The (NCES) commissioned the papers in this publication to address certain perplexing questions in education finance. Earlier papers in this NCES series focused on the nation's education finance information needs and concerned emerging education finance topics that posed statistical and measurement problems for the profession. This publication extends that tradition by first examining two policy related issues, whether money matters in education and the effect of state constitutional litigation. Certainly those involved with education policy have struggled to understand how money matters in education and must be astonished by educational research that finds no strong or consistent relationship between the two. Similarly, with more than 40 states having had the constitutionality of their state education funding systems challenged, and with a plaintiff success rate of about 50 percent, those involved with education policy must wonder if it is worth challenging a state funding system through the courts.
Additional papers explore three statistical and measurement problems that NCES has encountered. The first is how to measure resources at the student level rather than at the school or school district level, which is the custom today. NCES data bases on a sample of the nation's students contain impressive information about the education of the student, with the exception of the resources devoted to that student. Another statistical and measurement problem is what constitutes "good practice" when conducting education finance research using NCES data bases. Researchers tend to be idiosyncratic in their approach to preparing education finance data bases for analysis, but might a common approach be appropriate? The third, and most ambitious paper, struggles with a suggested research agenda for measuring educational adequacy. To obtain these papers, NCES turned to distinguished education finance researchers. NCES asked that they turn their knowledge, experience, and insights toward examining these issues, and to put their thoughts in publishable form.
Perhaps no more perplexing question arises than that of whether money matters in education. The first paper in this publication, by Lawrence O. Picus of the University of Southern California, explores the troubling finding that "there is no strong or systematic relationship between school expenditures and student performance" (Hanushek, 1989) and the more recent work that has questioned that conclusion (Hedges, Laine and Greenwald, 1994). Rather than focusing on the complex statistical analyses and econometric production-function approaches, Picus reviews existing studies that attempt to ask the question: "Does Money Matter?" and considers alternative approaches to the question of whether or not resources might enhance achievement. First, Picus briefly presents the overall pattern of expenditures for elementary and secondary education during the last century, along with evidence on how student achievement has changed over time. He then turns to the debate between Hanushek and Hedges, et al. Finally, he offers some conclusions and policy recommendations.
Few policy analysts realize that, according to Hanushek, expenditures on education have grown faster than spending for health care, and represent some 3.6 percent of the Gross National Product (GNP) in 1990 (compared to 1 percent a century ago). This money has gone predominantly to hire more teachers, pay higher teacher salaries, and lower class sizes. Although it is often thought that the class sizes are the product of educating children with disabilities, Hanushek and Rivkin suggest only one-third of the recent decline in class sizes is accounted for by special education programs. Where has the other money gone? Hanushek suggests it has been used to pay teachers increased benefits, such as teacher retirement, unemployment compensation, social security contributions, and group insurance, such as health insurance.
As Picus points out, school districts spend an average of 60 percent of their current funds on instruction, and they show little variation in that percent. As many have observed, the substantial increase in education funds cited by Hanushek has not been matched by higher student outcomes on the average Scholastic Assessment Test (SAT) or on the National Assessment of Educational Progress (NAEP). The evidence for graduates to find and keep good, high paying jobs, however, is more encouraging. Card and Kruger (1990) found that students with small classes and higher teacher salaries had higher earnings in their adult years.
Picus begins his review of production-function studies of whether or not money matters in improving education by pointing out that students' socioeconomic status (such as their parents' education, occupation, and income) may be more important in determining how well they do in school than are many of the kinds of inputs considered in these econometric studies. Typical inputs that are examined for their relationship to student achievement include per-pupil expenditures, pupil/teacher ratios, teacher education, experience and salary, school facilities, and administrative arrangements.
The debate between Hanushek and Hedges, et al., is whether or not Hanushek should have included all the studies which had insignificant results, or for which the direction of the effect could not be determined. Even when Hedges, et al., exclude the studies which they believe to be inappropriate, they conclude that while money might matter, class size and teacher characteristics may not. Hanushek then asks, if this is their finding, what factors do in fact matter, since teachers' salaries and class size are the two largest determinants of spending. In a recent answer to Hanushek's question, Ferguson (1991) conducted research in Texas that leads him to conclude that hiring more and better teachers does lead to higher student test scores.
Picus concludes by observing that what we don't know is what the impact on student performance would be if schools or school districts were to dramatically change the way they spend the resources available to them. Undoubtedly, this debate will continue, although as Picus suggests, it may be that the wrong question is being asked. Perhaps instead the question should be "How is Money Used in Education?"
The second paper, by G. Alan Hickrod, Distinguished Professor Emeritus at Illinois State University, examines the effect of constitutional litigation on educational finance. Some 44 states have experienced such litigation challenging their state aid to education systems. Hickrod reviews six empirical studies of either the effect of constitutional challenges, or studies that provide some data for the investigation of that topic. Hickrod (1992) studied the 1970-90 time period, and concluded that winning litigation in a state supreme court saw only a modest increase in per-pupil expenditures, but shifted the tax burden from state to local resources. Winning did seem to reduce disparities between school districts in spending per pupil. However, Heise (1995) did not find an increase in per pupil expenditures following the plaintiffs' successes in either Wyoming or Connecticut. Peternick (1995) studied 12 states where the state supreme court had found the state education aid system unconstitutional, and concluded that these decisions did stimulate growth in per-pupil expenditure. These studies differ in that Heise/Peternick assume some single, sharp increase, while Hickrod believes that the state legislature is faced with a winning court decision, compliance litigation, and increasing pressure to modify the state aid system in a way that is beneficial for the successful plaintiff school districts.
What makes the research so difficult is that plaintiff challenges might be characterized in different ways, such as "a clear win," or "a loss," followed by yet another filing, under yet another legal theory. In other states, litigation has been filed, but no hearings have been held or decisions rendered. Assigning "climate" to this body of litigation is not precise. Hickrod examines whether or not there might be the effect that if a state chooses to reduce disparity in spending between school districts (an equity goal), that perhaps the overall funding level of the state will suffer (an adequacy goal). He finds no significant relationship between adequacy and equity goals. He also finds that the data support the notion that the state supreme court decision in favor of the plaintiff school districts reduces expenditure disparity between school districts within states, even if they are not related to increases in average per-pupil expenditures in those states.
Hickrod concludes by examining anecdotal evidence from 11 states. He also addresses some unanswered questions. While Hickrod asserts that constitutional litigation is one factor in explaining the changes in state funding for education, it is not the only factor. However, he confirms that it is worth challenging the funding of K-12 education via the courts, albeit that the legal process can be quite expensive and very time-consuming. In addition, chances of losing the case are quite high, as only 17 states have successfully defended their funding system. Hickrod correctly observes that public education is like absolutely no other public service, in that it has its own article in every single state constitution in this nation.
The remainder of the papers explore statistical and measurement problems that NCES has confronted. A perplexing dilemma faced by NCES is how to measure student-level resources, perhaps to accompany NCES student-level education and personal characteristics data, when most financial data reside at the school-district level. Robert Berne is Vice President for Academic Development at New York University, and his coauthor, Leanna Stiefel , is professor of economics at the Robert F. Wagner Graduate School of Public Service at NYU. They are perhaps best known for their classic work in The Measurement of Equity in School Finance (1984).
The purpose of the Berne and Stiefel paper is to discuss the types of student-level resource measures that would be preferred if student-level resource data were collected on a routine basis. They begin by exploring the types of questions that could be answered with student resource data. They then review selected literature to show how some education finance researchers have attempted to answer such questions with extant data. They also introduce some cost accounting concepts that are useful in thinking about how to collect appropriate student-level resource data. They conclude by recommending alternatives for NCES to consider.
Berne and Stiefel believe there are three questions that student resource data could answer. The first has been termed production function questions, that is, questions of resource effectiveness and cost-effectiveness. Examples are: Do additional resources for children lead to additional outcomes? and What is the cost effectiveness of one program versus another? The second type of question that might be answered with these student resource data are equity questions. One examples is: Whether poor students receive the same (or greater) resources than other students in a school, or an educational program, such as preschool education. A third type of question is termed a resource intent question. If a student is entitled to specific resources, such as handicapped, bilingual, or compensatory education, do they receive the additional resources which were intended to be assigned to them?
Using illustrative studies, Berne and Stiefel demonstrate that "...even when researchers are careful to put together a comprehensive and unique data set, they cannot always obtain resource variables at the correct unit of analysis." The reason for this is that these studies often use administrative data because it is available, rather than because they prefer such data. In addition, data at the student level are preferable in production-function studies to data disaggregated from the school district level. This is particularly true because student outcomes might be sensitive to resources not currently available in administrative records at the school district level.
Examining equity studies, Berne and Stiefel conclude that data problems immediately emerge with administrative data from different sources, merged together for an equity study. For example, pupil counts of students in programs may not match students funded. Expenditures for employee benefits, transportation, school lunches, or utilities may be omitted.
Berne and Stiefel include four resource intent studies in their review. These employ cost accounting methods which are expensive, and use a bottoms-up approach that identifies the student and program of interest, and then assigns expenditures to a child in that program. The cost accounting method is valuable because it emphasizes the need to conceptualize the use of the expenditure data before the data are collected.
Turning to concepts of costing, Berne and Stiefel explain the distinction between departmental and product costing methods where distinctions are made between how full costs can be subdivided, how expenditures can be allocated to students, and the difference between "costs" and "expenditures." Departmental costing finds the costs of administrative units, such as school districts, departments, responsibility centers, etc. The primary purpose of departmental costing is to help managers administer units efficiently. Product costing, in contrast, finds the costs of producing various kinds of products. Berne and Stiefel argue that product costing is relevant because this is where most of the questions to be answered with the resource data are focused.
Berne and Stiefel also argue that these three questions require resources linked to students, rather than schools or school districts. Products are assigned their full costs if the total of all products equals the organization's total costs. Direct costs are those that can be assigned uniquely to one product. Indirect costs, on the other hand, are incurred because of the production of many products, and are frequently considered overhead costs. When students are the product, instructional supplies are an example of direct costs, while the principal's time is an indirect cost. Two systems are generally used for assigning direct costs to products. Job-order costing determines the costs of each individual unit of a product, for example, the cost of fine, hand-made, custom-ordered furniture. Process costing determines the costs of groups of identical units and then divides the number of units to obtain an average cost. An example might be the cost of a box of breakfast cereal. Job-order costing is more accurate, but much more difficult and expensive to collect. For education, most likely a mixed model would be used, such as assigning resource costs to types of students, rather than individual job order costing.
Berne and Stiefel conclude with recommendations, framed by the purpose for which the student-level resource measure will be used. They believe the student-level resource measure should be developed to assess the effectiveness of resource use, requiring the linking of resources with student outcomes. They argue that NCES should proceed to develop a student-level resource measure, regardless of whether the data are actually collected. They assert a student-based system (product costing) will be required. They also urge NCES to collect full costs, direct versus indirect costs, at different educational programs. The measure must include the sources of funding, such as local, state, or federal. Finally, for those costs that must be allocated, guidelines must be developed for the basis and method of allocation. This student-level resource measure, obtained on a sample of students (like many existing NCES student-level data sets), would permit NCES to begin to make progress on the question of effective use of resources in education.
The next paper on statistical and measurement problems that NCES has contended with is by Michael O'Leary of Coopers & Lybrand and Jay Moskowitz of Pelavin Research Institute. They explore some of the steps that were required to be taken before conducting various finance studies from NCES data sets. Although some of the problems they cite from the work that was undertaken in 1994 have been corrected in later editions of the NCES Common Core of Data (CCD) CD-ROM, the procedures they recommend are always appropriate to ensure accurate results. Although the authors acknowledge that individual research methods will (and should) always vary, if the data bases are constructed with the same sets of underlying procedures, researchers' conclusions can be debated on their merits, rather than on differences in underlying data sets. O'Leary and Moskowitz begin by pointing out that the school district finance collection undertaken by the U.S. Bureau of the Census typically contains a universe of more than 16,000 school districts in years ending in 2 or 7 (1992, 1997). In other years, the data are collected only for school districts larger than 15,000 students, or if the school district is fiscally dependent upon a county or city.
The number of school districts not processed by Census also varies by the year of the sample. NCES also requested that Census collect and process a universe of school districts in 1990, to accompany another NCES project, mapping decennial census demographic information (such as median income) to school district boundaries.
O'Leary and Moskowitz also explain that the NCES Common Core of Data (CCD) CD-ROM contains data at three different levels (state, school district, and school) from five different NCES surveys. The CCD CD-ROM typically contains five years of data for each of these levels. As a result of this complex structure, O'Leary and Moskowitz find three issues arise: sampling issues; school district types, and school district levels. Education finance researchers have sometimes spent considerable time and resources exploring why revenue data did not exist for many school districts over a decade. Other researchers weighted non-missing districts, or imputed values for missing districts, although the effect on the research results is unknown. There are also a variety of "special" school districts, including community colleges, vocational-technical school districts, correctional/custodial school districts, special education, and non-operating school districts. Non-operating districts have students but no buildings, and typically transfer their resident children to other school districts to be educated. These districts are typically not included in equity and education finance analyses. O'Leary and Moskowitz also explain that there are four basic "levels" of districts in the Census F-33 collection: elementary only; secondary only; unified K-12, and college-graded. Researchers are urged not to only deal with a single type, such as unified, because some states do not have this type of school district organization, and others have "mixed" models. Different researchers have approached this differently, although virtually everyone pupil-weights [see the classic work The Measurement of Equity in School Finance (Berne and Stiefel, 1984).]
O'Leary and Moskowitz also discuss unresolved data base creation issues, such as enrollment; special-needs pupils; property, poverty, and income data; and state and district finances. Although NCES and Census enrollment counts are now likely to be the same, it is still impossible to separate funds targeted for special-needs pupils from those allocated through basic state education aid programs to school districts. Regarding district finances, NCES has resolved "on-behalf' funds by imputation for current years, and the Governmental Accounting Standards Board (GASB) has included the collection and reporting of these numbers at the school district level, beginning in 1995. However, "outliers," erroneous and extreme values of per-pupil expenditures or revenues, will always plague data reporters such as NCES, especially when reporting values for more than 16,000 school districts. O'Leary and Moskowitz conclude their discussion with a model of "suggested best practice" when creating an analysis data set from NCES education finance data.
The last paper to explore statistical and measurement problems encountered by NCES is a suggested research agenda for exploring educational adequacy, by William H. Clune of the University of Wisconsin-Madison. Clune explains that the goal of educational adequacy, defined as high minimum educational outcomes for the disadvantaged, is constrained by the lack of a strong knowledge base, and that a well-formulated research agenda could make an important contribution to education policy. Clune argues that while better student educational outcomes have never been more highly valued, the link between these outcomes and educational reform and investments is widely doubted. This has the most poignant effect upon education for the disadvantaged, where poor children compose the majority of students with poor educational outcomes. Under such doubts, strains are developing in the traditional goal of equity in education finance. Traditional equity arguments guarantees equitable inputs regardless of outcomes. Now cost-effectiveness and productivity goals intrude. Clune argues that if society and the courts become satisfied that high minimum educational outcomes can be produced at minimum cost, then this strain will be minimized, and additional funds will become available.
Adequacy can only be assessed, Clune argues, when it refers to some outcome. Levin (1991) demonstrated that compensatory spending could be based on the benefits of good education and the social costs of a poor education. Unfortunately, it is difficult to be certain that compensatory spending in the present will reduce crime costs in the future, particularly for today's taxpayers. Fortunately, Clune asserts, standard setting for education is well established, providing some notion of educational adequacy. Turning to the available resources to establish adequacy, Clune explains that educational spending has grown faster than the cost of living for many decades, and maintains its relative position even in the midst of calls for austerity. Very few call for the outright repeal of educational subsidies, or the rollback of universal high school education. Clune remains optimistic "...that a substantial amount of extra money could be found for adequacy."
Clune is skeptical about disagreement in measuring adequacy, although courts frequently develop long lists of desirable outcomes for education (Rose v. Council for Better Education, Inc., 790 s. w. Ken. 2d 186, 212-13 [1989]; and Kentucky and California student tests result in many students unable to attain "proficient" scores (Kirst et al., 1995). He believes that there is more theoretical than practical confusion, as a simple working definition of adequacy includes literacy, numeracy and problem solving, and completion of high school in a manner sufficient to be eligible for further education. Clune then suggests that "high minimum achievement" might be defined as the attainment of average achievement by disadvantaged children. This high minimum achievement would be equal to the achievement of the average child in the nation. Policymakers, he argues, would benefit from a fleshing out of the various operational definitions of adequacy.
One problem that Clune touches on is the problem of educational adequacy in high-poverty schools. He asserts that there is still confusion regarding whether children that have not reached educational adequacy may, "within any reasonable range of resources" achieve high minimum performance. Although Slavin and his colleagues (Slavin, 1994), claim to be able to raise the reading achievement of elementary students to average grade levels, no data on costs accompany this claim. Clune argues that the maintaining of an evaluation data base (including, presumably, costs) for a group of "accelerated schools" would be " well worth the investment."
The basic elements of successful accelerated education have been known for some time, Clune asserts, and they are an accelerated curriculum, high expectations for student learning, a positive school climate, and a safe and orderly environment. The ultimate goal of educational adequacy is accelerated education, for if poor children learned at the same rate as other children, they would not be educationally disadvantaged. Thus Clune adds to the evaluation data base the element of educational process, or teaching technology.
We need a fine-grained understanding of what staff and students do differently in order to begin understanding the conditions under which such desirable behaviors can flourish. Careful investigation of schools adopting models of accelerated education are, therefore, a promising direction for research.
The New Jersey Supreme Court reasoned that a good estimate of the cost of an adequate education is the resources available to the state's most successful students, those in the highest socioeconomic school districts in the state. This highlights that the cost of an adequate education is not known. Assume that the additional costs to educate a child in poverty are $2,000, the question becomes, $2,000 in addition to what? As is apparent, future education finance research must conduct the types of studies of accelerated schools that could flesh out the costs for remediating a disadvantaged child. Clune also asserts that research is needed on school aid formulas, in order to provide sufficient "base" funding and to fully cover the additional costs of remediation. Clune also believes research is needed about the type of governance structures capable of tolerating and encouraging the reforms that will bring about educational adequacy for students in poverty.
Clune concludes by listing the ten research questions that need to be addressed by a research agenda focused on educational adequacy. Much of the agenda depends on gathering more data from existing accelerated schools, which, unfortunately, are not very common in the nation. Another strategy would be to turn to useful secondary data sources about educating students in poverty. Not all of this research needs be undertaken by government. Foundations and the voluntary cooperation of research scholars and research centers might undertake such challenging educational research.
Summary
The papers published here are but a single component of the continuing efforts of NCES to obtain and provide education finance data of interest and utility to the school finance community. The first two papers suggest that the debate regarding the efficacy of money for education will continue, and that NCES will be called upon to provide even more financial data that can inform the discussion. A crucial component of the "Does Money Matter?" argument revolves around "good practice" in constructing the data base for analysis. In addition, the controversy makes the need for a student-level resource measure even more critical. The courts, as well as the education finance community, are likely to attend not only to the efficacy of money, and the effectiveness of state constitutional litigation, but also to the development of a standard of educational adequacy.
The previous publication, Selected Papers in School Finance, 1994 , asserted that NCES also wishes to make known conceptual and methodological advances in the field of education finance, for researchers and students alike to emulate, replicate, disseminate, and enhance. That previous volume described the evolution of this series of papers, and other NCES products, such as the CCD CD-ROM, which have provided education finance data in a readily accessible form for all school districts in the nation. At that time NCES had also just started work on a geographic cost adjustment (Chambers and Fowler, 1995), and will soon release an even more sophisticated geographic and inflation adjustment (Chambers and Fowler, forthcoming).
New Developments
The education finance reporting needs of Congress, the Department, states, and the education finance research community are constantly shifting and expanding, as are financial reporting standards, rendering NCES surveys and reports outdated, although they were previously thought to be satisfactory for at least a future decade. This turbulent environmental press has created the need for NCES to have the capability to proactively undertake "developmental analysis."
The Education Finance Statistics Center is a specifically-designed Education Statistical Services Institute (ESSI) component to carry out the NCES need for developmental analysis to increase NCES' knowledge, capacity, data collection, and reporting in education finance. Concomitant with developmental analysis, the EFSC might share advances with states facing similar finance dilemmas, offering technical assistance to improve their fiscal reporting systems.
The EFSC is currently conducting three developmental activities: assessing how accounting changes effect NCES surveys and financial reporting; devising experimental measures in education finance, such as measuring the cost of educating students with special needs; and assessing the need for technical assistance to states. A small portion of its budget is to administratively carry out these three developmental activities.
The activities started by the EFSC, while addressing some of the immediate definitional, measurement, collection and reporting dilemmas and policy choices currently faced by NCES, do not capture the future developmental challenges in education finance. This requires proactively anticipating the next threshold in education finance: where the profession seems to be headed and what questions future public and policy demands will create.
Future EFSC activities need to revolve around developmental challenges in education finance, attaining an increase in theoretical knowledge, analytic capacity, data collection frequency/methodology, and financial reporting. For NCES to attain the next threshold in education finance, and to retain preeminence in the field, definitional, measurement, collection and reporting dilemmas will have to be explored through developmental analysis by the EFSC.
Among the projects NCES anticipates exploring in the next year are:
1. Anticipating changes in accounting standards
NCES fiscal surveys will be affected by changes in accounting standards promulgated by the Financial Accounting Standards Board (FASB), which changes the IPEDS postsecondary finance collection and the planned K-12 private school finance collection. Both the state-level National Public Education Financial Survey (NPEFS) and the school-district-level F-33 collection will be influenced when the Governmental Accounting Standards Board (GASB) acts at the beginning of 1997.
2. The acquisition, analysis, and reporting of school-level data
The dilemma is how to conduct nation-wide school-level analyses that will enable NCES to report on administrative overhead and program cost, either by collecting school-level finance data from state administrative records in those states that have such information or amending the NCES Schools and Staffing Survey (SASS) to contain resource allocation information.
3. Experimental measures of program- matic expenditures and costs
There is no agreement on how to obtain and measure programmatic expenditures in education. Programmatic expenditures are the expenditures for a program, such as compensatory education, bilingual education, or handicapped education. Expenditures do not necessarily represent the actual cost of an educational program- ways to explore differences between cost and expenditure need to be developed.
4. Using geographic and inflationary cost adjustments
NCES is in the process of developing both geographic and inflationary cost adjustments. However, how these indexes should be applied to education finance data requires further conceptual development. This project will examine whether the Digest of Education Statistics and the Condition of Education might use the geographic and inflationary cost adjustments in reporting financial data.
NCES has established its own home page at / and the EDFIN also has a home page on education finance at //edfin. From these sites, individuals can obtain products and publications, such as CD-ROMs or geographic and inflation cost adjustments; request specific data sets or download them; review frequently asked questions about the NCES education finance program; and email NCES staff (see graphic of EFSC home page).
NCES hopes these developments will further assist the education finance community in its research and data needs.
[First Paper]
