Section summary
In the House, Democrats holding any paid post almost never voted the reform side: 5 of 349 member-sessions. Of House Democrats who began a session without a stipend, 62% of those who cast no reform-side vote had one by the second year, against 14% of those who voted for reform. A member’s earmark money does not predict the vote, but among House Democrats district earmark money naming the town (first measure) is 9.5% higher per resident for each $10,000 of the member’s stipend and 19% lower, about $6.6 million over nine budgets, for reform-side voters. On 42 members that would seem to indicate a link, and the data cannot show which way any influence runs; the Senate shows neither pattern.
On February 25, 2025, the House voted on an amendment by Minority Leader Bradley Jones Jr. that would let the State Auditor audit the House. Nearly four months earlier, 72% of voters had said she could. The amendment failed, 24 to 127. One Democrat voted for it.
That vote fits a pattern. This report uses 152 recorded roll calls from 2017 to 2026 on the Legislature’s own rules, pay, transparency and leadership: term limits for the Speaker (Robert DeLeo until December 2020, then Ronald Mariano) and the Senate President (Stanley Rosenberg, Harriette Chandler, then Karen Spilka from July 2018), public committee votes, time to read bills, audits and the public records law. In the House nearly every reform failed by about the same margin, the Republican caucus plus a handful of Democrats.
The test asks whether what a legislator receives from the institution predicts how he or she votes on the institution.
What members receive
Every senator holds a paid post, most assigned by Senate President Spilka. In the House about 109 of 160 members do, most assigned by Speaker Mariano; the Minority Leaders, Sen. Bruce Tarr and Rep. Bradley Jones Jr., assign the Republican posts. In 2025 those stipends came to $5.43 million for 149 members. Most members also win local earmarks.
Stipends track votes; a member’s earmarks do not predict them
In the House, members holding any paid post almost never voted the reform side. Among House Democrats without a stipend, 17.9% cast at least one reform-side vote on a divided roll call. Among stipend holders it was 5 of 349 member-sessions, and none in the top two tiers. Because holders so rarely voted the reform side, the per-dollar estimate among Democrats, an odds ratio of 0.34 per $10,000 of stipend (95% interval 0.24–0.48), mainly restates that divide; it does not measure the effect of each added dollar. In the Senate, about one in five members of the middle tiers broke ranks at some point; none of the 40 top-tier member-sessions did.
The test also ran from money to vote: does a member’s earmark money predict his or her vote on the institution? The result was null. Whether counted by sponsor or by money naming towns in the district (first measure), a member’s earmarks are unrelated to how he or she votes on the institution (odds ratio 0.88 per $100,000 sponsored, interval 0.76–1.02). That test does not ask the reverse question, whether a member’s votes go with the money that later reaches the district.
Fewer promotions, not lost posts
House Democrats who cast a reform-side vote were not stripped of posts at a higher rate. What changed was promotion. Of House Democrats who began a session with no stipend, 62% of those who never cast a reform-side vote were drawing one by the session’s second year, 103 of 166. Of those who did, 14% were, 5 of 36.
Measured in dollars, reform voters’ stipends rose about $10,500 less than their colleagues’ over a session, in 2025 dollars. The gap holds when the starting stipend, seniority and each member’s dissent rate on ordinary bills are held constant, and when any one General Court is dropped.
In the House, district money goes with the post
The money-to-vote test was null. The reverse test asks whether the earmark money reaching a member’s district goes with the member’s post and votes. Each member-session from 2017 to 2026 is matched to the enacted earmarks naming towns in the district (first measure) in that session’s budgets, FY2018–FY2026. Among House Democrats that is 635 member-sessions, 1,143 district-years and $399 million.
- Stipend. Holding district population, seniority, election margin and the session constant, district money per resident is 9.5% higher for each $10,000 of the member’s stipend going into the session (p = 0.002). Unadjusted, the median district received $5.34 per resident a year when its member held no post and $15.33 in the 13 sessions of top leadership.
- Reform votes. On the same basis, districts of members who cast a reform-side vote received 19% less per resident (p = 0.015). Their 119 district-years drew $22.7 million, $4.35 per resident a year, against $8.34 for the other 1,024 ($376 million); at the adjusted rate the gap is about $6.6 million. It is 19% to 30% less in every cut tried: without Suffolk County, without the Speaker and the Ways and Means chair, in each half of the decade and with region fixed effects (p of 0.015 or less throughout). In the same cuts the stipend estimate runs from 6% to 11% per $10,000 (p of 0.033 or less).
- Promotion. When a member moved up a stipend tier during the session, the district’s money rose 12% against the prior session (p = 0.007; 224 promotions in 544 member-sessions).
- Senate. Nothing: stipend p = 0.67, reform vote p = 0.62, promotion p = 0.90. Every senator holds a paid post.
On the wider measures, which add organizations and state facilities placed in the district, the pattern runs the same way and is weaker: the stipend estimate is about 10% (p = 0.06 and 0.05), and the reform-vote gap, 14% and 16%, is not significant until Suffolk County is set aside (19% and 20%, p of 0.035 or less). The reform-vote result rests on the same 42 members as the promotion result, and only two of their reform-side sessions fall in 2023–2026. On that sample it would seem to indicate a link between reform votes and district money; it does not establish one. Nor can it show which way any influence runs: leaders may favor loyal members’ districts, or members whose districts already draw more may be likelier to stay loyal and to be promoted. It is consistent with section 6, where a House member’s documented break with leadership coincides with less earmark money for the town in later budgets.
What this cannot show
The causal direction cannot be separated. Leaders choose lieutenants who share their views of how the institution should run, and a member who wants the Speaker’s term limited is unlikely to be made his deputy. Rules votes are also party votes, and a stipend predicts loyalty on ordinary contested bills too. The institutional signal is several times stronger in the House and only modestly stronger in the Senate. The House result rests on 42 Democrats who ever cast a divided reform vote, most in 2019–2021; in each of the last two sessions, one did. On a sample that thin, the House pattern would seem to indicate a link between reform votes and slower promotion; it does not establish one. Nothing here shows that anyone traded a vote for a post. It shows that the members who vote to limit the institution are, with few exceptions, the members it does not promote.
The pool and the reforms
The reward pool is small. In FY2026 stipends came to $5.4 million. Enacted local earmarks came to $76.3 million naming the town, $106.9 million placed in the town and $158.4 million counting everything placed. With stipends the pool is $81.7 million, $112.3 million or $163.8 million: 0.14%, 0.19% or 0.27% of a $60.2 billion budget.
Three gaps the state has documented have a price for a single fiscal year:
| Documented gap | Year | Statutory or recommended level | Paid | Gap for the year |
|---|---|---|---|---|
| Rural School Aid at the Special Commission’s 2022 recommendation | FY2026 | $60.0M | $16.0M | $44.0M |
| Regional school transportation at the statutory 100% | FY2025 | $117.2M | $99.5M | $17.8M |
| TAFDC grant increase for families with children (Senate amendment 522), six months from January 2026 | FY2026 | $20.2M | $0 | $20.2M |
| Total | $82.0M |
Sources: DESE Rural School Aid allocations (FY2026 includes a $4M supplemental) and transportation reimbursements; FY2026-S3-522 amendment text (a six-month cost; a full year would be about twice that); the FY2026 pool, earmarks plus 2025 CTHRU stipends, is $81.7M naming the town, $112.3M placed in the town and $163.8M counting everything placed (Parcenomics analysis of the FY2026 General Appropriations Act, malegislature.gov, and CTHRU payroll, cthru.data.socrata.com).
Together they come to $82.0 million: 100% of the pool counting earmarks that name the town, 73% counting earmarks placed in the town, and 50% counting everything placed. The cash-assistance line counts only the TAFDC portion of the Senate’s FY2026 amendment 522, filed by Assistant Majority Leader Sal DiDomenico, which would have raised grants for families with children by $20.2 million for the six months from January 2026. The full amendment, including EAEDC grants for elderly and disabled adults, was $30.4 million. It had 20 signers, half of the 40 senators, and was rejected in a bundle without a recorded vote.
The comparison has limits. Earmarks pay for real things, a food pantry or a fire truck, and those programs have defenders. Moving the money would take a vote. Counting everything placed, the pool is about twice the cost of the three gaps; counting earmarks placed in the town, 1.4 times; counting earmarks that name the town, about the same.
If power barely moves the money statewide, what do the rewards buy? On this evidence, not the division of the budget as a whole. They line up with the institution itself: who rises and who waits, and in the House, how much earmark money reaches the district.