State and Federal Policies for School Facility Construction

Abstract

Background: The Ohio School Facilities Commission was set up in response to litigation compelling the state to achieve a more equitable distribution in the quality of school facilities. The American Recovery and Reinvestment Act (ARRA) was a federal policy to stimulate the United States economy and support school facility construction. These two programs provide an opportunity to compare a state-run program to a federal one. Purpose: This study analyzes the distribution of school and community resources across school districts in Michigan and Ohio, two otherwise similar states with very different policies to support school facilities. Using the year prior to the implementation of the ARRA as a baseline, this study then compares the allocation of Qualified School Construction Bonds (QSCB) provided under the ARRA. Research Design: Using Michigan as a control group because of the weak state support for school construction, the analysis shows the distribution and variation of school building capital, community resources, educational inputs, and student outcomes across districts in both states. Based on these indicators, districts that received QSCBs are compared with districts that did not. Findings: Ohio does appear to have a more equitable distribution of capital facilities but the QSCB allocations in Ohio do not appear to be noticeably more equalizing than those in Michigan. Moreover, the allocations in both states went, on average, to districts above the mean on a number of measures of community resources, educational inputs, and student outcomes.

Keywords

educational facilities education finance state and federal policy equity school construction bonds

The quality of school facilities in the United States is highly correlated with local community wealth. The U.S. Constitution makes no mention of education, so by virtue of the Tenth Amendment, the funding of schools has fallen on the states (Thomas, Cambron-McCabe, & McCarthy, 2009; Wood, 2007). And within states, a large portion of funding is tied to local property taxes. Roughly, 34% of all local government expenditures in 2010 were for elementary and secondary education and more than 75% of all local tax revenue during that period came from property taxes (Barnett & Vidal, 2012). The inequities in resources that result from this system have led plaintiffs to seek remedies that force states to increase funding to the poorest school districts in the hope that additional resources will boost student achievement (Rebell, 2009; Wood, 2007). Funding for improvements to school facilities is becoming a significant component of this litigation (Erlichson, 2001; Rebell, 2009; Sciarra, Bell, & Kenyon, 2006; Superfine, 2008; Wood, 2007). Rice (2004) explains the increased equity of centralized, state systems of school funding, but some claim that states are not doing enough. As a result, there is a call for a larger federal role in school facility funding (Filardo, Cheng, Allen, Bar, & Ulsoy, 2010). Critics dislike a larger federal role because they think local school boards are more responsive to voters (R. Jones, 2011) and more efficient at matching resources to individual preferences (Rice, 2004).

This article compares two states with fairly similar economic backgrounds, geographies, and school district organization. Yet in spite of their similarities, Ohio and Michigan have distinctly different policies regarding school facilities. Michigan has minimal state support for school renovation and construction whereas Ohio, as a result of litigation, has set up the Ohio School Facilities Commission (OSFC) to equalize the quality of school facilities across districts. This study uses standardized accounting data from publicly available comprehensive annual financial reports (CAFRs) to evaluate whether the efforts of the OSFC to close the gap in the quality of school facilities across the state has led to less intercommunity variation in school facilities compared with Michigan, which has no comparable program. In addition, it will analyze the distribution of federally subsidized Qualified School Construction Bonds (QSCB) provided under the American Recovery and Reinvestment Act. While it is too early to assess the impact of QSCBs on student outcomes, it is possible to track where the bonds were awarded in order to gauge the overall need of recipient districts compared with the districts that did not receive or apply for a bond. This study will first examine how the states were doing on their own with regard to the equity of school facilities and then compare how they distributed the federal stimulus funds.

Statement of Purpose

The attention given to the large variation in building quality across school districts is based largely on the assumption that there is a correlation between the quality of school infrastructure and student achievement. Studies have tried to examine possible correlations between school building conditions and student achievement on standardized tests through a number of statistical techniques, controlling for the socioeconomic status (SES) of the parents and other community variables. However, researchers are plagued by problems of measurement and data availability (Arsen & Davis, 2006; Earthman & Lemaster, 1996). School building quality is composed of numerous components, many of which are difficult to separate and measure accurately. In response, researchers either try to study each factor separately (as best they can), or they rely on some aggregate measure of building quality such as building age (McGuffey & Brown, 1978), engineering appraisal score (Earthman, Cash, & Van Berkum, 1995; Hawkins & Lilley, 1992), or residual value—either market or historic cost (Arsen & Davis, 2006). Those who employ production function models generally control for SES, student background variables, teacher effectiveness, as well as school building characteristics (Berner, 1993; Harter, 1999). They also require reliable measures of student achievement. Since many of these data are not available, researchers are forced to make due with only a small set of these potentially important factors (Picus, Marion, Calvo, & Glenn, 2005).

As a result of the data limitations, the evidence supporting the assumed correlation between school facilities and student achievement is mixed. Eric Hanushek (1996, 1997) has repeatedly called into question the link between school resources and student achievement. Picus et al. (2005) found no correlation between school engineering evaluations and student achievement in Wyoming. Woolner, Hall, Higgins, McCaughey, and Wall (2007) also noted a lack of empirical evidence linking the physical environment to student achievement in the United Kingdom. Roberts (2009) showed that the quality of teaching and learning environments was unrelated to engineering surveys but positively related to educators’ assessments of facilities, suggesting that the subjective impressions of teachers are more relevant than engineering surveys. Bowers and Urick (2011) used data from the 2002 iteration of the Educational Longitudinal Survey in a two-level hierarchical linear model and found no direct effect of facility disrepair on student mathematics achievement during the past 2 years of high school. This study directly addressed a number of issues that plagued previous studies, including subjective surveys; the failure to control for known covariates that affect student performance; small samples; and inadequate models based on building age, depreciated costs, and engineering checklists.

On the whole, however, one would have to conclude that courts have based their decisions on evidence from researchers who claim a positive correlation between the quality of facilities and student outcomes (Rebell, 2009). A number of researchers found a more direct correlation between some indicator of facility quality and student achievement (Berner, 1993; Duran-Narucki, 2008; J. Jones & Zimmer, 2001; Nelson & Zimmerman, 2011; Uline & Tschannen-Moran, 2008; Uline, Tschannen-Moran, & Wolsey, 2009). Other scholars have analyzed the impact of facility quality on mediating factors that are correlated with student achievement such as attendance (Branham, 2004), school completion (Duflo, 2001) students’ problem behavior (Kumar, O’Malley, & Johnston, 2008), and teacher turnover (Buckley, Schneider, & Shang, 2004). Researchers have also looked at the wider macro-social impact of high-quality school facilities on housing prices (Cellini, Ferreira, & Rothstein, 2010; Nelson & Zimmerman, 2011) and voters’ support for schools (Gronberg, Jansen, & Taylor, 2011).

Given that courts tend to accept the correlation between facility quality and student achievement and their decisions reflect a desire to level the playing field, the impact of litigation on the equity of school facilities is relevant for policy makers. This analysis will compare two states in which one has had a number of court decisions directing additional funding for school facilities while the other has not. The economies and demographics of Michigan and Ohio are fairly similar (U.S. Census Bureau, http://quickfacts.census.gov/qfd/states/26000.html) and both states have school districts that generally align with municipalities.¹ Moreover, the school districts in both states have some control over school facility funding by virtue of their ability to raise taxes for the construction and renovation of facilities (NCES-EDFIN, http://nces.ed.gov/edfin/state_financing.asp). The key difference is the existence of a school-infrastructure funding program.

In 1991, the Ohio Coalition for Equity and Adequacy of School Funding filed suit in the Perry County Common Pleas Court on behalf of Nathan DeRolph, a high school student. The coalition represented over 500 school districts in Ohio and the lawsuit alleged that the heavy reliance on local property taxes to fund schools violated the state’s constitutional mandate to provide an efficient education system. The initial finding of the lower court in favor of the plaintiffs survived numerous appeals² and in 1997, Senate Bill 102 provided $300 million to fund the Ohio School Facilities Commission, with the goal of rebuilding Ohio’s public schools (Superfine, 2008). Since the initial DeRolph decision instructing the Ohio legislature to make the school funding formula more equitable, there has been a great deal of follow-up litigation. Plaintiffs, seeing little improvement, filed suit to enforce the initial court decision three times resulting in DeRolph II, DeRolph III, and DeRolph IV, all of which found the state’s funding system unconstitutional (Superfine, 2008). During this litigation, the Ohio legislature continued to fund the facilities commission.

Michigan, on the other hand, has no state support for school facility funding beyond the School Bond Loan Fund, which allows qualifying districts to borrow under the umbrella of the state’s credit rating in order to secure favorable interest rates (Arsen, Clay, Davis, Devaney, Fulcher-Dawson, & Plank, 2005). The lack of state support for school facilities makes Michigan a good control group with which to compare the Ohio data. In 1994, Michigan voters passed Proposal A, which equalized revenues available to Michigan school districts by implementing a foundation allowance of $6,700 per pupil for operating expenditures, significantly increasing the revenue available to the lowest spending districts in the state (Arsen & Plank, 2003; Lockwood, 2002). It is possible that Michigan’s relatively equitable distribution of operating expenditures under Proposal A has had the unintended consequence of increasing interdistrict disparities in school facilities. Proposal A sharply curtailed districts’ ability to raise operating expenditures, yet districts with a large property tax base and strong voter support for education are not constrained on what they can spend on new facilities. Arsen and Davis (2006) showed that the top quintile³ of districts in Michigan together had 64% higher capital assets per pupil than the lowest quintile of districts; moreover, they were only taxing themselves at 37.7% the rate of the poorest quintile.

The analysis will investigate the following three research questions:

What are the differences in the average levels in per-pupil capital assets across the states?

What are the differences in the variation in per-pupil capital assets across the states?

Do the states’ distributions of federally funded QSCBs match the capital funding needs of the districts, their ability to pay, community resources, and prior student achievement?

Method

This study will compare the distribution of school building capital and QSCBs across school districts in Michigan and Ohio. First, the analysis employs a relatively new method for calculating the present value of the capital stock from the historical expenditure shown in school district CAFRs. The method uses the ratio of depreciation to historic expenditure to adjust the capital stock for inflation. Second, the study breaks the school districts in each state into quintiles based on the total taxable value of real property per pupil. The breakdown by quintiles enables a comparison of the mean levels of the resources available to school districts based on community wealth. Third, the article shows three common measures that reflect the variation in resources across school districts in Michigan versus Ohio. Finally, the study compares those districts that received a QSCB to the rest of the districts in each state in order to evaluate whether the bonds went to the districts with the greatest need.

Data Sources

The primary data on school capital assets were obtained from the tables included in the CAFRs filed by school districts in Ohio and Michigan. Ohio makes school districts’ CAFRs available through the website of the Auditor of the State of Ohio (http://www.auditor.state.oh.us/AuditSearch/Search.aspx). Michigan provides financial statements through the website of the Michigan Department of Education’s Office of Audits (http://web1mdcs.state.mi.us/nxt/gateway.dll?f=templates&fn=default.htm&vid=mofa:fa). Some school districts were dropped, however, because of missing or unusable data. For 2008, 389 of Michigan’s 552 traditional public school districts have online financial reports. The rest were gathered by hand. Three districts in Michigan had insufficient data and were dropped from the analysis. All of Ohio’s 614 traditional public school districts have their financial audits online, but 124 used a modified cash basis of accounting, which does not show accumulated capital assets. In the end, there were 549 school districts from Michigan and 490 from Ohio used in this study.

The building cost deflators came from the National Income and Product Accounts (NIPA) tables provided by the Bureau of Economic Analysis (BEA) at the U.S. Department of Commerce. On the BEA website (http://www.bea.gov/iTable/iTable.cfm?ReqID=9&step=1#reqid=9&step=1&isuri=1), Tables 5. 8.4A (years 1929-1997) and 5.8.4B (years 1997-2011) are labeled “Price Indexes for Gross Government Fixed Investment by Type.” Data on the taxable value of real property were obtained from the Michigan Department of Education Bulletin 1014 (http://www.michigan.gov/mde/0,1607,7-140-6530_6605-21514–,00.html) and the Ohio Department of Taxation (http://www.tax.ohio.gov/tax_analysis/tax_data_series/publications_tds_property.aspx).

Community characteristics data from the 2000 U.S. Census were obtained from the National Center for Education Statistics (http://nces.ed.gov/). Data on student characteristics, enrollment, and graduation rates were obtained from the State of Michigan’s Center for Educational Performance and Information (http://www.michigan.gov/cepi). The Michigan Educational Assessment Program (Grades 3-9) and Michigan Merit Examination (Grade 11) results are available through the Michigan Department of Education (http://www.michigan.gov/mde/0,1607,7-140-22709_31168—,00.html). The corresponding data for Ohio are all available through the Ohio Department of Education website (http://www.ode.state.oh.us/GD/Templates/Pages/ODE/ODEDefaultPage.aspx?page=1).

Estimating School Building Capital

This article estimates the capital assets for buildings in each school district in Michigan and Ohio based on the financial reports covering the 2007-2008 fiscal year. As mentioned above, some districts were dropped, but this analysis was based on a high percentage of pupils. Of Michigan’s 552 public school districts,⁴ 550 submitted figures for building capital and related depreciation in their CAFRs. The two districts in Michigan without building capital figures together contained only 85 students, giving a coverage rate of over 99%. A third Michigan school district did not provide a pupil count, but based on its other characteristics it was presumed small. Ohio contains 614 traditional public school districts, of which 490 showed accumulated capital assets, giving a coverage rate of over 88% of the pupils. For a detailed exposition on the capital asset calculation and more background on the Government Accounting Standards Board, see Arsen and Davis (2006). What follows is a summary of the method used to calculate school building capital and depreciation (Arsen & Davis, 2006).

The school districts’ financial reports are based on historic expenditure, so the capital assets must be adjusted for inflation. This calculation is complicated by the fact that school districts’ expenditures are spread across multiple years. Equation (1) is based on the simplifying assumption that all building expenditures occurred at a single point in the past. The CAFRs detail the assumed life of buildings and they use straight-line depreciation throughout this lifetime, denoted K-Life. One can then use depreciation as a percentage of historic expenditure to place the capital expenditure at some point on the K-Life timeline. The larger the ratio of depreciation to historic expenditure, the larger the necessary adjustment for inflation.

This “capital vintage” factor, t, for district i is calculated as follows:

t = 2008 - [(K-Life) \times (\frac{{depreciation}_{i}}{{depreciation}_{i} + {capital}_{i}})]

Figure 1 shows a histogram of the average capital vintages for the 549 Michigan school districts used in this study and Figure 2 shows a histogram for Ohio.⁵

Figure 1.

Frequency distribution for capital vintage in Michigan.

Figure 2.

Frequency distribution for capital vintage in Ohio.

The New Education Buildings price index from the U.S. Department of Commerce is used to adjust the nominal capital stock for inflation. Given t from Equation (1), the value of district i’s capital stock in 2008 dollars is given by

{Capital Stock}_{i}^{2008} = {capital}_{i}^{t} \times (\frac{{bldg_index}^{2008}}{{bldg_index}^{t}})

where ${capital}_{i}^{t}$ = The building capital shown in the district i’s 2008 CAFR, bldg_index²⁰⁰⁸ = The building cost index for 2008, bldg_index^t = the building cost index for year t, where t is given by Equation (1).

Finally, these ${Capital Stock}_{i}^{2008}$ values were divided by the pupil counts in each district to give a district’s per-pupil capital stock.

Breaking the Data Into Quintiles

Before removing the school districts with insufficient data, the districts in the samples were divided into quintiles based on the total taxable value of real property per pupil. The splits were made to produce roughly the same number of districts in each quintile, not equal numbers of pupils, so quintiles containing large urban districts contain more pupils. Approximately 20% of the districts on Ohio were omitted due to the accounting method they used in their financial reports, and districts in the lower quintiles were more likely to be dropped. Nevertheless, the sample of districts in each quintile was substantial, ranging from 73% in the lowest quintile up to 89% for the highest.⁶ Moreover, the omitted districts only contained 11.7% of the pupils. The analysis calculates the average capital stock per pupil for each state and for each quintile within each state.

Education Production Function Models

Student outcome data are notoriously noisy and researchers have had a difficult time showing a statistical correlation between school facilities and student achievement (Earthman & Lemaster, 1996; McGuffey, 1982; Weinstein, 1979). This study employs an education production function model to estimate the impact of school building capital on 11th-grade math scores and high school graduation. The math variable is the percentage of students who meet or exceed the state’s standards. To help eliminate some of the statistical noise, the regressions were run on the top and bottom quintiles only, with a dummy variable to indicate the top quintile.

The estimated model is given by the following equation:

\begin{array}{l} Y = β_{0} + β_{1} * ln (Capital / Pupil) + β_{2} * ln (COE / Pupil) + β_{3} * FRL + β_{4} * (SpEd / Pupil) + \\ β_{5} * (Pupil Growth Rate) + β_{6} * (Q5 dummy) + β_{7} * (Q5 dummy * ln (Capital / Pupil)) + ε \end{array}

where Y is the graduation rate or the percentage of 11th-grade students who met or exceeded the state’s benchmark for mathematics; Capital/Pupil is the capital assets per pupil as calculated by the algorithm used in this study; COE/Pupil is the current operating expenditure per pupil; FRL is the percentage of pupils in the district eligible for free or reduced-price meals; SpEd/Pupil is the state-administered federal funding for special education pupils; Pupil Growth Rate is the 5-year growth rate in fall membership for each school district; Q5 dummy is a dummy variable for the districts in the fifth quintile; and Q5 dummy * ln(Capital/Pupil) is a cross-product that estimates a different slope for the effect of capital for the districts in the fifth quintile.

The natural logarithms of capital funding and current operating expenditure reflect diminishing marginal returns to these inputs because the logarithmic transformation gives proportionally less weight to large values. Consistent with the notion of decreasing returns in economics (cf. Mankiw, 2011), an additional dollar per pupil spent in poor districts will have a larger impact than if it were spent in a wealthy district. Eleventh graders were chosen because they have had significant exposure to the school districts’ facilities, on average. Mathematics scores were chosen to minimize regional differences in English language ability.

Measures of Variation

The analysis will show three common measures of variation, the federal range ratio, the coefficient of variation, and the Verstegen Index. The federal range ratio is given by the difference between the 95th and the 5th percentile of a sample of a particular variable divided by the 5th percentile. The coefficient of variation is the standard deviation of a sample divided by the mean. The Verstegen Index sums all of the values of a sample that are above the median and divides that sum by the number of those observations multiplied by the median. In essence, the Verstegen Index divides the top half of a distribution by the value if all of those large values were at the median value (Odden & Picus, 2007; Wood, 2007). All three of these measures get smaller as the underlying distribution gets more equitable.

These three measures of variation are calculated for five indicators of community resources: taxable value of real property per pupil, capital per pupil (as calculated in this study), median house value and median income from the 2000 U.S. Census, and current operating expenditure per pupil. The taxable value of land reflects a community’s ability to pay for schools. The capital stock and median house value reflect accrued assets, whereas the median income is a more transitory value that reflects households’ income. Current operating expenditure reflects what a community is currently spending per year on schools.

Empirical Results

Table 1 shows a comparison of community and school characteristics between Michigan and Ohio. Ohio is further broken down by those districts included in the analysis and those excluded because of the accounting method. There is no reason to suspect that the districts in Ohio that were excluded are different enough to raise concerns over selection bias. The districts appear slightly poorer but their student outcomes are comparable. The Michigan districts are slightly wealthier than those in Ohio but the large difference in taxable value of land is offset by Ohio’s higher millage rates (Ohio Department of Taxation). In addition, Michigan achievement is higher in the lower grades but it falls significantly below the levels in Ohio by 11th grade. This is consistent with the lower high school graduation rate for Michigan school districts.

Table 1.

Summary Statistics.

	Ohio		Michigan
	Excluded	Included	Michigan
Districts, n (%)	124 (20.2)	490 (79.8)	549
Pupils, n (%)	196,431 (11.7)	1,476,530 (88.3)	1,541,215
QSCB awards, $ (%)	14,542,689 (6.1)	223,139,986 (93.9)	203,705,000
Taxable value per pupil, $	138,886	155,054	229,579
Current operating expenditure per pupil, $	9,738	9,828	9,690
Median house value, $	95,175	111,742	121,169
Median income, $	39,916	45,571	48,479
% Women with high school diploma	82.2	83.8	84.6
% Households with children <18 years	47.9	48.0	48.8
% Met standards 5th-grade reading	71.3	74.7	82.3
% Met standards 5th-grade math	61.3	64.3	75.1
% Met standards 11th-grade reading	92.1	92.7	61.0
% Met standards 11th-grade math	89.0	90.1	45.0
High school graduation rate	90.2	89.5	78.9

Note. QSCB = Qualified School Construction Bonds.

Breakdown of Average Levels of Resources by Taxable Value Quintiles

Table 2 shows the distribution of districts, pupils, and capital assets for Michigan. There is an enormous difference in the ability to pay for schools, reflected by the taxable value per pupil. The upper quintile has almost 3.9 times the taxable value of real property compared with the lowest quintile. This has translated into a 37% larger capital stock per pupil. The situation is rather different in Ohio.

Table 2.

Breakdown by Quintiles of Taxable Value for Michigan.

District Taxable Value per Pupil Quintile	Number of Districts	Number of Pupils	Taxable Value per Pupil ($)	Capital Stock per Pupil ($)
1 (low)	111	317,042	105,503	20,587
2	111	228,200	164,742	22,115
3	110	332,764	203,720	22,273
4	109	394,183	265,759	25,579
5 (high)	108	268,941	409,746	28,160
State	549	1,541,130	229,565	23,775

Table 3 shows that the difference in ability to pay is still quite large in Ohio, with the top quintile having almost three times the taxable value of land. In spite of this difference, the highest quintile is only about 4% above the lowest in capital stock per pupil. Clearly the Ohio School Facilities Commission is having an equalizing effect, at least for the poorest quintile of districts. It is also worth noting that the U-shaped distribution of capital assets across quintiles in Ohio shows the middle three quintiles well below their counterparts in Michigan.

Table 3.

Breakdown by Quintiles of Taxable Value for Ohio.

District Taxable Value per Pupil Quintile	Number of Districts	Number of Pupils	Taxable Value per Pupil ($)	Capital Stock per Pupil ($)
1 (low)	89	163,382	77,757	23,392
2	90	163,857	102,700	17,905
3	96	338,290	121,929	16,386
4	107	373,872	156,715	15,522
5 (high)	108	437,129	227,784	24,236
State	490	1,476,530	155,054	19,435

Education Production Functions

Table 4 shows the estimated regression coefficients for Michigan and Ohio. The dependent variable is the high school graduation rate. In both states, the log of capital per pupil is positively correlated with the graduate rate. In Michigan, this estimate is statistically significant at the 90% level; in Ohio, it is significant at the 99% level. The log of current operating expenditures is negatively correlated with graduation rates, indicating that at-risk students graduate at a lower rate in spite of higher expenditure. What is interesting is that the poverty rate, reflected by the percentage of pupils eligible for free and reduced-price meals, is negatively correlated with graduations rates in Michigan and positively correlated with graduation rates in Ohio. Both these results are statistically significant at the 95% level (higher in the case of Ohio). This seems to indicate that Ohio has broken the link between poverty and high school dropouts, controlling for the other variables in the model.

Table 4.

Education Production Function: Graduation Rates (Equation 3).

Independent Variables	Michigan (Standardized Coefficients)	Ohio (Standardized Coefficients)
Constant
ln(Capital/Pupil)	0.157* (1.709)	0.222*** (2.935)
ln(COE/Pupil)	−0.046 (−0.572)	−0.201** (−2.384)
FRL percentage	−0.409*** (−5.476)	0.147** (2.067)
SpEd/Pupil	−0.075 (−1.183)	−0.136** (−2.015)
5-Year pupil growth rate	−0.129* (−1.676)	0.077 (1.035)
Quintile 5 dummy	2.092 (1.398)	1.546 (1.558)
Q5 dummy * ln(Capital/Pupil)	−1.996 (−1.306)	−1.186 (−1.187)
n	200	197
R ²	.246	.293

Note. COE/Pupil = current operating expenditure per pupil; FRL = percentage of pupils in the district eligible for free or reduced-price meals; SpEd/Pupil = state-administered federal funding for special education pupils; Capital/Pupil = capital assets per pupil as calculated by the algorithm used in this study; Q5 dummy = dummy variable for the districts in the fifth quintile; Q5 dummy * ln(Capital/Pupil) = cross-product that estimates a different slope for the effect of capital for the districts in the fifth quintile.

p < .10. **p < .05. ***p < .01 (t statistics in parentheses).

Table 5 shows a similar regression analysis where 11th-grade math scores are the dependent variable. Again, Ohio shows a positive and statistically significant correlation between math scores and poverty. It is worth noting that the signs and significance are similar to those in Table 4. In addition, the production function explains considerably more variation for math scores, as reflected in the R² statistic, than it did for graduation rates.

Table 5.

Education Production Function: 11th-Grade Math Scores (Equation 3).

Independent Variables	Michigan (Standardized Coefficients)	Ohio (Standardized Coefficients)
Constant
ln(Capital/Pupil)	0.147** (2.041)	0.097 (1.417)
ln(COE/Pupil)	−0.227*** (−3.534)	−0.320*** (−4.196)
FRL percentage	−0.595*** (−9.901)	0.224*** (3.495)
SpEd/Pupil	0.005 (0.104)	−0.125** (−2.060)
5-Year pupil growth rate	−0.091 (−1.493)	0.021 (0.313)
Quintile 5 dummy	−1.136 (0.926)	0.395 (0.44)
Q5 dummy * ln(Capital/Pupil)	1.332 (1.065)	0.095 (0.105)
n	200	197
R ²	.524	.423

p < .10. **p < .05. ***p < .01 (t statistics in parentheses).

There is a difference in the way Ohio and Michigan administer their high school mathematics exams. Students first take the Ohio Graduation Test for mathematics in the spring of 10th grade. Those students who fail to achieve a proficient score take the exam again in 11th grade. Therefore, the scores used in this analysis for Ohio are cumulative: The percent proficient includes those who were proficient in 10th grade as well as those who achieved a proficient score on their second attempt, in 11th grade (Ohio Department of Education, Office of Assessment).⁷ In Michigan, the students take the mathematics exam for the first time in 11th grade.⁸

Variation in School Resources

The measures of variation use all the school districts in each state’s sample together, not by quintile. In Michigan, the taxable value has the largest value of the three inequality measures, as you might expect. Figure 3 shows that the capital per pupil is the next most unequal, followed by the median house value, median income, and current operating expenditure per pupil. This result is consistent across the measures. Surprisingly, Ohio shows the greatest inequality in the capital stock per pupil, followed by taxable value, median house value, median income, and current operation expenditure per pupil. It is likely that this is a result of the facility commission’s directing resources to the poorest districts as indicated by the U-shaped distribution of capital assets across the quintiles in Ohio (see Table 3). It appears that the reallocation of Ohio’s resources to the poorest districts had created a paradox where the average level of capital assets across the quintiles is relatively equal (at least with respect to the highest and lowest), yet the variation across districts is quite high. Figure 4 shows the equity statistics for Ohio.

Figure 3.

Equity statistics for Michigan.

Figure 4.

Equity statistics for Ohio.

As a final comparison, Figure 5 shows the Verstegen Indices for Michigan and Ohio. As the Verstegen Index approaches 1, the underlying distribution of the variable is more equitable. Apart from the taxable value per pupil, the states are remarkably similar with regard to the measures of equity. The school capital stock and the community housing stock are the more unequal of the measures. The flow variables, represented by income and annual expenditure for current operating expenses, were far more equal. The numbers supporting the graphs are provided in Tables 6 and 7.

Figure 5.

Verstegen indices for resources across states.

Table 6.

Equity Statistics for Michigan School Districts.

Index	Taxable Value per Pupil	Capital per Pupil	Median House Value	Median Income	Current Operating Expenditure per Pupil
Federal range ratio	5.68	4.33	2.11	1.29	0.71
Coefficient of variance	2.31	0.51	0.38	0.27	0.29
Verstegen Index	2.36	1.48	1.40	1.27	1.22

Table 7.

Equity Statistics for Ohio School Districts.

Index	Taxable Value per Pupil	Capital per Pupil	Median House Value	Median Income	Current Operating Expenditure per Pupil
Federal range ratio	2.46	4.90	2.00	1.38	0.63
Coefficient of variance	0.49	0.57	0.36	0.27	0.19
Verstegen Index	1.48	1.57	1.35	1.25	1.17

The Distribution of Qualified School Construction Bonds

Michigan and Ohio made available the distribution of $203,705,000 and $237,682,675 in QSCB, respectively, in 2009. The distributions in Michigan were sent to the author directly; the distributions for Ohio are available through the Web. The QSCBs provide interest-subsidized loans to school districts for the construction and renovation of school facilities. These loans come with a number of conditions, such as Michigan’s requirement that 10% of the funds be spent within 6 months and 100% spent within 3 years of the date of the issuance of the bond (http://www.michigan.gov/mde/0,1607,7-140-5236_6048-217328–,00.html). It is conceivable that a number of high-need districts may not have had the resources to consult with legal council to discuss all of the conditions and submit a proposal. While it is too early to assess the impact of the bonds on student achievement, particularly since some of the construction has yet to be completed, it is possible to compare the resources available to recipient and nonrecipient districts.

The first thing to note is the number of students who were affected by the bonds. In Michigan, the allocations for 2009 went to school districts that contained 5.1% of the state’s traditional public school pupils. In Ohio, the recipient districts contained 6.8% of the pupils.⁹ Table 8 shows a comparison of community characteristics and school resources between recipient and nonrecipient districts in Michigan and Ohio.

Table 8.

Comparison of School Resources and Community Characteristics.

	Michigan		Ohio
	Received QSCB	Did Not Receive QSCB	Received QSCB	Did Not Receive QSCB
Pupils	74,152	1,466,978	94,116	1,382,414
Taxable value/pupil	$251,887	$228,436	$175,184	$153,684
Capital/pupil	$20,612	$23,935	$18,708	$19,484
Median house value	$134,784	$120,484	$123,670	$110,930
Median income	$53,115	$48,246	$50,393	$45,242
% Women with high school diploma	84.3%	84.6%	86.0%	83.7%
% Households with children <18 years	47.0%	48.9%	47.8%	48.0%

Note. QSCB = Qualified School Construction Bonds.

In both states, the recipient districts had lower values of capital per pupil. It is also clear that they had greater resources available to them. Demographically, the recipient groups did not differ significantly from the nonrecipients, as shown in the women’s education and percentage of households with children younger than 18 years. But the recipient districts did have higher median incomes, house values, and substantially more taxable value of real property. This disparity suggests that perhaps these districts could have met their capital needs locally.

The goal of equalizing school facilities across districts is to improve student achievement. Table 9 shows a comparison of student achievement between recipient and nonrecipient school districts in the two states. In Michigan, the math scores and graduation rates are higher in the recipient districts—the graduation rate by quite a bit. In Ohio, all the test scores are higher in the recipient districts as is the graduation rate. There are too few scores to infer statistical significance, but 10 of the 12 test scores, graduation rates, the taxable value of land, the median house value, and median incomes are all higher in the recipient districts.

Table 9.

Comparison of Student Achievement.

	Michigan		Ohio
	Received QSCB	Did Not Receive QSCB	Received QSCB	Did Not Receive QSCB
5th-Grade reading	83.2%	82.2%	79.3%	74.4%
5th-Grade math	76.9	75.0	67.4	64.1
8th-Grade reading	77.2	77.3	84.4	81.1
8th-Grade math	75.2	72.0	79.1	75.3
11th-Grade reading	60.6	61.0	93.5	92.7
11th-Grade math	46.1	45.0	91.7	90.0
High school-graduation rate	85.3	78.6	92.7	89.3

Note. QSCB = Qualified School Construction Bonds.

Implications for Policy, Practice, and Research

Similar in many ways, Ohio and Michigan offer an opportunity to compare the effects of two different policies toward school facilities that appear to have countervailing incentives. Michigan has centralized the funding for school operating expenditures through the state’s foundation allowance, while at the same time it has left the funding for capital investment to the local communities. Litigation has compelled Ohio to address the inequities in school facilities across communities. This study provides a direct comparison of school facilities across the two states in order to shed some light on the efficacy of Ohio’s policy to improve and equalize the quality of school facilities. In addition, the analysis compares the districts that received federally subsidized school construction bonds with those that did not in order to further assess the equity of school facility funding in the states and to inspire debate about whether the tight filing restrictions for the program may have discouraged resource-poor districts from applying for the bonds.

Looking at average levels by quintile of taxable value, Michigan has large disparities in the ability to pay for facilities and the average capital stock across quintiles. These differences give some credence to the claim that while Proposal A equalized operational expenditures across Michigan school districts, wealthy districts are investing more in facilities than poor districts. Ohio shows a very different pattern: The disparities in ability to pay are there, but the top quintile has only about 4% more school capital per pupil than the bottom quintile districts.

The root of most court cases that include facilities is their impact on student performance. Student test scores provide a snapshot of student performance for a given year whereas graduation rates represent more of a stock variable reflecting the culmination of students’ relationships with the school district. This analysis employed a parsimonious production function model to show the impact of school districts’ capital stocks on both these variables in order to compare the strength of the correlations for the two different policy environments. The correlations between school capital and student graduation rates were positive and statistically significant in both states, though the correlation was stronger in Ohio. The correlations between capital and 11th-grade math scores were positive, but only statistically significant for Michigan. What was remarkable about the production function models was the fact that Ohio appeared to have broken the nexus between poverty and student performance, controlling for the other variables in the model. This deserves further research.

The variation in per-pupil school capital was high in both states, particularly compared with other measure of community and school resources. In Ohio, this has not necessarily translated into large discrepancies in the average level of school capital per pupil, as is the case in Michigan. It appears that the Ohio School Facilities Commission has to some extent leveled the playing field and also weakened the link between poverty and the quality of school facilities.

The Ohio School Facilities Commission also has the responsibility of administering the QSCB program. While the main goal of the QSCB was to stimulate the economy, it appears that the distribution of construction bonds could have been more equalizing. In both states, districts that received the bonds had higher wealth, a greater ability to pay, and better student performance measures. A subsequent analysis involving more states would show how states’ disbursement of federal construction bond subsidies affects the equity of school facility funding.

This analysis employs a straightforward method to calculate the capital stock using data that are publicly available for school districts nationwide. The Government Accounting Standards Board directive has standardized accounting procedures so that a direct comparison is possible, unlike previous studies that relied on ad hoc facility surveys or engineering reports. Moreover, the method generates plausible estimates that future researchers can employ in studies of school finance adequacy and to analyze the determinants of school district capital assets. A comparison of Michigan and Ohio would be both costly and time consuming if facility data had to be collected in person. This method allows researchers to use the Web to gather all the data. In addition, the accountability provisions of the ARRA have made those funds fairly easy to track.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Thomas E. Davis, PhD, is an assistant professor in the College of Education at the University of Maryland. His research interests include education policies that affect the funding for school facilities and educational technology as well as the financial burden of providing employee benefits to public school employees.

References

Arsen

Clay

Davis

Devaney

Fulcher-Dawson

Plank

(2005). Adequacy, equity, and capital spending in Michigan schools: The unfinished business of proposal A. East Lansing: Michigan State University, Education Policy Center. Retrieved from http://education.msu.edu/epc/forms/Arsen_et_al_2005_Adequacy_Equity_and_Capital_Spending_in_Michigan_Schools.pdf

Arsen

Davis

(2006). Taj Mahals or decaying shacks: Patterns in local school capital stock and unmet capital need. Peabody Journal of Education, 81(4), 1-22.

Arsen

Plank

(2003). Michigan school finance under Proposal A: State control, local consequences. East Lansing: Michigan State University, Education Policy Center. Retrieved from http://education.msu.edu/epc/forms/Arsen_et_al_2003_Proposal_A.pdf

Barnett

Vidal

(2012). State and local government finances: Summary: 2010. Washington, DC: U.S. Census Bureau. Retrieved from http://www2.census.gov/govs/local/summary_report.pdf

Berner

(1993). Building conditions, parental involvement, and student achievement in the District of Columbia public school system. Urban Education, 28, 6-29.

Bowers

Urick

(2011). Does high school facility quality affect student achievement? A two-level hierarchical linear model. Journal of Education Finance, 37, 72-94.

Branham

(2004). The wise man builds his house upon the rock: The effects of inadequate school building infrastructure on student attendance. Social Science Quarterly, 85, 1112-1128.

Buckley

Schneider

Shang

(2004). The effects of school facility quality on teacher retention in urban school districts. Washington, DC: National Clearinghouse for Educational Facilities. Retrieved from http://www.ncef.org/pubs/teacherretention.pdf

Cellini

S. R.

Ferreira

Rothstein

(2010). The value of school facility investments: Evidence from a dynamic regression discontinuity Design. Quarterly Journal of Economics, 125, 215-261.

10.

Duflo

(2001). Schooling and labor market consequences of school construction in Indonesia: Evidence from an unusual policy experiment. American Economic Review, 91, 795-813.

11.

Duran-Narucki

(2008). School building condition, school attendance, and academic achievement in New York City public schools: A mediation model. Journal of Environmental Psychology, 28, 278-286.

12.

Earthman

Cash

Van Berkum

(1995). Student achievement and behavior and school building condition. Journal of School Business Management, 8(3), 26-37.

13.

Earthman

Lemaster

(1996). Review of research on the relationship between school buildings, student achievement, and student behavior. Paper presented at the annual meeting of the Council of Educational Facility Planners International, Tarpon Springs, FL.

14.

Erlichson

(2001). New schools for a new millennium: Court-mandated school facilities construction in New Jersey. Journal of Education Finance, 27, 663-682.

15.

Filardo

Cheng

Allen

Bar

Ulsoy

(2010). State capital spending on PK-12 school facilities. 21st Century School Fund. Retrieved from http://www.21csf.org/csf-home/Documents/FederalStateSpendingNov2010/StateCapitalSpendingPK-12SchoolFacilitiesReportNov302010.pdf

16.

Gronberg

Jansen

Taylor

(2011). The impact of facilities on the cost of education. National Tax Journal, 64, 193-218.

17.

Harter

(1999). How education expenditures relate to student achievement: Insights from Texas elementary schools. Journal of Education Finance, 24(Winter), 281-302.

18.

Hanushek

E. A.

(1996). School resources and student performance. In Burtless

(Ed.), Does money matter: The effect of school resources on student achievement (pp. 43-73). Washington, DC: Brookings Institution Press.

19.

Hanushek

E. A.

(1997). Assessing the effects of school resources on student performance: An update. Educational Evaluation and Policy Analysis, 19, 141-164.

20.

Hawkins

Lilley

(1992). Guide for school facility appraisal. New York, NY: Council of Educational Facility Planners, International.

21.

Jones

Zimmer

(2001). Examining the impact of capital on academic achievement. Economics of Education Review, 20, 577-588.

22.

Jones

(2011, August 18). Please, Mr. President, don’t build us new schools. Dallas Morning News. Retrieved from http://dallasmorningviewsblog.dallasnews.com/archives/2011/08/please-mr-presi.html

23.

Kumar

O’Malley

Johnston

(2008). Association between physical environment of secondary schools and student problem behavior. Environment and Behavior, 40, 455-486.

24.

Lockwood

(2002, December). School finance reform in Michigan proposal A: Retrospective. Retrieved from http://www.michigan.gov/documents/propa_3172_7.pdf

25.

Mankiw

(2011). Principles of microeconomics (6th ed.). Mason, OH: South-Western Cengage Learning.

26.

McGuffey

(1982). Facilities. In Walberg

(Ed.), Improving educational standards and productivity (pp. 237-281). Berkeley, CA: McCutchan.

27.

McGuffey

Brown

(1978). The impact of school building age on school achievement in Georgia. Education Facilities Planner, 2, 5-19.

28.

Nelson

Zimmerman

(2011). The effect of school construction on test scores, school enrollment, and home prices (Discussion Paper). Bonn, Germany: Institute for the Study of Labor (IZA). Retrieved from http://ftp.iza.org/dp6106.pdf

29.

Odden

Picus

(2007). School finance: A policy perspective (4th ed.). Boston, MA: McGraw-Hill.

30.

Ohio Department of Education, Office of Assessment. (2008). March 2008 OGT statistical summary. Retrieved from http://education.ohio.gov/Topics/Testing/Statistical-Summaries-and-Item-Analysis-Reports

31.

Picus

Marion

Calvo

Glenn

(2005). Understanding the relationship between student achievement and the quality of educational facilities: Evidence from Wyoming. Peabody Journal of Education, 80(3), 71-95.

32.

Rebell

(2009). Courts and kids: Pursuing educational equity through the state courts. Chicago, IL: University of Chicago Press.

33.

Rice

(2004). Equity and efficiency in school finance reform: Competing or complementary goods? Peabody Journal of Education, 72, 134-151.

34.

Roberts

(2009). Measuring school facility conditions: An illustration of the importance of purpose. Journal of Educational Administration, 47, 368-380.

35.

Sciarra

Bell

Kenyon

(2006). Safe and adequate: Using litigation to address inadequate K-12 school facilities. Education Law Center. Retrieved from http://www.edlawcenter.org/assets/files/pdfs/publications/Safe_and_Adequate.pdf

36.

Superfine

(2008). The courts and standards based reform. New York, NY: Oxford University Press.

37.

Thomas

Cambron-McCabe

McCarthy

(2009). Public school law: Teachers’ and students’ rights (6th ed.). Boston, MA: Pearson.

38.

Uline

C. L.

Tschannen-Moran

(2008). The walls speak: The interplay of quality facilities, school climate, and student achievement. Journal of Educational Administration, 46, 55-73.

39.

Uline

C. L.

Tschannen-Moran

Wolsey

T. D.

(2009). The walls still speak: The stories occupants tell. Journal of Educational Administration, 47, 400-426.

40.

Weinstein

(1979). The physical environment of the school: A review of the research. Review of Educational Research, 49, 577-610.

41.

Wood

(2007). Educational finance law (3rd ed., Monograph No. 75). Cleveland, OH: Education Law Association.

42.

Woolner

Hall

Higgins

McCaughey

Wall

(2007). A sound foundation? What we know about the impact of environments on learning and the implications for building schools for the future. Oxford Review of Education, 33, 47-70.