← Newsroom

What does union political money actually buy?

Unions have spent members’ money on politics for as long as there have been unions, on a simple theory: spend, and you get ballot measures that go your way, allies in office, and laws. We followed that money through all five links of the chain — $175.4M of ballot-measure spending across 12 states, 5,361 candidates, 55 legislator-by-legislator comparisons. Two links hold. Two break. One we could not settle at all.

Five claims, tested one at a time

The one place the money plainly works

A ballot measure is the cleanest test available. There is no incumbency, no party label, no candidate biography — just two sides, a disclosed budget each, and a certified result. We swept twelve states’ campaign-finance records from 2014 onward for committees campaigning on a named measure. 54 of them took union money, $175.4M of it. Forty-nine have been decided.

“Won” and “lost” describe the union’s side, not the measure. On four of these six, the union was campaigning against a measure, so a rejected measure is a win. Share of the campaign is union money as a proportion of everything that committee raised. Each measure links to its full background; results can also be looked up at California’s Secretary of State and Washington’s results archives.

Two entries deserve reading twice. Proposition 22 is the sharpest case in the project. Labor paid for 94.8% of the No campaign — roughly $18M, essentially the entire opposition — and was outspent about ten to one by the app companies. It passed with 59%. And Proposition 15 is the largest single union bet anywhere in the data: $35.8M, two-thirds of its campaign, and it failed by four points.

So the win rate is real and the dollar-weighted picture is more sober. Where unions are one funder among many in a modest campaign, their side usually prevails. On the handful of nine-figure fights that decide the most, being the whole opposition has not been enough.

Fund a legislator, and pro-labor bills appear

The next link is legislative. Take every state legislator a union gave money to, compare them with legislators of the same party in the same chamber who got none, and count the pro-labor bills each one put their own name on as primary sponsor.

The obvious objection is that unions simply fund prolific legislators — people who introduce more of everything. That is testable, and it fails. The same funded legislators write only 1.44× more bills of all kinds against 4.17× more pro-labor ones, and their pro-labor output beats their own overall baseline in 10 of the 10 cases we tested. The effect is specific to labor, not a general busyness effect.

One methodological note that matters: we count primary sponsors only. Adding your name to someone else’s bill is cheap; being the person who files it is not.

Where the chain breaks

More bills written should mean more bills passed. It does not, to any degree we can find.

Look at it per comparison and it gets stranger. Funded legislators run +5.2 points ahead of their colleagues on the passage rate of all their other bills — and −8.8 points behind on their pro-labor ones. These are effective legislators, doing worse than average on exactly the bills the money is about.

That has a mundane explanation. A pro-labor bill in a chamber that does not want one is a hard bill to pass, whoever files it, and the extra bills a funded legislator files are disproportionately the ambitious ones. Money buys introduction. Introduction is not passage.

None of which makes the spending pointless. Filing a bill puts a position on the record, forces a hearing, builds the version that passes two sessions later. But if the theory is we fund champions and they deliver statutes, the delivery step is where we lose the trail.

On the floor, the money barely moves anyone

The last legislative link is the one most people assume is the whole story: does a legislator who took union money vote with labor more often than one who did not?

Comparing within party is the whole point. A Democrat and a Republican differ on labor bills by far more than two points, and union money goes overwhelmingly to Democrats. Compare across parties and you will measure the party and call it the money.

Two points is not nothing on a knife-edge vote. It is a long way from the transaction the phrase “union money buys votes” describes.

The claim we could not settle

Which leaves the first link in the chain, and the one we most expected to confirm: does union money get allies elected? We matched union contributions to 5,361 candidates in the six states where certified results can be joined to campaign finance by name, and looked at races decided by under ten points.

The six states do not give the same answer, and they are not close.

The spread is 42 points wide, and it runs backwards against sample size. Washington — the deepest close-race sample, the longest campaign-finance history, and the only one of the six that publishes its own certified export — shows nothing whatsoever. The three largest gaps come from the three thinnest samples.

The largest sustained bet on a single candidate anywhere in the file sits outside those six states. The Chicago Teachers Union gave $5,487,921 to Friends of Brandon Johnson between February 2019 and January 2025, on the Illinois State Board of Elections disclosure record; Johnson, a former CTU organizer, won the Chicago mayoralty in April 2023 with 51.4%.

They get the bills written. Passing them is the hard part.

For a union treasurer: the two links that hold are the ballot box for measures and the sponsorship line for bills. The link that does not is passage — which is an argument for spending differently rather than spending less. The marginal dollar that produces one more introduced bill appears to buy less than the same dollar spent on whatever gets an already-introduced bill to the floor.

For anyone who assumes union money buys votes: the within-party gap on final passage is about two points, and it is the most durable number in this project. Legislators who take union money were already voting with labor. That is largely why they took it.

For everyone else: five links, tested separately, is the only way this question gets an honest answer. Treated as a single claim, “union political spending works” is both true and false depending on which link you had in mind.

How this was built

Whose money

Form LM-2 filings from the Labor Department’s Office of Labor-Management Standards, which every union above $250,000 in annual receipts must file. The fiscal-2025 file we worked from holds 4,599 LM-2s. Locals are ranked on the dues they spent on political activity, not on their PAC. The two are legally separate pots — federal law bars dues money from funding federal candidates — and treating them as interchangeable is the single easiest way to get this subject wrong.

Where it went

Campaign-finance filings from twelve states — Washington, Illinois, California, Arizona, New York, Massachusetts, Oregon, Maryland, Michigan, Ohio, Minnesota and Pennsylvania — plus the Federal Election Commission. Committees are matched to unions from the committee’s own registered name rather than the FEC’s affiliated-organization field, which is unreliable enough to label a national fund after a single local branch.

What the legislators did

Bills, primary sponsorships and recorded floor votes from LegiScan. 1,969 bills were classified by direction — 575 pro-labor, 54 anti-labor, 1,340 neutral. That direction is our judgement, from keyword rules plus a model reading the bill, and it is not a record of any union’s stated position on any bill.

Who won

Candidate outcomes come from certified returns, never from our own copies of election data: Washington’s own certified export for that state, and OpenElections for Arizona, California, Oregon, Michigan and New York. Ballot-measure outcomes were looked up one measure at a time, each with its own source recorded against it. Nothing about a result is derived.

The comparison

Every result compares union-funded legislators with unfunded ones of the same party in the same chamber. One comparison is one union in one state, split by party; there are 55 of them. A candidate who stood in more than one of these elections is counted once for each. Where a state’s campaign-finance record begins later than its election results, the earlier years are dropped rather than scored — otherwise every candidate in those years reads as unfunded.

What this cannot tell you

Stated up front, because a finding is only as good as the boundary around it.

Association, not cause

Every result here is an association. Nothing in this data shows that spending caused any outcome.

This one runs backwards

Of all five links, bill-writing is the one where the arrow probably points the other way. Writing a pro-labor bill is the visible act that earns the money — a union reading a legislature does not need to buy a champion when it can identify one. Read that finding as union money and union-friendly lawmaking travel together, never as money produces champions.

What a win rate can’t tell you

A union chooses which fights to enter. Two in three is partly a statement about the measures they picked, not only about the money they spent. Unions also turn up on both sides of the same question: they funded the yes and the no campaigns on California’s Proposition 30 in 2022 and Proposition 36 in 2024, and on the two rival death-penalty measures on the same 2016 ballot — in that case police associations against everyone else.

Twelve states, not fifty

And not a random twelve — they are the states whose campaign-finance records we hold. Six of them can be joined to certified candidate results. Nothing here describes the other thirty-eight.

Most of the money is off this map

Across these six states, only 6% of union political giving — $126M of $2.04B — reaches a state legislative candidate at all. The rest goes to governors, mayors, city councils, judges, ballot committees and other unions’ own funds, none of which we can score against a legislative record. Part of that $2.04B is also transfers between union committees rather than giving. Everything in this article describes the sliver of union political money that is scoreable, not the whole of it.

Direction is our call

Whether a bill is pro-labor, anti-labor or neutral is a keyword-and-model classification made by us. A union that considered a given bill unhelpful will not have been consulted, and no union’s stated position enters the calculation.

The thinnest result here

Only three of the union-and-state comparisons clear our evidence bar on passage — operating engineers in Washington, service employees in New York, electrical workers in Maryland, all on the Democratic side. Three is not enough to call this settled in either direction. It is enough to say that the large, clean effect on bill-writing has no counterpart here.

No spending sweet spot

Sorting measures by how much a union spent appears to reveal a best band. It dissolves under checking — what it is actually measuring is campaign size and the state of California. We report no optimal amount because we could not find one that survived.

---