The Forced Sale
The rule reformers want Congress to pass already exists one branch over. Executive officials sign ethics agreements that can compel them to sell, on a clock they do not control. It works. It is also attached to the one disclosure form in Washington that cannot tell you when it applied.
The campaign to stop stock trading in government is aimed at Congress. The rule it asks for already exists one branch over. Executive-branch officials sign ethics agreements that can compel them to sell named holdings on a fixed clock, and the compulsion is visible in the record. GovGreed parsed 5,675 transactions from 323 OGE Form 278-T periodic transaction reports filed by 81 people between January 2020 and May 2026, across three administrations. Frank Bisignano ran Fiserv. The Senate confirmed him to run the Social Security Administration on May 6, 2025; an ethics agreement gave him 120 days to sell every share. He finished in 56. Our parse of those 22 sales brackets to $244.0M–$617.0M, and the $558M reported in contemporaneous press coverage sits inside that range. On October 29, 2025 Fiserv fell 44.0 percent in one day — roughly $30 billion — after his successor withdrew earnings forecasts Bisignano himself had issued. The forced sale spared him approximately $300 million. He could not have timed it: the clock started on the Senate’s vote, and the 2025 high of $237.79 on March 3 passed while he was still awaiting confirmation, leaving his first sale 29.9 percent below it.
The windfall is not the finding. The finding is that the form cannot tell you when the rule applied. Bisignano was legally compelled to sell his own company. Jared Isaacman’s ethics agreement does not require him to divest Shift4, the company he founded, and he has stated he intends to remain its largest shareholder — yet his filing also shows a $50M sale of it, arising from a share-class conversion. On a 278-T the two are indistinguishable, because the document that would separate them is a different PDF that nothing links to the transaction. Add brackets instead of amounts, an unbounded top bracket, and a ticker on only 34.9 percent of rows. Meanwhile, same stock and same year, thirteen members of Congress traded Fiserv and seven bought after January 2025, riding it from $237.79 to $63.80. That we can tell you to the day — because Congress publishes a database. The branch with the strict rule is the branch that cannot be audited.
- Key findings
- An ethics agreement gave Bisignano 120 days from his May 6, 2025 confirmation to sell his Fiserv holdings. He finished in 56 — first sale May 16, last July 1, sixty-four days inside the deadline.
- He could not have timed it. The clock started on the Senate’s vote, not a decision of his. Fiserv’s 2025 high — $237.79 on March 3 — passed while he awaited confirmation. His first sale, at $166.66, is 29.9% below it.
- On 2025-10-29 Fiserv fell 44.0% in a single day after his successor withdrew forecasts Bisignano had issued. The forced sale spared him ~$300M. Two House members have referred him to the SEC; two senators opened a probe. This paper does not answer that question.
- The form cannot tell a compelled exit from a voluntary one. Bisignano was required to sell Fiserv. Isaacman was not required to divest Shift4 and intends to remain its largest shareholder — but both appear on a 278-T as a sale of the filer’s own company.
- Only 34.9% of the 5,675 transaction rows carry a ticker. Municipal bonds are just 7.1%; 32.7% name a real public company in free text that never got mapped — inconsistently, within a single filer.
- Same stock, same year: 13 members of Congress traded Fiserv, 7 bought after January 2025, riding $237.79 → $63.80 (−73%). No divestiture rule applies to any of them, because Congress has never passed one for itself.
1. The rule Congress won’t pass already exists next door
Every argument about stock trading in government is an argument about Congress. It has the famous statistics, the famous names, and the reform bill that never moves. The proposal at the centre of it — that officials should be made to sell the holdings that conflict with their power — is treated as an aspiration.
It is not an aspiration. It is current practice, one branch over. An incoming executive-branch official negotiates an ethics agreement with the Office of Government Ethics and their agency’s ethics officials before confirmation, and that agreement can require them to sell named assets within a fixed number of days of taking office. Not a blind trust. Not a recusal promise. A sale, on a deadline.
Congress has never passed anything comparable for itself. A member of Congress may hold whatever they like, for as long as they like, on any committee.
So the interesting question is not whether the executive branch is worse. It is what the stricter rule actually produces — and whether, from the public record, you can see it produce anything at all.
2. Data and methodology
When an executive-branch official buys or sells a reportable asset, they file an OGE Form 278-T, a periodic transaction report. It is the executive twin of the Periodic Transaction Report a member of Congress files under the STOCK Act, and the resemblance ends at the concept.
We parsed 5,675 transaction rows from 323 such filings by 86 filer names, which normalise to 81 people — the same official appears under more than one spelling, so every figure here aggregates on a normalised name rather than the raw string. Coverage runs 2020-01-03 to 2026-05-22 and spans three administrations: the first Trump term, the Biden administration, and the current one. That is not incidental. It is what keeps this paper from being about one cabinet.
Values are brackets, not amounts, so every dollar figure below is reported as a floor and a ceiling rather than a point estimate. The top bracket, Over $50,000,000, has a floor and no ceiling — a filer who sold $400M and one who sold $60M produce the same row.
One methodological choice needs stating up front because it inverts a natural reading. A buy/sell count on this data does not measure conduct. Bisignano’s 127 transactions are 89 purchases against 38 sales, which looks like accumulation and is the exact opposite: the sales are Fiserv, and the purchases are New Jersey municipal bonds, a Vanguard tax-exempt bond index ETF, PIMCO Income, and Nuveen, Invesco and Franklin municipal funds. The purchases are the divestiture’s other leg. A conflict regime pushes officials out of single names and into diversified funds, so the buy count measures compliance rather than appetite. Counting rows would have produced the wrong story about the single most important filer in the dataset.
Only 34.9% of the 5,675 transaction rows carry a ticker. The obvious explanation — that the assets are municipal bonds and funds, which have no equity ticker — is only partly true. Funds and ETFs are 25.3%; municipal bonds are just 7.1%. The remaining 32.7% (1,856 rows) name a real public company in free text that our parser did not map.
It is not uniform, and that is worse than if it were. Chris Wright’s Liberty Energy sales are 41 rows: 28 carry the ticker LBRT and 13 give the same asset name with no ticker at all — same person, same asset, same transaction type. Any ticker-keyed view of this dataset undercounts silently and non-uniformly, which is why the analysis below reads the asset names rather than the ticker column. The gap hides exactly the rows that matter most: Fiserv (20), Shift4 Payments (20), Codiak Biosciences (47) — each filer’s own former company.
3. The case: a confirmation vote, and a 120-day clock
Frank Bisignano was the chief executive of Fiserv, the payments company behind a large share of American card processing. In December 2024 he was named as the incoming administration’s choice to run the Social Security Administration. The Senate confirmed him on May 6, 2025. Under his ethics agreement he had 120 days from that date to sell his Fiserv holdings — a deadline of roughly September 3.
He did not use it. Table 1 is the sequence as the filings and the tape record it.
Table 1 · The clock, the sales, and the collapse| Date | Event | Fiserv close | vs. 2025 high |
|---|---|---|---|
| 2025-03-03 | Fiserv’s 2025 high. Bisignano is awaiting confirmation. | $237.79 | — |
| 2025-04-24 | First earnings drop. | $176.90 | −25.6% |
| 2025-05-06 | Senate confirms him. The 120-day clock starts here. | $184.95 | −22.2% |
| 2025-05-16 | First Fiserv sale — ten days after the vote. | $166.66 | −29.9% |
| 2025-07-01 | Last Fiserv sale. 22 sales, done in 56 days. | $172.68 | −27.4% |
| 2025-09-03 | The 120-day deadline. He finished 64 days early. | — | — |
| 2025-10-29 | Fiserv falls 44.0% in one day (~$30B) as his successor withdraws the forecasts Bisignano issued. | $70.60 | −70.3% |
| 2026-07-10 | Latest close. | $63.80 | −73.2% |
Prices are daily closes. Sale dates and counts are from Bisignano’s OGE Form 278-T filings; the 22 Fiserv sales bracket to a floor of $244.0M and a ceiling of $617.0M. The confirmation date, the 120-day requirement, the ~$558M reported sale total and the October forecast withdrawal are from contemporaneous press reporting. Verified July 15, 2026.
Our parse of the 22 sales gives a range of $244.0M to $617.0M — the brackets are that wide. Press reporting puts the total he and his wife sold at approximately $558 million. That figure sits inside our range, which is the strongest confirmation this form permits: not a match, a containment.
Then the reinvestment. Between June 18 and August 4 he filed 89 purchases, bracketing to $458.3M–$912.5M, into municipal bond funds and index ETFs. Sell the conflict, buy the exempt. That is what the regime is designed to produce, and it produced it.
In October, Fiserv’s new chief executive withdrew earnings forecasts that Bisignano had issued as CEO. The stock fell 44 percent in a single day. Because he had been made to sell four months earlier, at prices between roughly $159 and $173, the rule spared him something on the order of $300 million.
Two House members have since referred him to the SEC over the divestitures; two senators have opened a probe. The question they are asking is whether he knew the forecasts were unreachable while he was selling. This paper does not answer it, and does not have the evidence to. What it can establish is the shape of the thing: he did not choose the date, he did not sell at the top — the top was in March, while he was still a nominee — and he finished the sale sixty-four days before he had to.
4. Who had to, and who didn’t — and why the form can’t say
Read the 278-T corpus by asset name rather than ticker and a pattern jumps out immediately. Filer after filer, the single largest disclosed sale is the company they walked in from. It is tempting to call that the divestiture regime working, and that is the mistake this section exists to prevent.
Table 2 · The largest own-company sale per filer — and whether an agreement required it| Official | Role | Largest own-company sale | Floor | Required to divest it? |
|---|---|---|---|---|
| Frank Bisignano | Social Security Administration | Fiserv — 22 sales | ≥$244.0M | Yes — 120 days from confirmation |
| Chris Wright | Energy | Liberty Energy — 41 sales | ≥$13.9M | Yes — confirmed after agreeing to divest |
| Eric Lander (Biden) | Science & Technology Policy | Codiak Biosciences — 47 sales | ≥$6.2M | Yes — agreed to divest work-related holdings |
| Howard Lutnick | Commerce | BGC — 5 sales | ≥$151.1M | Sold — requirement not verified here |
| John Phelan | Navy | MSD Hospitality Partners LP — 4 sales | ≥$52.0M | Partial — divested Red Cell; retained Dell holdings |
| Jared Isaacman | NASA | Shift4 Payments — 1 sale | ≥$50.0M | No — not required; intends to remain largest shareholder |
Largest own-company sale position per filer, by bracket floor, from OGE Form 278-T filings; asset names read directly because the ticker column is unreliable (see section 2). The “required to divest” column is not derived from the filings — it comes from separately published ethics agreements and contemporaneous reporting, and is marked unverified where we could not confirm it. Verified July 15, 2026.
The last row is the paper. Jared Isaacman’s ethics agreement does not require him to divest Shift4, the company he founded. He has stated he intends to remain its largest shareholder. The $50M sale that appears on his filing arises from a share-class conversion — a restructuring of how he holds the company, not an exit from it.
Now put his row next to Bisignano’s and look only at what the form shows. Two officials. Two sales of the company each of them ran. Same document type, same brackets, same fields. One was under a legal obligation with a countdown; the other was under no obligation at all and is still the largest shareholder.
That is the audit gap, stated precisely. It is not that the executive branch hides its trades — these filings are public, and we parsed them. It is that the transaction record and the obligation record are two disconnected systems. The ethics agreement is where the answer lives. It is published separately, per person, as prose, with no identifier that ties it to a transaction line. To know whether any given sale was compelled, you must find a human being’s agreement, read it, and match it by hand.
We did that for six officials. There are eighty-one in this dataset.
5. The mirror: the same stock, the other branch
Fiserv in 2025 is a rare gift to this comparison, because both branches are in it at once.
While Bisignano was being compelled out of the stock between May and July, thirteen members of Congress were trading it freely. Seven of them bought after January 2025. The full congressional record on the ticker runs to 111 disclosed trades — 49 purchases and 62 sales — and it rides the entire slope from $237.79 down to today’s $63.80. That story is told in full in the companion paper on the congressional side; what matters here is the asymmetry.
No divestiture rule applied to any of the thirteen. None exists. And none of them is under investigation for their Fiserv trades.
This is not a complaint about the investigation. The question in front of the SEC is a legitimate one, and it exists because a $558M sale ahead of a 44 percent collapse is exactly the pattern a regulator should look at. The observation is narrower and stranger: the man who had no choice about selling is the one being examined, and the members who had every choice are not — because they lost money, and losses attract no scrutiny.
6. The measurement we could not make
The obvious next question is whether Bisignano is a fluke. If the divestiture regime systematically pushes officials out of stocks before they fall, that is a finding. If it systematically pushes them out before stocks rise, that is a different and equally interesting one — a hidden cost of public service.
We tried to measure it: for every 278-T sale with a resolvable ticker, the stock’s return from the sale date to a fixed as-of date, benchmarked against the S&P 500 over the same window. It ran over 1,421 sales by 53 people. We are not publishing the result, because the result is wrong, and the way it is wrong is worth more than the number would have been.
The output claimed a stock rose 3,527% in fourteen months, another 1,147%, another 920%. None of that happened. The cause was not in the disclosures — it was in our own price history. 438 tickers are missing more than four fifths of 2025, and the standard “nearest prior close” lookup has no staleness guard: when the series has a hole it silently returns a price from a different market era instead of returning nothing.
The tell was available before the absurd numbers were: the mean excess return was +27.4% and the median was −25.2%. Mean and median disagreeing in sign is not a skew to interpret. It is an alarm. A handful of rows with a stale baseline were dragging the average, and reporting either figure — “divestiture costs officials money” or “divestiture protects them” — would have been a fabrication with a citation attached.
The question stands. The answer requires a price backfill we have not done, and a ticker resolver that does not exist. Until then, Bisignano is a case, not a class, and this paper says so rather than implying otherwise.
7. Limitations and caveats
Brackets, not amounts. Every dollar figure is a floor and a ceiling. The top bracket has no ceiling at all, so eight rows in this corpus — all Lutnick’s, floor $351.0M — are lower bounds with nothing above them. A naive sum over the ceiling column returns a total below the floor, because SQL skips nulls rather than propagating them. Every figure here uses the floor where no ceiling exists, and is rendered as “at least.”
The ticker column is unreliable and, as section 2 shows, unreliable non-uniformly. This paper reads asset names.
Coverage is self-selected. A 278-T exists only for officials who transacted after taking office. Someone who arrived already divested files nothing and is invisible here. This is a census of transactions, not of people.
The President is in this dataset and is not comparable to the rest of it. His 278-T shows 702 transactions, 700 of them purchases. It would be easy, and wrong, to contrast that with the 70 percent sell rate of the other eighty officials and call it a natural experiment in what conflict rules do. We know his 278-T captures only a fraction of his activity, because we have also parsed his annual filing and it holds far more transactions; the other officials’ annuals are not parsed. The gap may be coverage rather than conduct, and it is not measurable until they are.
Twenty-five transaction rows carry no value at all — both bracket bounds null. They are excluded from dollar figures and included in counts.
The “required to divest” column in Table 2 is hand-verified, six deep, out of eighty-one. That is the honest scope of the compliance claim in this paper, and it is small on purpose. It is also the whole point: establishing it for six people took manual reading, because no machine-readable link exists.
8. Conclusion
The reform Congress will not pass for itself is already law next door, and it bites. It made a man sell $558 million of the company he had run, on a clock that started with a Senate vote, and it finished the job sixty-four days early. Whatever else is true, that is not a paper rule.
But a rule you cannot verify is a rule on trust. The executive branch publishes brackets instead of amounts, leaves the top one open-ended, names assets in free text, and keeps the only document that says whether the rule applied in a separate file with no thread back to the transaction. So a compelled sale and a voluntary reshuffle arrive in the same shape, and telling them apart is manual work, one human at a time.
Congress gets accused of insider trading constantly. That is not because Congress is uniquely compromised. It is because Congress publishes a deadline, a penalty, a database and an API — enough machinery to be checked, and therefore enough to be doubted. The branch with the real conflict rule is the branch nobody can check.
The fix on the congressional side is a rule. The fix on the executive side is a foreign key.
Data availability
Primary sources. Transaction dates, types, asset names, owners and brackets come from OGE Form 278-T periodic transaction reports filed with the Office of Government Ethics, parsed into structured rows: 5,675 transactions from 323 filings by 86 filer names normalising to 81 people, 2020-01-03 to 2026-05-22. Fiserv prices are daily closes. The congressional side comes from STOCK Act Periodic Transaction Reports, deduplicated and joined to the sitting-member roster.
Not from the filings. The confirmation date, the 120-day divestiture requirement, the ~$558M reported sale total, the October 2025 forecast withdrawal, the SEC referral, the Senate probe, and every entry in the “required to divest” column of Table 2 come from separately published ethics agreements and contemporaneous press reporting. They are external claims and are labelled as such throughout. A machine-readable export of the 278-T parse is available on request — see “Sourcing this for a story?” below.
Reproducibility & verification
This is an independent working paper. Produced by GovGreed Research; not externally peer-reviewed. The corpus counts (5,675 / 323 / 86 → 81 / 2020-01-03–2026-05-22), the ticker-resolution breakdown (34.9% resolved, 25.3% funds, 7.1% municipal, 32.7% unmapped free text), Bisignano’s 22 Fiserv sales and their $244.0M–$617.0M bracket range, his 89 reinvestment purchases and their $458.3M–$912.5M range, Wright’s 41 Liberty Energy rows split 28/13 on the ticker column, the Fiserv price series, and the congressional counts (13 members, 111 deduplicated trades, 7 buyers after January 2025) were derived live and verified as of July 15, 2026.
Pinned parameters. As-of date 2026-07-10 for the latest close. Congressional trades are bounded by filed date, not trade date — the STOCK Act’s filing lag means late disclosures keep revealing older trades, so a trade-date bound silently changes the past. Congressional member counts are joined to the sitting-member roster; an unjoined count includes departed members and inflates. Executive filers are aggregated on a normalised name, not the raw filer string.
External check. Our parse of Bisignano’s Fiserv sales returns a bracketed range of $244.0M–$617.0M. The ~$558M reported independently in press coverage of the same filings sits inside that range. This is a containment, not a match — which is the strongest confirmation a bracketed disclosure permits, and it is stated that way deliberately.
What we withheld. A cohort measurement of post-divestiture returns versus the S&P 500 was computed over 1,421 sales and is not published; section 6 explains why and shows the diagnostics that condemned it. No figure from that measurement appears anywhere in this paper.
Conflict of interest & funding
GovGreed is a commercial congressional-trading-intelligence platform; GovGreed Research is its analysis function. This paper received no external funding, and no official named here was given prior review. It uses only public federal records and public reporting, and is released free to read, quote, and reproduce under CC BY 4.0 with attribution. Nothing here alleges wrongdoing by anyone. Frank Bisignano complied with a legal obligation on a schedule set by the Senate’s confirmation calendar; this paper makes no claim that he possessed material non-public information, and expressly declines to answer the question now before the SEC. Jared Isaacman is under no obligation to divest Shift4 and is named here precisely because his filing is lawful and ordinary — he is the control case that demonstrates what the form cannot distinguish. The thirteen members of Congress who traded Fiserv broke no rule; no divestiture rule applies to them. The finding is about two disclosure regimes, not about any person’s conduct.
Revision history
v1.0 · 2026-07-15 — Initial publication. Corpus counts, ticker breakdown, Bisignano’s sale and reinvestment brackets, the Fiserv price series and the congressional counts derived live and verified as of July 15, 2026. Pre-publication correction, recorded because it changed the thesis: an earlier draft of Table 2 and of the companion X post presented the own-company sales as a roll-call of divestitures — “Lutnick sold BGC, Isaacman sold Shift4, Loeffler sold Intercontinental Exchange… a conflict rule working as written.” Checking each ethics agreement before publication rather than after showed that claim to be false for Isaacman, who is not required to divest Shift4 and intends to remain its largest shareholder; unverified for Loeffler, whose relationship to Intercontinental Exchange we could not confirm and who was therefore removed; and partial for Phelan, who divested Red Cell while retaining Dell holdings. The correction produced this paper’s actual argument: if a transaction row cannot distinguish a compelled exit from a voluntary restructuring, the disclosure is the defect, not the official. The rule now applied throughout: a row is not a reason — never assert a divestiture from transaction data alone.
Frequently asked
Sourcing this for a story?
Every figure derived from the filings is reproducible from public federal records, and the pinned parameters are stated in the back matter. Machine-readable exports of the 278-T parse, the per-filer own-company sales, and the congressional Fiserv record are available to journalists on request. Note that the “required to divest” column is hand-verified from separately published ethics agreements for six officials only — please do not extend it without doing the same. Methods across the series are documented at /research.
References & data sources
- OGE Form 278-T periodic transaction reports — Office of Government Ethics. 5,675 transactions from 323 filings, 86 filer names normalising to 81 people, 2020-01-03 to 2026-05-22, spanning three administrations.
- Ethics agreements — published separately per official by OGE and the relevant agency ethics office. The source for Table 2's compliance column; hand-read for six officials.
- STOCK Act Periodic Transaction Reports — the congressional Fiserv record: 13 members, 111 deduplicated trades, 49 purchases and 62 sales. The full account is in GGR-WP-2026-18's sibling coverage of congressional trading in falling stocks.
- Daily price series — Fiserv closes: $237.79 (2025-03-03 high), $184.95 (2025-05-06), $166.66 (2025-05-16), $172.68 (2025-07-01), $70.60 (2025-10-29, −44.0% in one session), $63.80 (2026-07-10).
- Contemporaneous press reporting — for the confirmation date, the 120-day divestiture requirement, the ~$558M reported sale total, the October 2025 forecast withdrawal and ~$30B single-day loss, the ~$300M avoided, the SEC referral and the Senate probe. These are external claims, not derived from our parse.
- Image credit — Social Security Administration headquarters, Woodlawn, Maryland — Coolcaesar, CC BY-SA 3.0 via Wikimedia Commons.
Keep pulling the thread
Every disclosure, every trade, every bill — cross-referenced.
Start Free — See Today's Top 10 SignalsNot financial advice. All data from public federal disclosures.