Analysing the Economic Assumptions in the Productivity Commission's Housing Report
The Productivity Commission's (PC) Housing Supply Regulation Interim Report (July 2026) sets out a wide-ranging agenda for building more homes where people want to live. Its treatment of heritage controls, however, does not meet the standard the rest of the report sets for itself. Public submissions close on 30 September 2026 and the heritage analysis demands scrutiny.
Heritage Workshop support the Commission's objective. Enabling density in well-located, well-serviced places is sound policy, and the 'build mindset' the report adopts is the right starting point. Our concern is narrower. Reform Direction 2.5 would replace heritage conservation areas and heritage overlays with listings for specific buildings or small areas, each justified by cost-benefit analysis and exposed to delisting. That proposal rests on a misapplication of heritage economics and on data that is assumed rather than estimated.
The Misapplication of Heritage Economics
Section 1.1 of the interim report states that "a key role for regulation is to account for externalities" – costs or benefits that fall on people outside a transaction and are therefore not priced. It adds that "where a site has cultural or historical significance, regulation can help preserve it for the future". That is the correct economic frame. Demolition of a significant place destroys value held by neighbours, visitors, people who never visit but value its existence, and future residents. None of them is party to the sale. Heritage controls exist to correct this externality, and on the Commission's own Principle 2 they satisfy the market-failure test for regulation. The live question is calibration, not whether to regulate.
Reform Direction 2.5 answers the calibration question with the wrong instrument. It recommends that "targeted heritage listings should be justified by cost-benefit analysis and there should be more robust mechanisms to review and delist buildings". Throughout the report, the cost side of that ledger is the development benefit forgone on the listed site (Table 1.1; Box 2.2).
Forgone development value is a distributional effect, not a social welfare cost. The reasoning comes largely from Peter Tulip's paper for the Centre for Independent Studies, Reform of heritage legislation, which the Commission cites for the proposition that costs "vastly exceed" benefits. Tulip's example is a heritage-listed house in Gordon worth about $4 million that would sell for about $10 million as an apartment site; he counts the $6 million difference as the cost of listing. As Marcus Spiller and James Lesh set out in The economics of heritage laws and housing supply, a developer deterred by a listing does not leave the Sydney housing market. They buy the next-best unlisted site, and the premium that would have gone to the Gordon owner goes to another landowner instead. The welfare cost to the community is the difference in housing utility between the two sites – perhaps $1 million in the Gordon example – plus, possibly, some deferred infrastructure outlay at the urban fringe that is counted in five figures per dwelling and may never be incurred if infill capacity is ample. A cost-benefit test keyed to site-level land value uplift therefore overstates costs by an order of magnitude. It also makes protection inversely proportional to land value, so the places in highest demand receive the least protection.
The test also ignores irreversibility. An over-applied heritage control can be corrected at the next scheme amendment; a demolished building cannot be recovered. Where a loss is irreversible and future preferences are uncertain, standard welfare economics attaches a quasi-option value to retention – the value of keeping the choice open until more is known. The report's own objective "gives weight to the interests of both current and future residents", and future residents are precisely who a per-site delisting mechanism disenfranchises.
The Commission implicitly accepts this logic elsewhere. It treats environmental and hazard overlays as standing protections against "current or future costs" (Principle 1), proposes no mechanism to review and delist them, and front-loads flood, bushfire and biodiversity studies into strategic planning under Reform Direction 3.1. Heritage is the only protection for which the report proposes dismantling area-wide controls and adding delisting mechanisms.
There is a further inconsistency. Principle 4 asks for decisions that are "timely, predictable and simple", with "clear decision pathways, limiting discretion where possible", and Table 1.4 names "multiple consultant reports with conflicting advice" as the opposite of good practice. Listing today is a criteria-based assessment against published thresholds. A contestable cost-benefit test for every individual site would require site-specific feasibility modelling, quantified non-market values and competing valuation evidence for every decision. That is a new regulatory thicket, built in the one part of the system that is currently robust. Planning systems already balance multiple objectives when introducing and managing heritage overlays.
In our view, heritage economics supports assessing heritage once, at precinct scale, as now, through the strategic planning the Commission already favours – and then codifying the result.
The Data Disconnect
Box 1 (note a) sets the Commission's evidentiary standard: it gives "more weight to studies that estimate the effect of new supply using causal methods and capture long-term city-wide effects, rather than short-term or local effects". None of the heritage evidence in Box 2.2 meets that standard.
The headline figure – that removing broad heritage controls could lift infill development within 10 km of Melbourne's CBD by around 17 per cent – rests on a non-causal, cross-sectional comparison in the Grattan Institute's More homes, better cities (Coates, Moloney & Bowes, 2025). Grattan found that heritage-covered sites were about half as likely to have hosted projects of 10 or more dwellings between 2017 and 2022. The analysis controls for location, lot size and zone, but it cannot control for the characteristics on which heritage areas are selected: intact, structurally sound and high-value existing dwellings on small and irregular lots.
On the Commission's own feasibility logic (Box 2.3), demolish-and-rebuild is marginal wherever existing improvements are valuable, whatever the heritage status. Grattan also notes that "data are not available for smaller projects" – the rear additions, dual occupancies and secondary dwellings through which heritage precincts actually add homes. The Commission's caveats – that the estimate "should be treated as indicative" and that "some differences in infill rates may reflect other site characteristics" – are accurate. They do not survive into the Box's headline, into Reform Direction 2.5, or into the Overview.
The second problem sits inside Grattan's capacity model. Appendix B records that the authors excluded heritage items from the Sydney analysis altogether and "reduced capacity on residential-zoned sites in heritage conservation areas … by 40 per cent", with about 20 per cent in mixed-use heritage areas – an assumption carried over from a 2016 consultancy report. The same appendix concedes that "the level of protection can vary within HOs and HCAs" and that spatial data to tell them apart are not widely available. The heritage effect in the model is an input, not an estimate. Any capacity uplift attributed to removing heritage controls in a model built this way is an echo of the assumed parameter. The Commission has said it "will further consider the potential impacts of heritage controls on feasible housing supply" for the final report. If that modelling inherits the parameter, it inherits the assumption.
The third problem is the conflation of coverage with constraint. Box 2.2 reports that heritage protections cover 21 per cent of residential land within 10 km of Sydney's CBD and 29 per cent in Melbourne. These figures measure the extent of a permit trigger, not its effect. A heritage overlay requires that works be assessed for their effect on significance; it does not prohibit them. In the Commission's own typology (Overview, p. 5; Section 1.1) that makes it a regulatory burden, not a regulatory ban – yet Chapter 5 groups heritage overlays with "restrictive zoning rules" as controls that hold land back, and Section 2.4 says they "prevent development".
In Sydney Heritage vs Housing: Beyond the 84,000 Headline we examined an Australian Public Policy Institute paper that converted conservation area coverage into a theoretical yield of 84,771 dwellings. The interim report repeats that paper's central error: it treats overlay coverage as capacity forgone without modelling the housing capacity heritage areas already deliver under current controls, or the capacity they could deliver under compatible upzoning. Until both are modelled, a coverage statistic says nothing about supply.
Zoning as the True Binding Constraint
Chapter 2 is explicit that "the most restrictive control often determines what can be built", and it names that control the binding constraint. Applied to heritage precincts, the Commission's own logic points away from the overlay.
Heritage overlays are layered on top of zones; they do not replace them. The interim report records that 75 per cent or more of residential land in Brisbane, Sydney, Perth and Adelaide is zoned for one or two storeys, and Grattan finds that about 80 per cent of residential land within 30 km of Sydney's centre and 87 per cent in Melbourne is restricted to three storeys or fewer. In a precinct zoned for 1–2 storeys, with front and side setbacks, site coverage limits and – in New South Wales – a maximum floor-space ratio, the zone and its built-form controls set the ceiling. The overlay operates beneath that ceiling. Removing the overlay while leaving the zone in place delivers little. Relaxing the zone may deliver most of the available yield with the overlay retained, provided the overlay regulates significance rather than density.
Grattan's own results point the same way. The councils where its modelled uplift in commercially feasible capacity is largest – Sutherland, Northern Beaches and The Hills in Sydney; Monash, Banyule and Glen Eira in Melbourne – are middle-ring, low-density-zoned areas, not the inner heritage precincts (Figure 5.3).
The Commission proposes to relax exactly these binding controls through Reform Directions 2.1–2.6: three storeys across residential land, mid-rise near transport, smaller lots, no maximum floor-space ratios, and modified setbacks and site coverage. What heritage controls would constrain after those reforms are applied is the only question that matters for Reform Direction 2.5. The interim report has not estimated it. The final report should model the marginal effect of heritage controls conditional on Reform Directions 2.1–2.6, publish the method behind the 17 per cent estimate, and test the counterfactual for sensitivity. Until it does, Reform Direction 2.5 rests on an unknown.
A Constructive 'Conserve and Build' Agenda
Heritage conservation and housing supply are not opposed. The inner-city terrace, semi-detached, walk-up and shop-top precincts that make up Australia's largest heritage conservation areas are working examples of the compact 'missing middle' forms the Commission wants more of. Older stock is also the naturally affordable stock in the Commission's own filtering logic (Box 1); wholesale replacement with new premium product removes lower-cost dwellings in the near term. Retention avoids embodied carbon and uses existing infrastructure, which the report's own evidence shows can cost two to four times more to provide in greenfield than infill locations (Chapter 4).
Rather than unravelling heritage precinct controls and inviting site-by-site contestation, we propose a 'conserve and build' agenda in place of the heritage limb of Reform Direction 2.5:
Codify (more) permit exemptions for low-impact works. Like-for-like repair, services, solar installations, internal works to non-significant interiors and rear additions and in-fill within clear envelopes may not require a heritage permit. Exemption schedules can be standardised within each state, made more flexible by local authorities, and published as machine-readable data. This is a genuine quick win under Information Request 5.1 and serves Principle 4 directly.
Decouple heritage overlays from low-density zoning. Heritage controls should regulate what carries significance; historic, aesthetic, social, scientific and spiritual values; fabric, form, streetscape and subdivision pattern. Heritage is not neighbourhood character and significance is evidence-based and replicable. Standards can then apply for rear and upper-level additions, roof form, street presentation, and 'bluefield' in-fill housing. The zone should govern density and yield. Reform Direction 2.1 already frames its exception around "heritage listed buildings" rather than areas; heritage precincts should be included in any three-storey standard on that basis, with significance protected through the overlay's objective standards.
Micro-zone strategically. Direct additional height and density to non-contributory pockets within and adjoining heritage precincts – car parks, service lanes, recent infill, rear yards, single-storey commercial and depot sites. This delivers yield in well-serviced locations without the speculative demolition pressure that blanket upzoning of intact precincts would invite. It is the constructive version of Grattan's suggestion that new protections be offset by commercially feasible upzoning elsewhere.
Recognise adaptive reuse as a named supply pathway. The report identifies greenfield, brownfield and infill as sources of supply. It does not name the conversion of vacant commercial buildings or under-used upper floors on heritage main streets – a distributed, low-infrastructure-cost resource that supports Reform Direction 2.4. The City of Adelaide's Adaptive Reuse City Housing Initiative (ARCHI) already provides fast-tracked assessment and incentives for such conversions, including of heritage buildings, and the Council has made a submission to this inquiry. Heritage typologies like warehouses and industrial sites are canvasses for housing density interventions. The final report should name adaptive reuse, quantify it and recommend that the Adelaide model be generalised.
Value Calibration and the Final Report
The final report must align its heritage analysis with sound heritage and welfare economics: transfers distinguished from welfare costs, irreversibility priced, and the marginal effect of controls estimated after the zoning reforms are applied. It must also calibrate the benefit side properly. Box 2.2 presently cites Fifer (2025) as support for the claim that costs vastly exceed benefits. That study, The Value of Historic Cultural Heritage, is a benefit study: a discrete choice experiment across more than 4,000 New South Wales households finding willingness to pay of roughly $9 to $198 per household per year for conserving portfolios of heritage items, including items hundreds of kilometres away. Together with the Victorian willingness-to-pay research in The Value of Heritage (2018), it shows non-use value that is real, measurable and does not decay to zero with distance. The final report should present these estimates on the benefit side and show the netting, or withdraw the inference.
Beyond willingness to pay, the environmental, wellbeing and community values of retention are assembled in the Heritage Council of Victoria's 2023 synthesis, Why heritage? A synthesis of evidence for the social, economic and environmental impacts of cultural heritage. The interim report already cites the Heritage Council of Victoria (2023) for the proposition that heritage protections serve current and future generations. It should also draw on the evidence base the Council has compiled.
Disclaimer: Dr James Lesh has contributed to the submission prepared by Australia ICOMOS; however, the views presented in this article are our own.
