Friday 6 December 2013

Crowded house

Christchurch districts that experienced bigger increases in house prices from 2008 through 2013, the dates of easily available QV data, also experienced larger increases in the proportion of census households reporting two or more families living in the same household. This isn't particularly surprising. Where housing supply is inflexible due to Council regulatory constraints, increases in the number of families, whether due to kids reaching adulthood, immigration, or divorce have to be accommodated somehow. That somehow is multiple families living in the same household.

On Friday last week, the Christchurch Press called the University looking for a few folks to provide them commentary on the Census, released this Tuesday. Stephen and I put together the following as op-ed. The Press decided to turn it into a news story instead; the story appeared here. But our original piece is below.
Census 2013 focus: The hidden cost of housing regulation
Eric Crampton & Stephen Hickson

More Kiwi families are having to double-up as housing supply has failed to respond to demand pressures. The Census tells us how many families live in each household. In 2006, fewer than 40,000 households, or about 2.8 percent of all households, included two or more families living together. In 2013, that figure rose to just over 51,000 households, or about 3.4 percent of the total. Had the proportion of multifamily households stayed at the 2006 level, 9,100 fewer households would have had two or more families bunking together.
Economists call this an increase in the intensity of housing use. It’s one of the ways that people can respond to increases in housing costs. When housing gets more expensive, people buy less of it. One way of buying less housing is to share a house with one or two other families. The pattern of increased housing intensity suggests that cost pressures are behind it.
We paired district-level housing intensity increases between the two Censuses with Quotable Value data on residential house values from November 2008, the earliest month in their freely available data series March 2009 and March 2013, the month of the census.* The increase in two-or-more families living in the same household is strongly related to house price increases within each district (for the statistically minded, the correlation coefficient is 0.66). This is just a first cut as the Census has only just been released. But it is informative.

Selected Territorial Authority AreasNumber of households with 2 or more families, 2006 Number of households with 2 or more families, 2013 Percentage increase in multifamily households Percentage increase in QV property value, November 2008 – March 2013
Far North 546 639 17% -14.5%
Whangarei 648 672 3.7% -9.9%
Auckland Area 19977 27042 35.4% 22%
Waikato 612 684 11.8% 2.5%
Rotorua 744 735 -1.2% -2.8%
Napier 426 459 7.7% 0%
Wellington City 1215 1659 36.5% 5.4%
Nelson 216 303 40.3% 9.1%
Grey 51 39 -23.5% -10.1%
Waimakariri 225 420 86.7% 26.4%
Christchurch 2295 3132 36.5% 19.7%
Selwyn 204 369 80.9% 30.5%
Dunedin 519 621 19.7% 10.2%
Southland 75 81 8% -5.4%

We see this as one of the hidden costs of New Zealand’s very rigid approach to town planning. City Councils make it hard not only to expand housing supply out into the suburbs but also to increase density within town limits. When it’s hard for housing supply to respond as population increases, prices of existing houses increase. This brings supply and demand back into line, but by forcing families to make some pretty costly decisions by bunking together. Until Councils start taking housing supply seriously, expect the situation to worsen.
I emphasized to the reporter that demand side measures to try to hit housing prices really don't help. If we have fewer houses than families that want to be in houses, then we're going to have multiple families in a household regardless of any measures we might put in affecting house prices. Those measures can affect the price at which the market clears but don't really change which families wind up in which houses. Worse, where some measures targeted at demand can reduce building, they can make things worse.

* Looking back at it, Stephen had pulled QV data running March 09 through March 13. The latter is the month of the census; the former is the first March in the QV data. While November 2008 would be earlier, we then risk getting seasonal / month effects mixed in with things.


  1. If you take out Waimakariri, Selwyn - Places where housing broadly survived the quake but close to the large numbers of people that need housing. Also take out Christchurch, demand shock faster than capacity to increase housing boosting prices, do you still have a significant result? If so, is it just because of the existence of Auckland?

  2. Where I grew up, the county made this illegal. I forget the details, but you couldn't have more than X number of unrelated people in a single family house. It hasn't stopped people building illegal granny flats but has probably reduced their number.

  3. I remember those sorts of rules in Fairfax / Vienna / Falls Church too. They were pretty clearly designed to screw over Mexican migrant workers.

  4. There's no way that those are driving things. First off, this is very much a rough cut. I didn't weight things by population. Which means that while Auckland doesn't get anywhere near the attention it deserves, it also can't be driving the results.

    Lemmie check. Redoing, getting a correlation coefficient of 0.67 with all districts in. Dropping Auckland city: 0.68. Dropping Auckland and Christchurch and Waimak/Selwyn: 0.59.

    A lot of the action is in big reductions in overcrowding in places with declining housing costs - people moving away from small towns for the big smoke leaving more room in the small towns for those staying there.

    Again, this is a rough first cut. We got the data Tuesday morning and were aiming for a Tuesday afternoon Press deadline. But it's pretty easy to play around with yourself if you like. Get "Table 3: Household Composition by Territorial Authority Area", then match it to the QV website by district. We used March 2009 through March 2013 QV data to get a census-month end-point and to have a same-month earliest start point.

  5. What do you think is happening in Northland?

  6. There are dozens of districts; we just put up a few. Northland is one of the ones that's an outlier; so's Wellington. I've not been more than a half-hour north of Auckland, and haven't dug into any of the related data, so I'd be reluctant to start guessing.

  7. I am not sure about anything, i just know my prices for renting are going down right now, I do good deals but I have to fight. Christchurch is for tough people. They closed the roads down did you see that .

  8. Hopefully impossible to do that particular correlation on the data.

  9. Maybe it's all post hoc prompter hoc, but there seemed to be significant county-level tightening of rules and enforcement with increased Mexican migrant labour to the DC metro area. Neighbours see too many working-class cars and trucks outside a property, they complain....