A public-interest project. Not a company, and it sells nothing. What that means
autonomist

Results: the engine model

Generated 2026-11-20 by model/run.mjs from model/lec-core.cjs. Every input is a SCENARIO unless the code cites a source. The worked building is an 8-unit listing at a 6.27% cap rate. This is a mechanism model, not a forecast: it shows which design choices move risk and wealth, and by roughly how much. The tables regenerate on every run; the prose between them was written against these numbers, so reread it after changing an input.

1. One co-op, default governance

Stack: 70% senior at 5% over 30 years (repriced at year 10), 20% junior at 2% interest-only, 10% member equity ($31,400 a home). Members get the paydown (no appreciation), the share is indexed at 3%, 5% of market rent is saved (half to the federation pool), one home in seven turns over a year, and refinancing to pay leavers is capped at 75% of value (senior debt only). Rents and costs grow 3% a year.

yearvaluedebtmarket rent /mocharge /moaverage accountreserve
1$2,472,000$2,238,114$2,691$2,558$35,178$78,458
3$2,622,545$2,198,887$2,855$2,641$42,300$91,961
5$2,782,258$2,165,891$3,029$2,738$49,065$106,287
7$2,951,697$2,137,230$3,213$2,848$55,691$121,485
10$3,225,399$2,099,177$3,511$3,038$65,760$146,035
15$3,739,122$2,040,030$4,070$3,424$84,085$192,114

A member who joins at the start holds $68,345 after 7 years and $139,632 after 15.

2. Interest-rate risk at the senior reset (year 10)

rate at resetcharge /mo, year 10charge /mo, year 11market rent /mo, year 11
5.0%$3,038$3,108$3,616
6.5%$3,038$3,271$3,616
8.0%$3,038$3,443$3,616

Long fixed terms remove this risk; a 10-year reset is the conservative case.

3. A member with options: co-op vs renting vs a comparable condo

Same starting wealth (a 20% down payment plus 3% closing on a $350,000 condo, the sibling building's 2023 sale) and the same yearly budget; what a path does not spend on housing is invested. Condo: 6.5% 30-year mortgage, Norfolk's $1.23 per $100 tax (City FAQ, effective 2024-07-01), $450/mo fee, 1% maintenance, 7% to sell. Co-op: the run above. No income taxes modeled. Net worth at the horizon:

rent growthvalue growth (a year)portfolio returnhorizonrentercondo ownerco-op member
1%-0.5%6%7 years$148,171$62,379$173,671
1%-0.7%6%15 years$254,354$90,138$325,699
1%-0.5%8%7 years$166,921$62,379$186,218
1%-0.7%8%15 years$328,709$90,138$377,559
3%3.0%6%7 years$134,954$148,352$180,449
3%3.0%6%15 years$215,096$340,000$379,380
3%3.0%8%7 years$153,126$148,352$193,249
3%3.0%8%15 years$283,425$341,854$436,022
5%6.2%6%7 years$128,345$252,618$195,756
5%6.1%6%15 years$204,562$700,089$504,494
5%6.2%8%7 years$146,042$252,751$208,953
5%6.1%8%15 years$270,313$708,250$571,986

Read across a row: the co-op beats renting everywhere here. It beats the condo when values grow slowly or moderately, and loses to it when values boom, because the condo owner holds the appreciation with leverage.

4. The federation

24 co-ops of 6–24 homes, bought at cap rates of 5–7% only if they underwrite (DSCR ≥ 1.15), held for 15 years, 400 runs a case. Each co-op draws its own governance from a menu (senior share 60/70/80%, savings 3/5/8% of rent, pooled share 0–100%, appreciation credited 0/10/25%, paydown credited 50/100%). Shocks: interest rates wander (mean-reverting around 5%); a regional rent drop hits in 7% of years and recovers over three; each co-op faces repair bills of 1–8% of value (6% of years) and vacancy spells (5% of years). A shortfall is met by the co-op's reserve, then a one-off assessment on members (up to 10% of a year's charges), then refinancing within 75% of value, then the pool, or the co-op fails and its lender takes the building.

4.1 Architecture under stress

casehomes lost (mean ± s.e.)any co-op failsa quarter or more failworst run
base, isolated2.3% (±0.2%)36%0.3%27%
base, firewalled pool1.1% (±0.1%)23%0.0%16%
base, merged0.0% (±0.0%)0%0.0%0%
moderate, isolated11.6% (±0.9%)73%13.0%100%
moderate, firewalled pool5.3% (±0.6%)53%3.8%100%
moderate, merged3.8% (±1.0%)4%3.8%100%
severe, isolated15.2% (±1.3%)63%17.3%100%
severe, firewalled pool8.9% (±1.1%)44%7.2%100%
severe, merged9.0% (±1.4%)9%9.0%100%

Merged fails least often, but when it fails, it fails whole. The firewalled pool fails more often, in pieces.

4.2 One city or several (severe stress)

casehomes lost (mean ± s.e.)any co-op failsa quarter or more failworst run
one market, isolated15.2% (±1.3%)63%17.3%100%
one market, firewalled pool8.9% (±1.1%)44%7.2%100%
one market, merged9.0% (±1.4%)9%9.0%100%
four markets, isolated15.5% (±0.8%)83%24.3%88%
four markets, firewalled pool6.9% (±0.5%)58%8.5%60%
four markets, merged10.0% (±1.5%)10%10.0%100%

A firewall stops contagion, not a shock that hits everyone at once. Spread across independent markets, the firewalled pool beats merging in the tail; merging gains nothing from spread, because one default still takes all.

4.3 The fluid dial: how much each co-op pools (firewalled)

casehomes lost (mean ± s.e.)any co-op failsa quarter or more failworst run
moderate, pooled 0%11.9% (±0.9%)72%13.5%100%
moderate, pooled 25%6.7% (±0.7%)56%5.5%100%
moderate, pooled 50%5.3% (±0.8%)40%4.3%100%
moderate, pooled 75%4.2% (±0.7%)32%3.5%100%
moderate, pooled 100%3.5% (±0.7%)25%3.0%100%
severe, pooled 0%17.2% (±1.4%)66%19.0%100%
severe, pooled 25%11.6% (±1.4%)44%10.5%100%
severe, pooled 50%10.0% (±1.4%)34%9.0%100%
severe, pooled 75%9.6% (±1.4%)26%9.0%100%
severe, pooled 100%9.3% (±1.4%)21%9.0%100%

The first quarter pooled does the most. The model does not price what autonomy buys (local control of reserves, room to experiment), so it will always favor pooling more; that judgment stays with each co-op.

4.4 What else moves risk (firewalled, moderate stress unless noted)

casehomes lost (mean ± s.e.)any co-op failsa quarter or more failworst run
senior debt 60% at purchase3.7% (±0.8%)17%3.5%100%
senior debt 70% at purchase5.7% (±0.7%)42%4.8%100%
senior debt 80% at purchase15.6% (±0.9%)93%13.5%100%
pool seeded with 0% of purchase cost6.7% (±0.7%)58%4.8%100%
pool seeded with 1% of purchase cost5.3% (±0.6%)53%3.8%100%
pool seeded with 3% of purchase cost4.8% (±0.5%)53%2.5%100%
severe; members stay up to 100% of market rent8.9% (±1.1%)44%7.2%100%
severe; members stay up to 110% of market rent5.4% (±0.8%)37%4.3%100%

Leverage at purchase matters most: the patient junior money is the resilience lever, not only a cheap one. Members who will pay a little above market through a downturn (their accounts give them a reason) cut losses by more than half.

4.5 Networked learning

Every three years each co-op compares itself with a random peer and, half the time, copies a better peer's savings, pooled share and member credits (5% of the time it tries a random one). "Survivors" sees only co-ops still alive. "Public" also sees the federation's ledger of failures by policy and discounts a policy by its failure rate times what members stand to lose.

casehomes lost (mean ± s.e.)any co-op failsa quarter or more failworst runaverage account ÷ buy-insavingspooledappreciation credited
moderate, none5.3% (±0.6%)53%3.8%100%2.72×5.3%51%12%
moderate, survivors5.1% (±0.6%)52%3.3%100%2.90×4.5%50%16%
moderate, public4.4% (±0.5%)50%1.8%100%2.90×4.6%50%16%
moderate, merged, survivors4.0% (±1.0%)4%4.0%100%2.81×4.3%96%16%
severe, none8.9% (±1.1%)44%7.2%100%2.69×5.1%49%11%
severe, survivors10.6% (±1.3%)43%9.3%100%2.79×4.2%47%15%
severe, public10.3% (±1.3%)42%8.8%100%2.76×4.1%47%15%
severe, merged, survivors8.5% (±1.4%)9%8.5%100%2.75×4.0%92%15%

Learning from success drifts toward member wealth: less saving, more appreciation credited. Under severe stress that drift raises losses. The public failure ledger helps a little under moderate stress and not clearly under severe; the differences sit within about one standard error. Failures here come mostly from luck and from terms set at purchase, which co-ops cannot copy later, and 24 co-ops give the ledger thin evidence. A real ledger would carry causes, not just counts.

5. What the model leaves out