Equity

Equity is the older of the two doctrines in Ohio litigation and the one that lost first. In Walter the Supreme Court of Ohio held, 4-3, that disparity between districts did not by itself violate Article VI, Section 2. verified The disparities continued to widen under H.B. 920, whose reduction factors made a district’s revenue depend on its levy history and its voters rather than on any statewide standard. inference

Resources and compensation are measured in opposite directions # ↑ Contents

A rule this doctrine needs before any of its statistics can be read, and the one most easily got backwards. For spending, less dispersion is more equitable — that is the assumption behind a coefficient of variation or a restricted range ratio. For state aid, the sign flips. Aid is compensatory, so a wide spread means strong targeting and a narrow one means weak targeting.

Any dispersion statistic in this domain therefore has to state whether its subject is a resource a district enjoys or an offset it receives, because the direction of “better” is not the same. inference The same coefficient supports opposite readings depending on which it is, and the guarantee is the case where the difference is largest.

It trades against adequacy # ↑ Contents

Perfect equity is achievable by leveling down — every district equally underfunded satisfies equity and fails adequacy. inference Ohio has never proposed leveling down, but the tension shows up in a milder form every time a reform raises the floor without touching the ceiling: adequacy improves, dispersion may not.

What no correlation will settle # ↑ Contents

Whether Ohio’s compensation is aimed hard enough is not a question a coefficient answers. It is the adequacy question, and it needs a standard rather than an association. inference

What this repository computed # ↑ Contents

Not what Ohio publishes. Figures derived here from committed fixtures, each one citing the test that reproduces it.

Dispersion, FY2024 # ↑ Contents

Across all 606 traditional districts verified see crates/dispersion:

• Operating expenditure per pupil: median $15,646, coefficient of variation 0.202, restricted range ratio 1.844. The district at the 95th percentile spends about 1.85 times the district at the 5th.
• The distribution is right-skewed — the top half runs further above the median than the bottom half runs below it — so any mean-based summary overstates the typical district.

What 0.202 is large or small against # ↑ Contents

That coefficient of variation had no benchmark for as long as this node has quoted it. Ohio describing itself cannot say whether Ohio is unusual, and no other state’s distribution had been computed. Congress computes exactly this statistic for all 51 jurisdictions and pays states on it: under 20 U.S.C. 6337 a state’s equity factor is the enrollment-weighted coefficient of variation of per-pupil expenditure across its own districts, and the Education Finance Incentive Grant multiplies the state’s per-child amount by 1.30 minus that factor. verified 20 U.S.C. 6337(b)

On the Census Bureau’s FY2022 district file, computed for every state the same way, Ohio’s factor is 0.1987 and ranks 9 of 51 — against a median jurisdiction at 0.1582. Only 8 jurisdictions spread wider, and 3 of those have fewer than 50 districts. verified crates/dispersion

Two things follow. The federal government treats Ohio’s dispersion as a fault and prices it, at 3.55% of Ohio’s per-formula-child rate against a median jurisdiction — the doctrinal question this node is about already has a dollar consequence attached to it, in a channel Title I, Part A covers and the formula nodes do not. And the two independent measures of the same quantity agree: 0.202 from the department’s FY2024 operating expenditure and 0.1987 from the Bureau’s FY2022 current spending, on different populations in different years, 0.003 apart. inference

Wealth neutrality, which is more interesting than the dispersion # ↑ Contents

State revenue per pupil correlates with assessed valuation per pupil at −0.548: aid does fall as property wealth rises, so equalization is real, but property wealth explains only about 30% of the variation in state aid. verified

Meanwhile state revenue per pupil correlates with the economically disadvantaged share at +0.632 — more tightly than with property wealth. verified Ohio’s formula targets poverty better than it targets property. Under a design that adds resident income to valuation and weights disadvantaged pupils, that is the mechanism working as intended — and it means the classic wealth-neutrality test, run on valuation alone, understates how much targeting the system actually does. inference

Local revenue per pupil, by contrast, tracks valuation per pupil at above +0.7. verified That is the disparity the state formula exists to offset, visible in isolation.

One reason the wealth relationship is not stronger # ↑ Contents

The temporary transitional aid guarantee holds 48.3% of districts above what the formula computes for them, and its incidence rises monotonically with property wealth — 13% of the poorest quartile against 76% of the wealthiest. verified The local capacity measure reduces state aid to wealthy districts by design; the guarantee then restores it. The two mechanisms work against each other, and the −0.548 correlation is the net of them rather than a measurement of what the formula alone does. inference

Run separately: state aid against valuation per pupil correlates at −0.662 on formula alone and −0.605 as realized, so the guarantee gives up about 9% of the formula’s equalization. verified But the level effect is far larger than the correlation: median guarantee uplift is $0 for the two poorest quartiles and $1,154 per pupil for the wealthiest, which more than doubles what the formula gives that quartile. A component can leave the ranking nearly intact while moving a great deal of money. inference

How much of the local gap each level of government closes, in dollars # ↑ Contents

Everything above is a correlation, which says whether targeting exists and not how much of the disparity it removes. Quartiling the 611 comparable Ohio districts by local revenue per pupil in FY2022 and reading the other two levels against it — each cell the mean over the districts in that quartile, which is the reading the rest of this node’s dispersion figures are on: verified dispersion::national_peers::ohio_by_local_wealth, pinned by test

quartile   local/pupil   state/pupil   federal/pupil
Q1 poorest      $4,342        $8,970          $2,738
Q2              $6,678        $7,199          $2,231
Q3              $8,541        $6,232          $2,181
Q4 richest     $13,932        $4,522          $1,825

The local gap is $9,590 per pupil. State equalization closes $4,448 of it — 46% — and the federal channel closes $913, or 9.5%. $4,229 per pupil of the gap survives both. verified

Those are facts about the average district in each quartile, and the corpus states the average pupil’s separately. Summing each quartile’s revenue over its pupils rather than averaging its districts’ ratios — the same 611 districts in the same four quartiles, so nothing moves but the arithmetic — the gap is $8,928, the state closes $4,922 of it, or 55.1%, the federal channel closes $792, and $3,213 survives both: a quarter less than the district reading leaves open. verified dispersion::ohio_panel::equalization_by_year, pinned by test

Both numbers are correct and neither is a substitute for the other. The reason the corpus now carries both is the department’s own correction: the FY2024 District Profile Report’s Statewide Data sheet was reissued in December 2025 to change every figure on it from an unweighted mean of 606 districts to an ADM-weighted one, which moved its statewide Black enrollment share from 10.5% to 17.3%. A quartile holding Cleveland weighs the same as one holding four hundred village districts only on the first reading. inference

Two things follow that the correlations did not give. Federal money in Ohio is compensatory, monotonically across all four quartiles, which could have been assumed either way — Title I runs on a poverty count and IDEA Part B does not, so the aggregate behavior was not derivable from the parts. And the state’s contribution is not close to full equalization on any reading: nearly half of what the local base opens, the state closes, and the rest stands.

The year flatters both numbers. FY2022 is the peak of ESSER, so the federal offset here is the largest that will ever be measured, and those funds expired in September 2024. It is also the year in which state and local shares are both deflated by the inflated federal one. The honest reading is that this table is an upper bound on how equalized Ohio is, not an estimate of it. inference

And the level the state equalizes from has been falling for a decade # ↑ Contents

The quartile table is one year. Across the ten years the survey covers from FY2012, the state’s share of Ohio school revenue falls in seven of the nine year-on-year changes — 41.9% in FY2012 to 34.3% in FY2022, or three points between the two federally ordinary years FY2013 and FY2019 — while the local share rises. It rises twice, in FY2013 and FY2015; the survey has no FY2014, so the ten years are nine changes and not ten. verified dispersion::ohio_panel, pinned by test; the full series is on DeRolph I

A nominal freeze is a distributive act, and it runs the wrong way # ↑ Contents

Three of the plan’s dollar parameters have not moved since they were set — the preschool flat grant since FY2014, the DPIA amount and all five gifted figures since FY2022. Across the FY2027 model that costs districts $133.2m against what the same formula would pay at those amounts’ real value, and $104.3m of it would reach them once the guarantee is applied. verified

It does not land evenly. Ranked by economically disadvantaged percentage and cut into fifths, the shortfall per pupil is:

fifth	disadvantaged	shortfall per pupil
1 (wealthiest)	17.2%	$34.69
2	33.0%	$43.17
3	43.7%	$53.03
4	70.1%	$94.95
5 (poorest)	84.5%	$141.66

Monotone, and four to one end to end. verified crates/project

The mechanism is not subtle and that is the point. DPIA is the largest of the three frozen components and DPIA is paid on a poverty index, so letting its rate decay withdraws most from the districts the index was written to reach. Nobody voted for the gradient; it is what a nominal amount does when the formula around it is means-tested. inference

This is the state-side twin of H.B. 920. That statute freezes the local yield against inflation and the state’s response was a formula; here the formula’s own parameters freeze, and there is no further mechanism behind them. Both erosions are silent in the document that produces them, which is why both have to be measured to be seen. inference

It also bears directly on the equalization reading above. The quartile table measures how much of the local gap the state closes in one year; this measures a channel through which that share falls on its own, with no vote and no line in any fiscal note. The two findings are about the same trend at different resolutions. inference

That bears on equity in a way the correlations cannot show. Targeting is a property of how a fixed pot is distributed; this is the pot.

The gap grows, and whether the rate holds depends on which mean # ↑ Contents

verified dispersion::ohio_panel::equalization_by_year, pinned by test

FY      local gap    state closes    on ADM    federal    residual
2012       $5,727           42.4%     46.6%      10.2%      $2,713
2013       $6,030           44.3%     45.2%       4.6%      $3,084
2015       $6,599           40.0%     47.0%       6.4%      $3,537
2016       $6,768           44.0%     52.3%       6.5%      $3,352
2017       $7,041           45.8%     53.6%       5.8%      $3,408
2018       $7,558           46.5%     55.6%       5.1%      $3,661
2019       $7,784           48.8%     59.5%       5.2%      $3,576
2020       $8,572           48.3%     59.2%       5.7%      $3,939
2021       $8,808           44.6%     55.0%       5.9%      $4,360
2022       $9,590           46.4%     55.1%       9.5%      $4,229
2023       $9,810           44.9%     52.3%       9.7%      $4,457
2024      $10,527           43.9%     53.6%      10.8%      $4,772

The gap grows and the residual grows with it, from $2,713 per pupil to $4,772. A formula doing the same proportional job against a larger problem leaves districts further apart every year, whichever column the proportion is read off. verified

The rate is the part that depends on the statistic, and the corpus had only one of them. Read as a mean over districts, the state’s share does not move. It is 42.4% at the start, 43.9% at the end, and never outside 40.0% to 48.8% in any of the twelve years the survey holds — FY2014 is missing, as above.

Read on enrolled ADM, over the same panel and the same quartiles, it runs 46.6% in FY2012 to 53.6% in FY2024, never falls below 45.2%, and peaks at 59.5% in FY2019. That is a rise of seven points, fitting a trend of +0.79 points a year against the district reading’s +0.26. Cut the quartiles by pupils as well and it runs 40.4% to 50.0% at +1.06 a year. Both aggregate readings rise; only the district mean is flat. verified

So the corpus withdraws the flatness as a finding about Ohio and keeps it as what it is, a property of the unweighted mean. The honest statement is that Ohio’s equalization of the local gap has been improving on every reading that counts pupils — and that the gap has grown faster than the improvement, so the residual a district actually experiences grows anyway. The bleak half of the original finding survives intact; the stable half was the statistic, not the state. inference

This is the same arithmetic the department corrected in its own FY2024 aggregates, and finding it here is why the rule is written down once rather than re-decided per figure: a district mean is not a state figure. inference

The gap grows by 84% across it — $5,727 to $10,527 per pupil — so a constant rate leaves a growing remainder. The residual, what neither level closes, grows 76% to $4,772 per pupil.

What looked like a break in the last two years was one district # ↑ Contents

Until this revision the section here recorded that the band stopped holding — FY2023 and FY2024 both closing 37.2% of a gap 16% and 21% wider than FY2022’s, the residual up 46% in two years, and the cause open between Ohio’s reappraisals and the phase-in. Every one of those numbers was Kelleys Island Local SD.

It is a district on an island in Lake Erie that teaches between one and eighteen children against a tax base of resort property, and its local revenue is $214,400 per pupil in FY2023 — four times the next district in the state and ten times the tenth. In a quartile of 152 that single row adds $1,410 to the richest quartile’s mean. verified dispersion::ohio_panel, pinned by test

It is not in the survey’s comparable population every year. comparable is the Bureau’s own flag — AGCHRT != 1 and SCHLEV == 03 — and it moves as the agency is coded, so the district is out in FY2012, FY2013, FY2018 and FY2022 and in for FY2023 and FY2024. The population changed in exactly the year the break appears, and the break was the population changing. verified

The measure now takes a floor of forty pupils before it forms a per-pupil ratio. Forty is read off the data rather than chosen to fit it: across all fifteen years no comparable Ohio district has an enrollment between 18 and 63, so any number inside that empty band removes this district in every year and nothing else in any year. Its dollars remain in every aggregate; what it cannot carry is a denominator of five. verified dispersion::ohio_panel::MIN_ENROLMENT, pinned by test

With the floor, FY2023 and FY2024 close 44.9% and 43.9% — inside the band, which now runs the whole thirteen years to FY2024 rather than stopping at FY2022. The gap grows 9.8% across the two years against 5.3% a year over the decade before them, so it is on trend rather than off it. verified

The reappraisals did not widen the gap, and could not have # ↑ Contents

The cause this node held open was between Ohio’s FY2023 property reappraisals and the Fair School Funding Plan’s phase-in, and it recorded that separating them “needs the valuation series beside this one, and two observations cannot do it.” There are four observations, and they are staggered.

Ohio’s 88 counties revalue on a published cycle and Table SD-1 carries TY2021 through TY2024, so every district has one event year and two quiet ones. The arms separate without a threshold: median class-I valuation growth in a cohort’s own event year is +21.6%, +35.7% and +32.0% against +0.65% to +0.91% in its quiet years. verified crates/regime-diff, reappraisal_incidence, pinned by test

The premise fails before the measurement does. A reappraisal raises a district’s revenue only where H.B. 920’s reduction factors have stopped operating, which is the twenty-mill floor of R.C. 319.301(E)(2) — and the floor is where the poor districts are. 46.1% of the poorest quartile sits at it against 12.5% of the richest, falling monotonically across the four. Where a reappraisal reaches revenue at all it reaches the bottom of the distribution hardest, so the mechanism proposed for widening the gap is one that narrows it. verified crates/regime-diff, reappraisal_incidence::by_wealth, pinned by test

The arithmetic agrees. Local revenue per pupil grew 22.5% in the poorest quartile and 11.3% in the richest across FY2022-FY2024. Putting every district on its own quiet-year rate — removing the events and keeping the trend — moves the FY2023 gap by +$26 and the FY2024 gap by −$212. Reappraisal explains none of a $937 widening, and in the second year it was holding the gap down. verified crates/regime-diff, reappraisal_incidence::gap, pinned by test

Two cautions on the second number. Taxes charged for a tax year are collected across two fiscal years, so the counterfactual charges an event wholly to the first and overstates what it removes — and it still comes out the wrong sign for the hypothesis, which is why the direction is the claim and the magnitude is not. And the panel ends at FY2024, so the 181 districts whose counties revalued in TY2024 contribute two quiet revenue years: they are the control arm, not a hole. verified

What the phase-in did over the same years is a separate question and this does not answer it. It no longer has a break to explain, so what remains is whether state aid’s level moved with the phase-in in a way the share of a widening gap conceals. open

Ohio’s formula is doing the same proportional job against a larger problem. Both readings that were available before this run are wrong: equalization has not decayed, and it has not held. The rate held and the outcome worsened, because the rate was never the thing districts experience. A district experiences the residual. inference

This also reconciles with the falling state share above, which looks like a contradiction and is not. The state’s share of total revenue fell 7.6 points while its share of gap-closing held, which is possible precisely because state aid is targeted: a smaller pot distributed more steeply can close the same proportion of a wider gap. What it cannot do is close a growing absolute distance. inference

Both endpoints flatter the federal column — FY2012 carries the ARRA tail and FY2022 is the ESSER peak, each roughly double the ordinary ~5%. Read the middle years for federal and the whole window for state. verified

The outcome side sharpens the doctrine rather than settling it # ↑ Contents

The dominant correlate of measured achievement is not money. The Performance Index tracks the economically disadvantaged share at −0.846 — 71.6% of the cross-district variance. verified Against property valuation per pupil it is only +0.321.

Read beside the aid findings the picture is coherent. Ohio’s formula targets poverty (+0.632) more tightly than property (−0.548), and poverty is what measured achievement mostly tracks. So the compensation is aimed at the right variable. inference

One consequence for how equity claims get argued. Spending per headcount pupil correlates with the Performance Index at −0.337 (FY2025) and −0.388 (FY2024, independent source). verified Quoted bare, that reads as money failing. It is the compensation showing through: Ohio spends more per pupil where need is greater, and need is what the Index measures. The resource-versus-compensation rule again, arriving from a new direction.

Wealth predicts level and does not predict growth # ↑ Contents

In a need-adjusted model over 606 districts — controlling for economic disadvantage, English-learner and disability shares, district size, and per-pupil spending — assessed valuation per pupil predicts attainment level (t = +3.12) and does not predict growth at all (t = +0.06). verified

Property wealth tells you where a district’s students are. It does not tell you how far they moved. That is a sharper wealth-neutrality result than the aid correlations above, because it is about outcomes rather than dollars, and it cuts both ways: the disparity wealth produces is real and visible in attainment, and it is not visibly reproducing itself in the rate of learning. Whether that is the funding system compensating, or a growth measure normed in a way that removes level differences by construction, cannot be told from one cross-section. open