Why Is Education Reform so Hard?
Eric A. Hanushek*
Stanford University and National Bureau of Economic Research
* This paper was originally presented at Louisiana State University during a visit as a Reilly Distinguished
Public Policy Fellow at the Reilly Center for Media and Public Affairs.
Why Is Education Reform so Hard?
By Eric A. Hanushek
Reform has been the operable word in education for many decades. We are continuously
undergoing moves to reform, whether because the Russians launched the first satellite,
because we wish to cure the problems of poverty, or because we feel are spending is not
bringing desired results. Unfortunately, while reform of schools may well be
appropriate, the approaches we have taken have been ineffective – and are likely to
continue being ineffective unless significant changes are made.
This discussion begins by setting out the need for reform. It then outlines the
alternative approaches to reform that exist. Finally, it provides recommendations for
actions based on the ideas of increasing the performance incentives in schools and
providing better information about the performance of schools.
The need for reform
Understanding the current state of U.S. education is important both for motivating reform
discussions and for diagnosing possible reforms. The story begins with the pattern of
student performance. Figure 1 tracks achievement of 17-year-olds over three decades.
These data come from the National Assessment of Educational Progress, or NAEP, and
are generally recognized as a reliable indicator of how performance compares over time.
From this figure, reading and math performance are slightly up, while science and writing
are significantly down. In simplest terms, student performance looks flat over the last
three decades of the twentieth century.
Figure 1. Performance of 17-year-olds on the National Assessment of Educational
Progress (NAEP), 1970-1999
Being flat might not be bad if in fact this performance was high. But, by
international comparisons, it is hard to say that this performance is particularly good.
Table 1 displays the position of the U.S. on the Third International Mathematics and
Science Study (TIMSS), an international test taken in 1995 by a large number of
developed and developing countries. The performance of U.S. students, while near the
center of the science distribution in the eighth grade, falls to the bottom grouping by
twelfth grade. This performance actually came as no surprise, because the United States
has performed similarly on each of the international tests that have been given since the
Table 1. Country Ranking of Performance on Third International Mathematics and
Science Study (TIMSS), 1995
8th Grade Performance
12th Grade Performance
Mathematics Science Mathematics Science
Singapore Singapore Netherlands Sweden
Republic Sweden Netherlands
Switzerland Austria New
Netherlands Hungary Australia Austria
UNITED STATES Lithuania
Canada UNITED STATES Lithuania
Note: bold – significantly about United States; italics – significantly below United States.
One observation at odds with these comparisons is the fact that the U.S. led the
world in economic growth during the twentieth century, and most economists believe that
education and human capital is an important part of growth. Part of this appears to be
simply that the U.S. has substituted greater amounts of schooling for quality. Thus, by
having higher levels of school attainment, we have managed to get around the lower
content of each grade. Three other possibilities could also contribute to explaining our
better than expected economic performance. First, the U.S. higher education system,
frequently regarded as the best in the world, is able to overcome the poorer training in
primary and secondary schools. Second, even though the U.S. education system does not
provide general skills to the population, it could encourage and develop creativity –
creativity that comes through in inventions and technological change. Third, the
exceptional growth performance of the U.S. economy could reflect factors other than
schooling – an open labor market that encourages adjustment to new technologies, a well-
oiled capital market, an economy relying more on private decisions than governmental
decisions, and so forth. While we cannot currently distinguish among these explanations
for the performance of the U.S. economy, I suspect that all contribute.
But, just because all enter does not mean that we can ignore the performance of
our students and our schools. Other nations of the world, for example, are rapidly
increasing the educational attainment of their populations. They also appear to be doing
this in many cases without compromising the quality of their schools. Thus, the ability of
the U.S. to substitute high quantity of schooling for lower quality in comparison to other
countries is much less possible today than it was even twenty years ago.
There is another important aspect of the flat performance of U.S. schools: it does
not reflect lack of trying. The U.S. has substantially increased the resources devoted to
schools. Figure 2 shows the increase in real spending per pupil between 1890 and 1990.
Over this period, spending per pupil grew at 3 ½ percent each year after allowing for
inflation. This is a tremendous growth rate to sustain for a century.
Figure 2. Real spending per pupil, 1890-1990
1990 $'s $3,000
$0 1890 1910 1930 1950 1970 1990
Moreover, in the relevant recent period this change was accomplished in exactly
the ways commonly advocated. Table 2 shows the decreases in pupil-teacher ratios and
the increases in teacher degree levels and teacher experience from 1960 on. These
elements each contribute importantly to the cost of schooling, so it is not surprising that
the real spending calculations at the bottom of the table show a tripling between 1960 and
Table 2. Public School Resources in the United States, 1960-1995
% teachers with master's
degree or more
median years teacher
Source: U.S. Department of Education (1997)
These data suggest that the resources available to school have not been used in
very productive ways. Increases in spending have not translated into increases in student
Other factors can explain part but not all of this. Specifically, outside changes
have put cost pressures on schools, implying that the education provided in the past may
be more expensive to achieve today. First, kids may come to school less well prepared
today than in the past. We know, for example, that the prevalence of single-parent
families has increased over time. Tied to this, poverty rates among children have grown.
These would be expected to work against student achievement – and would imply that
schools have to work harder to achieve the same results now as in the past. But,
offsetting these, the education level of parents has increased, and family sizes have fallen.
These would be expected to work in favor of student achievement. Although it is
difficult to weigh the opposing factors with any precision, the best guesses suggest that
they about evenly balance each other.
Second, other cost factors enter. Since the incorporation of requirements for
serving handicapped students, the special education population has grown substantially:
from eight percent of students in the late 1970s to 14 percent by the late 1990s. School
programs for special education students on average cost more than twice those for regular
education students. Similarly, there has been growth in students from non-English
speaking backgrounds, entailing other programmatic expenditures.
Third, other industries have been competing for college graduates – particularly
women – with increasing vigor. The salaries of college graduates have risen
dramatically, meaning that schools must pay more for teachers just to have a chance at
the same group of graduates that they did in the past.
Each of these latter factors does in fact contribute to the cost of school operations.
Nonetheless, by any reasonable accounting, these factors explain only a portion of the
increases in spending on schools. If these factors were removed from the calculations,
substantial growth in real school spending would remain.
The aggregate trends are also confirmed by a large amount of detailed statistical
study of what goes on in classrooms and schools. These studies, which attempt to
uncover the effects of differing amounts of resources on student achievement, provide no
evidence of a consistent or systematic impact of added resources. To be clear, some
studies suggest that resources help, others actually find that more resources student
achievement, but most provide no confidence that there is any relationship between
resources and student performance.
These results do not say that money never matters. Nor do they say that money
cannot matter. On the other hand, they indicate quite clearly that unless other more
fundamental aspects of schools change increasing resources cannot be expected be
expected to lead to noticeable improvements in student performance. Thus, these results
reinforce the previous trends, where increases in overall resources did not correlate with
changes in student outcomes.
In short, there is a clear need for reform. Desires to improve student outcomes in
the past have not been realized, even with substantial injections of resources into the
Issues in improving student performance
Three factors appear to be extraordinarily important in explaining the current situation.
First, there are no significant incentives within the system that push toward higher student
achievement. Second, a variety of vested interests operate to distort decisions. But,
third, and I think often neglected, people like their schools. I will quickly describe these
ideas and then move to their implications for reform.
Incentives. Even though we are interested in raising the achievement of students,
student performance has little or nothing to do with the pay, employment, or careers of
most people in schools. To be sure, many people have chosen teaching careers because
they want to help children, and they may take some real satisfaction from seeing students
learn. Yet, almost everybody is motivated by a complex set of factors. In a situation
where rewards are not aligned with the outcomes we are looking for, it should not be
surprising that student performance is little moved by the series of reforms, programs,
and resources that we have directed toward schools. Another aspect of the lack of
incentives relates directly to efficiency of schools. It is not possible to identify anything
that looks like direct incentives to conserve on resources or, more importantly, to use
Vested interests. Many of the people making decisions about school policy have a
direct interest in the policies themselves, regardless of their impact on student
performance. Thus, for example, teachers’ unions have interests in increasing their
membership – say through lowering class sizes – and these interests come independent of
any potential impact on student outcomes. Similarly, the teachers themselves find that
fewer students imply less work. Building administrators find that cooperative teachers
who volunteer for a variety of activities outside of the classroom make their jobs easier.
Such personal interests can in fact influence the kinds of policies and operational
decisions that are made.
People like their schools. Regular surveys of attitudes toward schools reveal that
everybody likes their own school. The median person thinks that his or her school gets a
grade of B+. But, interestingly, those same people think that schools other than their own
are, perhaps, graded at a C- or below. The disjuncture between the two is interesting and
informative. The grades for schools by a random sample of people should provide a
measure of the quality of schools in the country, i.e., of the schools other than each
person’s own school, but they do not. This is a classic case of Lake Wobegone, where