# He Ara Angitū — Financial Case

## 1. What is being requested now

He Ara Angitū is seeking approval and funding for **Stage 0 feasibility and co-design**, with a central estimated cost of **$5.3 million**.

Stage 0 would complete the legal, data, delivery, evaluation and economic architecture required for a full business case. It is **not** a request to approve the three-year pilot or national rollout.

## 2. What each stage and scenario costs

| Stage or scenario | Current central estimate | Decision status |
|---|---:|---|
| Stage 0 feasibility and co-design | **$5.3 million** | Immediate decision sought |
| Three-year pilot | **$97.6 million over three years** | Conditional on Stage 0 |
| Universal core only | **$825.1 million per mature year** | Scenario comparator |
| Universal core plus targeted accelerator | **$875.0 million per mature year** | Leading testable structural scenario |
| Mature full national design | **$1.526 billion per mature year** | Planning comparator only |

The annual national costs are planning scenarios. They are not current appropriations, approved commitments or proof of affordability.

## 3. Why the broader cohort matters

He Ara Angitū may create economic value not only by preventing severe sustained outcomes of being **not in employment, education or training (NEET)** among a smaller high-risk group, but also through smaller improvements in engagement, attainment, transition, completion, employment continuity and earnings across a much broader group of young people.

The model aligns one mature annual programme cost with an annual eligible cohort of approximately **62,349 people**. Approximately **1,403 averted sustained-NEET cases** are modelled separately, leaving a broad non-overlap cohort of approximately **60,946 people**.

After adjustments for the affected share, domestic retention, programme attribution, persistence, deadweight, worker displacement, general-equilibrium effects and employment probability, the mature central sensitivity contains approximately **8,584 effective affected employed people**.

A **1% earnings uplift** for that affected population is equivalent to approximately **0.162% when expressed as an average across the full annual eligible cohort**. These percentages describe the same modelled value using different population denominators; they are not separate benefits.

At the **2% Treasury-aligned social discount rate**, the 1% affected-population example produces approximately **$196.0 million in lifetime earnings present value**.

These figures are modelled sensitivities, not demonstrated programme outcomes.

## 4. Three financial perspectives

### Participant perspective

Verified participation may provide immediate financial support and protected Learning Capital. Under the central assumptions:

- five-year protected contributions total approximately **$9,750 per participant**; and
- the projected nominal closing balance is approximately **$10,682**.

Investment returns are uncertain and are not guaranteed.

### Crown fiscal perspective

The model counts attributable tax, ACC, benefit and valid student-loan effects.

The **modelled student-loan fair-value offset using the official cost rate** is approximately **$58.4 million per annual cohort**. This applies an official fair-value cost rate to modelled student-loan borrowing displacement; it is a scenario estimate, not an official government forecast.

The model also includes approximately **$52.6 million** in consequential account-administration and induced-study costs, leaving a **net Learning Capital Crown offset of approximately $5.8 million**.

### National social and economic perspective

The model values causal, non-overlapping benefits to people and New Zealand. Social value does not automatically become Crown cash and cannot be used as though it directly funds the programme.

## 5. What the mature national model currently concludes

For the mature full national design at the **1% affected-population earnings sensitivity**, using the **2% lead case**:

- **Full-expenditure social-investment BCR:** **0.382x**
- **Conventional net social ROI:** **–61.8%**
- **Crown fiscal coverage:** **15.3%**
- **Conventional net Crown fiscal ROI:** **–84.7%**
- **Net Crown fiscal cost:** approximately **$1.293 billion**

A BCR of 0.382x means the model includes approximately **$0.38 of social benefit for each $1.00 of full public expenditure** under this sensitivity. It does not mean that the programme returns 38 cents to the Crown.

The broader earnings pathway materially improves the economic case, but modest earnings sensitivities do not make the mature full national design positive on the full-expenditure view.

The mature design reaches full-expenditure social break-even at an affected-population earnings uplift of approximately **5.8%** under the 2% social case. Crown fiscal break-even is substantially harder, requiring approximately **21.2%** under the same 2% lead case.

The detailed model also shows 4% intermediate and 8% high discount-rate sensitivities.

## 6. Alternative treatment of participant transfers

The detailed workbooks retain a:

> **Provisional alternative economic treatment of participant transfers — not Treasury-confirmed**

That technical sensitivity distinguishes the cash-transfer component of participant payments from national resource costs. It may produce a materially different economic result, but it does not yet incorporate a validated marginal cost of public funds, deadweight cost of taxation or complete behavioural treatment.

It is therefore **not used as the principal public value-for-money result**.

A transfer-adjusted economic result would not:

- reduce the cash appropriation required;
- make participant payments fiscally free;
- represent Crown profit; or
- prove that national rollout is affordable.

The method requires Treasury or independent public-sector economic confirmation.

## 7. The leading testable structural scenario

The universal-core plus targeted-accelerator scenario has a lower annual cost of approximately **$875.0 million** and reaches full-expenditure social break-even at approximately:

- **2.27% affected-population earnings uplift**; or
- approximately **0.405% expressed as an average across its full exposed cohort**.

Under the current **100% retained high-risk-effect assumption**, its full-expenditure social BCR is:

- **0.687x** at a 1% affected-population earnings uplift;
- **1.180x** at a 3% uplift; and
- **1.672x** at a 5% uplift.

This scenario is financially promising because it assumes substantially lower universal cost while retaining strong high-risk effects through targeted support. That retained effect has **not** been demonstrated.

The workbook therefore tests retained-effect sensitivities of **25%, 50%, 75% and 100%**.

The model also does not yet prove:

- a broad earnings effect;
- long-run persistence;
- delivery costs at scale;
- genuine student-loan displacement;
- labour-market responses; or
- the absence of material displacement, oversupply or wage-suppression effects.

## 8. Stage-gate safeguards

National rollout remains conditional. Stage 0 and the pilot must establish:

- legal, data-sharing and operational feasibility;
- causal changes in engagement, attainment, transition, completion and sustained-NEET outcomes;
- the distribution of effects across different groups;
- actual delivery and administration costs;
- genuine Learning Capital use and student-loan displacement;
- equity, implementation quality and absence of material harm;
- labour-market capacity, displacement and general-equilibrium effects; and
- performance against pre-agreed social, fiscal and stage-gate thresholds.

Failure of the causal, financial, equity, implementation or harm tests can lead to redesign, narrower targeting, delay or stopping the programme.

## 9. Downloads

- **Professional Financial Model:** the complete technical model, including scenarios, assumptions, formula-linked outputs and validation controls.
- **Simplified Guided Reader Workbook:** the same controlling model with a shorter guided entry point and plain-language public summary.
- **Website Financial Metrics CSV:** structured public metrics for website cards, tables and data-driven components.
- **Private Technical Audit:** reconciliation, formula checks, methodology notes and quality-control results.