# Step 300: conditional demand support based on lagged investment and debt

The new rule requests support equal to 50% of the previous year's underlying demand gap, capped at 5% of current output. It uses the existing reconstruction of the gap as observed shortfall plus requested support in the previous model year. It requires a decrease in realized real capital investment between the two preceding completed investment years and positive room under the 90% net debt ceiling at the end of the previous year, using that year's output. Year zero is initialization, so the rule can first activate in year three. The debt test is a lagged trigger. It does not guarantee debt compliance later. The model uses real capital formation rather than the accounting residual before demand adjustment.

Compare no support, 50% support at any gap, 50% support when the effective lower bound binds, and the conditional rule. Use 432 matched cases per cell, four investment decays, two interest sensitivities and shared levy ceilings of 10%, 15% and 20% (41,472 simulations). Each case uses the same financing ceiling as its no-support comparison.

The real benefit shortfall is calculated entitlement less payment, divided by the model goods price and the cohort mass. The outcome sums the annual shortfalls after the initialization year for each displaced cohort separately. Changes use matched completed runs. All 432 pairs completed in each cell. A decrease is an improvement relative to the model entitlement. The entitlement itself may change with policy.

Results at a 15% ceiling and interest sensitivity 0.60, counts out of 432. The last column requires both ordinary thresholds and no increase in the cumulative real benefit shortfall for either displaced cohort.

| Investment decay | Rule | Consumption | Debt | Both | Both with no additional real shortfall |
|---:|---|---:|---:|---:|---:|
| 0 | None | 298 | 356 | 297 | 297 |
| 0 | Any gap | 298 | 356 | 297 | 282 |
| 0 | Lower bound | 298 | 356 | 297 | 297 |
| 0 | Conditional | 298 | 356 | 297 | 297 |
| 0.518 | None | 278 | 356 | 269 | 269 |
| 0.518 | Any gap | 284 | 359 | 280 | 279 |
| 0.518 | Lower bound | 278 | 356 | 269 | 269 |
| 0.518 | Conditional | 278 | 356 | 269 | 269 |
| 1.5 | None | 56 | 92 | 22 | 22 |
| 1.5 | Any gap | 79 | 157 | 66 | 66 |
| 1.5 | Lower bound | 77 | 144 | 58 | 58 |
| 1.5 | Conditional | 87 | 145 | 62 | 62 |

In the last-column screen, the conditional rule improves on no support in 15 settings and ties in nine. It beats any-gap support in five, ties in six, and trails in 13. It beats lower-bound support in five, ties in 11, and trails in eight. The conditional rule also retains every baseline pass under the v299 screen, which separately tests rationing frequency and depth.

At the historical stress anchor and central rates with a 15% ceiling, conditional support increases the real shortfall in 147 scenarios for displaced households without debt and 146 for those with debt. Any-gap support does so in 252 and 231, respectively. These are counts among all 432 cases. The joint screen can improve while many other cases have larger shortfalls. At the post-2012 anchor and the central rate sensitivity, the rule activates at least once in 239 cases but yields the same joint count as no support. An activation count alone cannot establish effectiveness. Scenario counts describe the tested model grid.

Run `python scripts/israel_conditional_support_v300.py` for the full results.