What Lunenburg spends for each pupil, and what that number hides

Written by the Lunenburg Budget Project. Not an official town or district document. Every figure here is recomputed from sources/data/lunenburg.db by scripts/verify_peer_spending.py, which fails the build if one drifts.

What this establishes

1. Lunenburg is in the bottom quarter of Massachusetts districts by total per-pupil spending, and it has been in every one of the seventeen years the state publishes. In FY2025 the figure is $18,027. The statewide median is $23,520, and Lunenburg sits $5,493 below it. Of the 318 districts with a published figure, Lunenburg ranks 310 — 8 districts spend less.

2. Against five neighbouring districts it has never been higher than fifth of six. FY2025 puts it sixth, at $18,027 against $18,142 for Ashburnham-Westminster — a difference of $115, which is 0.6%. Over the seventeen years its rank in that set is fifth or sixth and nothing else. The word "last" is doing less work than it looks like it is doing; the durable fact is the quarter, not the position.

3. Most of the spread between these six districts is the denominator, not the money. From FY2012 to FY2025 every one of the six increased spending within a narrow band — 43.8% at the bottom (North Middlesex) to 59.0% at the top (Groton-Dunstable). Lunenburg's was 48.7%, in the middle of that band. What separates them is pupils: North Middlesex lost 23.5% of its FTE pupils over the same span, Groton-Dunstable 14.0%, Harvard 18.7%, and Lunenburg 5.6%. A per-pupil figure is a ratio and both halves move. Give FY2025's money to each district's FY2012 pupil count and Lunenburg is $17,025, fifth of six, and North Middlesex — currently third — is last at $16,474.

4. The gap is not spread evenly across the budget. It is concentrated in six categories. Lunenburg's in-district spending is $17,785 a pupil against $19,888 for the median district in the set, Ayer Shirley — a gap of $2,103 a pupil, or $3,299,397 across 1,568.9 in-district FTE pupils. Decomposed exactly across DESE's eleven function categories: Teachers −$654, Pupil Services −$614, Instructional Leadership −$415, Operations and Maintenance −$235, Instructional Materials, Equipment and Technology −$205, Professional Development −$154 — with Other Teaching Services running +$158 in the other direction.

5. Against the whole state the two smallest lines are the extreme ones. Professional development is $48 a pupil against a statewide median of $172 — 27.9% of it, and only 25 of 317 districts spend less. Instructional materials, equipment and technology is $226 against a median of $626 — 36.1%, with 13 districts below. Lunenburg is in the bottom quarter of the state in 7 of the 10 categories DESE publishes a per-pupil figure for. The exception is Other Teaching Services, at $2,010 against a statewide median of $2,031 — essentially at the median, ranked 162 of 317. That is the paraprofessional, tutor, substitute and instructional-support line, and it is the one place Lunenburg's spending is ordinary.

6. Lunenburg pays near the top of this set for teachers and employs the fewest of them per pupil. Average teacher salary $97,233, second of six behind Harvard's $99,136. Teachers per 100 in-district FTE pupils 6.70, last of six, against 7.41 to 8.55. The two multiply to the per-pupil Teachers figure — but that is DESE's own construction rather than a discovery, and the section below says so plainly.

7. A tenth of what DESE counts as Lunenburg's school spending is not general-fund money, and that is the smallest share in the set. Of $30,031,242 in FY2025, $3,044,898 — 10.1% — came from grants and revolving funds. Harvard's share is 24.0%. This matters for reading everything above: none of these totals is the town's appropriation.

8. The two Chapter 70 standings that look like a contradiction are not one. In FY2026 the state requires Lunenburg to fund 60.4% of its foundation budget against a median district's 74.1%, and the town and state together put in 1.2004× the minimum against a median of 1.3119×. Both are "less", from different starting points. The requirement is low because the formula's own target for Lunenburg is 68.7% and the requirement sits 8.29 points below it — a $2,278,929 shortfall still being phased in. It does not say the state judges the town poor. What cannot be resolved from any of these documents is whether the smaller margin above the minimum is a choice or a constraint, and that is registered as a gap rather than answered.

What this is not

It is not the school appropriation, and it is not what a Lunenburg household pays. DESE's measure is all funds. It counts grants, revolving funds, school choice receipts and gifts, none of which appear in the budget Town Meeting votes. It also counts town-paid insurance and retirement attributed to the schools — $3,459 a pupil for Lunenburg in FY2025, which is a line the school department does not carry. The difference between DESE's $30,031,242 and the town's school appropriation is two definitions, not hidden money, and reconciling them is a separate piece of work that has not been done.

It is not a scorecard. Spending less than a neighbour is not a failure and spending more is not a success. What this document reports is what the money is and is not; what should follow from it is a judgement the reader makes.

The organised data

Where Lunenburg sits statewide, year by year

DESE publishes the distribution of total per-pupil spending across every district, with quartiles. That is what the headline rests on, deliberately: the six-district set below is this project's choice and no document records the criterion for it, so no claim on this page is allowed to depend on it.

Lunenburg has been below the statewide first quartile in 17 of 17 years, FY2009 to FY2025. Its rank moved from 279 of 328 in FY2009 to 310 of 318 in FY2025, with a visible improvement in FY2023 (282 of 320) that did not hold.

The six districts, FY2025

The set is six in every year, but not the same six throughout: Ayer and Shirley regionalised into Ayer Shirley for FY2012, so FY2009–FY2011 hold Ayer instead. Every cross-district comparison in this document therefore starts at FY2012.

Money against pupils, FY2012 to FY2025

The identity is exact: (1 + spending growth) ÷ (1 + pupil growth) = (1 + per-pupil growth), checked for every district by the verifier. The last column is the same arithmetic run once more — FY2025's dollars over FY2012's pupils — and it is arithmetic, not a claim about what would have happened had enrolment held.

The gap, by category, FY2025

Against Ayer Shirley, the district whose in-district per-pupil figure is the median of the five comparison districts. Against that one district rather than against a category-by- category median, because a median of medians does not add up: the eleven per-category medians sum to $20,093 while the median district's total is $19,888, a difference of $205.

Teachers: pay against numbers, FY2025

The multiplication is DESE's own construction and not a finding. The Teachers function's spending divided by teacher FTE equals DESE's published average teacher salary to within 0.52% in the worst of the six. So saying "salary times ratio equals per-pupil cost" is close to restating a definition. What it is for is that it says which half moves: on this measure Lunenburg's teacher spending is low because of how many teachers there are, not because of what they are paid.

Students, FY2025

Low-income share runs from 9.7% to 33.2% across the six. That is a wider spread than the spending, and it is the first reason no relation between spending and results can be read off a table of six districts.

The two Chapter 70 measures that look like a contradiction

DESE publishes two standings for Lunenburg against every district in the state, and in FY2026 they appear to point in opposite directions.

• Required local contribution as a share of the foundation budget: 0.6041. The median district is at 0.7412. Lunenburg ranks 248 of 374, so 247 districts are required to fund a larger share of their own foundation budget.
• Net school spending as a share of what is required: 1.2004. The median district is at 1.3119. Lunenburg ranks 217 of 340, so 216 districts spend a larger multiple of their minimum.

Read carelessly that is the state asks Lunenburg for less than most towns, and Lunenburg still puts in less above the minimum than most towns. Read carefully it is two different denominators, and one of them is not what it looks like.

The required share is a phase-in position, not a judgement that the town is poor. The formula's own target local share for Lunenburg in FY2026 is 68.7% — DESE sets Lunenburg's target aid share at 31.3%, against a statewide target aid share of 41%, which means the formula treats Lunenburg as comparatively wealthy. The requirement of 60.41% sits 8.29 percentage points below that target, a shortfall of $2,278,929, and FY2026 closes $275,365 of it. So the low required share does not say the state thinks Lunenburg cannot afford more. It says the formula has not finished asking.

The two are still both "less", and that is as far as the data goes. What cannot be told apart from these documents is whether the smaller margin above the minimum is a town choosing to spend less or a town unable to raise more — the levy limit, two failed overrides, the district's proposed budget and Town Meeting's vote all resolve into the same single number, and the number is the outcome of all of them at once. That limit is registered in money_gaps rather than argued away here.

Rule 1 applies and is observed. The net school spending measure carries a stage in DESE's own basis column — actual in 31 of the 33 published years, budgeted in FY2025 and FY2026 — and the two are different quantities under one name. Nothing here differences across it. The FY2026 Chapter 70 figures and the FY2025 spending figures are also never subtracted from one another: one is a budgeted formula calculation, the other is what districts reported after a year closed.

What people in Lunenburg have already said about this

Rule 15a: for every category this report calls low, search what residents said about that thing in the same year. The search covered 8,899 of the 12,015 meeting documents this archive holds — 74%. The rest are image scans awaiting OCR. An empty search result would not have meant nobody said it.

"According to Massachusetts State reports Lunenburg per pupil expenditure is listed 361 out of 401 districts in Massachusetts. This means that we're in the top 20% for school performance while being in the bottom 10% for spending" — public comment, School Committee, 24 January 2024

The spending half of that claim holds on DESE's own distribution. The 401 in that sentence and the 318 in this document are different sets of districts, and both put Lunenburg in the bottom tenth. The performance half cannot be checked here at all: this archive holds MCAS for six districts and no statewide distribution of it, so no state percentile can be computed for any results measure. That is a registered gap.

"Lunenburg has a long and well established history of spending little per pupil and still having an amazing school district" — public comment, School Committee, 24 January 2024

The measurable half of that is exactly right, and it is the strongest finding here: seventeen consecutive years below the statewide first quartile.

"Teachers have learned to spend school budgeted money at the start of the year otherwise they and their students are penalized because a frozen budget in November means an inability to purchase needed materials later in the school year. Teachers and parents donate supplies" — the president of the Lunenburg Education Association, School Committee, 7 February 2024

Said in the same period in which the instructional materials, equipment and technology line is 36.1% of the statewide median, with 13 districts of 317 below it. The two are printed together because a reader with this concern will find the number anyway. They are not evidence that either caused the other, and nothing in this archive tests that.

"The North Middlesex superintendent reached out to me about potentially sharing resources as we have done in the past. Professional development with north Middlesex, they were looking for other opportunities to share resources" — the Superintendent of Schools, School Committee, 16 October 2024

Said about the line this document finds the lowest in the set. A district spending $48 a pupil on professional development that is sharing it with a neighbour is a different fact from one that is not, and neither the DESE figure nor this document could have told you which. Rule 8: the district is visibly doing the sensible thing here, and it is nowhere in the numbers.

"the remaining $17,000 will be dedicated to contracted professional development. This week our legislators have approved one-time earmarked funds, $36,000 is going toward touch view screens at the Primary School, $7,000 will fund IXL for our middle school students" — the Superintendent of Schools, School Committee, 4 February 2026

Rule 11 made concrete, in the district's own words: grant and earmark money buying professional development and instructional technology, which are the two lines this document finds furthest below the state. DESE counts it in the totals here and the town's appropriation does not, and neither source says how much of either line is grant-funded in any given year.

"neighboring districts such as Lancaster, Ayer/Shirley, Groton, Pepperell, Townsend, Leominster, and Fitchburg have faced similar challenges while achieving stronger outcomes" — public comment, School Committee, 24 June 2026

The comparison run the other direction, after the override failed. Two of the districts named are in the set here. The MCAS figures this archive holds are below, and nothing in them tests any relation to spending.

MCAS, FY2025

Printed because residents are arguing about it in both directions, and for no other reason. Nothing on this page establishes any relation between what a district spends and how its students score, and nothing in this archive could.

Per cent of students meeting or exceeding expectations. Three years are published on the site; one year is shown here. The two districts above Lunenburg on every measure are the two with the lowest low-income share in the set — 9.7% and 10.8%, against Lunenburg's 23.3%. That is a correlation across six points with no control for anything, offered as a reason not to read the table as a spending result rather than as an alternative explanation for one.

What this changes for planning

Rule 8: a finding arrives as what it means for planning, never as what somebody got wrong. Four things follow, and none of them is a recommendation about how much to spend.

A per-pupil comparison is always at least a year behind the budget being voted. These are DESE's end-of-year figures, reported after a fiscal year closes. The latest here is FY2025 and the Chapter 70 figures are FY2026. A board voting a budget today is looking at a comparison of years that have already closed, and every district in it has since moved. Quoting it as though it described the year being voted is the easiest mistake to make with it.

Enrolment moves this number as hard as money does, and it moves on its own. 22.2% of the rise in Lunenburg's own per-pupil figure since FY2009 is the denominator. A district that loses pupils without cutting proportionally climbs this table without deciding anything; a district that holds its enrolment falls down it while spending more every year. A plan built on the ratio rather than on the two halves is planning against an artefact.

The lines furthest below the state are the ones grants are already covering. 10.1% of everything counted here is grant and revolving money, and the district's own reports describe grant and earmark money buying professional development and instructional technology — the two categories furthest below the statewide median. A plan that treats the general-fund line as the whole of either category is planning against a fraction in both directions: the service is larger than the line, and it is less secure than the line, because a grant ends and an appropriation is voted again.

Where the district is visibly doing the sensible thing, it is not in this data. The professional-development sharing arrangement with North Middlesex is in the minutes and nowhere in the numbers.

What this does not show

• Any relation between spending and results. Six districts, one year, no control for demography, and a low-income spread wider than the spending spread.
• What the money buys. A Teachers line is not a class size; an Instructional Materials line is not a textbook count; a Paraprofessional FTE is not a person or a caseload.
• Whether the figure is a choice or a constraint. See above.
• That these are the right six districts. The set is ours. DESE's own workbook covers all 421 Massachusetts districts and this archive extracts six because the whole file takes the published database past a hosting limit. Every headline above is stated against the statewide distribution for exactly that reason.
• That any of this is the town's bill. Rule 11. All funds, not the appropriation.

A note on the instrument

DESE prints one column headed per-pupil and computes it over two different denominators: the district total over total FTE pupils, and every other row — the in-district rollup and all eleven categories — over in-district FTE pupils. Verified here against DESE's own dollar totals on 6,153 rows across seven districts and seventeen years, with no exceptions.

This archive's own derived table dese_function_statewide carries a per_pupil_basis column reading per pupil, in-district FTE on every row, and on the district-total row that is wrong. The string is ours, written by scripts/extract_dese_finance.py, not DESE's. Nothing on this page or its web version quotes it; the basis is asserted from the arithmetic on every build instead. It is recorded here rather than quietly corrected because it is the exact shape of the defect this project keeps finding — something derived written down, and then quoted as though it had been observed.

What would settle the open questions

All four are rows in sources/data/money-gaps.csv and render at /what-we-cannot-answer.

Sources

Every one is mirrored at /docs/state-dese/<filename> with its sha256 in sources/data/archive-manifest.csv, and the extracted rows are in the published database.

Generated data: /data/peer-spending.json, written by scripts/build_peer_spending.py, which refuses to write if any assertion in it stops holding.