# V106 — First annual-to-monthly cohort bridge

29 October 2026. Controlled replay of a frozen annual path through the v105 cash ledger. This is a first adapter, not a fully monthly macro model or empirical calibration.

## Scope and reference
Source is v100's default 20-year transition, concentrated ownership, income-share saving, owner-free/stable housing, zero corporate credit and zero workout trigger. The unchanged source path completes. Only the displaced_free cohort (mass .10) is replayed; this cohort has no personal loan or mortgage but can accumulate shares and has maintenance expenses.

The source implementation is scripts/corporate_workout_model_v100.py, blob 19df03e2e2794620fa8540b520cd80d0400877cb, present at main 7ddc2d549523fc01b4324d20684b6eccb8138dab. The output fingerprints imported local research modules and the adapter with SHA-256 for reproducibility.

Year zero supplies opening cash, equity units and consumption reference only. Flow years 1–20 become 240 artificial months, labeled 0001-01 through 0020-12. These are simulation labels, not historical observations or an inferred payment calendar.

## What is reconciled
Per-person nominal income and consumption are recovered by multiplying source real values by cohort mass and the annual price level. Source equity units are already aggregate, so they are not multiplied by mass again.

Wages use the source cohort employment rule; dividends use beginning equity holdings and the source after-tax payout. In this restricted cohort/tenure there are no lender, government-interest or landlord receipts. Transfer income is the remaining source income after wages and dividends; nonnegative and total-benefit bounds are checked. Broader configurations, incomplete paths and source liquidation are rejected instead of silently netting incompatible flows.

The reference replay pays wages/dividends and transfers uniformly, spreads target nonhousing consumption and maintenance uniformly and executes the source equity purchases in December after spending. The desired spending path is the source's realized annual consumption, not a newly estimated unconstrained demand curve. Actual nonhousing spending is constrained by cash after maintenance.

Each month invokes v105 with kernel [0]. Source transfers are already included in the spending target, so no second transfer response is added. The wrapper separates maintenance from nonhousing consumption and adds source equity purchases with an explicit market cash counterparty. It carries cash and share units across months; no asset sale or new credit is activated to cover a shortfall.

For the source opening-liquidity reference, annual nonhousing consumption, household closing cash and equity units reproduce the original path within 6.2e-15 in their respective normalized units. The largest cash-conservation discrepancy across the four replay cases is below 3e-13. Numerical reconstruction checks the adapter; it is not independent validation of the annual model.

## Timing experiment
All cases retain the same annual benefits, other income, prices, desired spending and scheduled equity purchases. Within each opening-liquidity pair, only benefit receipt timing changes. The low-liquidity pair starts with zero household cash instead of the source opening cash; other opening accounts remain unchanged. That changes the initial endowment, so it is not a wealth-preserving causal liquidity experiment.

| Opening cash | Benefit timing | Months with unmet desired nonhousing consumption | Months below 90% reference |
|---|---|---:|---:|
| Source opening cash | Uniform monthly | 0 | 168 |
| Source opening cash | December | 11 | 170 |
| Zero household cash | Uniform monthly | 0 | 168 |
| Zero household cash | December | 39 | 185 |

Each denominator is 240 modeled months. Counts are deterministic outcomes of four declared cases, not frequencies over households or estimated probabilities. The annual benefit total is unchanged in each timing pair.

The first shortfall measure compares actual spending to that month's frozen target. The second compares actual monthly real nonhousing spending with .90 times source year-zero annual real nonhousing consumption divided by twelve. These are distinct: the default annual path already has many months below the preservation benchmark even without a timing-induced cash shortage. No claim is made that all eleven new shortfall months cause a benchmark crossing.

The total missed nominal target is about .9793 aggregate normalized units in the source-cash December case and 1.9811 in its zero-cash counterpart, over the full twenty years. These are not currency amounts or welfare losses. Unmet desired spending is not automatically repurchased later; resulting cash remains a stock.

## Funding and feedback limits
Treasury starts with a reserved envelope equal to this cohort's entire source-path benefit bill. The external market account starts with enough cash to pay its frozen wages/dividends and receives maintenance, consumption and equity-purchase cash. This explicit synthetic funding avoids cash creation but does **not** reproduce the original treasury's annual balances, monthly tax collections, firm liquidity, investment or equity-issuer financial statements.

Future income/dividends and desired equity purchases remain frozen even if the timing experiment changes holdings or cash. Thus this is a partial-equilibrium payment-feasibility diagnostic. It cannot establish full-economy fiscal feasibility or a self-consistent new equilibrium, and it does not replace historical v100/v101 scenario results.

## Tests, artifact and next gate
Eleven new tests passed alongside v105's fifteen. The archived output is checked against a fresh build, including source fingerprints. They verify annual consumption/cash/holdings recovery, transfer-total equality across timing pairs, monthly counterpart reconciliation, nonnegative cash, calendar continuity, source units, monthly deflation, rejected unsupported/incomplete sources and archive coverage. Both suites run in research CI.

The public JSON archives all annual comparisons plus January/December and every constrained month. Full monthly rows and transaction tapes are reproducible through build(); they are not all included in the compact archive.

Run:
```
python scripts/test_monthly_transfer_v105.py
python scripts/test_annual_monthly_bridge_v106.py
python scripts/annual_monthly_bridge_v106.py --output data/processed/annual_monthly_bridge_v106.json
```

Next gate: replace the reserved envelope with explicit monthly treasury tax/spending flows and reconcile their annual totals before generalizing the timing result to fiscal feasibility. Debt-bearing cohorts, housing/asset liquidation and firm feedback need further adapters and checks. Empirical calibration remains closed, and the restricted-data access route remains unavailable.

[Results and source fingerprints](data/processed/annual_monthly_bridge_v106.json) · [V105 ledger](MONTHLY_TRANSFER_LEDGER_V105.md) · [V104 measurement contract](MONTHLY_TRANSFER_CONTRACT_V104.md)