Step 299: add benefit shortfall depth to the fiscal screen

Record the fraction of calculated benefits paid each year. For years after the initial period, measure the sum of unpaid fractions and the largest annual unpaid fraction. The sum is in full-benefit-year equivalents with equal year weights. It is not a cash or present-value measure. A case passes the depth screen if it passes the v298 consumption, debt and rationing-frequency screen, and neither depth measure exceeds the matched no-support value (tolerance 1e-8). Benefits and entitlements can vary across policy runs. The screen compares payment relative to calculated entitlement in each run.

Each cell is a count out of 432, in the order no support / 50% gap support / 75% gap support.

At either support intensity, net passes rise in 19 settings and fall in five. Both intensities lose at least one baseline pass in 11 settings. Relative to the v298 frequency screen, adding the depth conditions removes 45 case-setting passes at 50% support and 38 at 75% support. These totals count repeated scenario settings separately.

At the post-2012 anchor, central rate sensitivity and 15% ceiling, the 50% policy yields 280 ordinary passes, 279 frequency-protected passes and 279 depth-protected passes. The corresponding 75% counts are 285, 284 and 283. At the same financing and rate settings without investment decay, the 50% policy goes from 297 ordinary passes to 288 under the frequency test and 282 under the depth test. The no-support count is 297 under all three screens.

At the full-history stress anchor with central rate sensitivity, depth protection leaves the 50% policy counts at 59, 66 and 81 across the three ceilings, compared with 22 at each ceiling without support. These are model scenario counts under fixed levy behavior. The stricter metric records annual payment adequacy without assigning a monetary value to a shortfall.

Run python scripts/israel_investment_financing_v297.py to reproduce the three nested screens on the same 31,104 simulations.