VOTEM RESEARCH · AUGUST 2026 How public pension funds actually elect their trustees We went looking for the election procedure of every public pension fund in the United States that seats trustees by member vote. For nearly half of them, there isn't one you can find. A desk survey of 210 retirement systems, of which 153 run a member ballot. Compiled from public sources between 12 and 27 August 2026.
WHAT WAS COUNTED
210 systems examined 153 run a member ballot 31 seat every trustee by appointment A further 26 entries turned out not to be separate systems at all — they had merged into another fund or never existed under the name we started with. Everything below describes the 153 funds that put a ballot in front of their members. No fund is named next to a figure that reflects badly on it.
FINDING 01
Nearly half of these funds do not publish how their members vote For 73 of the 153 — 48% — we could not establish the ballot channel from anything the fund makes public. Not from the election notice, not from board policy, not from the minutes, not from the annual report. In most cases the fund publishes that an election is happening and who won, and nothing in between. Method established (80) Not publicly established (73) One cell per fund. 153 funds that run a member ballot. This is worth sitting with, because it is not a research failure — it is the finding. A member of one of these funds cannot find out how to vote in their own trustee election without calling the office.
FINDING 02
Where the method is known, electronic voting is already the largest group Among the 80 funds whose ballot channel we could establish, mail-only is no longer the norm. Some form of electronic voting — online, or online plus telephone, almost always alongside a paper option — now edges ahead of paper-only. Electronic in some form (online, or online + phone) 35 · 44% Mail or paper only 31 · 39% Delegate or convention vote 8 · 10% In-person polling place 6 · 8% The tail matters too: eight funds elect trustees at a delegate convention or annual meeting rather than by member ballot, and six use in-person polling places. Neither is a mailing problem, and neither is solved by moving a mail ballot online. One fund publishes its channel split. In a 2026 statewide election offering internet, paper and telephone, 56% of votes came in over the internet, 40% on paper and 3% by phone — on the first cycle where all three were offered.
FINDING 03
The legal barrier is much smaller than administrators assume The most common reason we heard for staying on paper is that the law requires it. We checked. Thirteen of the 153 funds have their ballot arrangements fixed somewhere above board level — but those thirteen are two different problems wearing one label. WHAT THE LAW ACTUALLY DOES FUNDS Prescribes paper. A statute or city code specifies the mailed ballot and return envelope. Changing the channel needs a legislature or a council. 5 Removes the ballot entirely. The seats are allocated to named associations or filled at a delegate convention. There is no member vote to move online in the first place. 6 Governing code could not be read from public sources. 2 Set those aside and look at the funds where we could establish who controls the ballot method. In 65 of 78 cases — 83% — the board can change it on its own authority or by amending its own election policy. No bill, no ordinance, no ballot measure. Board's own decision (65) Needs lawmakers (13) Of the 78 funds where the controlling instrument could be identified. For the remaining 75 it could not. One statutory case is unusually leveraged: a single provision of the West Virginia code sets the ballot method for all 53 municipal police and fire plans in the state at once.
FINDING 04
Almost none of them publish the result Only 18 of 153 funds — 12% — publish a vote count anywhere we could find it. Certification minutes routinely record that an election was held and a trustee was seated without recording how many people voted. Publishes a vote count (18) Does not (135) Three funds publish both ballots cast and voters eligible, which is the only pair that yields a real turnout figure. Those three came in at 4.2%, 4.4% and 16.1%. Three funds is not a sample and we are not going to pretend otherwise — but nothing we found anywhere in the survey suggested the picture is better where it goes unpublished.
FINDING 05
One fund in eight seated a trustee without holding a vote at all 20 of the 153 show at least one seat filled with no member ballot — a single candidate nominated, a seat carried by acclamation, or a rule that cancels the election outright when only one name is filed. At many funds this is automatic and written into the governing document: one candidate, no election, declared elected at the next board meeting. A seat filled with no ballot (20) No such case found (133) The scale this reaches is easy to underestimate. One statewide system with more than a million members had all three of its 2026 seats go uncontested — not a single member cast a vote that year. At another, one seat drew no qualified candidate at all. This is the quiet failure mode, and it is not a failure of election administration. A fund can run a clean, well- managed ballot for a decade and still seat trustees without a vote, because nobody knew the seat was open.
FINDING 06
The boards themselves are more alike than you would expect Across 143 funds where the seat counts are published, the shape barely moves: a board of nine, four of whose seats members elect, on three- or four-year staggered terms. MEDIAN Board size 9 seats Seats elected by members 4 Member-elected share of the board 44% Members elect an outright majority of the board at 41 of those 143 funds. On term length, of the 111 funds that publish one, 90 are on a three- or four-year cycle — which means a fund that runs its election badly gets to think about it once, then waits three years to try again.
IF YOU RUN ONE OF THESE FUNDS
Five things you can check without talking to anyone
Every one of these is answerable from documents you already hold, plus one federal rule. None of them requires contacting a vendor, and the fourth is one where the honest answer may count against buying an online ballot at all. What actually governs how your members vote? Find the instrument — a statute, a city ordinance, or your board's own election policy — and read what it says about the ballot channel. The answer sets your lead time for any change: a board policy is an agenda item, a procurement is a cycle, a statute is a legislative session, and not every legislature meets every year. Of the 78 funds here where we could identify the instrument, 65 could change the method by board decision alone. The reason to check before you need to: nine funds in this survey are currently locked into a channel by a live contract, and one statewide system went out to bid for roughly 139,000 mail ballots with no online or telephone option in scope — fixing the method for the length of the award. Knowing which instrument binds you is what stops your own next solicitation from deciding it for you. Finding 03. 1 Could a member find out how to vote — or how to run — without phoning the office? Open your own website and try it the way a member would. We ran exactly this check on 153 funds and 73 failed it. The cost is not only turnout. Twenty funds here had a seat filled with no ballot at all because only one member put their name forward, and a nomination packet nobody can find is one of the ways that happens. An uncontested seat means a trustee is seated without the member vote your governing documents intended — which is a governance outcome, not an administrative one, and it lands on the board rather than on you. Findings 01 and 05. 2 Does your certification record both ballots cast and members eligible? Pull the minutes from your last election. Those records are what you fall back on when a result is questioned, and the margins here are thin enough to make that a live risk: one county fund seated a trustee on a 15-vote margin out of 167 ballots cast. The count also drives procedure — where your rules require a majority to win, it is the count that determines whether a runoff is triggered, and several funds in this survey ran one. Only three funds here publish both numbers. Recording them is a governance question before it is a publication question — you do not have to publish a figure to want to know it. Without a baseline you also cannot tell whether anything you change later actually worked, including anything you buy from us. Findings 04 and 02. 3
Will your ballot meet WCAG 2.1 Level AA in time? The Justice Department's ADA Title II rule requires state and local government web content to meet WCAG 2.1 Level AA. In April 2026 the deadlines were pushed back a year: 26 April 2027 for entities serving 50,000 or more, and 26 April 2028 for smaller entities and for all special district governments — which is how many pension systems are classified. The standard did not change, only the dates. Worth saying plainly: an online ballot is web content too. This obligation applies to a vendor's voting portal exactly as it applies to your own site, and it is a fair question to put to anyone selling you one, us included. 4 What does a second round cost you? Runoffs are where budget and timeline actually move, and they are routinely left out of a cost estimate. One fund in this survey mailed roughly 18,500 ballots twice in a single cycle because the first round produced no majority. That is one observed case rather than a pattern we can size — but it is worth knowing your own answer before it happens. 5
METHOD AND LIMITS
Desk research only, from publicly available sources: fund websites, election notices and procedures, board minutes and agendas, annual comprehensive financial reports, state statutes and municipal codes. Nobody was interviewed and no fund was contacted. Fieldwork ran 12–27 August 2026. Election arrangements change; anything here can be out of date by the next cycle. "Not established" always means we could not find it, never that the fund does not have one. Several funds certainly publish material we could not reach — documents behind member logins, and codes on hosts that refused automated retrieval. Percentages are stated against the base they were measured on, which is often smaller than 153. Where the base is small we have said so rather than scaling up. The survey was compiled by Votem, which sells online voting to these funds. Read Finding 02 and Finding 03 with that in mind. We have published the denominators so you can discount them yourself. No fund is identified alongside a turnout figure, a vote count or any other number that could be read as criticism of its staff. The accessibility deadlines in the closing section are the only figures here that do not come from our own fieldwork. They are the Justice Department's ADA Title II web rule (28 CFR Part 35, 2024) as amended by the April 2026 interim final rule extending both compliance dates by one year. Check them against the current text before you rely on them for planning. Questions, corrections, or a fund we have wrong — we would rather fix it than publish it wrong. Votem · August 2026