Cascading, raw materials, and tiered GST recovery

Price cascading is the central mechanical worry when only part of GST on business inputs is refundable: a non-refundable wedge can stack at each stage of production, so the same economic value is touched by the "7% operational" logic more than once. One tempting fix is to zero-rate early-stage inputs—GST at 0% on the invoice—so tax never attaches to timber, grain, or steel on its way downstream. That does calm the snowball, but it reintroduces the very thing the 20/7 design tries to leave behind: fine-grained categories, border disputes, and accountant’s logic at every sale.

This note sets out a single coherent policy: keep a uniform 20% GST on the invoice everywhere, use a short statutory white list of true raw materials, and apply a two-tier refund—65% recovery on ordinary operational purchases (the usual 7% wedge) and 100% recovery on certified white-list supplies only. That choice deliberately favours 100% refundability over zero-rating for those inputs, and it can be operated either with automated e-invoicing or with a light manual two-bucket filing pattern.

Refund integrity: why tier-2 is narrow, and how up-front import GST interacts with carousel-style refund abuse, is developed alongside border policy in the risk framework.

1. What we are trying to fix

Under a flat 65% recovery on all business GST, each B2B stage retains a slice of the nominal 20% rate. On purely operational flows that is the intended 7% operational surcharge. On long physical supply chains (forest → mill → frame → builder → homeowner), the same "first mile" can be taxed in effect more than once unless something gives at the commodity end.

Illustrative story (orders of magnitude, not precise statutory arithmetic):

• Without relief on stage-one materials: The wedge can bite from standing timber through sawn timber, framing, and fit-out, producing a heavy cumulative burden on housing and food pipelines.
• With relief confined to a narrow white list of absorbed inputs: The wedge starts later—after bulk, physically identifiable feedstock—so cascade on housing and similar chains is materially shallower, while value-add and services still pay the 7% world.

The design question is not whether to protect that first mile, but how to do it without opening a fraud hole or a litigation industry.

2. Qualification: the absorption test (not "whatever feels strategic")

Standard economics separates intermediate goods (embodied in what you sell) from operating expenses (they keep the lights on but are not in the product). For 20/7, a purchase should only qualify for tier-2 (100% refund) treatment if it passes an absorption test: the thing bought must be physically absorbed into, or permanently form part of, the final sellable unit as a direct raw material.

The SaaS grey area (and why it stays in tier 1)

A developer may argue that AWS is "raw material" because the product cannot exist without it. For this policy, SaaS is not a raw material: it fails the tangible absorption test. If "digital inputs" could hit tier 2, every firm would have an incentive to re-characterise the office as a "digital factory," and Inland Revenue would inherit endless "is this code a raw material?" audits. Only physical materials that undergo a tangible transformation sit on the white list; software remains an operational tool at 65% recovery.

Pure logistics and professional services

In B2B commerce where nothing is physically transformed into a product for resale, the absorption framing has little traction. A logistics firm is moving goods, not embodying them; trucks, fuel, and tyres stay in tier 1 because there is no raw material to absorb into a sellable unit. Likewise, a law firm’s inputs (premises, IT, paper, coffee) are not turned into a physical product for sale, so those overheads remain on the 7% operational side. That is intentional: the service economy carries much of the wedge, while the white list protects the first mile of physical goods.

3. The white list (small, statutory, B2B-locked)

The model relies on a clear distinction between what drives a product’s physical existence (inputs that may qualify for tier 2) and what drives operations (the surcharge at 65% recovery). If the white list is too narrow, tax-on-tax still inflates cascades; if it is too broad, revenue from the 7% wedge erodes—the stream that funds the wider 0% income tax ambition in the full package.

Complexity is controlled by a statutory white list of categories, not ad hoc negotiation. If a good is not in a listed category, the purchase stays in tier 1 (65% recovery).

Construction: the “physical shell” test

For housing, the list should cover primary components of the permanent structure, not only “logs.” If a builder pays 20% on these and claims 100% back, the base cost of the shell stays closer to the economic cost of feedstock.

Often treated as tier 2 (illustrative):

• Primary shell: Timber (frames and trusses), concrete (foundations and slabs), steel (rebar and beams).
• Exterior envelope: Roofing (tiles or iron), brick or cladding, glass (raw panes).
• Core surfaces: Wallboard (plasterboard / GIB), plywood, flooring underlay.

Often tier 1 (operational or secondary finish):

• How the site runs: Scaffolding hire, portable toilets, power tools, builder’s insurance, project-management software.
• Secondary finishes: Paint, light switches, tapware. These are in the house but are treated as finishes rather than raw feedstock, so the 7% bracket still captures revenue from the luxury end of the build.

Manufacturing and heavy industry: the “atomic transformation” test

For B2B heavy industry, a useful question is whether the item undergoes a physical or chemical change to become the final product. The table below is illustrative, not a draft schedule; statute would compress and codify.

Illustrative statutory buckets often proposed:

1. Primary-sector extractive: Raw timber, unprocessed ores, quarried stone, raw wool or pelts.
2. Bulk agricultural (B2B): Grains, seeds, raw milk, livestock sold between registered businesses.
3. Industrial feedstocks: Raw steel, bulk resin or plastics, base chemicals in bulk.

B2B lock: White-list treatment applies when GST-registered businesses supply GST-registered businesses. A consumer buying timber at retail pays the same 20% as everywhere else; there is no consumer-facing 0% lane that invites impersonation of a business at the counter.

Hierarchy in one view: Raw resources → 0% net tax on the business leg (100% refund on qualifying tier-2 purchases); value-addition → 7% operational wedge (65% refund); final consumption → 20% consumer GST (0% refund). Zeroing tax on raw feedstocks where the test is met keeps New Zealand a place where houses, food, and hardware can be made without the same value being clipped repeatedly by the non-refundable slice.

In that light the white list is a kind of constitution of physical goods: it privileges atoms (the material NZ is built from) while the bits and processes (organisation, services, software) stay largely in the operational bucket. That targets the construction-style cascade without giving away the revenue base that services provide.

4. Zero-rating vs. 100% refund on the same shelf (why the invoice stays at 20%)

"Zero-rating" usually means a 0% GST line on the invoice: the buyer never pays the tax. "100% refundability" means the invoice still shows 20% GST; the buyer pays it and claims it all back on a return if the purchase qualifies.

Verdict for 20/7: Do not zero-rate raw materials at the invoice. Use 20% everywhere and 100% refund only for white-list, B2B, certified tier-2 purchases. That preserves one visible rate, pushes impersonation risk toward the refund gate where registration and evidence already live, and keeps downstream arithmetic aligned with the rest of the model.

5. Tiered recovery as the unified rule

The thesis condenses to three business buckets:

1. Tier 1 — operational surcharge (default): Rent, power, SaaS, legal, most goods and services. Pay 20% GST, claim 65% of the GST component → 7% effective wedge on that spend.
2. Tier 2 — B2B raw-material lock: Purchases that are on the white list and satisfy B2B and evidence rules. Pay 20% GST, claim 100% of the GST component → no wedge on that leg, stopping the worst snowball on first-mile physical inputs.
3. Tier 1E — essential-chain enhanced recovery (narrow): Certified, socially critical chain legs that are often not physically absorbed (for example, regulated medicine distribution). Pay 20% GST, claim 85% of the GST component → 3% effective wedge on that spend.

Optional refinement: smart invoices (industry or product codes) so accounting software routes each line to tier 1 or tier 2 automatically; failing that, a multi-bucket return.

Classification discipline matters: if a purchase is physically absorbed into the final sellable unit, it should generally be tier 2 rather than tier 1E. In construction, that usually means structural concrete, reinforcing/structural steel, and core roof shell materials sit in tier 2 (100%); tier 1E is reserved for certified essential links that fail strict absorption but where cascade pressure is still socially sensitive.

Targeted anti-cascade relief without breaking the refund model

The policy can reduce long-chain inflation pressure while preserving one visible invoice rate and refund-gate safeguards:

• Keep 20% on every invoice: avoid consumer-facing 0% lanes and supplier-side eligibility policing.
• Default to 65% unless proven otherwise: tier 2 and tier 1E are positive claims with evidence, not automatic entitlements.
• Use hard gates for tier 1E: product schedule, chain stage, B2B registration on both sides, and coded-evidence requirement.

Illustrative long-chain arithmetic (same base assumptions)

Directional comparison using the same modelling frame as this note: Current: base × 1.15. 20/7 baseline: base × (1.07^N) × 1.20. Refined tiering: base × (1.07^(N-k)) × (1.03^k) × 1.20, where k is the number of certified tier-1E legs after strict tier-2 reclassification of absorbed inputs.

Read directionally: tighter tier-2 classification plus a narrow tier-1E lane materially reduces cascade in long domestic chains, but does not erase all price pressure where many non-absorbed handoffs remain.

Calibration note: the detached-house row is aligned to the detailed line-item build model in Housing affordability stress-test. Other rows remain stylized chain examples for directional comparison, not sector-accounting forecasts.

6. Implementation: automated vs. manual (same policy, different workload)

Automated path (e-invoicing and coded lines)

When a registered business buys white-listed raw steel, the invoice carries a standard code. Xero, MYOB, or equivalent maps that code to the 100% refund bucket; everything else defaults to 65%. The owner does not argue grey areas at month-end—the software and the supplier code carry the classification burden, subject to audit on tier-2 claims only.

Manual path (three-bucket filing without full automation)

Most firms already file GST on purchases in aggregate. The return adds one conceptual split:

• Bucket A (tier 1): Surcharge inputs—rent, power, SaaS, legal, general supplies. Sum GST, multiply by 0.65, enter in the surcharge portion of the return.
• Bucket B (tier 2): White-list raw materials from a small set of regular suppliers (e.g. one or two timber merchants for a builder). Sum GST, multiply by 1.00, enter in the lock portion.
• Bucket C (tier 1E): Certified essential-chain non-absorbed legs (where statute allows enhanced recovery). Sum GST, multiply by 0.85, enter in the enhanced portion.

Default to 65%: Legislation can state that all input claims are processed at 65% recovery unless the taxpayer identifies the purchase as tier 2 or tier 1E with prescribed evidence (e.g. supplier industry code on the invoice, or raw-material certification on a narrow form). That makes the safe default Crown-friendly: firms that want the 100% path must positively sort a thin slice of invoices.

Supplier industry codes as a manual safety net: Primary producer or bulk mill codes nudge tier 2; telco, Microsoft, law firm default to tier 1. Spreadsheets and paper filers can use the same logic as large firms, just slower.

Comparison of effort: manual vs. automated

For a "man with a van" operation, the tiered model is comparable in spirit to today’s need to separate GST-inclusive, zero-rated, and exempt purchases—except the white list is short and tier 2 is optional upside, not the default.

7. Closing assessment

A tiny, physical, statutory white list plus B2B-only tier 2 gives most of the cascade relief zero-rating promises, without splitting the retail price into a maze of 0% / 20% lanes. 100% refund on 20% invoices beats 0% invoices on fraud, audit trail, and uniform pricing; the cost is temporary business funding of GST until refund. Default to 65% protects revenue if sorting fails. Automation moves effort to software and standards; manual filing stays viable with three buckets and a small set of tier-2 suppliers. Together, that is one integrated story: same 20% face to the economy, recovery tiers anchored by physical absorption plus a narrow essential-chain lane, and implementation that scales from spreadsheet to full e-invoicing.

For how this interacts with price risk, imports, and social licence, continue to the risk management framework.