Compare the parties.

The launch uses PREFU’s OBEGALx figures, which exclude ACC. The three-page plan allocates Budget allowances; it does not publish a revised whole-government cash and debt model.

Treasury shows OBEGALx of $4.016bn in 2028/29 and $11.693bn in 2030/31; OBEGAL including ACC is $2.062bn and $8.879bn. Staying within allowances can preserve that baseline, but the allocation arithmetic does not verify delivery. Debt’s GDP ratio peaks at 43.9% in 2027/28, while dollar debt rises until $232.763bn in 2029/30. The fall in the ratio is not an immediate fall in dollars.

The remaining operating and capital envelopes must cover later service pressures and investments. They are not spare cash or an independently demonstrated surplus.

The first future year leaves $588.5m operating. National explicitly assigns later allowances to education, social services, police pay and infrastructure. PREFU also discloses defence pressure of around $4.8bn across expenses and capital over the next two Budgets, rather than solely capital. Treasury already assumes baseline savings and underspending. Health’s first three increases follow Treasury’s non-binding indication; the further $1.55bn increment in 2030/31 has no explained estimate in this plan.

Crown-employer costs are $140.2m higher than the standalone paper. The new plan also counts a $506.1m net tax offset that the earlier paper deliberately excluded.

The original 2028/29–2030/31 employer costs are $84.3m/$94.2m/$104.3m; the plan uses $126m/$141m/$156m. The extra cost is included, so it is not missing funding. Page 3 identifies the tax offset as ESCT less reduced income tax, not gross ESCT receipts. The earlier paper warns that revenue depends heavily on new contributors’ income profiles. Ask for the changed employer assumptions and the tax model; double counting has not been established. The parental top-up’s separate $12.6m increase is explained in the later parental-leave paper by the longer entitlement.

The table puts $296m in 2026/27 with no future-Budget allowance in that column. The printed remaining amount is therefore −$296m.

The student-loan paper describes this principally as a change in the loan book’s accounting value. It is not a $296m cash payment to borrowers; slower repayments also affect cash and debt on a different path. The plan deducts it from the overall allowance envelope, but supplies no explicit bridge to the unchanged Treasury forecast. The $470m annual headline divides costs across five fiscal columns by four Budget rounds. That arithmetic works; its period needs explaining.

The plan uses $431.2m net aid savings and $583m net welfare savings. The main welfare saving is lower payments to long-term Jobseekers, rather than predominantly employment exits.

The welfare paper estimates $572m from lower payments, using PREFU recipient numbers, inflation indexing and a one-fifth offset into other assistance. Its employment-support model scales an existing programme and assumes a lower success rate, without publishing that exact rate. Implementation, hardship effects, collection yields and the tax-offset assumptions still need testing. A matching table does not certify those outcomes.

First Home Loan expansion, drug tests, some health services and other changes are explicitly assigned to existing baselines. Other pressures use future allowances.

The First Home Loan paper puts $4–6m within baselines, with no clear period; drug tests use around $1.5m annually from MSD baselines. Several screening commitments are already funded. The medicines narrative’s further $45.6m MFAT efficiency saving remains absent from the identifiable tables: they count fewer savings, not an unfunded expense. Road allocations partly rely on the National Land Transport Fund for additional costs; $100m for Mill Road route protection does not fund construction of Stages Two and Three. PREFU’s Budget 2027 capital total includes $100m of additional Chorus proceeds; National shows the $3.5bn base. That scope difference needs a bridge, not a missing-money allegation.