Latest figures from the Land Transport Authority (LTA) show that the private car population is at its lowest level since 2011, falling to 598,219 as at the end of January.
The shrinkage could continue, with analysts saying it is a sign that the supply of Certificates of Entitlement (COEs) is lagging behind actual replacement demand.
But a larger imbalance is on the horizon and could cause COE prices - which have been trending down - to spike by 2020 unless something is done to smoothen the annual COE supply.
That is because the supply of COEs in categories A, B and E is likely to shrink by 2020, when fewer cars are likely to be scrapped. The number of cars scrapped decides the number of COEs available in the following year.
LTA figures show that at the end of last year, the number of cars aged up to five years ranged from about 22,000 to about 41,000, while those aged six to 10 years ranged from about 68,000 to 114,000.
Thanks to the low COE supply since 2011, cars are being scrapped near the 10-year expiry of their COEs.
This contrasts with the years of bountiful COE quotas between 2004 and 2008, when the typical scrapped car was six to seven years old.
Thus, if most cars continue to be scrapped when their COEs are about to expire, we may expect a feast of COEs in the next five years.
But by 2020, low COE quotas are again likely and a painful consequence is high premiums, which have averaged about $64,000 in category A (and about $75,000 and $76,000 in categories B and E) since 2011.
In comparison, COE pre-miums in those categories averaged between $16,000 and $17,000 between 2004 and 2008.
One way to mitigate the roller-coaster ride in COE pre-miums is to smoothen supply. Back in 2013, anticipating the feast-or-famine cycle in the years ahead, Transport Minister Lui Tuck Yew asked the LTA to find "a practical way of putting aside some of the supply from the peak that we expect in the next few years, for the future when COE supply becomes tighter".
An analysis of COE figures shows that premiums shot up dramatically whenever the bi-monthly supply slipped below about 800, 700 and 500 in categories A, B and E.
One way to reduce volatility is to have at least those amounts of COEs in the three categories in the auctions held twice a month. That would amount to about 48,000 COEs annually.
To ensure this minimum supply from 2020 to 2024, when few cars are expected to be scrapped, COE supply from 2015 to 2019 would have to be adjusted. That would help ensure the car population does not exceed the longterm target of 620,000.
There are several ways to "bank" COEs in years of plenty.
One consists of setting a fixed cap, for example, no more than 72,000 COEs per year across categories A, B and E. Another is to set a variable cap by "banking" a constant fraction - say one-fifth - whenever the annual number of COEs in these categories exceeds 57,600.
These caps were arrived at assuming zero growth in the car population and the 2014 scrapping pattern.
How the caps would work
UNDER the fixed-cap scenario, if the combined annual quota for COEs in categories A, B and E is below 48,000, it will be adjusted upwards to 48,000.
If the combined quota is between 48,000 and 72,000, no adjustment will be made. But above that, the quota of COEs will be capped at 72,000.
This strategy will result in total savings peaking at about 88,000 COEs by 2019, but these will be more or less used up by the end of 2024.
A disadvantage of a fixed cap on COE supply is that in some years, it can lead to a mad rush to buy. For example, about 114,000 cars will reach 10 years in 2016. Supplying only 72,000 COEs then could lead to a scramble for them by owners who want to buy new cars to replace the ones they scrapped.
A variable cap is a way of sharing the "pain" more equally. Based on simulations, it would seem that only about 20 per cent of COEs need to be banked when total annual supply in categories A, B and E exceeds 57,600 between this year and 2019. That will guarantee at least 48,000 COEs in these categories and it will also result in total savings peaking at about 88,000 COEs, which will be more or less disbursed by the end of 2024.
The option of a flat supply of COEs has often been mentioned. It turns out that under the same assumptions as above, the annual flat quota works out to about 59,000 COEs across categories A, B and E from this year.
That, however, would result in total savings peaking at about 143,800 COEs by 2019, implying that only about 77 per cent of the 620,000 target car population will be on the road then. Hence, this option is the least practical.
In the 25-year history of the COE, there have been two roller-coaster cycles in the pre-miums.
Another peak in premiums is expected in about five years and that can be avoided by a bold management of COE supply. The earlier that is implemented, the less painful it will be for all stakeholders.
The writer is Associate Professor of Analytics at the NUS Business School.