The power of data, redux
There is a need to relook some observations.
As it turns out, the LTA has been giving us quite a bit of bad data. Yes, this is a Bad Thing. Because it means that several posts I made may have to have their conclusions slightly tweaked. And there are some truths, which the blog previously held self-evident, but which may now have to be questioned.
I’m not sure whether LTA staffers in a position to do anything are reading this blog, but I applaud them for realizing the error and fixing it. I wouldn’t have known it was an issue myself, if not for the heads up.
How it all came about
I returned from a lunch break to see this email from the LTA’s Datamall team:
Welp, that’s what we’ve been using all along to do things like evaluate the efficacy of the TEL, and perhaps also ask the question of just where our capacity on the 3-car lines are. And that they saw the need to go back and update previously released datasets shows that this was a pretty grave issue.
The issue seems to be with SimplyGo — technically named Account Based Ticketing. “CEPAS token card” may be the technical name behind the upgraded EZ-Link cards. In short, from my analysis of the data, it appears that SimplyGo users were SimplyGone (pardon the pun) from the data I was using all along. What may have looked like declines in ridership may be explained by people converting to use account-based ticketing mechanisms.
So of course, once the data was released, I had to go back and update all my findings with the new datasets. And of course I also have to update my assertions, because, unfortunately, it means the blog has been using faulty data. That said, spoilers ahead, it’s not all roses.
The medium-capacity lines
Even with the new data, things still look quite grim for the medium-capacity lines. Total entries at TEL non-interchange stations still remain below the 100k mark, and in July 2023, even despite RP still being in session, only TE6 Mayflower managed to pass the 10k mark in entries.
My earlier blog post suggested that this could be transient, as return-to-office expands and more need to use the MRT network. I’m not inclined to think that’s the case now, especially with the LTA’s supposed claim that only 160k daily passengers use the TEL, and even after accounting for growth since then.
In fact, Springleaf and Lentor doing so much more poorly than even DTL stations (see the next section for details) should give the LTA some pause. There are opportunities that should be taken to improve TEL bus connections, but they are not happening, and this also reflects in total ridership.
A similar story has played out on the DTL. Here, it appears that DT17 Downtown has replaced DT33 Tampines East as the DTL’s busiest non-interchange station, with the fixed data showing a 2.5x increase in passengers using the station compared to the broken data. These are professionals, they need to travel.
For DT17 Downtown station to receive such high ridership — only one-third of what Raffles Place gets, considering the much larger and much more frequent trains at Raffles Place— could very well vindicate my point that many DTL riders are in fact riding within this core section, whether to commute or to spice up lunch breaks. All the more since the walking experience to Marina Bay station is poorly designed. These statistics also reflects how the DTL is so affected by return-to-work, if the two CBD stations account for so much of the DTL’s passenger traffic.
As I suspected earlier, this also effectively means a large portion of MBFC has to take a DTL ride out before they can get to their MRT line, considering the poor walkability to Marina Bay station. Consequently, despite having three MRT lines, the ability to get a seat on all of them, and the opportunity to bypass Newton, Marina Bay is doing much more poorly than even Fernvale LRT; even Shenton Way is busier despite only TEL services available there.
Ah, the power of development, with Asia Square and Marina One next door.
Two steps forward, three steps back
I can now confidently say that building the TEL has not significantly helped increase public transport usage. Even bus ridership has stayed near the 4 million mark, a drop from 2019 levels even. But should I even be expecting any better considering how static the network has been?
A major question here is whether the SimplyGo rollout was already giving us bad data in 2019. Mastercard users were enabled in April 2019; Visa in June 2019; and by November, when NETS cards were enabled, 350k public transport trips were already made with SimplyGo. That can easily be 350k trips SimplyGone and we don’t know how much of them apply to each transport mode.
A quick sanity check against other 2019 data released by the LTA and more recent data from the public transport operators points to a majority of these “lost trips” being MRT trips. On average, each MRT line has only recovered to around 91–92% of their reported 2019 patronage numbers. The greater drop in NSEWL numbers compared to other lines could be partly due to the TEL.
How can this be? Some companies, as well as the civil service, have embraced flexi-work, be it fully remote working or a hybrid model, and this results in people staying home some days and not using public transport. Similarly, the increased accessibility of car-sharing and private hire services in theory can also mean less public transport users.
This will naturally mean public transport usage drops. Contemporary logic means that the MRT bears the brunt of these drops; considering that bus users also include students going to school, as well as senior citizens, housewives, and others working from home or not in the labour force travelling within their neighbourhoods. Only three towns have LRT, after all, so bus service is still a key part of public transport.
What does this tell me? This tells me that the 30km of TEL opened since then may not have had much of an impact in creating new public transport trips. People who own cars and make use of them well would drive anyway. So what may have happened is that people who don’t own cars, and who would have had to take a bus to an NSL station, now can walk to the TEL and save a bus ride. This may barely be enough to outweigh the lost ridership from flexi-work, remote work, and other cultural changes.
It will only get worse especially in 2024 after TEL Stage 4 opens along the heavily bus-dependent Marine Parade corridor, where demographics may point to less commuting demand compared to what was expected when construction started in 2015.
So, despite all these backgrounds, how else can we explain the difference in recovery between transport modes?
Fight the right battles
I can only think of one logical explanation — that there is competition between buses and trains for the same (smaller) pool of public transport users. The data seems to say that both are losers at the end of the day. Perhaps, then, the answer is to not fight this battle in the first place.
The expansion of MRT service to previously bus-reliant or bus transfer-reliant areas means that people ride less buses. Where someone living in Ang Mo Kio West going to Buona Vista might have had to first take a bus to Ang Mo Kio MRT before taking a train, they now have a TEL option with a change at Caldecott station; the bus ride falls off the statistics but not the MRT ride. Likewise, commuting patterns at Woodlands North are worth examining — are RP students walking to the TEL anyway when 902 isn’t running?
This should inform the kind of investments we should be making. Of course, you have cases like SBST reducing DTL peak service levels by around 12% between 2018 and now, despite the commensurate drops in ridership on the way to recovery. This should probably be one of the first things to be fixed. In this light, the data also seems to show stronger midday ridership in the weekends as well, so perhaps better midday service in the weekends can also be run.
But there is also a lot of investment, both in capital infrastructure and operations, that needs to be made, to improve bus service outside the Central Area. It is understandable why some people get visceral reactions when asked to take a feeder bus to the MRT, since bus infrastructure in the suburbs are not necessarily great. Frequencies too have taken a hit; while some may be deserved following reduction in passengers after new MRT openings, can we really say all of them are justified?
After all, more service reductions than improvements have happened in the four intervening years, which means people are losing direct service and being asked to change instead. If they are changing between bus services, the nature of the dataset in counting card validations should actually show an increase in ridership — or at the very least, show parity — but that is not happening. That we may also be losing MRT riders despite aggressive expansion of the network also does not bode well for our public transport future. Of course, better service can attract some back.
Yet, instead of making these investments in the rest of the network, the LTA has continued to spend its service hours on commuter-oriented CBD express and expressway services. The data shows whether this remains a wise decision or not. Service fees paid out to the operators has remained static, at around $1.7 billion; even despite fare revenues, which can be used as a proxy to determine how well services are doing, also dropping even more.
But for those that simply no longer need to commute, such as the new generation of flexi-work or 4-day work week converts? Maybe we can afford to give up on them; instead better serving their other travel needs by public transport.
While in theory there is the option, one must also consider the taxi companies’ and platform companies’ fare hikes, which may be felt more acutely by those actually opening their wallets to pay for the trips. And questions must also be asked about availability, with the increase in trips by private hire and taxis far outweighing the increase in vocational license holders, in 2023 compared to 2022. A key cause of the drop in satisfaction with such services also comes from increasing wait times; as tourism picks up and there is more demand for such services. Personally, I’ve also waited 10 minutes or more when using private-hire cars for work.
Even elsewhere in Asia, this has happened as well, so for those quick to blame the authorities, this is not unique to Singapore. Taipei has not reached parity with its 2019 numbers; and Hong Kong has barely done so despite opening the Shatin to Central Link; a new line through densely populated Kowloon and also providing additional cross-harbour capacity.
I aim to explore these trends and the policy response in future posts, but for now, it is what it is.
Like what you read? Join the Telegram Channel for updates!
