Legalising Groceries

The most protected aisle in retail

The Commerce Commission says two chains dominate the market and earn outsized returns. Both claims are contestable — and there is exactly one honest way to test them.

Start with the basic principle. Should it be legal to build a supermarket? Or should a supermarket be allowed only where a council has decided, in advance, that the neighbourhood's existing shops can spare the customers?

New Zealand has, for the most part, chosen the second answer. This site maps the consequences.

The Commerce Commission's 2022 grocery market study concluded that competition between the major chains was "muted", put their share of the market at around 82 percent, and estimated returns on capital of about 12.9 percent against a cost of capital of 5.5 percent — excess returns in the order of $430 million a year. Those figures have been repeated often enough to read as facts. They are estimates, and contestable ones. The market-share number depends on where the market's edges are drawn: Chemist Warehouse competes hard across a wide slice of the supermarket range, as do butchers, greengrocers, specialist food retailers and the meal-kit companies, and none of them count in the denominator. The excess-returns number is only as strong as its cost-of-capital assumptions — reasonable disputes about asset betas and the treatment of leases move it a long way. Eric Crampton and the New Zealand Initiative have been sceptical of both numbers since the study landed.

But the Commission's headline claims do not have to be settled on paper, because entry itself is the measurement. Picture consumer losses from concentration on a line running from zero to very large. Entry is costly — sites, stores, logistics, procurement, all real resources — and New Zealand's planning system adds a thick layer of regulatory cost on top. A chain enters only if the prize sits far enough along that line to cover the whole bill; entry, or its absence, therefore reveals where on the line the truth sits. Under today's rules, absence reveals little: the prize would have to be enormous to cover the regulatory bill, so non-entry is consistent with losses anywhere from trivial to substantial. Strip the regulatory cost to something minimal and the test sharpens. If entrants come, consumers win directly. If nobody comes even when entry is cheap, the honest inference is that losses from concentration are smaller than assumed — too small to attract anyone's capital, and therefore very unlikely to clear the far higher bar that would justify Commerce Commission structural intervention, which carries its own costs and its own ways of going wrong. Taking Demsetz seriously about what counts as a barrier, the only entry costs left standing at that point are real resource costs — the kind no competition authority can litigate away.

There is a subtler payoff, from Demsetz and William Baumol's contestable-markets work: entry does not have to happen to do its job. Where a rival could open across the road, an incumbent that prices like a monopolist is writing the entrant's business case; the credible threat disciplines prices without a single new store. Supermarket entry carries sunk costs, so the discipline is loosened, not perfect — but it exists only where entry is possible at all. Rules that make entry practically illegal do not just block the marginal store; they retire the threat everywhere, and with it the discipline. The policy prescription is the same on every reading: make entry legal, and watch what happens — or watch prices change because nobody needs to.

One piece of evidence needs no spreadsheet: the land. The Commission's first recommendation, ahead of anything about wholesale access or codes of conduct, was to change planning laws to free up sites, ban restrictive covenants, and watch the land-banking. The covenants were banned promptly; the planning half waited three years, until the Government's 2025 supermarket fast-track — Chapter VIII takes up what that package does and does not change.

Meanwhile, Marko Garlick counts, by his quick tally, about 22 Woolworths stores sitting next to another Woolworths nationwide — the famous Napier pair face each other across the street. The store data behind these maps lets that be checked for the six metros covered here: three Woolworths–Woolworths pairs sit within 500 metres of each other — in Johnsonville (259 metres apart), at Westgate/NorthWest (359 metres), and in central Lower Hutt (396 metres). Holding duplicate sites is expensive, and some pairs will have innocent explanations — chains acquired, legacy brands retained, leases that outlast strategy, deliberately different formats. But the pattern pays as a strategy only where the number of places a competitor could legally open is small enough to be worth cornering. Whatever one makes of the Commission's return-on-capital arithmetic, the two large chains' willingness to carry duplicate sites is suggestive evidence about how scarce legal supermarket sites are — and it points the same way as the rulebooks mapped below.

None of this is a story about anyone breaking the rules. It is a story about the rules. So the useful exercise is to read them — all of them — and draw the map the rules imply.

What the rulebooks actually say

District plans don't ban supermarkets. They grade them — on a ladder that runs from "yes" to "don't bother asking".

Every district plan assigns each activity, in each zone, an activity status. The maps on this site classify a new full-line supermarket against each zone's rules, using a reference store of 4,000 m² gross floor area — a standard Woolworths or PAK'nSAVE, and the midpoint of the 3,000–5,000 m² full-line range. Where a plan's threshold falls inside that range (Auckland's neighbourhood centres flip from discretionary to non-complying at 4,000 m²; Kāpiti's airport rations at 3,000 m²), the zone popup gives the split. One measurement warning: plans cap in different units — gross floor area, gross leasable floor area (Christchurch), building footprint (Hamilton), tenancy area — and these are not interchangeable; popups use each plan's own term. A future update will let readers toggle the reference size — an Aldi-format store of roughly 1,500 m², the 4,000 m² full-line store, a warehouse club — and watch the maps redraw; Chapter V's Aldi test previews how much that choice matters. The ladder runs:

One caveat on the bottom rungs: since late 2025, a qualifying grocery project can vault them through the Government's fast-track, which can consent even a prohibited activity — Chapter VIII covers what that does and does not fix. The ladder above is the law for everyone else, and for every application outside the lane. And the dots on the maps below carry the more important caveat in reverse: every existing store was sited under decades of these rules, long before any fast-track. The maps show the cumulative output of tight restraint, not a market that chose its own sites — a point Wellington's section makes concrete.

One more warning: the ladder reads as if each rung down is stricter — Restricted Discretionary milder than Discretionary. Auckland's own data says otherwise. Research in the council's Chief Economist Unit by Eilya Torshizian, written up in Housing Supply, Choice and Affordability (2015, tool 15), tracked eleven years of Auckland consents (to 2015, across all activity types — not supermarket-specific, and under plans since replaced) and found an application was about 20 percent less likely to be granted where the activity was Restricted Discretionary than where it was full Discretionary — same activity, same kind of location, significant at the 99 percent level, after controls. One city, one era; but it is the best evidence going, and it cuts against the ladder's implied ordering. The "restriction" turns out to restrict the applicant more than it restricts the council. Read the yellow and orange on these maps as different flavours of consent risk, not a reliable ordering.

Two features of the rulebooks deserve attention before the maps make sense.

First, "supermarket" is a defined term, and the definitions disagree. In Lower Hutt a supermarket starts at 350 m². In Hamilton and Kāpiti it starts above 1,000 m². Porirua requires foodstuffs to be more than 80 percent of retail floor space; Upper Hutt requires more than 90 percent — strict enough that a full-line store with a generous liquor and general-merchandise offer might not be a "supermarket" there at all, and would fall into the harsher "large format retail" rules instead. A national entrant designing one standard store faces nine rulebooks that cannot agree on what it is.

Second, the centres hierarchy. Most councils rank their shopping areas: the city centre at the top, then the big suburban centres, then local shops. Roughly two-thirds of councils run policies like this, on the Productivity Commission's count. The ranking is not just a description. The plan protects it: before a new store can open outside the favoured centres, the council asks what the store would do to the centres above it.

The law here is odd, and worth stating slowly. Since 1997 the RMA has told councils to ignore trade competition when deciding consents — protecting existing shops from a rival is not meant to be planning's job. But the courts (the Discount Brands line of cases) let councils consider effects on a centre's "vitality and viability": its liveliness, its foot traffic, its prospects. The courts maintain that a real line separates the two — ordinary trade competition excluded, significant consequential effects on communities and centre amenity allowed in — and as law, it does. The trouble is economic: the distinction rarely survives the arithmetic. A new supermarket's customers must come from somewhere, and wherever they come from is some existing centre's vitality. Refusing a store to protect vitality and refusing it to protect incumbents are the same refusal with different paperwork. In practice the test becomes demographic: applicants commission studies showing enough population growth to feed the new store without starving the old ones. Entry is welcome only where it displaces nobody — which is to say, the Schumpeterian kind of competition, where the better store wins the worse store's customers, is precisely what the test screens out. The competition test did not die in 1997. It was rezoned.

And the ranking usually does not start with the council. In four of the five regions mapped here — Auckland, Waikato, Wellington, Canterbury — the regional policy statement, the planning layer above the district plan, requires district plans to run a centres hierarchy; Wellington's and Waikato's name the protected centres outright. Dunedin's is the exception, a genuine council choice. Chapter VII returns to this layer, and the fixes in Chapter VIII have to reach it: telling councils to delete the hierarchy from their district plans achieves little if the layer above orders it put back.

New Zealand's zoning and consenting processes have treated competition as a harm to be mitigated rather than a benefit to be sought.— Eric Crampton, "Legalising groceries", Newsroom, 8 March 2022 — written the morning the market study landed

The atlas

Six metros, nine plans. Green means the land use is lawful — the building, transport and pipes still need their own consents. Everything else is a negotiation.

Click any zone for the controlling rule. Dots mark existing grocers, shaded and sized by store footprint where OpenStreetMap has a building outline (large ≳2,000 m²; wholesale-format grocers like Moore Wilson’s included) — a proxy for scale, not measured GFA. The Auckland, Christchurch and Hamilton maps also carry a toggleable layer of published water and wastewater constraint areas — the second gate, discussed in Chapter VI. Unshaded urban land is residential, rural, or open space — where a supermarket is non-complying or discretionary-by-default with a policy framework stacked against it. Each map shows only stores inside that council's boundaries. Some existing stores sit on land the maps leave unshaded — residential edges, special-purpose zones (the Woolworths inside Auckland Airport's zone; Waiheke's stores, governed by the separate Hauraki Gulf Islands plan). These are mostly lawfully established uses continuing under the RMA's existing-use rights: the store may stay, but could not be consented there today — the ratchet in miniature. These maps classify the land use; building-design consents, transport rules, and overlays stack on top. Compiled from the operative plans as at July 2026; a handful of rule numbers flagged in the popups could not be verified against click-walled e-plans. Screening tool, not legal advice.

Auckland: green centres, rationed precincts

Purple dashed areas are precincts that cap or forbid supermarkets on top of the base zoning. Auckland's Heavy Industry Zone (black) is the only place in these nine plans where applying under the district plan is itself forbidden.

The Unitary Plan reads generously at first: retail is permitted in the city centre and every metropolitan and town centre. Then the qualifications start. Supermarkets are singled out for tighter treatment than generic retail in five zones. In the Local Centre zone a store over 2,000 m² is restricted discretionary; in Mixed Use, discretionary; in a Neighbourhood Centre, a store over 4,000 m² is non-complying. In the General Business zone — the big-box zone, where generic large format retail is permitted — supermarkets specifically require consent, assessed for effects on the function of other centres.

The sharper action is in the precincts. A full-text sweep of all 232 published precinct documents found about twenty that cap or forbid supermarkets outright: Wairaka (one supermarket, at most 1,500 m², to protect Pt Chevalier's shops — the policy says so in terms), Westgate ("one only", to 5,500 m²), Hobsonville Corridor (no more than two), Silverdale 2 (supermarkets non-complying across 47 hectares), the Albany campus precincts, Takanini, and more. Auckland does not merely zone supermarket sites; in its growth areas it counts them.

Hamilton: permission means proving there was no alternative

Only the central city is green. Every other commercial zone routes a supermarket through an out-of-centre test.

Hamilton is the purest centres-hierarchy plan in the country — and it is not freelancing: the Waikato Regional Policy Statement names the protected centres and directs that district plans "shall" manage commercial development accordingly. A new supermarket is permitted in the downtown and city-living precincts of the Central City Zone, and nowhere else. Everywhere else that isn't a flat no — the suburban centres, the large-format zone, even ordinary industrial land — the consent test asks the applicant to show that "suitable land is not available within the business centres" and that the proposal reinforces the central city's primacy. The plan's explanation of its own five-tier hierarchy attributes the CBD's underperformance to an "unplanned dispersal of retail" — the diagnosis is competition, and the prescription is less of it.

The Base — the sub-regional centre Waikato-Tainui built and the council fought — operates under a total retail cap of 103,700 m², with a sub-cap of 34,300 m² on small-shop space to protect the CBD's specialty retail. And in the greenfields, Peacocke's plan provisions designate the local centre as "the only location for a supermarket within the Peacocke Structure Plan area", capped at 4,500 m² a tenancy. One growth cell, one supermarket site, chosen in advance.

Wellington City: permissive centres — but only since last year

The 2024 plan permits supermarkets in every centre zone with no supermarket-specific cap. The dots show the store network the old plan built.

Wellington's new plan is, on paper, the most liberal of the nine in its centres: a full-line supermarket is a permitted use in the City Centre, both metropolitan centres (Johnsonville and Kilbirnie), and every local and neighbourhood centre, subject to a design consent for the building. The supermarket-specific rules all sit outside the centres: in the Mixed Use zone a supermarket over 1,500 m² needs consent against a test of whether it would significantly harm the "viability and vitality" of the city centre or any other centre; industrial land is non-complying.

Read the map with its age in mind. The liberal regime is one to two years old — the intensification provisions were decided in March 2024 and operative from mid-April, the standard-track land-use rules from 14 July 2025 — against a restrictive baseline that ran from 2000 to 2024. Under the old plan, Plan Change 73 (operative 2014) made any supermarket over 1,500 m² in the out-of-centre Business 1 areas fully discretionary, assessed against whether it would harm the "viability and vitality" of the Golden Mile or any suburban centre (Policy 33.2.2.4); industrial land was an effective ban; and inside the Johnsonville centre — where supermarkets were at least permitted — heights were capped at 12 to 18 metres with anything taller non-complying. That envelope is why the consented Johnsonville mall redevelopment of 2017 was a two-storey scheme, and why the 11-storey mixed-use tower proposed in 2021 had to route around the plan entirely, through the COVID-19 fast-track. The new Metropolitan Centre zoning allows 35 to 42 metres on the same blocks.

The store network on the map, in other words, is the old plan's equilibrium, not the new plan's. Today's outcomes reflect twenty-four years of the previous rules; the new rulebook has had at most two years to work, and its test comes wherever supportive infrastructure and the new zoning overlap — Johnsonville, with rail, a motorway and 42-metre permissions, is the obvious place to watch. The remaining constraints are site assembly in a city short of large flat parcels, and water infrastructure with documented capacity problems and no published constraint map to plan around.

Lower Hutt: one green zone

The operative plan permits a big supermarket only in the CBD. Petone caps retail at 3,000 m² and demands a five-year economic impact study above that.

Lower Hutt still runs on its 2004 plan while the replacement grinds through hearings — a point worth being precise about, because Lower Hutt earned a national reputation for enabling housing. Plan Change 43 (operative 2020–21) and the intensification change PC56 (operative September 2023) genuinely liberalised the residential rulebook; their only commercial effect was to fold the old suburban commercial zones into a Suburban Mixed Use zone with taller height limits, leaving the retail rules untouched. The full replacement plan, notified in February 2025, is mid-hearings — the business-zones stream was heard in May 2026, with decisions to follow — and under the RMA only its heritage-protection rules have any legal effect yet.

On groceries, the operative rulebook is the region's most restrictive on paper. A large supermarket is permitted outright only in the Central Commercial area — but the Suburban Mixed Use centres are a quieter opening than they look: retail above 500 m² is restricted discretionary with no upper size bound, discretion confined to the store's scale, intensity, and functional requirements — no centres-vitality test — and applications presumptively non-notified. That is how Wainuiomata's full-line Woolworths sits comfortably where it does, and it is arguably the cleanest out-of-CBD consent path in the region.

Petone's big-box strip permits stores only up to a cumulative 3,000 m²; above that, the application must include an economic assessment of effects on the Jackson Street shops and the Lower Hutt CBD "over a minimum time period of 5 years" — a retail-impact study, required by rule, in 2026. Seaview and the other industrial zones are non-complying, with one grandfathered exception written for the Naenae supermarket site — a permission that outlived its store. Naenae has had no supermarket for years; the rule still points at the site where one used to trade. The proposed plan would liberalise much of this — a 1,500 m² permitted baseline in centres and light industrial — but a store built today answers to the old rules, and every grocer on this map already does: the dots are legacies of plans older still.

Porirua: what codified permission looks like

Porirua's 2025 plan writes supermarkets into the rules by name — mostly to say yes.

Porirua's brand-new plan is the tidiest of the nine. Supermarkets are permitted by name in the Metropolitan Centre and in the Large Format Retail zone — where, in a nice inversion, it is the small store under 450 m² that needs consent. In local and neighbourhood centres and the mixed-use zone, larger stores are restricted discretionary with public notification precluded: a consent hurdle, but not a years-long objection war. Industrial land is a hard no. The one bespoke restriction is at Plimmerton Farm, whose commercial centre is defined to exclude large format retail "except one supermarket and one trade supplier".

Upper Hutt: the accidental liberal

The Mixed Use zone permits large format retail with no size cap — the most enabling supermarket zoning in the Wellington region.

Upper Hutt's Mixed Use Zone permits retail and large format retailing outright, with no floor-area cap — subject to a landscaping standard, and with one carve-out now drawn on the map: the St Patrick's Estate development area overrides about 19 hectares of that permitted land with a restricted-discretionary gate and the plan's only mandatory retail-economics assessment. Even so, nothing else in the region comes close. The city centre permits the use with a design consent for the building; Silverstream's town centre allows up to 1,500 m² before a consent that asks whether the store would "undermine the role and function" of the CBD. Even the residential zones default to discretionary rather than non-complying. Whether the zone's land market can actually supply a five-thousand-square-metre site is another question; the rulebook, unusually, is not the problem.

Kāpiti: one green pocket, and a rationed airport

Kāpiti's only green is the Paraparaumu metropolitan centre core. The airport zone caps the district's grocery competition by rule.

Kāpiti's green is confined to a single pocket: the core precincts of Paraparaumu's metropolitan centre, where the use is permitted but wrapped in precinct building rules. Everywhere else in the district requires consent or forbids the store outright. The airport zone deserves a museum plaque. Its rules allow "one only supermarket with a maximum gross floor area of 3,000m²" as a discretionary activity; any other supermarket is non-complying; more than one store between 151 and 1,500 m² that "retails groceries" is non-complying; and the zone permits exactly one department store (AIRPZ-R28). The brand rule lives up the road: Paraparaumu's metropolitan-centre Precinct C allows one department store, of at least 3,000 m², whose brand "must not be in the District" (MCZ-R14). A schedule of permitted competitors, by format — and, in the district centre, by brand — written into a district plan.

Christchurch: the quota system

Permitted in the centres — with named-site exceptions, a central-city one-supermarket block, an airport quota already filled, and Homebase's ban-until-2031.

Christchurch permits supermarkets in its town centres, local centres, and city centre without size caps, and its Large Format Retail zone is genuinely open — with one loud exception. At Homebase on Marshland Road, no supermarket over 1,000 m² may open before 4 October 2031, save for a single negotiated exception of 4,300 m². That condition exists because commissioners, weighing a plan change, concluded a supermarket there could threaten the viability of The Palms mall five minutes down the road — and wrote the protection of The Palms' anchor tenants into the plan. The airport zone permits one supermarket of up to 2,700 m²; Spitfire Square's New World has used the quota, and the rule's own commentary cites "distributional effects on nearby commercial centres" for why there will be no second. The central-city mixed-use zone allows one supermarket, up to 2,500 m², in one named block.

Dunedin: permissive, with a transport toll-gate

Permitted in the CBD, the centres, and the big-box zones — where a store under 1,500 m² is, remarkably, non-complying.

Dunedin's plan barely says the word: "supermarket" appears four times in the whole document, and there are no site-specific caps anywhere. A full-line store is permitted in the CBD, the principal and suburban centres, and the trade-related and large-format zones — where the plan directs "large supermarkets" and where, in the strangest inverse cap on these maps, a food store under 1,500 m² is non-complying. Small-format grocers — the Aldi model — are locked out of the very zones built for grocery. Every store over 250 m² everywhere trips a "high trip generator" consent, but discretion is limited to transport matters. The centres-hierarchy policing is done with a scalpel: no single activity may occupy more than half of a neighbourhood centre, and dairies are capped at 200 m² so that none grows into "a destination supermarket".

Adding it up: permitted land per resident

Having seen the geography, the arithmetic. How much land does each district actually zone for a full-line supermarket, per person living there?

Land in zones where a full-line supermarket is a permitted use, per resident (June 2024 population estimates), with hard, mapped exceptions netted out (Auckland's quota precincts; Upper Hutt's St Patrick's Estate development area; Christchurch's Homebase, non-complying for supermarkets until October 2031; Lower Hutt's Jackson Street strip). Exceptions without a published boundary — Christchurch's central-city one-supermarket block, named-site rules — are disclosed in popups but cannot be netted. A rough comparator, not a measure of developable sites: it counts zoned land, not vacant land; the second column adds the consent-required commercial and industrial zones as mapped (Auckland's figure is inflated by its huge discretionary-by-default Future Urban Zone; Hamilton's by industrial land where the consent test demands proof that no centre site was available); the discretionary-by-default residential and rural catch-alls are excluded everywhere. Independent replications using different area projections may differ from these figures by a few percent; the published numbers follow the cos-latitude planar method in the data folder and reproduce exactly from the shipped GeoJSONs. *Dunedin's "permitted" land uniformly carries a transport-only RD consent for any store over 250 m² — a district-wide gate that applies on top of every colour, disclosed here rather than mapped, because mapping a uniform overlay would erase the within-city distinctions the map exists to show. Auckland has the most permitted supermarket land in absolute terms — 11.1 km² — and nearly the least per person.

One supermarket only

The clearest rules in these plans are not about where supermarkets may go. They are about how many there may be.

Reading nine plans end to end turns up a genre: the supermarket quota. Thirteen provisions across these plans set a numeric limit on the number of supermarkets in an area — usually one — and often a size for it. This is not a metaphor for restrictive zoning. These are counting rules.

Homebase belongs on the list too, as the limiting case: a dated prohibition — no supermarket before 4 October 2031 — imposed to protect a specific competitor, softened on appeal to permit exactly one store of exactly 4,300 m².

If those shopping malls agreed to that privately – to geographically not compete for ten years in East Christchurch – that's cartel conduct, collusion, a criminal offence. But if the council does it publicly, off their own accord, though [sic] resource management processes, we call that good urban planning.— Marko Garlick, recounting his comments on RNZ's The Panel, in "More stories of how councils accidentally on purpose banned competition", Yes In Our Backyards, 26 February 2025

Try to write down the economic model in which these quotas help consumers. The council must know the efficient number of supermarkets per suburb, decades ahead, better than entrants risking their own money. It must be right about catchment growth, formats not yet invented, and preferences not yet expressed. And it must gain more by preventing a redundant store than it loses by preventing a competing one. Nobody believes all three. The quotas survive because their costs are invisible — the store not built, the price not cut — and their beneficiaries are the incumbents already inside the ring.

Case files

The rules above are the system working as designed. Here is what it looks like case by case.

Homebase, Christchurch, worked through. In 2020 the Homebase centre sought a plan change to expand onto adjacent vacant land. The council's concern was that a supermarket there could "seriously threaten the viability" of The Palms — a mall five minutes away, owned at the time by Australian private equity, anchored by a Woolworths with roughly $6 billion in national revenue behind it. Commissioners granted the rezoning with a condition: no supermarket before 2031. On appeal the parties settled on the one-store, 4,300 m² exception, plus staging caps on total retail. The economic evidence traversed local incomes, spending, and travel patterns to establish whether East Christchurch could support the extra floorspace without pinching The Palms. Whether East Christchurch shoppers might simply prefer having two places to buy milk was not the question before anyone. That is the tell for the whole regime: the inquiry runs on protecting suppliers of retail space, and consumer interest appears nowhere in it.

Costco, Westgate — what entry looks like when it works. New Zealand's one large-format entrant this century took three years and three and a half months from announcement (June 2019) to opening (September 2022) — and the consenting was the fast part. Auckland Council granted the resource consents non-notified, in about three months, because Costco bought into the one sub-precinct of the master-planned Westgate precinct where a 14,740 m² single tenancy (the consented figure, as reported from the application) was a permitted activity — fitting under the 15,000 m² tenancy cap by 260 square metres. The sharper detail is definitional: in the town-centre sub-precinct next door, a supermarket is capped at 5,500 m². Costco's warehouse format landed in the big-box sub-precinct as generic large format retail rather than a "supermarket", so the cap that bound was 15,000 m², not 5,500. Definitions do work. The Overseas Investment Office approval took roughly eight months for a $23 million land purchase on sensitive land — apparently through adjacency to the neighbouring reserve, on the published summary — and the regulator now features the decision among the case studies on its own website.

What actually bound: the store needed a purpose-built precinct to exist at all (the product of nearly three decades of private master-planning at Westgate); two years of COVID-era construction; and — still unresolved — the West Auckland licensing-trust monopoly, which means the country's largest grocery store cannot sell wine. The sequel makes the pattern explicit: a consent obtained for a Costco at Rolleston by the landowner, not Costco, sits unused; the confirmed second store, at Drury, again buys pre-planned land inside a master developer's fast-track-consented project. Entry happens where someone has already built the permission. The atlas above maps how rarely that is.

The shorter files (drawn largely from Garlick's case notes and the decisions cited in Sources). Wairau Valley, Auckland: a PAK'nSAVE fought consent litigation — much of it from what is now Woolworths — for 18 years; the completed building sat empty from 2005 to 2009 awaiting judgments. Stonefields, Auckland: after nearby business associations objected, the town centre was limited to 4,500 m² of retail including "a superette of no more than 500m²" — the court heard that a real supermarket runs 4,000–5,000 m². Paerata Rise, Auckland: consent for a supermarket was refused as contrary to a precinct framework protecting Pukekohe's town centre, seven kilometres away; on appeal, the consent order recorded the applicant's undertaking to seek plan-change restrictions over its adjacent commercial land — a negotiated non-compete, run through public process rather than private covenant, extracted as the price of entry. Queenstown: the Frankton PAK'nSAVE was asked, along the way, to read as a fruit-packing shed and to justify its energy efficiency; the appeal was opposed by the owner of a competing centre 500 metres away that already hosted a Woolworths. And IKEA was kept out of Sylvia Park in 2008 on the ground that it would be too popular for the roads; New Zealand's first IKEA opened in 2025 — at Sylvia Park.

Aldi's Australian playbook is to open, on average, about 400 metres from an existing supermarket, and to say plainly that it intends to take the incumbents' customers. An obvious question is whether the format's small footprint — a store of roughly 1,500 m² — simply slips under the thresholds and never meets the vitality test at all. Run it through the nine rulebooks and the answer is: often, yes. At that size the store is permitted outright in Wellington's mixed-use and commercial zones (the cap is 1,500 m²), Porirua's mixed-use and large-format zones, Upper Hutt's town centre and mixed-use zones, Auckland's local centres (permitted to 2,000 m²), and Christchurch's town, local and city centres — though not its neighbourhood centres, which cap ground-floor tenancies at 1,000 m² GLFA and push even the Aldi format to consent. Where the format is permitted, the vitality inquisition never convenes.

Which makes the rules that still catch it the revealing ones. Hamilton's definition is format-blind — anything over 1,000 m² of footprint is a "supermarket", facing the same prove-there-was-no-alternative test everywhere outside the CBD. Dunedin's big-box zones run a minimum, not a maximum: food retail under 1,500 m² is non-complying, a floor that excludes the Aldi format by design. Homebase's 1,000 m² cap blocks a small grocer as surely as a large one. And Kāpiti's airport zone makes any second grocery store of between 151 and 1,500 m² non-complying — a quota drawn around precisely this format. The small store dodges the size gates. What it cannot dodge are the rules written to count grocers rather than square metres.

The second gate: pipes

A legal site still needs a wastewater connection. In parts of three cities, that is now the binding constraint.

Zoning is the first screen, not the last. In Auckland, under Watercare's capacity maps and Auckland Council's practice note as published in 2024–25, a supermarket-scale wastewater connection is out of reach in different ways in different places. In the "no capacity" areas — the Hibiscus Coast, Warkworth, Wellsford — new resource consents requiring a wastewater connection are not currently being granted at all, pending plant upgrades. In the "limited capacity" areas — Ōtara/Papatoetoe, Favona, Beachlands, East Auckland, parts of Waitākere, the lower North Shore — the council's guidance is that development demanding more than roughly five household-equivalents will be declined at consent. A third tier of "monitored" areas (Glendowie, Ōtāhuhu, East Tāmaki, Epsom/Newmarket) may see larger developments declined case by case. Capacity positions change as upgrades land; the constraint layer on the map carries Watercare's own categories and timelines, and any shortlisted site warrants a current check. In Christchurch, the post-quake vacuum sewer areas of Shirley, Aranui and Prestons are effectively closed to demand-adding development, with an unfunded replacement bill in the hundreds of millions. Hamilton publishes suburb-level capacity warnings and can refuse a network connection even to a permitted development.

Where councils publish the constraint polygons, they are now drawn on the maps above: toggle the "Infrastructure constraints" layer on the Auckland map (Watercare's network capacity areas, by severity), the Christchurch map (wastewater capacity constraint and limited sewer discharge areas), and the Hamilton map (the council's wastewater constraints layer). Wellington is the gap that proves the point: both its treatment plants run at capacity and the network constraints are real, but no constraint map is published anywhere — an entrant cannot find out where the no-go areas are without asking, site by site. The general point: a purple precinct rule can be appealed; a full pipe cannot.

The floor above the plans

District plans are the visible rulebook. The layer above decides what they must say — and insiders work at that layer.

There is also a floor above the plans. District plans must "give effect to" regional policy statements, and in four of the five regions mapped here the centres hierarchy is written into that higher layer. Auckland's regional objective directs commercial growth into "a hierarchy of centres" (AUP B2.5.1). The Waikato RPS names the protected centres in a table and directs that district plans "shall" manage commercial development accordingly, Hamilton's CBD primacy spelt out. Wellington's Policy 30 lists the region's protected centres by name. Canterbury's Chapter 6 directs new commercial activity "primarily… to the Central City, Key Activity Centres, and neighbourhood centres". Only Otago leaves the hierarchy to the council.

So national direction that reached only district plans would not finish the job — section 75(3) would keep pulling the plans back into line with the regional statements above them. The reform has to bind the whole stack, which it can: regional policy statements must give effect to national policy statements too. The replacement legislation now before Parliament makes the point sharper still: it would collapse district and regional plans into one combined plan per region — seventeen nationwide, in place of more than a hundred; two Bills before Parliament (the Planning Bill and the Natural Environment Bill, introduced December 2025), not yet law — so the stack becomes a single document per region sitting under national direction. That is one door for the discipline to guard rather than two. It is also one document for the old hierarchies to be copied into, drafted by the same planning community that wrote them — which is why the plan-content rules of the new system, not just its transitional instruments, need to carry the prohibition.

The regional layer also explains a piece of market behaviour the maps alone would miss. Sophisticated developers do not treat a hostile district plan as a constraint; they treat it as the document to be amended. Anyone may request a private plan change; the request is judged against the regional policy statement and national direction, not against the zoning it seeks to replace; and the outcome can be appealed to the Environment Court on the merits. That is how Drury acquired a 35-hectare metropolitan centre out of Auckland's Future Urban land over the council's own opposition — the first retail land sale went to Foodstuffs, for a New World — how The Base rose at Te Rapa, and how Homebase's owner litigated its way out of half the restrictions Christchurch first imposed. Since the Government's 2025 "plan stop" froze most council plan changes until the end of 2027, the private plan change has been, in much of the country, the only rezoning channel there is.

The hierarchy, in short, is negotiable for those with the money and patience to join it, and binding on everyone else. A challenger grocer needing twenty dispersed sites cannot run twenty private plan changes; a mall owner with one large site can run one. The maps above are the rules as most participants meet them. Insiders meet a different system, one floor up — slower, dearer, and far more forgiving.

And the channel is not priced only in money and patience. It is priced in know-how: which councils entertain requests, what the regional statement rewards, which experts panels find credible, when to settle and when to appeal. That knowledge lives in a small community of planning counsel, retail economists and development managers — the local fixers. An international entrant can import grocery expertise; what it cannot import is navigation of nine councils' folkways and two layers of planning documents, each with its own case law. Any serious entrant would need to hire that navigation locally before buying a single site. The de facto system is learnable only by participating in it, and the participants, so far, are the incumbents and the landowners who deal with them. That is one more fixed cost of entry the insiders have already sunk — and one more reason the visible rulebook mapped above understates the true asymmetry.

What would actually work

The Government built an express lane in 2025. The atlas above shows why the destinations are still missing.

The Commission said it in 2022: free up sites. Eric Crampton's version of the argument was operational — an entrant cannot assemble a network of sites when each one needs its own multi-year consent fight, and the Overseas Investment Act screens the very land assembly the Government says it wants. He proposed that a new entrant's full slate of stores be handled in one simultaneous process — rezoning, consenting, and overseas-investment approval together, decided in months.

Benno Blaschke turned that into full drafting instructions in May 2025: a Competitive Streamlined Planning Process for any genuine new entrant proposing ten or more stores, bundled multi-site rezoning plus consents, an independent economist on every panel, and statutory language directing panels not to decline or condition applications for effects on the "viability, vitality, amenity, hierarchy or function of any existing centre". Marko Garlick's version is a National Policy Statement on Supermarket Development: change the plans themselves, permanently and for everyone, not just the first entrant through the door.

The Crampton–Blaschke blueprint had two further limbs worth recording, because neither made it into the 2025 package. First, full mixed-use rights on approved sites, with height limits set aside: the supermarket as a podium, commercial or residential towers above, so the store buys its land jointly with the housing the same site can carry. Second, Kāinga Ora's landholdings as dealable currency — the Crown owns a great deal of well-located urban land, and could strike bargains in which an entering supermarket and a private developer build out a site together, with Kāinga Ora taking leases on a share of the apartments as its price. Both limbs answer the same economic problem. Downtown zoning is the friendliest on these maps, and downtown sites can generally already carry towers above a supermarket podium — that is not the constraint. The constraint is that downtown floorspace is the dearest in the country and parking dearer still, and the format that survives those economics is the small metro store, not the full-line grocer. Full-line supermarkets belong where land is cheaper and households actually live — the suburban rings, which is exactly where the zoning turns hostile. A rulebook whose most permissive zones are the places the full-line format least wants to be is a quieter way of not permitting it.

The Government moved in August 2025. The Fast-track Approvals Act now covers competition-enhancing grocery developments, backed by a Government Policy Statement on Grocery Competition; a single national building consent authority (Christchurch City Council, as it happens) handles supermarket building consents; the overseas-investment pathway has a grocery directive. Consents in under a year, on the advertised timeline.

One layer the package does not reach: alcohol. A full-service supermarket needs an off-licence as surely as it needs a building consent, and both Crampton and Blaschke flagged the case for bundling alcohol licensing into the same one-stop process — Blaschke's note names it as adjacent red tape without drafting the mechanism. Licensing runs on its own clock, before its own committees, with its own objectors. Leaving it outside the bundle leaves a full-service entrant one contested hearing away from opening a supermarket that cannot sell wine.

It is worth being precise about what the express lane does, because the mechanics decide how much it is worth. At the moment of decision, the fast track beats the plan: the panel gives the greatest weight to the Act's development-facilitating purpose (Schedule 5, clause 17), the non-complying gateway test is switched off, inconsistency with a district plan cannot on its own justify declining an application (section 85(4)), and a panel may even consent an activity the plan prohibits (section 42(5)(a)). Every centres-hierarchy policy mapped above is, for that one decision, demoted to a consideration.

Then the moment passes. The Act's exhaustive list of obtainable approvals (section 42(4)) contains no plan change: nothing a panel grants is ever embedded in a district plan. The consent is handed back to the council and treated "as if it were granted under the Resource Management Act" (Schedule 5, clause 31). Every later adjustment — expanding the store, reconfiguring the site, changing a condition — runs through ordinary RMA processes, decided by the council, against the same unchanged zoning, with none of the fast track's weighting. A condition change goes through RMA section 127; an expansion beyond the consented envelope is a fresh application in which non-complying status revives in full; and the fast-track lapse floor — and, since the 2025 Amendment Act tidied the drafting, its default — is two years, against the RMA's five. The plan learns nothing from the consent — no rule changes for the next store, or the next entrant. The Government adopted the consenting half of the blueprint and left the rezoning half on the table. The express lane crosses over the plans without changing them, and every store it delivers lands on the far side, governed thereafter by the rules the lane was built to bypass.

As of mid-2026, no third chain has committed. Aldi, Lidl and Tesco reportedly declined even to answer the Government's request for information. The old economists' joke says there are no $20 notes on the sidewalk — if there were, someone would have picked one up. One reading of the silence is that grocery's excess profits were never as large as the return-on-capital arithmetic suggested: the notes are counterfeit. The other reading is the one this atlas supports: for twenty years, checking whether the notes were real has itself been against the rules. An express lane is a process, and a process does not answer an entrant's first question, which is where.

The mechanics above cut both ways here. At the moment of grant, the lane genuinely overrides the maps: a panel can consent a store the plan makes non-complying, and the quotas and centre-vitality tests cannot on their own sink the application. But the override is case-by-case and is spent at the moment of decision — the consent it produces endures, as consents do; what does not endure is the favourable law under which it was won. The nine rulebooks mapped above remain the law of the land — governing every site an entrant does not fast-track, every expansion and condition change after a grant, and every assumption a network plan must be built on. A supermarket network is a multi-decade asset; an entrant weighing New Zealand is not pricing one consent but a portfolio of stores that must live, grow, and be refitted under rules the express lane leaves untouched.

The charitable reading of the consent-only design is expedience: a fast-track amendment was quick to draft, and the Government's stated expectation is that the full resource-management reform will embed competitive urban land markets properly, making bespoke supermarket machinery unnecessary. Fair enough — and it supplies a clean test of the reform itself. If the RMA's replacement genuinely delivers contestable urban land markets, the supermarket fast-track becomes irrelevant: nobody needs an express lane where the ordinary roads work. If entrants still need the fast-track once the reform is in force, the reform will not have gone far enough. The lane's traffic is the measure of the road network.

The remaining work is not mysterious, and it comes in a first-best and a second-best. First-best is to fix it at the source: write the resource-management legislation that replaces the RMA so that councils cannot run centres-hierarchy planning or retail-viability tests at all — in the single combined plan each region will prepare under the new framework, which inherits the work of both today's district plans and the regional policy statements that mandate hierarchies in four of the five regions mapped here. The new system simply should not recognise a competitor's lost customers, however described, as an effect to be managed. And because plans will keep finding new ways to embed the old outcomes, pair the prohibition with ongoing review: a standing check that plan provisions are not quietly rebuilding anticompetitive protections under new names, caught by an auditor rather than discovered by the next entrant's lawyers.

Second-best, while the reform is written: strip the centre-viability tests out of retail consenting by national direction, as the Commission's first recommendation implied and as Blaschke's draft clauses and Garlick's NPS-SD would each do. Delete the quotas — no plan should count supermarkets. Publish the infrastructure constraint maps everywhere, so an entrant can screen sites from a desk instead of discovering the full pipe after the land deal. And add the limb the 2025 package left out: a mechanism that changes the plans themselves — bundled multi-site rezoning for an entrant's whole network, or national direction making supermarkets permitted in every centre and mixed-use zone — so that permission outlives the consent that granted it. Entry is the only test of whether the grocery profits are real. Make the land legal, and run the test.

Sources

Authorship

This site was researched and written by Claude Fable 5 (Anthropic). The analytical framing draws on the published corpus of Eric Crampton and the New Zealand Initiative, Benno Blaschke's fast-track research note, and Marko Garlick's Yes In Our Backyards. Eric Crampton reviewed the text; for attribution purposes he is reviewer, not author. Errors are likelier to be the machine's than the reviewer's — and some surely remain: nine district plans, ~250 zone classifications, and a moving legislative target leave room for mistakes. Corrections and comments are invited via this form; submissions are reviewed periodically and material corrections are noted on the page.

Method

Zone classifications were compiled from the operative text of each district plan as at 3–6 July 2026: the Auckland Unitary Plan (chapter PDFs, including a full-text sweep of all 232 published Chapter I precinct documents), the Christchurch District Plan (post-PC14), the Wellington City 2024 District Plan (IHP decision documents; the live ePlan sits behind a terms click-wall and a few rule numbers are reconstructed — flagged in the map popups), Hamilton's operative plan (PC12 clean-version chapters), Dunedin's 2GP (live ePlan), the operative Lower Hutt plan, the Porirua District Plan 2025, Upper Hutt's operative plan (archived consolidation), and the Kāpiti Operative District Plan 2021. Zone polygons are the councils' own published GIS layers. Status-changing exceptions are drawn on the maps wherever a public boundary exists (Auckland's precincts, St Patrick's Estate, Homebase, Petone's two areas, Kāpiti's Meadows Precinct and per-centre town-centre thresholds); exceptions without a published boundary remain popup-disclosed, and only the mapped ones are netted from the comparison table. Every zone popup carries its controlling rule reference so the classification can be checked. The underlying datasets — classified zone polygons, the graded store inventory and its inclusion rule, and the per-capita calculations — are published in the data folder, alongside a companion note explaining, for human and machine readers alike, how the work was done and why the judgment calls went the way they did. This is a screening tool, not legal advice.

Reading

• Commerce Commission, Market Study into the Retail Grocery Sector — Final Report (8 March 2022).
• Eric Crampton, "Legalising groceries", Newsroom (8 March 2022); "Real supermarket competition will require planning reform", NZ Herald (3 April 2025); "The only way to find out if more supermarket competition is real", The Post (2 June 2025).
• Benno Blaschke, Fast-Track Supermarket Entry and Expansion Omnibus Bill, NZ Initiative research note (29 May 2025); "Breaking up supermarkets won't solve competition issues", NZ Herald (22 May 2025).
• Marko Garlick, "Supermarket YIMBYism" and "More stories of how councils accidentally on purpose banned competition", Yes In Our Backyards (February 2025).
• Hon Nicola Willis, "Express lane for new supermarkets" (27 August 2025); MBIE, "Supermarket fast-track pathway opening".
• Joel MacManus, "Why New Zealand can't get a third supermarket chain", The Spinoff (28 April 2026).
• District plans: Auckland Unitary Plan · Hamilton · Wellington City · Hutt City · Porirua · Upper Hutt · Kāpiti Coast · Christchurch · Dunedin 2GP.
• Case law and decisions: Landco Mt Wellington Ltd v Auckland City Council EnvC A035/2007 (Stonefields); Grafton Downs Ltd v Auckland Council [2024] NZEnvC 122 (Paerata Rise); Foodstuffs (South Island) Ltd v Queenstown Lakes District Council [2012] NZEnvC 135; the IKEA/Sylvia Park traffic decision, [2008] NZEnvC 8; the Homebase plan change decision and appeal consent order (CCC, 2022–23).
• Costco case study: LINZ overseas investment decision 201900422 (5 February 2020); Stuff on the non-notified consents (28 April 2020); Newsroom on the landowner-initiated Rolleston consent (7 December 2021); RNZ on the Drury second store (12 November 2025); AUP I615 Westgate Precinct.
• Productivity Commission, Better Urban Planning — final report (2017): centres-hierarchy prevalence and consenting case studies.
• Chris Parker (Auckland Council chief economist), Housing Supply, Choice and Affordability (October 2015) — tool 15, reporting Eilya Torshizian's Chief Economist Unit research on Restricted Discretionary vs Discretionary consenting outcomes.
• The regional layer: AUP Chapter B2 (RPS) · Waikato RPS Change 1 · Greater Wellington RPS (Change 1 appeals version) · Canterbury RPS · Otago RPS 2021 · Quality Planning on private plan changes · DLA Piper on the 2025–27 plan stop.
• Fast-track mechanics: Fast-track Approvals Act 2024 (esp. ss 42, 81, 85; Schedule 5 cll 17, 26, 31); Fast-track Approvals Amendment Act 2025; Government Policy Statement on Grocery Competition (Gazette, 17 December 2025); commentary from Bell Gully and Russell McVeagh.
• Wellington's previous plan: the 2000 District Plan's Centres objectives, Centres rules, and Business Area rules (Plan Change 73, operative 2014); Joel MacManus, "Who killed the Johnsonville mall?", The Spinoff.
• Infrastructure overlays: Watercare network capacity (GIS layer, 6 categories); CCC wastewater capacity open-data layers; HCC wastewater constraints layer. Existing supermarket locations © OpenStreetMap contributors (ODbL).

Disclosure: The New Zealand Initiative is funded by member organisations, which include both major grocery chains. The argument here — that entry should be legal — is the one position in grocery policy that incumbents have the least reason to fund. Design after Alex Tabarrok's The Ceiling Trap. Basemaps © OpenStreetMap contributors, © CARTO. Zone data © the respective councils (CC BY 4.0 where stated).