"Does School Quality Matter" Julian Betts

in a nutshell:

This paper has found that earnings of white male workers depend significantly on which high school they have attended. However, standard benchmarks of school quality explain very little of these differences between schools.

Does School Quality Matter? Evidence from the National Longitudinal Survey of Youth by Julian R. Betts
The Review of Economics and Statistics, Vol. 77, No. 2 (May, 1995), pp. 231-250

Abstract

The paper searches for links between school quality and subsequent earnings of students. Using data for white males from the NLSY, the paper rejects the hypothesis that workers' earnings are independent of which high school they attended. However, traditional measures of school "quality" such as class size, teachers' salaries and teachers' level of education fail to capture these differences. This result is robust to changes in specification and subsample. The paper contrasts the results with those of Card and Krueger (1992a), and speculates that structural changes may have weakened the link between traditional measures of school quality and student outcomes.


I. Introduction

A lengthy literature has attempted to de- termine which "inputs" to the schooling process affect student outcomes. In detailed surveys, Hanushek (1986, 1989) concludes that there is little evidence that standard measures of school quality have any effect on student performance.

The vast majority of work in this area, dating back to the widely read Coleman Report (Coleman et al., 1966), has used test scores to gauge students' achievements. However, the validity of test scores as a performance measure is questionable, given the low correlation typically found between test results and subsequent labor- market outcomes.

[snip]

Indeed, if earnings are the metric by which economists measure success in the labor market, it makes more sense to use wages or earnings to gauge the effectiveness of schooling.

[snip]

A second trait, shared by two-thirds of this literature, is the use of spending per pupil as the single measure of school quality. Murnane et al. (1991, p. 7) criticize this approach as "sterile," since it gives no indication of which of the many components of education spending should be increased to improve student performance

In summary, no published paper in this literature on school quality and earnings has yet measured school quality at the level of the school actually attended. In addition, most studies have measured "quality" in terms of spending.

[snip]

My findings indicate that while there are significant differences between the labor-market performance of students who attended different schools, these differences are not significantly related to standard measures of school quality. These results accord with the literature on school quality and test scores, as surveyed by Hanushek.


III. Results

It would be premature to search for the characteristics of schools which influence earnings without first determining whether earnings differ between workers who attended different high schools, ceteris paribus.

[snip]

[A]ll four regressions were repeated with the addition of father's and mother's years of education, a dummy variable indicating whether any family member possessed a library card when the person was 14 years old (as a measure of the intellectual environment of the family) and the imputed hourly wage of the father.'0 But in each of these regressions, even when these personal and family characteristics were added, the null hypothesis that schools have no effect on earnings was rejected...

[snip]

In summary, despite repeating the regression under many different specifications, and using various subsamples designed to eliminate poten- tial data problems, the same conclusion holds throughout: three commonly used measures of school quality--the teacher-pupil ratio, the relative salary of starting teachers and the percentage of teachers with Master's degrees or higher--in general bear no significant positive relation to the subsequent earnings of students.

[snip]

Overall, these results do not lend much strength to the notion that better high schools contribute to earnings by leading workers to acquire more education and/or training than they otherwise would have.

[snip]

However, as shown in table 5, subsequent earnings of students do appear to be positively and significantly related to the number of students enrolled at the school, although the elasticity at the means is only 0.044. This suggests that mildly increasing returns to scale may be at work. Table 5 also presents evidence that a school with a higher percentage of disadvantaged students is likely to produce graduates with lower earnings, at least for those students who achieve less than 12 years of education. (The interaction term between this variable and education is actually positive, so that students with more than 11.98 years of education are predicted to have higher wages as the percentage of disadvantaged students rises.) Similarly, the percentage of grade 10 students who drop out without completing grade 12 has a negative and significant effect on earnings, at least for those workers who obtain less than 13.4 years of education.2

[snip]

[T]he size of the school, as measured by enrollment, is related to earnings, and is a variable which policymakers can control. This is the sole measure of school "quality" which this study has found to be significantly related to students' subsequent earnings. [ed.: don't know how this relates to the "small school" movement...]


bureaucratization of the schools & school quality:

[I]ncreasing bureaucratization of public schools might have weakened the link between standard measures of school quality and educational outcomes. In a cross-sectional study using state-level data from 1984, Anderson, Shugart and Tollison (1991) find that states with larger educational bureaucracies tend to produce stu- dents with lower standardized test scores. In a 1987 survey of secondary school principals, fully 69% complained of administrative roadblocks caused by new State guidelines and requirements (National Center for Education Statistics, 1991, p. 92). The idea that increasing bureaucratization of the public schools reduces their effectiveness is gaining widespread acceptance.

[snip]

An implication of these hypotheses of "structural change" in the public school system is that standard school quality measures should still be significantly and positively related to wages of those who attended private and parochial schools. When the basic regression from table 1 was run on the subsample of white males who had attended such schools, the teacher-pupil ratio and the percentage of teachers with post- graduate degrees were indeed significantly and positively related to earnings. However, this intriguing result should be regarded with caution as it is based on 507 wage observations involving only 58 individuals.


growing the administration

Coming across this study today is a case of serendipity.

In the past five years, my tiny little school district (1848 student enrollment in K-12 projected for 2009-2010) has been massively bureaucratized. I've always been told that when you grow the administration school quality declines, but I'd never seen a study saying so.

Now I've got two.



Anderson, Gary M., William F. Shugart III and Robert D. Tollison, "Educational Achievement and the Cost of Bureaucracy," (pdf file) Journal of Economic Behavior and Organization 15 (1) (1991), 29-45.

Weintraub, Daniel M., "Wilson Calls for Schools to Set Own Rules," Los Angeles Times (Jan. 29 1993), Al and A26.