# Step 204: Israel's unemployed by occupation, before and after generative AI

Sources: CBS annual Labour Force Survey, table 5.3 (unemployed who worked in the past 12 months, by
occupation of last job), editions 2017, 2018, 2021, 2022, 2023 and 2024. The 2023 and 2024 editions
are in publication folders `2025/lfs23_1962` and `2026/lfs24_1988`. Employed persons by occupation
come from ILOSTAT (national source CBS). Exposure is ILO 2025, as in step 201. Archived in
`data/processed/israel_unemployed_occupation_v204.json` (CI run 42062838270).

Relative risk (RR) = share of the unemployed in a group / share of the employed in that group.

| Occupation group (most exposed first) | ILO exposure | RR 2017-18 | RR 2021-22 | RR 2023-24 |
|---|---:|---:|---:|---:|
| Clerical support workers | 0.549 | 1.35 | 1.60 | 1.34 |
| Professionals | 0.400 | 0.62 | 0.60 | 0.78 |
| Managers | 0.362 | 0.82 | 0.74 | 0.73 |
| Technicians and associate professionals | 0.351 | 0.97 | 1.01 | 0.93 |
| Service and sales workers | 0.238 | 1.34 | 1.28 | 1.17 |
| Skilled trades and operators (7-8) | 0.198 | 0.97 | 1.17 | 1.11 |
| Elementary occupations | 0.152 | 1.68 | 1.64 | 1.68 |

By year:

| Year | RR clerical | RR professionals | Professionals' share of the unemployed | Exposure gap (unemployed − employed) |
|---|---:|---:|---:|---:|
| 2017 | 1.36 | 0.57 | 15.8% | −0.018 |
| 2018 | 1.34 | 0.67 | 19.3% | −0.014 |
| 2021 | 1.60 | 0.56 | 18.1% | −0.014 |
| 2022 | 1.61 | 0.64 | 21.1% | −0.015 |
| 2023 | 1.26 | 0.72 | 23.8% | −0.018 |
| 2024 | 1.42 | 0.83 | 27.4% | −0.006 |

Reading:

1. **Clerical workers did not become more over-represented after generative AI.** Their RR fell
   from 1.60 in 2021-22 to 1.34 in 2023-24, the level of 2017-18. The rise seen in step 201 was not
   continued.
2. **Professionals moved toward parity.** Their RR rose from about 0.6 to 0.78, and their share of
   the unemployed rose from 16% in 2017 to 27% in 2024. Group 2 includes software developers; this
   is consistent with the Taub Center finding of a higher exposed share among the unemployed
   (Debowy, Epstein and Weiss, 2026).
3. **No exposure gradient.** The slope of the change in log RR on exposure is −0.12 against 2021-22
   and +0.04 against 2017-18. The unemployed remain slightly less exposed than the employed; the
   gap narrowed in 2024 (−0.006) because more professionals became unemployed.

Limits:

- 2023-24 includes the war from October 2023: reserve call-ups and sector shocks change who is
  unemployed, independently of AI.
- Two post-AI years, seven occupation groups, published tables without standard errors: this is a
  description, not a test.
- Exposure is measured at the major-group level; a change inside group 2 (for example among
  software developers) cannot be separated from the rest of the group.

No causal claim is made.