This is the third installment in an ongoing series on the politics and economics of zoning reform. Last week’s post took a look at the outcomes from the Unitary Plan process, which included a mix of political decision-making and technical assessment. The data in this post raised a few interesting questions, including: Why did councillors take a relatively conservative approach to the notified Unitary Plan?

In his fantastic book Zoning Rules! The Economics of Land Use Regulation, which I reviewed at the start of the year, William Fischel argues that zoning decisions generally respond to demands from politically active “homevoters” who agitate to minimise risks to their own property values.

While this doesn’t necessarily hold true in all cases, it’s a useful heuristic for understanding councils’ decision-making. And it suggests that when seeking to explain zoning decisions we should begin by asking: Who submitted on the plan?

After notifying the Unitary Plan to the public in September 2013, Auckland Council received submissions from around 9,300 individuals or organisations, who made a total of 93,600 unique requests. The submissions are all available online if you want to read them in detail.

I didn’t want to bother reading them in detail, so I started with some simple statistical analysis. Approximately 5,900 submitters provided an address, which allowed Auckland Council to match them to one of the city’s 21 local boards. This allowed me to identify which areas of Auckland had more or less submissions. Here’s a map showing submissions per capita.

UP submissions per 1000 residents map

Interesting map. We can immediately see three things:

  • People in rural areas – especially Rodney – submitted on the plan at higher rates than people in urban areas.
  • Two urban local boards stand out as having high submissions per capita – Orakei and Devonport-Takapuna. They are both relatively well-off coastal areas.
  • Submission rates in the west and south were much, much lower – as shown in the yellow band through the city.

All in all, people who lived in Rodney or Orakei were over ten times as likely to put in a submission on the Unitary Plan than people who lived in Henderson-Massey or Mangere-Otahuhu.

This is an extraordinarily large amount of variation, and it’s not matched by other indices of civic participation. For instance, voter turnout wasn’t ten times as high in Orakei as it was in Henderson or Mangere. So what factors are driving the differences?

Ideally, we’d be able to take a look at the decisions made by individuals – for instance, by surveying people on their decisions about whether or not to submit. But that’s a bit over the top for a blog post, so I’m going to take a quick look at some demographic factors that might underpin different submissions rates at a local board level.

More specifically, I want to investigate three hypotheses that are (loosely) derived from Fishel’s “homevoter hypothesis”. They relate to the time and money that people have to get involved in the process, and people’s sense of “ownership” over their neighbourhood:

  • Hypothesis 1: Areas with higher median incomes are more likely to submit at higher rates
  • Hypothesis 2: Areas with a higher share of home-owning households are more likely to submit at higher rates
  • Hypothesis 3: Areas with a higher share of people over the age of 65 are more likely to submit at higher rates

All demographic variables were measured at the local board level using data from the 2013 Census. (NB: These weren’t the only factors I considered, but they seemed to be the most relevant ones.)

The following table summarises results from a simple OLS regression, which measures the correlations between multiple explanatory variables (the local board demographic variables) and a single outcome variable (Unitary Plan submissions per 1000 residents). The statisticians among our audience will find it pretty intuitive; for the rest, here is an explanation of the findings:

  • At a local board level, a higher median personal income was associated with a higher rate of submissions on the Unitary Plan. This correlation was highly statistically significant (1% level). On average, every $1000 increase in median personal income was associated with another 0.51 UP submissions per 1000 residents.
  • A higher share of residents aged 65 and over was associated with a higher submission rates. This correlation was also statistically significant (5% level). On average, every 1% increase in the share of seniors was associated with another 0.47 UP submissions per 1000 residents.
  • After controlling for incomes and age, there was no statistically significant relationship between the share of households in rental accommodation and submission rates.
  • These factors “explain” about 56% of the variations in submission rates between local boards. In other words, income and age are quite important to overall outcomes.
Outcome variable:UP submissions per 1000 residents
Explanatory variablesModel coefficients
Median personal income ($000s)0.507***
Percent 65 years and over (%)46.973**
Percent renting (%)7.693
Model statistics
Adjusted R20.557
Residual Std. Error2.455 (df = 17)
F Statistic9.365*** (df = 3; 17)
Note:*p<0.1; **p<0.05; ***p<0.01

In short, areas with more wealthy people and more retired people tended to submit on the Unitary Plan at considerably higher rates. While there is an idiosyncratic story lurking behind every submission, demographic factors seem to have played a crucial role in shaping the submissions that Auckland Council received on the Unitary Plan.

This raises several questions:

First, how much stock should policymakers place on submissions as opposed to inputs gathered from other sources, like demographically-representative surveys? In asking this question I am not dismissing submissions entirely – the people who submit are also more likely to have relevant knowledge about the issue at hand. But is it plausible to think that people in wealthier and older areas are ten times more knowledgeable than people in poorer and younger areas?

Second, how should planning processes account for the preferences and needs of people who are invisible in the consultation process? If you care about the substance of democracy as well as the form, as I do, this is an important question. There are in fact many things that can be done to divine people’s underlying preferences for various things in cities. Perhaps we should invest more in this sort of research.

Third, to what extent did submissions on the Unitary Plan affect the process at different stages? For instance, were the areas that Auckland Council chose to “downzone” between the draft and notified versions of the Unitary Plan concentrated in local boards with higher submission rates? And what about the outcomes from the final Unitary Plan recommended by the hearings panel? These are obviously hard questions to answer in full without an extremely in-depth analysis of the 93,600 unique requests that people made. But there may be some simple insights we can get from a higher-level analysis; stay tuned…

What do you make of the data on Unitary Plan submission rates?

Share this


  1. Fairly logical, those with the bigger vested interests, I.e. they own property etc, are more likely to respond than those who don’t have financial skin in the game. That applies anywhere in life.

    1. Indeed.

      Nonetheless, it’s always good to look at the data to make sure one’s hypotheses are not bollocks, statistically speaking. One might argue that the traffic engineering and planning professions should do a little more statistical analyses of this nature?

      And of course, as Peter notes, this evidence simply motivates another, more challenging, question: To what degree does this income/demographic bias in democratic processes adversely affect decision-making?

    2. I would expect that the *individuals* who submitted would be disproportionately likely to own property. However, if we look at variations in submission rates between different parts of the city, home ownership didn’t play a significant explanatory role.

      As discussed in the post, once I controlled for age and incomes at the local board level, home ownership rates bore no statistically significant relationship to submission rates within local boards. (This wasn’t a multicollinearity issue either – I checked.)

      This suggests that demographics – age and income – shape consultations to a significant degree.

      1. “After controlling for incomes and age, there was no statistically significant relationship between the share of households in rental accommodation and submission rates” Presumably you have looked for correlation between %renting and age and income. I would guess you might show a strong correlation there. ie a good model of ownership would be to regress age and income. If that exists then your statement that “after controlling for age and income there is no relationship” becomes a simple truism because you have specifically excluded any relationship from the maths.

        1. Yes, I took that into account. I’m not the best econometrician in the world, but I’m aware of common pitfalls!

          There is some collinearity between these variables, but it’s not high enough to invalidate the results. 40% of the variation in home ownership rates between local boards is *not* explained by age and income. If home ownership had a strong impact on submission rates, over and above age and income, then that remaining unexplained variation would be sufficient to show a meaningful relationship.

          To put it another way: If I drop either the age or income variables from the model, the coefficient on the “percent renting” variable *still* isn’t statistically significant even at a 10% level. And if I run a bivariate regression of submission rates on percent renting, the R2 is only 0.12, indicating that, even under the most optimistic assumptions, home ownership rates explain only a very small share of variation in submission rates between local board areas.

        2. Sounds like you are on to it. Most people check for correlation between individual predictors and dismiss each as low but forget that in combination it can be high. The model within a model problem. The estimate of the dependent variable will still be ok but economists in particular dont use OLS for that they like to know the contribution of each predictor. Still I am surprised as I would have thought age and income would be great predictors of home ownership.

        3. there might be little variance in home ownership rates due to the large size of the wards in this sample; it might be something that you need more fine-grained spatial units to pick up.

        4. Actually, there was a surprisingly large amount of variation between local boards. The share of households in rental accommodation ranged from 26% (Hibiscus and Bays) to 61% (Waitemata).

          Like I said, because this was a counterintuitive finding I interrogated it more closely!

        5. Yah, indeed.

          I strongly suspect that if you looked at the *individuals* who submitted on the UP you’d find that they were disproportionately likely to own homes. But not all homeowners are equally likely to put in submissions! Submission rates varied between areas to a much greater degree than you’d expect from homeownership rates, and the variations appear to be correlated with income and age structure.

          Or to put it in quantitative terms, if (say) 1% of home owners in Auckland put in submissions on the UP, and 0% of renters did, then we would expect Orakei (70% home ownership rate) to have around 1.5 times as many submissions per 1000 residents than Otara-Papatoetoe (46% home ownership rate).

          In actual fact, Orakei had almost 30 times as many submissions per 1000 residents than Otara-Papatoetoe!

  2. I really like this post.

    The quality of political institutions (and by extension democratic processes) seems to be an important determinant of long run socio-economic outcomes. If people are interested in this sort of thing, then I’d recommend taking a read of Acemoglu and Robinson:

    Very interesting stuff.

    Specific comment: The statistical significance of your coeffiicients on income and age are even more impressive when you consider the relatively small sample size (n=21).

  3. Great post. My take on it is this from an average person as to why I did not make a submission, having participated, pointlessly, in Resource Management consultations in my area a few years ago;

    I know nothing about the inner workings of Auckland Council, I know no one in any council and furthermore have no connection with those who have executive dealings with them, I work for a living, I have virtually no idea how this process works, How do you make a submission?, In what format should it be made and where do you send it because will an incorrect form or layout render my submission invalid?, What can I submit?, What is the most effective way to say what I want them to take note of?, Who in AC if anyone looks these and and how will they use them?, Will I be required to make an appearance at any hearing? Will this cost me?, Do I have the time to research all of the above?

    Last but not least after going to all the effort above is there really any point? Will my submission end up in a shredder by invisible bureaucrats who either don’t care or have a set agenda, I mean we just don’t know. I had the distinct feeling the Unitary Plan outcome was preordained especially with English and Smith threatening to take over the council if it didn’t do what they wanted. Central government have made a mockery of “consultation” and due process in recent years (we didn’t even get to choose if we wanted the Super City for starters) with the one glaring exception being the rather farcical flag referendum where they seemed to bend over backwards to make everything seem really democratic and user friendly. Perhaps these are common reasons why plenty don’t bother and why that is so ideal for those pulling the levers of power!

    1. All very good questions. There seem to be reasonably high barriers (possibly as much social and psychological) to participation. Hence why my analysis here tried to look at factors like age and income that might give people more resources (time, money, familiarity) to overcome those barriers.

      1. In this context I would like to acknowledge the role of advocacy groups in lowering the barriers to making a submission (sometimes to the extent of making it ridiculously easy, e.g. Gen Zero’s online forms).

        Whatever your interest or concern, and whatever your outlook on the issue, there is a dedicated group of volunteers to help one way or another: TB, Gen Zero, Bike Auckland, the AA, Residents Associations, Business Associations, Rate Payers Groups, Forest & Bird etc etc.

        From my limited understanding of working with agencies, submissions are not just treated as a raw count, trading off one suburbs against another, akin to votes (though the numbers matter, and surely the greater the numbers the better if what you want is a picture of what people think). Where demographics are captured by the process, these are noted. And numbers aside, the commentary that submitters include is often valued for what it is: put it this way, if _no-one_ mentions some unforeseen issue consequence (good or bad) about a proposal, then the agency isn’t going to consider it; but if one or more people raise the issue via a submission, the process captures it, the issue will likely be acknowledged, the proposal may be reviewed, and an improved outcome may follow.

        Make no submission: you get what you get, and you can have few complaints. Make a submission: you might get something better. It’s not so difficult, with a little help. It’s taken me longer to write this comment that it has to make some submissions on issues that I am reasonably confident have been taken into consideration.

        1. Perhaps in respect of the ease you find submission writing but I recall that people who made submissions against the chopping down of Pohutakawa’s at the St Lukes Rd, Great North Rd intersection to make way for yet more roading, were ruled invalid because of the use incorrect forms or format. Or at least that was the case until it got media/blog site attention. That example meant the democratic process was being railroaded and confidence in it corroded by bureaucratic cuteness and convenience, which puts people off participating.

        2. That’s a great example of an advocate group helping, albeit after the event. Generally I find out about things to submit on via advocacy groups rather than the agencies directly and they generally point you to exactly the right web page and tell you what reference numbers to use and so on..

    2. The question of whether there is any point depends on how much you expect to achieve. Plans are always about the detail and getting changes to the higher levels objectives is next to impossible. Councils set those and independent bodies usually seem to accept the overall goal. But if you want to make a change to a specific rule you stand a great chance as you are likely to find you are the only person focused on that specific thing. I saw the process as a once in a lifetime chance to intensify my own site, all it really took was a few days of researching the S32 material, gathering site sizes from the council GIS and drawing some histograms to show what alternatives might actually achieve. Oh and the knowledge at the time of submissions to ask for a defenceable area to be rezoned and not just my own site as they never approve pocket zones. These things are always fought on details and Panels and council committees are usually quite generous in accepting advice from people who have done the detailed work even if you are not a planner or lawyer. Is there any point submitting on the major issues and telling them your personal preferences? About as much as signing a petition.

  4. “…how should planning processes account for the preferences and needs of people who are invisible in the consultation process? If you care about the substance of democracy as well as the form, as I do, this is an important question”

    Democracy is about everyone having the opportunity to have their say, and considering the massive amount of publicity surrounding the matter, democracy was fully served.

    It is NOT about trying to guess what people who made no effort to have their say may have wanted if they had given a toss.

    1. I think you’re simplifying things, and in the process missing some interesting and important issues.

      The first thing to note is that NZ is governed via representative democracies, which operate at both the central and local level.

      Moreover, my understanding of RMA and possibly LGA/LGAA legislation is that people making decisions (e.g. commissioners and councillors) must seek to balance the views of current residents with the needs of future generations.

      I also think you miss a key point: Evidence suggests that democratic participation is important to long-run socio-economic outcomes. If you have a process, such as local government, which experiences low levels of participation, then i think we should at least ask the question “why”, as Peter is doing here, rather than writing it off as people choosing not to participate. Unless of you’re happy to see the city go down the plughole.

    2. “Democracy is about everyone having the opportunity to have their say”

      And younger, poorer people clearly did not have as much opportunity to have their say.

    3. One thing that political science teaches us is that the formal practice of democracy is not the same as the substance of it.

      For instance, the Soviet Union had elections. People turned up at ballot boxes ever few years, and voted for candidates for office. But in spite of this nobody would describe the Soviet Union as a democratic society: they lacked essential elements like competing political parties, freedom of speech and assembly, and all the other things that allow everyday citizens to participate in the process and hold bad leaders to account.

      This is not to say that public consultation in NZ is akin to the Soviet Union – far from it! However, we do still need to ask some hard questions about whether we’re getting the substantive elements right.

  5. “….Third, to what extent did submissions on the Unitary Plan affect the process at different stages? For instance, were the areas that Auckland Council chose to “downzone” between the draft and notified versions of the Unitary Plan concentrated in local boards with higher submission rates?”

    Conversely, were there areas that Auckland Council chose to “upzone” between draft and notified versions in areas with low submisison rates?

  6. I live in one of the “poor” areas of Auckland. I didn’t submit because it seemed like a huge effort to submit only to be completely ignored. I tend to think the whole system is rigged and one person doesn’t matter. But it is also why I support organisations such as Generation Zero because they have a voice that people listen to.

  7. It’s far more likely to send in a complaint than to lodge an endorsement – it is an example of a self-interested vocal minority, not an example of democracy.

  8. Great post.

    It is interesting how closely the map of submitters per capita correlates with the areas that received higher density zoning. The large upzonings in west Auckland clearly align with the low rate of submissions, while the reverse is true for Orakei. This might be a reasonable outcome if the number of submitters in Orakei was anything close to a majority. But even in Orakei where the rate of submission was relatively high the number of submitters is a very small minority of 12 per thousand residents. So the system gives a very heavy weighting to a submission rate of 12/1000 and a low weighting to a submission rate of 1/1000 despite the fact that both numbers amount to a tiny minority of people in the area.

    1. Thanks Frank! I am planning on taking a look at the relationship between submission rates and outcomes from zoning decisions in a future post. (Assuming I can get some decent data.)

  9. It could be related to personal disposable time and education level.

    People who are retired has a lot of time.
    People who are wealthy doesn’t work long hours, so they have some time.
    People who has high income level has more literacy around council matters. So they are more likely to submit.

    People who are working on minimum wage have to work long hours to survive. So they have very little free time.

    People who are working on minimum wage are less likely to be educated. They would not understand the council jargons and documents. Nor they understand abstractions, logical fallacies and complex matters.

    1. personal disposable time and education level are both likely to be correlated with income, so I’d expect the latter to pick up some of their effects.

      Note also that in a sample size of 21, there is limited scope to include additional explanatory variables (like you suggest), because you don’t have many statistical degrees of freedom.

    2. Yes, those are likely to be relevant factors. That informed my selection of demographic variables – ie “share of people over 65” is a proxy for people who have more free time, while “median personal income” measures financial resources and (to an extent) skill levels.

    3. I think you are right. It is a matter of opportunity cost. If you are struggling then you are better off spending your time on other things like food, shelter or your kid’s medical bills. If you have an easy life then you probably have nothing better to do than make submissions, or join heritage societies or promote cycling or type things into comments sections of blogs.

  10. Interesting post.

    Minor statistical note: I’d use a poisson regression here with offset for population I think, as the residuals are unlikely to be normal due to the positivity of the rates. This may not change the results all that much though – depends how much out it is.

    Also, remember the ecological fallacy (pretty sure you do given above posts!) 🙂

    1. Good point on the residuals. Given that I used OLS I probably should have reported heteroskedasticity-robust standard errors but they’re a bit of a pain to construct in R.

      Ecological fallacy is also an excellent point. I think I was careful to discuss the results in terms of what they tell us about variations between *areas*, not *people*. But it’s easy to slip between the two…

Leave a Reply

Your email address will not be published. Required fields are marked *