Dispelling Some Urban (limit) Myths

August 2016

The debate around the benefits and costs of urban limits rolls on – or perhaps it rolls out. The view that urban limits have had a major effect on housing prices seems widespread, among politicians and across the media. A number of reports have concluded that urban limits have acted to inflate land prices, with flow on effects on housing prices, and these – particularly the Grimes and Liang 2009 paper which claimed the Metropolitan Urban Limit (MUL) had a strong “boundary effect” on prices - have received considerable attention in political statements and in the media. Both National and Labour have stated that they want urban limits gone, primarily because they see them as pushing up the price of housing. In contrast, research on urban limits which offers views and evidence to the contrary has to date received little publicity.

Nevertheless, the Auckland Independent Hearings Panel (IHP) recommended retaining a defined long term urban edge through the Rural Urban Boundary (RUB). However, they recommended placing it at District Plan level rather than at Regional Policy Statement (RPS) level - a change which means that once the Plan is operative there will be much greater scope for future challenge by those seeking to have the RUB extended to include more rural land. Auckland Council has voted to accept this change. Notably, the IHP recommendations themselves include several references to the Grimes and Liang paper, apparently demonstrating the Panel’s view as to the effects of urban limits on land prices.

In the wake of the IHP recommendations, commentator Gordon Copeland1 thundered against urban limits, holding the MUL responsible as “the evil which causes land banking”, claiming that “those who own the land just inside the line have become multi-millionaires overnight” since their “land has increased in value by 1000 per cent because of the urban limit.” Mr Copeland cites the Arthur Grimes’ study, and its conclusion that “the land just inside the metropolitan urban limit was 10 times the value of the land just outside it.” Mr Copeland prays that the Government will “outlaw the creation of the new artificial boundaries in the Unitary Plan for Auckland.” 

Heavy stuff, indeed. Against that backdrop, in this issue of MEMO we take another look at urban limits, including their pros and cons, and how they influence the costs of cities. Drawing from research into the key processes involved in the conversion of rural land to urban land, we have examined in more detail the Grimes and Liang 2009 paper and others following which - based on the frequency with which they are referenced – have had considerable effect on the views of politicians and decision makers, and which appear to underpin the Productivity Commission’s perspectives on urban matters.

Pros and Cons of Urban Limits

While not new, the debate about the effects of urban limits is steadily becoming better informed, as more robust research is published.

The core rationale offered for having urban limits is to achieve more efficient and sustainable cities. The main direct effect of limiting a city’s geographical extent for any given size of the urban economy and population, and the speed with which a city expands, is to achieve a relatively compact urban form, and enable efficient and co-ordinated delivery of core infrastructure. This compact form arises from the combined effects of specifying where new urbanisation can and cannot occur, and identifying the desired balance between outward expansion and intensification of established urban land. Rarely are urban limits applied by themselves to manage urban growth, and most commonly they are part of a wider suite of planning provisions that seek to increase the opportunity for (residential) intensification within an existing urban area2. 

Generally, a more compact urban form offers greater efficiency in the provision and utilisation of service infrastructure (waters and transport), and more efficient utilisation of urban land inside the urban edge and rural land outside the edge. A more compact urban economy will function more efficiently than a more dispersed city, because key transaction costs – of moving people and goods – are lower. Agglomeration benefits – another fundamental driver of urbanisation - are enhanced, so that the urban economy is more productive. A more efficient urban economy delivers benefits at the macro-level, especially through enhanced overall efficiency and sustainability, and at the micro-level through the lower costs of operating and living in the city, for businesses and households.

Common arguments against urban limits are that they do not necessarily reflect the preferences of landholders and commercial markets, and most especially that defining the urban edge as a limit acts to constrain the amount of land which is available for greenfield expansion. This supply constraint is seen to act to push up the price of land, and inflate the cost of housing, thereby harming housing affordability. 

Urban efficiency and sustainability – which generally underpin the rationale for urban limits - are multi-faceted and long term issues, and it is important when assessing their effects to do so with an appropriately long term and comprehensive perspective. Awareness of this is resulting in less research emphasis being placed on specific and often shorter term effects – such as on land prices per se – and greater emphasis being placed on combinations of effects over time and space – such as including households’ costs of transport and housing over their life-cycle, the provision of public and private infrastructure and facilities, and the operation of the whole urban economy, into the long term. At the heart of the debate are the trade-offs between any long term effects on land and housing costs, and the wider effects of urban form on a city’s efficiency and sustainability.

Understanding these effects derives from not just knowledge of the core economic processes which underpin urbanisation, but recognition of the influences of space and location, and the time over which they operate. It is not always easy to take proper account of how economic effects will accumulate and evolve over time. This is especially the case in long term planning, when a projected outcome - such as a 25-year population projection - may easily be given a false immediacy which can act to over-emphasise the end point itself, and under-emphasise the evolution of economic influences which will have effect throughout the intervening period. That is, economic effects within that 25-year time frame will not be simply spread pro rata across space and through time, but will be strongly influenced by the current urban spatial economy, and will progressively accumulate throughout the future, so that outcomes in the end 5 or so years may be quite different from those evident after the first 5 years. 

The debate on urban limits is also characterised by quite different perspectives. Some see urban limits primarily as a constraint to growth, which limit and distort the operation of commercial markets, and (unnecessarily) limit the outward expansion of cities. Others see them as a quite appropriate method for a council to carry out their RMA responsibilities, since land use is a core driver of effects. In the simplest terms, effects arise from “what happens where, and when”. The direct and flow on effects of land use patterns can be readily anticipated, at the macro-level as well as at the micro-level. A core role of territorial authorities is to manage effects by identifying appropriate urban form outcomes.

Close to the heart of the debate is the philosophical perspective. Some consider that commercial markets with minimum direction or constraint will in aggregate produce the overall outcomes which society desires, as well as serving the aspirations of individuals and entities within the community. Others consider that better outcomes are achievable when the wider community sets the limits – including spatial limits – within which the commercial market may operate. New Zealand’s statutory base clearly reflects the latter position, not least by the existence of the RMA. That is, if commercial markets by themselves were expected to deliver the outcomes which society desires, then there would be no need for the RMA. The planning and governance legislation of many nations suggests that New Zealand is not alone in this view.

But enough philosophy. At a practical level, it is very important that any debate is well informed, and is able to draw on a good evidence base. In this context, we have examined some key papers which appear to have strong influence on political stances and decision-making around urban limits.

Grimes and Liang 2009

The paper by Grimes and Liang3  (2009) Spatial Determinants of Land Prices in Auckland: Does the Metropolitan Urban Limit Have an Effect? has been widely quoted. The paper sought to “examine whether the existence of this growth limit affects land prices” (p23). The authors concluded that they had found “a strong zoning boundary effect on land prices.” and that “..land just inside the MUL is valued (per hectare) at approximately 10 times land that is just outside the boundary.” (p23). It further noted that “All our other estimates of the boundary effect find a boundary land value ratio of between 7.9 and 13.2…” (p43). 

These conclusions have been widely interpreted as showing that the MUL was responsible for a 10-fold increase in land prices, including most recently by Mr Copeland, and that notion is cited by the IHP as a major reason for the recommendation to place the RUB at District Plan level4. It is relevant that the Grimes and Liang analysis considered the MUL rather than a RUB located to accommodate 25 or so years’ of growth, and in their paper they noted “If a city’s current and prospective expansion is well within the growth limit, no citywide effect should be experienced and little local effect will be apparent.” (p24).

Having examined the Grimes and Liang paper, including the data used and the model applied to reach those findings, our core conclusion is that their analysis did not establish any clear effect of the MUL on land values. 

It did show that land values inside the MUL were about 10 times those just outside the MUL. However, the analysis did not establish that the MUL was the cause of those substantial differences in land values immediately inside and outside the MUL. The conclusion that a “strong boundary effect” existed was based on the authors’ belief that the MUL was the cause.

This critical difference between our conclusions and current popular belief, and the significant effect that popular belief has on our urban planning, demands some further explanation. 

To start, it is important to understand what went into the Grimes and Liang model, and how it was applied and interpreted. It is also necessary to consider the equally important question of which economic drivers were not examined. 

The authors modelled relative land value in some 8,000 meshblocks “as a function of distance from the coast, distance from the CBD, distance from other key nodes, TA effects, impacts of being inside or outside the MUL, plus a ‘rural’ variable.”(p28). That is, they sought to explain the patterns of land value in terms of the spatial distribution of those values. This research structure is important, because it denotes some key underlying assumptions. Of particular note, in the base model, all of the possible explanatory variables which were examined are indicators of location – several variables of the land’s distance from key locations in the urban economy, plus a TA variable (to show in which TA the land was located), and a variable from StatisticsNZ denoting meshblock locations as being “rural”, were analysed. An “extended” model was also applied including three ‘social’ variables - income, relative deprivation and population density in the meshblocks – to see if these added to the explanatory power of the model by accounting for some of the variations in land values. However, those ‘social’ variables did not capture other key indicators such as land zoning, the presence of service infrastructure, parcel size, land use and so on. 

That model design, and the exclusion of other potential influences on land value, meant that the analysis effectively predetermined that if any variables could be identified as the key driver(s) of differences in land values – assuming the results were statistically significant – then they would be some aspect(s) of location. Equally, if the results were not statistically significant, then the potential explanatory variables examined would be deemed to not offer sufficiently strong relationships with variations in land values. 

That model scope and structure raises a number of concerns. First, there was no scope in the model design to consider other potential explanations for the differences in land values other than the selected locational variables and the three social variables.

Second, and very importantly, the “inside or outside MUL” indicator used is effectively a binary variable. A common feature of regression analysis is that it tends to impart high significance to variables which are associated with a systematic variation in the dependent variable. In this case, land values are considerably higher inside the MUL than outside it. This meant that the model would show – quite accurately in statistical terms – that location inside or outside the MUL was closely correlated with land value. However, that is all it would show. It could not establish that the MUL was the cause of the difference in land values – attributing causality is the responsibility of the modeller when interpreting the results of the analysis. Nonetheless, the results were interpreted as showing “a strong zoning boundary effect on land prices.” 

Before concluding that there is any boundary effect, let alone a strong one, it would be necessary to consider the possible contribution to urban land values of other key drivers, rather than – apparently - attribute all of the value differential to the MUL. Which is clearly how the paper has been interpreted by many readers.

Professor of Economics Tim Hazledine is refreshingly unequivocal about this5, stating “Grimes and Liang found, approximately, a 10-fold difference in per hectare land values just inside and just outside the MUL, and it is this figure that has been simplistically misinterpreted to imply that, without the limit, housing land prices would drop by a factor of 10. But that truly is to compare apples and oranges, or coconuts and peanuts, to get the magnitudes realistic.”

In addition to assigning this causality – that the MUL was driving the differences in land values – when it is really an assumption, the analysis itself raises some questions. In particular, the coefficients suggest the possibility of “omitted variable bias”, which can arise when important variables are omitted from the model specification. That can mean that the coefficients of the dummy variables (notably the DMULs) are biased because they contain the effects of other, omitted, variables.

Nonetheless, the Grimes and Liang conclusion that a strong boundary effect exists has been picked up by others. The same methodology has been subsequently applied and extended by Guanyu Zheng in a report for the Productivity Commission, baldly entitled “The effect of Auckland’s Metropolitan Urban Limit on Land Prices” (March 2013). That papers reiterated that Grimes and Liang found “that the MUL has had a significant impact on land prices in the city, with the price of land just inside the MUL around 10 times higher than land just outside the MUL. “, and it went on to conclude that “Auckland’s MUL has significantly increased land prices in general…”piii). Because it replicated the Grimes and Liang approach, the Zheng report also replicates the flaws in that earlier study.

The Grimes and Liang conclusions have also been widely cited in reports prepared by the Productivity Commission.

Influences on Land Values at the Urban Edge

An obvious question is whether the Grimes and Liang assumption that the value differential was caused by the MUL is justified, and whether other factors un-tested in the analysis may have been the real explanation.

A key part of interpreting statistical results is to examine the economic processes leading to the outcome shown. Equally important is consideration of other processes which could also have led to that same outcome. This is critical for establishing the conceptual basis for any “explanation” suggested by the statistical analysis. A well acknowledged risk in spatial analysis is that spatial association does not necessarily show causal connection – a caveat drummed into geography and planning students early in their education path. 

This makes it important to understand the core processes by which the value of land inside the Auckland MUL came to be around 10 times the value of land immediately outside it. Such understanding may lead to different conclusions and interpretation of the statistical analysis.

A fundamental consideration is that the urban edge is not static, and that the difference in land values is arising in a location where the land is transitioning from rural or semi-rural to urban use. This transition from rural to urban is not a simple and instant process. It involves a number of changes, in the course of which the initially rural land typically incurs a number of costs, and the addition of implicit values, and commonly changes in ownership, all of which are reflected in its end value as urban (residential) land. 

This process of transition is widespread and unremarkable, and it is well understood and documented. A number of studies have shown the process of change, the sequence of commonly occurring steps, and the costs and value changes which occur in that sequence, in order to understand the changes in land value which arise as rural land becomes urban land. For example, David Mead (a director of Hill Young Cooper) in his 2014 submission on the Productivity Commission’s Housing Affordability inquiry drew on his research into urbanisation processes on the Auckland urban edge to show that the original value of rural land would account for only about one-twentieth the value of the same land when it was serviced and ready for market. Significantly, Mr Meads’ analysis does not show evidence that the presence of a metropolitan limit had an effect on the change in value as land became urban, nor does it show any evidence of substantial leaps in value during the process. It does show each step in the process, so it does not leave any unexplained gaps where value shifts may have occurred. 

Likewise, Professor Hazledine from his research on urban limit issues in Canterbury notes that “Urban land is surveyed, subdivided and serviced. It is close to roads, shops, public transport, schools. For the price that you pay for a house building lot, you get thrown in for free a comparable quantity of well-maintained public land - parks, reserves, footpaths, verges, schoolyards and indeed the roads themselves. Housing land costs more than farmland because it has had a heap of value added to it.” and that on the edge of rural villages around Christchurch “…serviced house building sections were being sold for around $200,000 - about 100 times the price the developer had paid for the farm land in the first place. Nobody was getting cheated here, and there wasn't an MUL in sight. It's just what things cost.”

More recently, from the research done for the Unitary Plan process, there is considerable information which offers insights into how this transition affects land values in the current Auckland land market. These effects on value arise from the process of change from rural land to urban land. Importantly, they would apply whether or not there was a MUL or any other urban limit. Commonly, the transition process involves (see Figure 1): 

Figure 1 - Components of urban Land Value

1. Purchase of a rural or residential property outside the urban edge, for development. On average, land value on properties in the proposed Future Urban Zone (FUZ) on the northern and southern edges of Auckland was $512,000 per ha at the 2014 valuation, although allowing for a 20% increase since then would give a starting value of about $615,000 per ha. This land had an average improvement value of around $185,000 per ha, which is part of the initial cost for a developer;

2. For urban uses, the land needs infrastructure. The 2015 FULSS Report shows a cost of $16.9Bn for bulk infrastructure to service 12,120 ha or so of FUZ land, or about $1,400,000 per ha. Add this to the cost of the un-serviced land and its improvements, and the cost gets to about $2,200,000 per ha, before any allowance for financing costs of the infrastructure;

3. Financing costs, which apply to both the property purchase itself, and the infrastructure to service it - for example, even at a 5% cost of finance, and allowing for very rapid cost recovery over 2.5 years, the financing cost of bulk infrastructure by itself would be in the order of $150,000 per ha. The financing cost for the initial property purchase would incur higher interest rates for a property developer compared with an infrastructure provider, and over a longer period, and may equate to in the order of $300,000 per ha (8% pa over 4 years);

4. In addition to the bulk infrastructure is local infrastructure and service reticulation, including roads and other assets usually vested in the local authority, plus land development costs, including professional fees, subdivision costs, and earthworks;

5. Allowance also needs to be made for the net land yield from the initial land purchase, which is generally 70-75% of the gross land area. This effectively applies a multiplier of 1.33 to 1.45 times the accrued land value;

To this point, the land able to be urbanised has accrued a value per ha of 6 to 7 times its original value per ha as rural land. That is not the end of it, as further values and costs accrue to the land, not necessarily in the order shown:

1. The effect of a zoning change from rural to urban has a major influence on land value simply because zoning dictates what uses may be made of the land, and the returns able to be generated. Since urban zoned land generally has a wider range of enabled uses than rural land, and can potentially generate higher returns, then additional value accrues to the land;

2. The effect of subdivision to smaller land parcels is important, because it is generally associated with an increase in value per m2 of land area. This is commonly evident on urban residential zoned land, where much of the value arises on a parcel basis simply because of an owner’s ability to build a dwelling on it;

3. These effects on land value are cumulative, and on top of them a key part of the process involves ownership change at one or more points along the transition path, where a development margin is usually added, which is commonly in the order of 15-20% (based on advice received from development entities in the course of the expert conferencing on the Unitary Plan).

Once this sequence of costs incurred and values added in the transition path from rural to urban is taken into account, then it is quite logical and explainable that the value of urban land is in the order of 10 times or more the value of nearby rural land. Nor should it be a surprise to anyone who has studied the underlying economic processes, and the outcomes.

Of particular note, the costs and value changes which accrue during the transition process generally do not include the sorts of quantum leaps in land value along the way which Mr Copeland rails against. One important reason for this is that the points at which value may be realised in the transition from rural to urban are well understood by most in the development sector – simply, the market is typically well informed, an important condition for efficient markets.

Implications

These matters highlight the importance of understanding what factors do contribute to land and dwelling price growth, how they contribute, and to what degree. Much of the heavy blame directed at the MUL for pushing up housing prices has relied on the conclusions of the original Grimes and Liang 2009 paper, and the resulting conventional wisdom that a 10-fold increase could be attributed to the urban limit. 

Our assessment of that paper indicates that this “conventional wisdom” may be more accurately described as an urban (limits) myth.

This finding emphasises the need to focus clearly on the key drivers of housing price growth - including the ready availability of finance, high consumer confidence, tight dwelling supply and the slowdown in construction following the GFC, and record levels of in-migration, especially to Auckland. 

The challenges posed by high dwelling prices, continued strong growth in those prices, and the major under-supply of new capacity in the lower price ranges are all pressing. They have a better chance of being addressed effectively if more attention is directed at the main causes, with less effort directed at further maligning urban limits.

For further information or comment on this article, please contact Douglas Fairgray (doug@me.co.nz or 09 915 5514)


Notes:

1. NZ Herald, 12 August 2016

2. The core rationale for the growth strategy for the Auckland Plan, and the Auckland Unitary Plan.

3. Spatial Determinants of Land Prices: Does Auckland’s Metropolitan urban Limit have an effect? Applied Spatial Analysis (2009) 2:23-45

4. While the Grimes and Liang paper is quoted in several places in the IHP recommendations, Dr Grimes did not present evidence in the Unitary Plan hearings.

5. Urban limit helps make region a liveable place NZ Herald 17 August 2016.