Barnabé Monnot

Research scientist @ Robust Incentives Group, Ethereum Foundation.

Research in algorithmic game theory, large systems and cryptoeconomics with a data-driven approach.

Publications + Talks

Posts

Reviews

Teaching

News

About + CV

Collected thoughts on transportation in Singapore I: Cars

Last week I had the pleasure to attend the Science, Technology and Policy workshop / summer school at the CREATE tower of NUS. They house the Singapore MIT Alliance on Research and Technology (SMART) who developed lately a prototype of self-driving car (which we tested!) and the Singapore branch of ETH’s Future Cities Laboratory. The workshop was tying up these two forces as well as locally-based people and asked us, students and young researchers from these institutions, to think about what Singapore as a smart nation could look like.

Thinking about cities and how we move in them has been somewhat of an obsession for me ever since I left my home in (near) Paris. Walking and busriding and trainriding and ridehailing around Hong Kong, Tokyo, Kyoto (those I know a bit better) and scattered spots in the Philippines, India, Thailand, the US has in many ways shaped my enjoyment of these places we live in, work in, explore and use. I hope to expand on all these aspects sooner or later, but for now I’ll use this past workshop to collect some thoughts on transportation in Singapore and what’s (possibly) next. These notes will be split in several parts, starting with cars.

Cars

  • Singapore is a strange beast when it comes to transportation. Cars are “prohibitively” expensive. I put prohibitively in quotes since, well, cars are still there to a large extent. Of the total percentage of vehicles on the road, over 60% are cars indeed. Roads make up almost 12% of the usable land in Singapore. First, why are they so expensive? Because the car itself is taxed over 100% by the state and drivers need to bid for a Certificate of Entitlement (COE) tied to the car that allows them to use it. The COE has a duration of 10 years and the bidding system has reached a maximum of 72,000SGD a few years ago. Meaning that it could be as expensive as 100,000SGD to drive a simple Toyota. Local banks have developed loans to help families finance their ownership and it is not rare to see the more affluent to drive shiny german cars (Singapore has the highest percentage of millionaires of any country in the world).

  • Ownership costs are through the roof, but Singapore is (mostly, relatively) congestion-free and the road infrastructure is excellent. Most roads have two lanes, with some non-freeway roads being four lanes (one of them a bus lane usually). Potholes are rare too. But if I was paying that much for my car, I would not expect less from the government to which around 80% of the money put up to buy the car is going to. The flip side of the coin is that people will use their car for anything. This is seen in some data collected by LTA, but also something we more or less saw in our own experiments. Short trips (sometimes less than a km) that could be made by foot or using bicycles say will be done with the car.

  • Which brings me to my next point, the first-mile last-mile problem, a favorite of transportation researchers lately. Singapore does not (yet) have an exhaustive rail-based system. The current system, called MRT, can be sometimes far from homes and to reach it, people rely on using the car (maybe a ride-sharing app) or taking a / some bus(es). In that sense the connectivity is not perfect, though this may be fixed by the appearance of a few new MRT lines (regular openings are planned until 2030) and better infrastructure for alternative means of transportation between large housing blocks and train stations (such as the growing Park Connector Network (PCN)).

  • Coming back to car, ownership and costs, we had a nice talk by Prof. Costas Courcoubetis, based on a paper written with a fellow ESD PhD candidate. The model looks at a world where car owners can rent some seats in their car and non-owners can decide to buy a car, use a ride-sharing app, or take public transportation (this is an extremely simplified presentation). One of the consequences though is that increased ownership costs don’t necessarily decrease the population of cars. One reason is that renting out extra space effectively dampens the impact of increased costs. On the other hand, usage costs (think tolls or expensive gas) tend to hurt that much more, so a policy that wishes to reduce congestion and the number of cars on the road would do well to increase usage costs instead of ownership. Not modeled in their paper, but perhaps also a factor: the psychological effect of sunk cost fallacy, which dictates that once someone has already put up a large sum of money, they should continue doing so simply because of the amount already on the table. In this context, the sunk cost is the heavy price tag on buying a car in Singapore, but it works in reverse too: usage costs now make any trip “costly” (even if the tolls are small), and you would think twice before using your car.

  • This could find an interesting outlet if tomorrow (more like in a few years), the cost of transportation becomes so low, perhaps because of autonomous vehicles. But if that’s the case, we can very certainly expect a lot of congestion, or at least for the transportation market to find an equilibrium after which the level of congestion make it unprofitable to go on the road. I would argue here that this is still not what we should aim for. On a mostly subjective level, four-lane roads and highways everywhere are not good urban planning. In the case of Singapore, where it is especially hot, it makes for a painful experience sometimes to simply walk around. Buildings are designed with cars in mind, with large driveways and not necessarily paths provided for pedestrians. Cars have priority too, something that contrasts with Japan for example, where cars mostly respect the pedestrians’ right-of-way. Waiting for sometimes a good 60 seconds before an uninterrupted file of cars turns at an intersection where no traffic light is regulating, and where it is not immediately clear if they will end up turning or not since the indicator is definitely optional can be annoying, but also dangerous under the hot sun.

  • In my opinion, smaller roads make for a more fluid interaction with the environment. Hong Kong is a majorly congested city, by design in fact (something I learned was called induced congestion). But on the island and parts of the peninsula, smaller roads make it easy to cross to a neighboring shop, and this organic flow seems to translate in a higher density of businesses and street hustle. There is a “tug-of-war” relationship between the pedestrians and the drivers, who have indeed priority but will know to stop before a crossing if they expect to be stopped there. Singapore on the other hand can only be designed with large distances in mind, because it is not clear that pedestrians on one side would (or could) simply cross to the other side to see something. Shops tend to cluster in malls that are attached to large train stations, or at the bottom of subsidized housing blocks where they know they will get some throughput. In turn, this makes for some “urban desert” spaces, where few points of interest will be around while you walk from point A to point B, along walls.

  • But back to costs, how do we make drivers abandon their cars? In this article, Kevin Webb argues that private transportation knew a small revolution when real-time information started flowing through the network, in the form of dynamic routing algorithms used by Google Maps or Waze. But information is not enough, as it can only make the system more efficient, though not necessarily with lessened demand for roads. One way to achieve so is with incentives, something Singapore has experimented with for some time now. They were one of the first cities to introduce congestion control mechanisms such as the ERP, an automatic tolling system installed on the major arteries of the road network around the city center, that dynamically adjusts rates every 30 minutes at peak hour. A car entering the central business district at 9am might incur a 2SGD fee for example. And the new version of the system, ERP2, is planned for 2020, and will continuously track every vehicle with a GPS-assisted device. This opens new ways of incentivizing users to leave their cars, though the LTA Master Plan writes that they would encourage cars to make shorter trips. They may mean that cars should be used less, but it is not clear. I would assume that you would instead want someone to use their car should they be in need of making a long distance trip, and incentivize them to leave the car if their trip can be done using different means of transportation (maybe bicycle and train for example). Perhaps a concave cost function would do the trick, where every additional mile would cost less than the previous one…

Read on for more thoughts about cycling

Back to posts