מסמך טכני מיומן המחקר · באנגלית, כפי שנכתב בשלב העבודה
Steps 231-233: households set against the CBS household expenditure survey
Three household inputs of the model are compared with the CBS Household Expenditure Survey 2021, General Summary (publication 1924): table 24 (households in owned dwellings, by net-income decile) and table 25 (households in rented dwellings). The tables were downloaded in CI run 42407324049 and archived with URL and SHA-256 in data/processed/israel_hes2021_v231.json. 2021 is the latest year with a published general summary; spending that year was still affected by COVID-19. Generated by scripts/israel_households_v231.py; archived in data/processed/israel_households_v231.json. Model files are not edited; constants are set in memory and restored. Illustrative, not a forecast.
Step 231: consumption share
The model has households spend 90% of the income left after housing payments. In the survey (monthly NIS per household):
| Net money income | Housing payments | Non-housing money spending | Share | |
|---|---|---|---|---|
| Owners | 18,689 | 1,751 (mortgage 1,378, other housing loans 89, other housing 284) | 12,300 | 0.73 |
| Renters | 13,355 | 3,463 | 8,320 | 0.84 |
| Weighted by tenure | 0.76 |
The share falls with income: in the bottom decile reported spending exceeds reported income (ratio above 2), in the top decile it is 0.53 for owners and 0.46 for renters. The survey share is a cash measure and excludes saving through pension contributions, which come out of gross pay.
Rerun of the 576-case grid with the Israeli macro core of step 230:
| Consumption share | Completed (of 576) | Both displaced cohorts kept 90% | With Israeli corporate tax (of 144) |
|---|---|---|---|
| 0.73 (owners) | 128 | 52 | 24 / 6 |
| 0.76 (survey, weighted) | 123 | 46 | 32 / 9 |
| 0.84 (renters) | 160 | 72 | 48 / 22 |
| 0.90 (model) | 207 | 88 | 43 / 22 |
| 1.00 (no saving) | 240 | 72 | 8 / 3 |
This input matters, contrary to the expectation that it would cancel out. At the survey share, fewer paths complete, and most of the additional stops are corporate liquidity failures (333 against 225 at 0.90): households spend less, firms collect less revenue, and more firms run out of cash. With no saving, firms survive but the government runs out of cash more often (180 against 144). Pass-through, preserved/completed at the survey share: 18/32, 17/40 and 11/51 at 0, 0.5 and full pass-through (model value: 30/64, 32/69, 26/74). Under the Israeli corporate tax only doubled productivity still completes, and the cohorts kept consumption in 9 of 144 paths against 22.
The survey share is lower than the saving rate the model value stands for (the Ministry of Finance reported net household saving of 12.1% of disposable income for 2016), so 0.76 and 0.90 bracket the plausible value. The result is reported as a sensitivity of the model to household saving, not as a recalibrated headline.
Step 232: housing tenure of workers in exposed occupations
Not measured. No published CBS table crosses tenure with the occupation of the head of the household. The step instead reports what the model implies for each income decile, using that decile's own tenure mix (tables 24-25) and its renters' rent burden (rent / net money income), combined with the step 205 results (mortgage share 37%, burden 29%, half fixed and half reset).
| Decile | Own | Rent | Renter rent burden | Preserved at full pass-through, goods-indexed | Housing-indexed |
|---|---|---|---|---|---|
| 1 | 34% | 58% | 44% | 14.9% | 33.8% |
| 2 | 47% | 46% | 34% | 18.9% | 35.7% |
| 3 | 51% | 41% | 28% | 20.5% | 33.4% |
| 5 | 61% | 32% | 24% | 27.2% | 35.5% |
| 7 | 73% | 20% | 23% | 29.9% | 38.2% |
| 10 | 80% | 16% | 17% | 30.4% | 37.6% |
The gap between deciles is a housing-tenure gap: the lower the decile, the more renters and the heavier their rent. If workers displaced from exposed occupations come mostly from the lower deciles, step 205's bottom-decile result (15.3%) is the relevant one; if they come from the middle (clerical support workers) the all-household result (26%-29%) applies. Which deciles they come from is an inference that requires the survey microdata (CBS research room), and is left open.
Step 233: share of owners with a mortgage
Not published for 2021. The average mortgage payment over all owners is 1,378 NIS. Divided by the average payment of those who pay one, it gives the share of payers: 3,400 NIS in the 2017 survey (as reported by Calcalist) gives 41%, and an assumed 4,000 NIS for 2021 gives 34%. Central value 37%; step 205 used 39.3%.
| Mortgage share | Preserved, all households, goods-indexed (0 / 0.5 / full) | Housing-indexed |
|---|---|---|
| 34% | 44.4 / 43.5 / 27.3 | 45.9 / 39.5 / 32.5 |
| 37% | 43.9 / 43.1 / 26.8 | 45.6 / 39.5 / 32.4 |
| 41% | 43.4 / 42.6 / 26.2 | 45.2 / 39.5 / 32.3 |
| 39.3% (step 205) | 43.6 / 42.8 / 26.4 | 45.3 / 39.5 / 32.3 |
The share of mortgage holders hardly matters: the result moves by about one point across the range.
Limits
- One survey year, 2021, with COVID effects; money spending and income as reported by households.
- Consumption share is a single constant in the model; the survey shows it falls steeply with income.
- Tenure by occupation needs microdata; step 232 is a decile sensitivity, not a measurement.
- The payers' average mortgage payment for 2021 is assumed.