What question did this study set out to answer?

The aim is to establish a standard that enhances the clarity and stability of institutional variables in research.

February 12, 2026Open Access

A Recomputable Institutional Disclosure Standard Making Platform Institutions Comparable, Identifiable, and Falsifiable

Key Points

The aim is to establish a standard that enhances the clarity and stability of institutional variables in research.
Development of a Minimal Verifiable Disclosure standard (MVD) and enhanced MVD+.
Creation of an Institution Identifiability Score (IIS) for cross-platform institutional comparison.
Introduction of a Minimal Research Template to upgrade empirical assumptions.
Validation through across-instance tests, ablation tests, and score calibration.
Implementation resulted in improved institutional variable stability and reproducing capabilities.
The identifiability score successfully categorized institutional sets across varying levels of identifiability.
Calibrated test results indicated the significance of maintaining finality and versioning for hypothesis testing.

Abstract

Institutional research has struggled to become a cumulative science. The same “named institution” can drift in meaning across platforms and versions (definitional drift); rules can be hot-patched and parameters tuned within the research window; and ex post rewriting plus non-verifiable enforcement makes the “institution-as-variable” unstable. As a result, identification often collapses into narrative plus weak, unverifiable assumptions. This paper proposes a Minimal Verifiable Disclosure standard—MVD (and an enhanced MVD+)—designed to restore recomputability and falsifiability to modern, mutable institutions. The standard is accompanied by: (i) an Institution Identifiability Score (IIS) that screens “identifiable sets” of institutions for cross-platform comparison; and (ii) a Minimal Research Template (template + locked specification) that upgrades familiar empirical assumptions (e.g., no-anticipation) into design-induced testable implications derived from institutional semantics. The core semantics are: finality (a verifiable end-of-history boundary), version bundles with timelocks (policy bundle + timelock), a recomputable hard budget redline (hard budget invariants), and OnlyDown monotone enforcement (one-direction verifiable consequences). Unlike declarative transparency manifestos, we provide a validation program: across-instance validation, ablation tests, and score calibration. We compare recompute-error distributions and pass rates across a strong–medium–weak identifiability gradient; we remove finality / OnlyDown / versioning to show where key implications become untestable or systematically fail; and we calibrate IIS bins against how many template tests (T1–T6) pass, fail, or become “not testable.” Our contribution is not a new estimator. It is to make the object—the institutional variable—stable, recomputable, and falsifiable in modern versioned environments, thereby moving institutional research from narrative comparison toward reproducible, scalable, falsifiable cumulative science (North, 1990; Williamson, 1985; Manski, 1995; Hart Tirole, 1999; Christensen Gelman & Loken, 2014).

KI fragen

Bookmark

View Full Paper