These are not decorative caveats added to appear thoughtful. Any one of them could weaken, narrow or kill the proposal if Stage 0 and the pilot cannot answer it.
Students already doing well may receive public money for behaviour they would have shown anyway. Others may experience the system as transactional or resent peers receiving more support.
Test: compare motivation, participation quality, substitution and distribution—not just enrolment.
A weak system will be exploited. An overbuilt system may punish honest students and consume school time.
Test: proportionate evidence, random audits, human review, appeals, burden measurement and fraud sensitivity.
Much of the wider cohort may receive support without changing behaviour or outcomes. That is why the universal intensity and break-even hurdle matter.
Test: causal incremental effects and alternative lower-cost designs.
The accelerator could expose need publicly, create resentment or encourage schools to classify students around funding.
Test: confidential criteria, appeals, participant research and monitoring for labelling or perverse incentives.
Investment returns are uncertain. Approved-use rules may be too narrow. Account administration may absorb value.
Test: legal design, fee caps, hardship access, portability, investment choices and real participant use.
Sectors could encourage oversupply to weaken bargaining power or gain privileged access to recruits.
Test: pooled funding, independent forecasting, union and worker representation, wage and job-quality monitoring.
Evidence for mentoring, incentives, participation and pathway support does not prove this exact combination works in New Zealand.
Test: independent pilot design capable of distinguishing implementation from causal effect.
At the 1% affected-population earnings sensitivity and 2% social discount rate, the full-expenditure BCR is 0.382x and net Crown fiscal cost is about $1.293b.
Response: do not market national rollout as affordable. Test lower-cost structures and publish the evidence hurdles.
Attendance support, transport, counselling, careers advice, teaching quality, income support or targeted mentoring might deliver more value with less machinery.
Test: compare alternatives in Stage 0 rather than defining success as preserving the original concept.
Payments create constituencies. Platforms create sunk costs. Ministers dislike admitting failure.
Test: expiry clauses, staged appropriations, independent reporting and predetermined stop rules.
That is normal and dangerous. Personal commitment can create confirmation bias and resistance to unwelcome evidence.
Safeguard: the founder must not control evaluation, adjudicate his own model or hold a veto over redesign and stopping.
What exactly causes the failure—price, burden, stigma, incentives, capacity, capture or something else?
Bring data, research, operational experience or a clearly reasoned counterexample.
Should the design change, narrow, be compared with another option, or stop?
