This paper reviews and interprets the literature on the effect of
school resources on students‘ eventual earnings and educational
attainment. In addition, new evidence is presented on the impact
of the great disparity in school resources between black and white
students in North and South Carolina that existed in the first half
of the 20th century, and the subsequent narrowing of these resource
disparities. Following birth cohorts over time, gaps in earnings
and educational attainment for blacks and whites in the Carolinas
tend to mirror the gaps in school resources.
education production function
This paper evaluates evidence on the effectiveness of elementary and secondary schooling in the
U.S. Contrary to popular perception, most standardized test scores have not declined in the last
quarter century, and the National Assessment of Educational Progress (NAEP) data show a
modest upward trend. In addition, school spending per pupil and NAEP scores exhibit a positive
correlation in the aggregate data. A review of the Tennessee Student-Teacher Achievement Ratio
(STAR) experiment indicates that students who were randomly assigned to smaller classes
performed better on standardized tests. Both the aggregate NAEP data and the STAR experiment
suggest that a 7 student reduction in class size is associated with a 5 to 8 percentile point increase
in test scores, on average. Evidence also suggests that students from low socioeconomic status
families learn just as much during the school year as students from high socioeconomic status
families, but students from low socioeconomic status families fall behind during the summer
months. Finally, the earnings of high school graduates relative to high school dropouts increased
substantially between 1979 and 1996, suggesting that high schools provide skills that are valued
by the labor market.
Economists attempting to explain the widening of the black-white wage gap in the late 1970's by
differences in school quality have been faced the problem that recent data reveal virtually no gap in the
quality of schools attended by blacks and whites using a variety of measures. In this paper, we re-
examine racial differences in school quality. We begin by considering the effects of using the pupil-
teacher ratio, rather than the school's average class size, in an education production function since the
pupil-teacher ratio is a rough proxy, at best. Second, we consider the importance of using actual class
size rather than school-level measures of class size.
We find that while the pupil-teacher ratio and average class size are correlated, the pupil-teacher
ratio is systematically less than or equal to the average class size. Mathematically, part of the difference
is due to the intraschool allocation of teachers to classes. As a result, while the pupil-teacher ratio
suggests no black-white differences in class size, measures of the school's average class size suggest that
blacks are in larger classes. Further, the two measures result in differing estimates of the importance of
class size in an education production function. We also conclude that school level measures may obscure
important within-school variation in class size due to the small class sizes for compensatory education.
Since black students are more likely to be assigned to compensatory education classes, a kind of
aggregation bias results. We find that not only are blacks in schools with larger average class sizes, but
they are also in larger classes within schools, conditional on class type. The intraschool class size patterns
suggest that using within-school variation in education production functions is not a perfect solution to
aggregation problems because of non-random assignment of students to classes of differing sizes.
However, once the selection problem has been addressed, it appears that smaller classes at the eighth
grade lead to larger test score gains from eighth to tenth grade and that differences in class size can
explain approximately 15 percent of the black-white difference in educational achievement.
This paper analyzes data on 11,600 students and their teachers who were randomly assigned to
different size classes from kindergarten through third grade. Statistical methods are used to
adjust for non-random attrition and transitions between classes. The main conclusions are: (1)
on average, performance on standardized tests increases by 4 percentile points the first year
students attend small classes; (2) the test score advantage of students in small classes expands by
about one percentile point per year in subsequent years; (3) teacher aides and measured teacher
characteristics have little effect; (4) class size has a larger effect for minority students and those
on free lunch; (5) Hawthorne effects were unlikely.
This paper examines evidence on the effect of class size on student achievement. First, it is
shown that results of quantitative summaries of the literature, such as Hanushek (1997), depend
critically on whether studies are accorded equal weight. Hanushek summarizes 277 estimates
extracted from 59 published studies, and weights all estimates equally, which implicitly places
more weight on some studies than others. A small number of studies, which often present
estimates for several small subsamples of a larger sample, account for more than half of the
estimates. Studies from which relatively many estimates were extracted tend to find negative
effects of school resources, whereas the majority of studies from which relatively few estimates
were extracted tend to find positive effects. When all studies in Hanushek’s literature summary
are given equal weight, resources are systematically related to student achievement. In addition,
when studies are assigned weights in proportion to the “impact factor” of the journal in which
they were published -- a crude measure of journal quality -- class size is systematically related to
achievement. When studies are given weights in proportion to their number of estimates,
however, resources and achievement are not systematically related. It is argued that assigning
equal weights to studies, or weights according to quality, is preferable to assigning weights
according to the number of estimates extracted from the studies, because study quality is unlikely
to be related to the number of estimates taken from the study, and because researcher discretion
in selecting estimates is limited when studies are assigned equal weight.
Second, a cost-benefit analysis of class size reduction is performed. Results of the Tennessee
STAR class-size experiment suggest that the internal rate of return from reducing class size from
22 to 15 students is around 6 percent.