Skip to main content

Pulling a Rabbit Out of the Census Hat..

John Polkinghorne
Rabit out of a hat

Two confessions up front. One, the census is my bread and butter. I use census data all the time, in almost all my research for retailers and landlords.   Two, I’m a terrible liar.

Because of 1), I’ve got an incentive to say how great and useful the 2018 census results will be. But because of 2), I’ll only say it if I believe it. A lot of people have criticised the 2018 census, and a lot of the criticism is justified.

Even so, I’m confident that the 2018 census results can be trusted by the retail industry.

New stores are expensive, and retailers shouldn’t commit to them unless they’re confident in their market and the local demographics. Ideally, retailers should be basing those decisions on census data.

And after dissecting every word and number Stats NZ have released about the census, and other commentators’ responses, I trust the results I’m going to get. I’m not happy that I’ll have to wait until late September for them, but our industry will have what it needs. The stats that retailers and landlords should care about the most, like population size, incomes, age and family groups, will all be accurate.

Against all odds, Stats NZ have pulled a rabbit out of the census hat. Or at least, they’ll start pulling rabbits out from September.    

If you’ll just trust me on that, then you might not need to keep reading. I’ll get a bit technical below.

There were definitely mistakes made in the 2018 census. Stats NZ went too far down the ‘online’ route, and didn’t provide enough backup with paper forms and people who needed help or prodding to fill out their forms.

From a PR perspective, Stats NZ’s response has not been flash. They’ve had to delay the release of results twice, and haven’t been proactive in explaining technical issues.

By itself, the census did not do a good enough job of collecting the data it was supposed to. Stats NZ have acknowledged that, in a roundabout way.


Fortunately, Stats NZ can combine the 2018 census data with other government data – from the Ministry of Health, the IRD, and other sources including their own 2013 census. Stats NZ can do all this for each individual person, following very strict confidentiality and privacy protocols.

It’s a painstaking process, but once they’re finished, they will have a very high quality dataset. And this is what will be published as the “2018 census results”, starting in late September. For some topics, it won’t be as complete – but for the topics that retailers need to focus on, it will be as good as, or better than previous censuses.

Even though we’re all meant to fill out a census form, coverage is never 100%. For the 2013 census, Stats NZ estimate that they counted 97.6% of the people they should have (New Zealand residents who were in the country on census night). For the 2018 census dataset, they’ll actually do even better – counting 98.8% of the people they should have.

That doesn’t mean that all those people filled out individual census forms. That’s a different number, called the “response rate”. In 2013, it was 92.9%. In 2018, it’s likely to be a touch under 90%. The response rates are lower than the census counts above, because Stats NZ will count people if they’re satisfied they exist, even if they don’t fill out an individual form. As an example: a household form is filled out saying the household has five people, but only four people fill out their individual forms. Stats NZ will count all five people, even though it only has individual forms for four.

So, the response rates are lower this time around, meaning more gaps in the data.

Stats NZ can’t fill all the gaps this way. For example, Stats NZ have acknowledged that they will struggle to provide info on iwi affiliation for Maori. This is sad, and there may be other social or cultural info that isn’t well covered, but the key data for the retail industry will not be affected.

That key data includes population, which I’ve discussed above. That should still be very high quality, in terms of the total number and the age group breakdown. The other key data is income. This is an area where the 2018 census could give even better quality than in the past. Any gaps will be filled with IRD data, and they should have even more accurate info on incomes than Stats NZ do.

But, despite a long delay (which I’m not at all happy about), it looks like the results we’ll be getting from September this year can be relied on by retailers, landlords and others in our industry. Until then, I’ll keep a very close eye on each update and technical paper that Stats NZ release, so that you don’t have to. And I’ll let you know if my opinion changes. After all, I’m a person who put together a 125-year database of census population data… in my spare time.

In the Press

Local Media highlights from the past week...

KiwiBuild houses mostly snapped up by young buyers

KiwiBuild houses are mostly being sold to people younger than 35, figures released to Stuff show. This generation of buyers are the most likely to be locked out of the current market.


Goodbye 'McMansions', hello compact living: What life in New Zealand could be like in 2039

In 20 years where will we live? And what will we be living in? Brittney Deguara looks at what life could be like. Life as we know it will change.


Sold after seven years: NZ's most expensive home getting new owner

New Zealand's most expensive house for sale, valued last year at $24 million, has been sold, according to an Auckland real estate agent.

(Source: NZ Herald)

Tenancy Tribunal awards Hamilton tenant $2470 after landlord kept breaking in and damaging home

A tenant has won more than $2000 in compensation after a property owner repeatedly broke in and damaged the home she was living in.

(Source: NZ Herald)

Looking for a strategic design and property partner?
Let's talk

Looking to buy, sell
or lease a property?
Talk to RCGR