==========================================================================
WHAT (PARTIALLY) HELPS, AND WHERE CS SITS -- against a real override base rate
==========================================================================

The stage-1 finding stands: comparable commitment rules are overridden often,
near-universally in crisis. Nothing below reverses that. The same literature does,
however, identify factors that correlate with rules holding *somewhat* better:

  PARTIAL MITIGANTS (from the IMF fiscal-rules and CBI evidence):
   - A good TRACK RECORD of compliance raises the political cost of later breach
     (IMF 2022: "benefits of a good track record in abiding by the rules").
   - INDEPENDENT MONITORING (fiscal councils) improves transparency and was
     associated with more disciplined, better-justified escape-clause use in COVID.
   - WELL-DESIGNED ESCAPE CLAUSES (pre-specified triggers + a defined path back)
     channel crisis pressure into a bounded, visible deviation rather than an
     ad-hoc suspension -- override still happens, but on the record and time-limited.
   - DE FACTO independence depends on institutions, not just de jure rules; legal
     text alone is "no guarantee."

  WHERE CS HONESTLY SITS (stated as cost-raising, not prevention):
   - CS's issuance rule is formula-bound and auditable, which is the transparency/
     monitoring factor above, built into the mechanism rather than bolted on. By
     the evidence, that should raise the cost of override -- NOT prevent it.
   - CS has NO operating track record, the single factor most associated with
     rules holding. So on the strongest empirical mitigant, CS scores worst: an
     untested rule has no compliance history to make breach costly.
   - CS centralizes the lever, which the objection correctly flags as the failure
     surface. Auditability makes a breach legible; it does not remove the incentive
     to breach under crisis, which the base rate shows is when rules fall.

  HONEST BOTTOM LINE:
   The fig-leaf risk is real and substantial -- the data supports the objection,
   not the design. CS's auditability is a genuine but PARTIAL mitigant (it raises
   the cost of override by making it visible), and CS is WEAK on the strongest
   mitigant (track record) because it has never run. The honest claim is not
   "the rule is safe" but "the rule makes override visible, and visibility is
   worth something but is not protection." Capture remains on the unsolved list.

