Review of Education Matters by Derek Neal - August, 2002 Education Matters is a collection of essays on the economics of education by Alan Krueger, the Bendheim Professor of Economics and Public Affairs at Princeton University. The introduction provides an overview of the 15 previously published papers contained in the book, and discusses how these papers evolved in the context of the literature. My comments focus almost exclusively on the six papers in sections II and III of the book. These sections deal with the relationship between school resources and student outcomes. I assume that these materials are of particular interest to readers of Education Next. I feel the book is a worthwhile investment for graduate students and researchers interested in the economics of education. Several of the papers play an important role in on-going debates over education policy, and I find it quite useful to have them all in one place on my book shelf. Krueger is among the most prominent academic researchers advocating significant reductions in class sizes as a possible strategy for improving public schools, and his 1999 Quarterly Journal of Economics paper on the Tennessee STAR experiment is one of his more influential papers. The STAR experiment ran from 1985 through 1989. It involved over 11,000 elementary school children in 80 Tennessee public schools. Each school assigned students to one of three types of classrooms. Small classes had a target enrollment of 13-17. Regular classes had a target enrollment of 22-25. A third sample targeted regular enrollment size, 22-25, but added a full-time teacher’s aide in the room. This paper is a noteworthy contribution to the existing literature on class size and achievement. The data come from the only large scale experiment in the United States involving random assignments of students to class sizes, and Krueger’s analyses of the data as well as his presentation of results are quite clear and thorough. Further, the results stand in sharp contrast to most of the results in non-experimental studies because they indicate large achievement gains from class size reduction. Two sets of results generate the key findings in the paper. He estimates a two-stage model that uses initial random assignment to generate a predicted class size variable, and then he regresses achievement on predicted class size given controls for individual student characteristics. He does this separately for samples that are defined by the current grade and the year the student entered the study. The results indicate significant gains from smaller class sizes. However, students who enter the program in kindergarten gain little after the first year in a small class. Results from an ordinary least squares model that includes controls for initial class assignment and cumulative years in a small class show sizeable jumps in achievement associated with the first year in a small class. However, given controls for the composition of students in classes, there is only weak evidence that additional years in a small class further enhance achievement.
The STAR program did not generate perfect experimental data. Attrition rates are high, and the experimental design is compromised on other dimensions as well. However, Krueger does a good job of explaining the patterns in the data and placing the results in context. Another set of papers with David Card explores the returns to school resources by focusing on individual earnings rather than student test scores. Analyses of the effect of school resources of future earnings must confront a difficult statistical problem. Wages for workers of a given skill and occupation vary among different geographic labor markets. Thus, levels of school spending may be correlated with wages for reasons that have nothing to do with individual skill levels. Card and Krueger adopt a interesting strategy in an attempt to circumvent this dilemma. They examine wage variation among persons who currently live in the same labor markets but restrict their attention to persons who live in a state other than their state of birth. They estimate rates of return to education that are specific to the birth states of migrants in a given birth cohort, and then in a second stage, examine the relationship between these rates of return and the levels of school resources devoted to cohorts born in a particular states. Chapter 8 is a 1992 Journal of Political Economy paper that focuses on results for men born between 1920 and 1949 using earnings data from the 1980 Census. They find that longer school terms, higher teacher salaries, and lower pupil/teacher ratios are
associated with higher future weekly earnings among adults. A subsequent paper by Heckman, Layne-Ferrar, and Todd (1996) performs several specification tests on the empirical model estimated by Card and Krueger, and several results raise doubts about the maintained assumptions in the Card and Krueger empirical model.1 While the Heckman et al analyses and the responses of Card and Krueger are too extensive to be reviewed here, I would like to note one particular item. Heckman et al show that the Card and Krueger results are driven primarily by the high estimated returns to graduating college among students who went to elementary and secondary schools in states with high resource levels. Estimated returns to finishing high school are not higher among workers from states with well-funded high schools. I do not know exactly why this pattern of correlations emerges, but it does raises questions about the correct interpretation of Card and Krueger’s results.2 Chapters 6 and 7 of Education Matters contain two survey papers by Card and Krueger. Both appeared in 1996, and both provide useful discussions of the related literature on school quality and earnings, the rational for their methodology, and the questions raised by Heckman et al. Chapter 6 appeared in the Journal of Economic Perspectives. It is less technical and provides a useful starting place for understanding the issues in the literature. In a 1992 Quarterly Journal of Economics paper, Chapter 10, Card and Krueger examine the relationship between black-white earnings differences and black-white differences in measured school quality. These analyses are particularly interesting because they cover time periods in which many southern states provided black schools with far fewer resources than white schools. The analyses for the QJE paper employ earnings data from the 1960, 1970, and 1980 census files as well as data on school inputs in 18 segregated states from 1915 to 1966. The authors piece together data on term length, teacher pay, and pupil-teacher ratios from several sources and examine whether or not black-white school resource differences among states and within states over time help explain variation in black-white wage differences among adults born in the same states. One set of results used the same methodology employed in the 1992 JPE paper, except the analyses involved only blacks born in Southern states who moved to specific metropolitan areas in the North. The results indicate that better funded schools are associated with higher returns from schooling. It is possible that these analyses are compromised by non-random patterns of migration. However, given the low rates of college completion among southern blacks over most of the sample period, I doubt that the earnings of college graduates contribute much to the results. These results should be driven by links between the funding of elementary and secondary schools and the returns to elementary and secondary schooling. As a whole, the paper provides considerable evidence that, in many southern states, the practice of separate and unequal schooling for blacks and whites during the first half of this century contributed significantly to the economic deprivation of southern blacks. Further, the overall improvement of black schooling relative to white schooling, even before the effective integration of schools in the 1960s, contributed to the economic progress of southern blacks. Chapter 11 provides more details about the history of black-white differences in measured school quality. This paper with Shari Wolkon and Michael Boozer provides considerable detail about developments after the 1954 Brown decision, and it is a nice compliment to chapter 10. 3
1
For example, the Card and Krueger model maintains the implicit assumption that any differences among birth states in the distributions of migrants among destination states are orthogonal to the unmeasured characteristics of the workers. 2 Card and Krueger note that better K-12 schooling should enhance educational attainment in a state across the board. Thus, migrants with only 12 years of schooling who come from states with good schools may be less able on unmeasured dimensions than similarly educated migrants from states with poorly funded schools. This argument may help explain the Heckman et al results. However, one would expect these same forces to create a similar pattern of bias in the estimated return to finishing college because college students from states with good schools should be less select than college students from other states.
3
There is a significant and interesting literature on the history of black-white differences in school quality. See Bond (1934), Margo (1990), Smith and Welch (1989), and Donohue, Heckman, and Todd (2002).
The final chapter is a review article written for a conference at the Federal Reserve Bank of New York and published in the banks’s Economic Policy Review. Krueger discusses the performance of public schools and argues that there is no immediate need for dramatic reforms. He writes, “My personal view is that policymakers should be risk-averse when it comes to changing public school systems. To alter the institutional structure of U.S. schools radically without sufficient evidence that the ‘reforms’ would be successful is to put our children at risk.....careful experimentation and evaluation should proceed on a limited basis before wide-scale institutional changes are introduced, such as vouchers, magnet schools and charter schools.”p 468. I have sympathy for Kruger’s position that small experiments, evaluation, and incremental implementation are the prudent course with respect to vouchers and related reforms.4 However, I feel that a similar argument can be made concerning class size reductions, and the book generally treats reducing class sizes as a straightforward reform within the existing system of public schooling. For example, at the end of his paper on the STAR experiment, Krueger argues that, given current per-pupil spending levels nationally and his estimates of the achievement gains from reducing class size by roughly seven students, such class size reductions may be expected to enhance future earnings by an amount greater than or equal in present value to the current costs of smaller classes. Krueger admits that his calculations are rough, and he acknowledges that it is hard to place a dollar value on the potential gains from reduced class sizes 5 , but I am equally concerned that we do not really know how much it would cost to fund a large class size reductions while holding teacher quality constant. I contend that a much more thorough analysis of the costs of class size reduction would nicely compliment Krueger’s work on the STAR experiment and other papers in Education Matters. Personal experience as a student, teacher, and parent as well as evidence from the existing literature on teacher quality effects convince me that good teaching requires skills that many people do not possess. At the federal level or even the level of a large state, significant across the board reductions in class size could not possibly be accomplished over a short time horizon without compromising teacher quality. Further, unless the long run supply of elementary school teachers is much more elastic than I imagine, a large expansion of the elementary school teaching force would likely require a reduction in average teacher quality and/or a significant increases in salaries. Finally, given pay setting practices in public schools, any salary increases for elementary schools teachers would likely be enjoyed by secondary teachers as well even if secondary class sizes remained unchanged.6 In sum, vouchers and charter schools are not the only risky reforms on the table. Wide-scale reductions in class sizes may be quite risky as well. Krueger’s research provides some of the most credible evidence that small classes yield higher achievement, but more work is needed to demonstrate that the benefits of smaller classes outweigh the likely costs.
4
My own work on the performance of private versus public schools, Grogger and Neal (2000), does not support that view that all public schools perform poorly compared to neighboring private schools. Affluent whites, especially those in suburbs, often appear to receive respectable performance from their public schools. However, the literature on private versus public school performance in cities is quite different. Minorities students, especially black students, in many large cities attend public schools that often fail miserably compared to Catholic and other private schools in similar neighborhoods, and many big city school districts are plagued by gross corruption and mismanagement. I agree that we cannot know ex ante that vouchers or other radical changes will work, however I contend that in some large cities, like Chicago, public schools perform at such low levels that the risks of radical reforms are relatively low. 5 A forthcoming book by Krueger and James Heckman contains more detailed calculations concerning the economic returns from raising student achievement. Carnerio, Heckman, and Manoli (2002) is an example. 6 See Hanushek (1999) for further discussion of this issue.
References Bond, Horace Mann, The Education of the Negro in the American Social Order (New York: Prentice Hall, 1934). Carneiro, P., Heckman, J. and Manoli, D. (2002). "Human capital policy," in James Heckman and Alan Krueger, Inequality in America: What Role for Human Capital Policies. Cambridge: MIT Press, forthcoming 2003. Donohue, John J. III, James J. Heckman, and Petra E. Todd, (2002) “The Schooling of Southern Blacks: The Roles of Legal Activism and Private Philanthropy, 1910-1960,” Quarterly Journal of Economics, 117, 225-268. Grogger, J., and D. Neal. (2000) "Further Evidence on the Effects of Catholic Secondary Schooling." in BrookingsWharton Papers on Urban Affairs edited by W.G. Gale and J.R. Pack, Brookings, Washington, DC Hanushek, Eric A., (1999) “Some Findings From an Independent Investigation of the Tennessee STAR Experiment and From Other Investigations of Class Size Effects,” Educational Evaluation and Policy Analysis, 21, 143-163. Heckman, J., A. Layne-Farrar, and P. Todd (1996), "Human Capital Pricing Equations with an Application to Estimating the Effect of School Quality on Earnings", Review of Economics and Statistics 78: 562-610. Margo, Robert, Race and Schooling in the South, 1880-1950: An Economic History (Chicago, University of Chicago Press, 1990). Smith, James P., and Finis Welch, (1989) “Black Economic Progress After Myrdal,” Journal of Economic Literature, 27, 519-564.