# Stage 118 — A recurring-income collection ceiling

## Result

Of **324 cases**, **111 pass** and **213 fail** the four collection, payment, monthly-consumption and annual-consumption gates. In this grid the classifications coincide; the gates remain separately recorded.

With unchanged outside income and zero opening owner cash, a 5% collection ceiling fails, while 10%, 20% and 100% ceilings pass the tested floors and owner responses across all three bases. These are tested ceilings, not an exact minimum rate or a recommendation.

| Owner outside income retained | Dividend-only base | Wages + dividends | All recurring receipts |
| --- | --- | --- | --- |
| 100% | 10%, 20%, 100% pass all floors | Same | Same |
| 10% | Only 100% passes all floors | Same | Same |
| 5% | None pass | None pass | 100% passes the 90% floor only |

Each row also tests owner responses q=0, 0.5 and 1. A 100% ceiling permits collection up to the eligible month's receipts; it does not impose a flat 100% tax every month. Actual collection is capped by the outstanding assessment and available cash.

## Bases and cash rule

The isolated stage-116 engine variant reconstructs these annual owner income components from the exact source model: wages, dividends and government interest. Their sum matches the owner's frozen outside income. Actual private-interest receipts enter the broad base separately in the monthly engine.

The three bases are:

1. Dividends.
2. Wages plus dividends.
3. Wages, dividends, public interest and actual private-interest receipts.

All are measured after inherited wage/profit taxes and before maintenance and the supplemental levy. They are experiment definitions, not legal tax bases or claims about real tax policy.

Each month, collected tax is the minimum of outstanding assessment, the declared ceiling times eligible current receipts, and available owner cash. Unused monthly capacity expires. Tax and transfer arrears carry forward; the treasury cannot borrow or receive a top-up.

Opening cash, government principal redemption and private principal repayment are excluded from every eligible base. Those cash flows can coexist in the owner's account, but they cannot expand the numerical collection allowance. A dedicated test supplies a large cash balance and public principal with zero dividends: dividend-based tax remains zero.

## Owner resource audit

Undiscounted twenty-year sums in source-model nominal units:

| Receipt | Amount | Included in broad recurring base? |
| --- | ---: | --- |
| Wages | 96.0000 | Yes |
| Dividends | 639.9722 | Yes |
| Public interest | 36.0000 | Yes |
| Private interest | 14.4266 | Yes, actual monthly receipt |
| Public principal | 0.0000 | No |
| Private principal | 3.0554 | No |

Dividends account for about 81% of these recurring receipts in the original unstressed path. Public principal, although left unshocked in stage 117, is actually zero on this path. The remaining private principal is an asset exchange, not recurring income.

## Experiment and limitations

Three existing supplement schedules target 90%, 95% or 100% of original control-year-1 recipient consumption. The four ceilings are 5%, 10%, 20% and 100%. Owner outside income retains 100%, 10% or 5%; opening owner cash is zero in all policy cases. Private-interest receipts are not shocked. Recipient references and schedules remain unchanged.

The behavioral q rule still responds to the full accrued assessment even when collection is capped. This preserves the inherited assumption; unpaid assessment is not silently forgiven or treated as actual tax paid.

The caps are tighter than a broad cash-affordability test. Their success verifies only the incremental supplement account under the stated fixed-source assumptions. Existing government financing, recipient income, prices, dividends and purchase plans do not respond to redistribution. No empirical calibration, legal tax interpretation or full macroeconomic sustainability is claimed.

## Validation and reproduction

Six tests cover source-income reconstruction, uncapped recovery of the original engine, exclusion of cash and principal, monthly cap compliance and complete cash replay, all grid/arrears identities and archive reproduction. Earlier simulator files were left unchanged.

```sh
python scripts/income_capped_transfer_v118.py --output data/processed/income_capped_transfer_v118.json
python scripts/test_income_capped_transfer_v118.py
```

[Resource audit and all cases](data/processed/income_capped_transfer_v118.json) retain bases, ceilings, separate gates, balances and annual eligible income. Full monthly tapes and counterpart transactions are reproducible with run(). Repository CI remains the merge gate.

## Research implication

The earlier broad-resource margin is not equivalent to a low recurring-income tax ceiling. More importantly, the base is dominated by dividends that currently remain frozen after the tax and transfer intervention. The next priority is to connect the supplement experiment to investment, corporate cash and future dividends, with both policy and matched control recomputed. A conditional financing result should not be treated as evidence that the underlying income base sustains itself.