The evidence — a neutral, power-limited null (Dell DE→TX event study)

A neutral, power-limited null.

On the clean disclosure day, Dell’s fully-adjusted (market + sector + AI) abnormal return is , statistically indistinguishable from zero (p_BH = ). The only large cumulative returns belong to the AI-server rally, which the AI-controlled model removes.

What this is — and what it is not. This is a power-limited null, not a bounded null: the two one-sided equivalence test at the ±2 pp band returns p ≈ , and the minimum detectable effect is . A plausible 1–3% governance re-pricing is below detectability. The null is neutral — equally consistent with an efficient market pricing a governance-neutral move and with a foregone-conclusion vote. The market study refines the thesis; the arithmetic identity carries it.

Headline scoreboard

Every benchmark returns a non-significant disclosure-day reaction.

Six benchmark models on the disclosure-day window (, primary anchor T0 = ). All cells non-significant (p_BH = ). The TOST tile does not affirm equivalence.

Mean-adjusted

p_BH = · null

Market (S&P 500)

p_BH = · null

Market + sector (XLK)

p_BH = · null

Market + sector + AI

p_BH = · fully adjusted

Donor-weighted portfolio

p_BH = · null

TOST equivalence (±2 pp)

p ≈

does NOT affirm equivalence

What this shows. On the day Dell’s redomestication crossed EDGAR, the stock moved indistinguishably from zero under every benchmark. The fully-adjusted figure is . The admissible claim is “no detectable effect, MDE .”

The AI rally, not the charter

The large numbers belong to the AI-server cycle.

The big positive cumulative returns that surface under one alternative clock are Dell’s own AI-server returns — on May 6 and on May 8 — not a charter effect. They survive only when the window swallows those days and the model omits an AI control.

Under one clock, a big number appears…

Moving T0 to and using the donor-weighted model gives a May-5 CAR of (t ).
That window pulls in the May 6 () and May 8 () AI-server returns.

…and the AI control removes it.

Adding an SMCI/HPE AI factor dissolves the May-5 spike to (t , non-significant).
On the clean May-4 day the fully-adjusted return is already — there is no charter signal to remove.

The eight stress tests

Each test attacks a different way the null could have been Dell-specific.

If the clean-day null were hiding a real, Dell-specific signal, at least one of these would isolate Dell. None do. The full interactive battery — with Plain-English / worked-example / Academic methodology toggles for every test — is on the Stress tests page.

#	Test	Result
1	Event-date sensitivity (both anchors)	May-4 fully-adjusted ; May-5 spike is AI-rally driven
2	Window sensitivity	Wide windows confound-flagged (overlap AI rally + late-May earnings); never headlined
3	Donor leave-one-out	No verdict flips across all donors, both anchors
4	Placebo-in-time	pseudo-dates; May-5 AI-controlled p =
5	Placebo-in-space	Honest limit: min attainable permutation p = with donors
6	AI-loading cross-section	Dell sits on the line; screened R² = (preliminary)
7	GARCH(1,1)	Day-0 AI-controlled z = (p = ); corroborative
8	Confounder log & power	MDE ; May 6/8 are Dell’s own AI returns

Open the full stress-test battery →

The comparator

ExxonMobil’s vote, for contrast.

ExxonMobil’s NJ→TX Proposal 4 was approved on at — not of outstanding (Form 8-K Item 5.07, acc. ). ExxonMobil faced organized proxy-adviser opposition and adopted zero elective provisions; Dell, with a controller bloc, faces a vote whose outcome is an arithmetic identity. Two boards, two postures, one statute.