From Foresight, Vol. 10, No.1
published 2003
In 1989, the Kentucky Supreme Court found the state’s public school system unconstitutional for failing to provide all children an equitable and adequate education. To comply with that decision and provide an adequate education will cost about an additional $1 billion, according to a recent study sponsored by the Council for Better Education (CBE), Inc.(1) Raising this amount would not only require an immediate increase in education funding of about 23 percent to 25 percent, it would also increase the budget base by that amount for future funding. In all likelihood, the state will have to bear the burden of finding this additional money. The federal government has never been more than a modest source of funding for elementary and secondary education. Furthermore, the No Child Left Behind Act enacted by President Bush and the Congress in 2001 provides more mandates than monies for the states. At the local level, many Kentucky communities are “property poor” and therefore unlikely to be able to increase their local share of education funding. This implies that much of the proposed $1 billion increase would have to come from a state budget with a general fund of about $7 billion per year and little likelihood of increased revenue over the next several years.
For these and other reasons, some education advocates argue that only major reform of the existing state tax system can produce the necessary dollars to provide adequate funding for public schooling. They may be right. But before the state undergoes a major tax reform to generate additional resources for education, it is reasonable to ask whether it is likely that a significant investment of tax dollars will lead to higher student achievement and improved district and school performance.
All adequacy models, including the one used in the CBE study, assume strong and independent relationships between education resources (tax revenue and the human capital, programs, and services it buys) and organization performance. In other words, the models assume that resources produce results.
But is this a reasonable assumption? Research that systematically and empirically links education resources to results is in short supply. To help shed some light on this issue, I examine some available data about Kentucky school district performance. I begin with a comparison between the best- and worst-performing school districts on the 1999-2000 CATS accountability scores, taking into account the amount of resources available. I also consider other performance indicators such as drop-out and attendance rates and the number of students who go on to college, again within the context of resources available. I next take a longer look, determining how performance and resources have changed in all districts from 1993 to 2001. Finally, I take a prospective look, exploring how districts are projected to perform and how organization needs and resources relate to that expected performance.
Table 1 compares resources for the top- and bottom-performing school districts in Kentucky based on 1999-2000 CATS accountability scores. The table shows that although in 1999 the top-performing districts on average have more than double the amount of local revenue per pupil than the worst-performing districts, the worst-performing group has considerably more state and total revenue. These comparisons of the best- and worst-performing districts suggest a negative relationship between total revenue per pupil and performance on the CATS accountability (i.e., top performers have less total revenue), while the relationship between local revenue per pupil and performance is positive (top performers have more local revenue). Analysis of data for all 176 districts supports these two hypotheses in that the simple correlation coefficient between district local revenue per pupil and CATS score in 1999 is moderately strong and positive (r = .56; r is a statistical measure of how well one thing correlates with another, with 1.0 indicating perfect correlation and –1.0 indicating none), while the correlation between total revenue and CATS score is somewhat weaker and negative (r = -.26). Comparing these two coefficients, it is important to observe that on average total revenue per pupil is about four times the amount of local revenue per pupil ($6,469 versus $1,651 in 1999).
Table 1: Kentucky School District Resources and Performance (1999-2000)
Since Kentucky’s testing and accountability system has faced much criticism and undergone many changes, some might argue that other measures of district performance should be assessed. Table 2 provides simple correlations between indicators of district resources and additional measures of performance. The additional measures are taken from District Report Cards in 2000 and include dropout rate, attendance rate, transition to college, and unsuccessful transitions for the 171 districts with high schools. Indicators of resources include size or average daily attendance (ADA), local and total per-pupil revenue, and average teacher salary (all 1999); a composite measure of teaching resources that combines (1) proportion of classes taught by teachers with a major or minor in the subject area, (2) proportion of classes taught by teachers with professional development in the subject area, and (3) percentage of teachers with a master’s degree or higher; spending per student; and student-teacher ratio.
Table 2: Correlations Between School District Resources and Performance
The first conclusion drawn from an analysis of the data in Table 2 is that almost all of the relationships between resources and performance are quite weak. The only moderately strong correlations are for local and total revenue—local revenue per pupil moderately and positively relates to both higher attendance rates and more transitions to college, and less local revenue relates to more unsuccessful transitions. In contrast, total revenue per pupil is moderately but negatively related to attendance and transition to college—more total revenue relates to lower rates of attendance and fewer transitions to college. Teacher salary is related weakly only to transitions to college, while spending per student is related negatively and weakly to both attendance and transition to college—more spending is associated with lower attendance rates and fewer transitions to college. Finally, districts with more students per teacher (fewer teacher resources) tend to have slightly higher attendance rates and transitions to college and fewer unsuccessful transitions, but the relationships are very weak. Overall, except for local revenue per pupil, there is little evidence that several different types of resources relate positively to performance or results.
One conclusion that could be drawn from Tables 1 and 2 is that the top-performing districts already have adequate resources, and that this amount is less than the worst-performing districts. Does this mean that the highest-performing districts are doing a more effective job with fewer but still adequate resources? Or, conversely, does this mean that the lowest-performing districts are much less successful in using and applying their greater resources to education programs and services? Studies with more complex research designs that examine and test other factors that may be causing these performance differences are necessary to help answer these questions. However, the data in Tables 1 and 2 illustrate some of the difficulties in attempting to link revenues and resources to education performance and also suggest the need for caution in accepting estimates of revenue adequacy.
Tables 1 and 2 also begin to address the important questions of what resources are purchased with education revenue and whether they help produce desired results. Teacher salaries, teacher training and experience, and student-teacher ratios do not appear to make much difference in district performance, despite the fact that personnel expenses comprise a substantial proportion of district spending. Table 1 shows that average teacher salaries for the top and bottom groups do not differ much, and Table 2 shows that teacher salary and several other measures of teaching resources appear to have little or no positive relationship to performance. This suggests that if the top-performing districts have better teachers and if this teaching excellence helps explain their outstanding performance, then having higher teacher salaries and more teachers per student, regarded by many as key educational resources, may not significantly influence district performance.
Tables 1 and 2 examine district performance for one school year; however, the resource-performance relationship also can be assessed over time by asking whether districts with the greatest improvements in accountability performance since 1993 also had the greatest increases in revenues. Table 3 examines revenue change for the most- and least-improved school districts from 1993 through 2001. The data show that districts that improved the least from 1993 through 2001 had much less local revenue per pupil from 1990 through 1999, but had somewhat greater proportional local revenue increases in that period than did the most-improved districts. In contrast, the least-improved districts had almost identical total revenue as the most-improved districts in 1990 ($3,364 versus $3,326), somewhat more total revenue in 1999 ($6,589 versus $6,234) and a slightly higher rate of total revenue increase than the most improved districts (96 percent versus 87 percent).
Table 3: Per Pupil Revenue for Most- and Least-Improved Districts
These comparisons suggest that for all districts the relationship between increased revenue and improved performance is ambiguous or nonexistent. In fact, for all districts, the correlation between percentage change in total revenue per pupil (1990-1999) and percentage change in accountability score (1993-2001) is -.05, and the correlation between change in local revenue per pupil and change in accountability score for these same time periods is .01. These weak correlations indicate that, after more than a decade of KERA, improvement in district accountability performance has little or no relationship to increases in local and total revenue per pupil.
The previous three tables examine past and present district performance. I next explore how districts are projected to perform and how needs and resources relate to that projected performance. Table 4 provides data on projected successful and unsuccessful districts using a simple forecasting model from a previous paper that compares several models for projecting school accountability scores.(2) The two groups of districts with projected CATS scores in 2014 and actual score in 2001 are listed at http://www.uky.edu/~proeder/keraweb.htm. Table 4 also provides several measures of “need”—district size and change in size, and poverty and change in poverty, as well as several measures of “resources”—per pupil local and total revenue and average teacher salary and changes in these indicators over time.
Table 4: Needs and Resources for Projected Successful and Unsuccessful Districts
Several points should be made about needs and projected performance. Not only are projected unsuccessful districts much smaller than projected successful ones in 2001 (ADA of 2,007 versus 3,970), they also have declined in average daily attendance since 1991 (-10.2 percent) compared with a small increase in size for the projected successful districts (8.7 percent). Since the projected unsuccessful districts are quite small and have been getting smaller than the successful ones, does this indicate more or less need for resources? Another interesting indicator of need is poverty. Projected successful districts had a somewhat greater increase in proportion of children eligible for subsidized meals from 1992 through 2001 than unsuccessful districts (24.1 versus 16.7 percent); however, the projected unsuccessful districts had more than double the proportion of poor children in 2001 (66.4 percent versus 32.6 percent). Previous research has demonstrated that poverty consistently is the strongest predictor of school and district performance controlling for other plausible determinants, so it is not surprising that the projected unsuccessful districts have high poverty rates. On the other hand, it is interesting to note that the projected successful districts faced a slightly greater increase in poverty from 1992 through 2001 than the projected unsuccessful districts.
The findings for resources and projected successful and unsuccessful districts differ very little from those in the previous tables for other groups of districts. Projected successful districts had much more local revenue per pupil in 1991 and 2001, but a lower rate of increase from 1991 to 2001. Projected unsuccessful districts had more total revenue per pupil in 1991 and 2001 and a slightly higher rate of increase of total revenue in that period. Average teacher salary and increase in salary do not differ much for the two groups.
This simple analysis raises doubts about the fundamental assumption of most school funding adequacy models that a positive relationship exists between resources and results. Several indicators of district performance and resources for Kentucky school districts suggest that money alone does not buy results. The least successful Kentucky districts in 2000 actually have more revenue per pupil than the most successful ones, and district improvement in performance over the past decade has no relationship to increased revenues.
What are some possible conclusions from this analysis? Based on these data as well as many other studies, some will conclude that money doesn’t matter when it comes to organization performance. On the other hand, some studies find relationships between resources and performance. Money may matter, but demonstrating exactly how it matters has not been an easy task for education researchers and advocates.
An alternative conclusion is that substantially more revenue should be invested, but only in the worst-performing or most-disadvantaged schools and districts. Since Kentucky has achieved relative equity in financing but some school systems and individual schools continue to underperform rather substantially, it may be that to boost overall performance, most new state tax dollars should be allocated to the worst-performing districts and schools. But this conclusion still begs the question of whether the increased resources would likely achieve intended outcomes. The testing and accountability data certainly suggest the worst performers need something, perhaps much greater investments than the top-performing systems. But the question of how these increased revenues would be applied requires much more study. Unfortunately for this alternative, it is doubtful that a majority of policymakers, school leaders, and perhaps more importantly, taxpayers are prepared to support such a radical redistribution of resources, even for the relatively popular goal of equal opportunity.
A third conclusion is that substantially more money should be invested in all Kentucky districts but only for organization resources and programs demonstrated to be effective. This returns to questions about the fundamental assumption of resource adequacy models that “resource configurations/strategies are able to produce desired results.” The limited data analyzed in this paper do not necessarily demonstrate that resources don’t produce desired results, but they do suggest that advocates for substantially increased investments in public education based on the concept of revenue adequacy need to do much more work to show Kentuckians how the investment of $1 billion will lead to more effective school systems.
* Dr. Roeder is a Professor of Political Science at the University of Kentucky. Return to text.
1 D. Verstegen, “Calculating the Cost of an Adequate Education in Kentucky,” Feb. 2003. Return to text.
2 See Method C in “The KERA Endgame,” Nov. 2001, available at http://www.uky.edu/~proeder/keraweb.htm. Return to text.