FSFP Cost Input Refresh vs. Freeze

RUN. First result. Reproduce with cargo run --example input_year_refresh -p foundation.

The question is the one the FY2026-27 budget turned on: what is the difference between running the Fair School Funding Plan on refreshed cost inputs and running it on the FY2022 inputs H.B. 96 carried forward, at the same phase-in percentage?

What this run is still not # ↑ Contents

It perturbs one salary of ten; administrator, counselor, librarian, and clerical salaries have all risen too and are held fixed, so the computed increase remains a lower bound. And FY2027 has not happened — the department’s model is a projection, so enrollment, valuation, and income are forecast rather than observed. open

The perturbation is now concrete. Base cost is built by applying statutory staffing ratios to enrollment and pricing each funded position at a statewide average salary for a reference year, times 1.16 for benefits, plus average employer-paid insurance. verified Changing the reference year changes exactly those price terms and nothing else — the ratios, the minimums, and the ADM definitions are untouched. That makes this a clean single-parameter perturbation rather than a re-specification of the model.

There is also a measured precedent to calibrate against: H.B. 33 refreshed the reference year from FY2018 to FY2022, and the statewide average base cost moved from roughly $7,352 in FY2023 to roughly $8,240 in FY2025 — about 12%, of which the phase-in advance accounts for none, since phase-in scales the appropriation rather than the computed cost. inference A refresh of comparable magnitude is the natural first perturbation to test.

What this repository computed # ↑ Contents

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

THE ANSWER, on the department’s own published worked example district (20,342.13 base cost enrolled ADM, 43 buildings), perturbing the classroom teacher salary from the FY2022 reference of $68,022.22 to an FY2024 refresh of $73,777.08 — an 8.46% rise: verified crates/foundation

base cost per pupil    $8,198.40  ->  $8,529.55     +$331.14   (+4.04%)
aggregate base cost  $166,773,020 -> $173,509,173  +$6,736,153

Only the teacher sub-component moves; student support, district leadership, building operations, and athletics are unchanged by construction, because the perturbation touches one price term. An 8.46% salary rise produces a 4.04% base cost rise — the salary-driven part of teacher cost is just under half of base cost for a district this size. verified

Scaled to Ohio’s FY2022 base cost enrolled ADM of 1,499,918.26, a comparable refresh raises computed base cost statewide by roughly $497 million a year, from the teacher salary term alone. inference assumes this district’s cost structure is representative The state’s appropriation consequence is a different and smaller number; see the incidence result below.

Predicted shape, stated in advance so the run can contradict it # ↑ Contents

Freezing the reference year should bite hardest where salaries have risen fastest since FY2022, and the effect should be larger in state-share terms for districts with high state share — because state aid is the residual after local capacity, a reduction in computed base cost falls disproportionately on the state’s portion. inference That would make a freeze regressive across the wealth distribution despite being uniform in method.

Two effects will partly offset it and must be reported separately rather than netted: districts at the 10% minimum state share are insulated at the bottom of the state-share scale verified, and districts held on a guarantee are off-formula entirely and see no effect at all. inference If the run does not show the predicted shape, the reasoning above is wrong.

The incidence result is sharper than predicted # ↑ Contents

A share among the districts the formula pays, which is what the column says for itself rather than leaving to be carried in from further down. The 294 of 609 districts the guarantee holds capture none of an increase at any capacity on this scale, so the table is incidence across wealth and not coverage across Ohio. verified crates/foundation

local capacity/pupil        share of the increase a formula-paid district receives
$500 - $7,378.56            100%     the formula pays the whole residual
$7,527.58                    55.0%   floored before the refresh, on formula after it
$7,676.59 and above          10.0%   the minimum state share holds it in both runs

Because state aid is the residual after local capacity, a district on formula receives the entire base cost increase, whatever its wealth, and a district the 10% minimum state share holds in both runs receives 10% of it. verified

Between those two there is a band, and it is a gradient # ↑ Contents

The floor is a share of base cost per pupil, and this perturbation moves base cost per pupil. So the frozen run and the refreshed run reach the floor at different capacities — $7,378.56 and $7,676.59 — and the $298.03 between them is a band where a district is floored before the refresh and paid by the formula after it. Across that band the share it captures falls continuously from 100% to 10%; at the midpoint it is 55.0%. verified crates/foundation

Nobody falls off the floor, because it is a max rather than a cut-off. One part of the band is structural rather than a fact about this district: at its midpoint the captured share is exactly (1 + floor) / 2, whatever the two base costs are. The width is not — it is (1 - floor) times the base cost increase, so it scales with the perturbation. The runner derives its grid from the two thresholds rather than listing capacities, so a later factor vintage cannot move the band out from under the table.

So refreshing the input year is progressive and freezing it is regressive, and the mechanism is not that poorer districts gain proportionally more but that wealthier districts are cut out almost entirely by the floor. Given that 170 of 606 districts sit at the 20-mill floor — a different floor, but correlated with wealth — the share of districts insulated from an input refresh is a first-order question this run does not yet answer. open

Statewide, all 606 districts # ↑ Contents

verified see crates/foundation/tests/statewide_refresh.rs

Grade-band shares from the department’s October FY2024 headcount file, scaled by each district’s base cost enrolled ADM:

statewide computed base cost increase   $466.2 million per year
ADM-weighted average                    $323.97 per pupil
per-pupil range                         $318.26  to  $442.77

verified crates/foundation

The single-district extrapolation above gave $497 million; the real distribution gives $466.2 million. The gap is mostly denominators — the worked example’s grade mix is slightly richer than the state’s, and FY2024 ADM is lower than the FY2022 base cost ADM used to scale. The two agree to within about 7%, which is the useful check on the extrapolation.

And $466.2 million is within 0.3% of the $465.0 million the department’s own FY2027 model gives for the same refresh — see the guarantee section below. Those two figures are computed on different panels, from different sources, by different routes: this one applies grade-band shares from the October headcount file to FY2024 ADM across 604 districts; that one scales the department’s own published aggregate base cost across 609. Their agreement is the strongest single check this scenario has, and it is a check on the whole chain — the ratios, the benefit multiplier, the professional-development term, and the fixture — not on any one term of it. verified

The spread has a single cause, and it is the staffing minimum # ↑ Contents

The most-affected district gains 1.4 times per pupil what the least-affected does — $442.77 against $318.26 — with no policy choice involved. 155 of 604 districts are small enough that the six-teacher special teacher minimum binds, and their mean per-pupil gain is $346.28 against $322.87 for everyone else. verified crates/foundation

The spread, revised # ↑ Contents

Reading the pre-2007 file natively — see crates/spreadsheet — reproduced 604 of the 606 rows byte for byte and left the two suppressed districts’ bands blank, which is what a band containing a withheld grade actually is: unknown, not smaller. Both are now outside this panel, so the denominator is 604.

The spread is real and structural. It was overstated, and the overstatement came from the one convention this corpus had already written down and the pipeline predated. verified

This is the same structural feature that makes small districts expensive per pupil in the first place, now visible as an incidence result: a minimum that funds more positions per pupil also pays out more per pupil when the price of a position rises. inference So a refresh is progressive twice over — once through the state share residual, and again through the small-district minimums — and a freeze is regressive on both.

The guarantee halves it # ↑ Contents

verified the department’s own FY27 model, carried in crates/project; see crates/foundation/tests/department_model_fy27.rs

districts on the guarantee in FY2027   294 of 609   (48.3%)
their share of ADM                     54.1%
computed base cost increase            $465.0 million
absorbed by the guarantee              $222.9 million
ACTUALLY DELIVERED                     $242.1 million   (52.1%)

A district below its guarantee floor receives the guarantee, not the formula amount. When the formula amount rises, the top-up shrinks by exactly the same sum and total funding does not move. So 253 districts — two in five, holding 43.0% of Ohio’s students — would gain nothing at all from a reference-year refresh, and 356 districts do gain. 41 guaranteed districts would be lifted off; the median guaranteed district is held more than three times above what a refresh would give it, so the guarantee is not a marginal buffer but the operative funding mechanism for those districts. verified crates/scenario-delta

The three counts were published as 242 unmoved and 52 lifted off. Both sum to the same 294 on the guarantee, which is why the decomposition read as coherent while neither term matched the calculator — the arithmetic that would have caught it was the one already satisfied.

This does not overturn the incidence findings above — districts on formula do receive 100%, and the small-district minimums do concentrate the gain — but it shrinks the population those findings describe to roughly half the state. Any statement that a refresh is worth half a billion dollars to Ohio districts is wrong by a factor of two.

It also reframes the phase-in. In the terminal year of a plan designed to fund every district at its computed cost, nearly half of districts are funded by a hold-harmless from the previous regime instead. inference

Inputs verified # ↑ Contents

The FY2022 reference salary this whole perturbation is measured from is now the department’s own: $68,022.22, read off the FY27 TRAD State Foundation Funding Calculator through StatewideFactors::fy2027 rather than restated as a literal. verified The figures above are the first version of this result that perturbs from a number the workspace does not contradict elsewhere; see the revision below.

The state share floor is not a small correction, and the $242.1m above does not apply it # ↑ Contents

computed base cost increase        465.0
  plus the denominated categoricals +32.1   special education, English learners and
                                            career-technical are priced in base cost
                                            per pupil, so a refresh raises them too
computed in total                  497.1
  less the ADM denominators         -7.6   base cost averages three years;
                                           the state share is paid on this one
  less the minimum state share    -118.7   138 districts keep a tenth of it
  less the guarantee              -150.2   253 districts keep none of it
delivered as state aid             220.6   44.4% of what a refresh computes

verified see crates/scenario-delta/tests/who_a_change_reaches.rs, which asserts every line and that the four close on the total, so this is a decomposition rather than four separately plausible numbers

So the deliverable share is 44.4%, not 52.1%, and a third mechanism was missing from the accounting entirely: base cost is computed on a three-year average ADM while the state share is paid on the current year, which costs a shrinking state a further $7.6M. The two figures are not strictly comparable — the run above perturbs one salary per district and this one applies a uniform scale of the same statewide magnitude — so treat 44.4% as the corrected order of magnitude rather than as a replacement to four significant figures. inference the comparison verified each decomposition term

What actually cuts wealthy districts out # ↑ Contents

The direction of the finding is unchanged and its magnitude is sharper than stated: per pupil, the increase pays $305.63 to the poorest quintile and $20.81 to the wealthiest, a ratio of about fifteen to one. A refresh is progressive; the reason is mostly that wealthy districts are on the guarantee. verified

Single perturbation: the cost input reference year in the FSFP base cost calculation
[R.C. 3317.011].
  Baseline  = FY2022 statewide average salaries and per-pupil operating amounts, as carried
              forward by H.B. 96. inference confirm against the enacted act
  Perturbed = the most recent fiscal year for which statewide average salary and operating
              expenditure data are available.
Affects only the price terms: average salary by staff category, average employer-paid
insurance cost, and the statewide per-pupil amounts for building, supplies, technology, ITC
support, safety, and co-curricular activities. Staffing ratios, minimums, ADM definitions,
local capacity weights, and the phase-in percentage are all held at enacted values.

Worked example district, 20,342.13 base cost enrolled ADM, priced at the FY2027 factor set
— the FY2022 averages H.B. 96 carries forward, which is what "frozen" means over this
horizon. Only the enrollment is the FY2022 payment report's; the prices are not, so these
are not that report's totals.
  teacher salary input      $68,022.22 -> $73,777.08    (+8.46%)
  teacher base cost        $97,068,364 -> $103,804,517
  aggregate base cost     $166,773,020 -> $173,509,173  (+$6,736,153)
  base cost per pupil        $8,198.40 -> $8,529.55     (+$331.14, +4.04%)
All other sub-components unchanged. verified