Emergency powers ratchet

Why do powers granted for a passing emergency so rarely go back once the emergency has passed?

The bargain sounds reasonable. A crisis arrives, ordinary procedures are too slow, so the legislature grants extraordinary authority on the understanding that it is temporary. Everyone agrees it is temporary. It is often written down that it is temporary.

Then the crisis ends and the authority does not. The tempting explanation is bad faith — someone wanted the power all along. That is too easy, because the pattern also appears where nobody wanted it and everyone said so. What is the machinery?

Begin with who decides. The grant runs from a legislature to an executive. Who decides to end it? In most designs, either the executive itself, or a legislature that must act affirmatively to withdraw it. That is not the configuration that made the grant: ending requires an affirmative act by someone who either holds the power or must overcome the holder's resistance.

One case makes the asymmetry exact. The United States National Emergencies Act of 1976 was passed precisely to stop open-ended emergencies: it required periodic congressional review and allowed termination by concurrent resolution. Two things then happened. The reviews were largely not conducted. And in 1983 the Supreme Court's Chadha decision struck down the legislative veto, so termination had to proceed by joint resolution — presentable to the President and therefore vetoable. Do the arithmetic: overriding a veto takes two-thirds of both chambers, so a President wishing to continue an emergency needs just over one-third of one chamber. A rule meant to require a majority to continue became a rule requiring a supermajority to end (statute and case stated as recalled; the threshold follows from the two-thirds requirement itself).

Second element: what counts as the end? A tax expires on a date. An emergency ends when the emergency is over — a condition with no observable threshold, no agreed measure, and typically reviewed with great deference or treated as non-justiciable. No state of affairs proves the crisis finished, so "should this be renewed?" has no fact that answers it, and the answer falls to whoever bears the risk of being wrong.

Which brings the third element, and it does most of the work. The two errors are not attributed alike. Vote to end the powers and an attack follows, and the ending has a name on it. Vote to continue and the cost — surveillance of people who did nothing, detentions, an executive acting as no legislature would have authorised in calm — is spread thin, arrives without a date, and is attributed to nobody. That is the shape of the retention asymmetry in persistent-rules.md, and the difference matters: there the rule is defended by a concentrated private beneficiary, whereas here the beneficiary of continuation is the institution that must vote to end it, and the feared cost is not inconvenience but catastrophe — which makes the asymmetry far steeper and far harder to oppose in public.

Then the element where "ratchet" earns its keep. The emergency often does end; what survives is the content, transferred into ordinary law. France declared a state of emergency in November 2015, extended it repeatedly, and lifted it in late 2017 as much of its substance was written into ordinary counter-terrorism legislation. Britain's Prevention of Terrorism (Temporary Provisions) Act was renewed year after year before permanent legislation superseded it. On this route the emergency genuinely terminates, the powers persist under another name, and the metric everyone watches — is the emergency still declared? — reports success.

Why does the transfer succeed where the same grant would have failed in peacetime? Because the powers are no longer novel. Agencies are built around them, procedures and budgets assume them, and the argument shifts from "do not create this" to "do not remove what we rely on". Normalisation is how the baseline moved.

Two correctives. This is a tendency, not a law: many emergency measures have lapsed, including much pandemic-era regulation. And some retention is defensible, since crises reveal gaps ordinary law should have covered, and calling every survival a usurpation assumes the pre-crisis settlement was right.

The refinement connecting the pieces is that the ratchet lives in the termination rule, not the grant. A power expiring automatically unless renewed by a majority behaves quite differently from the same power granted indefinitely and terminable only by a supermajority — though both are called temporary. Whichever side gets the default gets the outcome whenever the legislature is divided, which is most of the time.

That is where the comparison does real work. The German Basic Law's emergency provisions, added after long controversy in 1968, are strikingly specific about what may be done and by whom, and have essentially never been used. Several Latin American constitutions cap a state of exception at a fixed number of days, with automatic lapse and legislative confirmation required to extend. Others provide almost nothing and rely on general executive powers. The interesting variation is not in the list of powers but in who must act, and by what majority, for the exception to persist. Whether tightly codified emergency rules restrain executives or merely give unusual action a lawful form is argued seriously on both sides, and I do not think it settled.

That yields a comparative test. If the termination rule drives the outcome, then across emergencies of similar type and severity, those with automatic sunset and simple-majority renewal should lapse markedly more often than those requiring a supermajority to end, holding the threat constant. The refuting observation: lapse rates indistinguishable across those designs and tracking instead whether the threat visibly disappeared — which would put the ratchet in threat perception rather than institutional design.

Finally, who benefits. The termination rule is set and amended by the same institutions whose position it protects: an executive rarely proposes that its own emergency authority lapse automatically, and legislators who would be blamed for ending it are not eager either. That is much of why these defaults are so stable across so many systems.

Every term the collection defines is gathered in the glossary.