מסמך טכני מיומן המחקר · באנגלית, כפי שנכתב בשלב העבודה
Stage 126 — corporate cash-flow attribution
Instrument the exact corporate cash identity at every committed annual row: opening cash after corporate debt service + retained payout + net household equity subscriptions - total investment outflow = cash before corporate credit. Keep the stage 125 stopping guard unchanged.
The last committed year is not the failed transaction year. Failure-year candidate cash must be audited separately before asserting an explanation for the stop. This initial instrument identifies the composition and path dependence of completed-year cash and sets up matched comparisons between retention rates.
Replay all 1,620 paths. No empirical calibration. Do not interpret scenario counts as real-world probabilities.
Failure-year diagnostic
The attempted but uncommitted cash transaction is now recorded on the failure object. The same accounting identity is preserved, with an explicit post-credit candidate balance. The failed year's household consumption, equity subscriptions, and investment are candidate plans, not realized flows; they must not be added to committed annual ledgers.
Comparisons across retention rates must match scenario keys and use failure-year candidates only to explain the guard, not as realized economic outcomes. A cash shortfall is a proximate cause; differences in spending and stock subscriptions are possible mechanisms requiring matched-path decomposition.
Executed results — GitHub Actions run 379
The full grid passed CI, including the failure-year bridge. Every failed path has an attempted, uncommitted transaction recorded.
| Retention | Failed paths | Attempts recorded | Sum of attempted negative closing cash (model units) |
|---|---|---|---|
| 0% | 27 | 27 | 302.965915 |
| 25% | 54 | 54 | 220.200406 |
| 50% | 54 | 54 | 349.231520 |
| 75% | 54 | 54 | 768.842329 |
| 100% | 54 | 54 | 606.700304 |
The sum is conditional on which paths fail under each policy, so it is not a matched causal effect and is not directly comparable as severity across retention rates. The failure-year amounts are candidate cash deficits, not realized monetary losses. In particular, a lower aggregate gap at 25% retention despite more stops does not imply greater economic resilience. This stage establishes cash accounting and stop magnitudes; the next stage must compare matched scenarios and decompose retained earnings, household equity flows, and investment outlays.