Now that’s a traffic jam

Think Auckland has a congestion problem, take a look at these images from China a few days ago on the Beijing-Hong Kong-Macao Expressway. It’s the result of people heading home at the end of a week long national holiday.

China Congestion

China Congestion 2

The bottleneck kind of reminds me of this image from Sydney – and which is equally appropriate for another road based harbour crossing



Transport CBA, housing supply, and the spatial equilibrium

In comments to a recent post I wrote reviewing recommendations from the Australian Productivity Commission’s review of public infrastructure investment, reader Brendon Harre raised an important question about transport cost-benefit analysis (CBA). He commented that:

“the benefits of providing a grid of urban transport options (without mode bias) in advance of development in order to keep land, commercial and residential property affordable is not measured”

This is an important issue that’s worth careful consideration. As a best guess, I think that Brendon’s point isn’t quite true. In a roundabout way, transport CBA does capture benefits associated with enabling development. However, the modelling tools available might over- or under-estimate the magnitude of those benefits in some cases.

Let’s start by reviewing how transport CBA works in New Zealand. Here are the key steps:

  1. A transport agency or council comes up with a land use forecast – i.e. a rough idea of where people are going to live and work in the future.
  2. The transport agency then identifies two (or more) futures scenarios for the transport network in the area. For example, they may consider one scenario in which no new roads were built, and one in which a new highway is built at the edge of town.
  3. The agency then models the transport network under the fixed land use forecast and multiple transport network scenarios.
  4. Based on the modelling, it then calculates how travel times (and vehicle operating costs, emissions, etc) differ between the scenario. It sums up the reductions in travel times (etc), multiplies them by the average value of time, and then uses the resulting dollar value as the numerator in a benefit-cost ratio (BCR).

This procedure obviously bears little relationship to what we observe in practice. In reality, there is significant endogeneity between the availability of infrastructure and land use outcomes. In other words, if you build it, they will come, and vice versa. You can’t assume that land use will remain fixed if transport options change!

Another way of saying this is that rather than “banking” travel time savings from wider or faster roads, people tend to “re-invest” them into other things, such as living in a larger or cheaper house in a different location. (Or re-scheduling trips from off-peak times, shifting modes from PT, walking or cycling, etc.) Public transport is different, as it doesn’t get congested, but the principle is somewhat the same – speeding up journeys allows people to travel more.

Economists call this phenomenon “induced traffic”. I’ve previously discussed this phenomenon from a slightly different angle, focusing on the implications of induced traffic for how we manage and invest in road networks. I’ve argued that we should stop telling ourselves the lie that increased road capacity will ever “fix” congestion and accept that all we can do is give people alternatives to participating in congestion and implement congestion pricing to free up the roads.

However, I think it’s also worth considering what induced traffic means from a housing supply perspective. It’s useful to start by thinking about how individuals might respond to the opportunities created by new transport infrastructure. Let’s use the City Rail Link as an example, as we’ve got a good idea of what it will do for travel times:


Suppose I’m currently living in Morningside (I’m not, but it’s a simpler example) and facing the following costs for transport and housing:

  • Rent of $250 a week, assuming I’m flatting
  • Public transport fares of $30 a week, as a single journey to Britomart costs $3 with a HOP card
  • Travel time costs of around $130 per week, assuming that I value my commute time at around $20 per hour. It currently takes around 40 minutes to travel from Morningside to midtown by train, including the walk at the end.

Now let’s consider what will happen when CRL is done. My travel time will be cut dramatically – after CRL, it will only take 15 minutes to commute from Morningside. This is a big saving in travel time. Under these assumptions, CRL will make me better off by around $80 a week (i.e. ~4 hours saved * $20/hour).

However, I’ve also got the option to live further west in search of cheaper housing. Let’s say I choose to move to Henderson, where I pay a bit more in train fares (around $4.80 per trip) and save a bit of travel time relative to my old location. This only makes sense to do if it enables me to save at least $80 in rents for a similar dwelling. Otherwise, moving further out has made me worse off than simply staying in place and “banking” the travel time savings.

What we learn from this example is that the perceived benefits from relocating following the construction of new transport infrastructure, including lower housing costs or better quality housing, should be roughly equal to the added travel time cost of doing so. Economists describe this concept as the “spatial equilibrium” – i.e. people trade off housing and transport costs. As I found when looking at housing and commute costs in NZ cities, we can observe this trend empirically.

(That being said, there are reasons to think that moving further out in pursuit of cheaper housing is not necessarily a great idea. In The Happy City, Charles Montgomery argues that people overestimate the benefits they get from a bigger house, and underestimate the misery of longer commutes. But let’s set aside the impact of cognitive biases for the moment…)

The upshot of this is that, the standard approach to transport CBA actually seems to capture many of the benefits of new housing supply following transport infrastructure development. This sounds perverse – didn’t I say that transport models didn’t reflect reality very well? – but it makes sense when you think about how individuals make decisions about where to live and how to get around.

However, there are two caveats to this point. The first is that individuals don’t internalise all the costs (or benefits) of their location choices – there are externalities. If one location is better at generating positive spillovers in production or consumption (“agglomeration”), cheaper to serve with publicly-funded infrastructure, or responsible for fewer greenhouse gas emissions, it might be better to build infrastructure that will encourage people to live there. This is captured imperfectly in transport CBA at present – but it doesn’t have much of an impact on housing supply.

A second, more subtle issue is that our capital budget may be too constrained to deliver enough transport capacity to enable a sufficient supply of housing. For example, we may be pursuing a costly and land-intensive approach to supplying peak transport capacity that results in diminishing returns from investments. If that’s the case, we need to ask whether we have cheaper opportunities to add capacity to the transport network. (Or, alternatively, start raising taxes, which is always a popular option.)

What do you think about the spatial equilibrium in our cities?

Public transport and congestion in Wellington

Last week, I took a look at some new research from the Netherlands that estimated the benefits of public transport for car travel times based on data from 13 “natural experiments” – public transport strikes. The Dutch researchers found that PT provided significant congestion reduction benefits – around €95 million per annum, equal to 47% of PT fare subsidies.

While the data was specific to Rotterdam, I’d expect to find similar results in most other cities with half-decent public transport networks. The whole thing got me wondering: Is there any similar evidence from New Zealand?

Fortunately for PT users and drivers, but unfortunately for researchers, potential PT strikes have mostly been averted over the last few years. However, Wellington did experience a “natural experiment” of sorts back in June 2013, when a major storm washed out the Hutt Valley railway line:

Source: NZTA

Source: NZTA

The Hutt Valley rail line was out for six days, including four working days. During that period, things got pretty ugly on the roads, as the motorway into downtown Wellington didn’t have enough capacity to accommodate people who ordinarily commuted in by train.

The Ministry of Transport (among others) very cleverly observed that this was a great opportunity to learn something about the impact of PT networks on road congestion. During the rail outage, they surveyed around 1,000 Wellington commuters about their travel experiences. According to their report, they found that:

  • The closure of the Hutt Valley rail line put significant pressure on the road network. Delays for commuters were most severe on the Monday following the storm. Traffic on State Highway 2 was severely congested, with morning peak hour conditions lasting two hours longer than usual
    • 80 percent of Wellington commuters from the Hutt Valley and Wairarapa experienced a longer than usual trip
    • 32 percent of them experienced delays of over an hour
  • the severity of commuter delays lessened over the week, with the number of commuters from the Hutt Valley and Wairarapa experiencing delays of over an hour halving by Wednesday 26 June

Essentially, what happened was that a bunch of people who ordinarily caught the train from the Hutt Valley couldn’t do that due to the storm damage. A quick eyeballing of MoT’s graph of daily rail patronage suggests that around 4,000 people had to make other travel arrangements:

Wellington storm daily train patronage

Almost half of the rail commuters from the Hutt Valley opted to drive instead, while the remainder chose to take replacement buses or to stay at home instead. This had a serious impact on motorway traffic, as shown on this graph of hourly southbound traffic volumes. On a normal day (the green or blue lines), traffic volumes peak at around 7-8am, and fall off sharply after that.

By contrast, on Monday 24 June, when the rail line was out, people were still travelling in (slowly) until almost 11am. That’s some serious congestion:

Wellington storm hourly vehicle flows

Based on survey data, MoT estimated that the storm damage increased average travel times during the morning peak by 0.329 hours (20 minutes) on Friday 21 June, 0.309 hours (18.5 minutes) on Monday 24 June, and 0.230 hours (14 minutes) on Wednesday 26 June. It then used those estimates of average delay for people travelling at peak time to estimate the added cost of congestion that arose as a result of the Hutt Valley rail line outage:

Wellington storm cost of increased travel time

In short, a four-day breakdown in part of Wellington’s public transport network cost morning peak travellers around $2.66 million in lost time. If we assume that there was a similar level of delay during the afternoon peak, when people are commuting out of downtown Wellington, the total cost would be roughly double that – $5.32 million.

This can give us a rough estimate of the value of public transport for congestion relief in Wellington. Extrapolated out over a full year (i.e. 250 working days), these results suggest that the Hutt Valley rail line saves drivers the equivalent of around $330 million in travel time (i.e. $5.32m / 4 days * 250 working days).

That is a very large number. According to an Auckland Transport report comparing Auckland and Wellington rail performance, Wellington’s overall rail network only cost $81.2 million to operate in 2013. 56% of operating costs were covered by fares, meaning that the total public subsidy for the network is around $36 million per annum.

On the back of these figures, it looks like Wellington’s drivers are getting a fantastic return from using some fuel taxes to pay for PT rather than more roads. The travel time savings associated with the Hutt Valley line alone are nine times as large as the operating subsidy for the entire Wellington rail network.

There are two caveats worth applying to these figures, one practical and one methodological.

First, it’s likely that the value of rail for congestion relief is unusually high in Wellington due to the shape of the city. Here’s a map of Wellington’s population density and infrastructure in 2001 and 2013 (from my analysis of urban population density). Dormitory suburbs extend linearly up the Hutt Valley and towards Porirua and the Kapiti Coast. Everyone travelling from those places to downtown Wellington are funnelled through a single transport corridor running along the shoreline of the harbour:

Wellington density 01-13 v2

In Wellington, losing the rail line means pushing everyone onto a single road. (Unlike Rotterdam, cycling isn’t especially viable due to the lack of safe infrastructure on this route.) In other cities, there tend to be a greater range of alternative routes, which spreads around the traffic impacts.

Second, these results aren’t as robust as the Rotterdam study, due to their use of survey data rather than quantitative measures of traffic flow and speed. They’re not likely to be totally wrong, but it’s likely that people over- or under-estimated commute times, or that the survey wasn’t representative of all travellers (which could invalidate MoT’s extrapolation to all morning peak travellers).

However, the increasing availability of real-time data on traffic speeds from GPS devices means that the next time this happens, it will be possible to measure the impacts in much greater detail and with greater precision. The Rotterdam study offers some good methodological insight into how best to do that – it looks at transport outcomes at specific locations over a long period of time, and controls for seasonal and weekday effects that may influence transport outcomes.

Lastly, it would be really interesting to see some similar analysis done for Auckland. I’m sure that there have been a number of full or partial rail network outages, either due to bad weather or scheduled track upgrades. Perhaps it would be worth taking a look at congestion on those days.

Congestion Charging in Wellington

Auckland may be the most prominent voice when it comes to discussing congestion charging in New Zealand but it appears other cities are keen to join in. Last week it emerged that Wellington are also wanting to look at congestion charging however unlike Auckland where it is being talked about primarily as another revenue source, Wellington say they need it to deal with the after effects of building new motorways.

The call for a toll on Wellington’s CBD is growing louder, with studies revealing the central city could be flooded with almost 12,000 more cars once its proposed new motorways are up and running.

Wellington recently joined forces with Auckland to lobby the Government for the law changes necessary to introduce user-pays charges as a means of reducing car use.

Some of the ideas being floated include a congestion charge, such as the one used in central London, and fees that ramp up the cost of long-stay commuter parking.

While Auckland’s chronic traffic congestion is already apparent, Wellington’s is expected to get worse once the Kapiti expressway, Transmission Gully motorway, and Petone-Grenada highway are all built, making journeys in and out of the capital by road significantly easier.

Recent studies by Greater Wellington Regional Council show that, even with continuing investment in public transport, there are expected to be 11,500 more cars entering Wellington during the morning rush in 2031.

This has prompted the council to also model how that scenario would change, with various user-pays charges in place, despite some of them currently being beyond the law.

It found a congestion charge would have by far the greatest impact on car use in 2031. Vehicles trips in and out of Wellington’s CBD would drop by about four million annually, while public transport trips would increase by the same amount.

Tolling the yet-to-be-built Transmission Gully motorway and Petone-Grenada highway, which is currently legal, would have a more “moderate” impact on car use, as would a levy on all-day parking, which is not currently legal.

A couple of thoughts immediately spring to mind.

  • So far from the RoNS addressing congestion as the government/road builders/lobbyists so often love to claim, they’ll actually be making it worse by encouraging more people to drive, some of which comes from people encouraging people off using public transport and into their cars – making both systems less efficient. Why then are we wasting well over $2 billion on the new Wellington RoNS which already had poor economic outcomes.
  • From memory the NZTA have already ruled out tolling Transmission Gully as their modelling suggested that very few would use it if they did so. My guess is that would rule out any individual road specific tolls.
  • Like Auckland it appears that Wellington is blighted by politicians who seem to have the attitude of not caring what gets built as long as the government are spending money in their neck of the woods. There also seems to be the general attitude that public transport is only viable a mode of last resort.

The last point is displayed very well at the start of this interview on Radio NZ with Paul Swain, the chairman of the Regional Transport Committee who seemed aghast at the slight possibility of not building some new roads. It is also appears to be the attitude that is taken by Wellington City Councillor Andy Foster later in the piece who appeared quite annoyed that the Basin Reserve Flyover was cancelled – and as per this excellent op-ed from Dave Armstrong it appears both were quite keen on it.

Or listen here

I also found Foster’s comments on public transport interesting. He’s obviously correct that Wellington has the highest use of public transport in the country however I’m not sure I would go as far as him in saying that Wellington has a good system. There certainly seems to be a lot more that could be done to make the system better and therefore increase patronage. Many of those are things that Auckland has done or are on the agenda such as integrated ticketing and fares and a better bus network and greater bus priority. I kind of get the feeling that Wellington won’t really wake up and realise how far behind it’s falling until in a few years (at current rates) when Auckland passes them.

Coming back to congestion charging, I was also amused by this press release yesterday by the Property Council which claims that reducing cars in to Wellington would have catastrophic effects.

Wellington Branch president Mike Cole says slowing down traffic flow into the centre of a city of only 191,000 and a region of 471,000 people is ridiculous.

“They are talking about methods used in central London; a city of 8.6 million people. London could use the drop in traffic-flow, while we are desperate to get our city centre thriving by getting more people in.

“I think the Council is totally oblivious to the catastrophic effect this would have on retail and employment. Why would we drive people away, when we are working so hard to get them in?”

Perhaps someone needs to tell them that it’s people that buy stuff, not cars. Flooding the CBD with cars will only make it a less attractive place for people to be and therefore they will be less inclined to work and shop there.

Who knows what the outcome will be on congestion charging in either Auckland or Wellington but it’s certainly interesting that both cities are now starting to talk about it much more openly. As has long been the case my personal position is that any form of congestion charging should be designed at least initially to be revenue neutral – substituting rates or fuel taxes. That would give the public a greater level of comfort that it isn’t just a revenue gathering exercise but rather a traffic demand tool. I also think it is something that should be implemented in advance of another wave of road building so we can see the real impact it has before committing billions to construction.

How important is public transport for reducing congestion?

In July, I started taking a look at the economics of public transport fare policies. In the first part of the series, I took a look at how traffic congestion can be a rationale for public transport fare subsidies. (Parts 2 and 3 dealt with different issues.) I observed that:

In the absence of congestion pricing (and in the presence of other subsidies for driving, such as minimum parking requirements), higher public transport fares can result in a perverse outcome – additional congestion and delays for existing road drivers. This is shown in the following diagram:

PT fares and congestion diagram

Effectively, a failure to price roads efficiently means that we have to provide subsidies for public transport to prevent car commutes from being even more painful than they currently are.

But how much congestion reduction can we attribute to public transport? How much slower would car commutes be if some people weren’t travelling by PT instead of clogging up the roads? And how much is that worth to us?

It’s not possible to test this experimentally – we can’t exactly build a bunch of cities that are identical except for their PT systems and see what happens. (Transport research budgets are not nearly large enough.) However, we can observe the outcomes from various “natural experiments” that disrupt public transport systems while leaving everything else unchanged, such as natural disasters and public transport strikes.

Stu Donovan pointed me towards a recent research paper that analysed traffic speeds during public transport strikes in the Dutch city of Rotterdam. The authors, Martin Adler and Jos van Ommeren, use detailed traffic flow and speed data to model how 13 PT strikes that occurred from 2001 to 2011 affected traffic speeds. Because strikes prevent people from using PT without impeding road traffic, the outcomes observed during strikes give us some indication of what would happen to congestion in the absence of PT.

If you’re interested in knowing a bit more about the topic or the methodology, I highly recommend you read the paper. (It’s an excellent paper!) Here, I’d like to focus on a few key findings from the analysis.

First, the authors found that PT helps to speed up car journeys by reducing the number of people driving:

We demonstrate that during a citywide strike, car speed within the city decreases by about 10%. For highways, strikes exhibit a much smaller speed reduction of about 3%. During rush hours, the reduction in speed is more pronounced. These results imply that during rush hours, public transit provision reduces car travel time on inner city roads by about 0.2 minutes per kilometer travelled, whereas it reduces car travel time on highways by 0.02 minutes per kilometer. Hence, for cities such as Rotterdam, travelers on inner city roads benefit much more from public transit provision than highway travelers.

Intuitively, these results make sense. The benefits of PT for drivers are much higher in busier areas, such as Rotterdam’s inner city roads. However, Rotterdam’s ring road highways still derive some benefits.

The second interesting finding is that the popularity and ease of cycling in Rotterdam – even though it’s not exactly leading by Dutch standards – cushioned against some of the negative impacts of PT strikes:

a full-day citywide strike increases bicycle flow by 24% implying that a large share of travelers switch to bicycle use (rather than car use), which presumably reduces the car flow increase and therefore the speed reduction of a strike. Bicycle ownership and use is much higher in the Netherlands than in other countries in the world, so this result is likely specific to the Netherlands.

In other words, the availability of multiple congestion-free networks – public transport and cycling – meant that the roads didn’t have to accommodate all of the people who couldn’t get on the bus on strike days. In other words, the availability of multiple transport choices enhanced network resilience.

Third, the authors calculated the value of congestion reduction benefits attributable to public transport in Rotterdam. Based on some plausible assumptions about journey lengths and the value of time, they estimate that:

The annual public transit congestion relief benefit is then about €95 million (assuming 252 working days), so about €79 per inhabitant. This excludes any benefits of public transit provision on weekends that we assume to be negligible, so this is likely an underestimate. Given 721 million public transit passenger kilometers (OVPRO, 2014), the congestion reduction benefit per public transit kilometer is €0.13. This benefit is substantial given that the cost per public transit kilometer is €0.46.

In addition to congestion welfare losses there are rescheduling costs to car travelers. [Note: only 55% of the reduction in PT trips on strike days was balanced out by the increase in car and bicycle trips, meaning that a large share of people chose not to travel.] We do not include these costs, nor do we include the loss to public transit ticket holders or any other external cost of car driving that are likely an order of magnitude smaller than the effect through congestion.

The costs of providing public transit in Rotterdam are partially covered by subsidies, about €0.28 per public transit kilometer. So, the congestion relief benefit is about 47% of subsidies.

This is a really interesting finding! It puts a monetary figure on the congestion relief delivered by PT. (For Rotterdam, at least.) And, interestingly, it’s a large enough figure to justify a good proportion of PT fare subsidies. There are also other rationales for fare subsidies that I haven’t discussed here, such as social equity for people without cars and various types of network effects in PT provision.

But even if we leave those aside, this finding suggests that drivers should be happy to spend some fuel tax revenues to subsidise public transport.

What do you think about congestion and public transport?

Supply and demand and regional airfares

Back in June, Stuff published a report on regional airfares, focusing on the way that prices are affected by major events such as concerts and sports competitions. Now, I’m no airline economist, but I’ve got a general interest in transport pricing so I figured that it would be worth taking a look at the topic.

The point of the article seems to be that airplane tickets are higher during periods of high demand. That doesn’t seem too weird, but this guy in Nelson is absolutely ropeable at the thought:

Nelson man Steffan Eden is furious about Air NZ’s fares from Nelson to Auckland and return for the weekend of March  5 and 6 when Madonna will give her first New Zealand concert at the Vector Arena.

Fares that the previous weekend cost $79 are twice that at $159 on the weekend of the concert, an $89 fare rises to $169; and a $129 fare becomes $209…

“Look at the fares the weekends before and after the concert, they’re normal fares. Then on the concert weekend they’re virtually double.  It’s quite blatant.”

Eden said the same thing happened when he wanted to go to the Cricket World Cup match between New Zealand and England on February 20.  “I wanted to take my kids but didn’t in the end because of the cost,” he said.

The man quoted in the article seems to argue that these jumps up in fares are due to uncompetitive or discriminatory practices by Air NZ. By contrast, the airline says that the price increases are just due to cheaper tickets selling out faster:

An Air New Zealand statement said it has been experiencing high demand for flights into and out of Nelson that weekend due to both the New Zealand Masters Hockey Tournament which is being held in Nelson from February 28 to March 5 and the Madonna concert in Auckland.

“As you will appreciate, where there are major events on flights tend to sell out well in advance, with the cheaper fares selling out the fastest, so booking as early as possible is recommended.”

Now, as an economist I’m always wary of the potential for companies with few immediate competitors to exercise market power over their customers. But in this particular case, I don’t think that’s happening. What we are seeing is the normal, and in fact beneficial, working of supply and demand.

Let’s start with the supply side. Air NZ doesn’t have an infinite budget for airplanes and staff. It faces constraints. If it wanted to run more services between Nelson and Auckland on particular weekends of high demand, it would have to either:

  • Pull airplanes off other regional routes, which would potentially satisfy Nelson’s demand but would in turn lead to similar stories about how unfair Air NZ was being to Napier or Timaru or what-have-you, or
  • Buy extra airplanes and hire extra staff that would sit idle most of the time and fly only during a few periods of exceptionally high demand. This is superficially appealing, but it would mean an across-the-board increase in fares to pay for a bunch of empty planes.
mojave desert boneyard1

This isn’t really related, but it’s an interesting picture (Source)

So that’s the supply side. What about demand?

Air NZ has observed, correctly, that demand for flights is not constant over time. Simply put, more people want to fly at some time periods than during others. Airlines can respond to this in a few different ways. The first would be to keep prices constant, regardless of demand. This would turn air travel into a first-come-first served game, which is great if you always buy tickets months in advance but terrible if you have to take a last-minute trip for work or a medical emergency.

The second approach, which Air NZ may be using, is to charge higher prices during periods of higher demand. This may seem less fair, but it’s actually better for (almost) everyone. It means that airlines aren’t constantly booking out flights well in advance or misallocating resources in a futile attempt to give everyone a cheap flight. Travellers also benefit – they get a choice between paying more to travel at their preferred time or finding a cheaper fare at an off-peak time.

I fly for work on a semi-regular basis so I’ve noticed some of the patterns over time. Between 4-6pm, departure gates fill up with suit-wearing men and women headed home from their meetings in time for dinner. Not surprisingly, prices are highest at this time. Later on, prices drop, planes get a bit emptier, and the suits get replaced with casual clothes. By the end of the night, most of the people who want to get home have gotten there, and for a price that they’re willing to pay.

Occasionally, this means that somebody decides not to go to a Madonna concert. But that’s not a flaw with supply and demand – that’s how it’s supposed to work! If the man quoted in the Stuff article didn’t go, it’s only because someone who valued being there more bought the ticket instead.

Finally, I have to ask: Why are people outraged when the principles of supply and demand are applied to airfares? Perhaps it’s because we routinely ignore those principles everywhere else in our transport system.

As numerous economists have observed, we manage our roads like a Soviet supermarket. The price to use roads is set at a single, low value – i.e. NZ’s comparatively low petrol taxes – and thus people queue up for ages to drive on them every morning and evening. The same thing happens with parking, where we have regulated to make it abundant and free and ended up in a situation where people can never get enough parking.

In economic terms, there is no difference between this:

SH16 traffic

And this:


They are both situations in which scarce resources, including people’s time, are misallocated due to poorly-functioning price signals. So rather than asking “why don’t we price air travel as inefficiently as roads?”, we should ask “why don’t we price roads as efficiently as we price airfares?”

A failure to price roads efficiently badly distorts our supply decisions. We are forever pouring more asphalt and concrete that accommodates a few more slowly-moving cars at peak times and sits idle much of the rest of the time. By contrast, congestion pricing would allow us to avoid many of these expenditures by giving people an incentive to travel differently.

What do you think about airfares – and transport pricing in general?

The economics of fare policies, part 1

A few months back, Auckland Transport put out its new fare policy for consultation. The draft policy, which they call Simplified Fares, has two main elements:

  • Standardised fare zones that ensure that journeys within or between zones cost the same regardless of whether you’re travelling by bus or rail [ferries are excluded]
  • No transfer penalties between services, which is a key element in enabling a frequent connective network.

Those are indeed simple principles, but developing and implementing a fare policy is seldom simple. So the whole thing got me thinking: Why do public transport fares work the way they do? And could we do things differently?

As I’m curious, I figured that I should take a quick look at the economics of fare policies. Part one of the series looks at the biggest-picture question: Why do we subsidise public transport?

First, some background. In most developed-world cities, public transport systems are subsidised by taxpayers. Users pay some of the operating costs – ranging from as low as 10% to as high as 80% – but seldom all. In New Zealand, the national farebox recovery policy requires all regional transport agencies to cover 50% of their public transport costs from fares. However, data from the Ministry of Transport suggests that some agencies are closer than others to this target:

MoT farebox recovery rates 2013-14

Is 50% the right number for all regions? I don’t know – and the answer depends in part on what other goals we’re trying to accomplish with public transport pricing. But it’s clear that some level of subsidy must be provided in order for the entire transport system to work efficiently.

To see why, we need to take a look at what economists call “second-best pricing”. According to Wikipedia, it can be desirable to impose a subsidy to “offset” for an uncorrected market failure elsewhere:

In an economy with some uncorrectable market failure in one sector, actions to correct market failures in another related sector with the intent of increasing economic efficiency may actually decrease overall economic efficiency. In theory, at least, it may be better to let two market imperfections cancel each other out rather than making an effort to fix either one.

In transport, we have a situation where people have multiple options for getting around. They can drive, take the bus (or train), cycle, etc. In this situation, a price change in one market – say, a fare increase for public transport – can encourage people to switch to another mode instead of paying more.

As I argued in a recent post on congestion pricing, road space is usually not priced “efficiently”. All road users pay fuel taxes or road user charges based on the total number of kilometres driven or litres of petrol used. But they don’t pay more to drive on busy roads, where they impose delays on other drivers. As this diagram from a 2012 UK study on the external costs of driving shows, the last 10-20% of car trips impose significant costs on society.


Public transport can play a useful role in smoothing off the big spike at the right hand side of that chart, by providing a more space-efficient option for travelling on popular, congested routes. Another way of saying that is that in the absence of congestion pricing (and in the presence of other subsidies for driving, such as minimum parking requirements), higher public transport fares can result in a perverse outcome – additional congestion and delays for existing road drivers. This is shown in the following diagram:

PT fares and congestion diagram

Effectively, a failure to price roads efficiently means that we have to provide subsidies for public transport to prevent car commutes from being even more painful than they currently are. Public transport subsidies are, in that sense, subsidies for drivers. By making your neighbor’s bus fare cheaper, they in turn make your drive to work a bit easier.

Finally, it’s worth considering how we got into this situation. 80 or 100 years ago, public transport systems tended to cover their operating costs with fares. For example, Auckland’s tram system was profitable, if in need of maintenance and refurbishment, up until its removal in the mid-1950s. (Mees ref?) This changed, in large part, due to the introduction of subsidised motorways.

This article by Joseph Stomberg at Vox describes how the US interstate highway system was developed in the 1950s as an explicitly subsidised – i.e. not tolled – transport mode:

The first step was changing how roads were funded. In the 1930s, there were already privately owned toll roads in the East, and some public toll highways, like the Pennsylvania Turnpike, were under construction. But auto groups recognized that funding public roads through taxes on gasoline would allow highways to expand much more quickly.

They also decided to call these roads “free roads,” a term that was later replaced by “freeways.” Norton argues that this naming shift was essential in persuading the federal government — and the public — to shift away from tolls. “It started with calling the roads drivers pay for ‘toll roads,’ and calling the ones that taxpayers pay for ‘free roads,'” he says. “Of course, there’s no such thing as a free road.”

In other words, the “original sin” of transport subsidies was the construction of non-tolled highways paid for out of general tax revenues. This choice led in turn to a situation in which we must adopt “second best pricing” in public transport, and offer an offsetting subsidy. I’m not necessarily opposed to this… but it does mean that I am skeptical to complaints that buses and trains are subsidised.

What do you think we should do about public transport pricing?

If congestion is so bad, we should price it

Last Thursday, the Government shut the door on the idea of road pricing for Auckland, saying that it would prefer to undertake “a year-long negotiation with the council on an agreed 30-year programme focusing on reducing congestion, and boosting public transport where that reduces congestion.”

The following day, the road/infrastructure lobby undertook a bit of a media blitz pushing for more construction. As part of that, we got sent this press release from the Auckland Chamber of Commerce:

Media Release

12 June 2015

Auckland – defined by congestion

The Auckland Chamber of Commerce strongly supports the initiative of Government to seek a negotiation with Auckland Council on an agreed 30-year programme focusing on reducing congestion, and boosting public transport where that reduces congestion.

Michael Barnett, head of the Auckland Chamber was responding to news reports that Transport Minister Simon Bridges and Finance Minister Bill English have sent Auckland Mayor Len Brown a letter proposing a negotiation and ruling out allowing Auckland to bring in motorway charges to help fund transport projects.

“The Auckland business community overwhelmingly agrees that immediate action to address the City’s transport congestion is required,” said Mr Barnett.

In short, congestion is bad. Really bad. It’s a crisis deserving immediate action… in the form of a year-long talk-fest between local and central government.

Of course, it’s difficult to find reliable empirical evidence that Auckland’s congestion levels really are that bad. Average commute times are a cruisy 25 minutes – well below many other cities. NZTA research has found that the actual cost of congestion is neither (a) largely a monetary cost for businesses or (b) anywhere as large as people claim. While people like to claim that congestion costs “billions” annually, a more realistic figure is $250 million. The one source that does claim that Auckland has world-beating congestion, the TomTom index, has serious methodological flaws.

Nevertheless. Even though its empirical basis is shaky, the Auckland Chamber of Commerce’s recommendations for projects are not crazy. In fact, they seem to be on Auckland Transport investment radar already:

A good outcome from Government and Auckland Council working together would be a package of fast-tracked projects aimed at:

  • Improving public transport services’ reliability and frequency
  • Getting as much use as possible out of the transportation system we have
  • Removing parking from major arterial routes to create more usable road space.
  • More high occupancy lanes to encourage a reduction of sole occupancy cars.
  • Strengthened integrated traffic management covering arterials and motorways.
  • Expanding park and ride facilities at main trunk rail and busway stations.

But even if the ideas are sensible, “fast-tracking” them will be expensive. We simply can’t build everything at once. Even if Government was willing to give Auckland Council more tools to raise revenue – which is unlikely given its refusal to consider road tolls – capacity constraints in the civil engineering business would make it hard to do much more.

To its credit, the Chamber seems to recognise this and agree that we need to prioritise use of our scarce resources:

“Good leadership is about partnership,” said Mr Barnett. “It is about understanding that we have limited resources, so we must learn to prioritise correctly,” he concluded.

Which leads me to my point. If congestion is such a big problem, why don’t we use congestion pricing to make sure that we’re prioritising use of our road network efficiently?

I find it very strange that business groups aren’t more enthusiastic about this idea. If congestion is really as bad as they say it is, why aren’t they loudly advocating a policy solution that would actually address it? (Road-building doesn’t work.) Surely freight companies and construction firms would benefit from the resulting reductions in traffic, even if they had to pay a bit for them.

In my experience, congestion pricing is one of those ideas that virtually all economists agree on. It’s like free trade in that regard – there might be some disagreement about the fine details, but most agree that it’s a good idea. But it hasn’t gotten as much attention in other quarters.

So here, for example, is William Vickrey, who won the Nobel Memorial Prize in Economics for his pioneering work on the topic:

Known among economists as “the father of congestion pricing,” Professor Vickrey sees time-of-day pricing as a classic application of market forces to balance supply and demand. Those who are able can shift their schedules to cheaper hours, reducing congestion, air pollution and energy use — and increasing use of roads or other utilities. “You’re not reducing traffic flow, you’re increasing it, because traffic is spread more evenly over time,” he has said. “Even some proponents of congestion pricing don’t understand that.”

He has admitted that his ideas have sometimes not been well received by those who set public policy because, “People see it as a tax increase, which I think is a gut reaction. When motorists’ time is considered, it’s really a savings.”

And here’s urban economist Edward Glaeser commenting that more megaprojects aren’t the best fix for transport issues:

Infrastructure investment only makes sense when there is a clear problem that needs solving and when benefits exceed costs. U.S. transportation does have problems — traffic delays in airports and on city streets, decaying older structures, excessive dependence on imported oil — but none of these challenges requires the heroics of a 21st century Erie Canal. Instead, they need smart, incremental changes that will demonstrate more wisdom than brute strength…

IMPLEMENT CONGESTION PRICING: We should expect drivers to pay for more than just the physical costs of their travel. We should also expect them to pay for the congestion that they impose on other road users. If you have a scarce commodity, whether groceries or roads, and you insist on charging prices below market rates, the result will be long lines and stock outs, like those that bedeviled the Soviet Union decades ago. Yet U.S. roads are still running a Soviet-style transport policy, where we charge too little for valuable city streets. Traffic congestion is the urban equivalent of a stock out.

And here’s economist Matthew Turner, who co-authored one of the most comprehensive studies of “induced traffic”, which I discussed here:

So what can be done about all this? How could we actually reduce traffic congestion? Turner explained that the way we use roads right now is a bit like the Soviet Union’s method of distributing bread. Under the communist government, goods were given equally to all, with a central authority setting the price for each commodity. Because that price was often far less than what people were willing to pay for that good, comrades would rush to purchase it, forming lines around the block.

The U.S. government is also in the business of providing people with a good they really want: roads. And just like the old Soviets, Uncle Sam is giving this commodity away for next to nothing. Is the solution then to privatize all roads? Not unless you’re living in some libertarian fantasyland. What Turner and Duranton (and many others who’d like to see more rational transportation policy) actually advocate is known as congestion pricing.

And here’s the OECD in its latest country report on New Zealand:

A just-released OECD economic survey blames years of under-investment in infrastructure for the city’s roading problems. It calls for a mix of tolls and congestion charges to alleviate peak-hour traffic pressure and help fund new roads and more public transport.

“Placing a cost on travel during peak periods could incentivise drivers to travel at different times (off-peak), if they are not required to be on the roads, or could encourage more carpooling and use of public transportation,” the report says.

In short, if you’re worried about congestion, you need to take congestion pricing seriously. There are undoubtedly reasons why we may not want to implement congestion pricing, ranging from technical feasibility to equity concerns. But in my view it’s ridiculous for business groups and politicians to get all up in arms about the issue – and promptly rule out one of the few realistic solutions.

What do you think about congestion pricing?

We need to stop lying to ourselves about congestion

Without getting back on the topic of pohutukawas or St Luke’s Road again, I did notice something funny in the statement that Greg Edmonds, Auckland Transport’s Chief Operating Officer, made in Metro Magazine in response to the issue:

The founding premise of the Auckland super city was that the city’s congestion was costing $1 billion a year in lost productivity and this had to change.


Auckland Transport (AT) was created to solve the congestion problem…

Some people might think that this is a slightly too narrow view of Auckland Transport’s mandate. Whatever. Fair enough.

However, there is actually a much more serious problem with Mr Edmonds’ comments. Simply put: the notion that we can “solve the congestion problem” is not at all realistic. (Unless we are willing to try out road pricing, which is unlikely given the tepid response to the last few studies of the issue.)

I don’t want to pick on Mr Edmonds in particular. It’s common to hear politicians, bureaucrats, and advocates from all over say similar things. We constantly hear that Project X or Project Y will “fix congestion” or “solve gridlock” or “save us [some unthinkably large amount of money] in congestion costs”.

As an economist, I’m baffled by these statements. The empirical evidence on congestion overwhelmingly shows that it is not possible to reduce it by building more roads. This is because people change their behaviour in response to bigger roads. They shift from walking to the store to driving there; they buy a house further out of town; they travel at different times.

Here’s what two North American economists, Duranton and Turner, had to say on the topic after undertaking a comprehensive, multi-decade study on induced traffic in US cities:

Our data suggests a ‘fundamental law of road congestion’ where the extension of most major roads is met with a proportional increase in traffic. Not only do we provide direct evidence for this law, but also show find evidence that three implications of this law; near flat demand curve for VKT, convergence of traffic levels, and no effect of public transit on traffic levels.

All earlier studies, such as this comprehensive 1998 study of 70 US metro areas over a 15 year period cited by walkable city advocate Jeff Speck, have come up with identical findings:

Metro areas that invested heavily in road capacity expansion fared no better in easing congestion than metro areas that did not. Trends in congestion show that areas that exhibited greater growth in lane capacity spent roughly $22 billion more on road construction than those that didn’t, yet ended up with slightly higher congestion costs per person, wasted fuel, and travel delay.

Consequently, all we can realistically do about congestion is to give people good alternatives to participating in it. Other modes, such as grade-separated rapid transit and walking and cycling, do not get congested in the same way as roads do. While the research shows that providing alternatives to driving does not necessarily reduce road congestion, it does give people a way to reduce their exposure to it.

In light of these fundamental economic realities, it is essential that transport agencies stop talking about “fixing congestion”. This is nothing more than a dangerous fantasy.

Suggesting that we can solve congestion creates unrealistic hopes among the public. Every time a politician or transport agency opens a new road and promises that it will reduce congestion or speed up people’s journeys, they are feeding expectations that can never fully be met.

The result of this is that transport agencies are constantly dealing with demands for more roads that will not actually deliver long-term solutions to the problem of congestion. This sets the transport profession up to constantly fail to satisfy people’s desires and demands. This has to be a tremendously disheartening situation to be in.

My personal view is that instead of talking about “fixing congestion”, transport agencies should instead promise to deliver outcomes that are actually achievable.

This could include, for example, committing to deliver transport choice to underserved areas of the city by investing in rapid transit infrastructure, frequent bus services, and safe walking and cycling infrastructure. While transport agencies would have to work hard to deliver on all this, they could expect that the end result would be more transport choice for residents.

Transport agencies could even commit to some traditionally roads-centric goals, like, say, building new roads to enable the development of a new subdivision at the edge of the city. At least, as long as they weren’t making unrealistic promises of fast, frictionless commutes to the future residents…

Driverless cars to increase congestion

Proponents of driverless cars often suggest the technology will make all sorts of significant changes to transport. Gone will be car ownership with people just hiring a car when they need one, like taxi’s only easier and cheaper. As such they say gone too will be the need for public transport, especially when you take away some of the benefits PT currently has like being able to do other things such as work, read, use a phone or even sleep. Further congestion will also be a thing of the past with these smart vehicles able to better work together rather than the randomness of humans. Of course the biggest and likely most accurate prediction will be safety as for a start these cars will obey the road rules so no speeding, no running red lights or any of the other bad habits human drivers have.

That all sounds wonderful however an article from CityLab highlights research showing that at least for some time driverless cars could actually make things worse on the roads.

A new simulation-based study of driverless cars questions how well these two big secondary benefits—less traffic and more comfort—can coexist. Trains are conducive to productivity in large part because they aren’t as jerky as cars. But if driverless cars mimic the acceleration and deceleration of trains, speeding up and slowing down more smoothly for the rider’s sake, they might sacrifice much of their ability to relieve traffic in the process.

“Acceleration has big impacts on congestion at intersections because it describes how quickly a vehicle begins to move,” Scott Le Vine of Imperial College London, who led the research, tells CityLab via email. “Think about being stuck behind an 18-wheeler when the light turns green. It accelerates very slowly, which means that you’re delayed much more than if you were behind a car that accelerated quickly.”

For their study, Le Vine and colleagues simulated traffic at a basic four-way urban intersection where 25 percent of the vehicles were driverless and the rest were standard. In some scenarios, the driverless cars accelerated and decelerated the way that light rail trains do—more comfortable than, say, riding in a taxi, but still a little jerky at times. In other scenarios, the cars started and stopped with the premium smoothness of high-speed rail.

Within these broad scenarios the researchers also tested alternatives that reduced speeds but improved smoothness, such as longer yellow lights or following distances. All told they modeled 16 scenarios against a baseline with all human-driven cars. The researchers then ran each simulation for an hour, repeated it 100 times, and calculated the average impact that scenario had in terms of traffic delay and road capacity.

In every single test scenario, driverless cars designed to create a comfortable, rail-style ride made congestion worse than it would have been in a baseline scenario with people behind every wheel.

So cars with fast acceleration and deceleration are obviously easy to make but that’s not what people are likely to want if you’re also trying to do some of the other activities mentioned earlier. Regardless traffic generally moves at the pace of the slowest vehicles so all it takes is one slow driver or driverless vehicle and many others will be slowed down too. I bet they won’t say that in the marketing brochures.

The interior of the Mercedes-Benz driverless car concept


I suspect this isn’t the only aspect of driverless cars that could create congestion. As an example the driverless taxi model that most people say will happen, is likely to result in a lot more vehicle movements as cars reposition themselves to pick up additional fares. That means that where roads are generally congested in one direction only, with driverless cars congestion could occur in both directions.