Decisions Government must take

The largest single saving comes from charging for land the State currently grants at no cost. If investment law forbids it, most of the improvement disappears.

Without land pricing the programme loses the bulk of its value.

If it scores as debt at face value, it does not relieve the budget; it relabels a liability and adds a discount on top.

If it scores as debt, a plain deferred payment is the cheaper choice.

An entitlement is worth what a named authority has committed to honour. Identified capacity that nobody has confirmed is worth nothing to the holder.

The coverage test refuses every issuance.

The programme deliberately builds no tender platform. It needs the existing one to handle settlement after award, without touching how bids are evaluated.

Settlement has no lawful route into procurement.

Several figures the case rests on appear only in a written brief describing a meeting whose recording the programme does not hold. A decision of this size cannot rest on an account nobody can check.

Figures that cannot be traced are struck from the appraisal.

A transferable claim on the State may fall inside the securities perimeter, which changes who regulates it and how it may be traded.

The instrument would need a different legal vehicle.

The modelling framework exists. The Iraqi data it needs — real receipts, real absorption, real timing — does not.

The fiscal position stays modelled rather than measured.

The case assumes investors would accept a discount at or below 35%. Nobody has asked them.

Above 35% the instrument is worse than paying cash competitively.

The programme promises occupied, affordable homes. Neither the published threshold nor the sites exist yet.

Outcome verification has nothing to test against.

Independent verification is the control everything else depends on. It needs qualified people who carry liability.

Verification becomes a bottleneck, or a formality.

No mechanism may be adopted without being compared against the others on identical assumptions.

Met. Fifteen candidate designs appraised on one assumption register.