Once you adjust population growth for the average size of a household, the change in the dwelling stock in both Selwyn and Christchurch over the past five years has been pretty much in line with what has been required. This balance of supply and demand helps to explain the only modest rises in property values in these areas since 2015. By contrast, Queenstown’s supply has lagged behind demand, and so it’s no surprise that it’s seen bigger value rises since 2015. At the same time, the COVID-induced falls in prices in Queenstown this year are now easing, probably also reflecting excess demand over supply.

Two weeks ago on the NZ Property Market Podcast (27th Oct.), we highlighted the latest subnational population estimates from Stats NZ, discussing how areas such as Queenstown, Selwyn, Tauranga, and Central Otago had seen the strongest growth in the year to June 2020. But how does population growth (and need for housing) compare to our figures on the actual change in the dwelling stock around different parts of the country?

Total population change for selected areas 2015-20 (Source: Stats NZ)
Total population change for selected areas 2015-20 (Source: Stats NZ)

If we start by looking back over a longer, five-year horizon, the areas noted above have generally held their places, with Queenstown (42%), Selwyn (31%), and Central Otago (22%) all recording population increases of more than 20% from June 2015 to June 2020 – double the national average of 10%. Of the main centres, Christchurch, Wellington, and Dunedin have seen below-average increases in population since 2015 (see the first chart).

Total dwelling stock change for selected areas 2015-20 (Source: CoreLogic)
Total dwelling stock change for selected areas 2015-20 (Source: CoreLogic)

Looking at the dwelling stock figures for the same areas, the rankings are generally similar, although the huge growth in Selwyn’s housing base (34%) really stands out (see the second chart). Since 2015, the national stock of dwellings has risen by 6%, with Wellington and Dunedin both below that mark. Although not shown here, it’s interesting that areas such as Rotorua, Invercargill, Gisborne, and Whanganui have all seen their stock of dwellings rise by only 2% or less since 2015.

Estimate of total housing shortfall as % of stock  2015-20 (Sources: Stats NZ, CoreLogic)
Estimate of total housing shortfall as % of stock 2015-20 (Sources: Stats NZ, CoreLogic)

Now if we use each area’s occupancy rate (people per household) to translate population growth into actual demand/need for housing, and compare that to the change in stock, we can see some interesting patterns around the country. For example, Selwyn’s dwelling stock has actually grown by more than its demand for housing since 2015, by a margin of around 1%, or roughly 300 dwellings (see the third chart). Christchurch has also been relatively balanced (1% fewer dwellings than required), but Queenstown has been at the other end of the spectrum.

At face value, then, it’s no surprise that Queenstown saw average property values rise by about 65% from mid-2015 to mid-2020, but Christchurch and Selwyn only saw increases of about 10%. However, it’s also worth noting that there’s not always a direct correlation between an adequate supply of housing and low price growth, with other areas of the country that have seen their stock rise steadily also experiencing large increases in values. This highlights how other factors matter too, such as regional economic performance and the starting point for affordability.

Dwelling consents versus change in stock (Sources: Stats NZ, CoreLogic)
Dwelling consents versus change in stock (Sources: Stats NZ, CoreLogic)

Finally, it’s also interesting to compare new dwelling consents to the change in housing stock a year later (to account for the building process). As the fourth chart shows, Auckland for example had about 55,000 consents over 2015-19, but the stock only changed over 2016-20 by about 34,000. The difference reflects factors such as demolitions (to allow infill housing) and the fact that some consents don’t actually get built.