# MacroVision v97: first-binding corporate financing boundary
29 October 2026. Hypothetical model diagnostics; no empirical calibration, default probabilities or causal claim.

## Finding
All **230** corporate stops in v96's original 576-case actual grid first breach **firm cash nonnegativity only**. At that first stop none breaches real-capital nonnegativity, household equity nonnegativity or outstanding-equity existence. This is an observed first-binding constraint, not proof that the other constraints would never bind if the path continued.

Every one of those 230 attempted annual steps has **positive residual investment and positive net equity issuance cash**. Issuance is positive but insufficient together with opening firm cash. No baseline corporate stop has aggregate net equity redemption. Therefore the combined v77 error message was too broad to explain these particular stops.

The existing source calculations are unchanged. A Python return-frame observer records the attempted state before the existing early stop discards it; it does not relax constraints or continue financing-deficient paths.

## Exact identity and interpretation
The inherited model sets:
```
investment = nominal output - existing government goods - total consumption
next firm cash = opening firm cash - investment + net equity issuance cash
next real capital = opening real capital + investment / price
```
For the 230 original corporate stops:
```
investment > opening firm cash + net equity issuance cash
investment > 0
net equity issuance cash > 0
```
The first attempted cash deficit ranges **0.0522–26.8360 normalized nominal units**. These are annual first-stop gaps in model units, not currency amounts, cumulative funding needs, required credit lines, loss amounts or household per-capita values. Stop years range from 7 through 20.

V77's explicit financial counterpart rules include no corporate bank-loan margin and do not ensure that the residual investment quantity can actually be financed. An accounting residual investment rule is not an investment demand function, lending market or equilibrium clearing mechanism. The diagnostic exposes an incompatibility between the real-use residual and available corporate cash/issuance in those steps; it does not establish which alternative closure is correct.

## Existing quote-rule sensitivity
Before inspecting results, the diagnostic script specifies all four already-existing quote rules over the same v96 factor grid. Each row contains 576 actual paths plus 576 matched-policy controls, 4,608 paths overall.

| Existing quote rule | Actual complete | Corporate cash stops | Government financing stops | Complete pairs | Consumer-debt-free displaced preservation | Consumer-debt displaced preservation |
|---|---:|---:|---:|---:|---:|---:|
| Constant real (v96 baseline) | 202 | 230 | 144 | 42 | 83 | 75 |
| Fixed nominal | 200 | 232 | 144 | 40 | 84 | 73 |
| Profit payout multiple | 200 | 232 | 144 | 40 | 81 | 76 |
| 50% real drawdown schedule | 211 | 221 | 144 | 44 | 89 | 80 |

All corporate stops in every tested rule bind cash only at the first stop. Every rule has 72 completed controls. Preservation retains v96's >=90% own initial nonhousing consumption in every annual observation, conditional on actual completion; denominators differ by row. Rules are inherited hypothetical valuations, not observed returns or estimated market clearing prices.

Relative to baseline, the drawdown rule completes **12 formerly failing actual cases but makes 3 formerly completing cases fail**, a net +9. Fixed nominal and profit multiple each lose two previously completing cases and rescue none. Thus even the highest completion total leaves 221 corporate stops and 144 government stops.

Do not infer that falling asset prices improve welfare: quotes change how many ownership units are issued/traded and future dividend distributions, and alter which paths complete. Cross-rule preservation totals are not comparisons on a common completed sample. No welfare ranking or probability is claimed.

## Reproduction and validation
```bash
python scripts/financing_diagnostic_v97.py
python -m unittest discover -s scripts -p test_financing_diagnostic_v97.py -v
```
The generator creates full local case data at data/processed/financing_diagnostic_v97.json and checked-in aggregate output at data/processed/financing_diagnostic_summary_v97.json. SHA-256 hashes pin the observer and every inherited model dependency. No participant data or network calls are needed.

Checks completed:
- Two tests verify observer output exactly matches the unobserved source model and restores any prior profiler; the captured cash identity holds.
- Government-financing failures are not mislabeled corporate failures.
- All **1,152 baseline compact actual/control paths** reproduce v96 exactly, case by case.
- Completion/failure counts reconcile to 576 for every rule.
- Each corporate candidate state's cash decomposition is checked at runtime.

The observer uses Python profiling only for its own synchronous run and restores the prior profiler in a finally block. It is a research CLI, not a production-server hook. First-stop diagnostics are never presented as completed year-20 outcomes.

## Next research decision
Prioritize an explicit corporate financing/investment closure, rather than adding unrestricted money or selecting a favorable quote rule. A valid alternative must specify all counterparties and preserve money and balance-sheet identities.

The next specification should distinguish:
1. A funding-limited investment rule, where unfunded residual output cannot silently disappear: consumption, inventories, output or prices need an explicit counterpart.
2. A corporate credit channel with a defined lender, cash transfer, loan claim, servicing, credit limits and loss allocation.
3. External finance, only if an explicit external sector and its claims/flows are modeled.

No alternative is implemented or endorsed here. An unbacked cash top-up would change wealth/money and would not be a mere numerical repair. Freeze one closed alternative and its accounting tests before running comparisons. Retain original early stops as the benchmark and continue reporting all government financing failures separately.

## Implication for the thesis
The earlier grid's 230 corporate truncations do not demonstrate capital exhaustion or corporate default. They demonstrate that the current residual-investment and financing rules cannot both be maintained in those attempted steps. Liquidity, debt and asset ownership remain relevant channels in the model, but broader claims about sustainable consumption under 20% unemployment require a defensible closure and external empirical evidence.

References: [v96 sensitivity](MACRO_SENSITIVITY_V96.md), source scripts v71–77 with hashes in the aggregate output. All parameters and scenario selections remain illustrative.