Does Auckland have a land deficit, rather than a housing deficit?
One of the interesting features of Auckland is its geography. Its narrow Isthmus form with harbours and bush clad ranges on two sides plus some hills to the north and south makes for a great living environment with these natural areas so close by. But this landform also limits developable land close to the centre. While it is true that there is plenty of land further out to the north, north-west and south-west, proximity is vital to the functioning of urban areas.
Someone once said that "anatomy is destiny", I think in relation to humans, but it could equally apply to cities.Along with a housing deficit, Auckland has a land deficit. How we respond to that deficit is matter of transport and land use planning. Leap frogging about is not likely to be the answer.
The effect of the land deficit on land and house prices is beginning to be noted, which is helpful. For example, Lees in his 2014 NZIER paper (note 1) said that that the impact of Auckland’s narrow geography is equivalent to adding about 900,000 residents or moving the city up to 2,200,000 residents – about the size of Brisbane. So in his view we should expect big city outcomes in terms of housing and commuting times even though Auckland’s population is much smaller than our trans Tasman peers.
Even the Auckland Transport Alignment Project acknowledges geography. The above picture is from the ATAP Foundation report.
The point is that it is easy to grab some other place like Houston or Montreal or Christchurch and say “look they have low house prices” while ignoring that their geography is different. As Lees notes, finding a place for people to live and work in Auckland is harder than many other cities.
A classic study on geographic constraints and house prices is a 2010 paper by A Saiz: ‘The geographic determinants of housing supply’ (see note 2). Saiz looked at house prices and land availability across a range of American cities. The maths is a bit hard to follow, but the gist seems to be the following graph of the elasticity of housing supply compared to city size, as organised by cities that are geographically constrained and those that are not.
A low elasticity means that there is a low response from house builders to increases in house prices. Geographic constraints see elasticities rapidly decline as city size increases. So when Auckland was 500,000 people, it was much easier to respond to population growth than when the city is 2 million people.
What is geographically constrained? Saiz took into account harbours, lakes and rivers and land steeper than 15 degrees, within a 50km radius of the centre. Cities ranged from the likes of Houston which had 9% of land constrained by these features, to San Francisco which had about 75% of its land not developable. Guess which city has cheap housing and which city has expensive housing?
Saiz found that most areas that are widely regarded as supply inelastic were severely land-constrained by their geography. Using a comprehensive survey on residential land use regulations, he also found that highly regulated areas tend to be geographically constrained. His hypothesis is that the geographic constraints drive up land values, which mean that there is more of an incentive for existing landowners and residents to seek to both protect the amenity of their neighbourhood, as well as the wider environment within which they live as often the geographic constraints are valued features like coastlines and mountains.
So the two go hand-in-hand. He also found that due to the low elasticities of supply for geographically constrained cities, demand shocks (like a leap in population) are not very good for the housing sector.
I think the subtext of his study is that mixing high growth with geographic constraints in a city over 1m people is a bit of a recipe for trouble. Sound familiar?
The relative lack of land close to the centre of Auckland does not explain all of the increase in house prices seen over the past five years, nor all of the slow housing supply response seen over the past few years. Geographic constraints do not automatically mean high house prices and limited supply. Constraints just mean that the city’s housing market and transport system needs to respond in a different way to less constrained cities.
But geographic constraints do help to explain high land prices, relative to other cities. Higher land prices reflect higher demand. The same number of people needing to live in a smaller land area, compared to the same number being able to spread out over a larger area is bound to raise land prices. In theory at least, in response to higher land prices, the land area per dwelling unit should reduce, with stand alone houses replaced by townhouses, flats and apartments. Density increases. This helps to moderate price increases of the dwellings. But to accommodate more people in the existing footprint, sections need to be redeveloped and infrastructure expanded. Amenity characteristics change. None of this is easy.
So how much of a land scarcity penalty does Auckland face?
If we break urban Auckland down into four rings, based on a simple radius from the centre (foot of Queen Street) then, roughly, we get the following figures on land area if there was no harbours or hills versus what is present. The figures on the actual area of land are a bit of an estimate based on stats NZ data, as later on I want to correlate these areas with dwellings.
At a 20km radius, about 50% of the area is sea or islands. So I would class Auckland as being geographically challenged.
Distance band
(kms) |
Theoretical
area (ha) |
Actual area ha
(estimate) |
Actual as percentage
of theoretical |
0 - 5 km
|
7,850
|
5,285
|
67%
|
5- 10 km
|
31,400
|
24,728
|
79%
|
10 -15 km
|
70,650
|
45,269
|
64%
|
15 - 20 km
|
125,600
|
61,141
|
49%
|
I have stopped at 20km. Auckland extends up to 30km to the north and south (Orewa and Pukekohe) but I wanted to keep the analysis to what might be a reasonable commuter shed. A 20km radius is probably more like a 30km on the ground trip from the edge of the 20km circle to the CBD. That in turn is probably about the 30 to 45min commute that most people will think is reasonable (on a good day). Don’t forget that the median distance travelled to work in Auckland in 2013 was 6.0m kms!
The amount of land that is not buildable varies by distance. Within 5km of the CBD, about 1/3rd of the theoretical area is water. By 15km out, about 35% of the land area is taken out by harbours.
If we look at the number of dwellings in these bands and the density of housing, as of 2013, then we get a broad picture of decline in density out from the centre. The density figures are gross density - it is total land area divided by the number of dwellings. The land area therefore includes parks, business areas and open spaces.
Distance band from
CBD (kms) |
Dwellings 2013
|
Land area (ha)
|
Dwelling density
(dwellings per ha) gross |
0-5 km
|
63,084
|
5,285
|
11.94
|
5-10 km
|
119,544
|
19,443
|
6.15
|
10-15 km
|
126,837
|
20,541
|
6.17
|
15-20 km
|
49,863
|
15,872
|
3.14
|
Total within 20km
|
359,328
|
61,141
|
5.88
|
The density gradient in the two middle rings (5 to 10km and 10 to 15kms) is essentially flat. It would be expected that the density of the inner middle (5 to 10km) would be higher than the density of the outer middle (10 to 15km).
The total number of dwellings within 20kms - 360,000 - is about 75% of the regional total.
If we then ask what would Auckland look like if there were not the harbours and the hills to contend with, then when can get an idea of the number of dwellings that could be accommodated in the 20km footprint, at current densities. That is, we assume that Auckland is on a flat plain, away from the coast, in which the area in each circle is a simple matter of maths.
I have cut back the theoretical land area by 20%. This is to take account that for most cities, there are always some features like rivers, streams and hills that limit developable land.
I get the following figures.
Distance band
(km) |
Theoretical Area
(ha) |
80% of theoretical
area (ha) |
Density of
dwellings (current) - Dw per ha |
Potential
dwellings |
0-5 km
|
7,850
|
6,280
|
1 1.9
|
74,961
|
5-10 km
|
23,550
|
18,840
|
6.1
|
115,836
|
10-15 km
|
39,250
|
31,400
|
6.2
|
193,889
|
15-20 km
|
54,950
|
43,960
|
3.1
|
135,103
|
Total within 20km
|
125,600
|
100,480
|
522,790
|
Based on existing densities, then Auckland could easily accommodate an additional 160,000 dwellings within the 20km radius if it was on a (mostly) flat plain.
That is, the ‘missing’ land area covered by Auckland’s harbours and hills could accommodate 160,000 dwellings (or about 50% of the current stock). Say 400,000 people. This gives an indication of the extra pressure on Auckland’s actual urbanised area, because of its limited land area. Basically we need to bunch up some to account for the constraints.
The above figures assume a pretty low density of development out on the edges, in the 15 to 20km band. It also assumes a fairly low density within the inner band - the 0 to 5km band. Arguably both should be higher.
In the inner area (0-5km band) heritage suburbs dominate. As discussed elsewhere, land values in the inner band would easily support mid rise apartment developments. If we assume a density of 20 dwellings per ha (equivalent to areas like Grafton with a mix of apartments and older dwellings), then the inner area could accommodate quite a few more houses,
In the outer band, a density of 4 dwellings per ha (gross) would still allow for mix of open spaces and housing (the kind of fretted urban edge that we see elsewhere).
If we take both adjustments, then we get the following figures:
Distance band from
CBD (km) |
Area ha
(80% of theoretical) |
Assumed dwelling density
(dw per ha) |
Potential dwellings
|
0-5
|
6,280
|
20.0
|
125,600
|
5-10
|
18,840
|
6.1
|
115,836
|
10-15
|
31,400
|
6.2
|
193,889
|
15-20
|
43,960
|
4.0
|
175,840
|
Total
|
100,480
|
611,166
|
So dwelling capacity within 20km could easily be 250,000 more than now if we didn’t have land constraints and heritage constraints. The 250,000 more dwellings is also close to the number of dwellings that need to be accommodated in the urban area over the next 20 years. It is not too far off the extra 900,000 people that Lees estimated.
So we probably would be a bit more like Houston or Montreal if we had the land. Planning would be easy:
- There would be a more intense level of development close to the CBD with no heritage suburbs to contend with.
- The middle rings of suburbia - the 5 to 15km range -would not need to change that much. This would keep the punters happy.
- Out on the edge of the city, subdivision and development could be fairly haphazard and scattered. It probably doesn't matter too much, as distance wise, there is no big penalty to development being separated and 'stringy' in form.
My estimate of ‘missing land’ and dwellings due to topography and heritage sits between these two figures. Perhaps a 30% reduction in house prices, or to put it another way, are land prices 30% higher than if Auckland sat on an empty plain and could redevelop its core? Currently median house prices in the Auckland region are around $850k (Real Estate Institute of NZ) compared to $450k in Christchurch. If Auckland was 30% lower, then the median house price would be closer to $600k, more in line with the other metros and household incomes.
But Auckland is not like that. We have three problems to contend with:
- First, all the houses that would normally be close to the CBD, but cant be built there due to the harbour and heritage issues. Very crudely this is about 60,000 dwellings. These would be households that would like to live close to the CBD, but can’t.
- Secondly are the households that would like or need to live on the edge in the 15 to 20km band that can’t find a house (probably a cheaper house) due to the restricted footprint ( due to harbour and hills in this case). Again a back of the envelope exercise would suggest that this is around 120,000 dwellings, if we assume that the expected density of the area that could be settled would be around 5 dwellings per ha (gross).
- Thirdly are the middle rings- 5 to 15km - here the restricted footprint from the two harbours sees about 70,000 houses less than might otherwise be the case.
In theory Auckland could make up for its land deficit by extending itself to the north and south, but to do that, it needs a radically different transport system. Beyond about 25kms from the central area, and travel times start to get well beyond a comfortable commute. The Marchitte constant kicks in strongly. Building more motorways north and south doesn't seem to help that much, as congestion quickly builds and travel times lengthen. Fast rail or transit services may be able to provide competitive travel times to new edge communities, rather than motorway extensions, but rail will only serve major destinations like the Central Area. That might suit the lawyers and financiers that work in the central area, but they will want to live by the coast, not in the back of Kumeu or somewhere near Pukekohe.
Notes:
1. Big city life? Challenges and trade-offs for Auckland city NZIER public discussion paper Working paper 2014/02, May 2014
2. https://mitcre.mit.edu/wp-content/uploads/2014/03/The-Quarterly-Journal-of-Economics-2010-Saiz-1253-96.pdf
3. Land Supply Constraints and Housing Prices in New Zealand November 2016 Technical Report 2016/038
