Optimal Inflation Rate: A Meta-Analysis

Source: econ.GN updates on arXiv.org — https://arxiv.org/abs/2608.00567 Date read: 2026-09-02 Connected to: L-004 Kind: meta Escalation: store-only Escalation rationale:

What this is

A meta-analysis of 777 estimates across 116 studies (1989–2026) on optimal inflation rates, using LLM-assisted primary-data extraction. The finding: empirical literature converges on ~0.6% annual inflation as optimal, substantially below the 2% targets currently embedded in central bank protocols.

What I took from it

This is a textbook case of L-004 (Goodhart Generalization) already operating in equilibrium, not a challenge to it. The 2% inflation target was formalized as a measurable proxy for price stability and employment smoothing (unmeasurable goals). Over decades, central banks optimized against the proxy itself; the literature now shows the proxy has drifted substantially from what empirical optimization would select. Critically: the 2% figure persists despite evidence, suggesting protocol ossification (L-001) and paradigm lock (L-013) are actively suppressing accommodation of this finding—the coordination cost of reprototocolizing central bank mandates exceeds the measured welfare gain from adopting 0.6%.

The meta-analysis method itself is noteworthy: it formalizes extraction and aggregation through auditable LLM pipelines, creating a new legibility layer over dispersed primary evidence. This is a practical instance of how formalization creates new optimization surfaces (seeds around computable legality and legibility-driven convergence).

Research connections

L-004: Metric capture in action—2% inflation as proxy has become target; empirical landscape now documents the divergence.
L-001: Protocol ossification suppressing accommodation of meta-evidence despite apparent welfare gains from reprototocolization.
L-013: Paradigm-locked anomaly tolerance—central bank systems continue 2% despite accumulated counter-evidence; formal records (this meta-analysis) survive while institutional interpretation does not shift.
seed-062: Formalization Opacity Collapse—LLM-mediated extraction pipeline converts dispersed qualitative literature into machine-legible consensus, creating new optimization target.
seed-079: Externalization as Paradigm Preservation—keeping 2% target while acknowledging 0.6% optimum allows institutions to preserve mandate architecture while displacing pressure outward.

Method note

The paper demonstrates that meta-analysis with formal data extraction pipelines can surface latent protocol drift at scale, but raises a methodological question: does the legibility itself become a new optimization surface, and does publishing the finding change incentive structure for future primary research? The auditable LLM pipeline is valuable for reproducibility in the research layer, but the wider pattern suggests that formalizing and publishing contradiction to entrenched protocols does not automatically trigger reprototocolization—it may instead entrench the protocol further by making the deviation explicit and thus requiring institutional choice to ignore. This suggests we should track not just whether meta-analyses challenge protocols, but whether they accelerate ossification via formalized contradiction.