Unaffordable Housing – the case against land use planning

Successive governments have largely bought into the view that the rise in New Zealand’s housing prices is attributable to inefficient land markets. One major consequence is that a primary focus of urban planning has been shifted, to solve the housing affordability problem, rather than broader aims for liveable and sustainable - and affordable – cities. The apparent belief is that the best way to solve housing affordability issues is to change the planning legislation, especially to lessen the role of statutory planning and apply a stronger ‘competitive urban land markets’ philosophy.

It’s a very big call. If planning is not the primary driver of house price growth, then change to the planning system will not be enough by itself to address the problem. Affordability issues will remain unsolved while our cities will not perform as well as they might have.

There is significant time pressure to get this right. The new statutes to replace the RMA are evolving at rapid pace. The recent Exposure Draft for the Natural and Built Environments Bill had limited detail with important gaps not addressed and is not yet able to do full justice to the Randerson Review. Ministries are working hard but are under strain (reportedly) to develop policy within the tight timetable.  Any external review work has faced very tight deadlines.

These high stakes highlight the importance of being very sure about the key propositions – first, that regulation and planning by councils has been the fundamental driver of housing price growth across New Zealand, and second that the competitive urban land markets (CULM) philosophy offers a suitable path both for solving affordability, and for planning our built and natural environments. Questions include whether the apparent solutions may be based more on philosophical considerations than on economics and urban science.

This paper explores these matters.

Planning’s Role

The core of the argument is that planning has pushed up housing prices by constraining housing growth, through imposing stronger standards-related requirements on new builds and not freeing up sufficient land for expansion, especially urban form strategies slowing outward expansion. With population growth driving demand for more land, councils’ inability to enable extra capacity caused shortfalls in housing supply, pushing up prices. 

Housing prices have risen substantially over the last two decades, increasing well ahead of growth in real household incomes. Dwelling ownership rates have declined. Housing construction has lagged, especially in the post-GFC downturn. The supply and affordability issues are high profile, with regular media reports that housing here is among the most unaffordable in the world.

The big picture is useful. Housing price rises were rapid from 2000 to 2008, then again from 2013 to 2016, and most recently 2019 to 2021. And those rises have occurred in every region. Price rises were fairly consistent across the country in both magnitude and timing, particularly in the GFC lead up (when around half of the total price growth since 2000 occurred).

That consistency in magnitude and timing across the country is important. Clearly, planning must have had major impacts, long-term and in every TLA area. All or most plans must have impacted prices to about the same degree, and at the same time. Otherwise, the extent and timing of price rises would have varied across different parts of the country.

This highlights a key issue. New urban development is the main mechanism through which planning constraints can impact on housing prices. If growth is fast, the effects of planning constraints will be relatively high, because of greater pressure for land and housing, with new builds a bigger share of the dwelling estate. But with slow growth, planning’s effects on prices would be lower and take more time. For TLA plans to affect prices to the same degree at the same time, not only would the impacts of every plan need to be the same, but also the rate of demand growth would need to be broadly similar across the country.

However, growth rates were not similar across the country. The big picture (Figure 1) shows for all 67 TLAs housing price growth %pa (red line) and housing demand (dotted line) since 2000. One clear feature is TLAs’ consistency in housing price growth, mostly in the 6%-8%pa band. The other is the variation in housing demand[1],  some TLAs growing 2-3 times as fast as others. If planning did affect housing prices to the degree claimed, then price rises should have been higher in faster growing TLAs with high demand and therefore high impact from planning constraints, and lower in slower growing TLAs.

Nor does the observed pattern reflect some failure of housing supply lines, for example where land supply constraints caused low consenting volumes. Household growth, the core driver of housing demand, also appears un-related to price growth (Figure 2). Over the last 20 years housing price rises have occurred largely irrespective of TLA growth rates.

Figure 1: Housing Price Growth and Housing Build Intensity by TLA 2000-2020

Figure 2: Housing Price Growth and Household Growth by TLA 2000-2020

The Theories

These results are significant. The evidence that house price rises have occurred irrespective of TLA growth rates conflicts directly with the asserted relationship that housing price rises are driven by planning constraints on growth capacity. Price rises were similar right across the country, but demand growth – the catalyst for planning constraints to impact on prices – varied substantially. This anomaly shows there is more to the price rises than just regulation and constraints on urban growth. The search for causes of housing price growth needs to be wider, with closer attention to the theoretical and research base relied on to date.

Brief History

Academic and government interest in housing price growth ramped up from the early 2000s, especially in USA and Europe. There was major focus on substantial differences in land values either side of the urban edge, where the obvious contrast between urban and rural saw efforts to identify what urban land values ‘should’ be. That focus has flowed through to New Zealand, with attention to large differences between urban and non-urban land values around our cities and towns.

Two main lines of thought, with contrasting rationales, emerged on what urban land values should be in an efficient market.

The Marginal Opportunity Cost (MOC) theorem contends that urban land values are driven by the surrounding non-urban (rural) economy and the value of urban land should derive from the value of its past use (farming or lifestyle living), plus the costs of infrastructure needed to sustain its new use (urban), plus any value from accessibility to urban opportunities (jobs, goods, services) and amenity. A critical element is that land value is not influenced by its current urban use, or location. “In the efficient market …. Price is not determined by the higher value use of that land (residential vs agricultural)”. Land market performance may be assessed from differences between urban and non-urban values, with large differences showing the land market is inefficient, while housing prices should be very near dwelling construction costs plus some allowance for land. These assumptions are core to the CULM understanding of how urban economies work.

This CULM view had strong influence on policy in New Zealand, through the NPS-UDC (2016) and now the NPS-UD (2020), especially through its provisions such as Objective 2 for planning to “support competitive land and development markets.”  It has guided views of what constitutes a well-functioning urban land market, and the related contention that regulation and planning has been the main driver of rising house prices (reflecting especially the views of US economist Ed Glaeser). That influence is in place through the Price Efficiency indicators to be used for NPS-UD compliance, and directive intensification provisions of Policy 3, and seems behind recent efforts to consolidate CULM concepts into new planning legislation. The underlying view is that the main route to lower housing prices is through greatly increasing land capacity.

The other main paradigm has the city as an Urban Spatial Economy.  Based more directly on central place theory and urban geography concepts, this sees land values as driven by the combined influence of land use and location and time (markets developing cumulatively, retaining much of their history). Fundamentals of urban function - economies of scale, transaction costs, externalities – are major elements, in contrast to the CULM position.  Cities do not reach a ‘spatial equilibrium’ condition, instead urban economic processes are equilibrium-seeking, with land values driven by current and potential use, as markets look forward rather than back.

The urban economy itself drives land markets and urban values, not the surrounding non-urban economy. In an efficient market, urban values are much higher than non-urban values, because of higher intensity of use and corresponding returns, plus the high costs of urbanising land. Major differences in land values between urban and rural reflect efficient land markets and show an efficient urban growth path. This is clearest at the urban edge - as a city expands, high-yielding urban uses displace lower-yielding non-urban uses which cannot compete for land.  The costs of urbanisation are high, and urban dynamics mean that (as well as intensifying developed areas), the most attractive (highest value, lowest cost) new location to urbanise next is that just beyond the urban edge, and least distant from the centre. The most efficient outward path is incremental growth, taking up the minimum extra land and infrastructure needed to accommodate growth.

The Urban Spatial Economy approach thus offers very different interpretations of the key indicators drawn on to support the view of planning as the main driver of housing price rises.

Assessment

Conveniently, the two contrasting positions on planning’s impact may be succinctly[2] assessed by examining the two Price Efficiency indicators referenced in the NPS-UD (presumably as being the best tools available to show this). They are part of councils’ mandatory assessments under s3.23 of how “planning decisions and provision of infrastructure affects the affordability and competitiveness of the local housing market”. The indicators are the Rural-Urban Differential (encapsulating the MOC approach), and the Price Cost Ratio (the positions advanced by Glaeser). Both are cited in NPS-UD guidance as indicators of the need for TLAs to provide for more housing capacity.

Assessing their validity offers a direct route for conclusions about planning’s impact on housing prices. It also offers further basis for examining the CULM focus on providing much more land capacity as the solution to housing affordability issues.

The Rural-Urban Differential

The Rural-Urban Differential (RUD) is based on the MOC theorem. It compares land values either side of the urban edge, on the proposition that a substantial difference between urban and non-urban values is attributable to how land-use regulations are “constraining urban development capacity …[and] how much urban residential land values are being elevated because of these regulatory constraints.”

However, examination of the RUD method and its core assumptions shows that interpretation is not valid. The fundamental concern is the RUD position that land value is not influenced by use or potential use, or by location. That position is directly contrary to standard economic concepts and theory, which hold that the value of the land resource – and any resource - is driven predominantly by the potential returns generated by its use. For example, the Total Economic Value approach, widely used in resource economics, identifies five drivers of resource value, three of them Use values, and two Non-use values. That is consistent with how markets function, where land value arises from the price a potential purchaser is willing to pay in an open-market transaction. The value to a purchaser reflects the anticipated return from utilising and owning the land. Accordingly, land value reflects especially its potential use, arising from inter alia its characteristics and location, combined with zoned enablement.

That potential is the core reason why urban land is more valuable than non-urban or rural land – it can sustain a greater economic return. This is no mystery. Urban residential land, supported by infrastructure, carries many more dwellings than rural residential, and its value reflects that difference in earning potential – for example, in Auckland both carrying capacity (dwellings per ha) and value ($/ per ha) of urban residential land is over 30 times that of rural lifestyle blocks.

It means that in an efficient market, urban land values should be much higher than non-urban values. A strong differential does not mean that regulatory constraints are inflating prices, rather it shows how the market itself places higher value on urban land because it can generate greater returns, the higher intensity of use sustained by urban infrastructure. In addition, urbanisation involves substantial costs, from bulk and local infrastructure, land development, subdivision, legal and sales costs, ‘loss’ of land area for roading and reserves, plus development margin, financing and contingencies.  Studies into urbanisation costs commonly show the initial value of the land in its pre-urban use represents 7 to 12 per cent of its final value as urban residential land.

These underlying economics of land use and land value are well encapsulated by the valuation sector, with its core responsibility to attribute value to land.  The Generally Agreed Valuation Principles (GAVP, of the Property Institute of New Zealand) hold that potential use is a fundamental driver of land value. The principle of highest and best use of land, which underpins land valuation is defined as “the most probable use of a property which is physically possible, appropriately justified, legally permissible, financially feasible, and which results in the highest value of the property being valued”.

More issues arise from the logic of MOC theorem - initially land value derives from its use for farming, but thereafter land use is excluded as an influence on value – and from the assumption that farming represents the opportunity cost for land – which can apply only where there is no potential for higher value urban use, which means a city has no growth potential. Neither condition is likely.

Further, the coarse application of the RUD – its 2km swathe inside the urban edge picks up large parts of cities like Hamilton and Tauranga so it includes land urbanised for decades, while the 2km outside the edge includes non-urban land whose urbanisation potential may lie decades in the future – means it shows value differentials far greater than those at the current urban-to-non-urban interface.

 Its weak conceptual foundation, and evidence base mean the RUD is not robust as a method to show the effects on values of constraining land supply. A strong differential between urban and non-urban values instead shows an efficient urban economy, in which the land market itself confers higher value on urban land because of its potential to be utilised more intensively than non-urban land, including its more central location. This pattern is consistent with the urban spatial economy paradigm.

As such, the basic premise of the RUD does not stand up, and the characteristic value differential across New Zealand’s urban network does not show that land use regulations are constraining growth.

Price Cost Ratio

The second Price Efficiency indicator is the Price Cost Ratio (PCR). Picked up from work by Glaeser, the PCR is calculated from the land value share of total value expressed as a ratio. A threshold value of 33% (for a standalone dwelling) was chosen. Where land value is 33% or less of total property value producing a PCR of 1.5 or lower[3], then the land market is deemed to be “operating well”. A PCR above this 1.5 threshold indicates “..it appears there are constraints on the supply of infrastructure-serviced sections relative to demand.”

The PCR is calculated for all TLAs on the MHUD Dashboard to indicate how regulation and planning are affecting housing values, and to inform councils’ s3.23 reporting of planning’s impacts. Currently the PCR values of all Tier 1 and Tier 2 cities are above the 1.5 threshold, which would indicate that values throughout the urban network are being pushed up by constraints on land supply. The PCR values for Tier 1 and 2 cities, shown in Figure 3, indicate relative uniformity among all cities, especially for shifts over time which appear heavily influenced by national trends in housing prices. That consistency highlights issues raised earlier.

Figure 3: PCR Trends in Tier 1 and Tier 2 Urban Economies 1993-2021

Direct examination of the PCR shows critical shortcomings. These severely constrain its usefulness as an indicator of market performance because it applies strict normative conditions and assumptions about housing markets both city-wide and long term. Moreover, they show it is not a robust method to assess the impacts of planning on prices.

As an indicator of market performance key limitations include:

• It is based on a ‘normative’ assumption that the land value should be 33% of a property’s total value. The chosen 33% is well below the 40-43% share identified from research of new dwellings across the urban network[5].

• It assumes property value should derive from construction cost (current dollars) plus land (the underlying view is that land and dwellings are commodities). However, housing values reflect other site improvements not just a dwelling replacement cost, and land value is influenced by location, time, site size, and potential site use, rather than only the value of the dwelling.

• Itassumes land should be the same fixed share of total property value (updated by construction PPI) throughout its economic life, and this applies to properties city-wide. However, the value structure of an individual property changes over time, as the built structure ages, and the urban economy evolves.

These assumptions are especially problematic when applied to entire cities[6]. A property’s PCR value is typically lowest when the dwelling is new, and it increases over time. The evidence is that urban growth sees housing land being used more intensively, and the land share of property values increases as land values – driven by land’s potential use - grow faster than (depreciating) built improvements. This potential is realised as the economy grows. At any point in time, a PCR for the whole city will include older dwellings of smaller size, on larger lots, and in higher value more central locations than more recent builds. These conditions mean the average PCR value will be well above the normative, new dwelling value, and the effects are not offset by inflation-adjusting replacement costs. For a city, a PCR value (1.50) based on a ‘normative’ new dwelling is not a suitable benchmark.

There are also critical conceptual problems with the PCR formula itself, meaning it is not a robust basis for assessing market performance. The PCR specifies that land is a fixed share of total value, for every property. The land value and the total value of every property is therefore determined entirely by the replacement cost of the existing dwelling. There is no scope for other factors to influence land value. Consequently, the entire development potential of every site is assumed to be fully captured by the current dwelling. This means every residential property is assumed to be already developed to its maximum potential. There is no scope for that potential to change over time. At the city level, this means there is no potential for any growth to occur by intensifying the already built-up area (a problem similar to the Alonso-Mills-Muth model). Any housing growth must be by take up of additional land.

This is problematic because the real-world conditions in New Zealand cities are quite different. Research shows a key driver of urban growth is intensification of already urbanised land.  On the Auckland isthmus, for example, over two-thirds of residential properties show potential for more intensive use.

In addition to these matters, the more basic issue is that the PCR is not robust as an indicator of the impact of planning on land values.First, it is presented as an indicator of only “the supply of infrastructure serviced sections”. However, the nexus is very weak. It is relevant at most to dwellings recently built or in the pipeline whose value will be affected by the price of new sites, but the PCR calculation does not include vacant or unimproved sites which are without dwellings. The apparent rationale is that the mean PCR value of all sites with dwellings including those built many decades ago is able to show the effects of a land supply shortfall. There are many influences on the value structure of an urban dwelling estate. Assuming a supply shortfall is the sole or even a primary driver is a very long bow.

Second, the interpretation offered for PCR values is not accurate. Rather than indicating a shortfall of new sites, a high PCR is an indicator of potential supply of more housing capacity through intensification. There is a strong nexus - a high land value relative to total property value (high PCR) indicates the potential for feasible intensification. That land value is conferred because the market recognises the development opportunity, including the low opportunity cost represented by existing built improvements. To illustrate, a large site with a small dwelling has potential to intensify by developing terrace houses, with a high PCR because the market confers value based on the land’s potential, not on the existing dwelling. Since land value reflects development potential, lots with high potential to intensify and add more dwellings (including from up-zoning) will have high PCR values. A high average PCR for a city will reflect predominantly its intensification potential, including for developments consuming less land area per dwelling, for a more efficient land market.

The problem is that high PCR values are consistently misinterpreted as showing a constraint on land supply when instead they more accurately indicate the potential for additional housing capacity. Examination shows the PCR is not a reliable indicator of either a shortfall in section supply, or of the impacts of planning on housing prices. The view that a high PCR shows high prices because of a shortfall in section supply due to planning is not accurate.

Other Research

Other research findings do not support the position that planning has had major impact on prices:

• The Auckland Chief Economist Unit’s comprehensive evaluation into the effects of the Rural Urban Boundary (RUB) on land values established a quite low impact, finding “a price premium of at most 5.2%”. The findings contrasted with works by Lees.

• Papers examining housing price trends have identified a range of influential factors, including the ease of accessing finance and high consumer confidence (especially in the lead-up to the GFC), constraints on construction capacity, supply shortfalls, strong inward migration (especially in the 6 years post-GFC), substantial investment from overseas into New Zealand’s housing market (until 2018), interest rates, and the taxation environment.

• Most recently, assessments for individual TLAs for NPSUD s 3.23 compliance have found that at the TLA level, housing price growth has been driven predominantly by national trends, rather than local effects.

Conclusions

This examination shows neither of the two main Price Efficiency indicators is a suitable tool to assess a shortfall in capacity for housing, or the effects of planning on housing prices. Nor do they show that a TLA should provide for more housing capacity.

The wider issue is that both indicators are based on particular views of how cities function, which do not align well with economic concepts and understanding of urban spatial economies and land markets. Nevertheless, together with the directive intensification provisions, they are an important part of the CULM position, underlying the NPS-UD’s impetus to enable much more capacity as the primary solution to housing affordability.

That highlights wider concerns about how CULM position interprets the operation of (urban) economies and land markets, as a basis for future planning of the built and natural environments.

Implications for Planners

These findings are important because papers, website material and public pronouncements about the impact of planning have drawn from these two indicators. The contention that regulation and land use planning has been the major driver of rising house prices has been widely advanced for a long time and has become the ‘conventional wisdom’ in some quarters.

However, if the impact of planning on housing prices has been significantly ‘over-cooked’, then it is important to look very carefully at established or proposed provisions or policies which have relied on that ‘over-cooking’. That applies particularly to the forthcoming Natural and Built Environments Bill, and to the strong initiatives to implement the CULM approach on the basis that regulation and planning have been the dominant driver of housing price rises, and that reducing the role of planning will solve New Zealand’s housing affordability issues.

These findings highlight the need to look more widely for causes and solutions, and especially the importance of having an evidence-based approach going forward.

Article written by Dr Douglas Fairgray and published in the New Zealand Planning Quarterly Issue 222/October 2021.

Endnotes

[1] Housing demand estimated as housing build intensity or annual dwelling consents per 1000 resident households

[2] A process rather faster than examining the whole evidence base. The NZ literature and research base on these matters is a combination of reports and published papers, plus (lower-profile) avenues of expert evidence, and technical studies for TLAs’ NPSUD compliance.

[3] The formula: PCR = 1 / (1-LV%); Threshold PCR of 1.5 = 1/(1-.33)

[4] A study of over 65,000 new dwellings in Tier 1 and Tier 2 cities built 2013-2017 showed a norm of 40-43% for feasible new stand-alone dwellings, using data from Corelogic 2018

[5] The Dashboard calculations are based on recent sales, so represent the wider estate.