מסמך טכני מיומן המחקר · באנגלית, כפי שנכתב בשלב העבודה
Step 294: paired policy comparisons across investment decay
The v293 grid is extended to 14 decay settings from 0 to 1.5. Each setting uses the same 432 cases, two interest sensitivities and three support rules, for 36,288 simulations. The support variants retain the shared 15% positive-profit levy ceiling and the 5% of output spending limit. A joint pass requires both displaced groups to maintain at least 90% of their initial real nonhousing consumption in every year and peak net public debt to stay at or below 90% of output.
Each case is compared with itself under each support rule. A gain is a joint fail that becomes a joint pass. A loss is the reverse. The joint counts below are ordered as no support / support at any demand gap / support only at the zero rate. Gains and losses compare gap support with no support.
| Decay | Joint counts, rate sensitivity 0.60 | Gains / losses | Joint counts, rate sensitivity 0.15 | Gains / losses |
|---|---|---|---|---|
| 0.0 | 297 / 289 / 297 | +0 / −8 | 265 / 262 / 265 | +4 / −7 |
| 0.1 | 294 / 289 / 294 | +3 / −8 | 264 / 264 / 267 | +7 / −7 |
| 0.2 | 294 / 289 / 294 | +3 / −8 | 261 / 263 / 261 | +13 / −11 |
| 0.3 | 291 / 287 / 291 | +4 / −8 | 272 / 281 / 275 | +20 / −11 |
| 0.4 | 290 / 283 / 290 | +4 / −11 | 275 / 283 / 280 | +20 / −12 |
| 0.5 | 272 / 281 / 275 | +16 / −7 | 264 / 283 / 275 | +30 / −11 |
| 0.518 | 269 / 281 / 269 | +19 / −7 | 261 / 283 / 275 | +33 / −11 |
| 0.6 | 266 / 280 / 266 | +21 / −7 | 252 / 278 / 267 | +34 / −8 |
| 0.8 | 225 / 254 / 230 | +29 / −0 | 180 / 247 / 213 | +67 / −0 |
| 1.0 | 154 / 213 / 179 | +61 / −2 | 89 / 191 / 151 | +102 / −0 |
| 1.2 | 73 / 164 / 145 | +91 / −0 | 40 / 115 / 93 | +75 / −0 |
| 1.27 | 60 / 134 / 119 | +74 / −0 | 27 / 91 / 81 | +64 / −0 |
| 1.4 | 31 / 89 / 79 | +58 / −0 | 15 / 56 / 52 | +41 / −0 |
| 1.5 | 22 / 60 / 58 | +38 / −0 | 8 / 41 / 37 | +33 / −0 |
- At rate sensitivity 0.60, gap support first exceeds both alternatives at the sampled decay of 0.50, after trailing both at 0.40. At sensitivity 0.15 the first sampled value is 0.20, following a tie with no support and a loss to zero-rate support at 0.10. It remains ahead of both alternatives at every higher sampled value for each sensitivity. The sampled grid does not identify a continuous threshold.
- At the post-2012 anchor of 0.518 and rate sensitivity 0.60, the net gain of 12 contains 19 gains and seven losses. At 1.5, gap support gains 38 cases against no support without losses, but compared with zero-rate support it gains three and loses one.
- At decay 0.80 and rate sensitivity 0.60, gap support gains 29 cases and loses none against no support. At 1.00 it gains 61 and loses two. A higher joint count can coexist with individual losses even when the decline in investment is stronger. Grid counts describe the cases tested and do not estimate probabilities.
Run python scripts/israel_investment_policy_v294.py for separate consumption and debt counts and all paired comparisons. The common settings reproduce the v293 counts, and gains minus losses reconcile with the difference in joint counts.