Israel's 257 local authorities · 2024 budget execution

The Municipal Shekel

How much each city, local council and regional council spends per resident, where that money comes from, and how it lines up with the socio-economic map. Official CBS figures, cross-checked against Budget Key.

CBS · actual spending nominal ₪ unless marked no causal claims

Short version (cities compared with cities)

Explore

The atlas: every authority against its peers

Each bubble is a local authority, sized by population and coloured by how far it sits from the median of its peers: same kind of authority (city, local council or regional council) and same CBS socio-economic cluster. Pick a measure, filter, then click any authority to print its receipt.

Measure

Kind

District

Socio-economic cluster

Find

View

vs peer median

below peers±7.5%above peers

Grey dots are the 1,458 localities CBS lists — they draw the country. Regional councils sit at the population-weighted centre of their villages.

Furthest above peers

Furthest below peers

1Money in · cities only · 2024

Better-off cities spend more, and it is mostly their own money

actual · CBSnominal ₪ 2024pooled: Σ₪ ÷ Σresidents

Table view

Cluster 1 holds a single city (Modi'in Illit), so it is shown but not discussed. "State, earmarked" is everything the ministries transfer except the Interior Ministry's general (balancing) grant; balancing loans (≤ ₪991 per resident anywhere) are left out of the stack.

2Business arnona · cities only · 2024

Same cluster, wildly different tax base

actual collectionsnominal ₪ 2024

Table view

Business arnona (ארנונה לא למגורים) is the property tax a municipality collects from offices, shops, industry, hotels and the like. How much a town gets depends on where industrial zones, malls and plants happen to sit — history and planning decisions, not the residents' income. This chart describes the gap; it does not explain it.

3Residential arnona · cities & local councils · 2024

Billed is not collected

actual collectionsratio, not ₪

one authoritymedian of the clustermiddle half (25th–75th percentile)

Table view

CBS ratio "collected as % of all net charges": everything collected during 2024 (current year + arrears) divided by the year's net charges plus the opening debt balance, after exemptions and discounts. Regional councils are left out because much of their residential arnona is billed through village committees.

4Education · cities only · 2024

Per child, the gap widens

actual · CBSnominal ₪ 2024denominator caveat

Columns are pooled (Jerusalem dominates cluster 2, Tel Aviv-Yafo cluster 8); the prose quotes the median city. Both are in the table view.

Table view

Read with care. The denominator is residents aged 0–17, not pupils in municipal schools. Where many children attend institutions financed partly outside the municipal budget (for example parts of the Haredi education system), per-child municipal spending looks lower without saying anything about total spending on those children. Compare like with like: use the atlas peers.

5Welfare · cities & local councils · 2024

Need differs sevenfold; welfare spending by 40%

actual · CBSnominal ₪ 2024one proxy for need

Welfare spending per resident (₪, median)

Income-support recipients per 1,000 residents (median)

Table view

Two separate charts on purpose: different units, so no shared axis. Income support (הבטחת הכנסה, National Insurance, end of 2024) is only one signal of need; welfare departments also serve the elderly, people with disabilities and children at risk. Welfare services are largely co-funded (roughly 75% state, 25% municipality), so a municipality's ability to pay its share is part of the picture.

62015 → 2024 · cities only

Ten years on: more money everywhere, the gap a little narrower

actual · CBS 2015 & 2024real · 2024 prices

Prices

20152024

Table view

2015 figures come from Budget Key's copy of the CBS 2015 local-authority profile (budget.lamas_muni), matched to 2024 by authority code (regional councils re-coded to the 2024 55XX scheme). Both years are grouped by the 2024 file's cluster (2021 index), so groups hold the same authorities. Excluded: boundary changes (Mateh Yehuda / Tzur Hadassah, Sha'ar Shomron), authorities with no 2015 figure, and those whose population more than doubled or shrank by over 20%. CBS revised population estimates after the 2022 census, which slightly affects per-resident comparisons across years. Deflator: CBS CPI annual average, 2015 = 87.28 when 2024 = 100.

7Planned vs actual · Budget Key vs CBS · 2021–2024

Approved budgets are a floor, not a ceiling

planned · Budget Keyactual · CBSsame-year nominal ₪

Table view (including the excluded rows)

A handful of documents match the CBS result almost exactly (e.g. Herzliyya 2022, Ma'ale Adummim 2021); those may already be year-end revisions rather than the budget approved in advance. Budget Key's own "revised" and "executed" columns are too sparse to tell. Sample: whatever municipalities uploaded — not a random draw.

Why the main analysis uses CBS, not Budget Key's municipal budgets

Authorities with any line in budget.muni_budgets, and with a full budget document (all four expenditure chapters), per year — against the 257 authorities CBS covers.

any budget linesfull budget document

Table view

8Emergency operations · 2024

The war shows up in a few budgets, heavily

actual · CBS% of own regular spending

Table view

CBS added the line "emergency operations" (פעולות חירום) in 2023 for Iron Swords war activity. These authorities' per-resident totals are inflated by it, which is why the atlas flags them. Populations are registered residents at the end of 2024, including people who were evacuated.

How this was made

Method, sources, SQL and freshness

Sources

Pipeline

Read the shared Hasadna mirror (hasadna.duckdb) once, read-only, copying only budget.muni_budgets and budget.lamas_muni to local Parquet — no writes, no locks held while another dashboard reads it.
Inspected both: muni_budgets covers at most 41 authorities in any year, with full budget documents for 0–23 and almost no executed values; lamas_muni is the CBS 2015 profile (255 authorities, 2013 socio-economic index). Neither supports a current nationwide comparison.
Acquired the targeted official files: CBS Local Authorities in Israel 2021–2024 (budget execution + population + 2021 socio-economic index), the CBS localities file (English names, coordinates) and the CBS CPI series.
Parsed the multi-row Excel headers by label, not position (layouts shift between editions); every selected column must match exactly once or the build stops.
Joined, computed per-resident figures and peer medians in DuckDB SQL, and exported only aggregates: one row per authority plus group summaries.

Definitions

Spending = CBS "total expenditure in the regular budget" (תקציב רגיל): audited actual execution. Development (non-regular) budgets are excluded.
Per resident = amount ÷ CBS end-of-year population of the authority — the same denominator CBS uses for its own per-capita column (checked: identical for every authority, 2022–2024).
Group figures are pooled (Σ shekels ÷ Σ residents), so big cities weigh more; medians are labelled as medians.
Cluster = CBS socio-economic cluster from the 2021 index (1 = lowest, 10 = highest), as printed in the 2024 file. Two new councils (Tzur Hadassah, Sha'ar Shomron) carry a cluster but no rank.
Peers = other authorities of the same kind in the same cluster; if that group has fewer than six members, clusters ±1 are added. Fewer than three peers → no peer comparison shown.
Real vs nominal: 2024 cross-sections are nominal 2024 shekels; any comparison across years is deflated with the CBS CPI (annual averages, chain-linked across base periods, 2024 = 100) and marked "real".
Planned vs actual: "planned" = the budget document Budget Key holds for that year (usually the approved budget, sum of expenditure chapters 6–9); "actual" = CBS execution for the same year.

Checks that passed

Sum of authority regular spending 2024 = ₪99.19 bn = CBS national total; per-resident ₪9,932 matches CBS's own figure.
Population in the budget sheet equals the population sheet for every authority (2022–2024); CBS's printed per-capita spending equals spending ÷ population (2022–2024; 2021 differs by <₪5 at 13 small authorities from rounding).
Revenue identity: own income + state transfers + balancing loans = total regular revenue (residual never negative); expenditure chapters never exceed the total.
2015 ↔ 2024 code crosswalk: all 255 authorities of 2015 match; the 26 name differences (24 spellings, 2 renames: Nazareth Illit → Nof HaGalil, Emek Lod → Sdot Dan) were checked by hand; 6 local councils became cities (grouped by 2024 status).
Pre-2024 regional-council codes collide with local-council codes (e.g. 43 = Metula and a regional council), so older files are keyed on code + status — a naive join duplicated 12 rows and was caught.
Nine authorities reported nothing for 2021 (all zeros in the CBS file) — treated as missing, not as ₪0.

Not used, on purpose

CBS's "authority activity during the Iron Swords war" column: it contains #VALUE! and 0 entries and at least one implausible label, so the war is read from the emergency-operations spending line instead.
Budget Key muni_budgets for cross-authority comparison: partial uploads (e.g. Haifa 2024 = one ₪57 m line), mixed hierarchy levels that double-count if summed naively, and executed values almost always empty.

Limitations

Descriptive only. Differences between clusters are associations; cluster, peripherality, population mix, age structure and history overlap, and nothing here isolates a cause.
Cities, local councils and regional councils provide different services at different scales (regional councils share them with village committees), so the findings compare like with like and the atlas only compares peers.
Small authorities (under 5,000 residents) swing a lot per resident; the receipt flags them.
"Per child" and "income support" are proxies with known blind spots (see findings 4 and 5).
The socio-economic index is from 2021 and peripherality from 2020 — the latest CBS published in this file.

The SQL

The core of export.sql (DuckDB). The full pipeline: extract.py · cbs_parse.py · etl_cbs.py · cpi.py · export.py.

loading…

Downloads

authorities_2024.csv — one row per authority, every figure on this page.
data.json — everything the page draws (aggregates only; no raw tables).