New dwellings and population growth

I hadn’t really intended to write anything today –  tempted as I was by the topic of so-called “ethical investing” – but yesterday’s post on how best to look at new building consents relative to population (growth) sparked a surprising number of comments so I thought some brief follow-up comments and charts might be in order.

My single main point yesterday was that new building permits per capita, whether compared across time or across TLAs, is not a particularly useful indicator of anything.  There are substantial differences in population growth rates –  both across time and across TLAs – so that simple comparisons of consents for new dwellings relative to the current stock of population won’t tell observers anything useful about how supply/demand balances are unfolding in particular markets, or how responsive land use and building regulation allow markets to be in particular times and places.

For either purpose –  and perhaps particularly for the latter – one really probably needs a more formal empirical model that can capture more of the idiosyncracies of particular times and places, and some of the two-way causation that can be at work (eg population growth generates demand for housing, but a readily responsive housing supply might also make such a locality more attractive to more people).  Fortunately, in comparing across TLAs in a single country we can treat a lot of things as constant (applying similarly across all TLAs) –  eg the same tax system, the same interest rates, the same banking system, the same trends in divorce rates, or childbirth rates (the latter two have clear implications for the number of houses demanded per capita).  But there are still local idiosyncratic features that need to be taken into account at times.  The most obvious of these in recent New Zealand history is the impact of the Canterbury earthquakes, which led to the loss of a lot of existing houses, especially in Christchurch city and Waimakariri (Kaiapoi).    Even if the population of those places didn’t change much at all, one would expect a lot of new dwelling consents in the years following such destruction simply to re-establish the previously desired volume of housing.  Seeing a lot of new dwelling permits in those (and neighbouring) localities might not tell one much about the responsiveness of the regulatory systems in those council areas, but simply about the specific nature of the shock.  And –  fortunately –  we don’t know how other localities (and their regulatory systems) would have responded to a natural disaster of that sort.

Building permits per capita don’t tell us much at all.  Building permits for new dwellings per person increase in population tells us more, but it is still a far from perfect measure –  especially when, as around Christchurch, there is a sudden need to replace existing lost houses.  So in my post yesterday I used the SNZ national data on housing stocks, and compared the (estimated) change in the housing stock to the (estimated) change in population.  This was the resulting chart.

housing stock

At a national level, the net increase in the number of houses has been very weak relative to (estimated) population growth, and there is no sign of any improvement.   It isn’t a perfect indicator –  changing birth rates or divorce rates might affect the desired number of people per house – but it is less bad than anything else we have.

What about at the TLA level?  We don’t have annual housing stock estimates (that I’m aware of) and the latest annual subnational population estimates are for June 2015.  So we are pushed back to using new dwelling consents.  Comparing consents with population growth produces silly answers in places with falling populations –  where there is usually some new building just to slowly replace the existing stock –  or even places with very low population growth rates.  So in what follows I’m just going to focus on places that are

  • relatively large, and/or
  • have had reasonable population growth

but with a particular focus on Auckland, greater Wellington, greater Christchurch, Hamilton and Tauranga.  The readily accessible data go back to 1996.

Here is an easy-to-read chart comparing the experiences of Auckland and Hamilton.  Both cities have had around a 40 per cent increase in population over the period.

akld and hamilton

But in only one year of these nineteen were more new houses being built per each new resident in Auckland than in Hamilton.  There might be some underlying demographic differences  –  as I said, ideally one needs a fuller empirical model –  but on the face of things it doesn’t reflect very favourably on the land use and building restriction of the Auckland council(s).  At least up to June 2015, there was no sign of the gap closing.

Tauranga has actually had faster population growth than either Auckland or Hamilton over the 20 years.  Here is what the chart looks like when we add Tauranga.

akld hamilton tuaranga

Pretty consistently higher (apparently more responsive to changes in demand) than Auckland in particular.  But what really stands out is the final four or five years on the chart.  Auckland and Hamilton are seeing less new building (relative to population growth) than they used to, while activity in Tauranga has held up at around the average for the previous 15 years.

What about Wellington and Christchurch?  The population of greater Wellington (Wellington, Upper and Lower Hutt, Porirua, and Kapiti) has grown by only 17 per cent over this period.  I never voluntarily defend Wellington local authorities.  Perhaps –  quite probably –  in a climate of heavy land and building regulation it is easier for building to keep pace with more modest population growth.  But for the full period, here is the number of new dwelling consents per person increase in population.

Auckland 0.32
Hamilton 0.38
Tauranga 0.45
Wellington 0.49

Greater Wellington has actually seen more building, relative to population growth, than even the least bad of those northern cities.

Christchurch is a story complicated by the loss of houses as a result of the earthquakes.  One would simply expect to see a lot more permits in that region following the earthquakes even if the population changed little.  Greater Christchurch encompasses three TLAs –  Christchurch city, Waimakariri and Selwyn.  The Selwyn council has a reputation for having facilitated growth –  including the otherwise improbable meteoric post-quake growth of Rolleston.

If we split the sample and look at the years up to June 2010 (ie before the first earthquake), the number of new dwelling permits in greater Christchurch relative to the (quite strong) growth in population had been higher than in Auckland, Hamilton or Tauranga over the same period –  but still a little behind Wellington.

The loss of existing houses muddies the post-2010 data.  If we take the full period (1996 to 2015) in the table above greater Christchurch comes out at 0.66 –  far above the other large cities.  But, of course, greater Christchurch lost lots of existing houses –  so the high numbers tell one nothing about supply/demand balances, or responsiveness of councils.

But one interesting angle is to look just at Selwyn.  Queenstown apart, Selwyn has had the highest population growth rate of any TLA in New Zealand over the last 20 years (107 per cent).  And Selwyn had few houses destroyed in the quakes. This is the chart of new dwelling consents per person increase in population in Selwyn.


It is certainly a better experience than Auckland’s, but nothing to write home about.  In fact, in the sub-period prior to the quakes, the rate of new dwelling consents per increase in population was a little lower in Selwyn than it had been in Christchurch city itself. Of course, an open question is to what extent people moved to Selwyn because of a responsive regulatory system –  in turn pushed to its limits –  and to what extent because the land itself was more stable, and the new motorway made places like Rolleston very easy to get to and from.

And what if we add fast-growing Queenstown into the mix?

New dwelling consents per person increase in population (June years 1997 to 2015)

Auckland 0.32
Hamilton 0.38
Tauranga 0.45
Wellington (greater) 0.49
Christchurch (greater) to 2010 0.50
Queenstown 0.53

Of course, much of Queenstown’s construction is likely to be holiday homes, but nonetheless the contrast –  in a town with very rapid population growth – with Auckland (and even Hamilton) is striking.

As a final caution, do note that the sub-national population numbers for the period since the 2013 census are estimates, themselves derived from national population estimates.  In a couple of years’ time, after the next census, some of the recent population data could look quite different, affecting the interpretation of some of these recent construction numbers.  But in most cases, the patterns were well in place before even the 2013 census.


18 thoughts on “New dwellings and population growth

  1. Very interesting posts, both today and tomorrow. As someone building a new house right now in Selwyn, my experience is that Selwyn prices are now stagnant, if not about to drop. I’d suggest that whatever the ratio of house building to increase of population is, the ratio in Selwyn is about what is needed to keep house prices level, save for external factors influencing (I suspect Aucklanders moving to Tauranga is one such external factor influencinc BoP prices).


  2. I’m pretty certain that I’ve seen a few papers which looked at quantifying the elasticity of supply of residential dwellings in different cities, as a way of quantifying the effect of restrictive (or accommodative) urban land use policies. The rationale from the authors was that since we can’t identify and categorise the impact of many thousands of different planning restrictions, we can observe its effect in the market.

    That’s the issue that you’re getting at here: how the supply of new dwellings in different regions has responded to changes in demand (taking into account inter-regional differences in demand for lots of valid reasons), and whether we can consider that as ‘adequate’ by some measure.

    I think Ed Glaeser did some work on this, maybe ten years ago. Try Urban Growth and Housing Supply, by Glaeser, Gyourko and Saks (2005).

    More broadly – the only other bit I would add here is that I would also expect demand for dwelling size (and perhaps even number of dwellings) to be a normal good, and rise with income over time. So if my income doubles or is twice my parents income, I’d like maybe double the space (or maybe a space twice as nice) to live in. So while lots of planners might say that we should be living in denser and better planned communities than we did in the past, that’s not what my preferences are saying – they’re saying the opposite (subject to my value of time spent commuting). This may be an underappreciated factor in big price impacts in places like Auckland or Sydney.


  3. Thanks. The OECD has also done some good cross-country empirical work on supply responsiveness.

    Space is certainly a normal good. I look around my own suburb and see the number of houses (including all three I’ve lived in) that have had additional rooms added to them over the years – not “new dwelling consents”, but certainly larger better houses.


  4. Michael
    Wellington looks to be the winner.
    Coincidentally I was talking to a highly successful Wellington property investor/developer about Wellington’s addition of residential dwellings and he noted that in the last few years 15,000 (I think he may have meant residents not residencies) had been added to the Te Aro Flat area from zero, with another 15,000 on their way. His key point was that these dwellings had a total cost that was very low relative to edge of town dwellings because much less incremental infrastructure is necessary.
    Ive no idea of the total costs associated with say Wellington’s high density high growth area in Te Aro Flat and comparable areas of Auckland. Nor of how many people per hectare the comparable areas house, but it is hard not to believe that Wellington’s success (measured by new dwellings per new person) probably does owe a lot to the low cost densification of Te Aro Flat.
    As the developer also noted, and I 100% agree, such areas have to be attractive and in this regard Wellington City Council has done a great job encouraging and supporting inner-city cultural activities. This area of the city is its most culturally vibrant and I’d bet that encouraging such activities is a lot cheaper than building roads and laying pipes and cables to new edge of town suburbs, which then have to linked to jobs, schools and activities by cars rather than buses, walking and biking
    Tim Brown


    • I think for a certain demographic all those city apartments have been a really good development.

      Then again, Wgtn house prices aren’t cheap by international standards, and Wellington hasn’t had to grapple with the challenges of rapid population growth in the way that the northern cities have. Whether the housing outcomes would be much different if it did is probably something we’ll never know.


      • Relative to population growth, perhaps the new subdivisional housing that has gone in in Churton Park, Woodchester, Woodman Drive, and now Aotea, has been not too bad? I do not believe that intensification is ever the cause of “better balanced housing supply”.

        Houston is infamous for the ease of its sprawl, but it is also doing the most intensification and apartment development of any US city. You need the “option values” to keep the price expectations of site vendors everywhere, more honest – and when the prices are as anchored as Houston’s are, then investors look at actually building living space to get an honest return. The constrained, land-value-Ponzi cities end up with less development of all types, because as Arthur Grimes and Andrew Aitken concluded in a 2010 paper that should have made them world famous, “the profit potential from redevelopment is impounded in rising site values”.


      • I would argue that it makes more sense for Len Brown to spend $2billion for intercity rail from Britomart to Mt Eden if Mt Eden and Mt Roskill was rezoned to 18 levels to 50 levels of highrise apartments rather than the current Unitary Plan that maintains the status quo of low rise and low density in most of central Auckland due to its 57 sacred volcanic mounts. $2billion to Mt Eden when there isn’t even a decent shopping mall.


  5. Interesting about Te Aro and kinda demonstrates what is needed in Auckland with increasing density. Of course they need lots more rail as well to make it work. An hour and a half the other day from Manakau to Greenlane. The overhead costs for business up there must be horrendous.

    One of the things about Tauranga is that there was never a shortage of sections so if you wanted a new house it was just a matter of where. Tightened up a bit now though.


    • No, there is no city anywhere in the world where “increased density” is the cause of improved affordability. See my comment just above. It is always the ability to add more land, that counts. Rail does not do this, because the land added is in long narrow ribbons. On the other hand, automobility adds potential land supply on a mathematical basis of something like “distance ex the urban fringe able to be traveled in 30 minutes, times PI, SQUARED”.

      If density caused affordability, Hong Kong’s median multiple would not be 17 for its 26,000 people per square km, and Atlanta’s would not be 2.9 for its 700 people per square km. Land values are in fact exponentially elastic to population density.

      Every city in the world that has been attempting to increase density as a means of increasing housing supply in the face of escalating prices, has never been able to do any better than apartments at double the price that the typical market median suburban family homes used to be. NO option ends up remotely affordable. In contrast, sprawling freely keeps the land values so low that affordability is correlated with more spacious living.


      • Auckland already sprawls over 129km from Leigh north to Pukekohe south. Houston with 6.2 million people from the Woodlands north to Texas city south is only 118km.


  6. The only time I would use the consents issued per extra person metric is to look at impacts on occupancy rates. On average we have a comfortable 2.4 or so people per dwelling across the country. That rate of 2.4 translates to about .41 on your graphs. Demographic factors might shift that rate around a little but that is the general zone. Basically it is logical for the graph to trend towards 0.4 – 0.5.

    Auckland has a less comfortable 3.0-ish which is incredibly unevenly distributed so that is rises to 4 in Otara-Manukau while remaining comfortable in some other parts of the city, Since 2011 Auckland has not built enough dwellings to make a dent in its over-occupancy rate. And its useful to know that whenever the Smith & Brown show proclaim “more buildings since…”

    While consenting rates don’t mean much in the usual Stuff article context, if you know what you are looking for, they still have some stories to tell. You can’t get the kind of consenting rates that Waimakariri and Selwyn have just on liberal planning alone – in fact I wouldn’t class Waimakariri as even slightly liberal in its planning approach. It is a whole-of-system metric. Since high growth rates means putting in some infrastructure in advance of growth that means a tolerance from the whole community to the higher rates that that will inevitably impose.

    When you get population growth rates as low as Wellington councils can get away with being more reactive and any rates impacts tend to be in the order of noise. And we forget the majority of the country really doesn’t struggle with accommodating new population. The system, flawed though it is, works as long as growth doesn’t get over that 1.75%p.a. – 2%p.a. level. Then all the fragilities and weaknesses of a clunky development system get exposed.


      • What I meant to say about Wellington is that it gives us a glimpse of what conditions may be like 30 years from now in today’s fast-growing areas. If the demographers are right (and aren’t they always) our natural increase will have turned negative in 20-30 years. Our national population then will depend entirely on our immigration policy and the settlement desires of the 24m Australians and Kiwis abroad entitled to live here anytime they like.

        Uncder those conditions we will have way more choices about quality rather than quantity of our built environments.


      • Maybe, altho I’d argue that we have already for some time been at the point where population growth is largely explained by immigration policy. Given the trend net outflow of NZers and the decreasing natural increase, with even modest non-citizen immigration we’d already have trend flat national population.


      • When it comes to urban land and housing prices, a vital point is how much “competition for more space” there is at the margins of each income cohort. There has to be an explanation for why Liverpool had a shrinking population for decades and yet its land prices remained so high that even its small, poor quality median house was far more expensive than benchmark US cities large, high quality median house.

        The answer is almost certainly that Liverpool’s population density in 1950 was so high, that even when it had halved 50 years later, the space per household was still low enough to be provoking the “competition for more space” effect among the population. It is an absence of this effect, that explains affordable cities with freedom of choice all being low density. Planners can claim as much as they like, that they have “surveyed” people’s preferences and therefore they “know” just how much of each housing unit type needs to be planned for to keep prices affordable. Actually prices are “revealed preferences” and they are what should be used as the metric to determine whether people are “getting the space they want”.

        I say that the typical new-suburban density seen in 1960’s NZ, Canada, France and Germany, are about what true free market choice will deliver along with systemic affordability. The absurdly low densities of many US cities suburbs are a distortion to the free market in the other direction – but even then this does not cause an affordability problem. The price of urban land is exponentially elastic to population density, and this works to depress land values under mandated low density, just as much as it works to boost land values under permitted higher density.

        The affordable urban area for decades in NZ, Canada, France and Germany, tended to have a population density freely trending down towards 1500 – 2000 people per square km. The overall density could be higher in the case of old-world cities with dense urban cores. As long as these cores were leaking population to lower density suburbs, they did not suffer from the high land values associated with density. But Liverpool under the Town and Country Planning Act, still had a density of around 4000 people per square km in 2000; and its addition of “lower” density suburbs was heavily constrained relative to all the first world countries that achieved affordability during the post-WW2 decades.


      • What I am saying is that Wellington, even with its low population growth, only has relatively “healthy” housing supply and prices not as severely unaffordable, because it had decades of typical suburban development as its main means of supply, and even now, there is a “healthy enough” supply of suburban family homes going in on the northern fringes. Donald is right; if growth passed the magic trigger point, Wellington’s planning regime would quickly become exposed by rampant inflation. In fact this may already be happening now. Speculation can be an unpredictable input too.


    • Auckland airport handles 17 million arrivals and departures every 12 months. The traffic and accommodation pressures is more likely due to travellers international and from the regions coming into Auckland than actual population growth from migration of 14k gross migrant arrivals a year or net 68k if we include 120k international students and 30k returning kiwis.


  7. Woodhouse, immigration minister on Q & A yesterday again points to an inability to lower the 50k policy target for residences. He also indicated that kiwis choosing foreign spouses as a factor. This is the second time that factor has been commented in different interviews. This is the likely reason that we have so many work visas issued for general labour. If we lower the policy target we will end up with a 100% unskilled migrant workforce as kiwis love relationship needs comes before the skilled migrant.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s