Methodology

SMU Cox Corporate Governance Initiative · v3.84-rev5z · 2026-04-30 · Back to dashboard · Methodology · Corporate Law History · References · Legal & litigation · Legal & litigation · How to Cite

Methodology

The Reincorporation Tracker follows public-company proposals to change state or jurisdiction of incorporation beginning with Tesla's June 2024 move from Delaware to Texas. Each row is independently classified by legal status, primary-source evidence, and effective-date verification. Completed status requires both an actual effective date and a primary-source filing.

Source hierarchy. Each row is anchored to the highest-confidence source available, in this order:

• SEC EDGAR proxy filings — DEF 14A, PRE 14A, DEFM14A, DEFA14A. These provide announcement, meeting, and vote-result documentation.
• SEC EDGAR 8-K filings — Item 5.03 (amendment to articles / certificate of incorporation), Item 5.07 (vote results), Item 1.01 (material agreements). These document effective dates.
• State filings — secretary-of-state certificates of conversion or merger; Texas SOS, Nevada SOS, Delaware Division of Corporations.
• Company investor relations — used as secondary corroboration; never the sole source for completed-status classification.
• Press releases / rumored filings — flagged as RUMORED with source_confidence=RUMORED_NO_FILING; excluded from completed counts and empirical panels.

Status taxonomy (controlled vocabulary):

• COMPLETED — Vote passed AND actual effective date populated AND primary source verified. Counted in completion totals and timeline.
• APPROVED — Vote passed but no recorded effective date yet. Counted in approval totals; not in completion totals.
• SCHEDULED — Meeting date in the future; vote pending. Anticipated effective dates carried in proposed_effective_date_iso, never actual_effective_date_iso.
• REJECTED — Shareholders voted against; documented in 8-K Item 5.07.
• WITHDRAWN — Proposal withdrawn before vote; documented in DEFA14A or 8-K.
• RUMORED — Reported in press without a corresponding SEC filing. Monitored but not counted.
• LEGAL_CONTEXT / MERGED-INTO-TRIP — Legal-setting benchmarks (e.g., Maffei v. Palkon firms; Liberty TripAdvisor merged into Tripadvisor) carried for analytical context, flagged with is_legal_precedent=1 and public_tier=LEGAL_CONTEXT.

Source confidence (per row):

• VERIFIED_EFFECTIVE — actual effective date plus EDGAR accession on file
• VERIFIED_VOTE — vote outcome documented but effective date not yet sourced
• VERIFIED_PROXY — proxy filing on EDGAR; vote outcome pending
• COMPANY_RELEASE — IR or press release; not yet corroborated by SEC filing
• RUMORED_NO_FILING — no primary source; press chatter only
• NEEDS_MANUAL_REVIEW — flagged for follow-up research

Effective-date convention. The tracker distinguishes actual_effective_date_iso (legally effective; populated only when status=COMPLETED) from proposed_effective_date_iso (anticipated date from press release or proxy filing). The timeline visualization uses only actual dates. Firms whose effective date predates the current date by more than 60 days have their forward meeting dates suppressed because subsequent annual meetings at the new domicile do not constitute reincorporation events.

Bucket architecture — how we group firms for comparison. Plain-English label first, internal code second.

• Pre-SB29 completed move (Bucket A) — Delaware-outbound to Texas or Nevada with effective date before Texas SB 29 (May 14, 2025).
• Post-SB29 completed move to Texas (Bucket B1) — Delaware-outbound to Texas with effective date on or after May 14, 2025. Also includes selected non-Delaware outbound moves to Texas (e.g., NY→TX, MT→TX) retained as comparators.
• Post-SB29 completed move to Nevada (Bucket B2) — Delaware-outbound to Nevada with effective date on or after May 14, 2025. Also includes selected non-Delaware outbound moves to Nevada (e.g., MI→NV).
• TX-incumbent (Bucket C) — firm has always been Texas-incorporated; never reincorporated from Delaware. Tracked for legal-context comparison; not counted as a move.
• Pending / not yet effective (Bucket D) — vote scheduled, pending, or passed but no recorded effective date. Also includes rejected proposals.
• Comparator (Bucket Y) — Delaware-outbound to a destination other than Texas or Nevada (e.g., DE→Maryland, DE→Cayman). Retained as a revealed-preference comparator; not counted in the core Texas/Nevada wave.
• Out of scope (Bucket Z) — withdrawn proposals, name-only changes, merger-only events, and other non-reincorporation transactions. Carried in the master research database but excluded from the public dashboard.

Public tracker vs. research panels. The public dashboard records every public-company reincorporation since June 2024 — the "DExit phenomenon" as a journalistic / transparency deliverable (49 rows in the registry as of 2026-04-30; counts auto-derived from data.json on each build). The empirical claims associated with this project are tested using separate research panels constructed to answer specific causal questions; those panels are maintained in the master research database (DExit_Central_Database_v3.X) and described in the companion academic memo. The two are intentionally distinct: the tracker is curated for readability and breadth; the panels are curated for causal inference.

Research panels (chronological).

• Panel A — TX-incumbent universe (pre-SB29). Firms Texas-incorporated as of Texas SB 29's effective date (2025-05-14). N = 70. Includes TSLA (TX since June 2024), LUV (always TX), ATO, CNP, MTDR, EQPT, and other long-time TX firms. Test: did SB 29 affect firms that were already in Texas? Panel B in the locked spec; renamed to Panel A for chronological clarity.
• Panel B — DExit movers. Firms that reincorporated from Delaware (or other origin) to Texas or Nevada. Pre-SB29 and post-SB29 waves combined. N ≈ 30. Test: did the act of reincorporating produce abnormal returns?
• Panel C — Texas statutory bylaw adopters. Subset of Panel A (and Panel B post-completion) that affirmatively adopted Texas's two new opt-in regimes:
  • §21.552 (SB 29) — derivative-action 3% ownership threshold (Tornetta-response statute). 7 verified adopters as of 2026-04-30: TSLA, DDS, UAMY, CNP, LUV (under review), LEGH, HSCS. Reconciliation note: Panel C is a research-panel construct that includes §21.552 adopters whether or not they appear in the public tracker; the public dataset summary (panel_c1_21_552_adopters) reflects only adopters that are also tracker rows (currently 3). Counts will diverge by design.
  • §21.373 (SB 1057) — shareholder-proposal threshold for "nationally listed corporations" with principal office in Texas or listed on a Texas exchange. Effective 2025-09-01. To submit a proposal, the proponent must hold the lesser of $1M in market value or 3% of voting shares, for at least 6 months pre-meeting and through the meeting date, AND solicit holders of at least 67% of voting power. Director nominations and procedural resolutions are exempt. Adoption is opt-in via charter or bylaw amendment with proxy notice.
    Legal status (as of 2026-04-29): §21.373 is structurally more vulnerable to challenge than §21.552. Two principal theories: (1) Internal Affairs Doctrine — particularly when applied to companies headquartered in Texas but incorporated elsewhere; (2) dormant Commerce Clause — limits on interstate-commerce burdens. SEC Rule 14a-8(i)(1) permits exclusion of proposals "contrary to state law," so §21.373 may function operationally through Rule 14a-8 rather than face direct federal preemption. The SEC Chair signaled potential compatibility in Oct 2025, but no court ruling has issued. As of 2026-04-29, no firm has been verified as having adopted §21.373 by primary-source 8-K Item 5.03; firms appear to be in a wait-and-see posture pending legal clarity. Tracker shows §21.373 status as "Pending verification."
  Test: did the specific governance mechanisms — not just the state — produce abnormal returns at the moment of adoption?
• Panel D — Staggered DiD on movers. Completed reincorporators with treatment-timing variation. N = 13. Causal effect of moving with staggered treatment.
• Panel E — TX-vs-DE valuation cross-section. N = 3,483. Tobin's Q discount regression on TX indicator + controls.
• Panel F — Three-regime cross-section (TX + NV + DE). N = 3,477. Joint regression on three-state indicator.

A note on bucket vs. panel. A firm may appear in both a public registry bucket and a research panel — the lenses differ but the firm is the same. Tesla, for example, is Bucket B1 (pre-SB29 mover, 2024 wave) on the public registry AND Panel A (TX-incumbent at 5/14/25) AND Panel C (§21.552 + §21.373 adopter). Each per-firm page surfaces all three labels.

Reconciling the 49-firm registry with the n=47 announcement-day cohort. The Tracker's public registry includes 49 firms total. The cohort event study's announcement-day specification reports n=47, computed as 49 minus two firms excluded for non-canonical specification reasons: LTRPA (Liberty TripAdvisor Holdings — excluded because the Delaware-to-Nevada move was bundled into a merger-driven redomestication rather than a stand-alone reincorporation; LTRPA is retained on the registry as a legal-precedent reference firm), and TEM (Tempus AI — excluded for heterogeneous specification owing to extreme variance distortion in the announcement-day window). A third firm, BNZI (Banzai International), is flagged provisional in the audit log because of a temporal inversion between the written-consent execution date and the 14C disclosure date, but BNZI is retained in the n=47 cohort with the flag disclosed. The outcome-day cohort (n=31) further excludes pending and very-recent firms whose effective date or vote outcome falls outside the outcome-window measurement period as of 2026-04-30. The n=47 announcement-day, n=37 microcap-floor (≥ $50M), and n=31 outcome-day specifications are reproducible from data.json by filtering on headline_event_study_included and the cohort-eligibility flags documented in the audit log.

Database versioning. The canonical Excel database (DExit_Central_Database_v3.X.xlsx) is the single source of truth. Every revision is logged in the audit log with rationale and corrective action; the dashboard regenerates from the database on each release. Reviewers can request the current Excel file and the change log directly via sgoodwin@smu.edu.

Known limitations. EDGAR primary-source links are populated for verified-effective firms; for some scheduled or rumored rows the link is to the company's general EDGAR profile rather than a specific filing. Aggregate market-value figures use the most recent database export and are not point-in-time at the announcement or effective date. Logo images use Google's favicon service as a stable fallback; the tracker does not host independently-licensed brand assets.

XOM event-study figures — methodology

Question. Did the March 10, 2026 announcement that ExxonMobil would reincorporate from New Jersey to Texas produce a measurable abnormal return on ExxonMobil's stock?

Estimation window. 2025-03-25 to 2026-03-09 (240 trading days, ending the trading day before the announcement). All coefficients and synthetic-control weights are estimated on this pre-announcement window; the announcement day and post-announcement days are out-of-sample tests.

Six figures, six lenses on the same null hypothesis.

1. Synthetic-control tracking. 21-firm energy-sector donor pool (CVX, COP, EOG, SLB, OXY, MPC, VLO, PSX, DVN, FANG, HAL, BKR, CTRA, OVV, APA, EQT, WMB, OKE, TRGP, LNG, ET). Solver-derived weights: CVX 0.549, EOG 0.198, SLB 0.094, WMB 0.084, COP 0.036, OXY 0.029, CTRA 0.010 (other 14 = 0). Pre-period RMSPE = 0.746%; pre-period correlation = 0.886; pre-period R² = 0.771.
2. Cumulative-gap chart. Daily gap = (XOM cumulative return) − (synthetic-XOM cumulative return), with a ±1.50 pp 95% pre-announcement uncertainty band. Post-announcement readings stay inside the same range the gap occupied before the announcement.
3. In-time placebo histogram. 100 random pseudo-announcement days drawn from the 220 trading days before the real announcement; full SC machinery re-applied to each. Real-day gap = +0.02% sits at the 8th percentile of the absolute placebo distribution. Placebo p-value = 0.92 (i.e., 92% of pseudo-days produced a larger absolute gap).
4. Bayesian posterior. Posterior over the true announcement-day effect, weakly-informative N(0, 5²) prior, conditioned on day-0 gap and pre-period RMSE. Posterior mean = +0.02%; 95% credibility interval [−1.48%, +1.53%]; P(true effect ≤ −2%) = 0.40%; P(true effect ≤ −3%) < 0.10%.
5. Oil-factor importance (nested OLS). Spec 1: XOM ~ SPY → R² = 0.219. Spec 2: XOM ~ SPY + BNO → R² = 0.549 (+33 pp). β_BNO = 0.488, t = 12.6, p < 10⁻²⁷. F(1, 217) = 158.5 (p < 10⁻¹⁶). This justifies the oil-augmented market model as the preferred parametric benchmark for an integrated oil major.
6. Nine-firm Day-0 forest plot. Hand-picked panel of 9 firms — five SB 29 (§21.552) adopters and four high-profile Texas redomicilers — with day-0 abnormal returns from XOM~SPY single-factor market model. Sign tally: 6 negative / 3 positive (binomial sign test p = 0.508). Three firms individually significant at p<0.05: LEGH, DDS, TCBI negative; ARCB positive.

Replication. All six figures are byte-identical to the figures shipped with the FINAL_BLUE_SKY analysis package (SHA-256 verified at deploy time). Source numbers come from data_and_code/xom_rerun_results.json; placebo distribution from placebo_distribution.json; estimation script rerun_xom_primary.py. Per protocol, the underlying daily-close panel and Fama-French factors are bundled with the replication kit.

Interpretation. Five of the six figures point at a single conclusion: XOM's announcement-day move is statistically indistinguishable from the energy-sector benchmark. The Bayesian posterior puts <0.5% probability mass on a −2% governance discount and <0.1% on −3%. The data are not consistent with the disenfranchisement thesis that Texas reincorporations should produce a sizeable negative announcement effect for a firm of ExxonMobil's profile.