# Retained earnings and corporate investment finance — v124

This is a **scenario experiment**, not an empirically calibrated forecast.

## Accounting contract
For year t > 0, positive after-tax potential payouts P are split into retained earnings R = s max(P,0) and household dividends D = P-R. Firm cash increases by R, household cash income decreases by the same R. Retained earnings are not new credit, government transfers or created money. The equity price rule is re-evaluated using actual distributed dividends, which may cause liquidation failure at 100% retention for profit-multiple pricing. The goods-constrained investment and corporate-financing guard of v123 remain active.

## Protocol
Retention s: 0%, 25%, 50%, 75%, 100%. Replay 324 original scenario combinations for each s (1,620 paths); compare within matched alpha, investment rule, floor, owner response and collection ceiling. The zero-retention case is the reference. Do not interpret counts as probabilities. Measure completion, qualified consumption floors and fully funded transfers, firm liquidity, capital, and minimum consumption retention. Keep failure paths separate from completed paths; never compare year-20 outcomes when a path failed earlier.

## Limitations
Prices and employment remain scenario-driven; capital-output feedback is inherited from v74, without depreciation. No debt or dividend contractual restrictions are modeled. The treatment changes owner dividends and equity pricing; it is not a causal estimate. The output JSON is produced by running the script and must be reviewed before claims of economic success.

Reproduce: `python scripts/retained_earnings_v124.py --output data/processed/retained_earnings_v124.json`

## Executed results (GitHub Actions run 361, commit 61b9bda62ca6237d39624ca355f91915358e5987)

The full 1,620-path grid completed in CI. Each retention rate contains 324 matched scenarios.

| Retention | Completed | Qualified | Failed |
| --- | ---: | ---: | ---: |
| 0% | 297 | 267 | 27 |
| 25% | 270 | 210 | 54 |
| 50% | 270 | 171 | 54 |
| 75% | 270 | 111 | 54 |
| 100% | 270 | 54 | 54 |

Overall: 1,377 completed, 813 qualified. Maximum absolute money identity residual across groups is 2.14e-13; goods residual is at most 7.11e-15. These are numerical conservation checks, not empirical validity.

All failures in this output are categorized under the combined guard "Corporate liquidity, real capital or equity-redemption capacity exhausted"; the guard does not distinguish cash shortages from redemption or capital constraints. Consequently it would be incorrect to claim all 54 are strictly corporate financing failures without a finer decomposition. The new grid shows a worsening in completion and consumption-floor qualification with retention; it does not establish causal mechanisms or general equilibrium welfare effects. Especially at 100% retention, the equity quotation rule may amplify changes in liquidity.

**Research conclusion:** Reject the unconditional hypothesis that corporate profit retention by itself resolves the v123 funding gap or preserves household purchasing power under this model. Next: decompose failure guards and test a financing contract with explicit equity valuation and payout expectations before making policy claims.

Source: GitHub Actions run 361, research job 124336788119, artifact retained-earnings-v124 (ID 13015857167).