Model archive: Who pays for the quiet hour? A formal model of how residential regimes allocate sleep harm

Institutional rules are usually treated as settings that change how much sleep people get. This paper treats them as allocation rules that change who loses sleep. Three governance regimes of compulsory co-residence are modelled: autonomy (no institutional rule; the room's quiet hour is the outcome of bargaining), cutoff (room power is cut at a fixed hour) and separation (noise-generating activity is relocated to designated rooms). A public census of 132 Chinese institutions shows the regimes are not stylisations: 43.5% impose a nightly cutoff (median 23:00), 68.1% a night-time curfew and 19.6% neither. In an agent-based model of 800 occupants across 40 buildings, run for 200 replications, the regimes rank on total harm in the order autonomy > cutoff > separation (mean nightly sleep loss 1.05, 0.34 and 0.03 h) and in the reverse order on its concentration (Gini 0.61, 0.69, 0.89). The rules also produce three shapes of disadvantage: a predictable gradient on chronotype under autonomy and the cutoff (the earliest-sleeping quartile bears 58% and 73% of all loss), and a lottery on room location under separation (the worst-affected tenth of rooms bear 76%). Perceived fairness nevertheless ranks separation highest: the judgment tracks whether the arrangement has an addressee, not how the harm is spread. Two methodological results follow: adaptation and exit jointly remove 0.38 h of measured loss from the regime with the highest harm, so cross-sectional means are regime-specifically selected; and a 0.34 h difference in sleep projects onto a composite sleep-quality index as a contrast indistinguishable from zero, which may explain the literature's small and inconsistent dormitory effects. The autonomy penalty is shown to be a bargaining failure rather than rule absence: under full compromise it falls from 0.99 to 0.29 h. The paper closes with falsification conditions and the field test that would adjudicate them. This deposit contains the complete model archive accompanying the manuscript "Who Pays for the Quiet Hour? A Formal Model of How Residential Regimes Allocate Sleep Harm" submitted to the Journal of Artificial Societies and Social Simulation (reference 2026:407:1): the agent-based model source code, ODD documentation, the coded institutional census, and all scripts that generate every table and figure. Re-running step14_analysis.py 200 reproduces the reported results from the seeds given in the manuscript.