Brookline Town Finances: Reference Document

This is the full reference document behind a shorter public article. It is allowed to be denser and more complete than the article; the article gets distilled from this, not the other way around.

---

Part 0: Why this document exists

In May 2026, Brookline voters approved a $23.25 million override, the fifth since 2005. If you live here, you have a tax bill, and you just voted on one.

This document exists to answer two questions that follow from that. Does this keep happening? And if we would rather it didn't, what would actually have to change?

What a long-range forecast is for

A budget and a long-range forecast are different documents doing different jobs, and treating them as the same thing is one of the easier ways to misread a town's finances.

A budget is a commitment. It covers one year, it has to balance by law, and it is deliberately conservative, because the cost of being wrong in the optimistic direction is far higher than the cost of being wrong in the other. A town that overestimates revenue and comes up short in March has to cut services people are already relying on. A town that underestimates ends the year with a surplus and a much easier problem. Good budgeting builds that asymmetry in on purpose, and Brookline's Financial Plan does exactly that, as it should.

A long-range forecast has a different job. Nothing is appropriated from it and it commits the town to nothing. Its purpose is to show where the current path leads if nothing changes: far enough out that the structural forces have time to become visible, and early enough that there is still room to choose differently. It should use the best available estimate of what is actually likely to happen, not the most cautious one, because it exists to inform a choice rather than to guard against a shortfall.

That distinction is why this document sometimes uses different assumptions than the Town's own plan, and it is not a disagreement with the Town. A forecast built for planning and a forecast built for understanding should not look the same. Part 4 shows both, side by side, and explains every place they differ.

The argument, stated plainly

Brookline's budget gap is arithmetic, not mismanagement. Costs have grown at about 3.65% a year. The revenue that arrives on its own, without anyone voting on anything, has grown at about 3.52%. That difference of roughly a tenth of a percentage point has to be closed every single year, and over twenty years it is what five overrides were for.

The useful way to think about it is to sort both sides of the budget into what Brookline controls and what it does not. The spending side is where most of the choices are.

On the spending side, Part 4's forecast tests each assumption one at a time against the room the May 2026 override leaves, about $12 million a year by 2040. Sorted by who sets each one:

The same sort holds looking back. Of the $177 million increase in annual General Fund spending between FY2011 and FY2025, about 36% came from Brookline's own decisions, 44% was set outside Brookline (mostly inflation in pay, prices and premiums), 11% was paying down past pension underfunding, and 5% came from new homes and commercial space (Part 2b).

On the revenue side, Brookline controls less. The 2.5% cap is state law and state aid is the Legislature's to set; state aid grew only 1.99% a year, and statewide, unrestricted local aid has fallen 25% in real terms since 2002. What Brookline does control there is new growth, the fees it sets for its own services, and, by asking voters, overrides, and those have been carrying the load: new growth and overrides together supplied 57% of all levy growth since 2005. Brookline should keep pressing on state aid, and the Massachusetts Municipal Association's campaign for roughly $351 million more in unrestricted aid deserves support. But a plan that depends on winning that argument is not a plan.

That leaves the levers Brookline can actually pull. The usual list has three: growth that widens the tax base, overrides that ask existing taxpayers for more, and cutting what the Town provides.

Cutting deserves an honest word. It is not off the table, and this document takes it seriously. The Town's own Expenditures and Revenues Study Committee spent a year looking for savings and found real ones. But at the scale that matters, cutting means reducing the quality of things residents actively value: schools, libraries, parks, public safety. Brookline has consistently chosen not to, and that is a defensible choice rather than a failure of discipline. We like our town and what it offers, and most of us would like to keep it that way.

The fourth lever is the one the list leaves out, and it is the largest: what the Town pays for the services it already provides, which is most of the table above. Pay settlements, the health plan, how staffing follows enrollment, and what the pension savings are used for. None of these cuts a service. Each is a decision Brookline makes anyway, every year, in bargaining, benefit design and the school budget, and each is worth more than any plausible difference in how fast the town grows. They deserve to be made deliberately, with the numbers in front of us, rather than by default.

The numbers make the case. With the May 2026 override, revenue stays about $10 million to $19 million a year ahead of spending through the 2030s. That room is real, and it is small next to these four decisions: pay settlements (the teachers' contracts expired in August 2026 and are being negotiated now); health insurance; school staffing as enrollment changes; and the pension turn in FY2031, when about $37 million a year of payments toward past underfunding ends and the Town's schedule moves that money into prefunding retiree health, its largest remaining unfunded liability. Two of them together, pay and premiums both running high, would use up the room by about 2033.

Growth still matters, and Part 3 shows which kinds pay for themselves: renovation, commercial space and higher-value homes do, while a typical new rental apartment roughly breaks even. But growth is the slow lever. Its value compounds over decades, which is the case for starting early rather than for expecting it to settle a budget in any given year.

The Town's own plan points the same way. The Select Board's 2030 Roadmap, adopted in May 2026, opens its fiscal goal by saying Brookline "faces a structural deficit driven by rising fixed costs." This document's forecast shares that concern and is more specific about its conditions: with the override, and with costs behaving as they have, revenue stays ahead of spending through 2040; with pay or premiums running high, the deficit the Roadmap describes arrives in the mid-2030s. Several of the decisions above already have a place on the Roadmap's schedule, and Part 4c lists where.

So the question this document puts to all of us is not only "growth or overrides." It is: how do we want to use the room the override created, and which of these decisions do we want to make deliberately rather than by default? That is not a crisis, and this document does not argue that it is one. Making those choices visible is what it is for.

How this document is organised

• Part 1: How the money actually works. Where every Town dollar comes from and goes, what Proposition 2½ permits, and why rising home values do not give the Town more to spend.
• Part 2: What actually happened, FY2005–2025. Twenty years of revenue and spending, why costs outran the cap, and where growth did and did not happen.
• Part 3: What kind of growth pays for itself. Not all construction affects the Town's finances the same way, and the biggest single factor is not the one the debate usually turns on.
• Part 4: Where this leads, and what could change it. A forecast through 2040 built from Brookline's own cost record, the decisions that move it most, and an honest sizing of every lever available.

A note on what this document is not doing: it does not argue that Brookline should grow, or that it should not. It argues that the trade-off is real, that weighing it takes solid data and a structure for thinking it through, and that this document is an attempt to supply both.

---

Part 1: How the money actually works

Before any of the history makes sense, it helps to know the machine: where the Town's money comes from, what it is allowed to collect, and why the answer to "our house is worth more, so the Town has more" is no.

1a. Where every Town dollar comes from and goes

Start with the funds. Not all of Brookline's money is the same money. The Town keeps several separate pools, and knowing which one is in play makes it much easier to follow any conversation about Town finances. The General Fund, about $440 million a year, roughly 92% of everything the Town spends, is what pays for schools, police and fire, public works, libraries, and every other core Town service. It's also the subject of this entire project. Two more funds, the Water & Sewer Enterprise Fund and the Golf Course Enterprise Fund, are entirely separate and paid for by the fees people using those services pay rather than by property tax. A Recreation Revolving Fund works along similar lines for recreation programs, though not quite to the same standard, program fees cover about 87% of its costs, with the remaining ~13% coming from the General Fund (the Town's own figures; see below).

Three qualifications, verified against the FY2027 Financial Plan and the FY2024 audit (2026-09-04), that a strict "100% self-supporting, no subsidy" framing would miss:

1. Money flows from these funds to the General Fund, not the other way. All three reimburse the General Fund for staff benefits and overhead they consume: Water & Sewer $3,000,604, Recreation Revolving $709,574, Golf $309,320 in FY2027, about $4.0 million together. The General Fund transfers nothing into any of them.
2. Golf is not covering FY2027 costs from fees alone. Its budget draws $450,000 from its own accumulated retained earnings for a one-time landscape capital project, against $3,150,250 of fee revenue and $3,600,250 of spending. That's a drawdown of golf's own past surpluses, not a taxpayer subsidy, but "fees cover 100% of costs" is not literally true in FY2027. (The FY2024 audit shows Golf ran an $82,191 operating loss that year, too.)
3. One small, real General Fund subsidy to Golf does exist: an interest-free $500,000 note issued in 2012 for a maintenance shed, repaid at $25,000/yr, with $225,000 still outstanding at 6/30/2024 (FY2024 audit, Note 15). The principal repayment sits inside Golf's overhead reimbursement, but the foregone interest is a genuine, if modest, transfer of value from the tax levy.

And since 2021, Brookline has also collected a Community Preservation Act surcharge, an extra 1% added to property tax bills, legally restricted by state law to open space, historic preservation, community housing, and recreation projects. It cannot be spent on schools, salaries, or anything else in the General Fund, no matter how large a gap the General Fund is running. When someone says "the Town has a surplus" or "the Town needs more money," it matters a great deal which of these five pools they mean. This document is about the General Fund specifically.

1b. Inside the General Fund

The General Fund is about 90% of the Town's total financial activity and essentially all of what people mean by "the Town budget." It is worth seeing on its own.

Where the General Fund's money comes from. About 80% of it is property taxes, $332.5 million of the $419.6 million the Town expects to collect in FY2026. The rest comes from state aid (roughly $26 million, mostly Chapter 70 school funding and other formula grants Brookline has limited control over), local receipts (about $35 million, motor vehicle excise, building permits, meals and rooms taxes, and similar fees), and smaller transfers from other available funds and free cash. Because property tax is so dominant, and because it is the one major revenue source almost entirely within local control, it's the piece this project focuses on.

Two of those terms are worth defining, because they appear throughout this document and in every Town budget discussion.

Local receipts are everything the Town collects that is neither property tax nor state aid. Motor vehicle excise is the largest piece, and its rate is set by the state. Meals and rooms taxes were a local decision to adopt, at rates the state caps. The rest, building permit fees, parking, licences, ambulance charges, investment income, and similar service charges, Brookline sets for itself. Local receipts are recurring revenue: they arrive every year and are budgeted for in advance, deliberately conservatively, since a shortfall against an over-optimistic estimate has to be absorbed mid-year.

Free cash is the opposite. It is a one-time figure: what was left over when a fiscal year closed on June 30, meaning budgeted money that went unspent, plus receipts that came in above estimate, less unpaid receivables and any deficits. Nothing can be spent from it until the state's Division of Local Services certifies the amount. Once certified, state law lets a town appropriate it for anything, but Brookline's own Fiscal Policies, approved by the Select Board in September 2024, are stricter: free cash "shall not be used for Operating Budget purposes." It goes, in order, to the budget reserve; to keeping reserves at or above 20% of revenue; to the liability fund; to the capital plan, until total capital funding reaches 8% of the prior year's net revenue; to the Affordable Housing Trust Fund when its balance is below $5 million; and then to one-time uses such as the trusts for retiree benefits. The timing matters: free cash usually funds the budget two years after the year that produced it, not the next one. By the time a year's surplus is certified, the following year's budget has already been adopted. The surplus at the end of FY2024 (June 30, 2024) was certified at $25.1 million as of July 1, 2024, and $21.3 million of it went into the FY2026 budget. Some free cash can be spent sooner, at a fall Special Town Meeting, but the annual budget runs two years behind. Brookline's certification has run between $12 million and $25 million a year recently. One labelling note worth carrying: DLS labels free cash by the fiscal year that begins on that July 1, while Town budget documents label the same dollars a year later, so a DLS "FY2024" figure and a Town "FY2025" figure can be the same money. The distinction between free cash and local receipts matters because free cash is a result rather than a revenue source: a large certification means the prior year ended better than budgeted, and using it to fund recurring costs would convert a one-year surplus into a permanent obligation, which is what the Town's policy is written to prevent. Part 4b returns to this when it looks at reserves.

1c. Prop 2½: what the Town is allowed to collect

What actually limits how much property tax Brookline can collect: Prop 2½. Massachusetts voters passed a law in 1980, known as Proposition 2½; that caps how much a town's property tax collections can grow in a given year, regardless of what happens to individual property values. This is the single most important mechanic in this entire story, so it's worth laying out carefully.

The cap applies to a number called the levy limit, the maximum total dollar amount the Town is allowed to raise from property taxes townwide. Each year, that limit is allowed to grow by exactly three things added together:

1. 2.5% of last year's levy limit, automatically, every year, no vote required. This is where "Proposition 2½" gets its name.
2. "New growth", the tax value added by anything genuinely new since last year: a newly built house, an addition, a converted commercial space, a subdivided lot, or new taxable business equipment and utility property (what assessors call personal property). New growth is measured in dollars (the newly added assessed value, taxed at last year's tax rate) and gets added to the levy limit permanently, on top of the 2.5%.
3. Any override voters have approved that year, a permanent increase to the levy limit that requires a townwide ballot vote. Brookline voters have approved five of these since 2005: $6.2 million (FY2009), $7.7 million (FY2016), $6.6 million (FY2019), $12.0 million (FY2024), and $23.25 million (approved May 2026, phased in over FY2027-2029).

There is a fourth category of property tax dollars, debt exclusions, that sits entirely outside this cap. When Brookline borrows to pay for a specific capital project (a new school building, for example) and voters approve that specific borrowing at the ballot, the resulting bond payments are added to everyone's tax bill dollar-for-dollar, on top of the levy limit, for as long as that bond is being paid off, and then they disappear when the bond is retired. Debt exclusions are about $28.6 million of Brookline's current (FY2026) property tax bill; they don't compound, they don't count against the 2.5% cap, and they're tied to a specific, voter-approved project with a defined end date.

The Community Preservation Act surcharge described in 1a is also on the tax bill and also outside the levy limit, but state law restricts it to open space, historic preservation, community housing and recreation, and it never reaches the General Fund. The General Fund property tax figures in this document, such as the $332.5 million in FY2026, include debt exclusions and leave the CPA surcharge out. So overrides and new growth remain the only ways to grow what the General Fund can collect.

Rising home values, on their own, do not give the Town more money to spend. Here's why: the 2.5% cap applies to the total levy limit, the town-wide pot, not to any individual property's tax bill. If every home in Brookline appreciated by exactly the same percentage in a given year, the total amount the Town is allowed to collect would still only grow by 2.5% (plus new growth, plus any override). Nothing else would change in that scenario, since everyone appreciated equally and everyone's share of the fixed pot stays the same. In the more realistic case where some properties appreciate faster than others, what happens is a redistribution, not new money: a property that appreciated faster than the town-wide average ends up paying a larger share of that same fixed levy limit, and a property that appreciated more slowly pays a smaller share, even though the total amount the Town collects didn't change at all. Rising values only change how the existing, capped amount gets divided up among property owners.

That single fact is the hinge this entire project turns on, and it's worth sitting with before moving on: since market appreciation alone can't grow what the Town is allowed to collect, there are really only two ways Brookline's property tax revenue can grow faster than 2.5% a year, asking voters directly for an override, or generating new growth through construction and renovation. Everything about "why does new growth matter" traces back to this one mechanic.

One more piece: how the tax bill splits between homeowners and businesses. Once the total levy limit is set, the Select Board decides each year how much of it falls on residential property versus commercial, industrial, and personal property (business equipment) combined, a combination the state calls "CIP." State law caps how far a town can shift the burden toward CIP: a commercial property's tax rate can be at most 175% of what a single, unclassified rate would be. Brookline's Select Board has, in practice, pushed this shift almost to its legal limit, the FY2026 shift factor is 1.746, just 0.2 percentage points shy of the 1.75 ceiling: meaning Brookline's commercial and industrial property owners are already paying close to the maximum share the state allows, and there is essentially no room left to shift more of the burden their way through this particular lever without a change in state law.

1d. Where the money goes

Where the money goes. On the spending side, FY2026's roughly $421.5 million General Fund budget breaks down as: schools, $142.8 million (34%, the single largest line); departmental services (police, fire, public works, libraries, and the rest of Town government), $98.9 million (23%); employee benefits, pensions, retiree health insurance (OPEB), and active-employee health insurance, $92.7 million (22%); debt service, $43.5 million (10%, split between the excluded, bond-matched portion described above and a smaller regular portion funded like any other expense); and other non-departmental costs, $43.7 million (10%). Two of these categories, benefits and schools, have been growing noticeably faster than the rest of the budget, which is exactly why the next section matters.

1e. What "sustainable" actually means here

Everything in Part 1 comes together in one chart, which is also the evidence for the definition below: twenty years of Brookline’s actual revenue and spending, against what the levy limit alone would have done.

It’s worth being precise about this, because two different comparisons are easy to conflate. Actual total revenue and actual total expenses don’t need a gap between their growth rates, they’re required to balance every single year, by law. Massachusetts towns cannot run a structural deficit; whatever a year’s appropriations are, that year’s revenue (property tax, state aid, local receipts, and any free cash drawn down) has to cover them. So the fact that the “total revenue” and “total expenses” lines in the chart above track each other fairly closely isn’t a target being hit, it’s a legal requirement being satisfied, year by year, using whatever levers (an override, a free cash draw, a mid-year cut) it takes to make it true.

The comparison that actually matters for “sustainable” is narrower: does the revenue the Town gets automatically, 2.5% plus new growth, with no override and no vote, grow fast enough to keep up with real cost growth on its own? If it does, the budget balances itself indefinitely without asking voters for more. If it doesn’t, the gap has to be closed some other way, year after year, and since free cash and reserves are finite, that eventually means either an override, cutting services, or generating more new growth. Over the twenty years already lived through, automatic revenue grew at 3.52% a year against operating costs at 3.65%: a shortfall of about a tenth of a percentage point, every year, which is what four overrides were for. Section 2c takes that apart and finds most of it in one place. Whether the balance holds going forward depends heavily on whether new growth keeps pace with whichever expense categories are running hottest, recently, that’s employee benefits at 5.5%/yr, well above the blended 20-year expense average, which is why Part 3’s scenario menu treats “how much new growth” as a real, quantifiable lever rather than a vague aspiration.

---

Part 2: What actually happened, FY2005–2025

Part 1 described the machine. This is what it did over twenty years.

2a. Twenty years of revenue and spending

Start with the founding fact behind this whole project: Brookline's General Fund revenue grew from $168 million in FY2005 to $390 million in FY2025, more than doubled, while spending grew right alongside it, from $169 million to $365 million. Neither number moved smoothly or automatically. Both were shaped, year after year, by the same tension Part 1 described: property tax growth is capped by state law at 2.5% a year plus whatever "new growth" (new construction, renovation, conversion) adds to the tax base, and Brookline's new growth has chronically fallen short of what similar towns manage.

That shortfall is real and measurable, not a rounding error, Brookline's tax base has grown from new construction and renovation at 1.37% a year since 2005, in the bottom third of the state. Section 2c takes that apart properly.

Restacked as dollars rather than rates, the same three counterfactuals show where the $175 million of FY2005–FY2025 property tax growth actually came from, and how much of it came from asking voters.

Four times since 2005, that gap grew faster than new growth and the 2.5% cap alone could close it, and voters were asked to close it directly with an override, a permanent increase to the town's allowed tax levy, approved at the ballot:

A fifth override, $23.25 million, phased in over FY2027-2029, split $17.9 million for schools and $5.3 million for town departments, passed in May 2026, just after this twenty-year window closes; see Part 4 for how it is treated in the forecasts.

Three of the four are explicitly schools-driven, and that's not a coincidence; it lines up with a separate, real wave of school construction and expansion happening on the capital side over the same twenty years. That side is funded differently: not by overrides, which permanently raise the tax base, but by debt exclusions, which raise taxes only for the life of a specific bond and then expire. Brookline voters approved debt exclusions for the Edward Devotion School expansion (2016), high school renovations (2019), and the Driscoll School expansion (2021), and, most recently, two large ones still underway: the Pierce School's full PreK-8 rebuild (approved 2024, with roughly $114 million of already-authorized borrowing beyond what even this project's own balance-sheet forecast currently accounts for) and new fire station construction (2025). Operating overrides and capital debt exclusions are two different mechanisms, but they're pointed at the same underlying cause: a growing, aging set of public facilities and a student population the town has had to keep building and staffing for, even as, per the next section, enrollment itself has recently started to fall.

The other major driver of the gap sits on the benefits side, not schools, and it moves on its own separate, slower clock. Brookline's net pension liability swung from $184.5 million (2014) to as low as $125.3 million (2021, a strong investment year) and back up to $170.7 million (2023), genuinely volatile, not steadily rising or falling, and driven mostly by investment returns rather than the town falling behind on its own contributions. Net OPEB liability (retiree health benefits, the town's largest unfunded liability) tells a similar story: $254.5 million and only 12.9% funded in 2017, down to $189.6 million and 32.6% funded by 2023, then back up to $201.5 million in 2024, a reminder that this liability doesn't decline in a straight line either. The town's own funding schedule has the pension system reaching full funding by 2030, with the contribution capacity that frees up then redirected to OPEB, a real structural turning point Part 4's forecast relies on directly.

None of this reads as crisis or mismanagement on its own, the pattern is mechanical and explainable, and Part 1 laid out the machine that produces it. What this chronicle adds is the when and why now: four overrides, clustered mostly around school-driven need, layered on top of a tax base that grew more slowly than peer towns' because new growth chronically underperformed. The next section goes to the parcel level, to look at where, and where not; that growth actually happened.

2b. Why expenses outran the cap

Since 2005, Brookline's operating spending has grown at roughly 3.65% a year, above the 2.5% Proposition 2½ allows without new growth or an override. (That figure excludes debt service on voter-approved debt exclusions, which is matched dollar for dollar by its own revenue and so nets out on both sides. Including it, total spending grew at 3.92%.) That total is made of pieces growing at very different speeds. Employee benefits, pension contributions, retiree health insurance, and active-employee health insurance, have grown at roughly 5.5% a year. Education, the single largest line, at roughly 4.5%. Everything else, police and fire, public works, libraries, general government, debt service, combined, at roughly 2.5%, right in line with what the cap allows on its own.

That last figure is worth pausing on. Most of what the Town does has been growing at almost exactly the rate state law permits. The gap between what is capped and what is spent comes almost entirely from two places.

Debt service in that chart is split in two, because the combined line is misleading. Taken together it grew at 5.17% a year, which reads as a third fast-growing cost alongside schools and benefits. It is not. The part funded by voter-approved debt exclusions went from $1.7 million to $22.0 million, and every dollar of it is matched by its own revenue, raised outside the levy limit for that purpose. The part that actually competes with everything else inside the cap went from $11.6 million to $14.3 million, a growth rate of 1.06% a year, slower than every line on the chart except public works and general government. Brookline's school and infrastructure building programme is large and is not free, but it sits outside the operating problem this document is about.

All of it, sorted by cause

Growth rates say which lines grew fastest. They do not say why. A second pass answers that directly: take annual General Fund spending in FY2011 ($187.2 million) and in FY2025 ($364.7 million), and assign every dollar of the $177.4 million difference to one of nine causes. Each figure is the difference in one year's spending, not a total added up across fourteen years.

The window starts in FY2011 because in FY2010 about $6 million moved between the state's spending categories for Brookline: the five town department categories fell 12.9% that year while total spending fell 2.3%. Any rate measured across that year mixes a relabelling with real change.

Each budget line is assigned by three questions, in order. What would the FY2011 operation cost in FY2025 if nothing changed but pay and prices? Staff costs grow with the national Employment Cost Index for state and local government workers, other costs with Boston consumer prices, and the difference is categorised as Inflation. How many more staff were there, and how many would growth have needed? Staff needed for new homes and commercial space are categorised as New homes and commercial space, and staff beyond that as Staffing levels. Whatever is left is categorised as Pay and costs above or below inflation; it is a leftover, not a measurement of one thing. Pensions are split into the cost of pensions employees earn each year and the payment toward pensions already earned but not yet funded. The national pay benchmark does not understate Massachusetts: average pay for local government workers grew 3.40% a year in Massachusetts against 3.65% nationally from 2014 to 2024.

Three things follow.

Growth in the town itself is a small piece. About 1,640 homes and 350,000 square feet of commercial space added $9.5 million a year of service cost, 5% of the increase. New homes add real costs; they are small next to inflation, debt service and staffing levels. Part 3 takes up whether they pay for themselves.

About a third was Brookline's own choice. Staffing levels, routine and voter-approved borrowing, retiree health prefunding, and pay and benefits above or below the benchmarks add to $64.2 million, 36% of the increase, or 41% counting growth, which zoning and permitting shape. Most of the rest, 55%, was set outside Brookline (inflation, state assessments) or by obligations already incurred (the pension shortfall). The lines are judgement calls. Pay rates are bargained locally, so Brookline has some say over the part filed under Inflation too; only the part of pay growth that differs from the national benchmark is counted here as a local decision.

The staffing piece is almost all schools. The district had 6.75 students per staff member in FY2011 and 5.62 in FY2025. Had staff moved one-for-one with enrollment, it would have about 208 fewer positions, which at about $94,000 each in salary is $19.6 million. The endpoints hide the path: from FY2011 to FY2019 enrollment rose 18.5% and staff 34%; after COVID enrollment fell 10.6% and staff 5.0%. About 150 of the positions beyond enrollment were added while enrollment was rising, and about 73 are positions kept after it fell. The subsection below looks at what those positions do.

The pension piece has an end date. Nearly all of its growth is paying down past underfunding, and that payment ends when the system reaches full funding, which the Town projects for 2030. Part 4 returns to what happens to that money.

Schools: the town built for 7,855 students and now has 7,023

The question behind a 4.5%-a-year school budget is whether it reflects more children or the same children costing more. The answer is neither, quite, and the arc matters more than either endpoint.

The two ratio rows are calculated within a single year, so the peak column uses FY2020 for both, the year staffing peaked, rather than FY2019, the year enrollment did. One caution on the teacher-FTE series: DESE's published figure steps down from 555.5 in FY2007 to 477.1 in FY2008 while enrollment rises, which looks more like a change in how the count is compiled than a cut of that size. Comparisons that cross that boundary are worth treating as approximate; comparisons from FY2008 forward are on a consistent footing.

Enrollment rose by nearly a third between 2005 and 2019, 1,871 additional children. That is what drove the wave of school construction described below, and it is why three of the four overrides in this period were explicitly school-driven. Then it reversed: enrollment has fallen 11% from its peak.

Staffing followed, but not immediately, and the per-FTE ratios show the sequence. Enrollment dropped by 886 students in FY2021; teacher FTE rose that year, to its own peak of 674.6, and students per teacher fell to 10.21, the lowest in the twenty years. Teacher FTE has come down every year since, and by FY2025 the ratio is 12.91, higher than at any point since FY2015. The adjustment did happen; it ran two to four years behind the enrollment change that prompted it.

What did not reverse is the composition. Students with individual education programs grew from 999 to 1,270, up 27%, while total enrollment was falling. As a share of the district they went from 15.4% to 18.1%. Special education is more staff-intensive per student by design and by law, so a district can shrink and still need more people. Inside a declining total staff count, the mix shifted toward special-education support and away from classroom teachers: teacher FTE is down 18% from peak while total staff is down only 9%.

This is the honest answer to "why do school costs keep rising when there are fewer children." Measured from the FY2020 peak, the district employs fewer people. Measured from FY2011, it employs more per student: 5.62 students per staff member against 6.75, the Staffing levels piece above. Both are true, and the difference is the arc. Staff were added while enrollment rose and only partly withdrawn when it fell, and the mix shifted toward special-education support, to serve a student population with different needs, in buildings sized for a larger district. Against 17 peer districts Brookline is the 7th most staffed of 18 (5.6 students per staff member against a peer median of 5.9) with the 4th highest needs, so its staffing is somewhat above its peers and so are its needs.

Benefits: the slower clock

Benefits are the fastest-growing line in the budget and the one least responsive to any decision the Town makes in a given year. Between Brookline's two most recent actuarial valuations (January 2022 and January 2024), active employees in the pension system grew from 1,347 to 1,367 (+1.5%) while retirees and beneficiaries already drawing a pension grew from 891 to 929 (+4.3%), retirees growing nearly three times as fast as actives, over just two years.

The two large unfunded liabilities behind this do not move smoothly. Brookline's net pension liability swung from $184.5 million (2014) down to $125.3 million (2021, a strong investment year) and back to $170.7 million (2023). Net OPEB liability, retiree health benefits, the larger of the two, was $254.5 million and 12.9% funded in 2017, improved to $189.6 million and 32.6% funded by 2023, then rose again to $201.5 million in 2024. Both are investment-return-sensitive; neither is on a straight line.

That shape is worth sitting with. Neither line is a steadily rising debt the Town has been failing to pay down. Both move mostly with investment returns, 2021 is a strong market year on the pension line, and the OPEB line's improvement to 2023 reverses in 2024. A snapshot taken in any single year would support a much more alarming or a much more reassuring story than the decade actually tells.

There is a real turning point ahead. The Town's own funding schedule has the pension system reaching full funding in 2030, with the contribution capacity that frees up redirected to OPEB starting in 2031. It is the single largest piece of good news in the Town's medium-term finances, and what happens to the money it frees is one of the four decisions Part 4 finds matter most.

Everything else: not more people, more expensive people

Since "the Town keeps hiring" is a common and reasonable suspicion, it is worth testing directly against the staffing data. Across five departmental categories over the recent period, headcount grew in four, General Government (+0.99%/yr), Public Works (+0.38%/yr), Human Services (+3.61%/yr, the fastest, with essentially flat cost per employee, more staff, not pricier staff), and Culture & Recreation (+0.52%/yr). None of these is a story about runaway hiring.

Public Safety is the exception, and it runs the other way. Headcount shrank by 0.28%/yr, Police is down 7.5 FTE net, and the Town's own ERSC report footnotes that six officer positions were held open to balance the 2026 budget, while implied cost per employee grew at 3.89% a year, the fastest of any category. That is not more officers; it is the same or fewer officers costing more, driven by arbitrated wage settlements: a 2023 retroactive award, a certification differential stipend rising to 7.5% by 2027, and tenure-based step increases.

This matters for the lever menu in Part 4. Public Safety's 4.37%/yr cost growth is a real and ongoing pressure, but it is a wage-and-arbitration story on flat headcount, which is a very different problem from a service-expansion story and responds to very different remedies.

2c. Why revenue didn't keep up

Part 1 established that there are only two ways Brookline's allowed tax revenue can grow faster than 2.5% a year: ask voters for an override, or generate new growth. Section 2a covered the overrides, four of them in plain view, each one a ballot question. This section covers everything else.

Start by sorting the revenue side into what Brookline sets for itself and what the state sets for it.

Two dashed lines, two solid, and one benchmark. Read against operating expenditure, which grew 105% over the twenty years:

The levy did its job. With actual new growth and no overrides at all, the levy limit grew at 3.91% a year, faster than operating costs at 3.65%. Proposition 2½ plus Brookline's own construction and renovation was, on its own, enough to keep pace.

State aid did not. It grew at 1.99% a year, and the shape of that line is worth looking at rather than summarising. State aid fell in absolute terms after 2008, bottoming at 15% below its FY2005 level in FY2012, and did not recover to its starting point until around FY2015. Ten years of a large revenue source going backwards.

Blend the two together, weight them by size, and automatic revenue grew at 3.52% a year against costs at 3.65%. Had state aid merely kept pace with cost growth, it would have been $34.4 million in FY2025 rather than $24.9 million. That single shortfall of $9.5 million a year accounts for 87% of the entire gap between automatic revenue and costs.

This is not a Brookline story. Statewide, unrestricted local aid has fallen 25% in real terms since 2002, and its share of local tax revenue has gone from 11.5% in FY2007 to 5.4% in FY2024. Brookline experienced the Massachusetts pattern.

It is also not a reason to stop pressing. The Massachusetts Municipal Association is currently seeking roughly $351 million in additional unrestricted aid, which would be worth real money to Brookline. The argument of this document is only that a plan cannot depend on winning it.

Which brings the question back to new growth

Over FY2005–2025, Brookline's tax base grew from new construction and renovation at 1.37% a year. Checked against every one of the 351 Massachusetts municipalities' own new-growth filings with the Department of Local Services, that is the 26.6th percentile, the bottom third of the state.

What the distribution actually looks like

A percentile on its own is easy to wave away, so here is the shape of it. Across all 350 municipalities with a complete twenty-year series:

Brookline at 1.37% sits just above the 25th percentile. The 80th percentile, 2.03% a year, is the benchmark this document uses throughout, and it is worth saying why: it is not a ceiling and not an aspiration, it is simply what the better-performing fifth of the state managed. Sixty-nine other municipalities did at least that well over the same twenty years, under the same state law, the same Proposition 2½, and the same economy.

Named comparisons, including the unflattering ones

Percentiles hide who is in them. Here are eight towns and cities a Brookline resident would recognise as comparable, chosen to span the range, not to make a point:

Two things in that table matter more than the headline.

Brookline is not uniquely slow. Newton is barely ahead of it and Arlington is well behind, both dense, built-out, expensive inner suburbs facing the same constraints. Any account of Brookline's rate that treats it as a local failure has to explain Newton and Arlington too. An earlier version of this document cited only Cambridge, Boston, Needham and Lexington, which is a real comparison but a selected one; the full picture is that Brookline sits in a cluster of similar towns, and the cluster is slow.

But some very similar places are much faster. Somerville, Cambridge and Watertown are not less built-out than Brookline, not less expensive, and not less contested. They are two to three times faster at this. Whatever explains the gap, "there is no room" is not a sufficient answer, because places with no more room have done substantially better.

The recent five years, stated carefully

The trailing window tells a more complicated story, and it is easy to misreport in either direction. Over FY2021–2025 Brookline's rate fell, to 1.28%. Its percentile ranking improved, to the 40th.

Both are true because the bar came down. The 80th-percentile threshold fell from 2.03% to 1.97% and the median fell with it, from 1.62% to 1.40%. Brookline moved up the table by standing still while others slowed. That is worth knowing; it means the recent gap to strong performance is genuinely narrower than the twenty-year gap, and it is not a turnaround, and this document does not present it as one.

Why the shortfall compounds

New growth is not an annual bonus that resets. A percentage point added this year permanently enlarges the base that the 2.5% cap applies to every year afterwards, so a year of slow growth is not a year of lost revenue; it is a permanent reduction in the size of every future year's 2.5%.

Twenty years of running roughly two-thirds of a percentage point behind the 80th percentile therefore compounds into a meaningfully smaller tax base today than Brookline would otherwise have. Section 2e puts a number on exactly what that shortfall cost, by rebuilding the levy limit year by year under the alternative.

What this section cannot explain

It is worth being straight about the limits here. This section establishes that Brookline's new growth has run in the bottom third of the state and by how much. It does not establish why, and the honest answer is that the data can identify the composition of the shortfall but not its cause.

What the composition shows, set out fully in Part 3d, is that Brookline's new growth is overwhelmingly investment in existing buildings rather than new ones, and that new commercial space contributes almost nothing measurable. Whether that reflects zoning, process, land economics, or simple demand is a question this document deliberately leaves open, and Part 4d says so directly. It matters, because the growth lever in 4c is sized on the assumption that the rate could move, and nothing here proves that it can.

2d. Where the growth did and did not happen

The finding above, 1.37% a year, 26th-percentile new growth, statewide, is a townwide average. It says nothing about where in Brookline that growth did or didn't happen, or what it actually looked like on the ground: a torn-down house replaced with something larger, an old office building converted to housing, a kitchen renovation, or nothing changing at all. This project built a parcel-by-parcel map to answer exactly that (land-value-map/, a live, interactive version of everything below; it covers real estate only, while the town-wide new growth rate above also includes business and utility personal property), by comparing every one of Brookline's roughly 8,500 parcels' assessed value, year over year, against what that specific building or lot would be expected to do with no structural change at all, no addition, no conversion, no new unit, nothing but ordinary market movement. (251 of those parcels are permanently tax-exempt, schools, municipal buildings, religious institutions, housing authorities, and excluded from these growth figures entirely.)

Getting that "expected, nothing changed" baseline right mattered more than it might sound. Massachusetts' 2.5% cap limits the town's total tax levy, not any individual property's assessed value, a naive 2.5%-a-year assumption badly understates real organic appreciation (this project's own data shows townwide assessed value actually grew roughly 5.9% a year, compounded, over the period examined). The map instead measures every parcel's growth against Brookline's own observed, real appreciation rate among parcels already confirmed to have had no structural change, not a stylized formula. That distinction turned out to matter: parcels in the top decile of growth-above-that-real-baseline are 4.0x over-represented among properties this project's own classifier independently flagged as having undergone real new construction or renovation, meaningfully stronger evidence that something real happened there than a raw dollar-growth or raw percent-growth ranking gives (2.3-2.6x and 3.8x respectively), because it's the only one of the views grounded in the town's own audited tax mechanics rather than an approximation.

Running that classifier across eight recent years of assessor data (FY2017-2025) finds real, physical change on a meaningful but modest share of the town: 287 parcel-years of new construction and 467 of identified renovation, against roughly 61,000 parcel-years that saw nothing but ordinary market movement. That's the parcel-level face of the finding above: real construction and renovation genuinely happening across Brookline, just not at the pace, or possibly not in the mix; that would have closed the gap to peer towns.

Real stories, not just an abstraction. This project's own design principle is to make findings concrete wherever possible, a specific parcel, not just an aggregate. Nine parcels were picked across Brookline's range of outcomes, and each one's story was checked against the town's actual building-permit records, not just its assessed-value history alone. Seven are done:

• A single-family home, vacant since at least 1994, was newly built out under a 2017 permit. The lot itself was reclassified as developable roughly a year before construction started, once local zoning approval was granted, a reminder that the assessor's books can move ahead of a shovel going into the ground, not just behind it. Value went from $27,100 (bare land) to $3.1 million; the town's own books had substantially caught up with the construction within about 15 months of the permit being filed, several months before the certificate of occupancy was even issued.
• A single-family home was torn down and rebuilt sometime between 1999 and 2011 (before this project's own assessor data begins in 2009, so the exact date isn't independently visible), confirmed instead via its own sale history: a 2011 sale price of $2.105 million lines up almost exactly with the rebuilt home's own assessed value that same year.
• A single-family home was converted into a two-family, via a $611,000, zoning-board-approved gut renovation (a lowered, underpinned basement; a new roof; a new porch). Town records show it had already been informally treated as a two-unit property for years beforehand, which is why this project's own classifier initially filed it as an ordinary renovation rather than new construction; the permit record confirmed a real, substantial project either way, landing on the books about 22 months after it was filed.
• A former church rectory, owned throughout by the Episcopal Diocese of Massachusetts, stopped qualifying for its religious tax exemption on its own, not because of a sale, sometime before mid-2022, and then underwent a genuine $1.58 million renovation (kitchen, baths, a historic-district-approved exterior update) while the Diocese still owned it. Both events are real; only the second one is construction, and it landed on the books just six months after the permit was filed, a notably fast turnaround, since it was interior work on an already- standing building rather than new construction.
• A former private school, tax-exempt for over a decade, lost that exemption, was demolished within about five months of a 2021 construction permit filed by a healthcare/senior-living developer (Welltower/Balfour), and rebuilt as a 153-unit assisted-living building, the single largest new-growth story this project found so far. It took roughly 40 months from permit to substantially complete, about double the typical single-family timeline, consistent with a much larger, more complex project needing more time, not a sign anything went wrong.
• Two single-family homes showed no construction event of any kind across the entire period checked: no permits beyond routine maintenance, ordinary appreciation only, assessed value tracking the town's real baseline closely. These aren't gaps in the data; they're the genuine control cases every "did construction happen here" claim in this project needs, confirming the classifier isn't simply flagging every property as a story.

Two more of the nine, a hotel and a Chestnut Hill condominium redevelopment, are still being checked against real permit records as of this writing, and will be folded in once that's done; nothing in the finding above depends on them. A tenth parcel, checked before this list of nine was formalized, turned out to be the most instructive messy case: an ordinary single-family home where a $200,000 renovation permit stalled for years, the property changed hands mid-project, a new contractor picked the same job back up at a reduced scope, and even after all that, part of the eventual assessed-value jump turned out to be an unrelated townwide land revaluation, not the renovation at all. Real projects are not always as clean as the other nine examples above, worth keeping in view alongside them.

One honest limitation surfaced while doing this work. Brookline's own parcel identifiers occasionally change when a property is physically assembled from more than one lot, one real example found this way merged a neighboring property into an existing address's parcel entirely, with the town's own current records showing no link at all back to the absorbed parcel's old identifier. This project's map only catches that kind of change correctly when it happens to land on the same underlying lot number by coincidence; a genuine merger across two different lot numbers, like that one, can otherwise look like an unexplained value jump rather than what it actually is. Worth keeping in mind as a limit on precision in a small number of cases, not a limit on the overall finding.

Together, 1a and 1b tell the same twenty-year story from two directions, townwide numbers, and the real buildings and permits behind a sample of them.

---

2e. Would faster growth have avoided the overrides?

This is the counterfactual the whole argument turns on, and it is worth doing precisely rather than rhetorically. The quick version of the arithmetic is tempting and nearly right: overrides added about 0.8%/yr of levy growth, so 1.3% actual plus 0.8% needed is about 2.1%. But the levy limit is a compounding recursion, not a sum, so the answer has to be solved for rather than added up.

Rebuild the levy limit year by year from FY2005: compound it forward at 2.5%, add each year's real new growth, add each year's real override. That reproduces the actual FY2025 levy limit of $293.1 million. Now remove the four overrides and leave everything else alone: the levy limit lands at $254.4 million instead. The four overrides were worth $38.7 million a year of permanent levy capacity by FY2025, considerably more than the $32.4 million they raised on paper, because an override permanently enlarges the base that then compounds at 2.5% every year after.

So the question becomes: what rate of new growth, with no overrides at all, would have reached the same place?

2.15% a year. Against an actual average of 1.33%, a gap of 0.82 percentage points, every year, for twenty years. The quick arithmetic above was right to within a rounding error, which is reassuring, but this figure is solved for rather than asserted.

That is the precise answer, and it comes with an important qualification. 2.15% is more than even strong peer performance would have delivered. The 80th-percentile rate across all 351 Massachusetts municipalities over this period was 2.03%. Growing at that rate, with no overrides, gets Brookline to a FY2025 levy limit of $286.7 million, 83% of the way to where the overrides actually took it, not the whole way.

The honest statement is therefore slightly stronger than "growth could have replaced the overrides," and slightly weaker than it might first appear:

Matching the growth rate of the best-performing fifth of Massachusetts towns would have avoided roughly four-fifths of what Brookline asked its taxpayers for over twenty years. Avoiding all of it would have required doing somewhat better than the 80th percentile, sustained, for two decades.

Both halves matter. Growth is a genuinely large lever, large enough to have absorbed most of a $38.7 million annual burden. It is not, on the evidence of the last twenty years, a complete substitute for ever asking voters for anything.

---

Part 3: What kind of growth pays for itself

Part 2 showed that new growth has been carrying a substantial share of the load, and that more of it would have absorbed most of the override burden. This part asks the question that follows: does all growth do that work equally? It does not, and the largest single factor is not the one the debate usually turns on.

3a. The asymmetry: students are expensive, employees are cheap

Part 1 established that new construction and renovation are one of only two ways (the other being an override) that Brookline's allowed property tax revenue can grow faster than 2.5% a year. This section makes that concrete: what does a real construction project actually do to the Town's finances, and why does the kind of project, not just its total dollar value, matter so much to the answer?

The revenue side is straightforward. When a new unit is built, an existing building is renovated, or commercial space is added or upgraded, the Assessor eventually captures that added value, and it becomes "new growth", permanently added to the levy limit, on top of the 2.5%, no vote required. This is genuinely new spending capacity for the Town, not a redistribution of an existing capped pool the way ordinary market appreciation is (Part 1 again). Since 2005, this mechanism has added real, measurable capacity to Brookline's budget every single year, but, at only 1.37% of the prior levy per year on average (FY2005-2025), the 26.6th percentile statewide, well below the 80th-percentile threshold of 2.03% and well below what comparable places like Somerville (3.80%), Cambridge (3.19%) and Watertown (3.12%) have generated. Section 2c sets out the full distribution, including the peers that are slower than Brookline.

The expense side is where it gets more interesting, and where the type of project matters enormously. A new resident, especially a school-age child, costs the Town real money, primarily through the schools, which is by far the largest marginal cost driver of new residential growth (using the RKG Associates figures behind Brookline's own tax classification hearings and the ERSC report, roughly $15,600–$26,700 per additional pupil per year, plus a smaller per-unit municipal service cost; this document's own forecast puts the marginal figure at $16,443 in FY2027, see 3c). A new commercial employee costs the Town comparatively little, since commercial space generates no school enrollment at all and requires a much smaller marginal service footprint per employee. This asymmetry, students are expensive, employees are cheap, from the Town's perspective, is the single biggest reason why "how much got built" is the wrong question on its own, and "what kind of project got built, at what value" is the right one.

Three things follow from that asymmetry, and 3c turns them into a test that can be applied to a specific building:

• A residential project's fiscal impact turns on how many school-age children it adds, and that number is not well predicted by unit count or bedroom count alone. The same building, same bedrooms, produces very different enrollment depending on who lives in it, which is 3b's subject and the largest single factor in the whole calculation.
• The same math can flip entirely once a project's value per unit is high enough: a smaller number of higher-value units can generate enough tax revenue on its own to cover its marginal cost, even where a similarly-sized project of smaller, family-oriented units would not. Getting this right depends on a project's own real, expected value, not a generic Town-wide average, a Town-wide median blends in many small renovations alongside new construction and can understate what a specific new project is actually likely to be worth, which is enough on its own to flip the sign of a fiscal-impact estimate from negative to positive.
• Commercial and renovation-only projects tend to be the most reliably positive, for the mechanical reason described above: commercial square footage adds no school enrollment, and renovating an existing building (rather than adding units) adds assessed value without adding residents at all. The Chestnut Hill Commercial Area project approved by Town Meeting in May 2026, for instance, is projected (independent consultant RKG's own, comparatively conservative estimate) to generate over $3.6 million a year in combined new property tax and local receipts once complete, commercial and hotel space specifically, which is exactly why a project like this reads so differently from a small residential infill project on the Town's books, even before accounting for its cost side.

The Town's own figures, and where they agree

In June 2026 the Planning Department put its own version of this calculation to the Select Board, which makes it possible to check the model above against the Town's rather than only against itself. The figures are annual net fiscal impact for recent development projects, by use type:

The right-hand column is not the Town's; it is what those figures imply when run backwards through this document's own cost figures (3c), holding the children per home the Town's figures imply. Three of them land close to figures derived here independently. The office row implies $491 per square foot against the $518 used throughout Part 3, a difference of 5%. The apartment band contains the $548,418 per unit that the Pleasant Street project was actually appraised at. And the Town's FY2026 new growth of $3.56 million sits within 5% of the $3.74 million this document's trend assumptions produce for the same year. None of these were fitted to each other.

The gap between the two residential rows is worth dwelling on, because it is larger than anything bedroom count does and it is not a cost effect at all: the marginal service cost of the two mixes differs by about 17%, while the fiscal impact differs by a factor of three to eight. It is entirely assessed value per unit. Condominiums are valued individually, close to what they sell for; rental buildings are valued by capitalising their income. The same mechanism appears in the permit data in 3d, where a dollar of residential construction converts to $1.31 of new growth and a dollar of commercial construction to $0.363.

One caution attaches to the Town's table, and it is the Town's own: both residential rows are marked "no deduction for pre-development value." They are gross of the tax the parcel was already paying. In a town where almost all new housing replaces something, that is the difference between what a project collects and what it adds. The Pleasant Street project's own figures put the gap at 14%; for a single-family teardown replaced by a two-unit building, this document's calculator puts the net incremental impact at −$12,690 against a gross of −$690.

3b. Tenure: renting or owning changes the answer

Everything in 3a turns on how many school-age children a project adds. That is one lever, and it is real. But working through Brookline's own assessment roll surfaces a second lever: whether the units are rented or owned.

Start with the town as it stands, from the FY2025 assessment roll joined to the Town's GIS parcel data, 1,088 condominium buildings and 127 apartment buildings:

Read the middle row against the top row: 4-to-8-unit apartments contain about the same space per unit as condominiums, 1,335 square feet against 1,302, and yet they are assessed at roughly half the value per square foot.

That comparison sets different buildings side by side, and Brookline's condominium and apartment buildings differ in more than tenure: location, age, condition, finish. The test that isolates tenure holds the building fixed. Between FY2009 and FY2025, 20 older apartment buildings, most built around 1900 to 1915, were converted to condominiums. Two years later their assessed value was a median 82% higher than comparable apartment buildings that stayed rentals (the middle half, 54% to 109% higher). Four had essentially no permitted renovation and still rose about 69%, while apartment buildings that renovated heavily but stayed rentals rose about 6%. So the jump comes with the change of tenure, not the construction. Two- and three-family houses that converted rose much less, about 25%, and 14% with little renovation, because they are valued on sales whether rented or owned.

Why converting an apartment building raises its value so much. Massachusetts assessors value the two using different, and equally standard, methods. A rental building is valued by the income approach: its net operating income, rent, less operating costs, divided by the yield an investor requires, which in practice values it at roughly seventeen to twenty times annual net income. A condominium is valued by sales comparison: what units like it have actually sold for. Owner- occupiers consistently pay far more than the capitalised rental income of the same unit, because they are also buying mortgage leverage, deductible interest, expected appreciation and control of their own home. Investors buying an apartment building are not. The gap is widest where rents sit furthest below what the units would sell for, which describes much of Brookline's older rental stock.

For a new building, the gap is probably smaller, and it is the open question. In a September 2026 conversation, the Town Assessor put the premium for condominium use at about 20% of a building's value. His own method supports that for new, high-end buildings. At the $5 to $6 a square foot monthly rent he considers reasonable for new luxury units, with his stated vacancy (about 3%), expenses (33%, excluding tax) and reserves (2%) and a 4.25% capitalization rate plus the tax rate, a new rental is worth about $720 to $860 a square foot, against about $970 for condominiums built since 2010: a premium of roughly 13% to 35%. The new rental buildings actually on the roll, though, assess at about $350 to $480 per finished square foot, well below that formula. Why is not yet clear: finished area may be measured differently for rental buildings and condominiums, the rents applied may be lower, or inclusionary units and a prior-year capitalization rate may pull values down. The Assessor's rent tables, which he has offered to share, should settle it.

So the statement the evidence supports has two parts. Converting an existing apartment building raises its assessed value by about four-fifths. For a new building, the difference between building it for rent and for sale is probably nearer a fifth. For existing buildings that makes tenure one of the largest single inputs into what a building pays; for a new one it is real but smaller than the effect of how many children it houses. Section 3c turns both into a test that can be applied to a specific building.

3c. The breakeven test, and how to apply it yourself

Everything so far has been about categories, rental against ownership, new against renovated. Residents and Town Meeting Members do not vote on categories. They vote on a specific building, on a specific street, with a specific unit mix, and the question in the room is always the same one: does this one pay for itself?

That question has an answer, and it is simple enough to work out on the back of an envelope. This section sets the test out so it can be applied to any project, rather than presenting a verdict to be accepted.

The test

A residential project covers its own costs if the property tax it generates exceeds the marginal cost of serving the people who move into it. Almost all of that cost is schools. Rearranged, the test is a single threshold, the number of school-age children per unit above which a project stops paying for itself:

Breakeven children per unit = (assessed value per unit × tax rate − municipal service cost per unit) ÷ cost per additional pupil

Three inputs, all of which are published:

• Assessed value per unit, the project's own expected value, not a town-wide average. This matters more than any other input, and using an average is the single most common way these estimates go wrong: a town-wide median blends small renovations in with new construction and can understate a specific new project badly enough to flip the sign of the answer.
• Municipal service cost per unit, everything that is not schools: police, fire, public works, general government. This document uses $866 per home per year (FY2027): $660 of Town department cost, the central estimate from comparing how department spending grew with housing across 350 Massachusetts towns, plus benefits for the staff involved. RKG Associates, the consultant behind Brookline's own fiscal impact modeling, puts it at $1,371 (FY2025), an average-cost allocation that is 88% police and fire; Brookline's public safety headcount in fact fell by five over FY2011–25 while about 1,640 homes were added. Brookline already fields more police and firefighters per resident than most peer towns: about 2.0 officers and 2.6 career firefighters per 1,000 residents, against peer medians of about 1.6 and 2.0. (The police counts are FBI figures for 2019, the last complete year before reporting gaps and before the 2020–22 attrition; at 2022's 116 sworn officers Brookline's figure is about 1.8.) Fire staffing moves in steps, a station or company at a time, rather than with each new home.
• Cost per additional pupil, either $16,443 (the marginal cost in this document's forecast, FY2027) or $26,683 (Brookline's full average cost per pupil including benefits, DESE FY2024). The marginal figure staffs the enrollment-linked two-thirds of the school budget at today's ratio: about one added position per 8.5 students ($12,092), plus supplies and transport ($2,477) and health insurance and payroll costs ($1,874). Which of the two is right is a real and unsettled question, and 3e takes it up properly.

The residential tax rate is 1.024% for FY2026; the FY2027 rate is set in December 2026.

Brookline's own numbers, plugged in

Using the assessed values per unit from 3b, the actual FY2025 roll, not a projection, the test gives this:

That first column compares different buildings, so it mixes tenure with everything else that separates Brookline's condominiums from its apartment buildings. For one new building built either way, 3b's evidence points to a smaller effect. A rental building worth $548,418 a unit (the Pleasant Street project's appraised value) breaks even at 0.29 children per unit when it opens. Built as condominiums at the Assessor's 20% premium, it would break even at 0.36; at the 80% premium older buildings show on conversion, 0.56.

This reproduces the Town's own figure independently, which is worth saying. The ERSC's own fiscal-impact appendix (March 2026) puts the breakevens at about 0.21 children per unit as rental and 0.66 as for-sale condominium. Rebuilding the calculation from Brookline's assessment roll and this document's own cost estimates, without using their result, gives 0.22 and 0.68. Two routes to the same pair of numbers means the finding is not an artefact of either method. The small remaining differences are the residential exemption and the precise value and cost basis each uses.

The same test, fifteen years on

A home's tax and its costs do not grow at the same rate, so a home that covers its costs when it opens does not necessarily keep doing so. Its property tax can grow only about 2.5% a year, because Proposition 2½ limits the levy on existing property. The cost of an added student grows with school pay and health premiums: in this document's forecast, from $16,443 in FY2027 to $27,706 in FY2040, about 4.1% a year. Town services grow with Town pay, from $866 to $1,271. So the breakeven falls over time, by about a fifth by FY2040.

Read the chart by finding a home's assessed value on the bottom axis and its expected children on the side. Below the solid line it pays its way when it opens; below the dashed line it still does in FY2040. Three things stand out.

A typical new home in Brookline roughly breaks even. The homes Brookline has actually been adding are mostly rental and are worth about $430,000 each, weighted by units. With the forecast's average of 0.244 children per home, such a home comes out about $500 a year short when it opens and about $2,000 short by FY2040. A home needs to be worth about $477,000 to cover 0.244 children in FY2027. That is close enough to zero that the answer for any specific building depends on its value and who lives in it, which is the point of the test.

How this squares with the Planning Department's figures. Its June 2026 development look-ahead puts recent apartment projects at +$1,750 to +$5,400 a home a year, before deducting the tax the site already paid. The difference is not the costs, which are lower here ($866 of Town services a home against RKG's $1,371). It is children and value. Run backwards through this document's costs, the Town's figures imply about 0.11 school-age children per apartment and values of roughly $418,000 to $775,000 a unit, where this test uses the town-wide average of 0.244 children and $430,000. At 0.11 children, a $430,000 home here would come out about $1,700 a year ahead when it opens. Which figure fits a given building depends on its bedroom mix and who lives in it, the input the test is built around.

Ownership homes pay their way with room to spare. An average condominium covers up to 0.68 children when it opens and 0.55 by FY2040, more than any of Brookline's recent developments produces. At the forecast's 0.244 it nets about $7,200 a year, rising to about $8,600.

The test charges each new home for its own students; the town as a whole looks different. Brookline's enrollment is shaped far more by falling births than by new housing, and the last subsection of this part sets that out.

The test also assumes each added student brings added staff. That is the right assumption for a growing school system, and Brookline's is shrinking: enrollment is projected to fall from 6,889 in FY2027 to about 6,000 by FY2040 while the base case, like the School Department's plan, keeps staffing at the FY2027 level. The FY2027 kindergarten-to-grade-8 classes already have about 945 open seats under the School Committee's class-size guidelines. While enrollment stays below the FY2027 level, a new home's children mostly fill seats the Town is already paying for, and the added cost per student is about $2,477 (supplies, transport and the other costs that move with each student) rather than $16,443. On that basis even a $430,000 rental home nets about $3,000 a year, and the seats that open as enrollment falls would take about 1,300 homes beyond the recent pace by FY2030 and about 3,600 by FY2040. The open seats could instead be cut to save money, so the two uses compete. The new homes page sets this out school by school, and the Forecast Explorer's Students from added homes setting runs it through the forecast.

This is why housing is a weak lever for the budget, town-wide. In Part 4's forecast, reaching a faster growth rate entirely through more housing, about 6,900 more homes by FY2040 than the trend, improves the FY2040 balance by $12.4 million a year, about $1,800 per added home. The same growth rate reached through commercial space improves it by $56.9 million.

How much each assumption moves the answer, for the two homes that bracket the range (breakeven children per home, when it opens and in FY2040):

The cost of a student is the assumption that matters. Whether it is the marginal or the average cost moves the rental breakeven by almost half; how fast school pay grows decides how far it slides by 2040. The Town services figure, the one most argued over, moves it by about 0.03 either way.

What to put in for children per unit

This is the input the whole calculation hinges on, and the one most often guessed at from bedroom count. RKG's own schedule, used in Brookline's fiscal impact models, is:

Apply that schedule with care, because the Town's own data shows it is only a starting point. Two of the ten Brookline developments in the ERSC's sample are three-bedroom townhouse projects. The one built as rentals generates 0.56 children per unit; the one built as condominiums generates 0.25, the same bedroom count, more than twice the enrollment, in the same town. The Committee's presentation says so directly: student generation is "highly variable based on bedroom size alone."

Where a comparable existing project can be identified, its actual enrollment beats any schedule.

What the test says about real Brookline projects

Applied to ten real developments built in Brookline since 2015, the result splits exactly along the tenure line: six of the ten clear the rental bar, and all ten clear the condominium bar. (The ten values are read from the ERSC report's chart, which does not publish them as a table: about 0.03, 0.06, 0.07, 0.13, 0.16, 0.18, 0.25, 0.25, 0.40 and 0.56 enrolled students per unit in Fall 2025. By FY2040 the sixth sits right at the rental bar, and the highest sits just under the condominium bar.)

That is the most useful single sentence in Part 3, and it is worth being careful about what it does and does not mean. It does not mean six projects were mistakes, a project can be worth building for reasons that have nothing to do with the Town's ledger, and 3e says so. It means that under rental tenure the test is genuinely binding, so which projects go ahead actually changes the Town's finances; while under ownership tenure the test has, so far, never bound at all.

It also means the fiscal case and the housing case can point in opposite directions, which is the uncomfortable part and the subject of 3e.

Births, enrollment and new housing

A common worry is that new housing fills the schools with children who cost more to educate than their homes pay in tax. The breakeven test above answers that one building at a time. The town-wide answer turns mostly on something else: Brookline is having far fewer children.

Births have fallen by almost 40%. Brookline residents had about 690 babies a year in 2008–13, about 590 in 2015–19, about 490 in 2020–22 and 422 in 2023, the lowest in records going back to 1980 (Massachusetts Department of Public Health). Each of those years becomes a kindergarten class five years later, so the effect on the schools is still arriving.

How enrollment is built. Every enrollment forecast, including both of the ones the Town uses, follows the same logic:

• start with the children born to Brookline residents each year;
• some of them move away or go to private school before kindergarten, and some families with young children move in: in NESDEC's figures each kindergarten class is about 85% the size of the births five years earlier;
• as each class moves up a grade, families keep moving in and out, so a class grows or shrinks a little each year;
• add pre-kindergarten and the roughly 300 METCO and other non-resident students.

A single number summarizes the middle steps: enrollment divided by the children born to Brookline residents in the thirteen years that feed kindergarten through grade 12. It rises when families with children move in or stay, whether into new homes or, far more often, into the 29,000 homes already here, and falls when they leave.

So far, families moving in have held enrollment level. From FY2021 to FY2025 the pool of Brookline-born children fell 4%, yet enrollment stayed near 7,000. Had enrollment per birth stayed at its FY2021 level, FY2025 would have had about 6,620 students, roughly 400 fewer. Families moving into Brookline made up the difference.

From here, they slow the decline rather than stop it. The smaller birth years have not reached the schools yet. Assuming births from 2024 on match the 2020–23 average, about 470 a year, the pool of Brookline-born children falls about 28% by FY2040:

Keeping enrollment level would take more families moving in than during the 2010s boom, by a wide margin. The School Department's own forecaster, Cropper-McKibben, projects exactly that, rising to about 7,400 by 2034. Its 2023 report attributes the rise to new apartments, empty-nest owners selling to families and continued arrival of young households, while assuming only about 1,500 new homes from 2020 to 2034 (4a); it needs enrollment per birth of about 1.1, a level Brookline has not reached in twenty years of records.

How Brookline compares with its neighbors. Other districts show that the level the School Department's forecast needs is not unheard of; it is how family-magnet suburbs work, and Brookline has never worked that way. Two measures, for districts the Town and the schools usually compare themselves with:

In the suburbs, families arrive once their children are born, so a kindergarten class can be twice the size of the births five years earlier, and classes keep growing as they move up. In Brookline and its urban neighbors, kindergarten is about the size of the births or smaller, and classes shrink as they move up, as families leave or turn to private schools. Brookline's own kindergarten-per-birth figure has moved between about 0.8 and 1.0 in recent years (1.01 for the kindergarten class of fall 2014, 0.81 for fall 2024 and 0.93 for fall 2025, a class of 453 against NESDEC's projected 412). The School Department's forecast amounts to Brookline starting to draw families the way Newton or Belmont does. That could happen, through housing that suits families or schools that pull them in, but it would be a change in what kind of town Brookline is, not a continuation of its record. (Births are from the Massachusetts Department of Public Health's 2020 and 2023 birth reports; enrollment is DESE's October counts. The ten-year cohorts span COVID, and the district figures include METCO students.)

Where new housing fits. New homes bring their students through the same in-migration steps as any family buying or renting an existing home. At the recent pace of about 166 homes a year, plus Chestnut Hill, roughly 2,500 homes by FY2040 would bring about 430 to 610 students, using 0.17 (the middle of ten recent Brookline buildings) to 0.24 (this document's average). That is real, and it is smaller than the decline from births. Those students arrive in a system with fewer students than it had: the district served about 7,850 students in FY2019 and is projected to serve about 6,000 in FY2040 under NESDEC, which already includes homes built at the recent pace.

What that means for cost. It depends on what the schools do as enrollment falls. If staffing stays where it is, as this document's base case and the School Department's own plan assume, students from new homes fill seats that would otherwise sit empty and add little cost beyond the forecast. If staffing would otherwise shrink with enrollment, students from new homes mean fewer positions cut, and their cost is the saving forgone, which is what the per-home test measures. Either way, the town-wide question is not whether new housing will crowd the schools. It is how quickly enrollment falls, and whether staffing follows it.

Two cautions. Births for 2024 and later are not yet published; another low year like 2023 would deepen the decline. And the number of children a building brings varies widely with its bedrooms and tenure (0.03 to 0.56 in the ten recent buildings), so the per-home test above still matters for any one project.

3d. Renovation: the largest source of new growth

Everything so far in Part 3 has been about what happens when something new gets built, how many students a unit generates, whether it is rented or owned, what that does to the Town's bottom line. That framing quietly assumes the question is about new buildings.

It mostly isn't. Most of Brookline's new growth does not come from new buildings at all. It comes from money invested in buildings that already exist: additions, gut renovations, finished attics, teardown-and-rebuilds on the same lot, conversions and reconfigurations. That is the largest single component of the Town's new growth; it has been so for as long as the data allows checking, and it rarely features in the public argument about growth in Brookline.

Two kinds of growth, two kinds of property

The useful way to organise this is a two-by-two. New growth can be residential or commercial, and, cutting the other way, it can come from investment in property that already exists or from creation of new property:

Splitting Brookline's actual trailing five-year new-growth rate of 1.28% a year across those cells, using the Town's own DOR-certified new-growth filings, with the residential and commercial splits derived from the parcel-level classifier described in 2d, gives this:

Set aside personal property, business equipment and fixtures, which is not property investment in any physical sense, and of the part that is actually real estate, 76% is investment in existing property and 24% is new property creation.

That separation is worth making explicit, because business equipment arrives in the same certification as building investment and is easily counted alongside it. Leave it in and the split reads as roughly 80/20. The rate of genuine new property creation is the same number under either treatment, 0.2502% a year; what changes is the denominator it is measured against, 1.28% with equipment included and 1.05% without. The 76/24 figure uses the second, because the question here is what Brookline's building stock is doing.

Nor is this one unusual year. Across FY2018–2025, renovation's share of residential new growth averaged 75% and never once fell below 53%, with a high of 89%.

What renovation actually does to the Town's finances

Part 3a established the asymmetry that drives everything here: new residents cost the Town money, mostly through schools, and new assessed value earns it. Part 3b showed that tenure moves where a given project lands on that ledger, a great deal for existing buildings. Renovation is the limit case of both, and it is worth being precise about how extreme it is.

Running the parcel classifier across FY2017–2025 identifies 467 parcel-years of renovation and 287 of new construction. Their fiscal profiles could hardly be more different:

Renovation added more assessed value than new construction did over those eight years, and it added no housing units. The net figure is slightly negative because some renovations consolidate units: two apartments combined into one, a two-family returned to single-family use.

That makes renovation the one category of new growth that needs no breakeven test at all. Every new housing project has to clear a bar set by the students it generates; that is the whole subject of 3a and 3b. Renovation generates none, by construction: the classifier's own definition requires that no unit be added. It is, in the narrow arithmetic sense this document has been using throughout, unambiguously net positive for the Town, and it is the only kind of growth of which that can be said without qualification.

It is also the kind of growth that costs a sceptical reader nothing. Someone genuinely worried about traffic, school crowding, or the character of a neighbourhood has no reason to object to a neighbour finishing an attic. "Make it easier and faster to invest in the house you already own" is not a growth argument that requires anyone to change their mind about growth.

What the building permit records show

Everything above is derived from the assessment roll, which records the result of renovation rather than the act of it. In September 2026 the Town's Building Department supplied its permit history: 68,354 records from the legacy system covering 2006 through mid-2017, and 58,907 from the current one covering mid-2017 through September 2026. The current system records an estimated construction cost for each permit, so for the nine complete fiscal years FY2018 through FY2026 the investment itself can be measured directly rather than inferred as a residual. That window aligns with the classifier's FY2017–2025 window, which is what makes the two comparable.

A renovation reaches the Town's budget through two channels, and they are not close in size. The first is the building permit fee, collected when the permit is issued, at $20 per $1,000 of estimated cost through FY2026 and $21 from FY2027. The second is the assessed value, which enters the levy limit as new growth and then compounds at 2.5% a year for as long as the building stands. Per $1 million of permitted residential renovation, at FY2026 tax rates:

The permanent channel overtakes the one-time fee in the second year for residential work and the fourth for commercial, and is roughly seven times larger over a decade. The permit fee is real revenue and it is not the reason renovation matters.

The assessed value appears two fiscal years after the permit is issued. Tracking 176 parcels that pulled a building permit of $500,000 or more, and measuring their assessed value against town-wide appreciation so that ordinary revaluation is netted out, the pattern is sharp:

Nothing happens in the year before the permit, the year of the permit, or the year after it. Then 24 points of value arrive in year two, 12 more in year three, and the effect is essentially complete by year four. The flatness on either side of the break is what gives the measurement its credibility: a method that found a lift everywhere would be measuring revaluation, not renovation.

Below roughly $500,000, renovation does not register in the assessment roll at all. Comparing each parcel's FY2017 to FY2025 value growth against a baseline built from unpermitted parcels of the same property type:

A parcel that spent under $100,000 on permitted work grew at the same rate as one that pulled no permit at all. This is the direct evidence for a claim this document has so far had to make indirectly: small renovation is genuinely invisible in assessment data, which is why the aggregate figure has to come from DOR's certified total rather than from adding up detected parcels. The invisibility has a location, and it is around $500,000.

Residential and commercial renovation convert to new growth at very different rates. Measured against DOR's own certified split, each dollar of permitted construction cost produces:

This sharpens the point the next section makes about commercial property. Commercial work is 55% of all permitted construction dollars in Brookline and only 18% of permits, and it is highly concentrated: the twenty largest commercial permits account for 40% of commercial construction spending, against 12% for residential. Yet it produces roughly a third as much new growth per dollar spent. Commercial property is assessed on capitalised income rather than on construction cost, so money spent fitting out a building raises its assessment only to the extent that it raises what the building earns.

Three limits on all of the above. Estimated cost is self-reported by the applicant, which adds variance to any ratio built on it, and is a reason to read these figures as central tendencies rather than precise multipliers; the residential figure above $1.00 should be read as "roughly dollar for dollar or a little better" rather than as evidence that renovation returns more than it costs. Permits join to parcels by street address, since the export carries no parcel identifier, matching 85% of permits and 82% of permit dollars; the unmatched remainder is concentrated in large rental complexes where permits are filed per unit but the complex is assessed as a handful of parcels, so renovation of large rental buildings is undercounted here. And the commercial figures span FY2020 through FY2023, when commercial values in Brookline were flat, so the commercial conversion rate is measured across a period when the underlying market was not rising.

How much more would it take?

Part 4 puts a value on this lever. Each added tenth of a percentage point of new growth, sustained, is worth about $9.3 million a year by FY2040 when it comes through renovation, which adds tax without adding service cost. Two tenths is worth about $19 million a year, more than the whole room the forecast leaves in FY2040.

Brookline's existing-property new growth runs at about 0.80% of the levy a year. Two tenths more would take it to 1.00%: about a quarter more of the renovation activity Brookline already does. Not a quarter more buildings. A quarter more of a thing that is already the single largest source of new growth in the town, and which adds no residents.

There is a useful way to size that against something familiar. The standard rule of thumb is that a property owner should set aside 1–3% of value each year for maintenance and renovation, 4% is often suggested for older housing stock, which describes most of Brookline. Measured against that rule, the ask is unmistakably modest.

Two things have to be got right before the comparison can carry weight, and getting them right strengthens it rather than weakening it.

The denominator is the building, not the property. Land does not depreciate and cannot be renovated. A reinvestment rule is really a statement about a structure, most cleanly framed as: a building with an effective life of about thirty years has to be substantially rebuilt over that period, which implies something like 3.3% of building value a year to keep it whole. Measuring against total assessed value quietly includes the land, and in Brookline that is a large distortion: land is 40% of the town's taxable value. Splitting the FY2026 roll gives $13.3 billion of land and $20.0 billion of buildings.

Added assessed value is not money spent. New growth counts only what the assessor recognises as added value. Most maintenance spending preserves value rather than adding it, so a reinvestment rate and a new-growth rate are never the same quantity, and "1.3% new growth" should not be read as "the 1–3% rule, achieved."

With both corrections, the numbers line up and mean something. Brookline's existing-property new growth of 0.80% of the levy works out to roughly $207 million of newly assessed value a year, which against $20.0 billion of building value is 1.04% a year. Two tenths of a point more takes it to about 1.30%.

Set that against the ~3.3% a year that a thirty-year building life implies, and the whole thing resolves into one sentence. Brookline is converting 31% of the reinvestment that rule of thumb calls for into new assessed value; the rest preserves value rather than adding it, which is what one would expect and is the honest reason the two figures were never the same number. Adding $19 million a year of room by 2040 does not require the town to reinvest more than the rule already suggests. It requires that share to go from 31% to 39%.

A quarter point, a natural target on the maintenance-rule reasoning above, is worth about $23 million a year by FY2040, more than the forecast's whole FY2040 room. The target does not have to be heroic to matter.

Two things this does not say

New commercial square footage and commercial property value are two different levers, and they are not close in size. Brookline's commercial new growth runs at 0.438% of the levy a year, but almost none of it is new buildings:

So "grow the commercial base" can mean two very different things. Adding value to commercial property Brookline already has is the larger channel by a factor of twenty. Adding new commercial square footage is barely measurable.

The permit records reach the same conclusion from the other direction. Commercial construction is not scarce in Brookline: it is the majority of permitted construction spending, $1.71 billion across FY2018–2026 against $1.38 billion residential. What is scarce is the new growth it produces, about 36 cents per dollar spent against $1.31 for residential. A great deal of commercial construction activity is renewal of buildings that already exist, and it moves the assessment only insofar as it moves the rent.

The 2020 Fiscal Advisory Committee put a number on what the second one would have to deliver. Its Appendix G estimated that servicing the debt on a single $100 million school would take roughly 465,000 square feet of new commercial development generating $6.3 million a year. Set against Brookline's own commercial new growth, that target is either about five years of everything the commercial base produces, or well over a century of identified new construction, depending on which of the two levers is meant.

That range is itself the finding. The parcel data carries no historical square-footage series for commercial property, so new commercial space is structurally much harder to detect than residential, and the 0.010% figure is a floor rather than a measurement. Part 4d treats this as one of the document's real gaps.

"Fiscally positive" is not the same as "good for Brookline." This is the same caution 3b ends on, and it applies here with more force, not less. The −103 net units is not a rounding error in the argument; it is the argument's cost. A town that meets its revenue needs entirely by intensifying investment in existing buildings, while adding no housing, is solving a budget problem in a housing-short region by not building housing. The largest single reason code in the renovation set is condominium record splits and merges, which, given 3b's finding that converting an existing apartment building to condominiums raised its assessed value by a median of about 80%, means some of this growth is conversion of rental housing to ownership housing, with no new dwelling created and one fewer rental available.

That is a real trade-off and this document does not resolve it. What it argues is narrower: renovation is the largest, cheapest and least contested source of new growth Brookline has; it rarely features in the public conversation; and it is by itself large enough to close the gap through 2040. Whether the town should lean on it alone is a separate question, and the answer is probably no.

3e. Three caveats, and what this part does not claim

Part 3's findings are unusually easy to over-read, and two of them, the tenure gap and the breakeven test, could be quoted in support of conclusions this document does not hold. Three qualifications, in increasing order of importance.

First: the residential exemption narrows the tenure gap. Brookline grants a residential exemption to owner-occupied homes, which reduces their taxable value below their assessed value. Condominiums are far more likely to be owner-occupied than apartments are. So the gap in tax actually paid is somewhat narrower than the gap in assessed value shown in 3b, and the condominium breakeven in 3c is correspondingly a little generous. An 80% conversion premium on an existing building is not overturned by an exemption. The roughly 20% premium on a new building is a different matter: for an owner-occupied unit, the exemption could offset much or all of it. Anyone quoting the precise breakeven numbers should know they sit at the optimistic end.

Second: the choice of per-pupil cost decides every marginal case. The two figures in 3c are not competing estimates of the same quantity; they answer different questions. $16,443 is the marginal cost, what one additional child costs when the enrollment-linked part of the budget grows with them at today's staffing ratio. $26,683 is the average cost, total school budget divided by total enrollment. The marginal figure is right for a single small project. The average figure becomes right once enough growth accumulates that the Town has to add a classroom, a teacher, or eventually a building.

Which means the honest answer is that the correct cost basis changes with scale, and no single number is right for every question. The consequence is concrete: at the marginal cost, rental apartments break even at 0.22 children per unit; at the average cost, at 0.12, below what the Town's own model assumes a typical unit produces. Under one defensible assumption most rental projects roughly cover their costs; under the other, most do not. This document does not resolve that, and any analysis that picks one figure without saying which and why should be read sceptically, including where this document uses the marginal figure, which it does throughout, because most individual projects genuinely are marginal.

Third, and most importantly: "cheaper for the Town" is not the same as "what Brookline needs." A $1.18 million condominium and a $434,000 apartment do not house the same families. Everything in Part 3 is arithmetic about the Town's ledger, and the ledger is silent on questions that many residents would weigh more heavily.

Two claims are easily conflated and should be kept apart:

• Condominiums are better for the Town's balance sheet. This is a finding about arithmetic, and it is well supported.
• Condominiums are what Brookline should build more of. This is a judgement about housing policy, and nothing in this document supports it.

The same distinction applies to 3d's renovation finding, and with more force. Renovation is the cheapest source of new growth Brookline has precisely because it adds no residents, which is another way of saying it adds no housing. A town that closed its entire budget gap by intensifying investment in buildings it already has would have solved a fiscal problem in a housing-shortage region by declining to build housing. That may be the right trade for Brookline to make; it is certainly a trade, and it should be made with the trade visible.

What Part 3 does claim is narrower and, within its limits, solid: net fiscal impact depends on the value and physical mix of what gets built, not simply on how much gets built. A conversation about encouraging growth that does not distinguish between a rental building and the same building sold as condominiums, or between new construction and renovation, is not specific enough to reach a reliable answer, in either direction.

One thing that is not a caveat

There is a version of the concern above that comes up constantly and that the evidence does answer clearly, so it should not be left hanging: does new growth cost existing residents money?

It does not, if anything, the reverse. Comparing the tax bill growth of property that already existed in FY2005 against the Town's total tax collections over the same twenty years, existing property's tax burden grew at 3.57% a year while the Town's total collections grew at 4.91% a year, a gap of about 1.3 percentage points every year, for two decades, covered by new growth rather than out of existing owners' pockets.

Run the same model forward to 2040 at Brookline's current pace of new growth and the gap persists: existing owners' bills growing at 2.34% a year against 3.72% townwide. At the peer-town growth rate Part 4 discusses, it widens further, to 2.30% against 4.36%. This is the mechanical result of everything in Part 1, new growth adds genuinely new capacity on top of the capped levy limit, so more of it means existing residents cover proportionally less of the Town's rising costs, not more.

It is worth separating that from the questions above. It is the one place in this Part where the fiscal arithmetic and the direction of the usual argument run opposite to each other, and the arithmetic is the reliable guide.

Part 4, next, turns to the two forward-looking questions: where does the current trajectory actually lead by 2040, and what real, honestly-sized levers exist to change it.

---

Part 4: Where this leads, and what could change it

4a. The forecast through 2040

Part 1e defined "sustainable" precisely: not whether revenue matches expenses in a given year, Massachusetts towns cannot run a deficit, so those two are always forced into balance by some lever, but whether automatic revenue, the 2.5% plus new growth that arrives without anyone voting on it, keeps pace with real cost growth on its own. Over the twenty years already lived through, it nearly did: 3.52% a year of automatic growth against 3.65% a year of operating spending, a shortfall of 0.13 percentage points a year.

Running the same question forward requires assumptions, and the assumptions turn out to matter more than any single result they produce.

The Town's own forecast

Start where the Town starts. The FY2027–FY2031 Long-Range Financial Plan, presented January 13, 2026, is a real, professional, published planning document, and it is the right first stop for anyone asking where Brookline's finances are heading. It shows a gap of $42.7 million by FY2031. Adding the May 2026 override, which passed four months after the plan was presented, and extending the plan's own figures mechanically to FY2040 produces this:

A deficit that opens almost immediately and never stops widening, reaching nearly a quarter of the budget by 2040, even with the override.

That deserves to be shown first, and taken seriously. It is also not the forecast the rest of this document is built on, and the reason is worth being careful about.

Why this document's forecast differs

The Town's plan is not wrong. It is a budget forecast, and a budget forecast is deliberately conservative, because being wrong in the optimistic direction is far more costly than being wrong in the other. That is the distinction Part 0 opened with.

This document builds its forecast differently. Instead of trending each budget line forward, it starts from the FY2027 budget Town Meeting adopted in May 2026 and moves it with the drivers Part 2b identified: how many staff, what each costs, what prices and health premiums do, and how many homes, students and square feet the town adds. Each assumption continues what Brookline actually did from FY2011 to FY2025, unless a published schedule says otherwise.

Running the Town's plan through the same model shows what its spending lines assume. For Town departments, pay growth of only about 1.5% a year with staff flat, below this forecast's 3.0%; on departments, the Town's plan is the more optimistic of the two. For schools, either pay growing about 6.6% a year with staff flat, or staff rising to about 1,350 positions while enrollment falls. Neither matches the record.

The school line explains most of the difference, and the reason is documented rather than inferred. Every school figure in the Town's plan is the School Department's own forecast, stitched from two versions: FY2027–29 from a December 2025 forecast, and FY2030–31 from a November 2025 forecast the School Department labelled "preliminary estimates of PSB costs prior to any analysis for potential efficiencies." Its stated assumptions are "Staffing levels remain constant, enroll remains flat," with salaries rising about 5% a year. The budget the School Department adopted in February grows about 4% a year instead.

From the Town's plan to the base case, step by step

The Town's plan ends in FY2031, so the walk compares the FY2031 balance, in dollars a year (a deficit is negative, room positive). Start from the Town's plan and apply each difference in turn:

The two biggest differences are the override that passed and school pay. Together, steps 1 and 4 are $41.7 million of the $58.5 million between the Town's plan and the base case. The override is simply a decision taken after the Town's plan was written. Schools are the real difference of view, and they turn on one number.

How the override reaches $27.3 million by FY2031. Voters approved three steps, $9.85 million in FY2027, $5.60 million in FY2028 and $7.80 million in FY2029. Each year the added levy is last year's grown by the 2.5% Proposition 2½ allows, plus that year's new step. Three smaller effects ride on top. The forecast counts new growth as a share of the prior year's levy, because a higher tax rate means each new building pays more, so new growth rises a little with the override. The forecast's building permit fees follow new growth, so they rise a little too. And a larger budget leaves a slightly larger year-end surplus, which comes back as free cash two years later.

The second row can be checked by hand: 9.85 × 1.025 + 5.60 = 15.70, then 15.70 × 1.025 + 7.80 = 23.89, then two more years at 2.5%. Columns may not add exactly because of rounding.

The Town's plan and the base case answer different questions. The Town's plan asks what happens if school costs follow the School Department's own pre-efficiency forecast. The base case asks what happens if costs and revenue keep behaving as they have. Which is closer depends mostly on one number: whether school pay keeps rising about 5% a year or something closer to the record. The teachers' contracts expired on August 31, 2026 and are being negotiated now. The Town's plan is kept live below as the stress case, because it answers a real question.

The recent record pulls in both directions. Brookline's average teacher salary rose about 4.7% a year from FY2018 to FY2025, and faster after FY2022 (DESE), which is closer to the School Department's 5% than to the FY2011–25 record of about 3.4%. The raises the last contract set were smaller: the 2023–2026 teachers' contract raised the salary schedule 2.5% in September 2023, 2.75% in each of the next two Septembers, and 1% at the end of August 2026. Teachers who have not reached the top of the schedule also move up one step for each year of service. A step adds about the same number of dollars at every level, $3,200 to $3,400, so it is worth about 5% early in a career and under 3% near the top, 3.6% on average; a teacher moving up a step received about 6.5% in FY2025 and FY2026 with the raise included, and 4.65% in FY2027. A teacher at the top step receives the raise alone, plus longevity pay of 1% to 4% after ten or more years. Over the same years the number of teachers fell by about a fifth, from 675 in FY2021 to 544 in FY2025, and a smaller, more senior staff raises the average salary without anyone's raise being larger. The data here cannot split the rise in the average among steps, the smaller staff and any one-time payments. So cost per position grows by the raise plus the steps, less what is saved when a senior teacher retires and is replaced by a new one lower on the schedule; in a workforce that holds steady, the two roughly cancel. The School Department's own FY2026 budget build prices the pieces: the 2.75% raise, steps worth 1.6% of school payroll and degree-column moves 0.1%, against budgeted turnover savings of 0.4%, about 4.1% a year per position with staffing held constant. Turnover saved more than budgeted in FY2025, when professional salaries ended $1.2 million (about 1% of payroll) under budget from vacancies and rehiring at lower steps, though some of that is positions left unfilled. At a 2.75% raise, then, cost per position grows somewhere between about 3.3% and 4.1%. The base case takes the middle of that range, 3.7%, rather than the FY2011–25 record of about 3.4%: the record includes a decade of hiring, when teachers rose from about 524 in FY2011 to 675 in FY2021 and new hires low on the schedule held the average down, and the School Department's own arithmetic for the current contract runs above it. The School Department's 5% sits above the whole range and would need raises of about 3.5% or more in the next contract. The next contract, and the count of teachers on each step, which the School Department has and this document does not, will decide where in the range it lands.

One revenue detail matters enough to name. The Town budgets local receipts (excise, meals tax, fees, permits) below what it expects to collect; actual receipts ran $11 to $15 million above budget in FY2022–25. The excess, with unspent appropriations, becomes the year-end surplus, which the state certifies as free cash and the Town spends two budgets later. This forecast budgets receipts the way the Town does, at about 74% of expected collections, and returns each year's excess as free cash two years later, at the 84% share the Town has appropriated recently. The check that this is counted correctly: FY2026 and FY2027, both adopted balanced budgets, come out at +$1.3 million and +$2.7 million.

The base case

The base case is deliberately not the most favourable one. A version in which school staff shrink with enrollment and pay holds exactly to target would show large surpluses, and neither of those happened in the last fifteen years. Both are shown below as decisions instead.

Read left to right: revenue runs about $10 million ahead of spending in FY2028, about $16 to $19 million a year from FY2031 through FY2038, and $12 million in FY2040, about 2% of revenue. No further override is needed before 2040 if costs keep behaving as they have.

The room peaks in the mid-2030s and then narrows. Once the override is fully phased in, revenue grows about 3.78% a year and spending about 3.98%, a wedge of about 0.2 points, the same kind of small, persistent gap Part 2 measured over twenty years. Without the override, spending the override funded would have run about $9 million to $14 million a year ahead of revenue through the mid-2030s; the override filled that and left room besides.

That room should not be read as a surplus that will accumulate. Brookline has consistently spent close to what it has, and the override's later steps were sized to the School Department's own faster forecast. The room is better read as capacity that the decisions below will use, one way or another.

What decides it

Four decisions each move the FY2040 balance by $20 million to $46 million a year, as much as the whole room or more, and three of them are Brookline's to make.

Pay settlements. One point of pay growth a year for all staff is worth $46 million a year by FY2040 and $272 million over the fifteen years. School wages at the School Department's own forecast rate, about 5% a year including steps, would cost $35 million a year by FY2040. Holding school and Public Safety pay to the Town's 3% target instead of their recent premiums would add $31 million.

Health insurance. Premiums settling at 10% a year instead of 8% cost $32 million a year by FY2040; 6.5% saves $20 million. Health insurance is the line to watch whatever happens: it more than triples, from $46 million to $143 million, and by FY2040 is nearly as large as all Town departments combined. Brookline controls its share of the premium and the plan design, not the rate.

School staffing as enrollment changes. If enrollment falls as NESDEC projects and enrollment-linked positions follow it down, FY2040 improves by $22 million a year. If staff fall half as fast as enrollment, as they did from FY2019 to FY2025, it improves by $16.5 million. If enrollment instead rebounds as the School Department's own demographer projects and staff follow it up, FY2040 is $18 million worse. The two forecasts disagree by about 1,300 students by 2033–34.

The pension turn in FY2031. When the pension system reaches full funding, about $37 million a year of payments toward past underfunding ends. The Town's schedule moves that money into prefunding retiree health, whose unfunded liability was $201.5 million in 2024. The forecast follows the schedule. Not moving it would leave $35 million a year of room; moving it builds down the Town's largest remaining long-term liability.

When the room runs out depends on which of these goes the wrong way:

Growth moves the answer less than any of these in the first decade and can move it more by 2040, depending on its kind. Reaching the 80th-percentile peer growth rate through commercial space adds $57 million a year by FY2040; through housing alone, $12 million (Part 3c explains why). Chestnut Hill, counted on top of trend growth from FY2032, is worth about $5 million a year by FY2040.

Where the money goes

The pension and retiree health lines swap places in FY2031, and total spending barely notices, which is also why a two-year delay in reaching full funding changes the fifteen-year total by only $3 million. FY2027's capital, reserves and other line includes a $3.6 million deposit to a new Override Stabilization Fund, assumed here to be drawn back into the budget evenly over FY2028–30; the Town documents reviewed do not state the plan.

How sure these numbers are

• The method's own error. Run backward over FY2011–25 with the actual staff counts, enrollment and price indices, it predicts Town departments 3.8% too high and schools 5.7% too low.
• Enrollment. The two 2023 forecasts disagree by about 1,300 students by 2033–34. The base uses NESDEC because births to Brookline residents fell from about 690 a year (2008–13) to about 490 (2020–22) and 422 in 2023 (3c sets out what that means), and UMass Donahue Institute projections have the 5–14 age group falling about 28% from 2020 to 2035; the School Department's forecaster has tracked the last three years slightly better.
• The school variable share. Two-thirds of the school budget is treated as moving with enrollment, RKG's judgement with Town staff rather than a measured response.
• One pay rate per group. Contracts differ by bargaining unit and year, and the teachers' contracts are open.
• Free cash depends on the Town continuing to budget receipts conservatively and appropriating about 84% of each surplus.
• A cross-check. This project's earlier forecast grew each budget line at its own historical rate. With the same revenue it shows +$11 million in FY2031 and +$22 million in FY2040. Total spending in FY2040 is close ($671.7 million against $684.7 million); the mix differs, with benefits about $15 million higher here and schools about $7 million lower.

One mechanical note: debt exclusions are left out of both sides throughout, because voters approved a separate tax that pays that debt service exactly.

4b. The balance sheet: liabilities, reserves, and the rating

Almost everything above is about annual flows, revenue in, spending out, the gap between them. The Town's balance sheet is a different question: not what it spends each year, but what it owes. And on that measure the next fifteen years look markedly better than the operating picture does.

Brookline's total liabilities stood at $914 million in FY2024, the most recent audited figure. Three items are 95% of it: long-term debt, net pension liability, and net OPEB liability. Projecting each forward on its own real mechanics rather than a blended assumption:

Total liabilities fall from $914 million to roughly $203 million by FY2040.

The pension line is the most reliable of the three, because it is not a projection in the usual sense; it is the Town's own actuarially determined funding schedule, which retires the unfunded pension liability to zero in FY2031. The debt line is similarly concrete: existing bonds run down on their real contractual amortisation schedules, with four already-approved borrowings layered in.

The OPEB line is the interesting one, and it is where the pension payoff shows up. Retiree health liability keeps rising through about 2030, because the benefit accrues faster than the Town is currently funding it. Then the pension full-funding milestone arrives, roughly $43 million a year of contribution capacity is freed, it redirects to OPEB, and the liability falls sharply.

Two honest qualifications. First, that projection has the OPEB liability reaching zero around FY2037, which almost certainly will not happen literally, a town would not keep depositing $43 million a year into an already-fully-funded trust. Treat it as a signal that the contribution schedule would get revised well before then, not as a forecast that the liability vanishes on a particular date. Second, and more materially: the Pierce School has roughly $114 million of already-authorised borrowing that is not in these figures. Including it would meaningfully raise the debt line. The $203 million figure is a defensible floor, not a confident point estimate.

Even with both qualifications, the direction is clear and it is the strongest piece of good news in this document. The operating outlook in 4a depends on decisions still to be made. The balance sheet does not; it is on a path to being substantially resolved by 2040, largely because of decisions the Town has already made and is already funding.

Reserves: the third lever, and the one that moves quietly

Liabilities are one half of a balance sheet. The other half is what the Town has set aside, and it matters for a reason the annual budget conversation tends to miss. Reserves are the third thing that actually closes a gap, alongside overrides and cuts, and unlike those two they can absorb a shortfall for several years before anyone votes on anything.

Brookline's reserve position has roughly doubled in five years:

"Available fund balance" is the measure credit agencies use, meaning committed plus assigned plus unassigned. For Brookline it equals total General Fund balance, because the General Fund carries no nonspendable, restricted or committed amounts. The narrower GAAP unassigned figure is lower, 15.33% in FY2024, and the two get quoted interchangeably more often than they should be.

The Town's own formal Reserve Policy sets a floor of 20% of revenue, with a stated goal of reaching the median of Massachusetts Aaa communities. Brookline now sits above that floor, at 22.91%. Certified free cash tells a consistent story, rising from roughly $7 million a year through FY2016 to $21 million to $25 million in recent years.

There is no FY2025 audit yet, so FY2024 is the most recent verified year.

What the rating actually turns on

Brookline holds a Aaa rating from Moody's, one of only fourteen Massachusetts municipalities that do. That rating is worth real money. The spread between AAA and AA municipal bonds was roughly 10 basis points at ten years and 20 at twenty to thirty years in September 2026, which on Brookline's projected debt is worth a few hundred thousand dollars a year.

Moody's names two credit challenges for Brookline, and neither is the one this document has spent the most time on. They are below-average reserves for the rating category, and elevated leverage from recent school construction. Pensions and OPEB, which occupy most of this section, are not among them.

That pattern is not specific to Brookline. Across every Massachusetts downgrade since 2010 whose stated rationale could be verified, the driver was reserve depletion or structural imbalance. Sharon lost a notch in 2013 with Moody's citing shrinking reserves and the town's resistance to raising its levy to capacity. Braintree lost one in October 2025, with Moody's describing "a structural imbalance though as growth in revenue is not keeping up with rising costs" after its available balance fell from 29.9% to 13.5% in a single year. Pension and OPEB liabilities appear in these reports as secondary factors, not as causes.

Two things are worth holding together here rather than choosing between them.

Against its own standard, Brookline is meeting the test. Reserves are above the Town's 20% policy floor, and above the level at which Moody's says it would look again.

Against its rating peers, Brookline is at the thin end. Moody's Aaa communities nationally run available reserves around 62% of revenue. Brookline's most recent published figure is 27.1%. On Moody's own scorecard this sub-factor scores an A where several Massachusetts peers score Aa or Aaa.

The resolution is that every Massachusetts Aaa community is thin by national standards. Wellesley sits around 30%, Concord at 27.0%, Wayland at 25.6%, and Moody's has told Wellesley and Concord the same thing it tells Brookline. Massachusetts towns run leaner reserves than Aaa communities elsewhere, and Brookline is unremarkable within that group rather than an outlier from it.

One comparison to treat carefully: Lexington, Hingham and Newton report stabilization funds inside their fund balance, which lifts their figures relative to Brookline's. Comparisons built on audited statements rather than on the agencies' own series will overstate the gap.

When reserves would come under pressure

Rolling the last audited balance forward and letting reserves absorb each year's operating result gives a usable picture of what they are insuring against:

On the base case, reserves are never drawn on, because revenue stays ahead of spending for the whole window. The second row is the stress case: if costs follow the Town's plan, reserves fall below the Town's own 20% floor in FY2028 and run out around FY2033. The decisions in 4a sit between the two; pay and premiums both running high would turn the operating balance negative from about FY2033, and reserves would start to absorb it from there.

This is not a prediction of a downgrade, and it should not be read as one. Agencies weigh a full scorecard, they disagree with each other about the same town in the same year, and a single ratio crossing a threshold is a flag rather than a mechanism. What the exercise establishes is narrower and still useful: on the current path reserves are not the binding constraint, and on the stress path they stop being available around FY2033.

It is also worth being clear about what a downgrade would and would not cost. The interest penalty is real but modest, in the low hundreds of thousands a year against a declining debt portfolio. The expensive part is the condition that produces a downgrade, not the downgrade itself. The rating is a signal of sustained deficits and falling reserves rather than the harm they do.

4c. The menu of levers

Section 4a set the terms these levers are measured against, and they are not a deficit. On the base case, the override leaves room of about $10 million to $19 million a year through 2040, and no further override is needed within the window. Against that, four decisions are each worth $20 million to $46 million a year by FY2040. So the question is not "how do we close a gap." It is: which decisions use the room, which protect it, and what would keep the next override away once the wedge reopens?

None of these is presented here as the answer, advice taken from two sources that don't usually get cited together: the ERSC Final Report's explicit caution against point estimates (citing the risk to public trust when a single number turns out wrong), and the Town's own 2030 Roadmap (approved by the Select Board May 19, 2026), whose Goal 3, "Increase the Town's fiscal health and stability", already names several of the levers below as adopted Town strategy.

What the Town pays for the services it already provides. This is the largest lever, and the one the usual list leaves out. It is not cutting: it changes what existing services cost, not what the Town provides. Its parts, each sized in 4a:

• Pay settlements, worth about $46 million a year by FY2040 for each point of pay growth. Brookline's average teacher salary, $132,000 in FY2025, is the highest of the 17 peer districts with data (peer median about $112,000), and school cost per staff member ran above the national public-sector benchmark over FY2011–25, as Public Safety did in FY2021–25. Those are measured facts about pay levels and growth, not a judgement on any settlement; pay is also how Brookline recruits and keeps staff.
• Health insurance, worth $20 million to $32 million a year by FY2040 either way, through the Town's premium share and plan design.
• School staffing as enrollment changes, worth about $22 million a year if enrollment-linked positions follow a falling enrollment. The ERSC's own School Model Analysis, a study of consolidating classrooms across school buildings, is this lever by another name; it recommended a FY2027 study rather than a number. K–8 classes now average 17 to 21 students against the district's guidelines of 22 to 25.
• The pension turn in FY2031, a $35-million-a-year choice between prefunding retiree health, as scheduled, and other uses.

The ERSC's quantified savings belong here too. Its Solar Initiative ($3.2 million capital investment, approved 2026-01-22) is projected to avoid about $390,000 a year in electricity costs; an Electric Demand Management Strategy was recommended for further study, not costed. The report also documents Fire Department overtime running roughly $800,000 a year over budget, which the Town attributes to leave rules in the FY2018 contract, injury leave up about 150%, and a 2022 arbitration award. Against a $146 million deficit these read as rounding errors. Against the decisions in 4a they are the right kind of item, and a small one: what moves the answer is the handful of recurring decisions above, each made in its own cycle.

New growth. Brookline's trailing FY2021–25 new growth has run at 1.28% a year of the prior year's levy limit. Each added tenth of a percentage point, sustained, is worth by FY2040:

The difference is cost. Renovation adds tax without adding residents; housing adds residents whose schooling costs about what their homes pay (3c). At the 80th-percentile peer rate, about 1.97% on the same trailing window, the FY2040 balance improves by $57 million a year if the added growth is commercial, $55 million at today's mix, and $12 million if it is all housing.

Growth is slow: value added in FY2027 is worth far more by FY2040 than the same value added in FY2038, and none of it helps in the year it is permitted. Its natural role is to protect the room against the wedge that reopens in the late 2030s and against the decisions that go the wrong way, not to settle any one year's budget.

New growth is modelled in dollars, and homes are a translation: dollars divided by the assessed value of a new unit. That divisor is $430,000, a units-weighted average calibrated to the Town's own permit count of 166 net new homes a year across 2022, 2023 and 2024. It is the centre of a range running from about $325,000 to $580,000 depending on which permit years are used, so unit counts in this document are approximate. No dollar figure depends on it.

The scale of this lever is best shown in the Town's own arithmetic. The June 2026 presentation that produced the per-unit figures in Part 3a put them to a second use: sizing how much development it would take to fill a future override, which the Planning Department put at $25 to $28 million a year. The answer, on the Town's own numbers, is 1,563 to 2,154 condominiums over three years, or 4,630 to 16,000 apartments, or ten to eleven office campuses the size of 1 and 2 Brookline Place, or fourteen to sixteen hotels. Every one of those figures reproduces exactly from the per-unit impacts and the dollar range.

Set beside them, on the next slide, is what Brookline actually built: 499 net new homes over the same three years.

Three years of the town's housing production, valued at the Town's own per-unit figures, covers somewhere between a thirtieth and a third of one override, and the condominium row is the generous reading, since most of what Brookline builds is rental. This document's own per-home test in 3c reaches the same conclusion by a different route: housing is not a substitute for the levy, and its fiscal case rests on value and tenure, not volume. Renovation and commercial space are where growth adds net revenue.

The 2030 Roadmap sets a citable target for this lever, $15 million to $20 million of new growth revenue between FY2026 and FY2030 (Goal 3 KPI). Checked against this document's own figures: trend growth alone implies $20.7 million over that window, and the peer-target scenarios $31.9 million, both above the Town's stated target. The Town's target reads as roughly a continuation of the FY2018–25 trailing average of about $3 million a year, so the trend already tracks close to what the Town itself has called sufficient.

A second Goal 3 target asks for site control and conceptual planning for projects that "would likely see $15 M of new growth upon construction completion." If that means $15 million a year of added levy, the usual sense of new growth, it is a large ambition: about three projects the size of the approved Chestnut Hill plan (about $4.5 million each), or roughly $1.5 billion of new residential value at today's tax rate, or $875 million of commercial. If it means $15 million of added assessed value, it is a small fraction of that. The wording supports either reading, and the difference is worth confirming with the Town.

Permitting and zoning reform. This is not a separate lever so much as the mechanism that produces new growth: streamlined permitting and zoning are how the investment described above actually happens. The 2030 Roadmap names this too (Goal 3, Strategy B: "increase revenue through development," including a streamlined permitting process and an annually-updated five-year economic development plan).

The permit-lag case studies show real permit-to-substantially-assessed timelines running from roughly 15 months (a straightforward single-family new build) to 40 months (a 153-unit institutional conversion). The question this lever has to answer is specific and tractable: would compressing those timelines change the timing of enough value to be worth a tenth of a point, about $9 million a year by FY2040 if it comes through renovation? Because growth compounds, timing is not a second-order detail: value that lands three years earlier compounds for three more years. Named here as a real lever with a real data foundation; quantified in a later pass.

Overrides. The lever every real historical gap has actually been closed with, four operating overrides since 2005, plus the fifth in May 2026. On the base case this lever has already been pulled, and far enough: no further override is needed before FY2040. If pay or premiums run high, the room is gone in the mid-2030s, and the next override arrives then rather than after 2040.

The stress case shows what relying on this lever alone looks like. Taking the Town's own figures extended as-is, and asking what size override, called every three years (matching the ERSC's own finding that override intervals have compressed), would restore balance each time: four new overrides, FY2029 ($7.6M), FY2032 ($18.7M), FY2035 ($32.0M), FY2038 ($42.1M), each roughly double the one before, compounding to $113.2 million a year of permanent added levy capacity by FY2040. Roughly three times the entire FY2005–2025 override total ($38.7 million compounded, Part 2e), raised in about three-quarters of the time, and even then the budget is not closed: two years after the last, the gap has reopened to −$32.8 million. That is the "compressing intervals, growing amounts" pattern the ERSC flagged, shown as a quantified consequence rather than an observation.

Putting them side by side:

Three observations follow. First, the largest levers are decisions Brookline already makes every year, in bargaining, benefit design and the school budget, and none of them requires cutting a service. Second, they interact: the room the override created is one pool, and a decision that uses it on one line is not available for another. Third, growth is the slow lever: value added early is worth a multiple of the same value added late, which is the practical case for starting now even though the base case needs no new revenue before 2040.

Where these decisions are already on the Town's calendar

Most of these levers already have a place in the Select Board's 2030 Roadmap. The start dates are from the Town's public Goal 3 action tracker, which is updated quarterly and showed no progress entries as of September 2026.

Three things stand out.

The pension turn is already scheduled for decision. Action 3.D.3 will set how much the Town pays toward retiree health above the actuarial minimum, and by when it reaches full funding. That is the FY2031 choice 4a sizes at up to $35 million a year. The Roadmap puts the policy in front of the Select Board from mid-2027, three budgets before the money arrives, which is early enough to decide it deliberately.

On pay, two things are worth holding together. The Roadmap commits to competitive compensation, and pay is how the Town recruits and keeps staff; vacancies have costs of their own, as the Police Department's recent years of running 14 to 19 officers below authorized strength showed. And each point of pay growth a year is worth about $46 million a year by FY2040. The Roadmap does not put a price on its commitment; 4a does, and the two belong in the same conversation.

Two of the four largest decisions are outside Goal 3. Health insurance and school staffing do not appear in it. School staffing is the School Committee's to set, and premium growth is largely set by the market, so this may be deliberate scope. They are also, with pay, the lines that move the forecast most. The Town–School Partnership, which action 3.A.1 asks the Select Board to consider alongside the Expenditures and Revenues Study Committee, is the natural place for them, and the five-year priority list is the natural place to weigh all four together.

4d. What this document does not answer

A forecast that lists only its findings is easy to over-trust. This section lists what is missing, in rough order of how much it could move the conclusions, so a reader can discount accordingly rather than having to guess where the soft ground is.

What the teachers' contracts will settle at. This is the largest single number in the forecast, and it is being decided now. The base case assumes school pay grows 3.7% a year; the School Department's own forecast assumes about 5%. The difference is about $35 million a year by FY2040, about three times the room the base case leaves. Nothing in this document can predict the outcome; it can only size it.

Whether Brookline's zoning can actually accommodate the growth in 4c. This is open deliberately. Section 4c's growth lever is stated as a rate, valued per tenth of a percentage point, with implied physical equivalents in homes and square feet. What it does not establish is whether current zoning permits that, where it would go, or what would have to change. The scenarios treat the implied unit and square-foot counts as an output of the arithmetic, not as a claim about what Brookline would allow.

That gap matters most for the renovation lever in 3d, where it cuts the optimistic way: an addition or a finished attic is far less constrained by zoning than a new building is, which is part of why renovation is the cheapest channel. It matters most against the growth lever generally, where it cuts the other way: a rate that is arithmetically modest may still be physically or politically unreachable, and this document cannot say which.

Two physical limits the growth lever would run into. The Planning Department's June 2026 look-ahead flags both, and neither is modelled here. The first is the electric grid: Eversource's summer capacity maps show only some parts of town as "likely better for 40+ unit buildings," so where large buildings can go depends partly on where the grid can serve them. The second is open space: the deck contrasts about 1.5 acres of playground and field per 1,000 people near the Baker School with about 0.7 near the Ridley School and Winthrop Square, against a published range of 1.25 to 2.5 for a community recreation park. Both bear on where added housing could go and what it would need to come with.

Roughly $114 million of authorised Pierce School borrowing is not in the balance sheet. Section 4b projects total liabilities falling from $914 million to about $203 million by FY2040. The Pierce School's full PreK-8 rebuild is approved and its borrowing authorised, but it is not in those figures. Including it would raise the debt line materially. The $203 million is a defensible floor, not a point estimate, and 4b says so, but it is worth repeating here because the balance sheet is the most encouraging finding in the document and it is the one most exposed to a single omission.

Which per-pupil cost basis is right at scale. Section 3e covers this properly, and it is repeated here only because it propagates: the difference between $16,443 marginal and $26,683 average per pupil decides every marginal residential project, and the correct basis shifts from the first toward the second as growth accumulates. This document uses the marginal figure throughout, which is right for individual projects and progressively too generous as the volume of growth rises. A scenario that delivered the full peer-target growth rate would need the average figure for at least part of it.

The FY2027 revaluation is not modelled. Every assessed-value figure here rests on the FY2025 and FY2026 rolls. Massachusetts requires periodic full revaluation, and Brookline's next one will move values, including, potentially, the condominium-versus-rental relationship in 3b that carries much of Part 3. Nothing suggests it will reverse that relationship, but the specific breakeven numbers in 3c will need re-deriving afterwards rather than being carried forward.

Commercial new growth is measured as a floor, not a quantity. As 3d notes, the parcel data carries no historical square-footage series for commercial property, so commercial expansion is structurally much harder to detect than residential. The 0.01%-a-year figure is what could be identified, not necessarily what happened. This is the one place in the document where a number could be materially understated for reasons of measurement rather than substance.

Timing effects from permitting are named but not sized. Section 4c documents real permit-to-assessment lags of 15 to 40 months and declines to convert them into a scenario, because doing so credibly requires a target to size against. 4a now supplies that target, so this is a next step rather than a permanent gap, but as of this document, the permitting lever's value is asserted to be real and not quantified.

Two mechanical exclusions, stated for completeness. No future override or debt exclusion beyond the May 2026 one is modelled anywhere; that is by design, since those are the levers being sized rather than assumptions to bake in. And the forecast charges new growth its costs, $660 a year of Town services per home plus the schooling of its students, but at average rates: it does not model the step costs of a new classroom, fire company or school building, which arrive in lumps rather than smoothly.

None of these is a reason to discard the conclusions. The two central findings, that the override leaves real but modest room through 2040 and that a few recurring decisions about what existing services cost outweigh everything else, survive all of them, because they turn on growth rates and cost trends rather than on any single asset, project, or valuation. What the list should change is a reader's confidence in the precise numbers, which is lower than the arithmetic's tidiness suggests, and their confidence in the direction, which is high.