Three separate tests: companies that moved to Texas, companies already incorporated in Texas, and firms that affirmatively elected the §21.552 derivative-standing threshold.
There is no single "did Texas matter" question. We split it into three: the companies that moved their legal home to Texas, the companies that were already incorporated in Texas when the law changed, and the firms that affirmatively elected the 3% derivative-standing threshold of TBOC §21.552. Each group has its own companies, its own event, and its own test. The clean Group-3 read — the already-Texas firms that affirmatively elected the threshold — shows no statistically detectable announcement-day reaction under current power; the Group-1 mover and Group-2 incumbent-shock estimates remain provisional pending the full CRSP / Fama-French re-run and the queued cross-firm tests. A result statistically indistinguishable from zero is neutral, descriptive evidence — not a verdict on whether any of this is good or bad policy. The §21.552 election is detailed in Group 3 below.
Every count on this page is bound to one canonical object and hydrates from the live dataset; the denominator and any contested figure are flagged in the data-lock panel at the foot of the page. The 3% figure is the statutory ceiling under TBOC §21.552(a)(3) — a charter or bylaw may set a lower floor.
The 3% derivative threshold (§21.552) is only one item on a much larger menu. The interactive matrix below shows, firm by firm, all twelve Texas governance amendments each company elected, declined, or left to the statutory default — the §2.115 exclusive Texas forum, the §2.116 jury-trial waiver, the §7.001 officer exculpation, the §21.373 shareholder-proposal threshold, ch. 171 arbitration, the §21.419 business-judgment rule, and the statutory defaults. Click any column header for a plain-English explanation; click any cell to see the firm's status and its EDGAR filing.
There is no single "did Texas matter" question. There are three distinct ones, each with a different group of companies, a different event, and a different test. Set up the groups first, then drill down. Pick one to jump to its section.
Each group is reported on its own and never blended — a company that moved to Texas, a company that was already in Texas when the law changed, and a company that elected the 3% derivative threshold are three different experiments. The counts above hydrate from the canonical dataset. Day-0 abnormal-return figures throughout are illustrative, pending the full CRSP / Fama-French re-run on the corrected panel.
Did the companies that changed their state of incorporation show any market reaction?
One analytical cohort: every company that announced or completed a change to its state of incorporation — most to Nevada, the remainder to Texas and other states — measured together as "the movers" and split by direction below. The event is each firm's own reincorporation, not the statute. We never blend a move into Texas with a move out of it.
When a company changes its legal home — whether to Texas, Nevada, Delaware, or another jurisdiction — does the stock move on the day investors learn of it? If changing the state of incorporation created or destroyed value, the announcement-day returns of the movers would cluster away from zero. We line each mover's return up against what the broad market did that day and look at the gap.
Mover roster. The 7 movers that also elected the 3% derivative floor on the way in — Tesla, Dillard's, Eightco, Forward Industries, United States Antimony, Coinbase, TTEC Holdings — appear again in Group 3, where the adoption event (not the move) is the thing being tested. This roster hydrates from the canonical Texas Governance-Amendment Matrix (v8) and the validated adoption funnel — the names above are written by the page from the roster object, not hand-typed. The full mover roster, direction splits, and per-firm event dates will live on the cohort event study (coming soon) →, which carries the announcement→effectiveness two-event design. The two mover counts reconcile: 85 is every state-change move tracked in the registry; 77 is the analytical cohort that survives the event-study screen — the 8-firm gap is moves dropped for insufficient clean price history, a non-public destination, or a failed event-window screen.
Did the new Texas laws (SB 29) move companies that were already in Texas?
A different group and a different event. Here the companies do not move at all — the 62 Panel-A Texas-incumbent firms (Southwest among them) were already incorporated in Texas when the legislature changed the law. The "event" is the passage of Senate Bill 29 itself: the statutory shock to firms already inside Texas, independent of any single reincorporation.
If Texas's new governance package made Texas a better (or worse) place to be incorporated, the companies that were already there should have repriced when the law passed — no move required. So we hold the Texas-incumbent group fixed and watch their stock prices around the days SB 29 cleared each hurdle. The market could have priced the news at first passage, at final passage, or only at the Governor's signature — so we test every date, not just the ceremony.
The market could price at first passage, final passage, or signing — each is a separate event date. News does not always wait for the ceremony: a bill that clears its first chamber decisively is already a credible signal. So each of these dates is a candidate Day-0 for the Texas-incumbent group, and we report the reaction around each rather than cherry-picking the signing. SB 29 was signed 2025-05-14 and took effect immediately. Note: the 2025-09-01 "effective" date some sources attach is not SB 29 — it is when SB 1057 (the §21.373 shareholder-proposal threshold), a separate bill, took effect. Dates and recorded votes verified against the enrolled-bill certification (SB00029F) and the Texas Legislature Online bill history. (LegiScan's tracking feed reports the Senate votes as 29–2; the enrolled-bill certification — the authoritative record — states 30–1.)
Did electing the 3% derivative threshold reprice the firms that chose it?
A third, separate grouping: not who moved and not the law passing, but the firms that affirmatively elected the 3% derivative-suit floor (various ways) versus those that declined it. The event is each firm's own adoption. This is the question the rest of this page drills into in full detail.
SB 29 is an opt-in: a Texas company has to write the 3% floor into its own charter or bylaws — moving to Texas does not do it automatically. So the cleanest test of whether the threshold itself matters is to watch the firms that chose it on the day they flipped the switch, and contrast them with firms that cited the statute and declined (ArcBest) or adopted then reversed (International Bancshares). If electing the floor changed firm value, the adopters' Day-0 returns would move.
Roster arithmetic: 8 incumbents + 7 migrators + 1 opt-out (ARCB) + 1 rescinded (IBOC) = 17 roster rows; the 15 "active 3% floor adopters" KPI is the 8+7 standing electors, excluding ArcBest and International Bancshares. Drill all the way down. The sections below are the full Group-3 file: what the threshold does, the adopter taxonomy, adoption-over-time, the firm × mechanism landscape, the Day-0 forest plot, the power analysis, the full primary-source roster, the §21.419 drafting guardrail, the separate §21.373 proposal threshold, and the hostile audit. Every adopter is tied to its own EDGAR filing.
Set up the groups first, then run every test on each. Below is the full battery and, honestly, which pieces have run and which are queued for the re-run on the corrected panel. A status chip marks each.
The three groups call for three identification strategies. Group 1 (movers) is an event study on each firm's reincorporation date with a Heckman correction for the self-selection of which firms move (the HQ-state selection equation). Group 2 (incumbents) is a difference-in-differences around the SB 29 dates — Texas incumbents are the treated group, matched non-Texas firms the controls — plus an event-study CAAR at each candidate Day-0 with FF3/FF5 factor controls and placebo dates. Group 3 (adopters) is a per-firm adoption-date event study. Across all three we use cluster-robust / Newey-West standard errors, Romano–Wolf step-down p-values for the multiplicity of dates and cuts, and a synthetic control for the largest single names. Inference is read against Brown & Warner (1985), MacKinlay (1997), Patell (1976), and Corrado (1989).
Honest status. The structure and the clean-incumbent reads are in place; the cross-firm estimators marked queued await the full CRSP / Fama-French re-run on the corrected 65-firm cohort panel (the event-study analysis set — TX incumbents plus matched controls — defined in the data-lock note below; distinct from the Panel A universe). No abnormal-return number on this page should be cited as final until that run lands.
SB 29 added §21.552 to the Texas Business Organizations Code. It raises the bar for who can bring a derivative suit — a suit filed by a shareholder in the company's name against its own directors.
Before a shareholder can sue the board on the company's behalf, a Texas company can now require that the shareholder — alone or in a group — own at least a set percentage of the company, up to 3%. A shareholder holding less than the elected threshold would not have standing to bring such a suit. The rule is permissive, not automatic: the company must affirmatively write the floor into its charter or bylaws. Most Texas companies have not done so. The ones that have are listed below, each tied to its own filing.
§21.552(a)(3) lets the corporation set a derivative-standing threshold measured on outstanding shares; the statute caps the electable figure at 3% (a ceiling on the corporation's election), and the elected figure becomes the plaintiff's minimum-ownership standing floor. Standing requires
where \(s_{\text{plaintiff}}\) is the plaintiff (or plaintiff group's) beneficial holding and \(S_{\text{outstanding}}\) is total outstanding shares at the time the proceeding is instituted. This is a distinct instrument from §21.373 (the SB 1057 shareholder-proposal threshold, measured on voting shares) — different denominators, different bills, different elections.
Election into §21.552 takes several shapes. Some firms were already in Texas; some moved in and adopted on the way; one cited the statute only to decline it; one adopted then reversed. The scoreboard keeps each category separate.
Plain-English gloss: 8 Texas companies that never moved have written the 3% derivative floor into their bylaws; 7 companies adopted it as part of moving to Texas — Tesla, Dillard's, Eightco, Forward Industries, United States Antimony, Coinbase, and TTEC Holdings; United States Antimony, for example, completed a Montana→Texas conversion (effective 2025-08-18) and wrote the literal 3% floor into its new Texas bylaws (Art. XI §11.3); one company (ArcBest) deliberately rejected the floor while citing the statute; and one (International Bancshares) adopted it and then took it back. The two are §21.552 electors but are not counted among the 15 active 3% floor adopters.
Each step is one firm writing the 3% floor into its operative instrument, dated to the bylaw or charter effective date. The single down-step is International Bancshares reversing its election.
What this chart shows. Adoption arrived in a thin, stepwise trickle — one or two firms a quarter — with a single reversal (IBOC). Under either weighting, uptake never accelerates into a stampede. That shape is the signature of the Early-Adopter phase of normal diffusion (Rogers 1962; Moore 1991), not an abnormal market rejection of the statute. Hover any step for the firm, effective date, and EDGAR accession.
One grid, every firm that touched the Texas governance menu, every mechanism it used. Each filled cell links to the firm's own operative instrument on EDGAR. This is the structured replacement for a prose roster — sortable, and honest about the cells still in dispute.
What this grid shows. Adoption is real but narrow: the already-Texas incumbents plus migrators Tesla, United States Antimony, and Coinbase wrote the literal 3% floor directly; the remaining migrators (Dillard's, Eightco, Forward Industries, TTEC Holdings) reached it through the §21.419 election route (light cells) — which is why a plain "21.552" keyword search misses them. One firm opted out, one rescinded. Cap vs floor: §21.552(a)(3) caps the electable threshold at 3% (a ceiling on the corporation's election); a firm may elect a lower threshold, which becomes the plaintiff's standing floor. ArcBest's instrument drafts "§21.552(3)" [(a)(3)] by number (single-share opt-out); the others adopt via §21.551(2) standing + §21.419 election + a 3% bylaw. The §21.373 cells for DDS and FWDI are elected (DDS Cert. art. FIFTH(b); FWDI Cert. of Formation Ex. 3.1, art. VII §4), so the §21.373 adopter count is not zero.
For each adopter we measure the stock's Day-0 abnormal return around its §21.552 adoption date — the return left over after subtracting what a market model predicts. If electing the 3% floor mattered to investors, those points would cluster away from the dashed zero line.
Think of each adoption as flipping a switch and watching whether the stock reacts that day, controlling for what the whole market did. The cleanest read is Southwest — already a Texas company, no move muddying the picture — whose Day-0 abnormal return was about +0.09% (p ≈ 0.96): indistinguishable from zero. Every other clean incumbent lands on the same verdict. The pooled-mean diamond straddles zero. Tesla's point (grey) is shown but set aside: its election sat next to a much larger Delaware-to-Texas move, so its reading is confounded.
What this chart shows. Every clean incumbent's 95% confidence interval crosses zero, and the pooled-mean diamond is centred on zero. Under both the parametric test and the distribution-free Corrado (1989) rank test, no adopter shows a Day-0 reaction distinguishable from noise. The plot carries only TSLA among the migrators by design — the other migrator adopters each elected the floor adjacent to a redomestication, so their Day-0 reads are confounded and are not shown here. The result is statistically indistinguishable from zero under the tests applied — not "proven zero," because the cohort is small (see the power panel below).
For firm \(i\) the market-model abnormal return on the event day is \(AR_{i,0}=R_{i,0}-(\hat\alpha_i+\hat\beta_i R_{m,0})\), with estimation-window residual standard error \(\hat\sigma_i\). The parametric Day-0 test is the standardized statistic
Because Patell-type tests are sensitive to fat tails and event-induced variance, we also report the distribution-free Corrado (1989) rank statistic, which replaces returns with their within-firm ranks and is robust to non-normality. Pooling is equal-weighted across the clean incumbent reads; the confounded migrator (Tesla) is excluded from the pooled diamond. Inference is read against Brown & Warner (1985) and MacKinlay (1997); for the cross-firm multiplicity we defer to the Romano & Wolf (2005) step-down battery reported on the cohort page.
A small cohort can only "see" large effects. Before reading the null as "nothing happened," it is worth asking the honest question: what is the smallest effect this design could reliably have detected?
Plain English. The power panel runs on n = 7 clean incumbent reads (the eighth Tier-1 adopter, HeartSciences (HSCS), is excluded for insufficient clean price history). With only that handful of clean adopters, the test can reliably catch a same-day move of roughly ±1.5% or larger. The effects the governance literature expects — tens of basis points — sit below that floor. So a null is exactly what an underpowered design returns whether the true effect is zero or merely small. We report it as descriptive, neutral evidence, not a verdict.
The fix is not a cleverer single-day test on these n = 7 reads — it is the cross-firm battery, every sample cut, on the corrected panel. See the cohort event study (coming soon) →.
For a two-sided test at level \(\alpha\) with power \(1-\beta\), per-firm Day-0 standard error \(\sigma\), and \(n\) independent clean reads, the minimum detectable mean effect is
With \(z_{\alpha/2}=1.96\), \(z_{\beta}=0.8416\) (80% power), \(\sigma\approx1.45\%\), and \(n=7\) clean incumbent reads, \(\mathrm{MDE}\approx\) 1.54%. Any true Day-0 effect smaller than this is, by construction, undetectable here — so failing to reject \(H_0\) is uninformative about effects in the sub-percent range the literature anticipates. The argument follows the power discussion in MacKinlay (1997).
Every card is verified to the firm's own operative instrument on EDGAR — the bylaw section or charter article that contains the 3%-of-outstanding-shares language. Accession numbers link to the primary source.
The migrator adopters do not cite "§21.552" by number. They elect §21.419 and then write the 3% language directly. A literal "21.552" search misses them — which is exactly why a verification guardrail is needed.
Several firms — Dillard's, Eightco, Forward Industries — implement the threshold by saying "the Corporation affirmatively elects to be governed by Section 21.419" and then writing "at least 3% of the outstanding shares." So if you search a filing for the text "21.552" you won't find them, even though they have plainly adopted the 3% floor. The fix is to search for the §21.419 election together with the 3%-derivative-proceeding language.
§21.419 codifies the business-judgment rule; §21.552 sets the derivative-standing threshold. The two are routinely drafted together but are not the same instrument. The guardrail: a §21.419 election alone is not a §21.552 adoption. The operative instrument must also contain the 3%-of-outstanding-shares derivative-proceeding language for the firm to count as an adopter. Reading a bare "elects §21.419" as a threshold election is a category error — the kind the verification pass specifically caught (e.g., a §21.419-only instrument with no 3% floor is not an adopter).
For each verified adopter we line up the stock's return on the announcement day against what a market model says it should have done. The gap is the abnormal return. If the threshold mattered to investors, those gaps would cluster away from zero.
The cleanest single-firm read is Southwest Airlines — already a Texas company, so there is no reincorporation event muddying the picture. Its announcement-day abnormal return was about +0.09% with a p-value of 0.96: statistically indistinguishable from zero. Tesla's election is harder to read cleanly because it was close to a much larger Delaware-to-Texas move. Pooled across the verified adopters, the announcement-day reaction is null. This is a statement about one day's pricing, not about whether the governance change matters over the years it plays out in court.
SB 1057 added a separate, distinct provision: a shareholder-proposal threshold. It is tracked separately and has its own (much lower) uptake.
§21.373 incumbent adopters: 0. SB 1057 (effective 2025-09-01) lets a Texas company require the lesser of $1,000,000 in market value of voting shares or 3% of voting power — with a holding period — before a shareholder can force a proposal onto the ballot. It is measured on voting shares, not outstanding shares, and is a different bill from §21.552.
The §21.373 regime faces federal-preemption uncertainty under SEC Rule 14a-8 and the Supremacy Clause. With no court ruling yet, Texas incumbents appear to be in a wait-and-see posture. Low uptake among incumbents is consistent with the absence of a broad rush to adopt, but is not proof of market indifference. (Note: among movers, both Dillard's (DDS Cert. art. FIFTH(b)) and Forward Industries (FWDI Cert. of Formation Ex. 3.1, art. VII §4) affirmatively elect §21.373 in their own instruments — so the broader §21.373 adopter count is not zero; the 0 above is the Panel-A incumbent uptake, which remains zero.)
A referee at the Journal of Finance or JFE would press four objections before believing the null. We state each one in its strongest form, then give the honest status — what the design can and cannot answer.
The cohort is firms that actually wrote the floor into a charter or bylaw. Firms that considered it and declined, or quietly adopted without an 8-K trigger, are invisible. Adopters may be precisely the firms for whom the change was already priced in.
Open · biases toward null Status: disclosed and directional — the selection pushes the headline toward zero, so the null is conservative, not inflated. Cannot be closed without the full opt-in choice set.
With n = 7 clean incumbent reads (the eighth Tier-1 adopter, HeartSciences (HSCS), excluded for insufficient clean price history) the minimum detectable Day-0 effect is roughly ±1.5%. The governance literature expects effects an order of magnitude smaller. A null is the mechanical output of an underpowered design.
Open · power-limited Status: quantified in §06 (MDE ≈ 1.54%). We report "no detectable effect," never "proven zero." Resolved only by the cross-firm battery on the corrected panel.
Several adopters are controlled or founder-led (e.g., concentrated-ownership incumbents). A derivative-standing floor changes little for a firm whose insiders already dominate litigation risk — so a null could reflect the sample, not the statute.
Open · sample-dependent Status: acknowledged; the page does not condition on controller status. The cohort study carries the controller column and Heckman HQ-state correction.
Migrators elect §21.419 and write "3% of outstanding shares" without ever naming §21.552. A naive keyword tally under-counts; a careless one over-counts a bare §21.419 election as a 3% adoption.
Handled · guardrail enforced Status: every migrator cell requires the literal 3%-of-outstanding language in the instrument (§08). A §21.419 election alone is not counted. The §21.373 cells for DDS and FWDI are elected, verified to DDS Cert. art. FIFTH(b) and FWDI Cert. of Formation Ex. 3.1, art. VII §4.
Firms do not reincorporate at random. If the firms expecting a benign market reaction are also the ones likelier to move or to adopt, an uncorrected null is partly a selection artifact, not a clean read of the law's effect. The corrected re-run carries the battery below to separate the two.
Imagine only the companies that already expect investors to shrug ever bother to move or to elect the 3% floor. Then finding "no reaction" partly reflects who self-selected in, not what the change does to a random firm. The fix is to compare each mover or adopter to the look-alike firms that could have moved but didn't, and to model the choice to move itself.
| Method | What it does | Reference | Status |
|---|---|---|---|
| Propensity-score matching (PSM) | Match movers to look-alike non-movers on size, industry, controller status, HQ-state, institutional ownership, and litigation risk — selection on observables. | Rosenbaum & Rubin (1983) | Queued feasible first; runs as robustness |
| Heckman two-step | Stage-1 probit Pr(move) with HQ-in-Delaware as the exclusion restriction → inverse Mills ratio λ into the abnormal-return regression; a significant λ coefficient signals selection bias. | Heckman (1979) | Queued prerequisites: complete the controller-status audit + build the at-risk non-mover Delaware control set |
| Synthetic control / SDID | Build a custom "twin" from a weighted mix of non-movers for the marquee single firms (Southwest, ExxonMobil); placebo-in-space inference. | Abadie, Diamond & Hainmueller (2010) | Queued best for LUV / XOM |
| Instrumental variables (2SLS) | Exploratory only; no clean instrument exists here — every shifter of Texas-law attractiveness is entangled with the industry / regional shock. Documented, not run. | Stock & Yogo (2005) | Design only not pursued |
The same selection-bias battery applies to the cohort event study and the tracker; abnormal-return rows are reported as PROVISIONAL until the corrected-panel re-run and the selection correction are complete.
This page draws the line itself — separating the claims the filings settle from the items a reviewer should see flagged before citing.
data.json ↔ §150.8 ledger re-sync — the Panel A denominator hydrates live (canonical key panel_a_tx_incumbent_count, fallback 62, locked at the canonical 62-firm spine (2026-06-16); the 59 figure was the G4 source-gate metric, not the universe; the 65-firm figure cited in the abnormal-return status note is the separate cohort event-study analysis panel — TX incumbents plus matched controls — not the Panel A universe); Panel A is locked at the canonical 62-firm spine, and the broader §150.8 amendment-column merge remains a rev138 item.This page states its own limits first: which numbers are settled to the §150.8 ledger, and the one open data-reconciliation item a reviewer should see before citing the denominator.
data.json and is locked at the canonical 62-firm spine.Every number on this page is bound to a single object and hydrates from the canonical dataset; resolve the flagged item in the canonical workbook and every surface above updates at once. No adopter count is hand-typed into the layout.