The Economics of Inequality - Thomas Piketty
The Economics of Inequality - Thomas Piketty
The Economics of Inequality - Thomas Piketty
Thomas Piketty
Translated by Arthur Goldhammer
978-0-674-50480-6 (hardcover)
978-0-674-91558-9 (EPUB)
978-0-674-91557-2 (MOBI)
978-0-674-91556-5 (PDF)
Introduction
1. The Measurement of Inequality and Its Evolution
2. Capital-Labor Inequality
3. Inequality of Labor Income
4. Instruments of Redistribution
References
Contents in Detail
Index
Note to the Reader
This book was written and first published in 1997. It was subsequently updated for
several new editions, most recently in 2014. It should be noted, however, that the
overall structure has not been changed since 1997 and that the work essentially reflects
the state of knowledge and data available at that time. As a consequence, this book does
not fully take into account the results of the past fifteen years of international research on
the historical dynamics of inequality. In particular, recent research has demonstrated
that there are important historical variations in the capital-income ratios and the capital
shares in national income, and not only in the concentration of capital ownership at the
individual level. That is, the macroeconomic or functional distribution of national
income and national wealth is substantially less stable than what I was taught in
graduate school and what I report in this book. The large historical variations in top
income shares also receive insufficient treatment in the present book, because the
corresponding research became fully available only recently. Readers interested in a
detailed account of that more recent research and the lessons that can be drawn from it
are advised to consult the World Top Incomes Database (available online) and my book
Capital in the Twenty-First Century (Belknap Press, 2014).
Introduction
What orders of magnitude can we associate with contemporary inequality? Is the income
of the rich in a given country twice that of the poor? Ten times as great? Or a hundred
times? How does the income gap in one country or period compare with that in other
places at other times? Was the income gap in 1950 the same as in 1900 or 1800? Has
inequality with respect to unemployment become the major form of inequality in the
Western world in the 1990s?
Different Types of Income
What are the various sources of household income? Table 1.1 breaks down the income
of 24 million French households in 2000 into various categories: wages, self-
employment income (earned, for example, by farmers, merchants, doctors, lawyers, and
so on), pensions, other transfer income (family allowances, unemployment insurance,
welfare), and capital income (dividends, interest, rent, etc.).
What do we learn from Table 1.1? First, 58.8 percent of total household income
comes in the form of wages. If we add to this the 5.8 percent of income consisting of
self-employment compensation, we find that nearly two-thirds of total household income
is compensation for labor. Social income accounts for another 30 percent of the total,
and for more than two-thirds of retiree income. Finally, income from household wealth
(capital income such as dividends, interest, and so on) accounts for roughly 5 percent of
the total. As is well known, however, capital income is not accurately reported in
household income surveys. National accounts based on dividend and interest data
provided by firms and banks yield a higher estimate of the share of capital income in
total household income, on the order of 10 percent (INSEE, 1996b, pp. 26–29). In any
case, all sources agree that labor income accounts for at least six or seven times as
large a share of total household income as capital income. This is a general feature of
the income distribution in all Western countries (Atkinson et al., 1995, p. 101). But the
5 to 10 percent share of household income derived from capital underestimates the
share of capital income in total national income, since a substantial portion of the
capital income of firms is not distributed to households (see Chapter 2).
The importance of these various types of income is obviously not the same for rich
and poor. To analyze this further, it is useful to distinguish between different deciles of
the income distribution: the first decile, denoted D1 in Table 1.1, includes the bottom 10
percent of the household income distribution. The second decile, D2, includes the next
10 percent, and so on, up to the top decile, D10, which represents the 10 percent of
households with the highest income. To refine this description, we also use the notion of
centiles: the first centile includes the bottom 1 percent of households, and so on up to
the hundredth centile. Each decile includes a subgroup of the population: some 2.4
million households per decile and 240,000 per centile in the case of France in the year
2000. One can calculate various characteristics for each decile or centile: average
income, for example. This should not be confused with the notion of upper income limits
for each group. To capture this statistic, we use the letter P: for example, P10 represents
the level of income below which we find 10 percent of all households; P90 is the upper
limit below which we find 90 percent of all households; and so on. In Table 1.1, P90–
95 represents the subset of all households with incomes between the top of the 90th
centile and the top of the 95th centile, that is, the first half of the tenth decile, whereas
P95–100 represents the second half of the tenth decile, which includes the five top-
earning centiles.
Table 1.1 shows that the households in D1 consist largely of modest retirees and
unemployed workers: the wages they receive account on average for 18 percent of their
income, while nearly 80 percent consists of social income. The share of wages in total
income increases with income level, while the share of retirees and unemployed
decreases, until we reach the top 5 percent (P95–100), where capital income and
nonwage compensation account for a substantial share of the total (I make no distinction
between “wages” and “salary” throughout: both refer to income from labor). Nonwage
compensation is intermediate in nature between labor income and capital income, since
it remunerates both the labor of the farmer, doctor, or merchant and the capital invested
in his or her business. Still, labor income continues to account for a very large share of
the total income of households at the top of the distribution: the top 5 percent take more
of their income in wages than in income from capital, no matter how the latter is
estimated. One has to go even higher in the income hierarchy to reach a level where
labor income no longer accounts for the largest share (Piketty, 2001).
Wage Inequality
How are wages, which represent the lion’s share of household income, distributed?
Table 1.2 describes wage inequality among full-time private-sector workers in France
in 2000 (a group of some 12.7 million individuals).
The bottom 10 percent of the wage distribution (D1) earned on average an income
roughly equal to the minimum wage, or about €890 per month (net of taxes) in 2000. The
median wage (denoted P50, by definition the wage level below which lies 50 percent of
the sample) was €1,400. This was higher than the average wage of the fifth decile
(€1,310), since the fifth decile consists of workers between P40 and P50. It was also
lower than the average wage overall, which was €1,700 in 2000, because the top half of
the wage distribution is always “longer-tailed” than the bottom half, so that very high
earners inevitably lift the average wage above the median. Furthermore, the best-paid
10 percent, who earn at least €2,720 per month, earn an average wage of €4,030, or
nearly twice as much as the next lower 10 percent (€2,340).
One practical indicator of total wage inequality is the P90/P10 ratio, that is, the ratio
of the lower limit of the tenth decile to the upper limit of the first decile. In the case of
France in 2000, the P90/P10 ratio was 2,720/900 or roughly 3.0: to belong to the top-
earning 10 percent, one had to make at least three times as much as the least well paid.
This indicator should not be confused with the D10/D1 ratio, that is, the ratio of the
average wage of the tenth decile to that of the first decile, which is by definition always
higher and which in France in 2000 was 4,030/890, or 4.5: the best-paid 10 percent in
France earned on average 4.5 times as much as the worst-paid 10 percent. Table 1.2
also allows us to calculate the total wages paid to the top 10 percent: since the average
wage of D10 was 2.37 times the average wage (4,030/1,700 = 2.37) and the number of
workers in D10 is by definition 10 percent of the total number of workers, it follows
that D10 received 23.7 percent of total wages.
TABLE 1.2
Wage inequality in France, 2000
Average 1,700
D1 890 900 P10
D2 1,100
D3 1,110
D4 1,210
D5 1,310 1,400 P50
D6 1,450
D7 1,620
D8 1,860
D9 2,340 2,720 P90
D10 4,030
Note: D1 represents the worst-paid 10 percent; D2 the next 10 percent, and so on. P10 is the limit dividing D1 and
D2, P50 the limit dividing D5 and D6, and P90 the limit dividing D9 and D10. In other words, the worst paid all
earned less than €900 a month, with an average income of €890, whereas the best-paid 10 percent all earned more
than €2,720, with an average income of €4,030. These figures represent monthly wages excluding bonuses net of
social charges (and CSG/CRDS) for full-time, private-sector workers.
Source: DADS (Annual declaration of social data), INSEE, 2002, p. 10.
Other indicators are also used in order to capture the overall inequality of the
distribution and not just the gap between the extreme deciles: for instance, the Gini
coefficient or the Theil and Atkinson indices (Morrisson, 1996, pp. 81–96).
Nevertheless, interdecile indicators (such as P90/P10, D10/D1, P80/P20, etc.) are by
far the simplest and most intuitive. The P90/P10 indicator has the merit of being
available in reliable numbers for many countries, hence it will be cited frequently in
this chapter.
For a more complete view of wage inequality, one would need to include figures for
public-sector wages in addition to private-sector wages. In France, the 4.1 million full-
time employees of the public sector earn slightly more on average than private-sector
workers, and public-sector wages are significantly less widely dispersed: for example,
the P90/P10 ratio for civil-service workers was 2.6 (INSEE, 1996d, p. 55).
International Comparisons
Is a P90/P10 ratio of 3:1 typical of wage inequality everywhere? Table 1.3 gives the
P90/P10 ratio for fourteen OECD countries in 1990.
The table shows that France, with a P90/P10 ratio of 3.1 in 1990, occupied a middle
position between Germany and the Nordic countries on the one hand and the English-
speaking countries on the other. In the former, the ratio was generally around 2.5,
dipping as low as 2 in Norway, 2.1 in Sweden, and 2.2 in Denmark, while in the latter it
was as high as 3.4 in the United Kingdom, 4.4 in Canada, and 4.5 in the United States.
For all the countries shown, the figures in Table 1.3 concern only full-time employees.
This is an important detail, because including part-time workers (of whom there were
some 3.1 million in France in 2000) systematically leads to larger P90/P10 ratios. For
example, the OECD figures including intermittent and part-time workers in the United
States in 1990 give a P90/P10 ratio of 5.5, but only 4.5 when these workers are left out
(Katz et al., 1995, fig. 1; Lefranc, 1997, table 1), as is the case with other countries
(OECD, 1993, p. 173). In short, P90/P10 ratios range from 2 to 4.5, which is
considerable variation for countries at very similar levels of economic development.
TABLE 1.3
Wage inequality in OECD countries in 1990, measured by the P90/P10 ratio
Norway 2.0
Sweden 2.1
Denmark 2.2
Netherlands 2.3
Belgium 2.3
Italy 2.4
Germany 2.5
Portugal 2.7
Japan 2.8
France 3.1
United Kingdom 3.4
Austria 3.5
Canada 4.4
United States 4.5
Note: For example, in Germany the best-paid 10 percent earn at least 2.5 times as much as the worst-paid 10
percent.
Sources: OECD, 1993, pp. 170–173; US data, Katz et al., 1995, fig. 1.
Income Inequality
How does this inequality between workers translate into inequality of household
income? The answer is not simple, because one has to add nonwage compensation of
the self-employed (some 3 million individuals in France in 2000), social transfers, and
capital income, and then individual wage earners, non–wage earners, and their children
have to be grouped together to form households. Table 1.4 presents the results for
France in 2000.
The average monthly household income in France was €2,280, but 10 percent of
households had less than €790 of monthly income, while 10 percent had more than
€4,090, for a P90/P10 ratio for total household income of 5.2 (compare with the
P90/P10 ratio for wages of 3.0). The top 5 percent of households, with monthly income
above €5,100, had an average monthly income of €7,270.
TABLE 1.4
Income inequality in France, 2000
Average 2,280
D1 540 790 P10
D2 930 1,070 P20
D3 1,190 1,330 P30
D4 1,480 1,610 P40
D5 1,760 1,920 P50
D6 2,080 2,240 P60
D7 2,430 2,630 P70
D8 2,880 3,150 P80
D9 3,570 4,090 P90
P90–95 4,520 5,100 P95
P95–100 7,270
Notes: See Tables 1.1 and 1.2. The poorest 10 percent of households have income of less than €790 per month,
averaging €540. The wealthiest 5 percent have income above €5,100 per month, with an average of €7,300. Monthly
income calculated by dividing annual income by twelve, including wages, self-employment income, pensions,
transfers, and capital income. Income is net of social charges (including CSG/CRDS) but not of other direct taxes
(income tax, housing tax).
Source: “2000 family budget survey,” INSEE (author’s calculations).
The fact that inequality of household income is greater than wage inequality is quite
common, and it was made worse in France in 2000 by the fact that many households
included unemployed workers. In general, however, the explanation of this difference
depends on factors other than unemployment.
First, nonwage income and especially capital income are much more unequally
distributed than wages. Typically, the share of capital income claimed by the wealthiest
10 percent of households is on the order of 50 percent of total capital income, as is the
share of total wealth owned by the wealthiest 10 percent, whereas the share of total
wages going to the highest-paid 10 percent falls between 20 and 30 percent (23.7
percent in France in 2000). The share of capital income in total income is low,
however, so the share of income going to the wealthiest 10 percent of households was
just 26 percent of total household income in France in 2000. These very large
disparities of wealth, much greater than wage and income inequalities, are also much
less well known. Inequalities of wealth cannot be explained solely by inequalities of
present and past income. Behavioral differences with respect to savings and
accumulation also play a part (accounting for nearly half of wealth inequality in 1992,
according to Lollivier and Verger [1996]). These specific difficulties of accounting for
wealth inequality explain why measures of inequality are often limited to inequalities of
wages and income.
But the main reason why income inequality is always significantly greater than wage
inequality is totally different: it comes from the fact that a majority of low-income
households are households living on small pensions, often consisting of one person,
whereas high-income households are generally couples, often with two earners and
children living at home. If one were to calculate the P90/P10 ratio not for household
income but for household income adjusted for household size in order to measure
inequality in standards of living rather than of income as such, one would find a ratio of
4.3–4.4 rather than 5.2, depending on how the adjustment was made (INSEE, 1996b, p.
16). If one is interested in disposable income, then the effect of taxes on income also
needs to be taken into account, which the figures in Table 1.4 do not do. This would
reduce the P90/P10 ratio by about 10 percent, since the income tax paid by a household
with income at the P90 level, about €4,090 per month, would average about 10 percent,
while households at the P10 level pay no income tax (INSEE, 1995, p. 19). (On the
redistributive effect of taxes and transfers, see Chapter 4.) We would thus arrive at a
ratio for disposable household income adjusted for household size of 3.5–4, slightly
greater than wage inequality.
International Comparisons
How does a P90/P10 ratio of 3.5–4 compare with what we see in other countries?
International comparisons are unfortunately much more difficult to do for household
incomes than for wages: it is often difficult to include exactly the same income
categories in all countries. Culminating an ambitious effort to compile comparable data
for many countries, the Luxembourg Income Study (LIS) was published in 1995 at the
behest of the OECD (Atkinson et al., 1995).
The P90/P10 ratios in Table 1.5 are for disposable income, that is, income
accounting for taxes and transfers, and adjusted for household size. France is assigned a
ratio of 3.5 (and not 5.2 as in Table 1.4) for that reason, along with the fact that the
OECD study used 1984 fiscal data rather than the “Family Budget Survey” of 2000. The
same international contrast observed earlier in relation to wage inequalities occurs
again here: Germany, Belgium, Sweden, and Norway, which had wage ratios of 2–2.5,
have income ratios of 2.7–3, while the United Kingdom, United States, and Canada,
which had wage ratios of 3.5–4.5, have income ratios of 3.8–5.9, with the United States
topping the chart at 5.9. France again occupies an intermediate position.
TABLE 1.5
Income inequality in OECD countries
Sweden 2.7
Belgium 2.8
Norway 2.9
Germany 3.0
France 3.5
United Kingdom 3.8
Italy 4.0
Canada 4.0
United States 5.9
Notes: In Sweden, to belong to the top 10 percent in income, one must earn at least 2.7 times as much as anyone in
the bottom 10 percent. The table shows the P90/P10 ratio for disposable income adjusted for household size
(Atkinson et al., 1995). Years: 1984 (Germany, France), 1985 (Australia), 1986 (US, Italy, Norway, UK), 1987
(Canada, Sweden), 1988 (Belgium).
Source: LIS, Atkinson, Rainwater, and Smeeding, 1995, p. 40.
It is extremely difficult to compare these figures with the few available inequality
indicators from outside the developed world. Conditions vary widely: the South
American countries stand out for having even higher levels of inequality than the most
inegalitarian Western countries, whereas most Asian countries, as well as the less-
developed countries of Africa, generally have levels of income inequality equal to or
less than those found in the least inegalitarian Western countries (Morrisson, 1996, pp.
145–172). It is also difficult to compare levels of inequality found in the Communist
bloc, because income often includes compensation in kind in one form or another and is
difficult to quantify in monetary terms. The available indicators seem to show real
income gaps quite comparable to the average in capitalist countries and generally
greater than in the more egalitarian capitalist countries (Morrisson, 1996, p. 140).
Inequalities in Time and Space
Are P90/P10 wage and income ratios of 3:1 or 4:1 between people living in one
country at one point in time negligible compared with the gap between a citizen of an
advanced country in 1990 and of the same country in 1900 or a citizen of India in 1990?
Table 1.6 indicates the average purchasing power of a French blue-collar worker and of
an executive from 1870 to 1994 measured in 1994 francs (that is, accounting for changes
in the cost of living).
These figures should obviously be viewed cautiously: the further back we go in time,
the more problematic the idea of a synthetic cost index becomes, because modes of
consumption change so much. Still, orders of magnitude may be taken as significant:
between 1870 and 1994, the purchasing power of a worker increased roughly eightfold.
This spectacular increase in standard of living during the last century of capitalism was
more or less comparable in all the Western countries. For example, the hourly wage of
an American worker increased by a factor of eleven between 1870 and 1990, for an
average rate of increase of about 2 percent per year (Duménil and Lévy, 1996, chap.
15), which is approximately the same as in France if we take the decrease in annual
hours of work into account.
This 10:1 ratio between 1990 and 1870 is approximately equivalent to, or slightly
less than, the ratio of the average income of a Western citizen in 1990 to that of a
Chinese or Indian citizen, using the best available estimates of purchasing power parity
(Drèze and Sen, 1995, p. 213). The gaps in GDP per capita, which are often four to five
times greater, don’t actually make much sense, because they are expressed in terms of
official exchange rates with the advanced economies, and these rates are a very poor
gauge of actual differences in purchasing power. A 10:1 gap between the average
standard of living in the wealthiest countries and that in the poorest countries probably
comes closer to reality.
To sum up, inequality between the top 10 percent and bottom 10 percent in a given
country, as measured by the P90/P10 ratio, is on the order of 3–4, and this is two to
three times smaller than the gap in standard of living between the end of the nineteenth
century and the end of the twentieth century and than the gap between the richest and
poorest countries. These two forms of inequality are therefore not incomparable, even if
one is undeniably larger than the other.
The Historical Evolution of Inequality
Are these 3:1 ratios between top and bottom income groups in one country and 10:1
ratios between rich and poor countries destined to remain steady, increase, or decrease?
Although Marx and other nineteenth-century social theorists did not quantify
inequality this way, they were certain of the answer: the logic of the capitalist system
was to increase inequality between social classes—between capitalists and
proletarians—constantly. The gap between rich and poor countries would also grow.
These predictions were soon challenged, however, even within the socialist camp. In
the 1890s Eduard Bernstein insisted that Marx’s proletarianization thesis did not hold
because the social structure was clearly becoming more diverse and wealth was
spreading to ever broader segments of society.
It was not until after World War II, however, that it became possible to measure the
decrease in wage and income inequality in the Western countries. New predictions
were soon forthcoming. The most celebrated was that of Simon Kuznets (1955):
according to Kuznets, inequality would everywhere be described by an inverted U
curve. In the first phase of development, inequality would increase as traditional
agricultural societies industrialized and urbanized. This would be followed by a second
phase of stabilization, and then a third phase in which inequality would substantially
decrease. This pattern—of growing inequality in the nineteenth century followed by
declining inequality after that, has been well studied in the case of the United Kingdom
(Williamson, 1985) and the United States (Williamson and Lindert, 1980). In the latter,
for example, one finds that the share of total wealth owned by the wealthiest 10 percent
rose from 50 percent in 1770 to a maximum of 70–80 percent in the late nineteenth
century, before declining to about 50 percent, a level typical of contemporary wealth
inequality. The available sources suggest that the pattern was similar in all the Western
countries.
FIGURE 1.1. The fall of rentiers and the stability of the wage hierarchy in France, 1913–2005. Sources: Piketty, 2001;
Landais, 2007.
The most recent research on France and the United States (Piketty, 2001; Piketty and
Saez, 2003; Landais, 2007) shows, however, that the sharp decrease in inequality
observed over the course of the twentieth century is in no sense the consequence of a
“natural” economic process. Only inequality of wealth decreased (the wage hierarchy
showing no tendency toward compression over the long run), and this decrease was due
mainly to shocks incurred by wealth owners in the period 1914–1945 (wars, inflation,
the Great Depression). The concentration of wealth and capital income did not return to
the astronomical level achieved on the eve of World War I, however. The most likely
explanation involves the fiscal revolution of the twentieth century. The impact of the
progressive income tax (created in France in 1914) and the progressive estate tax
(created in 1901) on the accumulation and reconstitution of large fortunes seems to have
prevented a return to nineteenth-century rentier society. If contemporary societies have
become societies of managers—that is, societies in which the top of the income
distribution is dominated by the “working rich” (people who live mainly on their labor
income rather than on income derived from capital accumulated in the past), it is
primarily a consequence of particular historical circumstances and institutions. The
Kuznets curve is thus not the “end of history” but the product of a specific—and
reversible—historical process.
Note: To belong to the best-paid 10 percent in the United States, one had to earn at least 3.2 times as much as
anyone in the worst-paid 10 percent in 1970, compared with 4.5 times as much in 1990.
Sources: Germany, Italy, Japan, Sweden: OECD, 1993, pp. 170–173. France: INSEE, 1996a, p. 48. United States,
United Kingdom: Katz et al., 1995, fig. 1.
Capital-Labor Inequality
Since the industrial revolution, and in particular since the work of Karl Marx (1818–
1883), the question of social inequality and redistribution has been posed primarily in
terms of the opposition between capital and labor, profits and wages, employers and
employees. Inequality is thus described as a contrast between those who own capital,
that is, the means of production, and those who do not and must therefore make do with
what they can earn from their labor. The fundamental source of inequality is thus said to
be the unequal ownership of capital. Initially, the two terms of this basic inequality,
capitalists and workers, are conceived as homogeneous groups, and inequality of
income from labor is regarded as a secondary matter. This view of inequality as purely
a question of capital versus labor has exerted, and will continue to exert, a profound
influence on the way redistribution is conceptualized, even in countries that have not
gone so far as to abolish private ownership of capital.
The special attention accorded to labor-capital inequality should come as no
surprise. Indeed, the mere fact that a share of productive income goes to capital might
seem to contradict basic principles of justice and immediately raise the question of
redistribution: Why should a person who inherits ownership of capital receive income
denied to those who inherit only their labor power? Absent any market inefficiency, this
would amply justify a pure redistribution of capital income to labor income (using the
distinction between pure and efficient redistribution discussed in the Introduction). How
large should this pure redistribution be, and with what tools might it be accomplished?
What does the history of this kind of redistribution and of the capital-labor split tell us?
Social justice is not the only reason to worry about the capital-labor split, however.
Is the unequal distribution of wealth among individuals and countries not only unjust but
also inefficient, because it reproduces itself by limiting the ability of the poor to invest
and thus close the gap between themselves and the rich? If so, how can capital be
efficiently redistributed?
The Share of Capital in Total Income
The question seems simple. A nation produces what it produces using a certain quantity
of capital (machinery, infrastructure, etc.) and a certain quantity of labor (hours
worked). What determines the share of output going, respectively, to capital (in the form
of dividends and interest paid to the owners of capital) and labor (in the form of wages
paid to workers), and what can government do to redistribute these shares? This
question, and especially the role of the price system in determining the capital-labor
split, has given rise to unusually acrimonious intellectual and political controversy,
especially among economists.
The firm’s value added is defined as the difference between the revenue from sales and the cost of
intermediate consumption. The value added is thus equal to the sum of labor income and capital income.
When we calculate the profit and wage shares, we calculate the share of capital income and labor income as
percentages of value added. In other words, we deduct the cost of intermediate consumption. This is
perfectly legitimate, since the cost of intermediate goods purchased from other firms goes to remunerate
capital and labor in those firms, and we need to avoid double counting.
In addition to taxes levied directly on capital (such as tax on profits) and labor (such as social charges),
which are included in capital and labor income, firms also pay indirect taxes such as the value-added tax, the
amount of which does not depend directly on how value added is divided between capital and labor. Hence
these taxes cannot be attributed to either gross capital income or gross labor income. In calculating the profit
and wage shares, it is customary to omit these indirect taxes: in other words, we calculate these respective
shares as a percentage of value added net of indirect taxes. Thus the sum of capital’s share and labor’s share
equals 100 percent of net value added, as in Tables 2.1 and 2.2. This makes the results easier to interpret,
since indirect taxes do not depend directly on the capital/labor split.
Finally, another complication has to do with the treatment of self-employment income (farmers, merchants,
doctors, lawyers, and so on). Here, value added goes to pay both the labor of self-employed workers and the
capital they have invested, even though their accounts do not clearly distinguish between wages and profits.
Without correcting for self-employment income, one would find that the wage share of total value added has
increased significantly since the nineteenth century, simply because the percentage of non-self-employed
workers has increased so much (Morrisson, 1996, p. 78). The accounting convention used by the OECD is to
attribute to self-employed workers the same average wage as non-self-employed workers. This convention
has been applied to adjust the figures in Tables 2.1 and 2.2.
Allowance must also be made for taxes paid by firms on their profits before
distribution to shareholders. This factor is of limited importance, however, because
although the tax on profits in most Western countries is on the order of 40–50 percent,
receipts from this tax generally do not exceed 2.5–3 percent of GDP. In France in the
1990s, the figure was as low as 1.5 percent of GDP, even though the share of capital in
value added was higher than elsewhere (OECD, 1995, p. 78). This is because the notion
of taxable profits is much narrower than the notion of gross operating surplus, since
firms are allowed to deduct not just the estimated depreciation of their capital stock but
also the interest paid to creditors, provisions for anticipated risks, and so on. The tax on
profits is by far the most riddled with loopholes in the whole tax system.
Note, finally, that a significant portion of the sums listed under “wages” in Table 1.1
are in fact paid by government entities out of revenue from the tax on profits or on total
value added (in the form of a value-added tax, or VAT). This tends to increase the share
of wages relative to capital income in total household income compared with the
capital-labor split in the value added by firms. All these factors explain why, given
gross profits on the order of 32–34 percent of value added by firms, the share of capital
income actually received by households is typically only 10 percent of total household
income (see Chapter 1).
What the Constancy of the Profit Share Tells Us
Consider once again Table 2.1. How should we interpret the constancy of the profit
share across time and space? Leaving the matter of retained profits aside, the first
lesson to be drawn from this regularity is that the considerable growth of workers’
purchasing power in the twentieth century cannot be explained by changes in the capital-
labor split. In other words, the reasons why the purchasing power of the French worker
increased fourfold between 1920 and 1990 (see Table 1.6) was not that social conflict
reduced the share of income appropriated by capitalists. Indeed, the share of wages in
total value added was broadly speaking the same two-thirds of national income in 1920
as it was in 1990 (see Table 2.1). Furthermore, although the two world wars and
changes in nomenclature make it difficult to extend statistical series on the profit/wage
split beyond 1920 for France, we can use American statistics to go back all the way to
1869, and these indicate that the variation in the wage share was already limited to
between 66 and 68 percent in the nineteenth century (Duménil and Lévy, 1996, chap.
15). In other words, the profit-wage split has remained almost constant for more than
120 years, even though wages increased tenfold.
To be sure, it is important that capital appropriated a third of value added throughout
this period. If that income had all been distributed to labor, including the portion
devoted to capital depreciation, wages could have been increased by 50 percent, which
would have significantly improved the living conditions of workers in 1870 or even
1990, conditions that were in many cases miserable when compared with the opulence
in which many capitalists lived. At the same time, however, we must admit that this 50
percent increase in wages would have been only half as much as the wage increase that
actually occurred between 1870 and 1910 and less than a quarter as much as the
increase that actually occurred between 1950 and 1994 (Table 1.6). It is difficult,
moreover, to believe that the wage increases of 100 percent between 1870 and 1910
and more than 200 percent between 1950 and 1990 would have occurred if the capital
share had been reduced to zero in 1870 or 1950. Although our knowledge on this point
is limited, it is likely that the supply of capital would have decreased if there had been
that much redistribution, hence that the optimal capital/labor redistribution from the
standpoint of workers would have been much smaller, though surely larger than what
was actually instituted.
Who Pays Social Charges (Payroll Taxes)?
The second lesson to draw from Table 2.1 has to do with the question of fiscal
incidence. In the 1920s and 1930s firms paid little in the way of social charges,
whereas in the 1990s 45 percent of the wage bill consisted of social charges paid by
employers, while workers paid another 20 percent of their gross wages to support the
welfare state (see box). Who actually bore the burden of employer-paid social charges?
Certainly not the employers, since labor’s share of value added, which includes
employer-paid social charges, did not increase between 1920 and 1995. Similarly,
employer-paid social charges were much lower in the United States and United
Kingdom than in France in the 1990s, but labor’s share of value added was no higher in
France than in those two countries—indeed, the opposite was true (see Table 2.1). In
1996, the maximum rate of employer-paid social charges as a percentage of gross
wages was 7.65 percent in the United States and 10.2 percent in the United Kingdom,
with workers paying the same percentage, and total receipts from social charges (paid
by both employers and employees) represented 6–7 percent of GDP, compared with
nearly 20 percent of GDP in France (OECD, 1995, p. 79). If employer-paid charges
were actually paid by employers, we would therefore expect to find that labor’s share
of value added in France was at least 10 percent of GDP higher than in the United States
and United Kingdom.
It is therefore clear that social charges are not paid out of capital income. This is a
crucial fact, since it implies that modern systems of social protection, which are central
to today’s systems of redistribution and which were based on the idea of dividing social
costs between capitalists and workers, have not in fact redistributed income from
capital to labor; labor income has absorbed the full cost. This in no way undermines the
legitimacy of such systems, which do allow for considerable redistribution of labor
income and fulfill an insurance function that private markets are often incapable of
assuming (see Chapter 4). It does, however, profoundly challenge the implicit intention
to redistribute income from capital to labor that in many cases informed the creation of
the welfare state. This vision was closely linked to the classical theory of the capital-
labor split, according to which a better division could be achieved through negotiation,
for example, by setting a higher rate for employer payments than for employee
payments: social benefits were thought of as a supplement to the wage that capitalists
would otherwise have paid to workers.
In fact, the evidence appears to indicate that, as the theory of fiscal incidence would
predict, the only thing one needs to know is how the tax is assessed, that is, how its
amount depends on the level of wages, profits, and so on, and not what the name of the
tax is or who is officially supposed to pay it, that is, who writes the check to the
appropriate collection authority. It does not much matter whether social protection is
paid for by an income tax proportionate to wages or by social charges levied on both
employers and employees. In Denmark, there is no payroll tax, and the generous Danish
system of social protection is financed entirely by income tax (which in practice is
always mainly a tax on wages and social income in view of the limited importance of
capital income). Unsurprisingly, the share of labor income in value added is the same in
Denmark as elsewhere (OECD,1996, p. A27). Danish firms spend as much as French
firms on wages but pay it all to their workers instead of paying social charges; the
workers then pay tax on their income. More generally, the share of social charges in the
financing of European social protection systems varies widely from country to country,
with France and Denmark representing the two extremes, but the share of labor in value
added is more or less the same everywhere. The only relevant parameters are how the
tax rate, whether assessed in the form of an income tax or a social contribution, depends
on the wage level (in other words, how progressive is the assessment) and whether it
also depends on the level of capital income. In particular, the only way to redistribute
income from capital to labor is to tax capital.
A Cobb-Douglas Production Function?
With these lessons in mind, how can we explain the constancy of the share of profits?
The interpretation traditionally favored by economists is that over the past century of
capitalism, the Western economies have been reasonably well described at the
macroeconomic level by a production function of the Cobb-Douglas type, which is to
say, with an elasticity of capital/labor substitution equal to 1. Indeed, only if this
elasticity is 1 can one confidently predict that the respective shares of profits and wages
should remain constant over time regardless of changes in the available quantities of
capital and labor and regardless of political and economic shocks affecting the price of
each factor. An elasticity of 1 would also explain the observed fiscal incidence of
social contributions, whose burden falls on labor and therefore increases its price.
Even with a fixed-coefficient technology, one could of course argue that social and
political conflict always leads to the same result in all countries, with one-third of
income going to capital and two-thirds to labor. As Solow himself has noted, we would
need to know how much variation it would be natural to expect before describing the
constancy of the capital-labor split as surprising (Solow, 1958). Econometricians have
looked at how hiring by individual firms varies in response to variations in the price of
labor, however, and these microeconomic studies show considerable elasticity of
substitution between capital and labor. After comparing the results of several dozen
studies across all Western countries, Hamermesh (1986; 1993) found that most
estimates of the demand elasticity for labor suggest elasticities of substitution between
capital and labor of 0.7 to 1.1, leading to the conclusion that “the Cobb-Douglas
function seems to be a fairly good approximation to reality” (1986, pp. 451–452, 467).
Contrasting experience in various Western countries since the 1970s also suggests
substantial capital/labor substitutability (I will return to this point later). The data thus
seem to confirm the marginalist theory of the capital-labor split and therefore the
superiority of fiscal redistribution over direct redistribution.
Historical Time versus Political Time?
One should be careful not to underestimate the limits of this historical regularity,
however. Although the profit share of income is impressively constant over the long run,
it does vary over the short run, and the long run can seem very long indeed to the
individuals affected. Consider, for instance, the evolution of respective shares of profits
and wages in the OECD countries between 1979 and 1995.
Tables 2.1 and 2.2 show large variations in the profit and wage shares. Although the
wage share tended to increase in the 1970s, it was the profit share that increased in the
1980s and 1990s, in some cases substantially. These variations were widest in France,
where the wage share was 66.4 percent in 1970, rising to 71.8 percent in 1981, then
falling after 1982 to 62.4 percent in 1990 and 60.3 percent in 1995. How can we
explain the fact that more than 5 percent of national income was redistributed from
capital to labor between 1970 and 1982, while 10 percent went from labor to capital
between 1983 and 1995?
It so happens that the first period coincides with a period of substantial wage
increases inaugurated by the Grenelle Accords of 1968. Wages continued to improve in
the 1970s owing to the influence of social movements and substantial increases in the
minimum wage. The last major increment to the minimum wage occurred in 1981. After
1983, wages ceased to be indexed to prices, and increases to the minimum wage were
minimal. In fact, the purchasing power of the average net wage increased 53 percent
between 1968 and 1983 but only 8 percent between 1983 and 1995 (INSEE, 1996a, p.
48). True, GDP increased 44 percent between 1970 and 1983 but only 28 percent
between 1983 and 1995 (INSEE, 1996c, p. 34), and this growth had to finance a
growing burden of pension and health care expenditures, but the reduction of wage
growth relative to national income growth was nevertheless quite real. In other words,
over a period of twenty-five years, the predictions of the classical theory of the capital-
labor split seemed to be accurate: the profit share decreased when social militancy
enabled workers to win important concessions on wages, and it increased when
constraints were imposed on wages. Yet the increased profit share did not lead to the
promised creation of new jobs.
To be sure, these substantial variations over a twenty-five-year period do not change
the fact that over periods of fifty or a hundred years, wages have always accounted for
roughly two-thirds of added value, so that the increase in workers’ purchasing power of
250 percent since 1950 and 700 percent since 1870 cannot be explained by changes in
the capital-labor split. But why would that matter to the workers who lived through the
twenty-five years from 1970 to 1995? Their standard of living increased sharply from
1968 to 1982 and then stagnated from 1983 to 1995 while output continued to grow.
How could they not associate the improvement in their standard of living with
redistribution from capital to labor? The right-wing view that true improvement in the
standard of living can come only from growth and not from redistribution is valid only
in the long run, and politics, in which workers have a legitimate interest, operates on a
different time scale.
Furthermore, how could workers fail to associate redistribution from capital to labor
with social struggles and wage increases and thus with direct rather than fiscal
redistribution? Indeed, no fiscal redistribution had ever redistributed 10 percent of
national income in so short a period. To give an order of magnitude, the fiscal
redistribution measures adopted by the Socialist government after it came to power in
France in 1981, which were denounced at the time by the right as the height of “fiscal
bludgeoning” and which consisted essentially of a wealth tax and a surtax on top income
brackets, brought in less than 10 billion francs in 1981 (Nizet, 1990, pp. 402, 433), or
0.3 percent of national income. In theory, a government can achieve any level of
redistribution it wants via taxes and transfers, but in practice no transfer of comparable
magnitude has been accomplished in so few years. Inevitably, therefore, workers think
of and experience redistribution primarily in terms of social struggle and wage
increases rather than fiscal reform and transfer payments. It is not so much rejection of
the logic of fiscal redistribution and the price system that sustains left-wing skepticism
of redistribution through taxation; it is rather this historical reality. We will again
encounter this same historical reality and contrast between historical and political time
in the discussion of inequality of labor income in Chapter 3.
This level of variation in the capital-labor split over a period of ten to fifteen years
is not historically unique, moreover, even if certain specific features of French social
and political history helped make the period 1970–1990 particularly noteworthy. For
instance, the share of wages in the value added by American firms decreased from about
65 to 55 percent between 1869 and 1880, then rose again to 65 percent in 1885 and to
66–68 percent in 1890. The average wage rose by only 2 percent between 1869 and
1880 but then increased 27 percent between 1880 and 1885, a period marked by major
strikes and an unusually active trade-union movement (Duménil and Lévy, 1996, chap.
16). Over a period of ten to fifteen years, the marginalist understanding of the capital-
labor split can frequently seem rather absurd in the light of social realities. The same
can be said about the question of fiscal incidence: in the short term, social charges
assessed on employers are indeed paid by the employers and are not immediately
nullified by wage reductions, a fact that inevitably shapes the way many people see
fiscal incidence, even though it is true that over the long run social charges always end
up being paid by labor.
Why Has the Profit Share Not Increased in the United States and United Kingdom?
When it comes to the history of the capital-labor split in the period 1970–2000, things
are not so simple, however. The French pattern was reproduced in Italy, where the
profit share rose from 34.5 percent in 1983 to 42.5 percent in 1995, and to a slightly
lesser extent in Germany, where it rose from 28.2 percent in 1981 to 36 percent in 1995.
Strikingly, however, the United States and United Kingdom seem to have entirely
avoided this increase in the profit share in the 1980s and 1990s: the wage share of US
value added held steady at 66–67 percent throughout this period, while in Britain it
remained between 68 and 71 percent (Table 2.2). It is difficult to compare profit shares
at different points in time between different countries because of numerous differences
in accounting conventions, but comparing trends over time leaves no doubt: the capital
share increased by nearly 10 percent of value added in France, Italy, and Germany,
whereas it did not increase at all in the United States and United Kingdom. In contrast to
what happened with wage inequality, which increased sharply in the United States and
United Kingdom after 1970 (see Chapter 1), the countries where ultraliberalism won out
in the 1980s and 1990s are the only countries in which the profit share did not increase.
How do we explain this?
Part of the explanation is surely that there was catching-up to do: in France, the
profit share had decreased by 5–6 percent of value added in the 1970s owing to very
rapid wage increases. In the United Kingdom, the profit share decrease was much more
moderate, and in the United States it did not occur at all (Table 2.1). This cannot be the
whole story, however: by 1985–1986, the profit share in France had returned to its
1970 level, yet it continued to increase, while the profit share in the United States and
United Kingdom remained stable.
It is hard to avoid relating this to the fact that the United States and United Kingdom
were the only two countries to have created jobs in this period, thus increasing their
total wage bill, while wages stagnated in the other countries. Between 1983 and 1996,
the United States created 25 million jobs, an increase of about 25 percent (from 100.8
million to 126.4 million), while total employment in France increased by barely 2
percent (from 21.9 million to 22.3 million), and GDP in both the United States and
France increased by about 30 percent (OECD, 1996, p. A23). This is surely the best
proof that capital and labor can be combined in different proportions to increase
production by the same amount, hence that there is considerable opportunity for
substitution at the macroeconomic level. Between 1983 and 1996, the French economy
grew by employing skilled labor together with new machinery and infrastructure, while
American growth relied on intensive use of labor, especially relatively unskilled labor
in the service sector (restaurants, commerce. etc.) (Piketty, 1997b). This interpretation
is confirmed, moreover, by available data concerning the growth of the capital stock in
the period 1970–2000, which show that it grew much more rapidly in France and most
other European countries than in the United States (IMF, 1996). The data also show the
degree to which capital/labor substitution can be related to major intersectoral
reallocations (from industry to services, for example), and not just substitution of
workers for machines at the firm or sector level.
The simplest explanation might be that labor was not substituted for capital in
France, and jobs were not created, because wages were too high owing to very rapid
wage growth in the period 1968–1983. This would suggest that the long run, in which
the effects of marginalist theory would manifest themselves, is not as far off as the
individuals involved might have hoped. In order for lower wages to lead to more jobs,
however, the job creation effect would have to outweigh the wage effect, that is, the
elasticity of substitution between labor and capital would have to be greater than 1,
which, as mentioned, is higher than the usual estimates. Furthermore, if the average
wage in the United States rose by barely 5 percent between 1983 and 1996, it increased
by nearly 20 percent in the United Kingdom, compared with less than 12 percent in
France, yet total employment in Britain increased by nearly 10 percent in the same
period (OECD, 1996, A15, A19, A23). In the period 1983–1996, France thus appears
to have lost across the board, since wages and jobs stagnated, leading to an
exceptionally large decrease in labor’s share of value added.
In addition to the average cost of labor, two other factors might explain why the
wage share decreased in France and continental Europe while holding steady in the
United States and United Kingdom. One explanation might be that the variation of wages
with skill level increased in the United States and United Kingdom, and that this was the
sole explanation of employment growth in the 1980s and 1990s (see Chapter 3). A
second possible interpretation is that labor income includes a nonmonetary component
in the form of stability and guaranteed employment, and this component decreased in the
United States and United Kingdom but remained high in France and elsewhere in Europe
(see Cohen et al., 1996 for a comparison of France and the United States). We would
then need to explain why the price of this employment guarantee should have increased
between 1970 and 1995, and compare this to the incontestably high value placed on it
by workers.
The Dynamics of the Distribution of Capital
Why does capital/labor inequality command so much attention? Not just because capital
claims a significant share of national income. Even more striking is the fact that
capital/labor inequality often reproduces itself or even grows over time. It is this
reproduction that makes capital/labor inequality seem arbitrary, useless, and
incompatible not only with common sense and social justice but also with economic
efficiency: Why should capital-poor individuals or countries be denied the possibility
to invest to the extent their talents permit? In other words, capital/labor inequality
immediately raises the question of efficient (as opposed to pure) redistribution. The
time has come, therefore, to move from the macroeconomic division of total income
between the two factors of production, labor and capital, to the study of the income
distribution at the individual level. What are the dynamics underlying changes in the
income of individual workers and capitalists, and what investment opportunities are
open to them? Does the logic of the market economy lead to an inefficient reproduction
of inequality in the distribution of capital over time? If so, how can it be prevented?
* On this apparent regularity, please see the Note to the Reader at the opening of this book; recent research calls it
into question.
{THREE}
Although it is often thought that capital income is very unequally distributed while labor
income is not, the fact is that the lion’s share of income inequality today (and probably
for a long time in the past) is due to labor income inequality. For instance, it is the
increase in labor income inequality that is responsible for the reversal of the Kuznets
curve that has taken place since the 1970s: in the United States, the gap between the top
and bottom 10 percent of the income distribution has increased by nearly 50 percent. In
order to understand inequality as it presently exists and redistribution as it might exist in
the future, we must therefore give up the idea that labor incomes are relatively equal and
that inequality exists primarily between capital and labor. Instead, we need to analyze
the reasons for labor income inequality. The point of such analysis is to determine what
kinds of redistributive instruments might combat it. The goal is no longer to abolish
private ownership of capital, tax profits, or redistribute wealth. The instruments suitable
for dealing with labor income inequality go by other names: taxation of top incomes and
fiscal transfers to those with lower incomes; policies to improve education and training;
minimum wages; and measures to prevent employment discrimination, strengthen unions,
and establish wage schedules, to name a few. Which of these instruments are the most
justifiable? What arguments are invoked to justify (or reject) them, and how are we to
evaluate such arguments?
Inequality of Wages and Human Capital
The simplest theory of wage inequality is that different workers contribute different
amounts to a firm’s output. The computer specialist who develops a program for
analyzing a firm’s customer records quickly and efficiently is worth more to the firm
than the office worker who processes a certain number of files each day, and that is why
the firm pays the computer specialist more, lest competitors hire her away. Call this the
theory of human capital. It has long met with hostility. Why? No doubt because a theory
that declares that a computer specialist is worth more than an office worker because she
brings more human capital and therefore greater productivity to the firm is often
suspected of suggesting that this inequality of human capital actually measures an
irremediable inequality between two human beings and may even be used to justify a
considerable difference in standards of living. These suspicions are not totally
illegitimate, moreover, since it was Gary Becker and his colleagues at the University of
Chicago, known for their ultrafree market position, who developed and popularized the
theory (Becker, 1964). Of course, economists do not limit themselves to explaining
wage inequality by individual productivity. They also try to explain the origins of human
capital inequality and propose a theory of education and training that leads to rejection
of any ambitious form of public intervention.
It is nevertheless useful to examine these various questions separately, in order to
distinguish, as discussed in the Introduction, between pure redistribution (in the form of
transfers from high earners to low earners) and efficient redistribution (in the form of
interventions in the process of human capital formation). We begin, therefore, by taking
inequality of human capital as a given. Can observed wage inequality be explained
solely in terms of differences of productivity? What does observed wage inequality tell
us about the most efficient way of remedying the unequal standards of living that result
from unequal wages? We will then turn to the question of human capital formation.
Where does human capital inequality come from, and what instruments would enable us
to change it in an efficient way?
Instruments of Redistribution
The two previous chapters have tried to show how important it is to understand the
socioeconomic causes of inequality in order to know what instruments of redistribution
are most appropriate. In this chapter I will continue to analyze the most important of
these tools in the light of contemporary experience. As I have done throughout, I will
continue to distinguish between pure and efficient redistribution.
Pure Redistribution
The primary tool for pure redistribution is fiscal redistribution, which makes it possible
to correct inequality due to unequal initial endowments and market forces while
preserving as much as possible of the allocative role of the price system. I will focus
here on fiscal redistribution of labor income. In Chapter 2, I analyzed some of the
specific problems associated with the redistribution of capital income, which is less
important than labor income.
This conclusion also holds for households with children: family allowances of
course account for a higher percentage of supplementary income for low earners, but the
family quotient system (which reduces the tax on large families) results in a larger
decrease in income tax (as percentage of income) for high earners, so that the curve of
average effective tax rates remains fairly flat. To be sure, as the “infinite income” bars
in Figure 4.1 indicate, the effective average rate can go as high as 70 percent for very
high earners who surpass all the ceilings for deductions and family quotients
(applicable to those earning more than about ₣700,000 per year), who were liable to a
top marginal income tax rate of 56.8 percent in 1996. But this affects very few
households, so the actual importance of this top rate is much smaller than the symbolic
importance often attached to it in political debate. In particular, it hardly alters the basic
reality portrayed in Figure 4.1, namely, the absence of any substantial redistribution
between workers. Including other proportional taxes such as the value-added tax, whose
weight is twice that of the income tax, would further reinforce this conclusion.
These observations for France sum up the main feature of the contemporary fiscal
redistribution in all Western countries, in spite of national institutional differences:
there is no significant monetary redistribution between active workers. Taxation of
active workers is broadly proportional and transfers are minimal, so that the variation
in disposable income between active workers is quite close to the variation in wages
paid by employers. As noted in Chapter 1, the countries in which income inequality is
small are the countries in which wage inequality is low and vice versa. It is not the case
that fiscal redistribution between workers reduces initially high wage inequalities. In
addition to traditional expenditures (on justice, defense, highways, and so on), this
broadly proportional tax essentially serves to pay for unemployment insurance,
educational expenses, and above all pensions and health care. In some cases, these
expenditures are advantageous to workers at the low end of the wage scale, but this is
not always the case. I will say more about this later.
The U-Shaped Curve of Marginal Rates
Effective average rates of tax and transfer can measure the degree of redistribution that
actually occurs, but to measure the impact of this redistribution on individual behavior,
effective marginal rates are more useful. Figure 4.1 shows that effective marginal rates
are higher at both the low end and the high end of the income distribution than in the
middle, creating a rather marked U-shaped curve. It is not surprising that marginal rates
are higher at the top end: high earners are in the top income-tax brackets. Marginal rates
are also higher at the low end because a person who goes from zero wage income to
some wage income must not only pay taxes on his pay but also lose certain social
transfer payments available only to those who have no income from work. Consider, for
example, an unemployed person in France who receives €530 per month in guaranteed
basic income and housing allotment but who then finds an employer prepared to pay
€1,370 a month for his labor, presumably because his contribution to the production
process brings in at least that amount. The worker will actually receive just over €760
in net income each month after deducting all social charges. In other words, his gross
income goes from 0 to 1,370, but his disposable income goes from 530 to 760. The
worker’s additional income is just €230 per month, or less than 20 percent, which
yields an effective marginal tax rate of more than 80 percent, as shown in Figure 4.1 for
the first decile of the wage distribution. If the worker has children or was entitled to
unemployment benefits before finding work, the calculation would be slightly different,
but the effective marginal rate would still be on the order of 80–90 percent, and in some
cases more than 100 percent (for details, see Piketty, 1997a).
As it turns out, low-wage workers bear the highest effective marginal rates: a single
worker moving from the ninth to the tenth decile of the wage distribution faces a
maximum marginal rate on the order of 60 percent and 70 percent for an infinite income
in the top income tax bracket (assuming no special deductions), compared with a rate on
the order of 80–90 percent for a person moving from unemployment into the first decile
of the wage distribution. This U-shaped curve of effective marginal rates, with the
highest peaks at the bottom end of the curve, is the second major characteristic of
contemporary fiscal redistribution. Once again, it is a characteristic shared by all
Western countries: reserving social transfers for those who have no income from labor
while excluding low-wage workers is, in appearance at any rate, the least costly way to
fight poverty. It was this thinking that prevailed when national welfare systems were put
in place.
Adelman, I. and S. Robinson. 1989. “Income distribution and development.”Handbook of Development Economics,
vol. 2. New York: North-Holland.
Akerlof, G. and J. Yellen. 1990. “The fair wage-effort hypothesis and unemployment.” QJE 105: 255–283.
Arrow, K. 1963. “Uncertainty and the welfare economics of medical care.” AER 53: 941–973.
______ 1973. “The theory of discrimination.” In O. Ashenfelter and A. Rees, eds.,Discrimination in Labor
Markets. Princeton: Princeton University Press.
Atkinson, A. 1983. The Economics of Inequality. Oxford: Clarendon Press.
Atkinson, A., L. Rainwater, and T. Smeeding. 1995. Income Distribution in OECD Countries. Paris: OECD.
Atkinson, A. and J. Stiglitz. 1980. Lectures on Public Economics. New York: McGraw-Hill.
Banerjee, A. and M. Ghatak. 1995. Empowerment and Efficiency: The Economics of Tenancy Reform. Cambridge,
MA: MIT.
Banerjee, A. and A. Newman. 1993. “Occupational choice and the process of development.” JPE 101: 274–299.
Becker, G. 1964. Human Capital. New York: Columbia University Press.
______ 1991. A Treatise on the Family. Cambridge, MA: Harvard University Press.
Benabou, R. 1993. “Workings of a city: Location, education, production.” QJE 108: 619–652.
______ 1996. “Inequality and growth.” NBER Working Paper 5658, inNBER Macroeconomics Annual 1996,vol.
11.
Bewley, T. 1994. “A field study on downward wage rigidity.” New Haven: Yale University.
Blau, F. and L. Kahn. 1994. “The impact of wage structure on trends in US gender wage differentials.” NBER
Working Paper 4748.
Blundell, R. 1995. “The impact of taxation on labour force participation and labour supply.” OECD Jobs Study
Working Papers, 8. OECD Publishing. http://dx.doi.org/10.1787/576638686128.
Booth, A. and D. Snower. 1996. Acquiring Skills: Market Failures, Their Symptoms and Policy Responses.
Cambridge: Cambridge University Press.
Boudon, R. 1973. L’Inégalité des chances. Paris: Armand Colin.
Bound, J. and R. Freeman. 1989. “Black Economic Progress: Erosion of Post-1965 Gains in the 1980s.” In S. Shulman
and W. Darity Jr., eds., The Question of Discrimination: Racial Equality in the U.S. Labor Market,pp. 32–49.
Middletown, CT: Wesleyan University Press.
Bourdieu, P. and J.-C. Passeron. 1964. Les Héritiers. Paris: Minuit.
______ 1970. La Reproduction. Paris: Minuit.
Bourguignon, F. 1981. “Pareto-superiority of unegalitarian equilibria in Stiglitz’ model of wealth distribution with convex
savings function.” Econometrica 49: 1469–1475.
Bourguignon, F. and M. Martinez. 1996. Decomposition of the Change in the Distribution of Primary Family
Incomes: A Microsimulation Approach Applied to France, 1979–1989. Paris: DELTA.
Card, D. 1992. “The effect of unions on the distribution of wages: Redistribution or relabelling?” NBER Working
Paper 4195.
Card, D. and R. Freeman. 1993. Small Differences that Matter: Labor Markets and Income Maintenance in
Canada and the United States. Chicago: University of Chicago Press.
Card, D., F. Kramarz, and T. Lemieux. 1996. “Changes in the relative structure of wages and employment: A
comparison of the United States, Canada and France.” NBER Working Paper 5487.
Card, D. and A. Krueger. 1992. “Does school quality matter?” JPE 100: 1–40.
______ 1995. Myth and Measurement: The New Economics of the Minimum Wage. Princeton: Princeton
University Press.
Chamley, C. 1996. Capital Income Taxation, Income Distribution and Borrowing Constraints. Paris: DELTA.
Chassard, Y. and P. Concialdi. 1989. Les Revenus en France. Paris: La Découverte, “Repères.”
Coate, S. and G. Loury. 1993. “Will affirmative action eliminate negative stereotypes?” AER 83: 1220–1240.
Cohen, D., A. Lefranc, and G. Saint-Paul. 1996. “French unemployment: A transatlantic perspective.”Economic
Policy 12 (25): 265–292.
Coleman, J. 1966. Equality of Educational Opportunity. Washington, DC: US Dept. of Health, Education, and
Welfare.
Cooper, S., S. Durlauf, and P. Johnson. 1994. “On the transmission of economic status across generations.”ASA
Papers and Proceedings, pp. 50–58.
CSERC. 1996. Les Inégalités d’emploi et de revenu. Paris: La Découverte.
Davis, S. 1992. “Cross-country patterns of change in relative wages.” NBER Macroeconomics Annual 1992.
Douglas, P. 1976. “The Cobb-Douglas production function once again: Its history, its testing and some new empirical
values.” JPE 84: 903–915.
Drèze, J. and A. Sen. 1995. India: Economic Development and Social Opportunity. Delhi: Oxford University
Press.
Ducamin, R. 1995. Rapport de la commission d’études des prélèvements fiscaux and sociaux pesant sur les
ménages. Paris: Ministère de l’Économie et des Finances.
Duménil, G. and D. Lévy. 1996. La Dynamique du capital: un siècle d’économie américaine. Paris: PUF.
Ehrenberg, R. and R. Smith. 1994. Modern Labor Economics. New York: Harper Collins.
Eissa, N. and J. Liebman. 1996. “Labor supply response to the earned income tax credit.” QJE 111 (2): 605–637.
Erickson, C. and A. Ichino. 1995. “Wage differentials in Italy.” In R. Freeman and L. Katz, eds.,Differences and
Changes in Wage Structure, pp. 265–306. Chicago: University of Chicago Press.
Erikson, R. and J. Goldthorpe. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies.Oxford:
Clarendon Press.
Feenberg, D. and J. Poterba. 2000. “The income and tax share of very high income households.”AER 90 (2): 264–
270.
Feldstein, M. 1995. “The effect of marginal tax rates on taxable income: A panel study of the 1986 Tax Reform Act.”
JPE 103 (3): 551–572.
Fleurbaey, M. 1996. Théories économiques de la justice. Paris: Economica.
Freeman, R. 1973. “Changes in the labor market status of Black Americans, 1948–1972.” Brookings Papers on
Economic Activity, no. 1, pp. 67–120.
______ 1995. “Are your wages set in Beijing?” Journal of Economic Perspectives 9 (3): 15–32.
______ 1996. Disadvantaged Young Men and Crime. Cambridge, MA: Harvard University Press.
Freeman, R. and J. Medoff. 1984. What Do Unions Do? New York: Basic Books.
Friedman, M. 1962. Capitalism and Freedom. Chicago: University of Chicago Press.
Goldin, C. and R. Margo. 1992. “The great compression: The wage structure in the United States at mid-century.”
QJE 107: 1–34.
Goolsbee, A. 1997. “What happens when you tax the rich? Evidence from executive compensation.” NBER Working
Paper 6333.
Gottschalk, P. 1993. “Changes in inequality of family income in seven industrialized countries.” AER 83 (2): 136–142.
Goux, D. and E. Maurin. 1996. “Meritocracy and social heredity in France: Some aspects and trends.”European
Sociological Review 13 (2): 159–177.
Hamermesh, D. 1986. The demand for labor in the long run. In O. Ashenfelter and R. Layard, eds.,Handbook of
Labor Economics, pp. 429–471. Amsterdam: North Holland.
______ 1993. Labor Demand. Princeton: Princeton University Press.
Harhoff, D. and T. Kane. 1994. “Financing apprenticeship training: Evidence from Germany.” NBER Working Paper
4557.
Henriet, D. and J.-C. Rochet. 1988. “Équilibres et optima sur les marchés d’assurance.” In Mélanges économiques
en l’honneur d’Edmond Malinvaud. Paris: Economica.
Herrnstein, R. and C. Murray. 1994. The Bell Curve: Intelligence and Class Structure in American Life.New
York: The Free Press.
IMF. 1996. World Economic Outlook.
INSEE. 1994. “Un siècle de données macroéconomiques.” INSEE Résultats, nos. 303–304.
______ 1995. “Revenus et patrimoine des ménages, édition 1995.” INSEE Synthèses, no. 1.
______ 1996a. “Séries longues sur les salaires.” INSEE Résultats, no. 457.
______ 1996b. “Revenus et patrimoine des ménages, édition 1996.” INSEE Synthèses, no. 5.
______ 1996c. “Rapport sur les comptes de la nation 1995.” INSEE Résultats, nos. 471, 472, 473.
______ 1996d. “L’évolution des salaires jusqu’en 1994.” INSEE Synthèses, no. 4.
______ 2002. “Les salaires dans l’industrie, le commerce et les services en 2000.” INSEE Résultats Sociétés, no. 7.
Judd, K. 1985. “Redistributive taxation in a simple perfect foresight model.” Journal of Public Economics 28: 59–83.
Juhn, C., K. Murphy, and B. Pierce. 1993. “Wage inequality and the rise in returns to skill.” JPE 101: 410–442.
Juhn, C., K. Murphy, and R. Topel. 1991. “Why has the natural rate increased over time?” Brookings Papers on
Economic Activity 2: 75–142.
Kahneman, D., J. Knetsch and R. Thaler. 1986. “Fairness as a constraint on profit seeking.” AER 76: 728–741.
Katz, L., G. Loveman, and D. Blanchflower. 1995. “A comparison of changes in the structure of wages in four OECD
coutries.” In R. Freeman and L. Katz, eds.,Differences and Changes in Wage Structure. Chicago: University of
Chicago Press.
Kolm, S. C. 1972. Justice et équité. Paris: Éditions du CNRS.
Kramarz, F., S. Lollivier, and L. Pelé. 1995. “Wage inequalities and firm-specific compensation policies in France.”
Document de travail INSEE-CREST 9518.
Kremer, M. and E. Maskin. 1996. “Wage inequality and segregation by skill.” NBER Working Paper 5718.
Krueger, A. and L. Summers. 1988. “Efficiency wages and the interindustry wage structure.” Econometrica 56: 259–
293.
Krussel, P., L. Ohanian, J. V. Rios-Rull, and G. Violante. 1996. “Capital-skill complementarity and inequality.”
University of Rochester.
Kuznets, S. 1955. “Economic growth and economic inequality.” AER 45: 1–28.
Landais, C. 2007. “Les hauts revenus en France (1998–2006): une explosion des inégalités?” PSE Working Paper.
Lefranc, A. 1997. “Évolutions des marchés du travail français et américains: quelques éléments d’analyse
comparative.” Revue Economique 48 (5): 1041–1060.
Lemieux, T. 1993. “Unions and wage inequality in Canada and in the United States.” In D. Card and R. Freeman,
eds., Small Differences that Matter. Chicago: University of Chicago Press.
Lhomme, J. 1968. “Le pouvoir d’achat de l’ouvrier français au cours d’un siècle: 1840–1940.” Le Mouvement social
63: 41–70.
Liebman, J. 1996. “Essays on the earned income tax credit.” PhD diss., Harvard University.
Lollivier, S. and D. Verger. 1996. “Patrimoine des ménages: déterminants et disparités.” Économie et Statistique
296–297: 13–32.
Lucas, R. 1990a. “Supply-side economics: An analytical review.” Oxford Economic Papers 42, pp. 293–316.
______ 1990b. “Why doesn’t capital flow from rich to poor countries?” AER 80: 92–96.
Mankiw, G., D. Romer, and D. Weil. 1992. “A contribution to the empirics of economic growth.” QJE 107: 407–437.
Meyer, C. 1995. “Income distribution and family structure.” PhD diss., MIT.
Morrisson, C. 1991. “L’inégalité des revenus.” In M. Lévy-Leboyer and J.-C. Casanova, eds.,Entre l’État et le
marché: l’économie française de 1880 à nos jours. Paris: Gallimard.
______ 1996. La Répartition des revenus. Paris: PUF.
Mulligan, C. 1996. Parental Priorities and Economic Inequality. Chicago: University of Chicago Press.
Murphy, K., A. Shleifer, and R. Vishny. 1989. “Income distribution, market size and industrialization.” QJE 104: 537–
564.
Murphy, K. and F. Welch. 1993a. “Inequality and relative wages.” AER 83 (2): 104–109.
______ 1993b. “Occupational change and the demand for skill.” AER 83 (2): 122–126.
Neumark, D. and W. Wascher. 1994. “Employment effects of minimum and subminimum wages: Reply to Card, Katz
and Krueger.” Industrial and Labor Relations Review 48: 497–512.
Nizet, J.-Y. 1990. Fiscalité, économie et politique: l’impôt en France, 1945–1990. Paris: LGDJ.
OECD. 1985. The Integration of Women in the Economy. Paris: OECD.
______ 1993. Perspectives de l’emploi, July. Paris: OECD.
______ 1995. Statistiques des recettes publiques des pays membres de l’OCDE, 1965–1994. Paris: OECD.
______ 1996. Economic Outlook no. 59, June. Annual Projections for OECD Countries.http://stats.oecd.org/ Paris:
OECD.
______ 2000. Taux de chômage standardisés (www.oecd.org). Paris: OECD.
Phelps, E. 1968. “The statistical theory of racism and sexism.” AER 62: 659–661.
______ 1994. Structural Slumps: The Modern Equilibrium Theory of Unemployment, Interest and Assets.
Cambridge, MA: Harvard University Press.
Piketty, T. 1994. “Inégalités et redistribution.” Revue d’économie politique 104: 769–800.
______ 1995. “Social mobility and redistributive politics.” QJE 110: 551–584.
______ 1997a. “La redistribution fiscale face au chômage.” Revue française d’économie.
______ 1997b. “Les créations d’emploi en France et aux États-Unis: ‘services de proximité’ contre ‘petits boulots?’ ”
Notes de la fondation Saint-Simon 93 (December). See also Économie et Statistique 318 (1998–8): 73–
99.
______ 1998. “L’impact des incitations financières au travail sur les comportements individuels: une
estimation pour le cas français.” Économie et Prévision 132–133 (Jan.-Mar.): 1–35.
______ 1999. “Les hauts revenus face aux modifications des taux marginaux supérieurs de l’impôt sur le revenu en
France, 1970–1996.” Économie et Prévision 138–139 (Apr.-Sept.): 25–60.
______ 2001. Les Hauts Revenus en France au XXe siècle. Inégalités et redistributions 1901–1998. Paris:
Grasset.
Piketty, T. and E. Saez. 2003. “Income inequality in the United States, 1913–1998.” QJE 118: 1–39.
Rawls, J. 1972. A Theory of Justice. Oxford: Clarendon Press.
Roemer, J. 1996. Theories of Distributive Justice. Cambridge, MA: Harvard University Press.
Rosanvallon, P. 1995. La Nouvelle Question sociale. Paris: Seuil.
Rothemberg, J. 1996. “Ideology and the distribution of income.” Cambridge, MA: MIT.
Rothschild, M. and J. Stiglitz. 1976. “Equilibrium in competitive insurance markets.” QJE 90: 629–650.
Shavit, Y. and H. P. Blossfeld. 1993. Persistent Inequality: Changing Educational Attainment in 13 Countries.
Boulder: Westview.
Slemrod, J. 1995. “Income creation or income shifting? Behavioral responses to the Tax Reform Act of 1986.” AER 85
(2): 175–180.
Solow, R. 1956. “A contribution to the theory of economic growth.” QJE 70: 65–94.
______ 1958. “A skeptical note on the constancy of relative shares.” AER 48: 618–631.
Spence, M. 1974. Market Signalling: Informational Transfer in Hiring and Related Screening Processes.
Cambridge, MA: Harvard University Press.
Topel, R. 1993. “What have we learned from empirical studies of unemployment and turnover?” AER 83: 110–115.
Van Parijs, P. 1995. Real Freedom for All: What (If Anything) Can Justify Capitalism? Oxford: Clarendon Press.
Williamson, J. 1985. Did British Capitalism Breed Inequality? Boston: Allen & Unwin.
Williamson, J. and P. Lindert. 1980. American Inequality: A Macroeconomic History. New York: Academic Press.
Wilson, W. J. 1987. The Truly Disadvantaged: The Inner City, the Underclass and Public Policy. Chicago:
University of Chicago Press.
Wolff, E. 1992. “Changing inequality of wealth.” AER 82 (2): 552–558.
Young, A. 1995. “The tyranny of numbers: Confronting the statistical realities of the East Asian growth miracles.”
QJE 110: 641–680.
Contents in Detail
Hamermesh, Daniel, 49
Health insurance: adverse selection and, 115; justifications for compulsory, 115–117; as percentage of social
charges in 1966 France, 103
Herrnstein, Richard, 82, 87
Hidden underemployment, 25
Household size, income and, 12, 14, 22
Human capital: elasticity of supply of, 78–79, 82–83, 87, 107; measuring types of productivity and, 98; unequal
distribution of, 58–60; wage inequality and, 88–89, 92–99
Human capital, structural causes of inequality, 78–79; affirmative action versus fiscal transfers, 86–88;
discrimination in labor market and, 85–86; efficiency and, 79–81; inefficient social integration and, 83–84;
role of family and education expenses, 81–83
Human capital theory, 66–68; globalization and wage inequality, 73–74; historical inequalities and, 68–69; rise of
wage inequality since 1970, 70–71; skill-biased technological change and, 71–73, 76–77, 92; supply and
demand and, 69–70
Incentives: basic income and, 113; credit markets and, 60, 62, 114; effects of redistribution on, 105–110; of
households to save and invest, 35; human capital and investments, 78–88, 90, 93; of owners to accumulate
capital and invest, 28–29
Income: distribution by deciles and centiles, 5–8, 6t; household size and, 12; inequality of, 12–16, 12t, 15t;
inequality of, historical evolution, 17–25, 19f, 21t; left-right debate about inequality of, 1–3; types and
distribution of, 5–8, 6t. See also Capital income; Wage entries
Income, share of capital in: capital-labor substitution, 27–40; classic and marginalist theories, 40–55. See also
Capital income
Income tax, progressive, 19, 48, 64, 102–103, 106
India: human capital, 59–60, 80, 85; inequality, 16, 17
Individual self-interest, economic organization and, 39, 106
Inflation, as stimulus, 121
Information: credit markets and problems of, 61; insurance markets and, 114–115; prices and, 38
Inheritance of capital, 26; progressive tax on estates, 19, 64
Insurance society, myth of, 119, 121
Integration of schools, human capital and, 84
Intergenerational mobility, human capital and, 81, 83
Intermediate consumption, as share of firms’ costs, 42
Intertemporal markets, 60–61, 114, 115–116, 118
Italy: historical evolution of inequality, 21; labor participation rate, 25; profit share, 53; unemployment, 24
Katz, Larry, 96
Keynes, John Maynard, 40; demand management, 114, 119–121
Kolm, Serge-Christophe, 2
Krueger, Alan, 95–96
Kuznets curve, 18–20, 22, 66
Kuznets, Simon, 18
Labor income: as share of firms’ costs, 42; ways of redistributing, 74–77. See also Elasticity of substitution
between capital and labor; Wage entries
Labor market participation rate, 25, 108
Left-right conflict, about inequality and redistribution, 1–3
Lollivier, Stéfan, 13
Lucas, Robert, 59
Luxembourg Income Study (LIS), 14, 22
Marginalist theory, of capital-labor split. See Classical and marginalist theories, of capital-labor split
Marginal productivity of labor, capital-labor substitution and, 29–30, 34, 59
Marginal rates of redistribution. See Average and marginal rates of redistribution
Market economy. See Price system
Marx, Karl, 26, 30, 39; proletarianization thesis of, 17–18
Maximin principle, of Rawls, 2, 35, 106
McGovern, George, 112
Means of production, collectivizing of, 39, 62, 63–64
Minimum wage: EITC and, 109; health insurance and, 103; monopsony power of employers, 96; raising of, and
effect on level of employment, 95–96; redistribution and, 75, 94; unions and, 91; in US and France, 50, 110–
111, 117; wage distribution and, 8
Monopoly power, of unions, 89, 94
Monopsony power, of employers, 94–96, 113–114, 121
Moral hazard, credit markets and, 60–61
Murray, Charles, 82, 87
OECD countries: evolution of shares of profits and wages, 49–53, 50t; historical evolution of inequality, 21;
income inequality, 14–15, 15t; wage inequality, 10–11, 11t
Savings: behavioral differences and, 13; credit markets and, 56–57, 60; elasticity of capital supply and, 35–37;
Marx and, 39; pensions and, 115–116, 118
Self-employment: calculation of value added, 43; compensation as percentage of household income, 5, 6t; credit
rationing and, 62; income inequality and, 12
Self-interest, and economic organization, 39, 106
Singapore, 58
Skill-biased technological change, 71–73, 76–77, 92
Social charges: average and marginal rates of redistribution, 100–102, 108; employer-employee shares of, 46–
48, 104–105; fiscal incidence and, 52–53; as political issue, 76–77; price of labor and, 32, 42–44; shifting
burden of from low- to high-wage jobs, 34, 76–77, 111, 113, 119; unemployment and, 110–111
Social income: household income and, 5, 6t, 7, 105; income inequality and, 12, 23; taxes and, 47
Social insurance: efficient redistribution and, 114–116; as instrument of fiscal redistribution, 116–119; markets as
intertemporal markets, 61
Social integration, housing and educational outcomes and human capital, 83–84
Social justice: left-right consensus about fundamental principles of, 1–2; left’s skepticism about taxes and, 38,
52, 76; methods of redistribution and economic efficiency, 2–3, 32, 35–37, 55–57, 61, 64–65, 74, 105–106
Social origin, influence on human capital, 80–81
Solow, Robert, 30, 48, 59
South America: growth rate per capita, 58; human capital, 59; income inequality, 15
South Korea, 58
Stimulus programs, redistribution and, 121
Subsidized loans, as possible intervention in credit market, 62–63
Supply and demand: human capital and wages, 68, 69–70, 73–74, 88–89; unemployment and social charges,
111–112. See also Demand management
Sweden: historical evolution of inequality, 22; income inequality, 14; percentage of obligatory taxes in, 101; wage
inequality, 10
Taiwan, 58
Taxes: average and marginal rates of redistribution, 100–105; capital income and, 28–32, 36–37, 43–44, 64–65;
Earned Income Tax Credit, in US, 108–109, 112; effect on disposable income, 14, 22–23; Left’s skepticism
of redistribution through, 38, 52, 76; negative income tax, 1, 3, 112–113; progressive, on estates, 19, 64;
progressive, on income, 19, 48, 64, 102–103, 106; receipts as percent of GDP, 44; redistribution methods, 31–
35, 75–76; revenue and, 106–108. See also Social charges
Theil and Atkinson indices, 10
Time and space, income inequality and, 16–17, 16t
Value-added tax (VAT), 44–45, 103. See also Economic value added
Verger, Daniel, 13
Wage inequality, 8–11, 9t, 11t; historical evolution of, 20–21; human capital and, 66–74; human capital and,
structural causes of, 78–88; unemployment and underemployment, 23–25; in US and UK, 53
Wage inequality, social determination of, 88–89; efficiency wages and fair wages, 97–98; monopsony power of
employers, 94–96; national traditions and, 98–99; role of unions in setting wages, 89–94
Wages: binding wage schedules, 92–93; capital-labor substitution and, 30–32; labor income and capital share,
42; as percentage of household income, 5–7, 6t. See also Wage inequality entries
Welfare state: basic income and, 113; EITC and, 108–109; limits of, 108; redistribution and intent of, 47, 105,
114. See also Social charges
Women, labor market improvements and, 87–88