Friday, 8 March 2013

COE, ERP and the Sunk Cost effect

Tweaking the COE scheme to take into account what we know about behavioural economics.

Quote from this article:
As one driver told me, the high cost of his car makes him feel that he should use it every chance he gets, so he drives to the office every day rather than taking the bus that stops almost at his doorstep.
One unintended and unexpected effect from the COE scheme which was intended to reduce car ownership costs, and shift policy towards controlling car usage, was the "sunk cost" effect (or fallacy). The sunk cost effect is reflected in the quote above.

The problem with using the ERP to control or discourage usage is that the system did not account for what is now known as "behavioural economics".

Classical economic theory claims that sunk costs are irrelevant and it is only variable costs that should be relevant. So in classical economic theory a person who has bought a car will decide then that $1 ERP will make the car trip more expensive than a trip by bus or train and so rationally choose to take the bus/train.

Classical economists should get out more often.

Behavioural economics notes that sunk costs is very relevant to the (irrational) person (Hey! He bought a car! How rational can he be?). So a $1 ERP is nothing when the person has already spent $80k on a car (or on the COE alone!).

This has been borne out by this study.

The COE scheme needs to be tweaked to account for actual human behaviour (behavioural economics) and to make vehicle usage the issue (rather than ownership).

As paying the COE in one lump sum upfront is a sunk cost, the first thing to do is to reduce this sunk cost. The COE should be paid in installments based on usage.

Let's say the average driver worldwide drives about 1000 km per month. And say, the SG driver because of the "sunk costs" effect drives 1500 km per month.

On successful bidding, the car owner pays a deposit of the COE (say 10%, for illustration). So if the COE were $60k, he would pay $6k upfront. And he gets to register his car.

[Original proposal/post]
Every month where he drives less than 500 km, there is no further charge. However, the moment he exceeds 500km per month, he pays 1% of the COE. if he exceeds 1000km he pays 2% that month (or 3% if you wanna be progressively more punitive). And at 1500 km, he pays 5% that month (or more!).
Even at 1% per month (between 500 and 1000 km), it would be 12% a year, and 120% in 10 years in addition to the 10% first installment. So the $60k COE becomes $78k (130%) paid in total that the owner has control over (instalments) and so this would hopefully reduce the sunk cost effect and bring rationality (fat hopes!) back to the driver.

[Revised Proposal]

Every month the first 500 km driven by the car is "free" or rather pre-paid by the 10% of the COE "deposit" the owner paid to register the car (let's call this "Base COE"). However, when the car exceed 500 km, the driver may be charged according to these variations (or other variations):

a) 1% per 100km up to 1000 km, and 2% per 100km above 1000 km, 3% above 1,500km, etc.
b) 1% for the next 500 km, and 1% per 100km above 1000 km, 2% per 100 above 1500km, etc.
c) 1% for 2nd 500km, 2% for the 3rd 500km, etc.

The variations are in descending severity. Option A is very expensive. Driving 1500 km every month would incur 15% COE surcharge or 1.5 times the Base COE. So if the COE bid was $80,000, the owner would have paid $8,000 to register the car, and $12,000 to drive 1500 km per month. In a year, he would have to pay $144,000.

Option B would incur 6% COE surcharge, or $4,800 per month. In a year the COE surcharge would be $57,600.

Option C (is the original variation) and is the least severe. The owner would have to pay 3% more each month or $2,400 in COE surcharge. In a year, it would be 36% of the the Bid COE, or $28,800.

Other variations are possible. Option C is too lenient in my view. However, if we want to keep drivers to 1000 km per month, we could set the next 500km at a higher percentage, say 5%. 

People also tend to get desensitised to costs. If they are used to paying 2% for 500 km, they will no longer think about exceeding the limit. Unless the rate goes up.

What we hope is that either the driver spaces out his driving throughout the month, or if he fails to do that and exceeds the 500 km in one month, he would try to keep to the 500 km in the next month.

However there are administrative and implementation issues. How will monthly mileage be tracked? Can this be solved with tech? e.g. Electronic reporting by the IU? GPS tracking? Or just old fashion odometer readings during inspections.

There are also loopholes or flaws.

A rich guy with 7 cars can chalk up just below 3500km in 7 cars and not pay more than base COE (70% - 10% X 7 cars).

This is not an unsolveable problem. A simple revision of the deposit (or first installment) could address this. If the deposit is 50%, buying 2 cars may well be prohibitively expensive. To drive 3500 km with 7 cars would mean paying 350% COE.

A second approach would be to have progressively higher deposits. 10% for the first vehicle, 30% for the second, 50% for the third. And 100% for the 4th and subsequent vehicles. (Or any other rates 10% for first, 100% for second - drivers get a concession only for the first vehicle.)

The progressive approach helps by allowing those who simply need a car to afford one, while charging a premium for those who are "collecting" cars.

Of course, other loopholes will be harder to close. Registering a car under another persons name to avoid the higher Base COE charge. (Amendment added).

Another problem is what to do when you drive over to M'sia? To be fair, we would need to deduct distance travelled in M'sia? Odometer readings on entry and exit could be implemented.

A more difficult problem is the price of COE. As it is a bidding system and since successful bidders would only pay a portion of the sum bid, they may well bid the COE even higher?  Would the COE be bidded up to $200k? 10% is $20k which is lower than current COE. 1% is $2k for exceeding 500 km? But there is a propensity for temporal discounting so future costs are discounted (more behavioural economics). However these costs will be more germane and immediate when drivers hv to consider if they should incur those new costs (instalments).

Social relations will now change as drivers are less likely to give rides to friends and colleagues.

There will also be "surges" in usage. As you creep closer to the limit of 500 km, you will probably be more aware and reduce your vehicle use. But if you decide to exceed the 500 km mark (or inadvertantly exceed the mark), then you will suddenly increase your usage, because you have just "bought" another block of 500 km.

One solution is to reduce the "blocks". So first 500 km per month is covered by your initial deposit. The next 500 km is charged at 1%. After that, every 50 km could be charged 0.3% of the bid COE (equivalent to 3% for 500 km). 

There is another proposal to also charge by usage, but to allow COE to be purchase for blocks of distance, of about 25,000 km. There was no price stated for this 25,000 km block, which is equivalent to about 16 months of driving by an average Singaporean car owner.

Based on the status quo, 25,000 km should be charged at about 14% of prevailing or bid COE. However, if you want to reduce usage, it should be charged at 20% (or somewhat more than 14%). At 20%, the messaging is that 25,000 km should be 2 years of travel.

I will leave behavioural economist to analyse or study the merits of either approach or whether the approaches have adequately neutralised the "sunk cost" fallacy, as well as temporal discounting.


For commercial vehicles, the allowable distances could be set higher... or the Base COE could be higher - say 50%, and the allowable distance be set at 2500 km per month (if that is the typical distance needed for commerce). Or heavy vehicles needed for transportation of goods, could be exempted from this distance cap - this would keep costs affordable for businesses.

No comments: