As we start a new year, I thought it was a good time to review the past, present, and future of rail patronage in Auckland. Before diving into the data, I want to relay two relevant personal anecdotes.
First, my interest in Auckland’s rail patronage began in the early 2000s, when — as a mildly bored and nerdy young person growing up in rural south Auckland — I would download and analyse the latest monthly rail patronage statistics. I credit this experience with kickstarting my interest and career in transport more generally. So thanks to all the public servants out there who work hard to make this sort of information available, wherever and whoever you are.
Second, I was sufficiently enthused by the data to write to various politicians to express support for investment in Auckland’s rail network. I recall receiving a reply from Maurice Williamson in which he argued trains would go the way of the “slide rule”, and that in 20 years we’d be flying and driving everywhere. While Maurice’s opinion differed from mine, I was thrilled he’d taken the time to reply. So thanks also to all the politicians who take the time to engage with nerdy teenagers.
Now for some data. To set the scene for what follows, let’s first consider one of the founding documents of rail investment in Auckland, namely ARTA’s Rail Development Plan from 2006 (RDP), which included the following graph:
These patronage predictions were generated from the “Auckland Passenger Transport (APT) model”, which uses data on future land use and transport infrastructure/services to predict future transport choices and outcomes. The APT model predicted Auckland’s rail network would generate reach 16 million trips by 2016, which — as it turns out — was almost bang-on the money.
On the other hand, two things stand out from the above figure. First, from 2008 until 2015 actual patronage was below predicted patronage. The reason for the divergence becomes clear when you read the RDP, which made the following assumptions on implementation timelines for the core network upgrade (pg. 19).
Clearly, the RDP assumed upgrades would be implemented faster than what was achieved. With regards to electrification, page 9 of the RDP assumes there would be “28 trains in service by 2009 and 35 trains in service by 2011.” In reality, the first electric service did not operate until 28 April 2014, approximately five years later than planned. Delays in implementation seem likely to explain why actual rail patronage was less than predicted prior to 2015.
The second thing to stand out from the previous graph is that — in the two years since 2016 — rail patronage soared past 16 million trips, rather than flattening off as predicted by the APT model. The exact reasons for this discrepancy are unclear, but it may imply the original APT forecasts were on the low side. This is especially likely when you consider that Auckland’s electric trains are operating much slower than originally envisaged. Post-electrification rail services were expected to run from Papakura to Britomart in approximately 40 minutes, which is almost 15 minutes faster than what they currently achieve.
Why might actual rail patronage have grown more than originally predicted by the APT model? Some potential external explanations include:
- Faster growth (population, visitors, economic activity) than originally expected, especially in areas that are proximate to the rail network.
- Complementary transport investment and policy changes, such as city centre improvements (shared streets) and changes to parking policy.
- Unforeseen technological innovations, such as the increased prevalence of GPS-enabled smartphones, which have made it easier to use PT.
- Changing preferences, such as Millenials choosing to drive less and use non-car modes more.
You may be able to think of other factors that explain why growth in rail patronage has outstripped the original predictions. Aside from these external factors, it’s also possible that the parameters in the APT model were too insensitive to improvements in rail services. From the outside, there’s no way for us to know which of these explanations is most plausible. What we can say with some confidence, I think, is that the original APT forecasts for Auckland’s rail patronage appear to have been too low (in the order of 25-35%, if you compare 16 million to the current patronage of 20.3 million).
Let’s now predict future patronage using a simple model that is similar to those I first applied as a teenager. I assume that prior to the opening of the CRL, patronage growth declines (as per recent trends) in relative terms to settle at 3.5% and 7% p.a. in low and high scenarios, respectively. Based on these assumptions, Auckland’s rail patronage would sit between 24.5 to 28 million trips p.a. by 2023, as illustrated in the figure below.
This range of outcomes seems plausible given currently committed improvements, such as the New Network, electrification to Pukekohe, and ongoing rail service enhancements (e.g. improvements to frequency and span, especially following delivery of the next tranche of EMUs). These improvements seem likely to keep Auckland’s rail patronage growing faster than population growth, albeit at a slower clip than we’ve seen in the last 5-10 years following electrification and HOP.
And what comes after 2023? Well, the CRL is predicted to add around 20 million trips p.a. Let’s assume the actual impact is somewhere between 15 – 25 million trips p.a., and that all of this growth occurs by 2030. If we add this patronage to my pre-CRL predictions, then we end up with 40 to 55 million rail trips p.a. by 2030. If we assume an additional linear growth rate of 2% p.a. in the period from 2023 to 2030 (independently of the CRL), then the extra 5 million trips p.a. takes us to 45 to 60 million trips p.a. If I had to pick a mid-point, I’d suggest 50 million rail trips p.a. by 2030 is reasonable.
This discussion highlights an important issue: Statistical uncertainty. One of the problems with current transport modelling practices is that they often do not communicate the level of uncertainty associated with our predictions. In reality, strategic transport models are based on inputs and assumptions, all of which introduce considerable uncertainty.
This uncertainty, however, tends to be swept under the rug for two reasons. First, senior executives and elected representatives are often risk-averse and statistically illiterate; they simply don’t like uncertainty. Second, accepting predictions of the model helps people to avoid taking responsibility for their decisions; “the model made us do it” line-of-defence.
In contrast, once you start to understand how much uncertainty is associated with the inputs and assumptions on which strategic transport models are based, then you tend to place less emphasis on the models’ predictions, and greater emphasis on strategic considerations, notably up-side and down-side risks to those predictions.
Contrast transport planning practices with macroeconomics, where uncertainty is presented explicitly, as illustrated below.
Unlike macroeconomics, I think the transport profession has collectively failed to find ways to communicate the statistical uncertainty associated with our quantitative predictions. While sensitivity testing is sometimes undertaken, this testing doesn’t usually communicate how much uncertainty is expected. For example, testing the sensitivity of benefit-cost ratios to variations in demand doesn’t tell you anything about the degree to which we expect demand to vary. Based on the standard errors attached to the economic parameters used in strategic transport models, I’d be surprised if 70% confidence intervals more than 10 years into the future were within +/- 30% of the mid-point estimate. The uncertainty involved really is that large.
Returning to the question of rail patronage data, some may argue that my predictions are “pessimistic”. While I’d be happy to be wrong, I suspect it’s unlikely. The reason is that the CRL business case predicted an additional 20 million trips by 2040, which I have brought forward to 2030 simply because (1) most of the uplift in patronage can be expected to occur within the first 5-10 years and (2) the APT model seems to have been conservative in the past. So I’ve arguably already been optimistic with regards to the effects of the CRL.
Rather than haggling over what might happen, I think the more interesting question is what can we do to push Auckland’s rail patronage towards the upper-end of the range of possibilities? This links to Patrick’s recent post, which considered possible directions for Auckland over the next ten years. Personally, I think Auckland would do well to focus on continuous improvement across a wide range of areas. Rather than looking for “one hit wonders” (beyond the CRL), I prefer an “integrated but diversified” strategy of improvements. By this I mean:
- Invest in integrated transport/land use outcomes — especially in central areas and around major transport interchanges, such as Panmure, New Lynn, Otahuhu, and Manukau. Pedestrian access to rail stations is a strategic priority, especially in the city centre where we need to cater for the fountain of pedestrians travelling to/from Aotea and K Road stations in the post-CRL period. Initiatives like the Linear Park and Wellesley Street are key.
- Review operational practices to align with strategic outcomes — Post Unitary Plan, I would like to see an increased focus on aligning operational practices with strategic outcomes. Everything we do to make the areas around rail stations more pleasant, especially for those who use non-car transport modes, is likely to contribute to rail patronage. That includes measures like reducing vehicle speeds and improved street-lighting.
- Run trains faster, more frequently, and for longer span — while Auckland’s rail renaissance has been a huge success, there remains much room for improvement. Sorting Auckland’s painfully long dwell-times, while continuing to invest in frequency (perhaps even the 15 minute all-day frequency required to qualify for rapid service) and more span is likely to support ongoing growth. No shittake Sherlock.
- Turbo-charge Auckland’s post-CRL bus network — while the New Network is exciting, the opening of the CRL presents additional opportunities to improve the frequency, simplicity, and efficiency of Auckland’s bus network. We should start formulating plans now to cut direct peak-only services to the city centre, especially from the west. This will free up bus capacity in the city centre, and generate savings to be reinvested in higher frequencies elsewhere.
If we can implement a broad platform of complementary initiatives, then I suspect the experience of the last ten years may well be repeated again. That is, we may achieve less in the first 5 years than we want, while potentially achieving more over 10 years than we ever thought possible. What do you think?