence of the Patrimonial Middle Class

Three questions will concern us in the remainder of this chapter. Why were inequalities of wealth so extreme, and increasing, before World War I? And why, despite the fact that wealth is once again prospering at the beginning of the twenty-first century as it did at the beginning of the twentieth century (as the evolution of the capital/income ratio shows), is the concentration of wealth today significantly below its historical record high? Finally, is this state of affairs irreversible?

In fact, the second conclusion that emerges very clearly from the French data presented in Figure 10.1 is that the concentration of wealth, as well as the concentration of income from wealth, has never fully recovered from the shocks of 1914–1945. The upper decile’s share of total wealth, which attained 90 percent in 1910–1920, fell to 60–70 percent in 1950–1970; the upper centile’s share dropped even more precipitously, from 60 percent in 1910–1920 to 20–30 percent in 1950–1970. Compared with the trend prior to World War I, the break is clear and overwhelming. To be sure, inequality of wealth began to increase again in 1980–1990, and financial globalization has made it more and more difficult to measure wealth and its distribution in a national framework: inequality of wealth in the twenty-first century will have to be gauged more and more at the global level. Despite these uncertainties, however, there is no doubt that inequality of wealth today stands significantly below its level of a century ago: the top decile’s share is now around 60–65 percent, which, though still quite high, is markedly below the level attained in the Belle Époque. The essential difference is that there is now a patrimonial middle class, which owns about a third of national wealth—a not insignificant amount.

The available data for the other European countries confirm that this has been a general phenomenon. In Britain, the upper decile’s share fell from more than 90 percent on the eve of World War I to 60–65 percent in the 1970s; it is currently around 70 percent. The top centile’s share collapsed in the wake of the twentieth century’s shocks, falling from nearly 70 percent in 1910–1920 to barely more than 20 percent in 1970–1980, then rising to 25–30 percent today (see Figure 10.3). In Sweden, capital ownership was always less concentrated than in Britain, but the overall trajectory is fairly similar (see Figure 10.4). In every case, we find that what the wealthiest 10 percent lost mainly benefited the “patrimonial middle class” (defined as the middle 40 percent of the wealth hierarchy) and did not go to the poorest half of the population, whose share of total wealth has always been minuscule (generally around 5 percent), even in Sweden (where it was never more than 10 percent). In some cases, such as Britain, we find that what the richest 1 percent lost also brought significant gains to the next lower 9 percent. Apart from such national specificities, however, the general similarity of the various European trajectories is quite striking. The major structural transformation was the emergence of a middle group, representing nearly half the population, consisting of individuals who managed to acquire some capital of their own—enough so that collectively they came to own one-quarter to one-third of the nation’s total wealth.

Inequality of Wealth in America

I turn now to the US case. Here, too, we have probate statistics from 1910–1920 on, and these have been heavily exploited by researchers (especially Lampman, Kopczuk, and Saez). To be sure, there are important caveats associated with the use of these data, owing to the small percentage of the population covered by the federal estate tax. Nevertheless, estimates based on the probate data can be supplemented by information from the detailed wealth surveys that the Federal Reserve Bank has conducted since the 1960s (used notably by Arthur Kennickell and Edward Wolff), and by less robust estimates for the period 1810–1870 based on estate inventories and wealth census data exploited respectively by Alice Hanson Jones and Lee Soltow.11

Several important differences between the European and US trajectories stand out. First, it appears that inequality of wealth in the United States around 1800 was not much higher than in Sweden in 1970–1980. Since the United States was a new country whose population consisted largely of immigrants who came to the New World with little or no wealth, this is not very surprising: not enough time had passed for wealth to be accumulated or concentrated. The data nevertheless leave much to be desired, and there is some variation between the northern states (where estimates suggest a level of inequality lower than that of Sweden in 1970–1980) and southern states (where inequality was closer to contemporary European levels).12

It is a well-established fact that wealth in the United States became increasingly concentrated over the course of the nineteenth century. In 1910, capital inequality there was very high, though still markedly lower than in Europe: the top decile owned about 80 percent of total wealth and the top centile around 45 percent (see Figure 10.5). Interestingly, the fact that inequality in the New World seemed to be catching up with inequality in old Europe greatly worried US economists at the time. Willford King’s book on the distribution of wealth in the United States in 1915—the first broad study of the question—is particularly illuminating in this regard.13 From today’s perspective, this may seem surprising: we have been accustomed for several decades now to the fact that the United States is more inegalitarian than Europe and even that many Americans are proud of the fact (often arguing that inequality is a prerequisite of entrepreneurial dynamism and decrying Europe as a sanctuary of Soviet-style egalitarianism). A century ago, however, both the perception and the reality were strictly the opposite: it was obvious to everyone that the New World was by nature less inegalitarian than old Europe, and this difference was also a subject of pride. In the late nineteenth century, in the period known as the Gilded Age, when some US industrialists and financiers (for example John D. Rockefeller, Andrew Carnegie, and J. P. Morgan) accumulated unprecedented wealth, many US observers were alarmed by the thought that the country was losing its pioneering egalitarian spirit. To be sure, that spirit was partly a myth, but it was also partly justified by comparison with the concentration of wealth in Europe. In Part Four we will see that this fear of growing to resemble Europe was part of the reason why the United States in 1910–1920 pioneered a very progressive estate tax on large fortunes, which were deemed to be incompatible with US values, as well as a progressive income tax on incomes thought to be excessive. Perceptions of inequality, redistribution, and national identity changed a great deal over the course of the twentieth century, to put it mildly.

FIGURE 10.5. Wealth inequality in the United States, 1810–2010

The top 10 percent wealth holders own about 80 percent of total wealth in 1910 and 75 percent today.

Sources and series: see piketty.pse.ens.fr/capital21c.

FIGURE 10.6. Wealth inequality in Europe versus the United States, 1810–2010

Until the mid-twentieth century, wealth inequality was higher in Europe than in the United States.

Sources and series: see piketty.pse.ens.fr/capital21c.

Inequality of wealth in the United States decreased between 1910 and 1950, just as inequality of income did, but much less so than in Europe: of course it started from a lower level, and the shocks of war were less violent. By 2010, the top decile’s share of total wealth exceeded 70 percent, and the top centile’s share was close to 35 percent.14

In the end, the deconcentration of wealth in the United States over the course of the twentieth century was fairly limited: the top decile’s share of total wealth dropped from 80 to 70 percent, whereas in Europe it fell from 90 to 60 percent (see Figure 10.6).15

The differences between the European and US experiences are clear. In Europe, the twentieth century witnessed a total transformation of society: inequality of wealth, which on the eve of World War I was as great as it had been under the Ancien Régime, fell to an unprecedentedly low level, so low that nearly half the population were able to acquire some measure of wealth and for the first time to own a significant share of national capital. This is part of the explanation for the great wave of enthusiasm that swept over Europe in the period 1945–1975. People felt that capitalism had been overcome and that inequality and class society had been relegated to the past. It also explains why Europeans had a hard time accepting that this seemingly ineluctable social progress ground to a halt after 1980, and why they are still wondering when the evil genie of capitalism will be put back in its bottle.

In the United States, perceptions are very different. In a sense, a (white) patrimonial middle class already existed in the nineteenth century. It suffered a setback during the Gilded Age, regained its health in the middle of the twentieth century, and then suffered another setback after 1980. This “yo-yo” pattern is reflected in the history of US taxation. In the United States, the twentieth century is not synonymous with a great leap forward in social justice. Indeed, inequality of wealth there is greater today than it was at the beginning of the nineteenth century. Hence the lost US paradise is associated with the country’s beginnings: there is nostalgia for the era of the Boston Tea Party, not for Trente Glorieuses and a heyday of state intervention to curb the excesses of capitalism.

The Mechanism of Wealth Divergence: r versus g in History

Let me try now to explain the observed facts: the hyperconcentration of wealth in Europe during the nineteenth century and up to World War I; the substantial compression of wealth inequality following the shocks of 1914–1945; and the fact that the concentration of wealth has not—thus far—regained the record heights set in Europe in the past.

Several mechanisms may be at work here, and to my knowledge there is no evidence that would allow us to determine the precise share of each in the overall movement. We can, however, try to hierarchize the different mechanisms with the help of the available data and analyses. Here is the main conclusion that I believe we can draw from what we know.

The primary reason for the hyperconcentration of wealth in traditional agrarian societies and to a large extent in all societies prior to World War I (with the exception of the pioneer societies of the New World, which are for obvious reasons very special and not representative of the rest of the world or the long run) is that these were low-growth societies in which the rate of return on capital was markedly and durably higher than the rate of growth.

This fundamental force for divergence, which I discussed briefly in the Introduction, functions as follows. Consider a world of low growth, on the order of, say, 0.5–1 percent a year, which was the case everywhere before the eighteenth and nineteenth centuries. The rate of return on capital, which is generally on the order of 4 or 5 percent a year, is therefore much higher than the growth rate. Concretely, this means that wealth accumulated in the past is recapitalized much more quickly than the economy grows, even when there is no income from labor.

For example, if g = 1% and r = 5%, saving one-fifth of the income from capital (while consuming the other four-fifths) is enough to ensure that capital inherited from the previous generation grows at the same rate as the economy. If one saves more, because one’s fortune is large enough to live well while consuming somewhat less of one’s annual rent, then one’s fortune will increase more rapidly than the economy, and inequality of wealth will tend to increase even if one contributes no income from labor. For strictly mathematical reasons, then, the conditions are ideal for an “inheritance society” to prosper—where by “inheritance society” I mean a society characterized by both a very high concentration of wealth and a significant persistence of large fortunes from generation to generation.

Now, it so happens that these conditions existed in any number of societies throughout history, and in particular in the European societies of the nineteenth century. As Figure 10.7 shows, the rate of return on capital was significantly higher than the growth rate in France from 1820 to 1913, around 5 percent on average compared with a growth rate of around 1 percent. Income from capital accounted for nearly 40 percent of national income, and it was enough to save one-quarter of this to generate a savings rate on the order of 10 percent (see Figure 10.8). This was sufficient to allow wealth to grow slightly more rapidly than income, so that the concentration of wealth trended upward. In the next chapter I will show that most wealth in this period did come from inheritance, and this supremacy of inherited capital, despite the period’s great economic dynamism and impressive financial sophistication, is explained by the dynamic effects of the fundamental inequality r > g: the very rich French probate data allow us to be quite precise about this point.

FIGURE 10.7. Return to capital and growth: France, 1820–1913

The rate of return on capital is a lot higher than the growth rate in France between 1820 and 1913.

Sources and series: see piketty.pse.ens.fr/capital21c.

FIGURE 10.8. Capital share and saving rate: France, 1820–1913

The share of capital income in national income is much larger than the saving rate in France between 1820 and 1913.

Sources and series: see piketty.pse.ens.fr/capital21c.

Why Is the Return on Capital Greater Than the Growth Rate?

Let me pursue the logic of the argument. Are there deep reasons why the return on capital should be systematically higher than the rate of growth? To be clear, I take this to be a historical fact, not a logical necessity.

It is an incontrovertible historical reality that r was indeed greater than g over a long period of time. Many people, when first confronted with this claim, express astonishment and wonder why it should be true. The most obvious way to convince oneself that r > g is indeed a historical fact is no doubt the following.

As I showed in Part One, economic growth was virtually nil throughout much of human history: combining demographic with economic growth, we can say that the annual growth rate from antiquity to the seventeenth century never exceeded 0.1–0.2 percent for long. Despite the many historical uncertainties, there is no doubt that the rate of return on capital was always considerably greater than this: the central value observed over the long run is 4–5 percent a year. In particular, this was the return on land in most traditional agrarian societies. Even if we accept a much lower estimate of the pure yield on capital—for example, by accepting the argument that many landowners have made over the years that it is no simple matter to manage a large estate, so that this return actually reflects a just compensation for the highly skilled labor contributed by the owner—we would still be left with a minimum (and to my mind unrealistic and much too low) return on capital of at least 2–3 percent a year, which is still much greater than 0.1–0.2 percent. Thus throughout most of human history, the inescapable fact is that the rate of return on capital was always at least 10 to 20 times greater than the rate of growth of output (and income). Indeed, this fact is to a large extent the very foundation of society itself: it is what allowed a class of owners to devote themselves to something other than their own subsistence.

In order to illustrate this point as clearly as possible, I have shown in Figure 10.9 the evolution of the global rate of return on capital and the growth rate from antiquity to the twenty-first century.

FIGURE 10.9. Rate of return versus growth rate at the world level, from Antiquity until 2100

The rate of return to capital (pretax) has always been higher than the world growth rate, but the gap was reduced during the twentieth century, and might widen again in the twenty-first century.

Sources and series: see piketty.pse.ens.fr/capital21c

These are obviously approximate and uncertain estimates, but the orders of magnitude and overall evolutions may be taken as valid. For the global growth rate, I have used the historical estimates and projections discussed in Part One. For the global rate of return on capital, I have used the estimates for Britain and France in the period 1700–2010, which were analyzed in Part Two. For early periods, I have used a pure return of 4.5 percent, which should be taken as a minimum value (available historical data suggest average returns on the order of 5–6 percent).16 For the twenty-first century, I have assumed that the value observed in the period 1990–2010 (about 4 percent) will continue, but this is of course uncertain: there are forces pushing toward a lower return and other forces pushing toward a higher. Note, too, that the returns on capital in Figure 10.8 are pretax returns (and also do not take account of capital losses due to war, or of capital gains and losses, which were especially large in the twentieth century).