If planning constrains housing supply, how does it do it?
I want to know the counterfactual people have in mind
I’ve long puzzled over the exact mechanism by which planning regulations are supposed to constrain the rate of new housing supply. Sure, they constrain the locations at which different types and densities of uses can occur. That’s their purpose.
But density (dwellings per land area) and the rate of supply (new dwellings per period across all sites in a region) are very different.
It seems logical to me that landowners maximise their economic returns from choosing both
a density of development that maximises the residual (i.e. their revenue minus the cost of development), and
a rate of sales (and hence development) per period of time.
It is not at all clear that if more dwellings can be built on one site that this changes the optimal rate of sales per period for that site.
It is also the case that only landowners can choose to make planning applications, and that a huge majority of approvals are for projects that exceed coded density limits. We have a property market after all.
What I am puzzled about is that when someone claims that planning reduces housing supply, what sort of counterfactual pattern of supply do they have in mind?
I’ve put out a poll on Twitter asking this exact question, alongside some potential charts of what a more supply-friendly planning system might deliver. Please respond to the poll and share the tweet.
The three potential counterfactuals are shown below.
For me, the only one that possibly makes sense is C), that there is some maximum limit. But this then leads to the question of why private market actors don’t accumulate more approvals in downtimes to account for the expected limit. Why not invest in a buffer?
The first two choices make little sense in my view. If the planning system can approve 20,000 dwellings a quarter in 2015 based on the applications made, I don’t see why that wasn’t possible in 2013, or 2019, if similar applications were made.
I feel like thinking about counterfactuals is really important because many assumptions are necessary when looking at data on planning rules and approvals and making conclusions about supply. If planning rules work, they encourage development in the areas planned for it, and not in other areas. So you will usually see that rezoning of a high-value area for higher density leads to more development in that area compared to nearby areas.
A recent paper by Ryan Greenaway-McGrevy and Peter C.B. Phillips from the University of Auckland analysing the Auckland Unitary Plan is a good example. The charts below from that paper show a huge effect of rezoning on building approvals in their sample area (about 77% of the urban core of Auckland).
This looks great. But again, we have density, location, and rate of supply concepts to grapple with. Is this mostly location substitution? Maybe. Will these approvals result in additional total dwelling supply being constructed? Maybe. But it is really hard to eyeball any type of break in the trend of city-wide housing supply around 2016 when the new plan came into effect. The counterfactual here is important.
To be clear, many planning systems are unnecessarily cumbersome and in practice don’t achieve their intent. I have no view on whether Auckland’s new plan is good or bad from a city design perspective. The new mix of locations and densities could well be superior and represent the desires of locals. Great.
What I do have is a view about the effect of the change in zoning on overall pricing in the housing market. Private property markets are not in the business of minimising their economic returns by flooding the market, so I expect essentially no noticeable price effects. If this occurs, then we need to really take a closer look at the assumed counterfactuals of the analysis being used.
UPDATE [May 2022]:
That Auckland paper I think oversteps the mark on claiming that these additional approvals equate to supply. Especially this part from pp17-18.
This figure implies that consents issued per year have approximately doubled as a result of the policy, which averaged at 4,213 dwellings between 2010 and 2015. The additional 19,725 consents correponds to an increase of 3.74% in the cityís extant housing stock. Statistics New Zealand estimates that there were 530,300 dwellings in Auckland by the end of 2016.10 Because completions of consented dwellings range between 95% and 99% (outside of recessionary periods), the cumulative completed construction enabled by the policy implies is between 3.53% (= 19,725/530,300 0.95) and 3.68% (= 19,725/530,300 0.99) of the dwelling stock of Auckland. Consents in upzoned areas continue to trend upwards. So the full impact of the policy will likely not be known for several more years.
Applying the same approach to the sample based on Area Unit geographies with control variables (see Section 5.5.2 above) implies an additional 25,631 consents.
That number seems implausible, and not only because of the unchanged trend in the above image. Auckland simply is not that big, and the change in the dwelling stock is not that fast. In the four years prior to 2016, Auckland had only approved 32,000 dwellings, or about 8,000 dwellings a year. Fewer than that were built.
For there to be 26,000 dwellings in four years above the rising trend implies a massive construction program that seems implausibly high. The below chart shows a counterfactual that is 26,000 dwellings above trend. It looks nothing like the above chart.
All I see in the data are business cycles, as I think the below chart captures well. Given the recent interest rate moves and declining prices in Auckland, I suspect that many of these approvals will never get built. It is also likely that without the buyer demand, few property owners will seek approvals in the coming years and when I replicate this chart in four years’ time we will see another cycle in progress.
Thanks for reading Fresh Economic Thinking! Subscribe for free to receive new posts and support my work.