|View Quarterly by:
This Issue | Volume and Issue | Topics
|This article was originally published as the Executive Summary of the Research and Development Report of the same name.|
Research and Development Reports are intended to
Most people intuitively recognize geographic differences in costs and in measuring inflation. Efforts to compare the costs of exactly the same things in different geographic regions involve comparisons of the same market basket of goods in two geographic areas. The difference in the prices of the same market basket of goods is designed to reveal the differences in the cost of living in different geographic regions. Measuring cost differences in education, however, is difficult, since most of the costs are in personnel, rather than in supplies. This report attempts to explain the differences between education costs and expenditures, explain the differences in the unit price of teachers in different regions and differences over time in the level of inflation, examine existing indices that can be used to make judgments for these differences in costs, and outline a future plan of action to derive a precise, stable, and accurate index for school administrators and policymakers to use.
The cost of education can be defined as the minimum of what must be given up to accomplish some result. Expenditure is different from cost in that expenditures are not tied to results or outcomes and can exceed the minimum of what must be given up.
Education costs can be organized according to an allocation hierarchy where the lowest level is the unit cost of various inputs like teachers time, space, and supplies. At the next level, there are costs that occur as the individual inputs are combined to form education services within classrooms and schools. Finally, at the uppermost level are the actual outcomes of schooling, where costs occur because of the presence of students with specialized needs of various kinds. Resource allocation decisions are made at each of these levels, and it is useful to keep them distinct because this can allow us to determine the relative magnitude of each source of cost.
The purpose of a geographically based teacher price index is to determine the relative cost of engaging the services of comparable teachers. Some of the necessary components include teacher characteristics (level of experience, training, minority status, and gender), cost-of-living adjustments, regional amenities, employment amenities, nonteaching wages and employment opportunities in the region, union and collective bargaining, and demand for teacher quality. Several scholars have attempted to define a geographically based index. The Teacher Attribute Model is the result of Stephen Barros (1994) approach. Barro did not strive to include all of the components outlined above in order to minimize the number of assumptions based on incomplete data. His estimate focuses on interstate comparisons and estimates what each states average teachers salary would be if the state employed teachers with the same average experience and training as that found in the nation as a whole.
Another approach has been characterized by McMahon and Chang (1991) as the market-basket approach. This approach does not focus on school personnel but rather on costs that are outside of the schools control, such as wages in other sectors of the economy and geographically based differences in the cost of living. One reason for this focus is to prevent a feedback loop rewarding schools that increase salaries. The basic components of this model include the value of housing, per capita income, the percent change in population for the preceding decade, and variables representing regions of the country. It can generate cost-of-living indices at several levels of aggregation.
The hedonic model (Chambers 1998) is a more ambitious approach that deals explicitly with each of the influences addressed by the models discussed above. The model is called hedonic because it is sensitive to whatever it is that teachers find attractive or unattractive about a given career opportunity. The Teacher Cost Index (TCI) (Chambers and Fowler 1995) is an example of this approach. Using Schools and Staffing Survey (SASS) data, it includes teacher characteristics (ethnicity, gender, education, and experience), working conditions (class size), and salary information. Other data sources, such as FBI crime statistics and U.S. Weather Bureau climate statistics, were used to assess regional amenities. Cost influences that the school has control over were statistically controlled while other influences were allowed to vary. The Geographic Cost-of-Education Index (Chambers 1998) is a more recent application of this approach. In this model, the index was broadened to include other types of inputs (school administrators and noncertified school personnel), and the range of data sources was widened. Both approaches run the risk of relying too much on potentially questionable data sources and assumptions.
The production function (PF) models are perhaps the most ambitious because they focus on the costs associated with actually realizing gains in educational performance. Unfortunately, a lack of adequate data and complete theoretical specifications for these PF models have hindered widespread practical use. However, in recent years these models have been applied to several states. For an example, see the application to New York in Duncombe, Ruggiero, and Yinger (1996). There also have been applications to Wisconsin and Texas.
A comparison of the three main models for geographically based cost adjustments (the Teacher Attribute, market-basket, and TCI models) demonstrates that the indices are highly correlated, at over .70. Also, the more adjustments that are made, the more the degree of variation drops. Despite the high correspondence between these indices, there are certain geographic regions where there is disagreement between the indices. A comparison between the hedonic and cost-of-living (market-basket) models might indicate, for example, that this discrepancy is due to the regions attractiveness (such as San Francisco) or unattractiveness (such as nonmetropolitan Connecticut) to most teachers.
Adjusting for regional cost-of-living differences is only one of the challenges to producing a cost-of-education index. The other major challenge involves adjusting for cost-of-living differences over time. Different deflators can lead researchers to different conclusions.
The most common way of measuring inflation is the market-basket approach used by the Consumer Price Index (CPI), where the cost of commonly purchased items is tracked over time. The School Price Index (Halstead 1998) is one example of this approach that uses the Urban component of the CPI, the CPI-U. Unfortunately, this index can only be used at the national level. There are many problems with applying the CPI approach to education, especially the change of relevant products over time (item substitution) and the uneven growth of inflation for different occupational areas. Education is one of those occupations that has been strongly influenced by changes in technology. This makes it difficult to track inflation since the supplies bought today (such as the computer or VCR) are not really comparable to the supplies of a few decades ago (such as the typewriter or projector). The second problem is that some occupational areas (such as medicine) have seen strong inflation, while other areas have not. Rothstein and Mishel (1997) argue that factors such as the increase in quality due to smaller teacher/student ratios have made inflation greater for education. Their solution is to use the Net Services Index (NSI), which measures inflation by focusing on labor-intensive components of the CPI similar to education. However, they acknowledge that while the NSI is an improvement, it still produces an underestimate.
A second approach, the Inflationary Cost-of-Education Index, modifies the hedonic TCI to include school administrators and noncertified staff. However, given data limitations, this only provides a 6-year inflation index during the years SASS was administered.
The Employment Cost Index also avoids the market-basket approach by measuring the rate of change in employee compensation, which includes wages, salaries, and employers costs for employees benefits. It covers all occupations with the exception of federal government workers and is used extensively by the Federal Reserve Board as a measure of inflation. It has an education subscale and has separate data on salaries as well as fringe benefits. Of all of the indices, this one is the most attractive because it avoids the pitfalls of item substitution found in the market-basket approach and has a large time frame (1981 to 1996) available.
Both geographic and inflation cost adjustments suffer from many flaws. Overall, there is correspondence between different geographic indices; however, for a particular geographic area the results can be dramatically different. Given the political nature of these adjustments, such discrepancies can be as problematic as they are informative. While the addition of more adjustments leads to a reduction of variability and arguably greater accuracy, policymakers reluctance to use adjustments is understandable.
There are two primary goals for the future of geographic cost adjustments: improve the indices of cost variations and educate the public and policymakers about any progress that is made. The basic challenges are to make the indices generalizable across different levels (local, state, and regional), separate and distinguish influences that are controllable by the school, be careful of double counting when adding new adjustments, and address any political considerations.
The following are some guiding principles for policymakers to consider as they seek to take advantage of what has been learned about variations in the costs of education:
Barro, S.M. (1994). Cost-of-Education Differentials Across the States (NCES 9405). U.S. Department of Education. Washington, DC: National Center for Education Statistics Working Paper.
Chambers, J.G. (1998). Geographic Variations in the Prices of Public School Inputs (NCES 9804). U.S. Department of Education. Washington, DC: National Center for Education Statistics Working Paper.
Chambers, J., and Fowler, W.J., Jr. (1995). Public School Teacher Cost Differences Across the United States (NCES 95758). U.S. Department of Education. Washington, DC: National Center for Education Statistics.
Duncombe, W., Ruggiero, J., and Yinger, J. (1996). Alternative Approaches to Measuring the Cost of Education. In Helen F. Ladd (Ed.), Holding Schools Accountable . Washington, DC: The Brookings Institution.
Halstead, D.K. (1998). Inflation Measures for Schools, Colleges, and Libraries. Washington, DC: Research Associates of Washington.
McMahon, W.W., and Chang, S. (1991). Geographical Cost-of-Living Differences: Interstate and Intra-state, Update 1991 (McArthur/Spencer Series, No. 20). Normal, IL: Center for the Study of Educational Finance, Illinois State University.
Rothstein, R., and Mishel, L. (1997). Alternative Options for Deflating Education Expenditures Over Time. In William J. Fowler, Jr. (Ed.), Developments in School Finance, 1996 (NCES 97535). U.S. Department of Education. Washington, DC: U.S. Government Printing Office.