AEFP 45th Annual Conference

Toward a Meaningful Impact through Research, Policy & Practice

March 19-21, 2020

Application of the Comparable Wage Index for Teachers (CWIFT) to School District Education Finance Data

Malia Howell, U.S. Census Bureau,

School finance data help the public understand where and how funds are spent on education. However, comparisons between districts often do not take into consideration differences in costs which are beyond the control of the district. The cost of living for a given school district has a direct impact on the amount of wages the district must pay in order to attract and retain teachers. In fiscal year (FY) 2016, school districts spent an average of $11,669 per pupil, of which 81.6 percent were for salaries and benefits (Cornman et al. 2019). With the majority of current expenditures being spent on costs affected by the cost of living, the purchasing power for each dollar spent by a school district in a high-cost area is less than that for a school district in a low-cost area.
The Comparable Wage Index for Teachers (CWIFT) was developed by the National Center for Education Statistics (NCES), in collaboration with the U.S. Census Bureau, as a means for measuring geographic differences in labor costs. The CWIFT uses data from the American Community Survey to estimate the regional variations in the earnings of college-educated workers who are not educators after controlling for differences in job-related and demographic characteristics. The methodology assumes that teachers are similar to other college educated workers in terms of their preferences for certain wages, working conditions, and community amenities. The comparable wage index can thus be used to adjust district-level finance data to make better comparisons across geographic areas. By controlling for geographic costs, researchers can focus on the differences in spending patterns that result from the choices made by the district in staffing and resource allocation.
The School District Finance Survey (F-33) collects detailed financial data on revenues and expenditures for all public school districts serving prekindergarten through 12th grade students in the United States. All states report expenditure data for each of their school districts within survey categories utilizing standardized accounting guidance, which allows this study to compare spending for instructional salaries and total current expenditures for school districts both within and across states.
Purpose: The paper will demonstrate the application of geographic cost adjustments to education finance data and investigate the possible impacts of its application on research and policy studies. The paper will focus on two primary research questions:
1. How does accounting for geographic cost differences affect the measurement of disparity between high-spending and low-spending school districts within each state?
2. How does accounting for geographic cost differences help us understand the relationship between expenditures and policy choices such as class size?
Methodology: The study will apply the CWIFT to selected expenditure data from the FY 2016 School District Finance Survey (F-33), incorporating relevant statistics from the Local Education Agency (School District) Universe Survey for the corresponding (2015-16) school year. The study will cover all active school districts with student membership greater than zero, where a CWIFT was able to be calculated.
Preliminary Results:
After applying the CWIFT, disparity between the highest-spending and lowest-spending districts increased in 33 states, decreased in ten states, and remained about the same in six states. We will examine three states in greater detail to describe the CWIFT and its impact on disparities in school district spending within those states, including its impact based on the specific locale of the districts. The three states will include one where disparity increased, one where disparity decreased, and one where disparity remained about the same.
As class sizes get smaller, school district per pupil spending on instructional salaries increases. Controlling for differences in cost by geography strengthens the correlation between class sizes and instructional salary spending per pupil, which indicates that when we control for geographic differences in labor costs, per pupil instructional salary spending will increase as class sizes decrease at a greater rate than when we do not control for geographic costs.
The purpose of the poster presentation will be to facilitate discussion and elicit feedback from researchers, policy makers, and practitioners regarding the application of the CWIFT to per pupil spending data.
Cornman, S.Q., Ampadu, O., Wheeler, S., Hanak, K. and Zhou, L. (2019). Revenues and Expenditures for Public Elementary and Secondary School Districts: School Year 2015–16 (Fiscal Year 2016) (NCES 2019-303). U.S. Department of Education. Washington, DC: National Center for Education Statistics. Retrieved November 7, 2019 from



Can you suggest why using the CWIFT to adjust-spending has different effects in MO and SC? And are there other cost adjustments to which the CWIFT can be compared? For example, do cost function adjustments of the kind suggested by Duncombe and Yinger and others have different implications for disparity in adjusted expenditures. Bo Zhao at the Federal Reserve Bank of Boston has recently estimated cost-adjustments for CT. Might be interestng to compare his adjusted spneding numbers to those implied by CWIFT.

This is such an interesting and well-put together poster! The graphics you use nicely tell your story. Well done and thank you for sharing. Just a few questions/ideas for you as you move forward with this work: 1) It looks like the mean spending in VA stays about the same and the LEA range maybe even increases a little, although the distribution appears a bit more normal. This is also the case in the second example. What do you make of this? why do we see this pattern? 2) It seems like in each of the 3 states you use as examples in the PP instructional salary graphs has some serious outliers (esp CA). I wonder what you make of those outliers? What would your conclusions be if you used the interquartile range, or some measure of dispersion, rather than mean and range IPP overall and across urbanicity/poverty distinctions? 3) How do you pick your example states for each part of your analyses? I think it would also be helpful to show national trends.

Add new comment