From FY14 through FY24, Marblehead appropriated more for Group Insurance than it actually spent every single year. The average gap was about $2.56 million a year, or roughly 19% of the budgeted line. Over the eleven years that came to $28.2 million in unused appropriations. None of it was lost or hidden. Each year's leftover was certified as free cash and re-appropriated the following year, mostly to reduce the operating tax rate. The ACFR said so in plain language in nearly every year's MD&A section. Then in FY25 the pattern broke: the budget was de-appropriated by $1.1 million mid-year, the line came in 2.7% under final, and the standard insurance-surplus paragraph disappeared from the MD&A.
11 years of structural under-spend, then FY25 broke it. The line came in 12–22% below its final appropriation every year FY14–FY24. In FY25 the original $14.47 million budget was de-appropriated mid-year to $13.37 million, and actual came in 2.7% under that revised number, in line with peer-town discipline.
Peer towns book 0–5%. Swampscott, Cohasset, and Wellesley all come within 1.5% on the same line. Marblehead's FY14–FY24 pad was roughly four to fifty times larger than what comparable Massachusetts municipalities leave on it; the FY25 result puts Marblehead inside the peer range for the first time in twelve years.
The surplus had a destination. Through FY24, the auditor's own commentary ended in identical language year after year: the surplus "was offset by the use of prior year resources to balance the [next year's] budget."
The free-cash draw is shrinking with it. FY24 used $8.0 million of unassigned fund balance to balance the operating budget; FY25 used $6.5 million; FY26's budget was approved with a smaller planned draw still. The cushion that quietly balanced operations for a decade is no longer there.
The pattern
The two bars per year are the Town's final budgeted appropriation for Group Insurance (the outlined bar) and the actual amount spent by year end (the solid bar). The number above each pair is the percent of the budget that went unspent.
Budgeted (final appropriation) Actually spent
Source: Town of Marblehead Annual Comprehensive Financial Reports, FY14–FY25, "General Fund Schedule of Revenues, Expenditures and Changes in Fund Balances – Budget and Actual," Group Insurance line.
Where the surplus goes
The Town's audited financial reports describe the mechanism in unusually direct language. Five of the eleven MD&A sections use a nearly identical formulation. Here is FY20, on Marblehead's FY20 results:
Appropriations exceeded actual expenditures and encumbrances by $4.2 million mainly due to group insurance surplus of $2.8 million; these surpluses were offset by the use of prior year resources to balance the 2020 budget.
Town of Marblehead FY20 ACFR, Management's Discussion and Analysis
"Prior year resources" is the auditor's term for free cash. Unspent appropriations close to fund balance at the end of the fiscal year on June 30. The Massachusetts Department of Revenue then certifies a free cash amount each fall, and Town Meeting can re-appropriate that money the next spring. In Marblehead, the largest recurring use of free cash is a single warrant article titled some variation of "Available Funds Appropriated to Reduce the Tax Rate." That article reduces the amount that must be raised from the property tax levy to fund the next fiscal year's operating budget. Capital and stabilization use of free cash is much smaller; in FY22, the breakdown was roughly $8.79 million to reduce the tax rate versus about $51,000 split between special articles and prior-year bills.
Group Insurance is not the only source of surplus. Other under-spends and revenue over-collections also flow into free cash. Insurance is, however, the single largest contributor that the auditor calls out by name almost every year, and it explains roughly a third of total free cash appropriated to balance the next year's budget.
Source FY
Insurance surplus
Next FY free cash used to balance budget
Insurance share
FY14
$1.43M
$4.50M (FY15)
32%
FY15
$2.32M
$5.46M (FY16)
42%
FY16
$2.39M
$6.03M (FY17)
40%
FY17
$2.74M
$7.10M (FY18)
39%
FY18
$2.70M
$7.90M (FY19)
34%
FY19
$2.69M
$8.58M (FY20)
31%
FY20
$2.81M
$7.20M (FY21)
39%
FY21
$2.57M
$8.79M (FY22)
29%
FY22
$2.65M
$10.20M (FY23)
26%
FY23
$3.47M
$8.00M (FY24)
43%
FY24
$2.41M
$6.50M (FY25)
37%
FY25
$0.36M
FY26 in progress
n/a
Free cash totals come from the General Fund Budgetary Comparison Schedule in each year's ACFR. The Town Meeting article that re-appropriated each year's free cash is the "Available Funds Appropriated to Reduce the Tax Rate" article in the corresponding Annual Town Meeting warrant. FY25 free cash drops to $0.36 million from the prior decade's ~$2.5 million; the recycling chain is effectively cut.
The Marblehead Finance Committee flagged this dynamic in its 2019 report, noting the town's growing structural reliance on prior-year surplus to balance the operating budget. Six years later, the FY25 ACFR's general-fund narrative drops the standard "insurance surplus" disclosure entirely and replaces it with: "Operations of the general fund were consistent as revenues and expenditures approximated each other."
How this compares to peer towns
Marblehead's Group Insurance pad is far larger than what comparable Massachusetts municipalities leave on the same line. The bars below show the most recently available ACFR for each town. Wellesley and Cohasset are clean apples-to-apples comparisons because their schedules isolate a single Health Insurance or Group Insurance line. Hingham's broader Employee Benefits line and Lexington's Insurance line bundle in other small items, so their numbers slightly overstate the true health-insurance pad.
Each town's most recently available ACFR, Group Insurance or Employee Benefits line. Marblehead is shown twice: its FY24 single-year value (16.6%) and the average across all eleven years FY14–FY24 (18.7%). Sources: Swampscott FY24 audit p.91; Cohasset FY23 ACFR p.82; Wellesley FY24 Financial Report p.104; Hingham FY25 ACFR p.92; Lexington FY24 audit p.83.
Cohasset, Wellesley, and Swampscott all participate in the same state-administered GIC plan that Marblehead joined in July 2012. Their GIC-driven premium uncertainty is no smaller than Marblehead's, yet they true their budget up to within a couple of percent of actual. Swampscott is particularly worth noting: its FY24 Group Health Insurance line was originally budgeted at $7.53 million and was de-appropriated mid-year by $826,405 to a revised $6.71 million, almost exactly matching the eventual actual. That kind of mid-year true-up is what discipline on this line looks like; Marblehead's mid-year revisions on the same line are typically under $100K on a $14 million budget.
FY25 broke the pattern
The FY25 ACFR, published in June 2026, shows the line behaving differently for the first time in twelve years. Three things changed at once:
Mid-year true-up. The original $14.47 million Group Insurance appropriation was de-appropriated by $1.11 million during the year, taking the final budget down to $13.37 million. In the prior decade, mid-year adjustments to this line never exceeded $130K. The FY25 reduction is the kind of action peer towns like Swampscott book annually.
Actual came in close to revised. Spending was $13.0 million, or 2.7% under the de-appropriated final – on par with Cohasset (0.3%), Wellesley (1.5%), and Swampscott (0.0%). It is the first year in twelve that Marblehead's Group Insurance line lands inside the peer-discipline range.
The MD&A drops the recycling clause. Every year FY16 through FY23 included some variant of the sentence: "these surpluses were offset by the use of prior year resources to balance the [next] budget." The FY25 ACFR does not. Its general-fund narrative reads: "Operations of the general fund were consistent as revenues and expenditures approximated each other."
FY26 in-progress data is consistent with the change holding rather than reversing. Through May 29, 2026, Group Insurance has spent $14.55 million against a $15.20 million budget; linear extrapolation to year-end gives roughly $15.88 million, or 4% over the original budget. If the FY25-style mid-year true-up repeats in FY26, the final budget will be revised up to reflect the actual cost; if it does not, FY26 will be the first year in a decade where the line ends over rather than under. Either way the historical 18%+ pad is gone.
The likeliest mechanical drivers of the change: the GIC premium hike of 11.1% effective July 2025, plus tier-mix shifts and headcount changes, consumed the cushion that had been hidden in the prior decade's appropriation. Whatever the proximate cause, the operating budget no longer has a $2.5 million-per-year cushion running through this line.
Three ways to read this
The numbers are not contested. The framing is.
Defensible: this is how municipal finance works
Premium rates are set in February for a July–June fiscal year. Healthcare claims experience is volatile. A conservative appropriation that turns back to free cash and is re-approved by Town Meeting is a standard tool a town can use to manage uncertainty without a separate health-insurance reserve fund. Every dollar was audited, disclosed in the MD&A, and re-appropriated by an open Town Meeting vote. The FY25 mid-year true-up shows the town can also adjust the line down when it has better information.
Critical: a 19% pad is not actuarial uncertainty
Peer towns with the same GIC plan and the same February rate-setting timeline true their budgets up to within 2% every year. An 18.7% pad sustained for eleven years is not a prudent cushion against claims volatility – it is a deliberate budgeting practice that produces a predictable, recyclable surplus. The Finance Committee itself flagged this in 2019 as a structural rather than incidental pattern. That Marblehead's pad collapsed to 2.7% in FY25 once the budget situation became politically visible is consistent with the practice having been a choice, not a constraint.
Override-relevant: the cushion ran out in FY25
If the operating budget was balanced for eleven years using a $2.5 million-per-year surplus running through the insurance line, the right question about the FY27 override is not "are costs newly exploding?" but "is the surplus that used to balance the budget no longer there?" The FY25 numbers settle that question: the surplus is gone. The recycling that quietly funded the operating budget for a decade ended at the same moment the override appeared on the calendar.
Open questions
Why did the practice change in FY25 specifically? FY25 closed June 30, 2025, six months before the FY27 override became a public conversation. The mid-year de-appropriation happened during the year, so the decision to true the line up predates the override debate, but not the FY26 budget tightening that drove it. No public document explains the change in posture.
Why didn't the pad narrow sooner? Peer GIC towns track to within 2% of actual on the same line every year. The town's posted budget books since FY14 do not explain why Marblehead's pad ran four to fifty times larger for so long. The Finance Committee's 2019 report described the reliance on surplus but did not document an internal decision to maintain it.
Are other line items doing the same thing on a smaller scale? The FY22 ACFR breakdown of total general-fund under-spend ($3.3 million) versus the Group Insurance component ($2.6 million) implies about $700K of additional pad spread across other lines. Identifying those lines would refine the "where did the cushion come from" picture.
How will FY27 actual track against the $16.75 million budget? If FY27 also lands within a couple of percent of its final budget, the 2.7% FY25 result is the new baseline. If FY27 under-spends substantially, the pattern has not actually ended – it just paused for a year.
Sources and methodology
FY14–FY25 budget vs actual. Each year's Group Insurance row was extracted from the General Fund Schedule of Revenues, Expenditures and Changes in Fund Balances – Budget and Actual in the corresponding Town of Marblehead ACFR. Original budget, final budget, actual, encumbrances, and variance are recorded in data/group_insurance_budget_vs_actual_FY14-25.csv. ACFR PDFs are at github.com/agbaber/marblehead/releases/source-archive-v1.
MD&A quotes. Verbatim. The same clause about "use of prior year resources to balance the [next year's] budget" appears in the FY16, FY17, FY19, FY20, and FY22 ACFRs. The FY14, FY15, FY18, and FY23 ACFRs use slightly different wording but report the same mechanism. The FY25 ACFR drops the clause and instead says the general fund's revenues and expenditures "approximated each other."
Free cash flow. Free cash appropriated to balance each subsequent year's budget comes from the General Fund Budgetary Comparison Schedule in each ACFR, captured in data/general_fund_budgetary_FY15-24.csv. FY25 used $6.5 million in unassigned fund balance to balance the budget, down from $8.0 million in FY24 and $10.2 million in FY23.
Peer town extractions. Swampscott FY24 audit (Group Health Insurance line, p.91, final budget de-appropriated mid-year from $7.53M original to $6.71M); Cohasset FY23 ACFR (Health Insurance line, p.82); Wellesley FY24 Financial Report (Group Insurance: Expenses line, p.104); Hingham FY25 ACFR (Employee Benefits aggregate line, p.92); Lexington FY24 audit (Insurance line aggregating health insurance with workers' compensation transfer to internal service fund, p.83). Raw data in data/peer_group_insurance_budget_vs_actual.csv.
FY26 in-progress numbers. From the Town's checkbook export as of 2026-06-01, period end 2026-05-29, captured in data/budget_actual_FY26.json. The full-year projection is a linear extrapolation from month 11 of 12 and is approximate; insurance billings are roughly level monthly but not exactly.
Definitions. "Group Insurance" on the Town's budget schedule is the Town's own appropriation for its share of employee and retiree health insurance premiums, net of payroll-withheld employee contributions. It is not the same as the gross GIC invoice total, which includes the employee share that retirees and active employees pay themselves and which never flows through the Town's appropriation.