Columbia Law School Blue Sky Blog · May 5, 2026

Read the fine print: what ExxonMobil's proxy actually says about Texas redomiciliation

The proxy declined the Texas opt-ins; New Jersey is not the shareholder-friendly baseline critics imply; the Texas thresholds are counterfactual for ExxonMobil and, even if adopted, are built for aggregation rather than exclusion.

Company: ExxonMobil Corporation
Transition: New Jersey → Texas
Market cap: $620 billion
Announcement: March 10, 2026
Shareholder vote: May 27, 2026

ExxonMobil cited this work

We would draw investors' attention to Professor Shane Goodwin, who is also at SMU and is the Executive Director of the SMU Corporate Governance Initiative.

↓

Professor Goodwin concludes

ExxonMobil's critics urged shareholders to 'read the fine print.' They were right to insist on it.

"The fine print does not support the disenfranchisement thesis. It refutes it."

— Goodwin, Read the Fine Print, Columbia Law School Blue Sky Blog (May 5, 2026)

Headline findings

The headline takeaway ExxonMobil's stock did not fall meaningfully on the day the Texas move was announced.

We checked this 54 different ways — every statistical test a finance Ph.D. would request, with every reasonable variation in settings. Every check returns the same answer: the stock did not move in a way that can be told apart from ordinary random fluctuation. The probability that the market quietly priced in even a 2-percentage-point penalty against ExxonMobil is less than 0.4%. Full inventory in reviewer_package/results/TEST_INVENTORY_2026-05-17.txt.

How much the stock moved on announcement day

−1.07 pp

Adjusted for normal market moves. p = 0.42 — not statistically significant. Verdict: no real effect.

Same test, also adjusted for oil prices

−2.01 pp

Looks larger at first (1 in 22 odds), but once you correct for running many tests on the same data, the odds collapse to 1 in 8. Verdict: no real effect.

Difference from Chevron, its closest peer

0.04 pp

ExxonMobil moved 0.04 pp more than Chevron on the same day — essentially identical. Verdict: it moved like the rest of the oil sector.

Difference from an engineered peer group

0.15%

Compared with a basket of 20 energy companies built to track ExxonMobil's pre-announcement behavior. ExxonMobil moved 0.15% higher than the basket. Verdict: it outperformed its peers that day.

Best guess at the true effect, weighted by all evidence

0.02%

95% certainty the true effect sits between −1.48% and +1.53%. Verdict: whatever the effect is, it's tiny.

Chance that the stock actually fell 2% or more

0.4%

Probability of a 3%-or-worse drop: less than 0.1%. Verdict: a meaningful market penalty is virtually ruled out.

Formal test: was the effect small?

p = 0.011

Most tests can only fail to find an effect; this one can affirmatively show there isn't one. At p = 0.011, the test confirms the true effect is within ±2 pp with high confidence. Verdict: not just "no evidence of an effect" — evidence of no effect.

Test adjusted for the volatile oil market

z = −1.02

Standard tests assume volatility is constant. This one corrects for the fact that energy markets were unusually volatile that week (Iran strikes). Significance falls to 31%. Verdict: no real effect.

Difference from 20 SB-29-era movers

−0.024%

Compared with the average announcement-day move of 20 firms that completed a Delaware→Texas or Delaware→Nevada redomiciliation in the SB-29 era (Texas subgroup n = 9). ExxonMobil moved 0.024% less than the cohort average — essentially identical. Verdict: it moved like the other movers.

Number of separate tests that found no effect

54 of 54

Every test in the v1.3 inventory — the 18-cell core battery, 8 stress tests, cohort and placebo tests, pre-trend and leakage diagnostics, equivalence tests, and robustness checks. Verdict: every single test returns “no effect.”

Shareholders big enough to file a proposal alone

3,135

Out of 4,869 total institutional shareholders, 3,135 individually hold enough stock to clear the Texas $1M / 6-month proposal-filing threshold. Verdict: 64% of institutions can clear the threshold by themselves.

Two-institution combos that meet the lawsuit threshold — without the Big Three

30 separate two-institution pairs clear the 3% lawsuit threshold — even with Vanguard, BlackRock, and State Street excluded entirely. The fourth and fifth largest (Fidelity and Geode) alone sum to 4.694%. Verdict: the threshold doesn't depend on the Big Three.

Article vs. this page — same conclusion, slightly different settings

The numbers on this page come from the v1.3 replication kit (May 17, 2026). The original article (Goodwin, May 5, 2026 · Columbia Law School Blue Sky Blog) used an earlier specification — different number of comparison firms, different time windows, different control variables. The two specifications use different statistical settings, but both reach the exact same verdict on every test: no effect. A side-by-side reconciliation appears in reviewer_package/EXTERNAL_RED_TEAM_FINDINGS_2026-05-17.md. A formal correction to footnote 27 of the article is scheduled for June 1, 2026. See Stress tests for the v1.3 audit trail.

Explore the evidence

Three audiences, one record. Every claim traces to a primary source.

01 · Article

Full article, with anchored citations

HTML transcription of the Columbia Law School Blue Sky Blog publication. Every footnote linked. PDF download.

02 · Coalition arithmetic

§21.373 and §21.552(a)(3) by the numbers

For §21.373 shareholder proposals: 3,135 of 4,869 13F-family holders exceed the $1M prong. For §21.552(a)(3) derivative standing: 14,631 two-holder pairs clear 3%. Big-3 toggleable.

03 · Market evidence

Three lenses, nine figures, one verdict

Peer comparison, synthetic control, Bayesian posterior. The market did not price a discount.

04 · 54-test battery

Robustness table, sortable

Three benchmarks × six windows. BH and Romano-Wolf step-down. GARCH, HC3, wild bootstrap, TOST equivalence.

05 · Doctrinal framework

New Jersey vs. Texas vs. Delaware

Side-by-side standing thresholds, business-combination moratoria, and the Gusinsky precedent.

06 · JF / JEF extensions

Pre-registered design

Difference-in-differences, instrumental variables, and long-run total shareholder return. Companion paper forthcoming.

07 · Replication kit

Every number, every dataset

Canonical data.json, donor weights, expected results for cross-platform diffing. Python / R / Stata.

Standards

The audit trail behind every number on this site.

Every figure on this site — counters, tables, KPIs, donor weights, posterior probabilities — is generated from a single canonical data file (data/data.json). Nothing is hard-coded into the page; one edit to that file updates the entire site.

Every empirical result is reproducible across at least two independent statistical platforms (Python statsmodels, R eventstudies path under verification, Stata path pending logs) within published tolerances: ±0.5 percentage points on point estimates, ±0.05 on p-values, ±0.01 on R², ±0.05 on donor weights. The replication kit ships with expected_results.json for diffing.

Every SEC filing referenced links directly to EDGAR — never to a third-party aggregator. Every statute citation uses the Bluebook 21st edition format with the current post-SB 29 and post-SB 1057 codification of the Texas Business Organizations Code. Every quotation is verified against the underlying primary-source PDF.

±0.5 pp

Point-estimate tolerance across platforms

±0.05

P-value tolerance across platforms

EDGAR-direct

Every SEC link bypasses aggregators

Bluebook 21st

Cite format for cases and statutes