municipal-policy-methodology

name	municipal-policy-methodology
description	House methodology for evidence-based municipal policy development — the stage model, evidence standards, the municipal source hierarchy, the vires gate, arnona and budget discipline, testimony handling, and the think-tank writing style. Load whenever cleaning testimony, researching a city policy area, drafting a policy paper, or updating the program for the city.

Municipal policy development methodology

This project turns lived experience of a city into a credible program for governing it. Its only asset is credibility; these rules protect it.

The national version of this problem lets you propose almost anything and argue about whether it is a good idea. A municipal platform cannot. Most of what residents blame City Hall for is not City Hall's to fix, and a good half of what City Hall could fix tomorrow nobody asks it to. Establishing which is which is not a preliminary to the policy work — it is the policy work, and it is what the extra stage in this pipeline exists for.

The pipeline

Four stages, each with a gate (full detail in the repo's pipeline.md):

1. Testimony (testimonies/) — dated, cleaned, redacted first-person accounts of living in the city. Raw material never committed.
2. Research (policy/<area>/research/) — six documents: problem statement, jurisdiction map, municipal instruments, comparative city review, data annex, City Hall's own record. Three of the six have their own method skills: municipal-powers-analysis, comparative-city-research and city-hall-record-research.
3. Paper (site/public/papers/) — Typst + PDF using templates/policy-paper.typ.
4. Platform (platform/program-for-the-city.md) — the distilled meta-proposal.

Each policy area has a stage tracker at policy/<area>/index.md. Keep it current.

Paths above are relative to the repository's content root. This repository publishes its content as a site and puts the root at site/src/content/docs/ — see CLAUDE.md for the mapping table, and resolve every path in this skill against it.

The vires gate

The rule that distinguishes this project from a wish list, and the one to apply hardest:

No proposal enters a policy paper without a named instrument and a named holder of it.

For every recommendation, state which of these it is:

Class	Meaning	What the paper must show
A — Municipal, own motion	The municipality already has the power and the money	The instrument (bylaw section, tender clause, licensing condition, budget line, council decision), the department that administers it, and why it has not been used
B — Municipal, needs approval	Council can act but a ministry must sign off	The approving body, the statutory basis for its consent, and evidence about how it has decided comparable requests
C — Joint	Delivery is municipal, decision or funding is national	The split, precisely — who sets policy, who employs the staff, who pays what share — and the lever the municipality holds inside the arrangement
D — National	The municipality has no formal power at all	Say so plainly, then either drop it or reclassify it as an advocacy proposal with a named counterpart, an ask, and a realistic route

A class-D proposal presented as something a mayor can deliver is the single most common dishonesty in municipal campaigning. This project does not get to commit it. Advocacy proposals are legitimate and stay in — labelled as advocacy, with the mayor's actual leverage (convening power, statutory consultee status, planning consent, land, publicity) named rather than assumed.

Classification is done in 02-jurisdiction.md by the municipal-powers-analysis skill, and every recommendation in the paper carries its class letter.

Evidence standards

• Never invent a number. If you cannot find a source, write [UNVERIFIED — verify before publication]. Unverified claims block stage advancement.

• Municipal source hierarchy (prefer higher):

  1. The municipality's own published record — the approved budget (תקציב), council and committee protocols (פרוטוקולי מועצת העיר וועדותיה), bylaws (חוקי עזר), tenders (מכרזים), city plans lodged in the planning system, and the municipal auditor's reports (מבקר העירייה). These are primary and they are what a mayor would actually be handed.
  2. CBS (הלשכה המרכזית לסטטיסטיקה) — the local authorities file (הרשויות המקומיות בישראל), the social-economic ranking (מדד חברתי-כלכלי) and the municipality's own CBS profile. Neighbourhood-level statistical areas (אזורים סטטיסטיים) are the resolution most municipal arguments actually need.
  3. Jerusalem Institute for Policy Research (מכון ירושלים למחקרי מדיניות) — the Statistical Yearbook of Jerusalem, published annually since 1983 ahead of Jerusalem Day, and Jerusalem: Facts and Trends. It compiles CBS and municipal data at city and neighbourhood level. Treat it as an excellent secondary source and follow its own footnotes to the primary where a claim is load-bearing.
  4. State Comptroller (מבקר המדינה) — including the dedicated local-government audit (הביקורת על השלטון המקומי), which audits municipalities directly.
  5. Ministry of the Interior (משרד הפנים) — municipal financial reporting, the balancing grant (מענק איזון), and the district commissioner's (הממונה על המחוז) decisions.
  6. Knesset Research and Information Center (מרכז המחקר והמידע של הכנסת) — frequently has already done the cross-municipal comparison you are about to attempt.
  7. Bank of Israel, OECD and international official statistics, peer-reviewed research.
  8. Quality journalism — last resort, always flagged. Local Jerusalem press (Kol Ha'Ir, Jerusalem Post, Ynet Jerusalem desk, Ha'aretz) is often the only record of a council row, but a council row is reported, not proven; find the protocol.

• Legislation and subordinate legislation are cited precisely: official name (Hebrew and English), year, amendment number, specific sections. Municipal work leans heavily on subordinate instruments — bylaws, ministerial orders, licensing conditions, tender terms — and these get the same treatment: name, publishing authority, date, clause. Verify clause numbers against the actual text; do not cite provisions from memory.

• Testimony is evidence of experience, not prevalence. One resident's street proves something happen; only data proves it is . At municipal scale this trap is sharper than nationally, because a single neighbourhood's problem can be genuinely unrepresentative of the city while being completely real. Papers keep the distinction explicit and, where the data allows, say .

Money discipline

A municipal proposal that does not say where the money comes from is not a proposal. Every costed recommendation states its funding class:

• Within an existing budget line — name the line and the year's allocation, and say what gets less.
• Reallocation — name both lines. "Efficiency savings" is not a source.
• Arnona (ארנונה, municipal property tax) — the default assumption is that the municipality cannot simply raise it. Rates are updated annually by a nationally set formula, and a deviation (חריגה) requires the joint approval of the Ministers of the Interior and Finance. Any proposal relying on arnona income must establish the current rule and cite it, and must address the approval risk. Changes to classifications, discounts and exemptions (הנחות ופטורים) are a separate and often more available lever than the headline rate — check both.
• A government ministry — this makes the proposal class C or D. Say so, and name the budget and the decision-maker.
• A municipal company or external body — Jerusalem delivers a great deal through municipal and quasi-municipal corporations and through philanthropy. Name the entity, its ownership, and whether the mayor actually controls it.

State the estimate's basis and confidence. "Fiscally neutral" requires showing the offset.

Neighbourhood discipline

City-wide averages hide the thing municipal policy is for. Where the data supports it, report by neighbourhood or statistical area rather than city-wide, and state explicitly:

• whether the problem is concentrated or general;
• whether the proposal's benefit is concentrated or general, and whether those are the same places;
• what it does to the gap between the best-served and worst-served parts of the city.

In Jerusalem specifically, an analysis that does not disaggregate East Jerusalem, the Haredi neighbourhoods, and the rest of the city will be wrong about almost any service question, and will be seen to be wrong. Where a data source does not cover a part of the city, say so — missing data about a neighbourhood is itself a finding, and often the most important one.

Testimony handling

• Cleaning removes disfluencies and transcription artifacts; it never adds, softens, strengthens, or renumbers a claim. Ambiguity in the raw stays ambiguous or gets resolved by asking the author.
• Redact private individuals' names, exact addresses, third-party identifying details. Institutions and public figures stay named. Municipal testimony is unusually easy to de-anonymise — a street, a building, a school and a date will identify a household. Generalise to the neighbourhood unless the specific location is the point, and if it is, get the author's explicit agreement.
• Frontmatter schema: title, date (YYYY-MM-DD), policy-area (kebab-case, matches policy/<area>/), status (cleaned | synthesized-into-research), source (one line, where the raw is retained privately).

Writing style (papers and platform)

• Think-tank register: measured, concrete, mechanism-first. Anger belongs in the testimony; the paper's force comes from evidence.
• No promise without a mechanism: every recommendation says what instrument changes, who administers it, what it costs, who pays, and which class (A/B/C/D) it is.
• Distinguish what the council can resolve at its next meeting from what needs a ministry, and give the realistic elapsed time for each.
• Hebrew terms on first use: transliteration + Hebrew + gloss, e.g. "Arnona (ארנונה, municipal property tax)". Where a term has a standard Arabic form used in the city, give it too on first use in any East Jerusalem context.
• Footnote every factual claim. A reader should be able to check anything without asking.

Typst conventions

• Import from templates/policy-paper.typ: policy-paper (show rule), #recommendation[...], #tradeoff(objection, response).
• Paper numbers are sequential MP-NNN across the whole project (municipal paper).
• Must compile warning-free with typst compile; visually inspect page 1 via PNG render before committing. Commit both .typ and .pdf.