Option value and irreversibility in legal and regulatory monitoring

The general counsel's clock

A compliance programme is a bet on the current shape of the law holding still long enough to be worth building. General counsel who signs off on a policy is making an implicit claim: that the docket, the proposed rule, the enforcement posture of the relevant agency will not move in a way that invalidates the policy before it does its work. That claim has a shelf life, and the shelf life is rarely disclosed. Somewhere in most legal departments sits a compliance manual built on a rule that was superseded two quarters ago, unnoticed because nobody was watching the docket that mattered. The manual is not wrong because someone was careless. It is wrong because the intake regime that produced it had no mechanism for knowing when the ground had moved.

Two positions bracket this problem, and both are held by serious people inside real legal departments. Neither is foolish. Set them against each other properly before choosing.

Position one: waiting is cheap and informative

The first position says: build slowly, watch continuously, commit late. Regulatory rulemaking has a known structure. A Notice of Proposed Rulemaking opens a comment period, typically 30 to 60 days under the Administrative Procedure Act's framework. A final rule usually carries a further delay — commonly 60 days — before it takes effect. Enforcement actions telegraph themselves through consent orders, warning letters, and settled cases that read as precedent for the next target. None of this is noise. Each stage narrows the distribution of what the final rule will require, who the agency intends to pursue, and on what theory. Waiting through a comment period to finalise a compliance posture is not indecision. It is the correct application of an interval that carries real information, and economists have a name for the value of holding a decision open through such an interval: quasi-option value, the premium Arrow and Fisher priced into irreversible choices back in 1974, on the argument that delay pays for itself exactly when delay brings evidence.

On this view, a general counsel who commits early — writing binding internal guidance the week an NPRM is published rather than waiting to see what survives the comment period — is destroying option value for no reason beyond a preference for closure. The docket runs continuously: PACER filings, Federal Register notices, agency guidance letters, appellate decisions narrowing or widening a statute's reach. A department that treats these as a live stream, revising its risk register on each material filing, is doing nothing more exotic than pricing the interval correctly.

Position two: filtering cost eats the gain, and every revision is a small commitment

The second position takes the same facts and reaches the opposite operating conclusion. Continuous docket monitoring generates volume, not signal. A general counsel's team watching every filing across every court where the company has exposure will drown in filings that are procedurally routine, factually distinguishable, or filed by parties whose theory will not survive a motion to dismiss. The cost of triage — reading enough of each item to decide whether it matters — is the actual binding constraint, and it scales with intake, not with insight gained. A well-curated quarterly compliance digest, built by someone who has already done the filtering, may deliver more usable certainty about the one variable that matters — will the agency's enforcement priorities touch our sector — than a live feed the department has no capacity to read properly.

Worse: revising a compliance posture is itself an action with cost. Reissuing guidance to a sales force, retraining a call centre on a new disclosure script, amending a standard contract clause — each of these is a small irreversibility. Dixit and Pindyck's observation about real options cuts against continuous revision as hard as it cuts against premature commitment: acting on each new signal, even a genuine one, can destroy more option value through repeated switching costs than it recovers through better-timed information. A department that reissues guidance every time a circuit split narrows is not disciplined. It is thrashing, and thrashing is expensive in exactly the currency — organisational attention, credibility with the business units being governed — that compliance programmes run on.

"We watched every filing in three circuits for eighteen months and changed our disclosure language four times. The fourth version was worse than the first. We'd have done better freezing it after the first circuit ruled and living with the risk."

That is a real position, voiced by people who have tried continuous monitoring and been burned by it. It should not be dismissed as a failure of nerve.

Refusing the cheap resolution

The lazy synthesis says: do both, monitor continuously but act discretely. That is true and also empty — it describes the goal without solving the actual difficulty, which is that filtering cost and thrash risk are real costs, not rhetorical ones, and they scale with the volume of streams a legal department tries to hold live. The Large Language Model framing of this problem — a compliance manual as a document frozen at drafting, unrevised until the next scheduled review — has one clean virtue: it is honest about its own staleness. It does not pretend to be watching anything. The continuous-intake position has no such honesty by default; a department that claims to monitor every docket and fails to filter well is worse off than one that never claimed continuous coverage, because it has the confidence of currency without the substance of it.

Here the two positions actually narrow each other rather than cancel. The quasi-option value argument is correct in principle and unconditional: a rule that changed last quarter cannot be un-changed by a general counsel who simply didn't notice, and no amount of confidence in last year's manual repairs that. Position two is correct about implementation: dumping every stream into one undifferentiated feed and reacting to each item is a way of paying the cost of continuous intake without collecting the benefit. The resolution is not "monitor everything" but "separate the update from the commitment." A department can hold a live, provenance-tagged register of every docket entry, NPRM, and enforcement action relevant to its exposure — updated continuously, cheaply, without triggering any downstream action — and reserve the expensive act of reissuing guidance for the moments a defined threshold is crossed: comment period closes, final rule publishes, circuit ruling becomes binding in the relevant jurisdiction. That register is the mechanism the option-value literature calls learning without committing. The guidance reissue is the option being exercised, deliberately, and rarely.

Where the horizon actually falls

The table is the actual answer, and it is narrower than either opening position. Continuous intake is not generally superior; it is superior exactly where the decision horizon is shorter than the review cycle a static programme would otherwise use. For most of a compliance manual's contents — settled doctrine, routine filings, boilerplate — the frozen quarter costs almost nothing, because the correlation time of the underlying law is long. For the specific docket items where an agency's position is visibly in motion, freezing the record for a quarter is not caution. It is the two-quarter gap that eventually surfaces as a manual built on a rule that no longer exists, discovered only when an auditor, a plaintiff's lawyer, or the agency itself points it out.

The general counsel's real job, on this account, is not choosing between the frozen manual and the live feed. It is deciding, docket by docket, which items sit in the top row of that table and which sit in the bottom — and building the one live, provenanced register that makes that classification possible in the first place, rather than discovering it by omission.