This previous post considered wider socio-economic factors which might shape the development/deployment of transport technologies, namely 1) denser cities, 2) policy settings, and 3) demographics. In this post I now discuss some more specific issues that are relevant to transport technologies, specifically:

  1. Economies of density
  2. Costs: Fixed versus variable
  3. Complements versus substitutes
  4. Conclusions: The Spruce Moose?

The takeaway message from this post is that the future is complex and uncertain. D’uh … read on for enlightenment!

1. Economies of density

The Prime Minister’s recent speech, in which he announced funding for the City Rail Link and the East-West Link, is an example of how policy settings can respond to wider socio-economic factors, such as population growth. Auckland’s rapid growth means that public investment in underground passenger rail stacks up more now than it perhaps has done in the past. As a result changed, Auckland’s transport network is set to change in a rather profound way.

Why has Auckland’s growth helped make the case for the CRL? Well, passenger rail has relatively high fixed costs. This in turn means that passenger rail experiences so-called economies of scale (or more accurately density). The phrase “economies of density” is used to describe situations where marginal costs are below average costs. This means that average costs fall as demand increases. Fixed costs like rolling stock and train stations all contribute economies of density in passenger rail.

Last week saw another example of “economies of density” in the context of New Zealand’s transport system, although this time the investment originates with the private sector. Emirates announced that – from this March – they will begin direct flights between Auckland and Dubai (NB: Apparently Qatar Airways is considering similar moves). Direct flights to Dubai will shave approximately 3 hours off travel-times between New Zealand and a number of destinations in the Middle East and Africa.

Economies of density lead to virtuous/vicious cycles: Higher demand begets lower costs, which begets higher demand etc (and vice versa). That’s why they are sometimes called “positive feedback loops”. Domestic air travel in New Zealand provides a nice example of virtuous cycles at work. Here, a combination of falling fuel costs, higher visitor numberslarger and more efficient planes, and competition has seen airfares fall (NB: I’m waiting on updated inflation data before making a more definitive statement). Many transport technologies, especially those where passengers share vehicles and/or facilities, tend to experience economies of density.

Private vehicles also experience economies of density to some degree. High levels of car ownership, for example, makes it easier to find a petrol station when you need one. The key difference between private vehicles and other transport technologies, however, is the degree to which road networks experience congestion. Beyond critical levels of demand, traffic congestion increases rapidly, i.e. congestion is a convex (upwards sloping) function. Congestion externalities are an example of dis-economies of density, and it’s one that is apparent in most cities of Auckland’s size and larger.

Congestion externalities are also an issue that things like driverless and/or electric vehicles do not resolve. Road pricing does solve congestion, although this is achieved by suppressing demand. So when people argue that electric and/or driverless cars will reduce demand for PT I’m somewhat sceptical, because these technologies don’t seem to change the fundamental dis-economies of scale that afflict urban road networks.

And when one thinks of the growth that Auckland will experience over the next 20 years, then economies of density are actually rather important. Perhaps more important than new technologies themselves …

2. Costs: Fixed versus variable costs

Our previous discussion noted that economies of density/scale arise in the presence of fixed costs. In this section I want to distinguish more carefully between fixed and variable costs. One useful way to think about transport costs is using a matrix which has “fixed/variable” versus “private/public” dimensions (NB: The latter is not relevant to this discussion, but can be for discussions of transport pricing more generally).

Cost table

Of course, such distinctions involve some arbitrary judgements. Parking for example, is both a private and a public cost. I’ve categorised it as “public” simply because a proportion of parking costs are subsidised by society. Despite the need for such judgements, I think this type of cost matrix is quite useful for classifying costs and subsequently analysing how changes in transport technologies (or policy settings for that matter) might impact on demand.

Using this framework (and a bunch of heroic assumptions) I have constructed the following graph to illustrate how the costs of cars compares to taxis and public transport, and how costs vary with distance measured in kilometres travelled per annum. PT fares are capped at the price of 12 monthly passes at $220 per month).

C1 Total private costs (c)

Here we see that cars become cheaper than taxis for people who are travelling more than 2,500 kilometres per year. Of course, the orange line is just one possible instance of the cost curve for cars. Depending on their preferences and circumstances, some people may opt to buy a cheaper car with higher operating costs, which would reduce the intercept but bump up the slope of the orange line. Nevertheless, the general point remains: For people who travel a lot, owning your own car makes more sense than using taxis.

Now consider a future scenario where electric driverless vehicles (“robot cars” for short) are available. How might they impact on these cost curves? Well, let’s assume robot cars cost 50% more than conventional cars; are 50% cheaper to park; and cost 25% less per kilometre to operate/maintain than conventional cars. When robot cars are used for taxis, they remove the need for a driver and increase higher utilisation. For the sake of the argument, let’s assume that the cost of using taxis drops by 50% with the advent of robot cars.

Under these assumptions, the cost curves presented above can be re-calculated as follows.

C2 Total private costs (future)

In this scenario taxis suddenly become cheaper than private cars for people who travel less than 10,000 kilometres per annum. By way of comparison, that’s slightly more than the average New Zealander travels per year. By pushing up fixed costs but reducing variable costs, robot cars seem to dramatically increase the scope for cheap taxi services to meet the demand for car travel for a wider range of households.

So consider this “what if” scenario: What if the primary impact of robot cars was to reduce the cost of taxis? In such a scenario would we need more or less PT? This leads me nicely onto the next topic …

3. Complements versus substitutes: A or B versus A & B?

When a new transport technology hits the market, one of the key questions we should seek to answer is whether it will complement or substitute existing technologies. The answer is not always obvious, but it is always important.

To provide an example, consider whether taxis increase or decrease demand for public transport. Many people think taxi’s compete with public transport, i.e. the two are substitutes. Rodney Hide, for example, had this to say on the matter (source; emphasis added):

[Uber is] … a whole new way of doing business. It disempowers bureaucrats and puts customers and drivers in charge. I love it. The public transport of the future won’t be clapped-out trains but driverless cars and Uber. It may happen much quicker than we now can possibly imagine. We won’t own a car. A driverless car is always just a minute away, ready to whisk us to our destination.

According to U.S. researchers, however, taxis are complements for public transport. This article provides a nice synopsis of relevant issues, while more detailed analysis is available here. The last report identifies “primary factors” for taxi demands, including 1) the number of workers commuting by subway and 2) the number of households that don’t own vehicles. Of course, places like New York have some attributes which are – for better or worse – not relevant to the New Zealand context …

Given that Auckland doesn’t have the density or subway system which exists in NYC, could Rodney be correct with his crystal ball-gazing? The answer, I think, is both “yes and no”. More specifically, taxis will compete with public transport in some places, while complementing it in others.

I think this article puts the distinction rather nicely when it says:

It is likely that these [taxi] services will just exacerbate the differences that already exist in the quality of public transit in different areas.  The localities that invested and ran their systems well won’t be as impacted.  In the areas where frustration is higher, the problems will surely become worse when revenues and ridership decline.  And the biases and history that impact why certain systems are the way they are will become more apparent.

The complements versus substitutes issue is relevant to many debates about new transport technologies. Some people seem convinced not only that driverless cars are imminent, but that they will drag people away from PT. I’m personally not so sure.

Conclusions:  “The Spruce Moose”?

One of my favourite Simpsons episodes starts with Mr Burns opening a casino. At the end of the episode, when the whole sordid venture is crashing down, an increasingly erratic Mr Burns pulls a gun on his ol’ mate Smithers and orders him to get into the model bi-plane affectionately known as the “Spruce Moose”.


While hilarious, the behaviour of Mr Burns is not something we should emulate when it comes to transport policy. History is littered with unfortunate examples of politicians picking winners well in advance of demand. Indeed, such attitudes gave us 50 years of motorway building in Auckland, replete with well-documented examples of optimism bias and strategic misrepresentation. My point? Let’s not hold a proverbial gun to our own heads and over-commit to any particular kind of transport technology. Instead, let’s embrace the transport technologies we have, and be passionate about using them more efficiently, while keeping tabs on new transport technologies as they evolve.

There is no doubt that we live in exciting times: New Zealand is growing and seems to be experiencing economies of density, at least in the transport sector. Costs of air travel are falling, while other technologies, such as passenger rail, are more viable than they have perhaps been in the past. Robot cars and trucks seem likely to emerge onto the scene at some point, helping our car fleet to become smaller, quieter, cleaner, and generally more efficient than it is currently.

Meanwhile, rapid developments in telecommunications (which I intend to discuss in a later post) are changing the way that goods and services are delivered. This change is occurring at astounding speeds and in amazing ways. In a few years Uber has grown to become one of the world’s most valuable transport companies. In Europe, a start-up called “Blah blah car” has raised hundreds of millions in venture capital to expand its business, one which is already moving millions of people around the continent at very affordable prices. I can book any one of 5 trips from Amsterdam to Brussels tomorrow for about 10 Euro.

In such a dynamic and rapidly changing environment it’s hard to know what to do. And that’s kind of the point: There’s a risk that we over-adjust to new technologies, just like the motorway builders did in Auckland 50 years ago. What I would do, however, is ask some hard questions “around the edges” of what we are currently doing. And perhaps consider delaying some projects slightly in light of the uncertainty.

Indeed, when it comes to the CRL, my main criticism of the National Government’s requirements for accelerating funding was that similar conditions were not also attached to other large transport projects. Appropriately specified triggers would seem to be useful adjunct to existing benefit-cost analysis and provide a more precise understanding of when work on a particular project should go ahead (NB: Of course there are other factors to consider, such as interest rates).

In an upcoming post I will talk more about emerging technologies for non-car transport modes; that stuff is so exciting it deserves its own post.

Share this


  1. It seems like your whole premise is based on your “heroic assumptions”. What are they? This is needed to determine the accuracy of your graphs

    1. Not really. The graphs are illustrative of some important relationships between fixed and variable costs. Not the main message at all … which is the direction of change in which driverless cars will shift relative prices.

      But its not rocket science. You can easily estimate fixed costs of owning a car per year, e.g. insurance, depreciation, etc. Then estimate a per km rate, e.g. $0.50 – $1.00 would seem reasonable.

      1. That price estimate seems about right at first glance but having a look at some actual data via the Fuelly website (user reported, probably doesn’t cover all costs, parking insurance blah blah blah) it seems to be in the area below $0.10 per km (caveat here about different vehicles varying greatly). Personally it is around $0.06 but this doesn’t take into account parking as I ride rather than drive. Parking prices, especially in Auckland, bump it up but without a lot of data it is difficult to add to model effectively. The changes shouldn’t make a huge difference to the graphs though. Interestingly in my personal experience, after having taken the bus for untold years when I moved to riding (office moved, taking multiple buses drove me nuts) my daily commute cost halved (even with subsidies) although I have a schedule I can change to minimise my contact with Auckland’s traffic. Your graphs don’t seem to take that into account given how you have modelled the upfront/yearly depreciation costs of vehicles. It is a specific market obviously but there are at least learner motorcycles which hold up their value very well, assuming they aren’t written off. This is to the point that, with correct timing, the purchase and future sale can result in a little positive income.

        That sort of brings me to my second point. The graphs are all nice and good but not many people economically analyse these things before making their transport decisions. They just get a car and be done with it. What you eventually end up seeing is people using a mixed mode so they will have a car and use a bus or taxi sometimes. This brings about ‘instant price differences’ it feels different to pay $20 petrol and get say a weeks worth of travel vs. paying $20 on a single taxi ride. Even if the first works out more expensive long term, there is a difference in the perception of the cost. People are not generally rationalistic about overall cost of ownership.

        What is difficult to capture in this style analysis is the qualitative life difference when choosing different modes. As already mentioned, I have for years walked or bussed everywhere. Was pretty easy living on majorish roads. Usually going to central locations. There are some things that are very difficult with buses though like carrying heavy or bulky sports gear (and getting taxi’s for some of that, the drivers get very antsy over carrying certain goods). There are places I just wouldn’t bother going and I tended to stay in a lot. Having the freedom offered by my own vehicle has made a big difference.

        It would be interesting to see what difference vehicle access makes in life style choices. People bemoan the use of cars, getting around everywhere rather than walking or cycling but I would be interested in seeing analysis of how active people are via differing levels of access to things like team sports and remote locations. Playing something like cricket over summer would be a nightmare if having to rely on buses. Lots of gear and potentially early weekend morning starts when bus frequencies aren’t ideal (although that might only be little kid cricket where parental transport is more of an option). Auckland has some amazing bush walks, not that far out of town. But getting there and back on a bus? Potentially doable sometimes…

        1. Hi Nick, thanks for the comment and here’s a couple of points for you to muse over:

          1. You’re right, the changes you mention won’t change the graphs a lot. The main thing they seek to highlight is the direction of change in relative prices for cars versus taxis when driverless cars become available. It’s fairly clear that the latter will reduce the relative price of using taxis. The numbers can move around a lot, but the conclusion stays the same.
          2 I would be careful using that self-reported cost data, mainly because $0.06 per km is not plausible. The average NZ car uses 10 litres per 100km, or approximately $0.15 – 0.20 per kilometre. That is before any allowance for maintenance, use-related depreciation, and parking etc. Also worth considering who is most likely going to be in the market for driverlerss cars when they become available? The answer is, of course, people/businesses who were already looking to purchase a car. Things like scooters aren’t likely to be a relevant consideration for them …
          3. I agree with you that other factors influence people’s purchasing decisions, although I don’t see how that is relevant here. Again, the main message is that the advent of driverless cars is to shift relative price levels in favour or taxis compared to owning one’s own car. Yes there’s lots of non-price related qualities that impact on people’s final decision, but that doesn’t change the direction of change – only the size of it.

  2. Whole I agree with you that some of the opinions on driverkess vehicles are not necessarily that well thought through, one thing future technologies (especially driverless vehicles) do is increase the risks associated with returns on investment in long lived transport investments. Accordingly this should be considered when determining approproate discount rates and time horizons over which to assess projects.

    1. Looks like rereading your comment you have said something similar at the end.

      Another comment though, if I may:

      “Road pricing does solve congestion, although this is achieved by suppressing demand. ”

      Not clear what you are intoning by “although…”. It appears this may be a bad thing? This is a frequent position I see coming up across the political spectrum which is that transport is some kind of maximand as opposed to a service that should be consumed where its benefits exceed the cost of provision. Its never been clear to me why it would fall in the maximand category.

      1. My use of “although” reflects my view that policies which suppress demand are typically difficult to implement even if they are relatively worthwhile.

        1. To quote one of my favourite Homer Simpsonisms:

          “You say that, like its a bad thing”

          So, I take it what you really meant to say was something like this?:

          “Road pricing does solve congestion, by suppressing demand. Although this is often achieved with some difficulty, so it may not be as effective a solution as it appears it should be.”

          Which reads differently but makes your meaning clear(er)?

          My comment on that is that yes, but also point out that its difficult, often, if not mostly, because the political will is not there to implement, or keep enforcing it once implemented. Not because its technically impossible or that difficult to do.

          You could draw a direct parallel between road pricing debate and pricing water usage.

          Auckland used to have this issue with effectively unpriced water, via flat rate uniform annual charges.

          Because in part, when people paid almost nothing for their water supplies (fresh and waste water) they tended to overuse it.
          So much so, that we had to build the expensive pipeline from the Waikato to Hunua storage lakes to compensate for that and add additional capacity to the sewage system to take it all away afterwards. Now we have actual water usage metering, people now use water more wisely and will fix that leaking supply pipe to their house, or the leaking toilet cistern rather than letting it run. As a result the average usage of water in Auckland per household is now quite lot lower than it was 25 years ago. So we have avoided the need to overbuild our water supplies or the downstream sewage system.

          However, there are some things which are not suitable for this sort of price based demand suppression – fresh air for one.
          While you have a (human) right to fresh air, you don’t have one for uncongested roads. Your life won’t end if you have to drive in congestion for instance.

          Although the way some politicians act to any suggesting of road pricing you might think that congestion and road pricing were simply going to kill you.

        2. I guess that may be the case although I agree with Greg that it is never really tried. I think it is good you referenced congestion pricing in this post though as it does qualify as an emerging technology, or at least the tools associated with it have developed rapidly,. We are now in a position where GNSS TDP road pricing is technically feasible which, IMO is probably the most significant new “transport technology” of our life times, even more than driverless cars. It is certainly technically closer to being able to be rolled out than driverless cars in IMO.

          1. Mathew why don’t you write an article about that? In my experience Patrick, Stu and co are co-operative with well written articles that are evidence based.

          2. Well it would be an honour to post on such a high profile blog, however I have to disclose I am an opinionated punter with no real professional expertise in any such matters! I will see what I can do though…

          3. Matthew W – I completely agree with you on all points about congestion pricing being a highly relevant transport technology. That’s why it featured heavily in my first post on the topic of transport technologies, which I reference at the start of this post (NB: Look under the section on “policy settings”). I only mention it here in passing, because it’snot that relevant to this particular post. It will feature in a future post however … so stay tuned, and I agree with the others: Please do consider submitting a guest post on it.

            We have written on road pricing several times in the past (which you could reference for material/angles), but it’s always good to get new perspectives – mainly because it keeps things fresh and avoids echo chamber type environments.

      2. “Road pricing does solve congestion, although this is achieved by suppressing demand. ” We always hear about suppressed demand but to be more correct it is quantity that is suppressed by price. Demand stays the same as it is a function of quantity people want at each given price. Or even better it is an inverse function because Ricardo or some early economist got his axes around the wrong way. Demand only shifts when the entire curve shifts due to changes in opportunities.

    2. Surely the demand suppression aspect of road pricing is no problem if there are good alternatives available. Suppressing demand of something with high negative externalities (urban SOV use) while providing alternatives without them (well priced fast broadband, improved local amenity, Rapid Transit Network, cycling network etc) is surely the answer.

      1. Greg N nails it.

        Personally i dont have a problem withu using road pricing to suppress demand provided 1) its well signaled (so people can adjust in advance as much as possible, ie mitigate costs) and 2) revenues are applied in a way that is efficient and equitable (i.e. to worthwhile transport projects that deliver benefits to low income households).

        But i know a lot of people don’t like road pricing, and i respect their views. Hence i wouldn’t see it as something we needed to ram through fast, even if a bit of leadership and long term certainty on the issue would be very useful.

  3. Uber isn’t “new transport technology”; it is a giant, criminal rort using existing technologies in unregulated and unsafe manners.

    It’s like saying “new revolution in building costs!” where you mean unlicensed builders not building to code. Yes, cheaper, faster, and easier. Not good.

    1. Agree it’s not necessarily a new technology, but I don’t see how it is a giant criminal rort. It appears to be a good willing user, willing provider service, where people have decided that they don’t want a whole lot of red tape and bureaucracy adding to the cost of their trip.

    2. I think it’s fair to say that some more work is needed around how businesses like Uber operate within our regulatory frameworks, but I don’t think you’ve given a fair summation of their business. Their business model does away with a great deal of inefficiencies and overheads associated with the traditional taxi business model, and that leads to savings for customers (and, anecdotally, better pay rates for drivers as well). Innovation is a good thing.

      1. Innovation is not automatically a good thing (though blocking innovation is definitely a bad thing). I think that Uber definitely improves on the taxi transport service, but I am concerned at the whole “fake self-employed” shenigans. Its a giant company with lots of hired hands. They can cut pay going to drivers by 20% tomorrow, and what are those drivers gonna do to prevent them from doing so? Open up their own service and compete, at higher rates and without the advantage of the multinational giant back-end and name recognition? So I’d prefer regulation that clarifies that they are employees, not entrepreneurs.

        1. They’d probably switch allegiance to a competitor, or otherwise leave the company. The contracting structure doesn’t give Uber carte blanche to set their drivers wages; if it’s not lucrative for the drivers to drive for Uber, then the whole business will collapse as they’ll find themselves with noone to ferry their customers around.

        2. I disagree that Uber is a “giant company with lots of hired hands”. It’s a booking and payment system that allows customers to deal more efficiently with private operators. To take a local example, it’s just like Nobody is suggesting that bookabach should go out and buy all these houses and rent them out directly.

          1. And God help us if Uber gets intermediated by 3rd parties the way bookabach has by “agents” who are rapidly destroying its appeal by preventing the willing buyer & seller directly communicating and discovering the many and varying definitions of ‘willing’.

    3. Uber is a new technology, at least in an economic sense. It combines existing technologies to provide services in a more user friendly and efficient manner. Hence its a new technology, broadly speaking. Same with blah blah car. Note that new technologies dont have to mean new types of physical rocks and mortar. Most new technologies are simply new combinations of existing technologies with some small tweaks. E.g. electric bicycles.

  4. My favourite Simpsons episode is where Springfield is conned by outside experts into building a monorail but only Marge sees it for the scam it really is. Of course Homer gets a new job as the driver….

    1. Yes that’s a great episode too. I also think it was kinda funny with bronwlee invoked that episode when criticing the crl, even if i completely disagreed with him.

  5. Complements versus substitutes:

    A great example of taxis being a complement to PT is Bangkok where I spent the last week. To get from the centre of town to where I was staying I would first take the BTS to the end of the line since traffic is so bad walking is probably faster than driving. Then once at the end of the line (BTS Mo Chit/MRT Chatuchak) there are lines of buses, minivans and taxis waiting to take passengers to areas not covered by the metro systems.

    Most of the end of line/major area stations are like this and as the BTS/MRT extensions and new lines open up something similar will occur in those areas, the taxi drivers will like it because they will only have to do shorter runs in areas they know well and can stay away from traffic and without the taxis congestion will drop along the rail routes.

    Bonus: At BTS Udom Suk there is a free shuttle bus to Mega Bangna so you can go quaking at Ikea, just use the delivery service to send all your new furniture home for you.

    1. Great example – thanks!

      As an aside, I always think it’s strange that people like Rodney Hide and Don Brash (who lay claim to economic credentials/expertise) would make statements on taxis versus PT without considering the economic evidence. Does this suggest that the ideology to which they submit has subverted their critical thinking skills?And that the opinion they express are merely justifications for the positions they have already adopted, rather than based on evidence?

      If so then it’s quite sad to see a former Governor of the Reserve Bank stoop to this level … although if you read any number of psychology books it’s “all too human”.

          1. yes, just look at Australian cities (and Wellington) if you want evidence of the sorts of problems which emerge when parochial little fiefdoms strangle the metropolis.

            IMO Auckland’s system of governance, while certainly not perfect (councillors elected at large would be nice), is probably the best in Australia or New Zealand.

Leave a Reply