House Purchase Affordability: Where I Stand

Earlier this week, the Productivity Commission released its interim report on housing supply regulation. There is much to like here, and I find myself in broad agreement with much that is written.

For some time I have been thinking about how I view the house purchase market in Australia and how my views align with or disagree with some of the louder voices I encounter on twitter.

This note sets out how I think the market works, what would actually improve affordability, and where my view aligns with and departs from the main twitter commentators. Throughout, this is about the owner-occupier purchase price. While the rental/investor market is related, it is out of scope for this piece.

1. How the market works

(a) Housing is two goods

A home is two goods sold as one. Shelter is satiable, because one roof is enough. Land is scarce and positional, with no satiation in the places people want to live. The building depreciates; the land appreciates; the gain flows to the land.

Why it stays expensive: what is scarce is usually not houses but good locations, the places close to jobs, transport and services. You can always put up another building, but you cannot make more land near the centre. So when buyers get more to spend, from lower rates or rising incomes, it does not make homes cheaper; it gets bid into the price of the well-located land. Builders cannot compete that away by building more, because the planning rules cap how many homes each site can hold. Lifting that cap is what upzoning does: it lets a good location hold more homes by building up instead of out, the only way to make more of something you cannot spread sideways.

(b) The price sits in a band

The market clearing price for a house and land package sits inside a band. The floor is what it costs to build, construction cost plus the normal return a developer needs. That floor is a threshold for new supply, not a hard floor under existing homes: below it, new building simply stops, and existing homes can and do trade lower, as they do in declining towns. The ceiling is what buyers can pay.

The gap between floor and ceiling is mostly the rent on scarce, desirable land, and it opens because supply is slow, not because anyone is powerful. Even with a thousand fiercely competitive builders and no collusion, a fixed stock of good locations meeting rising demand clears at a price near what buyers can pay, and everything above build cost is land rent.

(c) What moves the demand side

The ceiling itself has been lifted over a generation by falling interest rates, wages growing faster than inflation, the growth of dual income households, long and heavily leveraged mortgage credit, and the tax privileges of the home (the capital-gains-free main residence, untaxed imputed rent, the pension assets-test exemption).

Demand for shelter is inelastic, because almost no-one chooses to exit housing altogether. For most people, the choice is renting or buying, not homelessness. For purchases, demand adjusts through hidden margins, households that never form, sharing, longer commutes, smaller dwellings, deferred purchase. So the market clears, but the unmet demand is invisible rather than absent, which is why “we build enough” ratios prove nothing.

Expectations work on both sides. When prices are expected to keep rising, buyers stretch further and bring purchases forward, which lifts the bid, and holders of developable land wait for more, which slows release. Either way the expectation of gain feeds the price.

(d) Why supply answers slowly

Supply is slow for structural reasons: much building is simply not permitted (zoning), and what is permitted is slow and discretionary (approvals and objection rights). Those are the locks that hold the short-run supply curve near vertical.

New building also rides the rate cycle as well as housing policy. When the RBA eases, approvals pick up with a lag; when it tightens, they stall. So the flow of new supply is partly hostage to macro policy and cannot be dialled up to meet a housing target on command.

(e) How demand often sets the price

Demand sets the bid and supply decides how far it runs. In desirable areas supply cannot answer quickly, so the bid is banked into price rather than competed away.

The mechanism is two clocks, not two players. The RBA can move in a day, and borrowing capacity reprices over the following weeks. Physical supply answers over years, sometimes decades. In the gap between those two clocks the price runs up toward capacity to pay, and it stays there for as long as supply cannot catch up, which in the desirable places is most of the time.

Interest rates pull in three places at once: the demand ceiling, the cost floor, and expectations. That triple role is why monetary policy moves building so violently, and why the cross-cutting drivers matter most. Historical charts of approvals, commencements and completions show the stop-start nature of the building cycle over time.

The surge is captured by whoever already owns. When rates fall the whole existing stock reprices at once, so the gain lands first with current owners and the land beneath them. Developers take a slice on the new stock they happen to be holding, but they are one seller among many, not the cause.

The drawdowns confirm the timing story rather than deny it. The major Australian falls have generally come through the demand side, tightening credit, recession or higher rates, rather than through a surge of completed housing, which is exactly what the timing asymmetry predicts. The falls are usually single-digit and clawed back within a couple of years. So under today’s slow supply the rent sits on the table almost always, and a visible fall needs a demand retreat. Faster supply would not stop the shocks; it would compete each demand-driven spike back toward cost sooner, making the rent transient rather than permanent, though never instantly.

(f) Who keeps the gain

A separate and smaller story sits on top: bottleneck power. Wherever a slow pipeline has a chokehold that cannot be bypassed quickly, whoever holds it can take a slice of the rent. Australian construction unions are the clearest live case. They own no land, but by controlling a scarce trade that cannot easily be bypassed they capture part of the surplus (Queensland Productivity Commission, 2025-26). Some developers with large landholdings stage release the same way. This sits as an amplifier on top of the timing asymmetry rather than being the cause of high prices, and it is a claim about specific bottlenecks, not about how many builders there are.

The core question is incidence: when something moves, who keeps the gain. A demand subsidy raises bids, and the tighter supply is, the more of it capitalises into price and lands with existing owners. A supply reform can be captured too, by whoever controls the scarce location or the timing of its release. A reform improves the buyer’s position only to the extent that it does one of three things: creates real substitutes for the scarce good, weakens the scarcity rent itself, or shifts the rent’s incidence away from the buyer. Competition, more supply, a land tax and a credit limit each work through one of those channels, and competition is one of them, not the master key. These are not the same as a lower sticker price: a land tax can change who keeps the rent without cutting the gross price, and a credit limit can lower prices while leaving a constrained buyer no better able to buy.

In one line: demand creates the bid, scarcity sets the rent available to capture, supply speed decides how much of the bid becomes homes rather than price, planning sets much of that speed, bottlenecks decide how much of the remaining rent is skimmed along the way, and tax decides who ultimately keeps the land rent. Each driver has its own job.

2. What we can do to improve affordability

(a) Make supply elastic

Supply works best when three things come together. I put these as reinforcing conditions rather than strict necessities, because broad upzoning can itself create much of what the third one asks for.

• Upzoning: permission to build more, and especially to build density, which reproduces a scarce location vertically.
• Development as of right: permission you can actually use on a timetable, rather than negotiate site by site while objectors run down the clock.
• Enough contestability that no developer can profitably restrict the response. Broad, as-of-right upzoning does part of this on its own, because once thousands of sites become developable, withholding any one of them stops paying. Where good locations stay scarce and lumpy, contestability has to be pursued directly.
• The strongest regime combines broad permission, predictable as-of-right approval, and enough contestability that no one can meter the market’s answer. My real disagreement with Tulip is not that his lever fails without mine. It is empirical: how often does broad upzoning deliver that contestability by itself, and how often must land-development contestability be treated as its own problem.

(b) Discipline the land tier

Discipline the land tier as well as the delivery tier, because competition cannot make more ground where land is fixed.

• A broad land value tax, ideally replacing stamp duty, falls on the unimproved value of the ground and continues to apply to the site whether it is developed or withheld.
• More experimental than the tax, and to be used with evidence and care: use-it-or-lose-it and time-limited consents, and vacancy and idle-land levies, which make metering costly rather than profitable. On concentration, target acquisitions that materially lock up developable land in already tight submarkets, aimed at anticompetitive control rather than at the site assembly that density often needs.
• Competition-policy scrutiny of land markets, and a public developer acting as a market-making entrant that disciplines incumbents and reveals costs, not as a standing subsidised provider.

(c) Lower cost and friction

Support the cost and friction levers, but know their limit: a cost saving does not guarantee an equal saving for buyers. Where development rights stay scarce it can be competed into land value or held at a bottleneck rather than passed through, even with competitive builders.

• Approval reform: deemed-to-comply standards, an end to the council veto, scrapping parking minimums, time-limited approvals.
• Construction productivity: prefabrication and modular methods, national licensing, a leaner building code. Dwellings built per hour worked have gone backwards for decades while the wider economy’s productivity rose.
• Trades contestability: broaden entry into the licensed trades, mutual recognition of occupational licences across states, and workforce mobility, so no single labour chokehold can hold up the pipeline and skim the rent. This is the remedy for the union bottleneck named in section one. Lead with the licensing and fragmentation the Productivity Commission points to rather than an industrial-relations fight, and keep the limit in view: clearing a labour chokehold lowers cost, but unless land contestability holds too, the freed margin is captured further down rather than passed to the buyer, and even full de-unionisation would not touch land-market power.

(d) What not to do

What not to do, because these capitalise into price or merely move the rent between captors:

• Demand-side handouts (grants, deposit schemes) substantially capitalise into price where supply is tight, helping the seller more than the buyer.
• Cutting migration to hit a housing target. It would lower prices against the counterfactual, but it spends a whole-economy lever on a housing job and sacrifices more than it buys.

(e) A test and a caveat

Test every lever with one question: where does the gain go? It helps the buyer only if the gain reaches them, not if it is captured on the way by land, a bottleneck or an incumbent.

The honest caveat is that this is a portfolio held together for decades, and the politics rarely sustains it. The supply-only frame is also the reform agenda most congenial to incumbents, since it disciplines planners rather than landholders, and an industry rarely campaigns for competition against itself, which is why the contestability and land levers, the parts that touch landholders, have no natural champion.

3. Where my view aligns and disagrees with others

(a) Murray, and the public-provision tradition

Agree with much of his market analysis. Cameron Murray is a Georgist, and his account of land rent, the slow supply response, staged release and economic rent extraction to the capacity to pay is largely right and close to my own (as set out above).

The sharpest version of this camp, put by Emily Sims on Murray’s Fresh Economic Thinking, argues that location is the original monopoly, so upzoning only reallocates development rights and the market, not zoned capacity, sets the pace of building. I share both premises: location is a monopoly, and permission is not delivery. But the conclusion does not follow. That developers can meter under narrow upzoning is an argument for contestability and land levers, not for abandoning supply reform and going around the market. Attack the monopoly directly, through broad as-of-right permission, competition and a land tax, rather than leaving it intact and housing people outside it.

On land-banking, the evidence is contested, not settled. Tulip and the Commission read it as weak; Sims points to greenfield estates drip-fed over decades, tens of thousands of vacant dwellings, and towers left largely unreleased after completion. I treat withholding as a real but secondary amplifier, live enough to watch and to design against, not the engine of high prices.

Disagree that the current market failure necessarily requires a public solution. Diagnosing a failure of the private market does not by itself say the state should build the housing for purchase (for example, along the lines of the Singapore Housing Development Board).

• Concede first that a public developer can change some constraints a private one cannot. It can acquire land, including by compulsion, coordinate infrastructure, assemble sites, finance on the government’s balance sheet, accept a lower return and retain the land value rather than pay it away. Murray’s HouseMate leans on exactly these, so the objection is not that public building has no advantage.
• The real question is narrower: what constraint does public ownership solve that competition and planning reform cannot solve more cheaply, and does that justify moving development and credit risk onto the state balance sheet at scale? I have not seen that case made.
• The stopping-rule problem is practical, not theoretical. Parliament could in principle set a target, eligibility and a budget, but from the inside I could never get a straight answer to what the market failure is, how large it is, and how we would know when it was fixed. A programme without that answer is an open-ended appropriation.
• It need not discipline private prices even at very large scale. If the objective is subsidised shelter, the very rules that protect that subsidy, one flat per household, owner-occupation, a minimum occupation period, income limits, are what stop the public tier from cross-competing with the private market. Singapore shows how far that can go: around eighty per cent in public housing, yet condominium prices among the world’s highest, fenced into a separate market by design. Public provision can house people very successfully without eliminating the land rent in the private tier.

The boundary with Murray, put concretely: I am not opposed to a government developer that behaves like any other builder, competing for sites and selling homes on the open market, so its presence keeps private developers honest on price, shows what a home really costs to build, and stops them holding supply back. However, good market reform should make such a public builder unnecessary. I am opposed to a government that becomes the main supplier of housing, or runs a large subsidised home-ownership scheme that houses most people outside the market, as Singapore does. The first uses the state to make the market work; the second replaces it. So my quarrel with Murray is about disciplining the market, not bypassing it.

(b) Tulip, and the supply-only school

The real divide is mechanism. Peter Tulip’s focus is shortage: the premium over build cost is a deficit of permitted homes, and building them competes it away. Mine is rent: the premium is economic rent on scarce location that fast demand and slow supply hand to land, which building can discipline substantially, competing away the scarcity that planning creates, though never the inherent premium on genuinely superior locations.

Agree on the big point: the supply side is where constructive intervention belongs. Supply is constrained, planning has steepened the curve, and freeing it helps. The New Zealand and Auckland evidence is real, and removing time and rule frictions does raise supply and soften price.

Agree in particular on upzoning and making supply elastic. That is his core, and it is the first leg of my package.

Note that Tulip would probably not accept my full package as necessary. The gap is precisely the contestability leg and the land-market levers, which he treats as unneeded rather than merely secondary.

• He rejects land-banking. He argues that postponing sales is not foregoing them, because otherwise inventories would grow without limit, which we do not observe, and he cites the Productivity Commission finding no evidence that land-banking is a material limitation on land supply. However, the Commission is a little contradictory here. The same report that finds weak evidence for outright land-banking also concedes the hold-back risk is “likely greatest where upzoning is narrowly targeted”, which describes Australia, and its own incidence discussion presents evidence of exactly this mechanism, land values absorbing the windfall, which is the substance of the concern under another name.
• He treats the industry as already competitive, pointing to the large number of residential building firms (ABS counts in the tens of thousands), which he takes to rule out supply manipulation.
• He discounts the tax levers I would keep. He puts the price effect of negative gearing and the capital gains discount at only about one to four per cent, and concludes it can practically be ignored, and he does not propose a land value tax as a remedy.
• So on his account, once planning is fixed the competition and land-banking problems largely dissolve. That is exactly why he would regard my contestability leg and my land levers as unnecessary rather than embrace them.

My bigger problem is with the model, and the headline it produces. Mechanically combining Tulip’s own inputs, a 42% Sydney zoning wedge and a 2.5% price fall for each 1% of extra supply, the figure summed across the capitals is around a million homes. Even this is soft: it is a linear extrapolation across variables whose real relationship is almost certainly non-linear, so at that scale the functional form does much of the work, and it treats the entire unexplained gap as zoning when other factors almost certainly sit inside it. I am with him on the direction; but the implied number I do not believe.

(c) Grattan, and the portfolio reformers

Agree: this is the closest solution space to mine. A portfolio rather than a single lever, and rules over discretion (deemed-to-comply, ending the council veto, removing parking minimums). Coates is also good on land economics, and Grattan does reach the land tier, backing windfall gains taxes on rezoning and winding back the capital gains discount and negative gearing. Good work, and I support most of it.

Where we differ is narrower than it first looks, and it comes down to two things.

• Instrument. Grattan’s land lever is betterment, taxing the uplift at the moment of rezoning, and it rates windfall gains taxes among the most efficient we have. Mine is a broad land value tax on the standing stock, which falls on all scarce land all the time and is harder to dodge. I would keep both, but the broad tax is the structural one; betterment only catches the rezoning event.
• Competition. Grattan is more relaxed than I am about developer market power and land-banking, treating them as not a binding constraint. Its builder-concentration data captures one factor that genuinely matters, but only one, and it does not reach control of developable land or the timing of release, which is the additional concentration margin I care about.

(d) The Productivity Commission

Agree with much of it, as I said at the outset. The PC lands remarkably close to the Grattan reform agenda, and Danielle Wood chaired Grattan before taking the Commission, so it reads as the official supply-side statement beside Grattan’s. It puts land-use reform and infrastructure coordination at the top, calls broad upzoning the single highest-impact lever, and says plainly that approvals reform alone “will not be sufficient”. I agree with all of that.

And it hands me evidence for my own argument, in places better than I had it.

• Permission is not delivery, in their numbers. The PC separates zoned capacity from commercially feasible capacity and finds the gap is large: New South Wales’s Low and Mid-Rise policy zones about 650,000 dwellings but only around 240,000 are feasible, and Victoria’s townhouse code zones 980,000 against 420,000 feasible. That is my “upzoning grants permission, not homes”, in the Commission’s own data.
• Land rent is the residual, in their incidence work. The PC finds a developer charge can fall on the landowner rather than the buyer: on the IPART result it cites, waiving water and wastewater contributions lifted land values but left house prices unchanged. The Commission notes the evidence is not unanimous, but this is the land-incidence mechanism my argument runs on, from the Productivity Commission itself.
• Supply shows up as slower growth, not a crash. The PC concludes that higher supply lowers city-wide prices against the counterfactual, and that this “appears as lower price growth, rather than an absolute decline”. That is my drawdown point in their words.

Where I part from the PC is less about what it says than about what it was told to leave out. Its terms of reference bracket out tax, so there is no land value tax, and its remit does not include a systematic examination of developer or land-market structure, or the demand side. So the Commission demonstrates the land-rent incidence that would justify a land tax and is then barred from recommending one. Its silence on my land and competition levers is a scoping line, not a verdict.

(e) Phillips, and the no-shortage school

In my reading, Ben Phillips focuses more on market analysis than on a solution for purchase affordability. His analysis is not far from Murray’s, or from my own, and I share his caution about treating a housing shortage as the source of high prices.

Agree more with Ben Phillips than with the supply-only camp on the headline question: the large shortage numbers do not look good against the data. His point that dwellings per adult is a better gauge than per capita is right, because it is adults who form households, and on that measure the stock is not far out of whack. On that aggregate metric, the stock does not show an obvious million-home deficit.

Agree, too, that much of the price rise is capacity to pay: lower rates, dual incomes and easier credit let buyers bid more. That is Phillips’s engine, and it is also mine.

Disagree that this settles it, on two counts.

• Aggregate adequacy is not a working market. Supply does not have to be in deficit to hand capacity to pay to landowners; it only has to answer slower than demand reprices. Capacity can jump in a day, from a rate cut or a second income, while building answers over years, and in that gap each surge capitalises into well-located land before construction can compete it away, however reasonable dwellings per adult looks over the long run. The problem is not a missing million homes; it is supply that arrives years late where it is needed. Balance in the stock and dysfunction in the flow can coexist.
• Phillips stops short of the Georgist step, even though his account leans on it. Rising ability to pay becomes higher prices rather than more homes only because good locations are scarce and slow to reproduce, so the extra capacity capitalises into land rent. That step is what squares an adequate stock with soaring prices, so it is already implicit in his own case. He simply does not name it, and so does not locate the cost where it actually sits, in the land rent.

And a caution about the metrics. The naive comparisons he is right to distrust, the deposit-to-income ratios and the like, still register real pain, even where part of what they capture is increased ability to pay. But the increased ability to pay is itself the pain, not a rebuttal of it, because that extra capacity is bought. It is bought by delaying the purchase until people have paired up, because a single income no longer reaches. It is bought by both partners working, and paying for childcare in order to. And it is bought by handing a larger share of disposable income to the mortgage than the previous generation did. A balanced ratio and a rising capacity to pay do not mean people hurt less; they are the record of how much has been given up to keep meeting the price.

(f) Cutting migration to fix housing

The serious case for cutting migration to fix housing is worth answering on its own terms. Tarric Brooker puts it plainly: the current intake “adds more to housing demand than it does the ability to build new homes”, so a lower intake would put us on a path to more affordable prices. I take that at face value, as a good-faith reading of the data.

Agree that migration adds real pressure. Since 2000 Australia has had one of the five fastest-growing populations in the developed world, so of course it pushes on the bid. But pressure on the bid is not the root cause. It becomes price rather than homes only because supply is slow, which is the whole argument of this note. With elastic supply the same inflow would produce building, not price.

The pressure is not uniform. In passing, permanent worker migrants add to both the demand and the supply side of the economy, while students and temporary flows are more demand-heavy and fall mostly on rents. How much each contributes is part of a large migration debate this note does not try to settle.

Disagree that cutting migration is good housing policy. Migration is a whole-economy lever, bearing on the labour market, skills, the budget, ageing and education exports. Setting it by the housing breakeven line mistakes one consequence of migration for the objective function of migration policy, aiming a whole-economy lever at a housing job. A lower intake would ease demand against the counterfactual, but that makes it the wrong tool, not a good one.

Put simply: migration policy should be evaluated on its own terms, including its impact on housing. But housing pressure alone should not drive the migration decision.

Where this leaves me

Set the camps side by side on two questions: who reads the problem right, and who reads the solution right. Almost no one scores on both.

On the problem, I am largely with Phillips and Murray. Neither treats it as a raw shortage of dwellings, and Murray names the land rent that Phillips leans on without labelling. Tulip I part from here. His focus is shortage, and he does not reckon with the economic rent that timing hands to land when demand reprices fast and supply answers slow. Grattan and the Commission I credit less, not because they are wrong, but because they never cleanly articulate the problem. They set out the reforms without naming the mechanism underneath them.

On the solution, the order changes. Phillips does not really fight on this ground; his work is largely analytic, it does not offer a program. Murray reaches for market failure, and then for public provision, with too much alacrity, when the failure he diagnoses argues for disciplining the market rather than replacing it. Tulip does not go far enough. Upzoning is the first move and I take it from him, but he stops there and will not add the contestability and land levers. Grattan and the Commission carry the best package on offer, and I broadly like it, though I would add the tax disincentives they leave out, a broad land tax above all.

Migration sits to one side of this. It needs examining, and its housing effect belongs in that. But I am not convinced it answers housing affordability on its own, not without weighing what immigration does across the whole economy. Housing pressure alone should not drive the migration decision.

Underneath all of it is one mechanism the camps keep missing. Demand moves in a day, supply answers over years, and in the gap capacity to pay becomes land rent long before it becomes homes. So the fix is not a number of homes. It is a working market: make well-located supply elastic, keep it contestable so the gain is not metered away before it reaches the buyer, and tax the rent that remains. Judge every lever by where the gain goes, not by whether the headline price ticks down.

The hard part is not knowing what would work. It is that the reforms that would reach the buyer attack the rents of the best-resourced interests in the country, and those interests have the most sway over the rules meant to discipline them. The fix is easy to name and hard to win.