David Ellwood and Christopher Jencks
John F. Kennedy School of Government, Harvard University
The following remarks were presented at the 1839th Stated Meeting,
held at the House of the Academy in Cambridge on November 8, 2000.
|Speakers David Ellwood and Christopher Jencks
David Ellwood, Lucius N. Littauer Professor of Political Economy at
the Kennedy School, is a labor economist specializing in poverty and welfare,
family change, and wage inequality. He previously served as assistant secretary
for planning and evaluation in the US Department of Health and Human Services
and codirected the initial welfare reform efforts of the Clinton
administration. Christopher Jencks is Malcolm Wiener Professor of Social Policy
at the Kennedy School. His recent research has dealt with changes in the
material standard of living over the past generation, homelessness, the effects
on children of growing up in poor neighborhoods, welfare reform, and poverty
This is the first in a series of Stated Meeting lectures on how
inequalities in race, gender, and income continue to divide American society.
The Academy has long been concerned about these problems. In the 1960s and
1970s a number of pathbreaking Dædalus
issues on such topics were published, and Academy studies led to several
volumes on race, ethnicity, and poverty in the United States by Fellows Nathan
Glazer, Daniel Patrick Moynihan, and James Sundquist. Since then the Academy
has conducted numerous studies resulting in publications on themes as varied as
race and schooling, ethnic pluralism and public policy, the meeting of social
needs through public-private partnerships, medical care for poor patients,
homelessness, and welfare-to-work programs. Through the 2000–01 Stated Meeting
series on inequalities, the Academy seeks to reevaluate what has been achieved
in the past quarter-century and assess the challenges that await us in the
To some in the media, the American economy is an unprecedented
success story; to others, the rich are richer while the poor continue to
struggle. Can both be true? How can we understand these conflicting voices? In
analyzing inequality, we need to take into account not only wages and incomes
but also family structure.
The confusion that arises in assessing "winners and
losers" is illustrated by two measures of economic performance: national
income per personthe amount of money that could be given to every
individual if all the income produced by the economy were divided equally
and the median annual earnings of full-year male workers. Between 1973, a peak
year, and 1996, another good year, national income per person rose 34 percent.
The median earnings of full-time male workers rose with national income in the
1960s, then leveled off and began falling; in fact, between 1973 and 1996,
earnings declined by 10 percent.
How can we account for these inconsistent patterns? The concept of
taking total income and dividing by total population is misleading because
Americans are having fewer children. As the ratio of children to adults falls,
the national income per person rises, even if the adults are earning no more
money. Therefore, it makes more sense to look at national income per adult.
In 1973 income per adult was $26,754; in 1996 it had risen to
almost $32,000a difference of $5,000. Where did the money go?
Approximately $1,800in the form of rents, profits, interest, and the
likewent to capital holders. A significant portion went to greater
compensation for the larger number of women working outside the homefrom
a $2,000 increase for women in the top third of wages to a $372 increase for
those in the lowest third. What about men? Although the wages of men in the top
third increased by $1,000, men in the middle and lower third actually earned
To provide more detail on the factors that contribute to
inequality, we need to examine what has happened to wages for people at
different places in the wage distribution. When we have shared prosperity, the
wages of people in different thirds of the wage distribution have gone up at
the same rate, which is precisely what happened in the 1960s. Then in 1973 the
country experienced the oil price shock, and the economy went into a tailspin
for about ten years; the result was shared stagnation. But starting in 1979,
patterns began to diverge significantly. Wages for men at the top began to
grow, with those in the 90th percentile earning 60 percent more than they did
in the 1960s; simultaneously, the pay for those at the bottom fell so much that
they did no better than they had in the 1960s.
What happened to wages of full-year, full-time women workers in the
same period? They shared the prosperity of the 1960s and the stagnation of the
1970s. But in the period between 1973 and 1996, women at all levels showed some
rise in income. One of the key factors underlining the rise in women's
aggregate earnings was the dramatic increase in the number of hours they
worked. Some 24 percent more women were in the workforce, and those who worked
averaged 15 percent more hours per week. More hours worked also led to more
experience accumulated, with greater rewards from employers. In addition, women
benefited from a changing economy, with its greater emphasis on service-related
occupations, and perhaps from reduced discrimination and affirmative action in
the workplace. As in the case of men, compensation raises were greatest for
those with the most education and those who were highest in the
wage-distribution scale; women in the top third had a hourly wage increase of
What have these figure meant for families? For husband-wife
families with children, incomes for those in the top third rose a dramatic 57
percentthe result of the combined effects of higher earnings for highly
educated husbands and much higher earnings for highly educated wives. For those
at the bottom, the fact that women worked more hours almost exactly offset the
declining earnings of men, meaning that for these less fortunate two-parent
families, earnings have been essentially flat for the past twenty years. Thus,
inequality in family income has grown considerably among husband-wife families.
Moreover, in this period there has been a significant increase in
the fraction of children living with one parentfrom 6 to 7 percent in the
1960s to almost 25 percent today. Single parents are disproportionately from
lower educational backgrounds, and the children in these families typically
have far lower incomes than those living with two parents. Here too there have
been major changes over time, with single parents in the top third of parental
education demonstrating a sizable rise in income.
The educational level of women is particularly important to our
understanding of what has happened to family income. If a mother is at the
highest educational level, the odds that she is a single parent have hardly
changed during these years. By contrast, the fraction of women who have become
single parents in the least educated group has grown enormously. As wages
spread out and family structure changes simultaneously, the two become mutually
reinforcingwith long-term implications, including fewer opportunities,
for the children. In the mass of data, one conclusion is clear: A mother at the
highest educational level is more likely to be in a two-parent family and more
likely to have a higher income. At no time has it been more important to choose
your parentsparticularly your motherwisely.
Whereas David has talked about changes in the distribution of
individual earnings, I will focus on changes in the distribution of household
income, which reflect the distribution of both individual earnings and
retirement income, as well as the ways individuals sort themselves into
households. Census data on the distribution of household income between 1967
and 1998 reveal that since 1980 the rich have become a lot better off, the
middle class has become a little better off, and the poor have become no worse
off. In conventional economic analysis, if some people gain and nobody loses,
the population as a whole has gained.
Is this conclusion correct? Or do changes in richer people's
incomes generate what economists like to call "externalities" that
affect the welfare of poorer people whose purchasing power has not changed? If
other people get richer and I do not, how does that affect me?
In sociology the standard answer is that other people's incomes
matter because their consumption level affects my subjective assessment of my
needs. If most teenagers start buying Nike sneakers instead of Converses, my
teenager will feel worse off if he cannot buy Nikes, and I will feel worse off
because I can no longer buy him what he needs to feel comfortable at school.
It's a short step from there to the idea that when other people get richer, my
objective needs may also change. When almost everyone can afford a car, public
transportation decays, so those who cannot afford a car have more trouble
Such externalities are real, but they are not the ones I want to
discuss in this talk. Instead I want to focus on some other mechanisms by which
the gains of the rich seem to influence what happens to those whose incomes do
First, I want to compare the educational attainments of students
whose parents are in the top and bottom quartiles of the income distribution.
If incomes rise at the top and stagnate at the bottom, what should we expect to
see? In the simplest account, we expect the children of the rich to gain while
the children of the less affluent don't change much.
In a more complex version, we would expect that since rising
inequality means that the payoff to college is rising, parents and students at
all income levels would invest more in schooling, and educational attainment
would rise for everyone. However, if those at the bottom have trouble financing
higher education, we might not expect their children to change as much as more
affluent children. When we compare high-school graduates in the early 1980s,
before the big jump in inequality, with those who graduated in the early 1990s,
after the big jump, we see that college attendance rose a lot among the more
affluent, a little among middle-income families, and hardly at all among the
poorest families. These findings imply that the poor are worse off only if you
think that schooling is a positional good whose value depends not on how much
you have but on how much you have relative to others with whom you are
competing. Most of the available evidence fails to support the positional
interpretation of the payoff to schooling, so one cannot argue that the poor
were worse off in 1992 than in 1982.
Now I want to ask a different question: Do other people's incomes
affect a child's educational prospects, independent of the child's own family
income? One way to consider this is to compare American states. Over the past
thirty years inequality has grown far more in some states than in others.
Whenever economic inequality grows, the increase can take two distinct forms:
neighborhoods can become more internally heterogeneous, with bigger income gaps
between neighbors; or people who get richer can move to better neighborhoods,
leaving the distribution of income in their old neighborhoods pretty much
In investigating this issue, Susan Mayer of the University of
Chicago has found that while income inequality increased between 1970 and 1990,
almost all of the increase took the form of greater inequality among, not
within, neighborhoods. This has important substantive implications. If
people mainly compare themselves with their neighbors, they are not going to
see themselves as falling farther behind, because the big winners in the
economic lottery will have moved elsewhere. Of course, people also compare
themselves with others outside their neighborhood, but the most important
vehicle for such comparisons is probably television, which has always projected
a vision of how Americans live that is far beyond the means of low-income
What are the effects of growing residential segregation by income?
First, it leads to increasing disparities in school qualitynot only
because the tax base for local support is more unequal, but also, even more
important, because the strongest single determinant of where able teachers
choose to work is the socioeconomic mix of the students they will be teaching.
The student mix is a far more powerful factor than salaries, so even when
districts with a large proportion of poor children pay a little more than
nearby districts, as they often do, they cannot compete successfully for the
Although there are no data linking the growth of economic
inequality to disparities in students' test performance, Mayer has examined the
impact of inequality on the length of time students remain in schoolan
equally important determinant of adult economic success. She finds that in
states with high levels of economic segregation, low-income students get less
education while high-income students get more than they would in more equal
states. In itself, this situation is not surprising; what is surprising is that
it remains true even when she compares students whose parents had exactly the
same level of education and exactly the same average income over a lengthy
So the moral is that if you want your children to stay in school
and attend college, and your income is above average, it is better to live in
an unequal state. Roughly speaking, a 10 percent increase in income inequality
is worth as much as a 40 percent increase in your own income. If you have
below-average income, however, you should live in a more equal state. The
difference seems to be linked to the amounts that states spend on both public
higher education and public elementary and secondary education, although I
should emphasize that these results are fairly fragile.
The strong link between income and health is present throughout the
income distribution, not just at the bottom end. At age 50, being near the top
of the distribution rather than in the middle is a strong predictor of how long
you will live.
There is a little evidence suggesting that as economic disparities
between rich and poor have widened. Other health disparities may also have
widened, but that is not well established. Since the early 1990s there has also
been an explosion of research on the relationship between an individual's
health status and the incomes of other people in the same state or the same
country. Initially, this research concentrated on comparisons among rich
countries and seemed to show a strong relationship between income inequality
and health. As more countries were brought into the analysis and different
periods were analyzed, however, that relationship seemed to disappear.
My own work has looked at how changes in inequality relate to
changes in life expectancy. I often find some relationship, but it is weak.
Increasing inequality by a tenth, which is about the increase evident in the
United States over the past generation, lowers life expectancy by about two
months. Life expectancy rises by about two months every year in rich countries,
so the impact of modest changes in inequality looks tiny.
A parallel literature on economic inequality in American states
seems to indicate that people in more equal states have a higher life
expectancybut the picture shifts dramatically when we look at changes
over time. During the 1960s the states in which inequality fell the most had
the smallest improvement in life expectancy; during the 1970s and 1980s changes
in inequality had no effect on life expectancy. I think we can conclude that
although inequality may affect health, the impact is fairly small.
Macro Effects of Inequality on Society as a Whole
I now want to consider three areas in which economic inequality
could affect everyone, rich and poor.
Crime. Increases in crime hit the poor harder than the rich, but they affect us
all. They have a direct impact if we are victims, but they also have huge
indirect effects because our efforts to avoid becoming victims make our lives
more inconvenient and less enjoyable. Several papers have found that crime is
higher in metropolitan areas where the distribution of income is more unequal,
but they have not looked at whether increases in inequality are associated with
increases in crime. I cannot address that question now, but I can say that no
such temporal correlation emerges at the national level:
Inequality fell during the 1960s. Crime rose.
Inequality rose during the early 1980s. Crime fell.
Inequality rose in the late 1980s. Crime rose.
Inequality was pretty flat in the 1990s. Crime fell.
These facts do not suffice to prove that changes in economic inequality play no
role in the crime rate, but they do suggest that any such role is swamped by
other influences, most of which we do not understand.
Economic growth. During the 1990s a number of reports suggested that high
levels of inequality lead to lower levels of subsequent growth. On closer
analysis, however, it becomes apparent that this finding applies mainly to
poorer countries where high levels of inequality are associated with extreme
poverty, preventing many parents from providing their children with even
minimal nutrition and education. In addition, extreme inequality in poorer
countries is often associated with public policies aimed at preserving the
privileges of the ruling class, and many of those policies are bad for growth.
The evidence for a link between inequality and subsequent growth in rich
countries is very soft; indeed, we would not expect such a link. A little
reflection suggests that both extreme inequality and extreme equality are
likely to lower growth. The problem is to find the optimal point between these
Looking across rich countries since 1945, it seems pretty clear that if
economic inequality has any influence on growth, its impact has been swamped by
other more powerful factors. Europe is a lot more equal than either the United
States or the other English-speaking democracies, yet Europe has grown about as
fast as the United States over the past fifty years, and Europe is only
marginally poorer than the United States today.
Political health of the country. No other stable democracy is as economically
unequal as the United States. Should this worry us? Five facts seem relevant:
The United States has practically the lowest electoral turnout of any stable
The disparity in turnout between rich and poor voters is higher in the United
States than in any other rich country for which I have seen data.
When turnout falls, the income disparity in turnout widens (although I should
note that some deny this).
Both theory and empirical observation suggest that when turnout among the poor
falls, politicians' interest in redistributive policies that mainly benefit the
poor also falls. Politicians do engage in occasional random acts of charity,
and rich voters do support some forms of redistribution. But when altruism is
the only source of support for redistributive policies, not much redistribution
The biggest determinants of economic inequality in rich countries are political
(e.g., centralized wage bargaining, strong left parties).
If we compare the United States with Germany in 1990, for example,
disparities between the top and bottom income deciles for both wages and
household income were about twice as high in the United States as in Germany.
That was also true for most other countries of the European Union. The
distribution of skills plays a minor role in explaining why Europe is more
equal than the United States. What differs is the political environment.
These facts raise a troubling question that I cannot answer: Can
the growth of inequality lead to the demobilization of voters, which then leads
to further growth in inequality, ad infinitum? The problem of political
withdrawal is especially worrisome in a country where there are few restraints
on the use of money to buy political influence. In such an environment,
political candidates pursue a two-part strategy that combines extraordinary
attention to special interest groups with an effort to woo the typical
inattentive voter by moving to the exact center of the political spectrum. This
convergence maximizes each candidate's electoral success, but it also leads to
a "tragedy of the commons," in which most Americans, who by
definition are not located at the center of the distribution, feel
unrepresented, alienated, and disinclined to vote. The situation poses huge
obstacles for anyone who hopes to reverse the growth of economic inequality or
reinvigorate the democratic process.
Remarks © 2000 by David Ellwood and Christopher Jencks,
respectively. Photo © 2000 by Martha Stewart.