System comparison — final-year results Baseline/CCO-only top tier = Comfortable; CCO+PTF = Flourishing
| Configuration | Poverty elim. | Median wealth | Gini | Median BLEI | Avg EDC | Stability |
|---|---|---|---|---|---|---|
| Run the simulation to see results. | ||||||
▶ 📋 Model Assumptions & Calibration Notes v3.1 — click to expand
📊 Empirical Inputs — sourced from real-world dataBase daily cost $68.33/day and CCO+PTH daily cost $31.67/day — US Bureau of Labor Statistics Consumer Expenditure Survey 2023. Components: rent/PTH + food + utilities + transport, divided by 30 days.
62–72% extractive drain coefficient for near-poverty US households — US CES 2023 rent-to-income and consumer debt ratios. Nordic Gini benchmark 0.28 — OECD 2023.
Starting wealth lognormal(10.5, 1.2) calibrated loosely against Fed SCF 2022 net-worth distribution for non-homeowners (median ≈ $36K). PTF ε range 1.50–3.00 grounded in Mondragon Eroski cooperative pricing data (Whyte & Whyte 1991).
r_a = 4%/yr base (range 2–7%). Calibrated against Community Land Trust long-run data (Davis & Demetrowitz 2003; Theodos et al. 2017 — 70% lower foreclosure rate vs. conventional). PTH payment 100% generative (zero EDC), consistent with Burlington CLT documentation.
Calibrated to simulation unit scale, not directly US CES ($60–72K BLS 2023). The operative ratio is wage×12 / ANNUAL_BASE_COST: <1 for baseline agents (wealth depletion), >1 for full CCO participants (accumulation). See Unit System note below.
PTF 18% share and target wealth $82K are framework specifications under active revision. An open goal of this simulation is to identify optimal CCO parameters — analogous to a structural engineer ensuring load capacity across all points simultaneously. Some earlier calibrations (e.g. PTF share ~30%) pre-date the v3.0 expense system and may need updating.
Two full years of basic living covered — capital commitment horizon required for entrepreneurial ventures to be viable without first-year break-even (BLEI §4). Flourishing requires CCO+PTF both active (generative system maturation); without them the top tier displays as "Comfortable."
Wealth poverty line $25,000 (nominal asset threshold — a retiree with $24K savings and $80K pension would incorrectly register as poor). BLEI poverty (% agents with BLEI < 30 days = Crisis+Precarious) is more theoretically grounded per BLEI §4 and is shown as a separate KPI. Both are displayed for multi-dimensional poverty assessment.
BLEI = cashDays + incomeDays + benefitDays = (liquid + γ·wage + buFood) / dailyCost. Three conceptually distinct buffers: stock (liquid savings), flow (income), and benefit (BU food basket). v3.1: income buffer uses actual agent wage, not wealth/60 proxy. γ = 0.12 baseline → 0.20 with established CCO floor.
Wages grow annually: 1% base + 0.8% BLEI stability premium + 0.3%/octave + CIP bonus. v3.1 AI automation: each agent has a heterogeneous automationRisk ∈ [0.2, 1.0] drawn at initialization — displacement = populationRate × automationRisk, creating realistic inequality effects (high-risk occupations affected more). Duke's note: labour markets cannot reabsorb past 30% displacement.
PTF modelled as cost reduction factor (not wealth injection — v3.0 fix for ex nihilo money creation). PTH covers housing component (35% of costs) and builds Acre Equity separately — no double-counting. CCO food basket reduces food component (20%). PTF has dynamic adoption: agents in economic distress or via diffusion join over time (v3.1).
Recessions: population-level, 10% annual probability, 2–4 year duration, 65–85% income multiplier (v3.1: renamed from "severity" — 0.65 means 65% of normal income, i.e. 35% loss). v3.1: agents can accumulate debt to −$10,000 floor (was $100 minimum, eliminating debt/bankruptcy). FBS gating per Mullainathan & Shafir (2013): advancement infeasible at FBS ≈ 0.
Ablation analysis estimates each system's BLEI contribution via removal (200-agent subsample). "Run 3×" button shows mean ± SD across three consecutive seeds. These are scenario comparison estimates, not formal Shapley values or Sobol indices — see CONTRIBUTING.md for rigorous sensitivity methodology (Latin hypercube, Sobol).
Annual costs compound at set inflation rate. PTF presence dampens effective inflation (community pricing stability; up to 50% of PTF share); PTH dampens housing-component inflation (10% reduction). US Baseline preset applies 3% to reflect historical CPI. Compassionism presets default to 0% — the price-stability hypothesis of cooperative essential goods provision.
wealth: nominal USD. Median ≈ $36K from lognormal(10.5, 1.2), calibrated against Fed SCF 2022.
wage: Simulation Income Units (SIU), not USD. Median ≈ 33 SIU/month from lognormal(3.5, 0.5). Annual income ≈ 396 SIU/yr, intentionally below ANNUAL_BASE_COST ($1,500) so baseline agents deplete savings over time — demonstrating why CCO intervention is needed. Do not interpret wage values as US median income.
ANNUAL_BASE_COST: calibrated to simulation scale, not directly US CES. The ratio wage×12 / ANNUAL_BASE_COST is the operative quantity (<1 = cost pressure; >1 = accumulation). This calibrated unit approach is standard in ABM (Epstein & Axtell 1996; Farmer & Foley 2009): parameters reproduce emergent dynamics of interest rather than matching absolute real-world values. Formal dimensional consistency is a future development direction noted for CONTRIBUTING.md.