The Dilemma of School Finance Reform Joshua Hall Department of Economics PO Box 6025 West Virginia University Morgantown, WV 26506-6025 Telephone: 724-288-7579 Fax: 304-293-7897 [email protected]

In many states school finance litigation has mandated the centralization of education funding in an attempt to improve educational adequacy. This study uses data on Ohio school districts to investigate the factors affecting school district performance and finds few aggregate variables that state education policymakers can change to improve student performance. I argue that this presents a dilemma for state policymakers under judicial order to improve educational opportunity through increased financing. Given the constraints imposed by the judiciary, the superior knowledge of local officials suggests that state policymakers should give local school district officials autonomy to craft local solutions. Keywords: School district performance; school finance reform JEL codes: 1210, 1280

The Dilemma of School Finance Reform 1 Introduction State governments have always played a significant role in education policy. Most state constitutions, for example, contain an education clause granting the state government the power to establish a system of common schools. These constitutional provisions are quite general in how education is to be provided. Some education clauses merely grant state governments the authority to establish a system of schools within broad constitutional guidelines. In Ohio, for example, Article 6 of the Ohio Constitution states that the General Assembly “shall make such provisions, by taxation, or otherwise, as … will secure a thorough and efficient system of common schools throughout the state.” This provision is analogous to the clause in the U.S. Constitution giving Congress the power to provide for the common defense (Owsiany, 2001) in that it gives the General Assembly the authority to levy taxes for a “thorough and efficient” system of common schools, but leaves decisions about specific education policies to the discretion of policymakers. Over time the role of state governments in education policy has expanded considerably in terms of financing. In 1919, state governments provided only 16 percent of all funding for elementary and secondary education in the United States. By 1950 that amount had risen to nearly 40 percent and in the year 2000 stood at 49.7 percent (U.S. Department of Education 2004, Table 156). Over the past thirty years a primary reason for the increase in state financing of education has been a wave of court-ordered school finance reforms that have significantly increased state spending on education (Murray, Evans and Schwab, 1998). There have been efforts in nearly every state to challenge the constitutionality of school finance systems that produce unequal levels of school funding and unequal school outcomes among school districts.

1

Even in states where the school funding system was ruled to be constitutional, litigation has spurred considerable changes. Since 1970, every state has changed their school finance system to create a more equal distribution of funding among districts (Hoxby, 2001). The most successful of these cases have focused not on equity of finances but rather on the concept of adequacy. At its core, adequacy in the context of education means adequacy in terms of the outcomes of education. In Ohio, for example, the plaintiffs filed their adequacy lawsuit on the grounds that the current system of school financing resulted in some districts having funding levels so low that students were not given the opportunity to academically succeed. Plaintiffs made their case that the current system was inadequate by presenting the courts with evidence of deteriorating facilities, lack of access to diverse and challenging curriculums, and low test scores on state proficiency exams (Hunter, 2000). As a result of the litigation, Ohio policymakers were directed by the Ohio Supreme Court to fashion a remedy that satisfy the “thorough and efficient” clause of the state constitution. With the exception of telling the Ohio General Assembly what the school finance system could not look like, the Ohio Supreme Court left the details of the reform up to the Ohio General Assembly. This type of mandate without details is typical of many judicial decisions related to school finance reform litigation although the degree to which the court micromanages state policymakers varies considerably by state. It is my contention that the mandate without details presents a dilemma for state policymakers because the idea of educational adequacy suggests that government can improve education by increasing spending, requirements and oversight of local school districts (Orfield, 1994). This perspective conflicts with a large body of education research showing no consistent relationship between spending and achievement (Hanushek, 2003). The result is state

2

policymakers under an “adequacy” mandate know the goal but are uncertain about how to get there (Hoxby, 2001). Faced with this dilemma, policymakers usually impose a package of reforms that increase the number of mandates on local school districts as well as financing because “he who pays the piper calls the tune.” Thirty years of school finance reform has shown this approach to be a failure as inequality of achievement has changed little. This is not surprising given that the effects of equalization on student achievement are generally weak (Hoxby, 2001). Using data on Ohio school districts, I confirm that there is little relationship in the aggregate between the level and usage of school resources and student performance. As Hanushek (1994) suggests, this rules out central direction in how to direct increased funding. In the face of the evidence that few policy variables seem to affect student performance, how should state policymakers proceed? Building upon the insight of Hayek (1945) with respect to the distribution of knowledge in society, this paper argues that policymakers should try to decentralize decision-making as much as possible given political and judicial constraints. Institutional changes giving parents and school officials the freedom to utilize their superior local knowledge to improve student performance gives state policymakers their best hope of overcoming the dilemma of judicially-mandated state education reform. The remainder of this paper is organized as follows. Section 2 presents data on Ohio school districts to illustrate the diversity that exists among school districts in a typical state. Section 3 then employs regression analysis to confirm that lack of a systematic relationship between policy-manipulable variables and student performance. Section 4 considers how decentralized decision-making can lead to institutional changes that increased the effectiveness of additional spending on schools. Section 5 offers some concluding remarks.

3

2 Data on Ohio School Districts Ohio is a very diverse state with eight metropolitan areas and numerous urban, suburban and rural school districts. Thus, Ohio is a good state to exemplify the dilemma facing state policymakers under judicially-mandated school finance reform. For the 1999-00 school year Ohio had 611 regular local public school districts. This year was chosen because it allowed for the integration of Ohio Department of Education data with data obtained from the 2000 U.S. Census. After excluding four small districts because of censored data, the final sample contains 607 districts. Summary statistics of Ohio school districts are presented in Table 1. Graduation rates and tenth grade math proficiency test scores are the two student achievement variables that will be analyzed in this study. There exists considerable variation exists among Ohio school districts in terms of performance. For example, the Cleveland Municipal School District only saw 33.7 percent of its current senior class graduated in four years while several districts achieved a graduation rate of 100 percent. The same variation can be seen in the passage rates on the tenth grade math proficiency exam, where East Cleveland City had a passage rate of only 44.5 percent and six districts had every eligible student pass the exam. As was the case with performance variables, there are large differences among districts with respect to teacher variables such as experience and pay. The average salary of classroom teachers in a district varies from a high of $58,434 in Cuyahoga Heights to a low of $27,609 in Eastern Local in Meigs County. While average salary of classroom teachers is one of the best measures available for measuring teacher compensation levels it is important to note that district averages can be somewhat misleading. For example, given two school districts with identical salary structures, the one with more experienced teachers will have higher average salaries by virtue of having more teachers higher up the salary schedule.

4

One of the most striking things shown in the table is the variation in teacher experience levels among districts. Looking beyond the summary statistics, most of the districts with extremely high rates of inexperienced teachers are small, rural school districts. In these districts, many with fewer than 100 teachers, several retirements in one year can drastically change teacher experience levels. Looking beyond the most extreme districts, however, it is clear that some large districts have a large number of inexperienced teachers as well. Columbus City, for example, had nearly 40 percent of its teachers with four years or fewer of classroom experience. While this could also be due to retirement trends, it is symptomatic of the high turnover and burnout that is frequently seen in urban education. The financial disparity between school districts in Ohio can also be seen in Table 1. For the 1999-00 school year the difference between the highest spending school district in the state (Beachwood City in Cuyahoga County) and the lowest spending school district (Western Brown Local in Brown County) was over $9,000. To put it another way, Beachwood City spent nearly three times what Western Brown Local did Education spending can be summarized as coming from either local, state, or federal sources. The average school district in Ohio received 46.9 percent of its revenue from local sources during the 1999-00 school year, 48.1 percent from state sources, and 5.0 percent from federal sources. Within those averages though, school districts tend to vary between extremes. For example, Trimble Local in Athens County receives only 10.2 percent of its revenue from local sources with 79.4 percent from state taxpayers and the remainder coming from federal sources. At the other extreme, Perry Local in Lake County receives 92.7 percent of its revenue from local sources (primarily due to a nuclear power plant being located with the district).

5

Finally, Ohio school districts vary considerably in their geographic size, number of pupils, median household income, poverty levels, education levels, and taste for private education. The average Ohio school district is 72.6 square miles in size. Sebring Local in Mahoning County is the smallest school district at one square mile and the largest school district is Adams County/Ohio Valley Local in Mahoning County which is 640 square miles in size. As can be expected, the size of a school district can have a large role to play in district resource usage. The most educated school district in Ohio (as measured by the percentage of district residents 25 or older with at least a bachelor’s degree) is Wyoming City in Hamilton County where over 68 percent of adults have at least a bachelor’s degree. In terms of poverty, the range is pretty dramatic, with some school districts having no students receiving free or reduced price lunches and Cleveland Municipal with over 80 percent of its students receiving free or reduced price lunches. The special school district tabulation from the 2000 Census provides private school enrollment numbers by district. There are several districts in Ohio with no district residents attending private schools and there are several districts with private school enrollment rates over 40 percent. The average school district in Ohio has just over 10 percent of its students attending private school. When school district size is measured in terms of number of pupils, the largest school district in Ohio was Cleveland Municipal with over 75,000 students. The average school district in Ohio had just over 3,000 pupils and the smallest in the state, Vanlue Local in Hancock County had just 311 pupils. The picture that emerges from looking Ohio’s school districts in this manner is one of diversity. Ohio has extremely large school districts and extremely small school districts. It has high income districts and low-income districts. It has districts that finance their schools primarily through local sources and districts that finance their schools primarily through state sources.

6

Some districts spent well over $10,000 per student and others make due with under $6,000. To be effective, state education policy needs to take these differences into account.

3

Empirical Model and Results

A major problem with relying on simple descriptive statistics when evaluating school districts is other factors that impact student learning are not held constant. A district that performs poorly in general might actually being doing quite well once the socioeconomic characteristics of the district population are taken into account. From the standpoint of proper education policy, what is important is examining the relationship between various school district inputs and student learning holding all other factors constant. In this section I try to achieve that objective by employing multiple regression analysis to estimate an education production function. Table 2 presents the results from a basic multiple variable regression model explaining a school district’s graduation rate for the 1999-00 school year. The model attempts to relate various school district characteristics to the measure of school district performance, in this case school district graduation rates for the 1999-2000 school year. A similar model utilizing school district passage rates on mathematics proficiency test scores was also estimated to check the robustness of these results and is presented in the Appendix. The information presented in Table 2 suggests that the most important factors impacting a school district’s graduation rate are the education level of adults in the district, the percentage of district students eligible for free or reduced price lunches, and a school district’s attendance rate. It is clear that there is a strongly negative relationship between non-school socioeconomic factors and a school district’s graduation rate. In terms of economic significance, the most important non-socioeconomic variable is a school district’s attendance rate. The higher a school

7

district’s attendance rate, other things equal, the higher the school district’s graduation rate. For example, the results in Table 2 suggest that a district raising its attendance rate by one percentage point could expect an increase in its graduation rate of around 2.3 percentage points. Recall that the thrust of the school finance litigations rests on the issue of inadequate spending in schools. While a relationship exists between expenditure per pupil and district performance, the relationship is negative and economically unimportant. Increasing expenditure per pupil seems to lower school district performance; however, the change from increasing expenditures is not economically meaningful. Caution should be taken on placing too much weight on one estimate but it should be clear from the evidence presented in Hanushek (2003) and here that there is not a clear relationship in the aggregate between spending and achievement. For example, an increase in the average expenditure per pupil by $1000 would not raise a district’s graduation rate by even a 10th of a percentage point. Again, this does not mean that money does not matter (Hedges, Laine and Greenwald 1994), just that it impossible to know how money might matter at the state level. The same is true for the other financial variable included in the regression – the amount of school district revenue raised from local sources. While the results in Table 2 are consistent with the theory that locally financed schools are more efficient (Fischel, 1997), the small coefficient combined with the large sums of money necessary to change the variable even marginally suggests that increasing the amount of revenue from local sources is not an easy way to improve school district performance (and is exactly opposite to the court’s wishes). To the extent that having a high percentage of inexperienced teachers is a result of high rates of teacher turnover, this result is consistent with the research by Vedder and Hall (2004) showing that teacher turnover plays an important role in school district performance. Upon first

8

glance, this appears to be a policy variable that state policymakers could use to influence student performance. While reducing the percentage of inexperienced teachers in a school district could be a good way to increase school district performance, however, studies of teacher turnover and mobility suggest that doing so might be extremely difficult. The reason why this may be so is because research suggests that working conditions, especially characteristics of students such as achievement levels, plays a prominent role in teacher mobility (Hanushek, Kain and Rivkin, 2001). School district spending priorities, not low salaries per se, play a significant role in causing new teachers to leave a district (Theobald and Gritz, 1996). While great hope should be placed on increasing school district performance by reducing teacher inexperience, that hope must be tempered with the knowledge that the factors influencing teacher mobility are complex and not easily manipulated by policymakers. Table 2 shows a negative relationship between a proxy for class size (student-teacher ratio) and a school district’s graduation rate. This finding is similar to meta-analyses of similar research that can find no consistent relationship, in the aggregate, between measures of class size and student performance (Hanushek, 1999). This does not mean that class size is unimportant, merely that the not all class size reductions are equal in their impact. A consistent negative relationship exists between school district size and school district performance exists, consistent with past research (Galles and Sexton, 1995). However, this relationship does not appear to be economically significant. So while larger school districts seem to do worse than smaller school districts other things being equal, the effect on a school district’s graduation rate is extremely small. A final interesting relationship is the negative relationship between the percentage of a district’s residents that attend private school and a school district’s graduation rate. This research is result is contrary to previous research on Ohio school districts showing that higher private

9

school enrollment, other things equal, increases student performance on achievement exams (Hall and Vedder, 2003-04). This suggests the need for further research into the interaction between competition and school district performance.

4

Policy Implications

Faced with the mandate to increase state aid to local school districts, state policymakers have reacted by demanding more accountability and state control. The rise in school finance litigation, not surprisingly, has coincided with the rise in the number of statewide education reforms such as achievement testing, all-day kindergarten, and class size reductions as state policymakers desire to ensure that the mandated funds are spent wisely. The contention of this paper is that this idea of centralism rests on the unrealistic assumption that individuals farthest the classroom have greater knowledge about the best use of resources. As Hayek (1945) points out, this is generally not the case. The ultimate decisions about the inputs into the education process need to be decentralized as much as possible to insure that knowledge unique to a given time and place will be utilized. Failure to use this knowledge will result in a misallocation of resources. Thus while class size reduction might be appropriate in certain circumstances, it might not work in the aggregate because it misallocates resources away from where they might be used more effectively. The aggregate relationships presented here tell us nothing about the wisdom of increasing teacher experience or reducing class size in a particular instance. That decision can only be made by individuals intimately involved in the classroom, with knowledge of the educational goals as well as the cost-benefit trade-off of any marginal change. The knowledge advantage of actors in the classroom and at the local school

10

district level is persistent and occurs naturally, or at a far lower cost than for state policymakers as is generally true for lower decision-making units (Sowell, 1980). The persistent cost disadvantage of state policymakers in acquiring the necessary information to make rational economic decisions with respect to educational policies implies that if state policymakers want to comply with the courts directive to increase spending and increase educational opportunity, the best way to do so is by giving as much flexibility to lower decisionmaking units as possible. This could be through block grants to local school districts. One possible concern about this approach might be the principal-agent problem between school officials and parents (Hoxby, 1999). This concern, however, exists with state policymaking as well. Institutional changes could be made that would increase parental exit options to help mitigate the principal-agent problem. Such changes include interdistrict enrollment, charter schools, or a voucher program. Regardless, any attempt to increase state control together with state financing is likely to fail to meet its goal of improved performance because state policymakers lack the necessary information to channel increased resource through productive channels.

5

Conclusion

The results of this investigation into the relationship between school district characteristics and student performance should not be surprising as they confirm an excess of evidence from previous research on education production. At the same time, however, a new look at the factors that influence Ohio school district performance is useful because it helps to refine the limits and scope of state policy. Namely, there appears to be few “one-size-fits-all” solutions to improving the performance of Ohio school districts. This suggests that the recent increased centralization of

11

education policymaking in the hands of Ohio state education officials might be misguided, as there are almost no manipulatable policy variables that have a consistent positive impact on student achievement. Faced with the mandate to increase spending, state policymakers should decentralize decision-making as much as possible to allow lower decision-making units to utilize their local knowledge.

12

References Galles, G. and R. Sexton. 1995. “Diseconomies of School District Size.” Journal of Social, Political, and Economic Studies 20: 241-245. Fischel, W. 1997. “Homevoters, Municipal Corporate Governance, and the Benefit View of the Property Tax.” National Tax Journal 54: 157-73. Hall, J. and R. Vedder. 2003-04. “The Impact of Private Schools on Public School Performance: Evidence from Ohio.” Journal of Economics and Politics 16: 77-92. Hanushek, E. 1994. "A Jaundiced View of 'Adequacy' in School Finance Reform." Educational Policy 8: 460-469. Hanushek, E. 1999. “The Evidence on Class Size.” In Mayer, S.E., & Peterson, P.E. (Eds.) Earning and Learning: How Schools Matter. Washington: Brookings Institution Press: 131-168. Hanushek, E. 2003. “The Failure of Input-based Schooling Policies.” Economic Journal 113: F64-F98. Hanushek, E., J. Kain and S. Rivkin. 2004. “The Revolving Door.” Education Next 4: 77-82. Hayek, F. 1945. “The Use of Knowledge in Society.” American Economic Review 35: 519-530. Hedges, L., R. Laine, and R. Greenwald. 1994. “Does Money Matter? A Meta-Analysis of Studies of the Effects of Differential School Inputs on Student Outcomes.” Educational Research 23: 5-14. Hoxby, C. 1999. “The Productivity of Schools and Other Public Goods Producers.” Journal of Public Economics 74: 1-30. Hoxby, C. 2001. “All School Finance Equalizations Are Not Created Equal.” Quarterly Journal of Economics 116: 1189-1231. Hunter, M. 2000. “Trying to Bridge the Gaps: Ohio’s Search for an Education Finance Remedy.” Journal of Education Finance 26: 63-86 Murray, S., W. Evans and R. Schwab. 1998. “Education Finance Reform and the Distribution of Education Resources.” American Economic Review 88: 789-812. Ohio Department of Education. 2005. Interactive Local Report Card Data Download [electronic file]. Columbus: Ohio Department of Education. Orfield, G. 1994. “Asking the Right Question.” Educational Policy 8: 404-413.

13

Owsiany, D. 2001. “The General Assembly v. The Supreme Court: Who Makes Public Policy in Ohio?” The University of Toledo Law Review 32: 549-561. Sowell, T. 1980. Knowledge and Decisions. New York: Basic Books. Theobald, N. and R. Gritz. 1996. “The Effects of School District Spending Priorities on the Exit Paths of Beginning Teachers Leaving the District.” Economics of Education Review 15: 11-22. U.S. Department of Education, National Center for Education Statistics. 2004. Digest of Education Statistics 2003. Washington: Government Printing Office. Vedder, R. and J. Hall. 2004. Effective, Efficient, Fair: Paying for Public Education in Texas. Austin: Texas Public Policy Foundation.

14

Table 1. Summary Statistics on Ohio School Districts Variable

Mean

Median

Maximum

Minimum

Std. Dev.

Graduation rate

86.4%

88.1%

100.0%

33.7%

9.0%

% students passing grade 10 math exam

84.9%

86.2%

100.0%

44.5%

9.2%

Attendance rate

94.8%

94.9%

98.6%

83.2%

1.3%

$39,320

$38,718

$58,434

$27,609

$4,909

22.8%

22.2%

55.6%

5.3%

8.0%

97.5%

99.7%

100.0%

65.9%

4.0%

Student-to-teacher ratio

18.3

18.4

23.3

7.8

2.0

Teacher attendance rate

96.0%

96.1%

99.8%

86.9%

1.0%

Expenditure per pupil

$6,663

$6,366

$14,684

$5,031

$1,142

% revenue from local sources

46.9%

45.7%

92.7%

10.2%

17.2%

$30,571

$29,491

$61,222

$15,820

$6,411

17.2%

12.8%

68.2%

2.2%

12.3%

21.2%

17.9%

80.9%

0.0%

14.3%

10.2%

8.5%

52.2%

0.0%

7.6%

Students per square mile

114.0

32.9

2023.7

2.0

192.5

School district size (head count)

3,009

1,869

76,367

311

5,274

Average salary of classroom teachers % of teachers with 0-4 years experience % of 9-12 teacher certified in teaching area

Median household income % of district residents 25 and older with a bachelors degree or higher % of students receiving free or reduced price lunches % of districts school-age children attending private schools

Source: Ohio Department of Education and author's calculations.

15

Table 2. The Relationship Between District Characteristics and Graduation Rates Effect of Incremental Change on Dependent Variable (Coefficient)

Variable or Statistic

School district size (head count)

-0.000002

***

Attendance rate

(-3.05) 2.319091

***

% of district HS teachers certified in teaching area

(8.05) 0.003705

% of district residents 25 & older with at least a BA

(0.057) 0.136910

***

% of district teachers with 0-4 years of experience

(4.32) -0.131137

***

School district expenditure per pupil ($)

(-3.83) -0.000007

% of district students eligible for free or reduced price lunches

(-1.46) -0.192143

***

(-6.54) 0.000007

**

(1.96) -0.084853

**

School district revenue from local sources ($) % of 5-17 year-olds within district boundaries attending private school Teacher attendance rate

(-2.22) -0.223493

Classroom teacher to student ratio

(-0.83) -0.003537

**

Average salary for classroom teachers in district

(-2.22) -0.003537

***

(-3.29) Constant

-0.844810 (-2.41)

R

0.55

2

Adjusted R2

0.54

F-Statistic

60.61

Note: * indicates significance at the 10% level, ** at the 5% level, and *** at the 1% level. Absolute t-statistics in parentheses, N= 607.

16

**

Appendix. The Relationship Between District Characteristics and Math Scores Effect of Incremental Change on Dependent Variable (Coefficient)

Variable or Statistic

School district size (head count)

-0.0000004 (-0.72) 1.770568

Attendance rate

***

% of district HS teachers certified in teaching area

(6.56) 0.143261

**

% of district residents 25 & older with at least a BA

(2.36) 0.064721

**

% of district teachers with 0-4 years of experience School district expenditure per pupil ($) % of district students eligible for free or reduced price lunches School district revenue from local sources ($) % of 5-17 year-olds in district boundaries attending private school Teacher attendance rate Classroom teacher to student ratio Average salary for classroom teachers in district Constant

(2.17) -0.101669 (-3.17) -0.000013 (-2.74) -0.308573 (-11.21) 0.000004 (1.33) -0.024379 (-0.68) 0.418031 (1.66) -0.003015 (-2.01) -0.003015 (-0.19) 0.000000 (-3.52)

R2

0.62

Adjusted R2

0.61

F-Statistic

80.43 Note: * indicates significance at the 10% level, ** at the 5% level, and *** at the 1% level. Absolute t-statistics in parentheses, N= 607.

17

*** *** ***

* **

***