The Specter of Inequality

Ein Gespenst geht um in Europa

The French economist Thomas Piketty has focussed the attention of professional economists and the public on the old Marxist question: are capitalist societies driven inexorably to states of extreme inequality?

The question itself has an interesting recent history. The work of the Belarusian-American economist and 1971 Nobel laureate Simon Kuznets, particularly his acclaimed 1955 study, Economic Growth and Income Inequality, dominated the postwar consensus.¹ Income inequality, he argued, is destined to fade with advanced economic development. Over time, levels of inequality will naturally conform to a curve in the shape of an inverted U. Inequality increases in the early phases of industrialization. It then decreases spontaneously in more advanced stages of economic development, as wages rise, new entrepreneurs generate competition, and the shift to more productive "industries creates new economic elites.

But Thomas Piketty offers us quite a different vision in Capital in the Twenty-First Century.² For the past fifteen years, he has led an international team of economists in an impressive study, one designed to enlarge the data set by mining evidence from long-forgotten archives. From this, he has derived three conclusions that defy the conventional wisdom:

1. First, the income inequality contraction phase ended in the 1960s. Since the 1970s, we have seen a strong resurgence of inequality, which is now approaching a record corresponding, more or less, to the very elevated figures that characterized the Belle Époque.
2. Next, if indeed there are mechanisms that reduce inequality in the context of capitalist growth, they have been offset, increasingly, by countervailing mechanisms that promote inequality. These mechanisms are linked to a capital accumulation and concentration process that is all the more powerful now that we have returned to weak economic growth.
3. Finally, these developments, in tandem, must ultimately and inevitably lead to great extremes of inequality. The figures are destined to exceed even those seen at the dawn of the Industrial Revolution, making it plausible to imagine a global and “potentially terrifying” political catastrophe: hence Piketty’s proposal to act prophylactically by raising taxes on the very wealthy and imposing a global wealth tax.³

Thus the postwar phenomena that we understood to represent a radical break from the economic and social universe of yesteryear—the rapid growth, the disappearance of the permanent rentier class, the growth-related downward trend in inequality, the decreasing primacy of inheritance, the increasing role of human capital, the correlation of advancement with work and merit—all of this was in fact a mere contingency related to the catastrophic consequences of the First World War. The entire edifice is no more than an illusion, one that is now crumbling before our eyes thanks to the figures revealed in this book.

But is this true?

The Art of Tracking the Data

Capital in the Twenty-First Century is well-written and well-constructed. Although thick, the book reads easily. The tone is compelling. The most visible academic authorities in the world of Anglo-Saxon economic science received it more than favorably.⁴ All the same, the thesis has serious shortcomings, both methodological and theoretical.

Two graphs play a prominent role in Piketty’s case. The first measures the evolution of income inequality in the United States.⁵ It traces the growth in income among the wealthiest 10 percent of American taxpayers between 1910 and 2010. This is a simple extension of the historical series established by Kuznets in the 1950s. From 1913 to 1948, we see the strong compression of inequality that Kuznets had observed; there is a redistribution of nearly 15 percent of total income to the disadvantage of the top decile of the population. (The top 10 percent held 45 to 50 percent of national income between 1910 and 1920, but only some 30 to 35 percent by the end of the 1940s.) Inequality then stabilized at this level from 1950 to 1970.

But we have seen a very rapid reversal of this trend since the 1970s. This change is linked to a 15 percent rise, which is considerable, in national income among the top decile. According to the author, this indicates that “... income inequality has exploded in the United States.”⁶ The concentration of wealth among those with the highest incomes has returned to the levels of the early twentieth century: now, again, they hold roughly 45 to 50 percent of it. “The shape of the curve,” says Piketty, “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.”⁷

Piketty takes the analysis further: during the thirty years prior to the Great Recession,⁸ he concludes, the richest one percent were the beneficiaries of three-quarters of total American growth.⁹ For 90 percent of Americans, average income growth was only 0.5 percent per annum.

If these figures are striking, they must nonetheless be viewed with much caution, especially where they concern the top percentile of the population, which is the most spectacular part of the display. Let us consider the method used to collect and create these data sets. Piketty and his team have aggregated them from a survey of a century’s worth of tax records, income tax having been introduced in the United States in 1910. Given that the analysis is based on tax records that range over such a long period, one has to wonder what effect changes and developments in the tax law might have had on the numbers. How did changes to tax rates, taxpayer categorization, and tax-assessment criteria affect the figures reported in these records, or the way they appear to represent patterns of national income and its distribution?

The postwar period was characterized by stability in American tax law. Nevertheless, many new laws were passed after President Ronald Reagan’s election, including the 1981 Economic Recovery Tax Act, the 1986 Tax Reform Act, the 1997 Taxpayer Relief Act, and the 2003 Jobs and Growth Tax Relief Reconciliation Act. These greatly changed the rules for taxing various types of capital income, as well as the ways various professional incomes were reported, the structure and level of tax rates, and the system of tax benefits associated with certain activities and investment choices.

Especially in the highest income bracket (where taxable income is highly elastic), the lowering of marginal rates dramatically increased the total amount of tax paid by those at the top of the income hierarchy. But this did not in reality amount to an equivalent enlargement of the gap in primary income. For example, when in 1981 the Administration reduced the maximum rate of tax on interest, dividends, and other investment income from 70 to 50 percent, the wealthiest investors swapped low-yield, untaxed assets (municipal bonds, for example) for riskier, taxable investments with much higher yields.

Similarly, until 1986, the Internal Revenue Service created incentives via the tax code for the highest-income taxpayers to manage income derived from commercial and professional activities through legal structures subject to corporate tax rates, which were more attractive than personal income tax rates. But the 1986 reforms inverted this, prompting the same taxpayers to report business-generated income on their personal tax returns instead, using so-called flow-through entities. The income of such an entity is treated, for tax purposes, as the income of the investors or owners. As a result, figures collected from tax returns on file with IRS show growth that does not in fact correspond to a real growth in the filers’ real incomes, and this creates the illusion that the highest percentile took a larger share of the wealth. In reality, this reflects a change in the legal regime.¹⁰

Conversely, since the 1980s, the IRS has taken an ever-increasing share of middle-class income out of taxable income (incomes from capital gains or real-estate windfall profits). Recall the introduction in 1997 of a US$500,000 deductible on capital gains tax from the sale of homes, and the introduction in the 1980s of tax-deductible 401(k) plans, as well as tax-sheltered retirement and college savings plans. Subtracting these tax-exempt incomes from the total income declared by middle-class households creates a widening gap between top and middle global incomes as recorded by IRS data. Here again, the widening of the gap does not reflect any increase in the difference in real earnings between top- and middle-income classes. The image of a widening inequality gap is just an illusion created by legal changes in the tax regime.

These are not trivial effects. In 2012, the economist Alan Reynolds calculated that these changes to the tax code explained more than half the nominal increase in the share of total income among the top one percent of taxpayers since 1983, and almost all of the increase recorded since 2000.¹¹ The technique of assessing the share of the richest taxpayers through the examination of national income tax records is further flawed in that it excludes—by definition, in fact—a proper accounting of supplementary and replacement income, as well as public or private transfers and subsidies. These nonetheless represent a growing proportion of real income in the poorest households.

The inevitable consequence of this technique is the overestimation of the growth of the top portion of the revenue-and-inheritance hierarchy relative to the rest of the population.¹² From 2000 to 2012, estimated income that is not measured by the IRS as part of American households’ total real disposable income has risen from 33 percent of that total to 38 percent. Piketty’s overestimate for the period from 1979 to 2010 is apt to represent about a quarter of the increase obtained by the one percent elite as a share of global national income.¹³

This analysis is far from trifling, insofar as the graph described above plays the key role in the unfolding nightmare scenario that Piketty projects into the decades ahead. When the same operations are performed on the European data, particularly the French data, very different results are obtained. The American U-curve disappears, giving way to an L-curve.¹⁴ The tendency toward the amplification of the gap between the top and the bottom of the income hierarchy seen in the United States, and to a lesser extent in the United Kingdom, is barely noticeable, even though the author senses that “... the trend is in the same direction.”¹⁵ It is in large measure by extrapolating from the trend he observes in the United States and applying it to other developed countries—even to emerging economies—that he derives his conclusion of a threatened, socially and politically unstable twenty-first century, one characterized by extreme inequality: a vision that leads Piketty to ask for worldwide mobilization to forestall a “terrifying” catastrophe.

Housing and Rent

The book’s second critical graph traces the evolution of the capital-to-income ratio in France, the United Kingdom, and Germany.¹⁶ It shows the change in the total value of private assets (real estate, financial, and professional, net of debt), expressed in years of national income, from the 1870s to the 2010s. Since 1950, we have seen a huge increase in total asset holding, so much that the numbers today appear to be approaching levels last witnessed only before the Great War. This high-amplitude U-shaped curve is central to Piketty’s argument. But again, there is cause for serious doubt about the probative value of this series.

In economics, we traditionally draw a distinction between productive capital (everything that contributes to physical production capacity) and unproductive capital (durable goods, primarily those intended for leisure or personal consumption). Yet in Piketty’s world, capital encompasses everything: houses and other dwellings, arable land, tools, equipment, machinery, financial instruments such as stocks and bonds, patents, and even intellectual property. This can only lead to trouble. He includes housing in the global measure of capital, for example, but assesses its value by means of conventional national accounting techniques that report transaction values. Housing, however, is a special kind of good. It is both a consumer good and an investment in which returns are determined by rental prices (both real, when received by landlords, and implicit, in the form of homeowners’ savings on rent they would otherwise be obliged to pay). Thus it is changes in the ratio of rental prices to disposable income, and more specifically, changes in the ratio of capitalized lease prices to household disposable income, that should be taken into account when calculating the capital-to-income ratios.

In principle, rents and real estate prices should grow in parallel, more or less, which should allow us to view them as much of a muchness. But that is not what happened in the postwar period. Take France, for example. There has been remarkable stability in the ratio of rents to disposable incomes over the past few decades. Real estate prices, on the other hand, have surged by 60 percent in relation to household disposable income.¹⁷ What this contrast tells us is that the spectacular rise in the value of housing assets—which appear, on Piketty’s graphs, as the key to the rise over the past twenty years, of the capital-to-income ratio—is a phenomenon related mainly to prices.¹⁸ It is a property bubble that owes its origin to monetary policy (low interest rates, the creation of the single currency in Europe) and also to something we often forget: the massive rise of land planning schemes that exacerbates land shortages by depleting supply. What we have seen in the housing sector reflects this more than it does the cumulative compounding of returns on capital. While it is true that bubbles have transfer effects, which exacerbate inequality, rising real estate prices do not augment the incomes of homeowners who are renting out their properties. Some people’s gains are offset by other people’s losses. These effects are quite different from those Piketty analyzes, and are not a cause of exponential divergence.¹⁹

A team of researchers from the Paris Institute of Political Studies, known as Sciences Po, recalculated the housing component of the capital-to-income ratio in Piketty’s argument, replacing numbers from national accounts with assessments based on the capitalized value of rents. Once they estimated the value of housing capital correctly, they found that the capital-to-income ratio is no higher today than it was in the 1950s.²⁰

They next applied these revised calculations to the other countries for which price and rents series are available: the United States, the United Kingdom, Canada, and Germany. They found that since 1970, “[a]side from Germany, where this ratio appears to be lower than in other countries (Germans do not own as frequently as in other countries), the ratio of capital over income has remained stable in the other countries.”²¹ Further research, conducted on OECD data, confirms this diagnosis.²² Moreover, quite beyond technical issues of measurement, we should not forget that 56 percent of all housing stock in France, and as much as 70 percent in Britain, is owner-occupied.

We are thus very far from late nineteenth-century England, where only 36,000 members of the twenty-million-strong population were homeowners. So even assuming that Piketty’s hypothesis about the statistics is correct—to wit, that we’re in a new phase of accelerating income inequality as revealed by changes in the capital-to-income ratio—the situation is nothing like that of the Belle Époque, even though the book refers to it over and over. Once we understand that housing accounts for more than half of reported national capital relative to income, we see that the increase in total asset value since the 1970s, and especially since the 1990s, in fact mostly reflects the relative enrichment of the middle class at the expense of those at the top and bottom of the income hierarchy.²³ There is no explosive trend toward the concentration of wealth among a tiny caste of super-rentiers.

The Reemergence of Capital

The book’s strength is not just a matter of the rich documentary record to which it appeals. The thesis that Piketty develops is embedded in a formal model, one expressed as a series of macro-economic equations. It is very difficult for a professional economist to be taken seriously without an apparatus of this sort. The model is expressed by an inequality and two identities:

1. The inequality r > g, where r designates the rate of return on capital and g, the rate of growth of income and production. This inequality, Piketty asserts, expresses the “central contradiction” of capitalism itself.²⁴
2. The identity (or the first fundamental law of capitalism): α = r × β, where α designates the income derived from capital as a fraction of annual income, r is again the rate of return on capital, and β the value of an economy’s accumulated capital stock as a fraction of annual national income.²⁵
3. The identity (or the second fundamental law of capitalism): β = s / g, where β represents the ratio of capital to income, s the savings rate, and g the growth rate of a given economy.²⁶

From these premises, Piketty derives the following conclusions:

1. If the rate of return on capital (r) displays a stable long-term tendency greatly to exceed the rate of growth (g) of income and production, then it is likely that the distribution of wealth will be skewed in favor of the fraction of national income derived from capital.
2. This leads inexorably to an increase in the fraction of income (α) derived from capital in an economy’s annual income flow, and thus to a cumulative process of enrichment and capital accumulation.
3. Given that the return on capital (r) is relatively stable over the long-term, any sustained slowdown in the growth rate (g) causes an acceleration in the rate of capital accumulation with respect to other economic factors. The return on capital (r) then gradually converges to a stable limiting value in virtue of the law of diminishing marginal utility and the relationship β = s / g.

In Piketty’s view, this process determines the value of both the figures and the shape of the curves in his book, especially those from which he derives his alarming projections.

Piketty’s model is simple, elegant, and consistent. No neoclassical model of growth had, before Piketty, integrated the production, accumulation, and distribution of wealth in such a pithy play of equations. Still, Piketty’s model, it is important to observe, must meet certain very specific conditions if it is to justify his dynamic predictions. The model demands that the reduction or fall in the rate of return on capital induced by the expansion of capital accumulation be neither too strong nor too fast. If the descent is too rapid, for example, the tendency toward accumulation will result, in only a few years—not decades—in a reduction of the share of capital to annual income, even if the share of total assets continues to grow. This is precisely the opposite of what is supposed to happen.

This in turn implies that the coefficient of elasticity between capital and labor must be greater than one. Only this way can there be both an increase in the capital-to-income ratio (inventory effect), and an increase in capital income with respect to labor income (flow effect). If the coefficient of elasticity is less than one, the increase in the capital/income ratio is offset by a reduction in the share of capital within the flow of income. The assumption of positive elasticity greater than one thus functions as a premise in Piketty’s model, a claim about both the past and the future.

Piketty is well aware of this. His response: “Over a very long period of time, the elasticity of substitution between capital and labor seems to have been greater than one ... On the basis of historical data, one can estimate an elasticity between 1.3 and 1.6.”²⁷ This estimate has been severely contested among English-speaking economists (especially Matthew Rognlie).²⁸ The literature is extensive, and various claims are very often confused. Nonetheless, very little of this body of work supports Piketty’s thesis that capital and labor elasticities are, or have been, greater than one. There is a rough consensus that the ratio has a mean value between 0.40 and 0.60.

Most of the estimates, it is true, are derived from gross capital measures. These include depreciation. Piketty, on the other hand, uses net capital measures throughout. These exclude depreciation. This does little to strengthen Piketty’s argument. Net elasticity is invariably less than gross elasticity. Piketty’s appeal to historical data is, moreover, restricted to data that he has himself calculated. Empirical observations, he argues, confirm the dominant trend of high elasticity. But this is only true if a specific attribute of the model, stability of the real price of capital, is also to be found in the real world. Given the heterogeneous nature of capital, this is a very risky assumption—as the above example of housing indicates.

The discussion (or debate) has just begun. No one is yet entitled to draw firm conclusions. These issues will be debated for years among economists. If firm conclusions are not yet possible, a certain degree of caution with respect to Piketty’s predictions is obviously desirable.

Management and Wealth

The note of caution applies in equal measure to Piketty’s statements about the rise, in the Western world, of a new generation of super-executives, and the bankruptcy of the traditional capitalist model of corporate governance. He defends the thesis that present-day inequality is owed primarily to the disproportionate explosion in high wages, and attributes this phenomenon to the structural deterioration of control mechanisms within very large companies: senior management, in his lights, has acquired power over managing directors and the shareholders who sit on salary committees. But does this correspond to the facts? Piketty has been so focused on reviewing tax data that it would seem he has not bothered to subject this surmise to a confrontation with the data, which researchers have recently updated from other sources.

Pundits, journalists, and politicians are convinced alike that there has been a stratospheric surge in the remuneration paid to CEOs by the largest corporations, and that this corresponds to the reality of a business world in which the gap between those earning the highest salaries and the rest only enlarges exponentially. They do not seem to be are aware that if, in fact, top-level salaries did increase dramatically during the 1990s, the average compensation of the CEO of an S&P 500 company nevertheless declined by 46 percent between 2000 and 2010.²⁹ Nor do they seem aware that during the same period, the median value of this remuneration has in turn risen by 8 percent since 2000, but fallen by 7 percent compared to 2001.³⁰ The convergence of the mean and median values implies that since 2000, boards of directors have been much less likely to pay their CEOs ever-more-exorbitant windfall emoluments.³¹

A similar follow-up study was performed on a sample of one thousand CEOs of smaller companies that are not part of the S&P 500. It shows comparable changes in patterns of remuneration, or certainly, it shows no obvious disparity. In both cases, the average CEO’s salary—which in the larger companies is still two hundred times the average annual income of an American household—returned, in 2010, to 1998 levels. It should moreover be noted that these figures apply to a new generation of CEOs, whose tenures at the top have, over this period, significantly decreased (from eight to six years, on average, if mergers and acquisitions are included). According to the economist Steven Kaplan, this reflects an increase in the cost of pursuing a top managerial career; it does not really correspond to the putative emergence of mega-managers who name their own salaries and control the very people who assign and define the terms of their jobs. Finally, a comparison of the average income of CEOs in the S&P 500 with the average income of taxpayers in the top 0.1 percent of the income hierarchy, measured by declared tax revenue—a comparison Piketty makes in the book—shows that this ratio, too, has very much declined since its peak in 2001. It has stabilized at levels corresponding to those of the mid-1990s.³²

In other words, the gain in their incomes vis-à-vis other professions in the same income cohort hardly amounts to a trend that would best be described as “exponential drift,” as Piketty would have it. And to put this observation in a longer-term perspective, note that in 2007, the ratio was about the same as it was in the late 1930s. To be sure, at that time J. K. Galbraith already worried about a rise in managerial power with respect to that of shareholders, but this was certainly not yet a real issue.

These observations are not sufficient forever to obviate the book’s thesis. No one doubts that we are confronted with a research topic destined to nourish professional debate for many years to come. But they do suggest that traditional factors—the market and competition, professional competence and efficiency (otherwise known as talent), and the technological changes that increasingly promote the efficient management of ever larger and more complex organizations—are indeed mostly driving the rise in higher wages, even in the largest companies. These observations thus do contradict Piketty’s assertion that from now on, economists should treat the neoclassical concept of marginal productivity as an obsolete research tool, at least when attempting to explain the remuneration of senior managers. Certainly, the observations here lend no support to the notion that we are witnessing the collapse of the traditional model of corporate governance. In reality, economies and markets do not function in the stylized and simplistic fashion they do in Piketty’s book, which reflects throughout his resolutely macroeconomic and deterministic approach.

Indisputably, the wages earned by a handful of the best-paid managers in the largest companies are astronomical in absolute terms: in 2010, three American CEOs in the S&P 500 earned more than fifty million US dollars per year apiece, compared to an average top salary for S&P 500 CEOs of about ten million a year. This is, in itself, a legitimate source of ethical and political concern. But it is surely not by throwing ourselves willy-nilly into the great Utopia of a frankly confiscatory global tax regime—Piketty’s proposals would result in an effective marginal tax rate of about 330 percent in the United States—that we will solve the problem of modern society’s extreme distaste for inequality.³³

It would be a modest improvement, perhaps, to begin by focusing our attention on the nexus of national laws, regulations, and tax schemes, which in our democratic societies—through collusion, mutual dependence, and chronic rent-seeking among politicians, bureaucrats, corporations, and civil society organizations (such as trade unions and NGOs)—keep generating a rising and endless search for virtually indestructible legal privileges. But this is never mentioned, not even as an afterthought, in a volume comprising almost seven hundred pages. That there is no mention of public choice theory and its relevance to the question of inequality is surely not the least of the defects in a book that styles itself as encyclopedic in its scholarship.

In Conclusion

Ultimately, the most remarkable aspect of Capital in the Twenty-First Century is the case it makes for the author’s political entrepreneurialism. The tremendous acclaim the book received, almost immediately, has less to do with the intrinsic quality of its content than its utility. Just as John Maynard Keynes did eighty years ago in the General Theory, Piketty offers a cloak of science and theoretical language to leaders who seek to legitimize the growingly intrusive ambitions of the modern state. In a radio program broadcast on France Culture in September 2013, Piketty offered forthrightly that his book was a political book.³⁴ Duly acknowledged.

Translated from the French by the Editors.