Exploring the Sensitivity of Education Costs to Racial Composition of Schools and Race-Neutral Alternative Measures: A Cost Function Application to Missouri

Abstract

This article applies the education cost function methodology in order to estimate additional costs associated with black student concentration and with alternative, race-neutral measures of urban poverty. Recent research highlights the continued importance of the role of race in educational outcomes, and how the intersection of peer group effects and teacher-quality distribution serve to simultaneously disadvantage black children attending racially-isolated black schools (Clotfelter, C., Ladd, H., Vigdor, J., 2006; Hanushek and Rivkin, 2007). But some questions persist as to whether race, in-and-of-itself should be considered a relevant factor influencing the costs of providing equal educational opportunity. I replicate cost models originally introduced in the context of school funding litigation in 2006, using data from the state of Missouri from 2001 to 2005, and then estimate updated cost models based on data from 2006 to 2008, including indicators of school district racial composition and also including race-neutral indicators of urban, concentrated poverty. I find substantive differences in the cost model predictions when race variables are included in the model, as opposed to race-neutral alternatives, identifying higher costs to achieve state average outcomes in districts where the majority of enrolled students are Black, all else equal. But, conflicting evidence from the race-sensitive versus race-neutral models, including varied prediction accuracy for individual districts makes it difficult to choose which measures better approximate the true cost drivers of Missouri public schools.

Notes

Duncombe and Yinger (2007a) pointed out that “it is not possible to estimate a production function at the school-district level using spending as a measure of ‘inputs’ without making the assumption that spending on every input is equally productive and that all input combinations are equally efficient. These are extreme, indeed, ridiculous assumptions, which the cost-function approach does not have to make” (p. 4). Further, Duncombe and Yinger (2007a) noted that “the production-function approach magnifies the problem of accounting for efficiency because it automatically incorporates inefficiency into the definition of the key explanatory variable, namely spending per pupil” (p. 4). Although this concern is typically overlooked in production function analyses, most recent cost-function studies have attempted to address, via varied methods, inefficiencies in the financial input–achievement outcome relationship.

Duncombe and Yinger (2007) explained potential public monitoring measures: “For example, voters may have a stronger incentive to monitor school officials, that is, to force them to be more efficient, if they must pay a high price for any additional school spending. As it turns out, the price of education varies a great deal across districts in most states. With a standard property tax, voters in some districts can shift the burden of additional property taxes onto commercial and industrial property whereas voters in other districts, where commercial and industrial property is limited, cannot” (p. 20).

Significant changes to the MAP assessment reporting were implemented in 2006, including realigning cut scores and performance categories more closely with NAEP. Previously, Missouri had very high relative cut scores for the math assessment in particular. An overview of changes is provided here: http://dese.mo.gov/divimprove/assess/revmapoverview.html

Factors used in determining the category for each college included median entrance exam scores for the 2001–2002 freshman class (SAT or ACT), percentages of 2001–2002 freshman scoring in the top 15% of the SAT or ACT, percentage of 2001–2002 freshman who ranked in the upper fifth and upper two fifths of their high school graduating classes, minimum class rank and grade point average required for admission (if any), and percentage of applicants to the 2001–2002 freshman class who were accepted. We assume relative stability over time to the highest categories of selectivity.

Notably, however, this spending margin is not a function of recent school finance reforms and instead a function of court imposed higher local property tax rates (see Baker & Green, 2009). At equitable tax rate, Kansas City's spending would fall below estimated need.

“The State contends that it would be inappropriate to require funding determinations to be based on those [highly successful] districts because, despite their educational success, they have ‘notable inefficiencies’ in their spending practices and, for that reason, the amount that they spend on education cannot serve as the measure of the amount necessary to achieve a constitutionally adequate education. Neither CEIFA itself, the record in this case, empirical evidence, common experience, nor intuition supports the State's position that inefficiencies explain why successful districts’ spending levels exceed what the State asserts is the amount needed to provide a thorough and efficient education” (Abbott v. Burke, 693 A.2d 417 (N.J. 1997) (Abbott IV)), p. 27).