# V107 — Cash taxes, government arrears and household payment feasibility

29 October 2026. Monthly fiscal settlement on the frozen v100 default path. No empirical calibration or endogenous macro feedback.

## What changed
V106 reserved twenty years of selected-cohort benefits at the start. V107 removes that benefit envelope. Treasury begins with the source opening cash (zero in this case) and receives only the cash taxes assigned to each month.

The adapter now settles **all** source government goods spending, government interest, benefits and scheduled bond principal, while replaying only the selected displaced-free, owner-free cohort's spending and equity purchases. Other benefit recipients are aggregated in a separate cash account. No government overdraft or new borrowing is permitted.

Annual reference: unchanged default v100 path, concentrated ownership, income-share saving, stable owner-free housing and zero corporate credit/workout. Source year zero supplies opening stocks; artificial simulation years 1–20 supply 240 months. These are model labels, not observed dates.

## Timing and payment rules
The experiment crosses three assumed tax-collection profiles (uniform monthly, quarter-end, December only) with two benefit due-date profiles (uniform monthly, December). Annual tax and obligation totals remain fixed.

Goods and interest become due uniformly each month. Scheduled source bond principal becomes due in December. Unpaid obligations are carried as explicit arrears, without added interest or penalties, and paid when cash arrives. Within a month the priority is:
1. outstanding and current goods obligations;
2. outstanding and current interest;
3. outstanding and current benefits, pro rata between selected and other recipients;
4. scheduled outstanding bond principal, capped by principal still owed.

This priority is a declared modeling rule, not an observed legal priority or an optimal policy. A benefit scheduled for December is not delinquent in January. Conversely, a monthly benefit unpaid until December is recorded as arrears during the intervening months.

The selected household receives its payment from the same treasury account through v105. The transfer is not duplicated into a separate prefunded household account. Source wages/dividends, prices, desired consumption, maintenance and equity-purchase plans remain frozen. Kernel [0] avoids adding an extra spending response to a target that already reflects source benefits.

## Results
Each row covers 240 modeled months, using the original household opening cash.

| Tax collection | Benefit due dates | Months with any government arrears | Months with benefit arrears | Months with selected-consumption shortfall |
|---|---|---:|---:|---:|
| Monthly | Monthly | 0 | 0 | 0 |
| Monthly | December | 0 | 0 | 11 |
| Quarter-end | Monthly | 160 | 160 | 0 |
| Quarter-end | December | 129 | 0 | 11 |
| December | Monthly | 220 | 220 | 11 |
| December | December | 220 | 0 | 11 |

All six cases reproduce source annual cash taxes, goods, interest, total benefits, bond principal, closing treasury cash and outstanding government bonds within 1.3e-12 normalized units. Arrears clear by each year end to numerical tolerance. Maximum monthly cash/bond identity error is below 1e-12. Material arrears and consumption-shortfall counts use tolerance 1e-9.

This is the substantive distinction for the thesis: annual fiscal feasibility can coexist with months of unpaid obligations. Household liquidity can sometimes absorb a payment delay, so benefit arrears need not immediately become a consumption shortfall. Equally, an on-schedule December benefit can leave earlier desired spending unfunded without ever being classified as delinquent.

A consumption shortfall compares realized spending to the frozen monthly target; it is not the v106 90%-of-reference criterion. Counts are deterministic outcomes of these profiles, not household frequencies, probabilities of default or causal estimates. No result establishes that automation benefits are sustainable.

## Accounting checks
Every cash movement has a payer and recipient. Treasury cash reconciles opening cash plus collected taxes minus paid goods, interest, benefits and principal. Outstanding government bonds match the holder's remaining principal claim. Each arrears category reconciles prior arrears plus new obligations minus cash paid. Bond redemption cannot exceed remaining principal.

The fully uniform case also reproduces selected-household annual consumption, cash and share holdings. A deliberately underfunded annual test retains unpaid obligations and fails reference reconciliation instead of inventing finance. A separate test prevents principal redemption beyond the outstanding claim.

Twelve new tests cover these identities, annual reference recovery, timing distinctions, payment priority/pro-rata allocation, nonnegative cash, calendar carry, archive coverage, invalid profiles and exact reproduction of archived results with current source fingerprints. Existing v105/v106 tests remain in CI.

## What remains outside the model
Taxpayers and the combined market account have explicit synthetic cash endowments sufficient for the frozen source receipts. This removes the treasury's prefunded benefit envelope, but does not derive monthly tax capacity from endogenous taxpayer production or consolidate the original firms' cash flows.

Unpaid public spending is delayed rather than assumed paid. Its effects on supplier solvency, services, labor income, output and prices are not modeled. Government interest obligations remain those of the annual reference; no penalty, rollover, credit repricing or interest-on-arrears model is supplied. The result is therefore a partial-equilibrium settlement diagnostic, not a complete monthly macro model.

Only one household cohort is replayed. Its future dividends and equity-purchase plan remain frozen even if timing alters actual holdings. Other households' consumption, private debt, mortgage service, defaults and asset liquidation remain outside this extension.

## Reproduction and next gate
```
python scripts/test_monthly_fiscal_bridge_v107.py
python scripts/monthly_fiscal_bridge_v107.py --output data/processed/monthly_fiscal_bridge_v107.json
```

The public artifact retains all annual comparisons, first-year months, every December and every selected-household consumption-shortfall month. Full monthly accounts and counterpart transactions are available by rerunning build(). Source-module SHA-256 fingerprints are included; original annual simulation rules were not edited.

The next controlled comparison should add a defined, explicitly financed treasury cash buffer and record the counterparty and financing cost. It must show whether timely payments improve without hiding the cost in an assumed initial endowment. Only after that should wider household debt and firm feedback support broader sustainability claims. No restricted records or new empirical estimates were used.

[Results and fingerprints](data/processed/monthly_fiscal_bridge_v107.json) · [V106 annual bridge](ANNUAL_MONTHLY_BRIDGE_V106.md) · [V105 monthly ledger](MONTHLY_TRANSFER_LEDGER_V105.md)