When one firm reports, the whole industry's options move

When a firm announces earnings, its own implied volatility moves — that’s well-known and structurally obvious. The interesting object is what happens to its peers. If JPMorgan reports and Bank of America’s at-the-money IV jumps half a vol, is that information transfer? Hedging flow? Index-level noise leaking through? Or just an artifact of all bank stocks tracking each other?

The answer in this paper is: it’s information, it’s industry-level rather than firm-level, and it lasts longer than you’d expect.

The puzzle

A naïve regression of peer IV changes on announcer IV changes will pick up a strong correlation, but you cannot interpret it. The same earnings day might coincide with a Fed announcement, a macro print, a sector-rotating ETF flow — anything that moves all firms in the cross-section gets attributed to the spillover unless you net it out.

The cleanest way to net it out is to date-fix the regression. If the spillover survives date fixed effects, you are no longer picking up market-wide common shocks; you have something that varies within a date, across announcer–peer pairs.

The data set: U.S. single-name equity options, 2010–2024, paired with GICS-based industry classifications to define peer groups. For every announcing firm on every earnings date, every other firm in the same GICS sub-industry is a peer; the regression asks whether peer IV around the announcement window covaries with the announcer’s IV move, holding the date constant.

The coefficient is positive and significant.

Industry-level, not firm-specific

The next question is whether the spillover transmits a firm-specific signal — “BoA reports, the news is bad, JPM gets re-priced” — or an industry-wide one — “BoA reports, the market updates its view of the whole bank-sector announcement season ahead”. The two stories are observationally similar in pairwise data, but they diverge sharply once you control for an industry-date mean.

Add a fixed effect for (industry × date). Now the regression is asking: among firms in the same industry on the same day, do peers closer to the announcer move more than peers further away?

The coefficient on the announcer-specific signal collapses. The signal is not running pair-by-pair; it is running industry-by-day. Whatever the option market is pricing, it is pricing it for the whole industry at once, not as a series of bilateral updates.

This is exactly the pattern you’d expect if the announcement updates beliefs about industry-wide announcement-season uncertainty rather than firm-specific fundamentals.

The signal predicts forward

If the spillover is just a contemporaneous co-movement that decays away by the next session, it’s not particularly interesting. The harder test is whether the size of an early-season spillover forecasts the IV behavior of peer firms when they report.

Two findings on that:

Spillover persists into the peer’s own announcement. Peers that received a larger spillover when an early-reporter announced have larger IV moves at their own announcement, conditional on industry-date controls. The effect is concentrated in peers reporting shortly after — exactly where forward-looking information should matter most.
The signal does not dissipate across the earnings season. A typical “shock that gets absorbed” story would predict spillover effects fade as the season progresses and uncertainty resolves. The opposite happens: spillover-induced IV remains elevated through the late-reporting peers, consistent with the option market continuing to price unresolved industry-wide uncertainty until the season actually closes.

Together, these are not the dynamics of a bilateral information leak. They are the dynamics of an aggregate industry-uncertainty premium that the option market starts pricing as soon as the first relevant announcement lands.

How to think about the mechanism

A clean way to summarize the result is in terms of the variance decomposition the option market is pricing. Write a peer firm’s option-implied variance as

\sigma^2_{\text{peer}} \;=\; \underbrace{\sigma^2_{\text{firm}}}_{\text{idiosyncratic}} \;+\; \underbrace{\sigma^2_{\text{industry}}}_{\text{industry-wide}} \;+\; \underbrace{\sigma^2_{\text{market}}}_{\text{macro}}.

Date fixed effects net out $\sigma^2_{\text{market}}$ . Industry-date fixed effects net out $\sigma^2_{\text{industry}}$ . The spillover that survives date FE but vanishes under industry-date FE is, by construction, the industry-wide announcement uncertainty premium — and the data say that’s where the action is.

The structural interpretation: investors learn something about industry-level earnings season uncertainty from the first firm to report, that information stays in the cross-section, and the option market repeatedly re-prices the remaining peers in light of it.

What this means in practice

A few useful takeaways:

For volatility surface modeling. If you fit a peer firm’s IV term structure during earnings season, you cannot treat individual peer announcements as independent draws. The market embeds an industry-level prior that updates with each early-season report. Models that ignore this will systematically misprice peers reporting later in the season.

For event studies. Standard market-model event-study residuals overstate firm-specific announcement responses for any peer firm that reports later in a clustered season. The “abnormal” IV move includes an industry-uncertainty component that has nothing to do with the firm itself.

For traders. This is not — to be clear — a directional vol-trading edge. The spillover is in the level of pricing, not in the cross-section of mis-pricing. Peers whose IV rises after an early-season announcement are not systematically mispriced; the market is updating correctly. The result is structural, not actionable.

Caveats

Three things this paper deliberately does not claim.

The signal is industry-specific, not market-wide. The spillover does not generalize to cross-sector pairings. Bank announcements move bank options; they do not (in this framework) move tech options. Anyone treating earnings-season IV as a single market-level factor is throwing away the structure that matters.

GICS is the right grouping for these tests, not necessarily the right one for prediction. The paper uses GICS sub-industry as a peer definition because it is exogenous and reproducible. Trading-style models that use textually-derived or co-movement-derived peer groups will pick up different spillover magnitudes; the structural finding does not change, but the size of the effect will.

The 2010–2024 sample covers a lot of regimes. I do not split the analysis by macro regime in this paper — that’s a deliberate choice to keep the headline result clean. Conditional patterns (e.g., spillover behavior during the 2020 COVID earnings season vs. the 2023 banking-stress quarter) are interesting but separate.

What I’d read next

Patell, James M., and Mark A. Wolfson (1979 / 1981). The original studies of option-price behavior around earnings; the empirical baseline that everything in this literature still echoes.
Dubinsky, Kaeck, and Seeger (2018), “Option Pricing of Earnings Announcement Risks.” Strong baseline on how single-name options price earnings risk in the cross-section.
Diebold, Francis X., and Kamil Yilmaz (2014), “On the Network Topology of Variance Decompositions.” The connectedness machinery that frames how to think about spillovers in volatility space, even outside the announcement context.
Hong, Harrison, Walter Torous, and Rossen Valkanov (2007), “Do Industries Lead Stock Markets?” Older but still useful — industry-level information aggregation in equity returns, the analog of what option markets seem to be doing in IV space.

The full paper is at /papers/cross-firm-iv-spillover.pdf. Currently under review at Journal of Financial Markets. SSRN draft: linked here. Comments welcome — simen.guttormsen@nmbu.no.