The Measurement Trap: How Backward‑Looking Measurements Destabilize the Systems They Were Designed to Protect

Abstract

Institutions routinely make consequential decisions from backward-looking measurements — smoothed averages, trend estimates, and filtered signals meant to separate noise from signal. This paper asks what happens when such a measurement feeds back into the system it measures, and derives a closed-form stability criterion for that loop: a policy that regulates a persistent variable by responding to a uniform trailing average of it is unstable exactly when bg(1 − φ^W)/(1 − φ) exceeds π²/2, where φ is the variable's persistence, W the averaging window, and bg the feedback strength. The constant π²/2 arises from a half-wavelength resonance at a sharp Neimark–Sacker bifurcation: below the boundary, measurement errors decay geometrically and the policy achieves its aim; above it, they grow geometrically, and the policy delivers the opposite of its aim, with no intermediate regime. We call this the Quantitative Lucas Criterion — it turns Lucas's qualitative warning that a policy rule alters the process it is calibrated against into a computable boundary for the backward-looking measurement rules institutions actually use. A mathematically equivalent boundary was derived independently in mathematical biology for continuous-time delay differential equations with uniform delay kernels (Campbell and Jessop 2009); the correspondence is treated as author-verified pending expert review. Held to a single pre-registered specification across six institutional domains, the criterion places the Basel III countercyclical capital buffer — which estimates the credit cycle with a one-sided Hodrick–Prescott filter over an effective twenty-year window — on the unstable side of the boundary for 42 of the 44 economies with sufficient BIS data. The instability is slow (a spectral radius near 1.008, an emergent oscillation on the order of thirty years), so it is invisible against ordinary fluctuation over a few years but structural over decades: a rule designed to dampen the credit cycle is, under this analysis, calibrated to amplify it. The paper reports what the data support and what they do not — including domains where a pre-registered materiality bound cannot exclude a genuine cost — and derives a design principle: for any window and persistence, there is a maximum feedback intensity beyond which the policy is self-defeating, so robust institutional design uses the shortest feasible window and the weakest feedback consistent with the intended countercyclicality. Verified rebuild under the Research-to-Publication Standard v1.9.1: every load-bearing number is registered in a machine-checked ledger (claims.lock) and regenerated on demand by verify.py on hash-pinned inputs, then independently reproduced from a clean checkout; a capped single-round adversarial review under a fix-or-rebut protocol and a public corrections log are committed in the repository. A dated, reader-runnable forward prediction is registered publicly as a prospective out-of-sample falsifier — that at the onset of the next credit contraction affecting three or more G7 economies simultaneously, no G7 economy will hold an effective countercyclical capital buffer at or above 2.0 percent of risk-weighted assets — to be resolved on finalized Bank for International Settlements data at the contraction onset, and carried forward if no qualifying contraction occurs before July 2031. The paper is licensed CC BY-NC-ND 4.0; the accompanying plain-English companion is CC BY-NC 4.0, and the reconstruction and verification code is MIT-licensed.