In this earlier post Matt discussed the proposed Mill Rd project in South Auckland.

I actually grew up in Franklin (Waiuku) and know the wider sub-region quite well. Mill Rd is, in my experience, an unsafe stretch of road within a somewhat disconnected/fragmented network. So there’s definitely some transport/land use issues in the area that deserve our attention.

Tick to AT on that front.

As a transport economist, however, one must always ask whether the benefits of a proposal are commensurate with the costs. In terms of Mill Rd, the economic benefits of the proposed project are shown below (NB: This is extracted from the scheme assessment report pg 79).

Project benefits

The benefit cost ratio of the project is stated to be 2.2. This means that the project receives a “medium” rating for economic efficiency under NZTA’s project evaluation policies. It’s worth nothing that in Peter’s post on the MoT’s analysis of capital spending on roads, local road projects were found to typically have BCRs in the range of 3-4. Mill Rd’s economic efficiency is, in comparison, somewhat underwhelming.

Digging a little deeper we can see that the largest 3 benefits attributed to the project are 1) travel time savings ($271m); 2) agglomeration benefits ($69m); and 3) vehicle operating costs ($27m). I thought the benefits ascribed to agglomeration were the most interesting (NB: For those who haven’t heard of “agglomeration” before, you might want to read some of our previous posts on the topic here and here).

In a nutshell the literature suggests the primary agglomeration economies typically arise from:

  • Regional economies of scale, i.e. larger markets for goods and services (especially more efficient labour markets). This can be achieved by either bringing more people/firms into the city and/or bringing existing people/firms closer together by reducing transport costs;
  • Local knowledge spillovers in dense urban environments that increase productivity; and
  • Efficiencies in the public provision of infrastructure/services.

When discussing agglomeration benefits the Mill Rd SAR notes (emphasis added):

“Agglomeration economies describe the productivity advantages that arise from the close spatial concentration of economic activity. There is a strong link between transport provision and the benefits that arise from the spatial concentration of economic activity. 

The contribution of the improved Redoubt Road-Mill Road corridor to the upgrading of the Auckland transport system qualifies for the agglomeration benefits to be taken into consideration. Economic Evaluation Manual, Vol.1, Section A10 provides the methodology for estimation of these benefits. 

The corridor provides an access route to Auckland CBD and Manukau City Centre. Both are major employment and commercial centres, which justify an adoption of the agglomeration benefits for the project. The value of these benefits was assumed at 20% of the total benefits, which is conservative as similar projects in the Auckland region use values in excess of 25% to 30%.

Having observed that agglomeration economies arise from the close spatial concentration of economic activity, the SAR then proposes to estimate agglomeration economies by applying a blanket 20% factor to total benefits.

At this point I emitted audible “hmmm”, a bit like Homer Simpson day-dreaming about warm pie (NB: Source).


The SAR is correct that agglomeration benefits can arise from spatial concentration of economic activity. Where the SAR seems to make a fairly large leap of faith, however, is by assuming that the Mill Rd project (and specifically the land use patterns it enables) will logically lead to sizable positive agglomeration benefits. It’s worth noting at this point that from a quick read the above extract is the only discussion of agglomeration economies in the entire (200 plus page) report, despite them supposedly representing the second largest source of benefits for the project.

Before discussing agglomeration economies in more detail let’s introduce an important concept” The “counter-factual”. This describes would happen without the Mill Rd project. In the case of agglomeration economies, we’re primarily interested in land use effects, i.e. what would those 10,000 households do in the absence of the Mill Rd project?

It seems plausible to suggest that some households would simply choose to locate in the area anyway, while enduring slightly longer travel-times. For these households the agglomeration effects are almost identical to the base case. Other households might instead choose to locate somewhere else, most likely in a more central location. This  would actually tend to generate larger agglomeration benefits than would have arisen had they located in the Mill Rd area, i.e. for these households the Mill Rd project can be considered to have a negative effect on agglomeration economies.

This discussion highlights two important points about agglomeration economies: 1) You can’t have a very informed discussion about them without first carefully defining the counter-factual (land use) scenario and 2) depending on this counter-factual scenario, it is possible that transport investment gives rise to negative benefits, because it encourages/enables development to spread out more (and create more congestion) than would have eventuated otherwise.

In this context, simply assuming a 20% agglomeration premium on total benefits strikes me as a tad presumptuous. Let’s now go back to the three micro-economic channels that contribute to agglomeration economies that we listed above and consider how they relate to the Mill Rd project.

First, it seems unlikely the Mill Rd corridor will, on its own, impact on regional economies of scale. It’s simply not a sufficiently large step change in accessibility that it would encourage more people/firm to migrate to Auckland than would have done otherwise. Indeed, with Auckland’s annual population growth running at approximately 50,000 people per annum, the total growth expected in this area over the next 30 odd years (25,000 people) is a veritable drop in the growth bucket. It’s primary effects seem to be bringing Papakura closer to Flat Bush and Botany Downs, and both of these centres slightly closer to Manukau. Unlike the SAR, I’m not sure how it provides a new access route to the city centre, at least to a degree that would have implications for regional productivity.

Second, it seems unlikely that the Mill Rd will contribute much to knowledge spillovers. In a geographic sense, the area is right on the periphery of the metropolitan area, and relatively remote from employment areas to the north. It is especially remote from the city centre, which is the source of most of the knowledge spillovers in Auckland. While secondary centres like Manukau do experience some agglomeration economies, these seem more likely to arise due to the two types of agglomeration discussed above and below.

Third, Mill Rd doesn’t seem to give rise to major efficiencies in the provision of public infrastructures and services. Indeed, the Mill Road transport project represents approximately $20,000 in CAPEX costs for every household that is expected to locate in the immediate vicinity. It seems plausible to suggest that these households could be accommodated for similar (if not lower costs) somewhere else in the region. To make the case for these agglomeration economies, we would expect to see evidence of surplus capacity existing in nearby health and education facilities.

So what’s the take-away message from all this econo-mumbo-jumbo-fiddle-faddle?

First, even taking the economic evaluation of Mill Rd project at face value, we find that it’s a fairly mediocre local road project. While there are transport issues in this area, addressing them in this way seems expensive compared to most local road schemes around the country.

Second, I’m not confident it is reasonable to assume the agglomeration benefits of the Mill Rd project equate to 20% of the total benefits. By extension, I question whether there is sufficient evidence to include them in the base economic evaluation. Instead, it seems more reasonable to treat them as a possible sensitivity, and even then 20% seems to be a very high premium given the location of the project. Removing these benefits from the evaluation causes the BCR for the project to drop from 2.2 to approximately 1.8. It’s important to note this is not necessarily a terminal issue for the project; the proposed scheme obviously still manages to address some importany problems in the wider area. It’s just probably not as urgent as it appears from the evaluation.

Finally, it highlights what I think is the most important and interesting policy issue: That is, the need for land use policies to be informed by transport outcomes, and possibly linked to financial incentives. In this case, it seems like the Mill Rd project will cost quite a lot. Perhaps, then, there is a need for AC to revisit the merits of enabling growth in this area. Indeed, AT can rightly say that they are to some degree simply responding to land use plans developed by AC. And the latter may want to consider whether development would even occur out this way in a hypothetical situation where the actual infrastructure costs were internalised into the costs of development and ultimately passed onto consumers?

What do you think is the best way to balance public/private interests when it comes to transport and land use outcomes?

Share this


    1. we’re all guilty of the “stuff this” approach at various times :). And yes agglomeration is rather complex. Important, but complex. Hence my suggestion the benefits are not included in the base evaluation unless there’s a robust methodology and bespoke calculation supporting the numbers.

      But ultimately not a huge issue given that this is at the SAR stage and so primarily focused on securing designation, after which detailed design/evaluation should be undertaken. Unless you’re a RoN of course, at which point BCRs are lasered to an invisible death by the largely circular “strategic fit” reasoning.

      1. I am not sure I am even convinced that agglomeration is an ‘additional’ benefit. The values we use for travel time are calculated in cities that have agglomeration, the travel time savings or travel time for new trips therefore includes benefits that reflect the fact that people and businesses derive value from that trip being made. It seems to me that if we go down the path of enumerating agglomeration benefits then we should also be pruning back and travel time benefits that we included using the basic procedures.

        1. Interesting comment.

          My intuition is that agglomeration benefits are largely additive. Reason being that there’s three distinct economic indicators where AE might be internalised: 1) wages (gross and net), 2) rents, and 3) profits.

          It seems the risk of double-counting with travel-time valuations that you point to may is most evident when considering net wages. But there’s a reasonably large body of literature that finds AE also accrue to gross wages (Venables’ tax wedge); rents (Donovan masters thesis); and corporate profits (which in a dynamic sense tend to lead increases in wages and/or rents).

          P.s. But I do agree with you: It may be worth taking a conservative approach with travel-time values in instances where AE are being estimated.

        2. Definitely an important question! Stu and I have both grappled with this one in the past.

          It’s instructive to read Kernohan and Rognlien’s 2011 NZTA research report (, which provided the theoretical justification for inclusion of agglomeration in the standard EEM procedures. My reading of the paper is that “wider economic impacts” like agglomeration economies should _only_ be counted where they arise from a market imperfection in labour or land markets.

          With that in mind, there are two broad stories about how agglomeration arises.

          The first is that agglomeration is a function of increasing returns to scale at the firm level. In this case, agglomeration is _not_ an externality – there are no inter-firm spillovers! But note that this does assume a market imperfection in the form of fixed costs of production/increasing returns to scale. (Which means that smaller firms will tend to be outcompeted by large firms – no level playing field!)

          The second is that agglomeration is an externality, or spillover, that arises when firms/workers co-locate. This may take the form of knowledge spillovers between firms (which you could think of as arising from the fact that workers are mobile and retain knowledge, or the “public good” nature of knowledge), or some other mechanism such as efficiencies in infrastructure and public service provision.

          Now, my intuition is that both stories are true, albeit to different degrees in different industries. There are multiple market imperfections that give rise to additional benefits related to agglomeration. But I’m a bit less confident in our ability to accurately forecast agglomeration benefits of individual projects.

  1. Right there is the main question of ‘Do We Need It – at such a scale?’ The answer is no.
    Do we need it at a lesser less gold plated scale? The answer is yes.
    Is Auckland Transport and more to the point Auckland Council via the Auckland Plan missing something here? The answer to that is definitely!

    For all purposes I am commenting on the current Northern Section of the Mill Road Corridor, not the Southern Section. And to keep things clear I live in Papakura five minutes away from the Mill Road corridor, I also use the corridor as well as a bypass link to Manukau when the motorway packs it (which is often) or to get to East Auckland.

    Now the Mill Road Corridor project’s original purpose when it was born back in the Manukau City Council days was to bypass a Southern Motorway that was not due for upgrading until 2025. Auckland Council AND Auckland Transport hold onto this legacy mantra despite thanks to Judith Collins and Calum Penrose the motorway upgrade starting this October. Thus with the motorway upgrade to get under way it negates the need for this 4-lane defacto motorway.

    All that is needed know is safety upgrades, provisions for bus lanes for when the area develops (in the case of Redoubt Road that would be now), and the provision of actual grade separated cycle lanes rather than painted ones.

    But reading Stu’s article it seems the entire exercise if we look at all the information presented by Auckland Transport leads to AT presenting a self defeating argument (bit like RoNS). Council’s support behind it for similar reasons leads to a self defeating situation as well.

    What Bryce showed me makes more sense along with the upgrades I mentioned earlier
    It reality it is east-west connections missing rather than a snaking north-south connection. The case for the Manukau Rail South Link becomes extremely clear cut using Bryce’s bus to rail links with AT and Council recognising Manukau City Centre as a major employment centre and people heading towards it from the South and South East. Anyone from Murphy’s Road area needing the Southern to go north should be using Te Irirangi Drive. For Manukau, Airport and Wiri this is where the Te Irirangi bus way or the Botany Line Sky Train would come in.
    I could go on but that is best put for my own post ( ) and eventual submission.

  2. BTW, my understanding is that they shifted the BCR rating categories recently and now anything less than 3 is ‘low’?

  3. “In a geographic sense, the area is right on the periphery of the metropolitan area, and relatively remote from employment areas to the north. It is especially remote from the city centre, which is the source of most of the knowledge spillovers in Auckland.”

    Is this really established?

    It was my understanding that agglomeration effects have been established for metropolitan areas in terms of their scale as a whole, but not density or activity in specific areas, e.g. the CBD.

  4. “Third, Mill Rd doesn’t seem to give rise to major efficiencies in the provision of public infrastructures and services.”

    These are fiscal externatilities – are they relevant? I’m not convinced.

    1. “These are fiscal externatilities – are they relevant?”

      If you care about your tax dollars being spent efficiently, then yes, they are highly relevant.

      1. If we choose to socialise the costs of x, we are choosing to spend our tax dollars based on other peoples choices. The inefficiency arises from the socialisation of costs. If we don’t want to pay for x for someone living in the back of beyond, the answer is not to socialise it.

        1. Yes I think they’re relevant. Publicly provided goods are almost by definition imperfectly functioning markets. If mill rd area has higher marginal costs in provision of public goods/services than other areas then not enabling development there is a fiscal and economic benefit. And vice versa.

        2. Publicly provided goods are quite different from public goods. In any case, I dont think just stating that something is an imperfectly functioning market is enough to answer the question at hand.

          I am not disagreeing there is a fiscal cost, but the fiscal costs are really associated with the government policy to subsidise a particular good or service. Even with the fiscal cost, it is not necessarily inefficient. We would need to know about the effects of government subsidisation of healthcare and education on the actions of the people who choose to live down Mill Rd. Without the subsidisation, it might nudge them to live somewhere more central, but who knows. That is the relevant question. It is not simply a matter of – does it cost the government more to service them?

        3. Yes I appreciate the distinction between publicly provided goods and public goods. I wouldn’t agree that they need to be treated differently from a BCA perspective, however.

          Public goods are of course non-rivalrous and non-excludable. Their marginal cost < average cost and they would be undersupplied by a private market. Publicly provided goods, e.g. health and education, are rivalrous (and arguably excludable, if you ignore United Nations conventions on human rights etc). However, publicly provided goods/services do often experience fixed costs and economies of scale, hence leading to variations in marginal costs and a form of agglomeration economies. These are, moreover "external" to the costs presented to "consumers" of said goods/services. I think the difference in our positions can be summarised as follows: I'm arguing that when considering a particular transport intervention, you need to take other public policy interventions (e.g. health/education) as given. In which case observing that the marginal cost of providing health/education varies across the city (and is external to consumers) is a economic benefit/cost that should be included in the evaluation. I acknowledge the presence of the externality in health/education arises perhaps due to a prior policy decision, but so what? It's an economic cost nonetheless. To ignore it would be, well, evaluate transport projects as if the government didn't also intervene in other markets. You (I think) are arguing that because publicly provided goods/services are subject to political whim, and that policies can change, then they are a somewhat ephemeral, or constructed, economic benefit. Theoretically your position may be defendable, but in practice I don't think it makes much sense. And it could quite plausibly lead to bad economic outcomes, e.g. transport investment leading to land use development that in turn required huge public investment in new schools and hospitals.

  5. This is an interesting topic.

    “I acknowledge the presence of the externality in health/education arises perhaps due to a prior policy decision, but so what? It’s an economic cost nonetheless.”

    The question I have is – how are you defining an economic cost? I am thinking of it in terms of economic efficiency i.e. efficient allocation of resources. If people would live down Mill Rd with or without the government subsidy of services then the marginal cost of providing the service to those people vs somewhere else is irrelevant. Just because it costs the government more or less, it is not inefficient in terms of allocation of resources – the same investments would have occurred with or without the government subsidy.

    What I am saying is, just because you have a higher cost to government from a given land use pattern, it doesn’t make it inefficient and therefore relevant when looking at the effects of a transport investment. The underlying reason is that people have individual preferences that may have been expressed the same way without the health/education subsidy.

    Now, transport investment itself may be inefficient if it leads to development that would not have otherwise occurred in the presence of people facing their true costs on the transport system. But that is a different issue.

  6. A prime example of the costs of the transport system not falling on the users is the uniform “targeted” rate for Auckland just announced. All households, large or small, bear the same cost, wheher they use private or public transport or not.

Leave a Reply

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