Rating Methodology
How we calculate charity ratings — and when we don’t
Rating Overview
CharityData NZ rates charities on a scale of 0-100, with letter grades from A+ to F. Our rating combines three equally-weighted components: Financial Health, Governance, and Transparency.
Grade Scale
Financial Health (33%)
Evaluates how effectively a charity manages its finances and deploys resources toward its mission.
Reserve Levels
Charities should maintain reasonable reserves (3-12 months of expenses is typical). Excessive reserve accumulation may indicate funds aren’t being deployed effectively. Too-low reserves may indicate financial instability.
Stable-revenue softening (v3.1.1): when a charity runs below 3 months of reserves but has 3+ years of consistent revenue (coefficient of variation 30% or less) and no detected red flags, the financial-health sub-score is floored at 50/100 instead of being dragged toward zero by the raw reserve formula. Rationale: service-delivery charities with predictable contract or grant income run cash-flow-matched models — low reserves with steady revenue is an operating choice, not financial distress. Any detected red flag disqualifies the softening (anti-gaming).
Revenue Stability
We assess revenue trends over multiple years. Consistent or growing revenue suggests financial sustainability. Volatile or declining revenue may indicate risk. Revenue stability also gates the low-reserve softening described above.
Operating Efficiency
Based on available expense data. Note: NZ charity reporting doesn’t require detailed program/admin breakdowns, so this metric uses available categories (employee costs, other expenses).
Program-spending-ratio API fallback (v3.1.1):where a charity hasn’t been manually categorised but the Charities Services API reports a cost-of-service-provision figure of 50% or more of total expenses, that value feeds an API-fallback efficiency score. The score is capped at 75/100 (vs 100/100 for manually-categorised charities) to reflect the lower-confidence source. Charities that book costs differently keep a null efficiency score and the engine reweights the remaining components — no penalty for accounting choice.
3-Year Weighted History (v3.1.1)
For operating charities, financial health and efficiency are smoothed across the last three years using weights of 60% latest / 30% prior / 10% prior-prior, falling back to 70/30 with two years of data and 100% with one year. This indicates sustained performance rather than a single-year snapshot, so one-off events (a large bequest, property sale, or grant-timing shift) no longer spike or tank the rating. Efficiency is re-normalised against the weight-sum of years that actually have a non-null efficiency score, so a charity with only one year of categorisation data isn’t pulled down by missing years. Endowment foundations keep the single-year branch because their metrics work on a different basis.
Governance (33%)
Assesses board composition, oversight quality, and governance practices.
Board Size & Composition
Effective boards typically have 5-12 members. Too few may indicate insufficient oversight; too many may indicate inefficiency. We also consider tenure diversity.
Officer Disclosure
Charities should disclose complete officer/trustee information including names and roles. Missing or incomplete officer data reduces governance scores.
Overboarding Check
We identify trustees serving on excessive numbers of boards, which may indicate limited ability to provide adequate oversight to each organisation.
Transparency (33%)
Measures how openly a charity shares information with the public.
Filing Timeliness
Charities that file annual returns on time demonstrate better accountability. Late or missing filings reduce transparency scores.
Financial Disclosure Quality
We assess the completeness of financial information provided in annual returns. More detailed disclosures receive higher scores.
Contact Information
Charities providing complete contact details (website, email, phone, address) score higher than those with limited public contact information.
Officer Integrity Check
Separate from the rating, every charity profile's Governance tab cross-references the charity’s current officers against four NZ public registers and surfaces any flags individually with explanatory context. The integrity check does not affect the charity’s rating — it’s a parallel KYC layer for any donor, funder, or grant officer running their own due diligence.
Editorial guards:we surface flags only for currently-active officers (the §36B field is sometimes used by charities as a bulk departure marker, so historical attestations are stored for audit but not displayed). Sanctions matches that aren't backed by a discriminating identifier (date of birth for persons, NZBN for entities) sit in an internal review queue until a human verifies them. Where a flag does surface, every flagged row carries a direct link to the right-of-reply form pre-populated with the affected officer’s context. See Disclaimer §7a for the full reasoning. Sources documented at Data Sources.
Red Flag Detection
In addition to scores, we identify specific concerns that may warrant further investigation:
- Excessive Reserves - Cash reserves exceeding 18 months of expenses (or 36 months for fundraising-driven charities). The penalty is graduated by distance over the threshold (within 12 months: low; up to 2× threshold: medium; beyond 2×: high). Property, facility and conservation trusts whose assets are the mission (not idle cash) are exempted automatically — see FAQ.
- High Admin Costs - Administrative expenses significantly above sector average
- Governance Concerns - Very small boards, overboarded trustees, or missing officer data
- Late/Missing Filings - Annual returns filed late or not at all
- Revenue Decline - Significant multi-year revenue decline
- Disclosed Non-Arm’s-Length Related-Party Loan — Fires only when the charity’s own annual return discloses a related-party loan on terms that aren’t market-equivalent (interest-free, repayable on demand, no formal terms, concessionary, etc.). The flag mirrors what the charity itself published; we never assert our own opinion on whether a loan is below market. Description rendered on the profile reads “Annual return discloses N related-party loan(s) totalling $X on non-arm’s-length terms — may warrant trustee review.”
Note: Red flags indicate areas for further research, not wrongdoing. There may be valid explanations.
When We Don’t Rate (v3.1)
Earlier versions of the engine assigned every charity a letter grade, which meant a charity with no balance-sheet data and no actual concerns was floored at F purely for lack of disclosure. From v3.1, explicit “Not rated” statuses are used instead. In each case the charity’s page still shows the filed figures and any red flags — we just don’t attach a letter to a number we can’t calculate.
Volunteer / community organisation
Revenue under NZ$250,000, no paid staff, and less than 50% government funding. The standard financial template assumes salaried operations and external fundraising; it doesn’t fit a small marae committee, volunteer fire brigade, or community garden trust. We show the figures, but no grade.
Stale data
Most recent annual return is more than three years old. The charity may be inactive or pending deregistration. Until a current return is filed, a rating is not assigned.
Insufficient financial data
Net assets and total assets are both missing from the most recent return, and no prior year within the two-year fallback window has them either. The financial-health score cannot be computed without these figures.
Dormant or inactive
The charity appears to have no recent financial activity — either no annual return has been filed with Charities Services, or the most recent return reported zero revenue and zero expenses with no balance-sheet figures. The charity may be dormant, in transition, pending deregistration, or simply behind on filings — the public data alone doesn’t tell us which. If a charity is actively operating, filing a current annual return will allow a rating to be calculated on the next refresh.
Structural balance-sheet
Liabilities have exceeded assets for two or more consecutive years. That pattern is common and structural for retirement villages (occupation-right-agreement bonds recorded as liabilities), iwi treaty-settlement vehicles (future-obligation accruals), and infrastructure trusts with Crown loan principal. It does not indicate financial distress, but our standard reserve-months scoring template does not apply. A single year of negative equity is still rated normally — it’s the persistence that signals a structural balance sheet.
Deregistered (DR)
The charity has been removed from the Charities Services register and is no longer obliged to file annual returns. Because the charity is no longer active under the Charities Act, the standard rating is withheld and the badge is DR rather than NR. The figures shown above reflect the last filings made while the charity was registered.
Summary-form filing
Charities Services Tier 3/4 reporting lets smaller charities file a summary annual return rather than separately audited financial statements. The return reports revenue, expense, and asset totals but not the program vs administration split or the line-item detail our template uses. When the template-derived grade would be D+/D/D-/F on a charity with revenue under NZ$1M, no detected red flags, and no manual categorisation from reviewed statements, a letter grade is withheld — the score in that case reflects filing format, not operations. Charities with any red flag, or with revenue at or above NZ$1M (where audited accounts are expected), stay on the normal scoring path.
Anti-gaming guard
A charity with revenue of NZ$1M or more that omits balance-sheet figures is still ratedand receives an “incomplete financial disclosure” red flag. Charities at that scale are required to file proper accounts — non-disclosure cannot be used to slip into the “Not rated” bucket. Detected red flags of any kind also disqualify a charity from the volunteer / community carve-out.
Prior-year balance-sheet fallback: if the most recent return is missing net assets but a return within the last two years has them, the engine reuses the prior year’s balance sheet to compute reserve months. The rating notes which financial year was used.
Different charities, different scoring templates
The default scoring template assumes a service-delivery operating charity — staff salaries, program-vs-administration spending split, 3–12 months of reserves. That template doesn’t fit every charity registered with Charities Services. Three classes of charity have explicit carve-outs so they aren’t penalised for not matching a template that doesn’t apply to them.
Fundraising charities
Cause-based, donor-funded charities (Cure Kids, Movember NZ, Royal NZ Foundation of the Blind, Heart Foundation, Auckland Rescue Helicopter, etc.). Their revenue is donor-cycle — most intake comes during a campaign window and is deployed across the year, so a bigger reserve buffer is operationally appropriate. The 18-month “excessive reserves” threshold lifts to 36 monthsfor these charities. Their staff are specialists (researchers, caseworkers, paramedics) so the “average salary” benchmark is interpreted with that context.
Crown entities
Established under the Crown Entities Act 2004 Schedule 4 (Maori Television Services, Crown Forestry Rental Trust, NZ Walking Access Commission, etc.). Disclosure is governed by the Crown Entities Act — in their own Annual Report — not by Charities Services. The “no executive compensation disclosure”, “low program spending”, and “excessive reserves” flags are skipped because they describe Charities Services-shaped concerns that don’t apply to a Crown entity’s legal regime.
Statutory regulators — not graded
Professional licensing bodies (Pharmacy Council, Nursing Council, Medical Council, Teaching Council, Veterinary Council, Midwifery Council, etc.). Their function is licensure, inspection, and discipline of a profession — their “programs” ARE regulation, so the program-vs-administration split makes no operational sense. From #2469 they are routed to the “Not rated — not applicable” status: the standard charity scorecard does not measure what these entities actually do. Red flags are still detected and shown if any fire; a letter grade is withheld so the entity is not misrepresented as a poor charity when it is in fact a profession-regulating body.
Subsidiaries of service-delivering parents — not graded
Property, trading, and investment vehicles held by a service-delivering parent charity (church property trustees, diocesan ecclesiastical goods trusts, university research trusts, professional-association property holdings, etc.). Their purpose is to hold or trade assets on the parent’s behalf — not to deliver programs to public beneficiaries on their own books. The standard scorecard’s reserve-months, program-vs-administration split, and fundraising-efficiency signals are meaningful at the parent level, not the subsidiary level. Tagged entity_type='subsidiary' and routed to “Not rated — not applicable”. Look up the parent charity by name for its rating. Functionally-endowment vehicles — iwi quota holders, Maori settlement asset companies, and similar entities that deploy capital for beneficiary distributions rather than holding for a service-delivering parent — are tagged endowment_foundation instead and rated under the endowment scoring branch.
Anti-gaming guard
The carve-outs only switch off flags whose methodologydoesn’t fit the entity type. They do notwaive the data-quality, governance, or red-flag detection that DOES apply. A fundraising charity holding $228M of reserves over 79 months is still flagged for “excessive reserves” even with the relaxed 36-month threshold — that level genuinely is high regardless of fundraising cycle. A Crown entity that self-deals or has dormant filings is still flagged.
Classification audit: each candidate charity is reviewed and assigned an entity-type tag at confidence ≥ 0.70. Below that threshold, the charity stays on the default operating-charity template — we don’t apply carve-outs we’re not confident about.
v3.2 — tier confidence and outcomes disclosure (2026-04-24)
Two calibration changes that address a specific gap: charities filing the simpler Tier 3 / Tier 4 Annual Return were previously judged on the same absolute program-vs-administration thresholds as charities filing a full Tier 1 audit. In practice the simpler returns don’t report the program-vs-admin split the template needs, so the engine was scoring an absent field as if it were a zero. v3.2 fixes that by reducing how much a thin filing drives the composite, and adds an upside for charities that go beyond the register and disclose verifiable outcomes.
Reporting-tier confidence weight (#2461)
The Charities Services Tier 1–4 regime has four filing formats keyed to revenue: Tier 1 (≥ $30M, full audit), Tier 2 ($2M–$30M, audit or review), Tier 3 ($125K–$2M, simple accrual), Tier 4 (< $125K, cash basis). Only Tier 1 / 2 filings reliably itemise program vs administration expenses, fundraising cost, and all balance-sheet lines the scoring template consumes. v3.2 weights the financial signal by tier confidence: Tier 1 × 1.00, Tier 2 × 0.90, Tier 3 × 0.75, Tier 4 × 0.50. The lost weight is redistributed to the governance component — when the financial statement is thin, the engine leans on what IS disclosed (board size, chair structure, officer turnover) rather than amplifying the gap.
Outcomes-based impact bonus (#2455)
A fourth, asymmetric scoring dimension: charities that publish verifiable outcome metrics in their Annual Report (beneficiaries served, hectares restored, procedures delivered, year-on-year comparisons) earn up to +15 pointson the final score. Sparse data incurs no penalty — the design goal is to reward disclosure, not punish its absence. Bonus scales with count and with presence of prior-year comparisons (which are the hardest to fake): 2–4 outcomes = +4 pts, 5–9 = +8 pts, 10+ = +12 pts, plus +3 pts when at least one metric carries a prior-year comparison, capped at +15. Applied after red-flag penalties and before the 0–100 clamp, so a well-documented charity can’t have its bonus erased by the penalty cap.
Peer-relative efficiency blend (#2456) — plumbing shipped, signal limited
An 80/20 blend of the absolute program-ratio score with a cohort-percentile score is wired into the engine, but measurement of the live data shows it has near-zero aggregate effect today: roughly 95% of NZ charities file Tier 3 / Tier 4 summary forms that don’t populate the program-spending-ratio field, so the cohort percentiles (p25/p50/p75) collapse to zero and there is no signal to blend with. The plumbing is kept in place — it will activate as more charities either voluntarily file at Tier 1 / 2 or as the extractor populates program/admin categorisation for more returns — but we’re transparent that it is not the load-bearing change in v3.2. The real F-overcount fix in this version comes from the tier-confidence weighting above.
v3.3 — rating eligibility aligned with Charities Services tiers (2026-04-24)
Who we rate is now aligned with the Charities Services Tier 2 / Tier 3 boundary — a methodology decision that matches established US practice and resolves the largest remaining source of F-grade noise. Critically, “not rated” never means “no information”: every not-rated profile carries a structural “Reasons to look carefully” tile when any donor-relevant concern fires.
Rating eligibility aligned with Tier 2 boundary ($2M revenue)
Charities Services Tier 3 filings ($125K–$2M revenue) use a simple accrual template that doesn’t itemise program vs administration expenses, fundraising cost, or separated balance-sheet lines. We were attempting to grade these filings against thresholds calibrated for Tier 1 / 2 audited statements. Under v3.3 a charity under $2M revenue that has zero detected red flagsis routed to “Not rated — limited disclosure” rather than being pushed to D or F by template-fit artefacts. This threshold now matches US best practice — Charity Navigator rates from ~$1M contributions, CharityWatch requires $1M+ with complete audited financials, GuideStar issues voluntary disclosure seals rather than grades. Our $2M threshold still rates a broader population than any of those.
Anti-gaming guard: detected red flags always rate
The $2M threshold only applies when nored flags are detected. A charity of any size with detected issues — excessive salaries, related-party transactions, missing balance sheet at ≥ $2M revenue, stale filings, governance abuse patterns — stays in the “rated” bucket and receives a letter grade reflecting those flags. Small size is not a shield.
“Reasons to look carefully” tile for not-rated charities
Every not-rated profile now runs through a structural-concern detector and surfaces the findings in a dedicated tile on the charity page. The v3.3 signal catalog covers:
- Thin board — 1–2 current officers listed (below best-practice 5+)
- No chair — board has 3+ officers but no chair role designated
- High reserves for operating size — reserves cover 60+ months of expenses on an operating charity with < $500K revenue
- Government-funding concentration — more than 80% of revenue from government grants (single-stream dependency)
- Current return overdue — latest annual return is 24–36 months old (the 36-month threshold triggers the “stale data” carve-out)
Signals never penalise the score — they are structural observations drawn from the same public data and are meant to help donors decide whether to look more carefully before giving. Language is deliberately neutral (“reasons to look carefully”, not “warnings” or “concerns”) to avoid implying wrongdoing.
Profile Analytics
Each charity profile includes computed analytics that provide context beyond the rating score. These are calculated server-side using peer data from the same sector and size band.
Revenue Diversification Score
A Herfindahl-Hirschman Index (HHI) across 6 income sources: government grants, donations, trading/services, investment income, rental income, and other. Charities with 80%+ from a single source are flagged. A normalised 0-100 score indicates diversification (higher = more diverse).
Sector Percentile Benchmarking
Ranks charities on revenue, expenses, surplus, assets, and rating score within their sector and size band. Shows P25/P50/P75 markers so you can see how a charity compares to its peers.
Trend-Based Red Flags
Analyses multi-year financial history to detect 8 trend patterns: revenue decline, expense growth, surplus deterioration, reserve depletion, revenue concentration increase, staff cost spikes, asset decline, and liability growth. Each flag includes severity (low/medium/high) and a sparkline.
5-Year Trend Dashboard
Interactive bar chart showing financial metrics over time. Anomaly detection flags years with > 40% year-on-year changes in any metric, helping identify unusual financial movements.
Staff Cost Benchmarking
Compares staff costs as a percentage of total expenses against sector and size band peers. Shows P25/P50/P75 percentile gauges so you can assess whether staffing costs are typical.
Peer Comparison
Auto-groups charities by sector and size band, then ranks on 5 metrics: revenue, surplus margin, reserve months, rating score, and staff cost efficiency. Shows rank badges and links to top peers.
Methodology Limitations
- Ratings are based solely on publicly filed data - they cannot assess program quality or impact
- NZ charity reporting doesn't require detailed program vs. admin expense breakdowns
- Small volunteer-led charities are intentionally not letter-graded — see the “When We Don’t Rate” section above. Financial figures are still shown.
- Ratings don't capture qualitative factors like mission effectiveness or community impact
- Data may lag - annual returns reflect the previous financial year
Frequently Asked Questions
Data sourced from Charities Services register and annual returns. Our datasets may not be complete. Automated analysis can produce errors. If you believe any data on this page is incorrect, please contact us at hello@charitydata.co.nz.
For informational purposes only. Not investment advice.