School finance researchers have been evaluating and comparing district- and school-level expenditures for decades, drawing largely on regression-based approaches which account for differences in needs and costs across settings, districts and schools.[i] With charter schools introduced into the mix over the past several decades, researchers have extended those methods to study differences in spending between district and charter schools serving otherwise similar student populations. The most thorough example, and specific application of this approach is the study conducted on behalf of the Maryland Department of Education in 2016.

Step 1: Matching the Dollars to the Students

The first step in the process is ensuring that the right revenues and expenditures (numerator) are attached to the right students (denominator) when calculating per-pupil resources. This is a fatal flaw – egregious error – repeated time and time again – in the often cited University of Arkansas Department of Education Reform charter funding gap reports. Where charter schools are fiscally dependent on public districts (as in most of the locations addressed by the authors), some revenues sent to and spent by districts are spent on services for children attending charter schools. If we leave those in the district’s funding numerator, but take those pupils out (as they are in charter schools) that overstates district per-pupil funding and understates resources to charter schools. Further, some district revenue sources may be dedicated to other services outside of their own schools—be they community services or students tuitioned elsewhere.

Step 1 is to get a comparable, comprehensive school site spending figure for both district and charter schools, likely excluding special schools or services (those served and the resources spent) from the comparisons. A detailed explanation of the process of achieving a comparable spending figure is explained on page 9, section 2 of the AIR study referenced in this report (http://marylandpublicschools.org/stateboard/Documents/01242017/TabG-CharterPublicSchoolFundingStudy.pdf).

Figure 1 is adapted from a report for the Maryland Department of Education, where the goal was to determine the “commensurate” expenditures of district and charter schools in that state. Maryland charter schools, like those in many states are fiscal dependents of local districts: they are funded by a pass-through system, where local tax revenues and state aid “pass through” the district and to the charter for each resident student enrolled. The district also retains responsibility for the direct provision of some services for charter schools, including transportation and system-wide enrollment management. While the Maryland report focuses on comparing charter and district school spending, its broader goal is establishing data standards and methods for better evaluating equitable allocation of resources across schools within districts, where that context includes a variety of “mission centers,” including charter schools.

As Figure 1 shows, the majority of public/taxpayer financing for the system goes first to the school district. Only a portion of that funding then flows to typical (“regular,” “traditional”) general purpose schools. Districts provide a wide array of services, including providing special schools, where necessary, for children with significant special needs, and alternative placements or schools for children removed from general purpose schools due to disciplinary actions (as is required in New Jersey). Often, students with significant special needs are placed in schools outside the district’s direct control; the district, however, still retains the financial responsibility and pays tuition for the student. Districts also provide an array of community services, make facilities available for community organizations, subsidize transportation and textbooks for private school children (in many states), and provide for transportation and special education supports for children attending charter schools (in many states).

Figure 1

When evaluating resource equity across schools, the goal is to have as complete and comparable a measure as possible of the resources actually available to specific schools, based on the characteristics of students served in those schools. Doing so requires knowing not only which resources are assigned directly to individual schools – per Figure 1– but also identifying which other district services provide support to which schools, or provide support to activities that aren’t connected with individual schools – like providing community services. It would be inappropriate, for example, to take the total revenues received by the district, divided by the district’s own pupils, and compare those to the funding allocated to the charter school, divided by the charter schools’ pupils. This is because some of that revenue received by the district (even after subtracting direct transfers to charters) includes district spending on activities that serve charters, and funds obligated for support of private school students. District revenues allocated to special education schools, which serve student populations that differ substantially from those enrolled in charter schools, will operate at higher per pupil costs than charters.

These complications do not relate exclusively to comparisons between district and charter schools; they relate to any attempts to evaluate the equitable distribution of resources across schools within districts. Once we have isolated the comparable per pupil resources for all schools within a geographic space there will still exist important differences across student populations that must be accounted for when comparing school resources. This is true even when comparing the subset of “general purpose” district-operated schools and charter schools (most of which in New Jersey are “general purpose”).

Additional explanation provided in this report: https://www.njpp.org/wp-content/uploads/2020/11/NJPP-School-Funding-in-New-Jersey-A-Fair-Future-for-All-Part-5.pdf

---

Step 2: Modeling Spending Variation with Respect to Cost and Need Factors     

The second step is to use that comparable spending figure (spending per pupil, school site) as the dependent variable in a regression model which accounts for a standard, well-known and frequently used set of factors. This is the approach used in the Maryland and California studies, as well as several peer-reviewed articles evaluating school site spending variation (whether focused on charter schools or not). The standard model is:

Spend = f(% Low Income, %ELL, % SWD LI/HC, % SWD HI/LC, % Grades 6 to 8, % Grades 9 to 12, Geographic Location, Year, Control*)

That is, spending is modeled as a function of the share of children from low-income families (using a measure set to an income threshold sufficient to capture variation across schools), % who are English language learners, % students with disabilities preferably at least in two groups by severity, % in different grade ranges such as to compare schools of similar grade range, and if beyond a single metropolitan area, some geographic indicator to capture labor cost differences.  To determine whether charter schools are funded differently than TPS, one can include a dummy variable on charter status (control).

Table 1 provides an illustration with Maryland data. Table 1 shows that a school with 100% children from low-income families spends about $1,500 more per pupil than a school with 0% children from low-income families. A school with 100% ELL children spends only about $360 more than a school with 0% children who are ELLs. Special education populations, in the aggregate are by far the largest driver of spending differences with a school having 100% children with disabilities expected to spend nearly $22,000 per pupil more than a school with 0% children with disabilities. Notably, however as the share of those children with disabilities who are in the mild/moderate category increases, the overall spending margin decreases. Finally, charter schools are spending approximately $630 more per pupil than district schools—in the same district (fixed effect)—and serving otherwise similar student populations.

Table 1. Model of Maryland School Site Spending 2013-2015, Includes LEA Fixed Effect [schools weighted for enrollment. Estimated with Robust Standard Errors clustered on School]

	(1)
VARIABLES	Commensurate Expense per Pupil
charter	630.360*
	(181.284)
% school enrollment in grades 6 to 8	850.170*
	(84.529)
% school enrollment in grades 9 to 12	558.609*
	(89.590)
Percent Special Education	21,929.519*
	(1,132.973)
% Students with Disabilities that are Non-Severe Disabilities	-1,212.161*
	(361.059)
Percent ESL	358.567
	(435.256)
Percent Low Income	1,515.191*
	(244.471)
year = 2014	183.814*
	(19.534)
year = 2015	263.468*
	(27.582)
Constant	8,475.939*
	(410.742)
Observations	3,966
R-squared	0.504
Robust standard errors in parentheses
* p<0.05

---

[i] Baker, B.D. (2009). Within-district resource allocation and the marginal costs of providing equal educational opportunity: Evidence from Texas and Ohio. Education Policy Analysis Archives, 17(3). Retrieved August 10, 2021, from https://www.redalyc.org/pdf/2750/275019727003.pdf

Chambers, J.G., Levin, J.D., & Shambaugh, L. (2010). Exploring weighted student formulas as a policy for improving equity for distributing resources to schools: A case study of two California school districts. Economics of Education Review, 29(2), 283-300.

Chambers, J., Shambaugh, L., Levin, J., Muraki, M., & Poland, L. (2008). A tale of two districts: A comparative study of student-based funding and school-based decision making in San Francisco and Oakland Unified School Districts. American Institutes for Research.

Atchison, D., Baker, B., Levin, J., & Manship, K. (2017). Exploring the quality of school-level expenditure data: Practices and lessons learned in nine sites. Office of Planning, Evaluation and Policy Development, US Department of Education. Retrieved August 10, 2021, from https://files.eric.ed.gov/fulltext/ED584614.pdf

Baker, B.D., & Weber, M. (2016). State school finance inequities and the limits of pursuing teacher equity through departmental regulation. Education Policy Analysis Archives/Archivos Analíticos de Políticas Educativas, 24(37), 1-36.

Baker, B.D., Libby, K., & Wiley, K. (2015). Charter school expansion and within-district equity: Confluence or conflict?. Education Finance and Policy, 10(3), 423-465.

Toutkoushian, R.K., & Michael, R.S. (2007). An alternative approach to measuring horizontal and vertical equity in school funding. Journal of Education Finance, 32(4), 395-421.

Kolbe, T., Baker, B.D., Atchison, D., Levin, J., & Harris, P. (2021). The additional cost of operating rural schools: Evidence from Vermont. AERA Open, 7. Retrieved August 10, 2021, from https://journals.sagepub.com/doi/pdf/10.1177/2332858420988868

Berne, R., & Stiefel, L. (1994). Measuring equity at the school level: The finance perspective. Educational Evaluation and Policy Analysis, 16(4), 405-421.

Stiefel, L., Rubenstein, R., & Berne, R. (1998). Intra-district equity in four large cities: Data, methods and results. Journal of Education Finance, 23(4), 447-467.

Stiefel, L., Rubenstein, R., & Berne, R. (1998). Intra-district equity in four large cities: Data, methods and results. Journal of Education Finance, 23(4), 447-467.

Knight, D.S., & Toenjes, L.A. (2020). Do charter schools receive their fair share of funding? School finance equity for charter and traditional public schools. Education Policy Analysis Archives, 28(51), 1-40.

There are days when I’m perplexed at how far we have NOT come in understanding (and conveying in policy circles) why we include measures of poverty and other student needs, along with a variety of “cost” factors in state school finance formulas. The theory and academic literature on this topic are well developed.[1] Some state legislatures have adopted school finance formulas with deeper, others not so deep understanding of these issues, just as some state courts have addressed these issues thoroughly while others have merely acknowledged their existence (others not at all). That is, that it is important, for some reason, to provide additional resources to schools and districts serving more children from low income families. Still, far too few legislatures or courts ask the key questions. Heck, too few of the consultants who advise them on a regular basis, or testify in court on these issues ask the right questions:

• Why? Toward what end?
• And how much will that really cost?

Most importantly, how do we get from A to B? And how do we use this information to reform state school finance policies?

The “Why” question is straightforward and embedded throughout every state’s education policy structures – primarily laid out in accountability systems (expressions of desired/preferred outcomes) – systems which dictate that all schools and districts must strive to bring their students to common outcome goals. These days, those common outcome goals are most often described in terms of College and Career Readiness, measured by interim assessments intended to predict whether a child is on track to succeed in college. 

Ideally, state school finance systems would be designed to provide sufficient resources such that all schools and districts could provide the programs and services necessary for all children to have equal opportunity to achieve those goals. And if the school finance system isn’t designed toward this purpose, the system is inherently unfair (most are).

So where does poverty fit into all of that?

Child Poverty is a form of “Risk” Factor

State policymakers often use the terms “at risk” and poverty interchangeably when talking about student need adjustments in their state school finance formulas. But they don’t often ask – “at risk” of what? And how does that relate to providing additional school funding, or how much funding? The tendency is just to say, well, they are “at risk” and we need to provide…oh… and additional 20 to 30% funding to help them out… you know, with a few extra school counselors and maybe some additional teacher aides in classrooms or something like that.

“At risk” is a useful term if we define “at risk of what?” And state policies already, in a sense, provide that, along with an empirical basis for evaluating it: At Risk of not meeting college readiness standards! (or whatever the desired common standards may be) More specifically, statistically less likely to meet these standards, all else equal.

So then, where does school funding come in? Additional resources are intended to mitigate (or equalize) the risk of not meeting those common standards!  In some states, this is actually the constitutional (court articulated) standard for evaluating the state school finance system- that the system must provide equal educational opportunity for all children to achieve constitutionally adequate educational outcomes.

In this context, poverty and poverty measures operate a bit differently than other student need factors in state school finance formulas. Other factors like children’s disability status, or having English as a second language are specific educational program needs of individuals, requiring specific programs and services assigned to those individuals who qualify for and require those programs and services. By contrast, school and district rates and concentrations of low income children, child poverty or other indicators of socioeconomic status capture a broader, community, schoolwide or districtwide need. We don’t (god forbid, I hope we don’t), for example, pull all the kids whose families fall just below the poverty income threshold and send them off to a separate room at 1:15pm every day to meet with the “at risk” counselor, hired with the money generated by the “at risk” weight. Rather, schools with higher poverty concentrations (compared to lower poverty concentrations), require broad based strategies in order to achieve that equal opportunity standard, compared with their more advantaged peers. What do I mean? Perhaps most importantly, high poverty schools need smaller classes (not just more counselors and aides), which means more teachers and classrooms and also likely need to pay a higher wage to recruit and retain teachers of comparable quality/qualifications.

Higher pay x More teachers = More money! (quite a bit more)

How much more? (there’s actually an article on that, using rigorous methods) Enough more so that kids in high poverty schools have equal opportunity to achieve the college and career readiness standards as their peers in lower poverty schools.

What poverty measure should we pick?

Given the framing above, the approach to picking the right measure of poverty as a “risk factor” is pretty straightforward. It’s not about identifying the individual child for some supposed educational need – programs and services – because his/her family falls below a certain income threshold, but rather about the extent to which school or district poverty rates increase “risk” that children don’t achieve the desired outcomes. So the measure you want to use is the measure which, at the district or school level, best predicts risk of not meeting the desired outcome standards? That is, the measure of child poverty, income status or whatever, that most tightly correlates with outcomes. Importantly, the measure which correlates strongly with outcomes across the full range of outcomes and economic status.

It can also be important to explore whether child poverty is associated with outcomes differently in small, remote rural districts and schools than in suburban or urban schools. If child poverty has different relationships to outcomes (different degrees to risk) at different concentrations or in interaction with other factors like population density.

Some cost benefit analysis can play into this. It may be that some measures are more cumbersome or costly to collect and update from year to year, and may perform somewhat better as predictors of outcomes, but not enough given the cost involved.

What then? How do we address the “how much? Question?

                The value of having:

1. Valid conceptual framework (equal opportunity to achieve common goals), AND
2. An empirical definition of “risk” and approach to selecting “risk factors”

…is that the next step – estimating the costs of mitigating those risk through state school finance systems flows logically into statistical modeling of the spending differences associated with achieving common outcomes at varied risk (poverty, income, whatever was selected) levels.

This next step – to follow “risk modeling” is “cost modeling,” and cost modeling sets out to very specifically estimate the costs of providing all children with what? Equal educational opportunity to achieve the desired, common outcome goals in question!  Cost modeling can then be used to directly inform the design of state school finance formulas including a) setting the overall level of spending needed in the average district to achieve the desired goals, and b) setting the weights and cost adjustments for a variety of factors (district size, competitive wage variation, student needs) to ensure that all children have equal opportunity to achieve those goals.

Using statistical models to relate risk and cost factors, to actual spending and outcomes – and, letting the data do the talking leads to the most accurate estimates – least political estimates – of the cost of providing equal educational opportunity. The relationships among the data are what they are, and are specific to each state context, measured outcomes and selected risk factors. Yes, there are judgments to be made in that process, by humans. But this process if far less subjective than a) asking local educators and other advisors what resources they believe they need to achieve a given set of outcomes (this is useful, but in other ways), or b) relying on consultants to provide their interpretation of what existing research says about how much (what staffing, programs, interventions) is needed to achieve the desired outcomes. Most of that research – even the best and most useful of it – was likely done in different contexts, toward different outcome measures. While it can be useful for making choices among, say, alternative reading programs, or other interventions, it is not particularly useful for estimating the costs of providing equal opportunity across lower and higher poverty settings.

The dangers of not having a guiding framework

In many states, the conversation has changed over time to adopt the logic and framework laid out here – that there is a connection between how we fund schools and what we expect of them. More broadly, that the state’s constitutional obligation to fund schools may be tied to the state’s obligation to ensure that all children have access to some common outcome goals, not just access to some equal dollar inputs to their schools.

When we – more importantly – when state legislatures (or courts) operate outside of or without any guiding framework, equal educational opportunity is unlikely to ever be achieved. It’s all well and good for state legislatures to let themselves feel progressive by saying – hey, we know kids from low income backgrounds need some more resources – so we’re gonna throw em’ a bone or two, and then to debate how they should count which kids get that bone and which don’t – sometimes picking the count method which requires buying the fewest bones, other times picking the count method which makes sure that those bones get spread most evenly – both being a political calculations. You can’t avoid political calculations. School finance formulas are political calculations.

But, when given a legitimate guiding framework and set of rigorous empirical methods for determining Who? Why? And how much?  We can hope to make school finance formulas better than they might otherwise be, in the absence of this framework and related empirical evidence. We can move the needle toward providing all kids equal educational opportunity to achieve common goals. But we’ve got to change the conversation in many places to get there!

============

My forthcoming book on Kansas will explain how the conversation changed in that state over time, and, as a result, why that state outpaces most of its neighbors in the region.

And these two recent reports apply the framework I outline here:

VERMONT

NEW HAMPSHIRE

---

[1] Baker, B. D., & Green, P. C. (2014). Conceptions of equity and adequacy in school finance. In Handbook of research in education finance and policy (pp. 247-259). Routledge.

Duncombe, W., & Yinger, J. (1999). Performance standards and educational cost indexes: you can’t have one without the other. Equity and adequacy in education finance: Issues and perspectives, 260, 261.