9. Transport Policies In Singapore

9. TRANSPORT POLICIES INSINGAPORE

Georgina Santos, Wai Wing Li and Winston T. H. Koh

1. INTRODUCTION

Being an island city-state that measures 42 km (26.1 mi) East to West and 23 km(14.3 mi) North to South, Singapore has 3,149 km of roads (1,955 mi) for apopulation of about 4.2 million people and 707,000 registered motor vehicles inthe year 2002.

Since gaining independence in 1965, Singapore has enjoyed sustained andrapid economic growth, with an annual average growth rate of about 8% untilthe mid-1990s. Rising affluence has brought an increase in demand for personalmotor transportation. In 1960, the level of motorisation in Singapore was 50cars per 1000 people, similar to levels in the U.K. and France then, but wassubstantially higher than other developing countries and Japan, where the figurewas 10 cars per 1000 people. By 1995, with a per-capita GDP of S$33,520 at 1995prices (or US$19,154, at a 2003 exchange rate of US$1 = S$1.79), comparableto the levels in U.K., France and Japan, the level of motorisation in Singapore, at105 cars per 1000 people, was about a quarter of the levels experienced in thesethree countries (Willoughby, 2000).

Successfully managing congestion was considered an important part of thecountry’s efforts to attract multinational corporations to locate their operationsin South-East Asia in Singapore. Like in other cities, urban congestion not onlycreates air and noise pollution, but also increases the cost of doing business. Thiswas a serious problem for Singapore in the early 1970s, when traffic during the

Road Pricing: Theory and EvidenceResearch in Transportation Economics, Volume 9, 209–235Copyright © 2004 by Elsevier Ltd.All rights of reproduction in any form reservedISSN: 0739-8859/doi:10.1016/S0739-8859(04)09009-2

209

210 GEORGINA SANTOS ET AL.

morning and evening rush hours in the central business district (CBD) slowed to acrawl of about 20 km per hour (12.5 mi per hour). Recognising the repercussionson foreign direct investments into Singapore, the government tackled the problemby formulating a transport policy, which has continued to evolve to this day.Besides a series of road pricing schemes designed to control traffic congestion,efforts were also made to develop and upgrade the transport infrastructure (e.g.the construction of a mass rapid rail system) to support the country’s economicdevelopment.

Singapore has adopted an integrated approach to traffic management, witha set of “carrot and stick” policies. The sticks raise cost of car ownership andmotoring through the imposition of various “ownership taxes” and “usage taxes.”As of 2003, the ownership taxes include a quota licence fee (levied throughan auction scheme), import duties, a basic registration fee and an additionalvehicle registration fee (calculated as percentage of car value). The usage taxesinclude tolls for entering the CBD or using expressways, petrol taxes andparking fees. The carrots to encourage the use of public transport include thedevelopment of reliable and convenient bus, rail and taxi services. The policiesthat Singapore has undertaken in the last thirty years can be summarised asfollows:

� coordinated traffic management with ownership taxes and usage taxes on vehiclesin order to achieve an optimal usage of the road network;

� expansion of the road network in order to provide access to every part of thecountry;

� development of a mass transit railway network;� improvement of the quality of buses and bus services;� coordination and integration of bus, rail and taxi services.1

As well as directly reducing congestion, Chapter 6 of this volume points out thatroad pricing produces a virtuous cycle in shifting motorists to public transport.Shifting people from cars to buses reduces congestion and allows buses to travelfaster; with more customers for both buses and trains, the frequency of servicesincreases, resulting in a better service and a further increase in demand, takingmore cars off the road.

This chapter has three objectives:

� to describe the transport policies in place in Singapore with special attention toroad pricing;

� to analyse their successes and shortcomings;� to draw lessons for other cities around the world.

Transport Policies in Singapore 211

2. OVERVIEW OF POLICIES

From 1960 to 1970, the private vehicle population in Singapore roughly doubled,from 70,100 to 142,500. The key factors behind the increase in car ownership wererising household incomes, a housing programme to develop residential districts inthe suburban areas away from the CBD, and unreliable public transport. Duringthis period, public buses and taxis increased by about 64%, while the total lengthof public roads increased by only about 35%. In 1970, the length of major arterialroads was 240 km (150 mi), and had only increased 12% since 1960.

With the number of private cars per km of arterial roads rising from 328 in1961 to 594 in 1970, traffic conditions were deteriorating fast. The railway systemwas not built until the 1980s, so the only modes of public transport were taxisand buses. Although there were ten bus companies, services were infrequent andschedules were uncoordinated.

In 1970 the government directed the bus companies to merge into four and tosubstantially expand their fleets; in 1973 these four companies were nationalisedand merged to form a new bus company, the Singapore Bus Services. This moveresulted in swift and significant improvements in both the quality and quantity ofpublic bus services. The national company’s shares were floated on the Singaporestock exchange in 1978 but the company remained strongly regulated.

After ten years of studies and cost-benefit analysis of different options, theSingapore government approved the construction of the Mass Rapid Transit(MRT) railway system in 1982. World Bank consultants and a team of economistsfrom Harvard found that the rail construction costs in the cost-benefit analysiscarried out by Wilbur Smith and Associates had been underestimated (Phang,2003). Mohring (1983) estimated that the measurable real rate of return wouldbe less than 2%. Notwithstanding all this, the project went ahead, mainly becausethe government wanted to attract foreign investments and talent to Singapore.

A consistent policy of the Singapore government in the last three decades hasbeen to ensure that Singapore’s urban infrastructure (transport and housing) andpublic services provide a conducive environment for businesses. It is this con-sideration that swayed the government to embark on the MRT project. The MRTnetwork has been continuously expanded since it came into operation in 1987,with a system of buses and the Light Rail Transit (LRT)2 serving as feeders to theMRT network.

Beginning in the mid-1970s, there was substantial expansion in the strategicroad network of arterials and motorways, with a six-fold increase in investmentsin road construction and upgrades between 1975 and 1980. Road density doubledbetween 1975 and 1985, and tripled by 1997 (Willoughby, 2001).

212 GEORGINA SANTOS ET AL.

An ownership tax on vehicle ownership, the Additional Registration Fee (ARF)has been in place since the late 1950s, when Singapore was still under Britishcolonial rule. Intended originally as a revenue-raising measure, the ARF is anad valoremduty on a vehicle’s open market value (OMV)3 payable by buyers ofnew motor vehicles, in addition to an administrative fee, referred to as the BasicRegistration Fee.

The publication of a study by Wilbur Smith and Associates (1974), highlightingthe dangers of uncontrolled private vehicle ownership (Foo, 2000), alerted theSingapore government of the urgency of reforming traffic management policies.The ARF was expanded in scope to control car ownership, and its rate was steadilyraised through the 1970s, reaching 125% in 1978 and 150% in 1980. However,as the ARF rate rose, it also discouraged existing vehicle owners from replacingtheir cars and encouraged new car buyers to buy used cars. Concerned with astock of aging vehicles, the government introduced a Preferential AdditionalRegistration Fee (PARF) to counterbalance the disincentives on vehicle renewal,when the applicable ARF rate was raised to 100% in 1975. The purchaser of anew vehicle paid a substantially lower PARF rate if he de-registered an old vehicle(i.e. by exporting or scrapping it) of the same engine category at the time of hisnew purchase. PARF rates varied according to the engine capacity; for examplebetween December 1975 and October 1983, they ranged from 35% (expressedas a percentage of the required ARF payment) for the smallest engine category(<1000 cc) to 55% for the largest engine category (>3000 cc). Since 1997, thePARF has been amended to a system where the applicable discount is a function ofthe age of the vehicle to be de-registered. As of 2003, de-registering a vehicle under5 years old qualifies for a 25% PARF rate and de-registering a vehicle between9 and 10 years old qualifies for a 50% PARF rate. Vehicles over 10 years old nolonger qualify for PARF treatment.4

The Area Licensing Scheme (ALS), essentially a usage charge, was formulatedin 1973 and implemented in 1975, and it was the first of its kind in the world.Vehicles entering the 7 km2 (2.8 mi2) restricted zone (RZ), which included theCBD, were required to purchase and display a paper area licence on their wind-screen. Enforcement was done manually by enforcement officers standing at theboundaries of the RZ. Offending vehicles were not stopped but issued a summons.Drivers could either appeal to the Traffic Police Department or pay the fine.

When the scheme was initially implemented, the restricted hours were from7.30 to 9.30 AM daily, except on Sundays and public holidays. Three weekslater, the restricted hours were extended to 10.15 AM in order to reduce theexcess traffic occurring immediately after 9.30 AM (Chin, 2002), as motoristsrescheduled their trips to just before and after the restricted hours and businessesdelayed their opening hours to avoid paying the area licence fee. In June 1989,

Transport Policies in Singapore 213

the scheme was further extended to the evening peak, from 4.30 PM to 7 PMon weekdays. The evening period was later cut back by half an hour to 6.30 PMto accommodate requests from residents who lived inside the charging area butworked outside, although this was subsequently extended back to 7 PM, becauseof increased traffic congestion (Chin, 2002). In January 1994, the operatinghours of the scheme were further extended to cover the inter-peak period, from10.15 AM to 4.30 PM on weekdays and the post-peak period of 10.15 AM to3 PM on Saturdays. The Saturday charging hours were later cut back to 2 PM.

A vehicle displaying the licence could enter and leave the RZ an unlimitednumber of times during the day. Police cars, ambulances, fire engines and publictransport buses were all exempt. In addition to that, in the beginning, taxis, goodsvehicles, motorcycles, and passenger cars carrying three or more passengers apartfrom the driver were all exempt. Two months after the implementation of theALS, taxis were required to purchase the permit as well. In 1989, motorcycles andgoods vehicles, were also required to purchase the permit, together with car-pools(Chin, 2002). No discounts or exemptions were given for residents living insidethe RZ. Driving inside the zone without crossing the boundary could be donefree of charge. In this sense, the scheme was a cordon system and not an arealicence system. “Area Licence” was not a good name, it should have probablybeen called “Entry Licence” or “Cordon Licence.”

The ALS was effective in reducing urban congestion during the morningand evening peak hours. The charges however might have been set too highinitially and might have reduced traffic by more than was necessary, leading tounder-utilisation of the road network and in the process, shifting congestion to theexpressways and non-restricted times. Holland and Watson (1978) for example,find that upon introduction of the ALS, the volume of cars entering the RZ duringthe restricted period fell by 73% whilst the volume of cars entering the RZ outsidethe restricted times rose by 23%. Average speeds went up from 19 to 36 km perhour (11.8–22.4 mi per hour), exceeding the government’s optimal flow target(Phang & Toh, 1997). McCarthy and Tay (1993) conclude that the initial price ofan ALS licence, which the government set based on its judgement rather than anycalculation of the congestion externality, was about 50% above the optimal level.The government appeared to be less concerned with short-term optimisation ofroad use than with long-term reduction of urban congestion, and raised the dailylicence fee for private cars from S$3 to S$4 six months after the introduction ofthe ALS, and raised it again to S$5 in 1980, where it stayed until it was broughtback to S$3 in 1989. This reduction in rates in 1989 was mainly because morevehicles were required to purchase licences (Chin, 2002). Li (1999) estimatesthe optimal toll using traffic count data and concludes that S$3 was the correctfee for 1990.

214 GEORGINA SANTOS ET AL.

In January 1994, an alternative “part-day” licence was introduced. Thus, vehiclescould purchase a licence to enter the RZ throughout the day, or a licence that wouldonly allow them to enter the RZ during the inter-peak period (Chin, 2002). Theinter-peak ALS charges were two thirds of the whole day ones. For cars, thefees were S$3 and S$2 respectively. Part-day licences were valid for use between10.15 AM and 4.30 PM on weekdays, and from 10.15 AM to 3 PM on Saturdays(Foo, 1997a).

In 1989, the Parliament of Singapore appointed a Select Committee to examinethe transport policy at the time. Several public hearings were held with themajority of views in favour of usage restraint rather than ownership restraint.However, the Select Committee recommended the Vehicle Quota System (VQS),to control vehicle growth5 (Olszewski & Turner, 1993). One of the concerns aboutvehicle ownership must have been the limited parking provision, mainly the resultof high housing density, as providing housing had been a key government policy.

The VQS was thus introduced in May 1990 and remains in place today. It isbasically an ownership tax applied in addition to the ARF,6 and another policythat was also first of its kind in the world. Prospective vehicle owners are requiredto purchase a ten-year quota licence called a Certificate of Entitlement (COE).The quota on COEs for each category is announced a year in advance, in Mayeach year. Initially the COEs were allocated in a monthly sealed-bid uniform priceauction. Each individual was allowed to bid only once in one of the categoriesand was required to pay a deposit equal to half his bid. All the successful bidderspaid the lowest bid that cleared the quota allocation, which is referred to asthe quota premium, and had three months to purchase a new vehicle using theCOE, failing which the deposit was forfeited.7 The deposits from unsuccessfulbidders were returned.

When it was implemented in 1990, the monthly quota was separated into sevencategories:

� Category 1: Small cars with engine capacity of 1,000 cc and below;� Category 2: Medium-sized cars with engine capacity of 1,001–1,600 cc, and

taxis;� Category 3: Large cars with engine capacity of 1,601–2,000 cc;� Category 4: Luxury cars with engine capacity of 2,001 cc and above;� Category 5: Goods vehicles and buses;� Category 6: Motorcycles and scooters;� Category 7: “Open.” A Category 7 COE may be used to purchase any type of

motor vehicle.8

In May 1999, after a government review of the quota system, Categories 1 and 2were merged into one category, as were Categories 3 and 4. The categories werere-named as follows:

Transport Policies in Singapore 215

� Category A: Cars, 1600 cc and below, and taxis;� Category B: Cars, above 2000 cc;� Category C: Goods vehicles and buses;� Category D: Motorcycles;� Category E: “Open,” for registration of all types of vehicles.

Under the VQS, when a vehicle reaches ten years of age, the owner is requiredto renew the COE for another five or ten years at the prevailing quota premium.No further renewals are allowed thereafter. Since November 1998, the prevailingquota premium is computed as a three-month moving average of the quotapremium for the category. Before November 1998, a twelve-month average wasused. If the motor vehicle is sold before the expiry of the quota licence, the quotalicence will be transferred to the new owner together with the vehicle. A pro-ratedrebate of the remaining validity period is paid if the vehicle is deregistered,scrapped or exported. Owners of vehicles over eight years old at the time ofintroduction of the VQS were given a grace period of two years, after which theywere required to purchase a COE at the prevailing quota premium. A discussionon the successive changes to the VQS until today follows in the next section.

In June 1995, a paper-based Road Pricing Scheme (RPS), operating in thesame way as the ALS, was introduced on an expressway (East Coast Parkway).This was later extended to other expressways. The aim of the RPS was to reducecongestion on the expressways during the office rush hours, and to familiariseSingaporeans with both linear passage tolls and road charging outside the CBD(Goh, 2002).

In 1998, Electronic Road Pricing (ERP) replaced both the ALS and RPS.Gantries were installed at all the approach roads to the ERP zone and on theexpressways. In contrast to the ALS and RPS, the ERP scheme charges vehicleseach time they cross a gantry. The system uses a dedicated short-range radiocommunication system. Vehicles wishing to enter the ERP zone when chargesapply need to install an In-vehicle Unit (IU), which is a radio transponder in whicha stored-value smart card is inserted. The fees are deducted from the smart cardwhen a vehicle passes under a gantry at the restricted times, and the IU displaysthe remaining balance on the smart card. ERP charges vary between differentgantries and different times of the day, depending on the level of congestion.9 Thecharges are reviewed quarterly and for the June and December school holidays toachieve an optimal flow of traffic. Optimal speeds in Singapore have been foundto be between 20 and 30 km per hour (12–19 mi per hour) for major roads and45–65 km per hour (28–40 mi per hour) on expressways.10

The basic principle behind the government’s transport management policiesregarding the ARF, petrol taxes, import duties and parking charges has alwaysbeen to maintain them at as high a level that is politically acceptable. The VQS

216 GEORGINA SANTOS ET AL.

and the ERP, still the only ones of their kinds in the world today, deserve a detailedanalysis; hence, the rest of this chapter will concentrate on these two schemes,within the wider framework of Singapore’s integrated transport policy.

3. THE VEHICLE QUOTA SYSTEM

Various problems encountered with the use of the ARF to control vehicleownership led to the introduction of the Vehicle Quota System (VQS), whichwas based, as explained above, on the recommendations made by the SelectCommittee appointed by the Parliament. Phang et al. (1996) note that the PARFmade used and scrap car prices dependent on the prices of new cars. For example,the Japanese yen appreciated 52% against the Singaporean dollar between 1984and 1989, raising the import prices of new cars and raising the absolute amountof the PARF discount from scrapping a car. In turn, this directly increased thevalue of old cars. Furthermore, increases in the ARF rate without concomitantchanges in the PARF also raised the price of old cars, since the ARF was set as apercentage of a vehicle’s OMV whilst the PARF was set according to categorieswith engine sizes. Phang et al. (1996) point out that an inferior car within one ofthe top engine size categories, bought in the late 1970s, could have been sold forhigher than the purchase price ten years later. This was because the old car couldbe sold to a potential car buyer, who could scrap it when buying an expensivenew car in the same engine capacity category, and thus qualify for the PARF.This “appreciating asset” argument was used by car dealers to stimulate sales,and many car buyers bought cheaper makes from Russia and Eastern Europe,expecting the scrap prices to be close to their initial purchase price.

Although the ARF increased the cost of vehicle ownership significantly,it proved to be relatively ineffective in controlling the growth of the vehiclepopulation. Phang and Chin (1990) estimate the income elasticity of demandfor cars in Singapore to be around 1 and the price elasticity to be around −0.45,indicating that the demand for ownership is relatively inelastic. Between 1980 and1989, rapid economic growth had produced a 65% increase in the motor vehiclepopulation in Singapore.

3.1. VQS in Theory and Practice

In principle, usage taxes (such as ERP charges) can fully internalise congestionexternalities, as these taxes directly affect the price of road journeys, which atthe margin are the source of the externality. Ownership taxes (such as the quota

Transport Policies in Singapore 217

premiums under VQS and the ARF), on the other hand, control the size of thevehicle population, and aim to indirectly reduce the amount of potential urbancongestion.

The main justification for ownership taxes, however, must have been thatgiven the investment in MRT and housing, and the scarcity of parking spaces,the government felt necessary to contain vehicle ownership. It is with thisconsideration in mind that the government decided to augment its plethoraof usage taxes with a new ownership tax that was directly set by the market.Furthermore, as Toh (1992) points out, one-time ownership charges are likely tobe less painful to consumers than recurrent usage charges.

The success of the VQS should therefore be judged by its ability to control thegrowth rate of the motor vehicle population. Between 1975 and 1989, the averageannual motor vehicle population growth rate was 4.4% with a standard deviationof 4.24%. Between 1990 and 2002, after the implementation of the VQS in 1990,the average annual motor vehicle population growth rate went down to 2.83%,with a standard deviation of 2.24%.11 Thus, the VQS was successful in loweringboth the rate of vehicle population growth and its volatility.

Bhagwati and Srinivasan (1969) show that the social utility-maximising policyto restrict consumption below a certain level is a consumption tax on the good.Since Singapore has no domestic motor vehicle manufacturing industry, andassuming perfect competition in the motor vehicle market, an auction quota isequivalent to an import tariff, hence it is an efficient consumption tax. However,Singapore’s VQS led to some unexpected and undesirable consequences.

3.2. The Open Category

The rationale for the open category licence was to allow for flexibility in the mix ofthe motor vehicle population, according to the demands of the market, so that, forexample, if consumer preferences shifted towards larger cars, the open categoryCOEs would be used to purchase large cars. As long as the open category is not toolarge,12 one would expect that its quota premium would be close to the highest-priced category, otherwise buyers of vehicles in this highest-priced category wouldbe better off bidding in the open category rather than in their own category. Hencethe open category COE would tend to be used in the highest-priced Categories3 and 4, and one would expect the shift in the vehicle mix to be towards luxurycars (Tan, 2001).

Using the latest available data (May 1990 to June 2003), Fig. 1 and Table 1show that the prices of an open category COE were most closely correlatedwith Categories 3 and 4 prices, the luxury car categories with the highest COE

218 GEORGINA SANTOS ET AL.

Fig. 1. Quota Premiums Under VQS: 1990–2003. Note:In 1999, Categories 1 and 2 werecombined to form Category A, while Categories 3 and 4 were combined to form CategoryB. Category 7 was renamed Category E. The quota premiums for Category 7 (E) were oftenclose to those of Categories 3 and 4 (B). Source:Land Transport Authority, Singapore.

www.lta.gov.sg.

prices. Data on the use of the open category COEs are not published, but a May1990 to April 199913 average shows that the ratios of new registrations to quotalevel for Categories 1–4 were 95, 113, 195 and 260% respectively. This waspossible only through the use of the open category COEs on the higher categories(Tan, 2001).

Phang et al. (1996, p. 148) observe that “by 1995, the Mercedes Benz hadovertaken Toyota as the most popular make of car registered in Singapore.” Theincreasing population of luxury cars gradually led to larger quotas for these cars.The shares of cars between Categories 1–4 in 1990 were 15, 67, 14 and 4%respectively. By 1999 these shares had changed to 12, 60, 20 and 8%. This trendcontinued even after the change to open bidding (which is discussed below). In

Table 1. Correlation Coefficient With the Open Category.

Cat 1 Cat 2 Cat 3 Cat 4 Cat 5 Cat 6

0.50 0.88 0.96 0.97 0.84 0.64

Source:Land Transport Authority, Singapore. http://www.lta.gov.sg.

Transport Policies in Singapore 219

2002 the share of the combined Category 1 and 2 was 68% and the share of thecombined Category 3 and 4 was 32%. Although sub-categorisation created theincentive to purchase larger cars, there was also a shift in consumer preference tolarger cars, particularly in light of rising income levels.

3.3. Sub-Categorisation

The rationale for having four categories for passenger cars, grouped by enginecapacity, was that low-income car buyers would not have to bid against wealthybuyers for a COE, and pay more as a result. Tan (2001) uses a two-categorymodel of high and low-income buyers, to show that low-income buyers will notnecessarily be forced out of the market, because even though their inverse demandfunction is everywhere below that of high income buyers, the supply price of theircars is also lower. Falvey (1979) analyses the case where there is substitutionbetween the categories, i.e. buyers of cheap cars switch their purchase to luxurycars because of the fall in the price of luxury cars relative to cheap cars. In thiscase, the absence of categorisation would not force low-income buyers out of themarket, because they would voluntarily upgrade to luxury cars.

One problem with sub-categorisation is allocative efficiency. It is very difficultto determine the correct quota for each category, possibly resulting in excessdemand in some categories and excess supply in others, and leading to high pricesin the former case and an under-utilisation of the total quota in the latter. This hasbeen the experience of the VQS, which produced wild swings in COEs’ prices.In December 1997 for example, the price of a Category 3 COE was S$64,100; inJanuary 1998 the price was S$50, as the number of bids was lower than the numberof quotas. In late 1994, a Category 4 COE topped S$110,000. At the same time,there was a collapse in demand for motorcycles, due to the imposition of strictemissions standards in October 1991, when not many motorcycles in the marketat the time met the standards, and improved models took some time to arrive. Asa result, for 27 months between December 1991 and February 1994, the price ofa motorcycle COE was S$1.

The evidence clearly suggests that sub-categorisation failed to ensure equity. Ofthe 106 auctions held between May 1990 and April 1999, Category 1 premiumsranked the lowest of the four car categories for 81% of the time, Category 2premiums ranked second lowest for 59% of the time, Category 3 ranked thirdlowest for 49% of the time, and Category 4 ranked highest for 53% of the time.The fair outcome with the ranking of quota premiums in the order of the categories1 < 2 < 3 < 4, on the other hand, occurred only 43% of the time. In other words,well over half of the auctions produced an outcome where it was cheaper to buy

220 GEORGINA SANTOS ET AL.

a COE for a larger car than a small car. Furthermore, in 13% of the auctions,Category 1 premiums were higher than Category 4 premiums.

The VQS reduced the difference in the relative prices between luxury and cheapcars. The relative price of luxury cars fell and there may have been a substitutioneffect raising demand for luxury cars. With similar COE prices, it was preferable topurchase an expensive car to preserve the resale value. This also explains, at leastin part, the improvement in the general quality and price of the Singaporean vehiclefleet. As can be seen from Fig. 1, the quota premiums of the four quota categorieswere very close between May 1990 and mid-1993, and have been close again sinceearly 1998. Even when the quota premiums for smaller cars were lower than thosefor larger cars (the fair outcome of category premium 1 < 2 < 3 < 4), for examplein January 1992, the implicit tax rates for Categories 1–4 cars were 119, 123, 75and 28% respectively, producing a highly regressive system (Tan, 2001).

Sub-categorisation is attractive in theory, but in practice, it is difficult to deter-mine the shape and position of the demand curves (Tan, 2001). The quota limitswere inappropriate most of the time, the quotas for Categories 1 and 2 should havebeen higher and the quotas for Categories 3 and 4 should have been lower. In May1999, the government merged Categories 1 and 2 into Category A and Categories3 and 4 into Category B14 to alleviate these problems. Although the situationimproved, COE prices were still regressive. In 40% of the auctions held betweenMay 1999 and June 2003 for example, the price of a Category A COE was higherthan that of a Category B COE. Even after the introduction of open bidding in2001, as explained in the next section, 36% of the auctions produced outcomeswhere it was cheaper to buy a COE for a larger car than to buy one for a small car.

In reality, the Singaporean government essentially set the quota limit for thevery first auction. The auctions that followed were determined by a pre-announcedtarget growth rate common to all categories, the number of de-registrations ineach category and the use of the open category COEs. The open category has,over time, led to larger quotas for luxury cars.

3.4. Transferability and Speculation

When the VQS started in May 1990, COEs were transferable once before beingused to purchase a vehicle. The COE had to be obtained at least one month beforethe actual purchase of the vehicle. Given the uncertainty surrounding the value ofthe quota licence at the time of the auction, the secondary market for COEs acted asa mechanism to smooth out demand. As discussed in Krishna and Tan (1997, 1999)and Koh (2003), transferability is an important consideration for social welfare asit provides flexibility in the final allocation of COEs.

Transport Policies in Singapore 221

Licence transferability meant that unsuccessful bidders with a high post-auctionvaluation could purchase a COE from a successful bidder with a low post-auctionvaluation. Car dealers who had been unsuccessful bidders could also purchaseCOEs from the secondary market and honour their sales contracts. Transferabilityimproved liquidity, specially for smaller car dealers with liquidity constraints,who may had decided not to tie up their funds in the deposit for a bid but purchasethe COE from the secondary market instead (Koh & Lee, 1994).

A transferable COE possesses an option value, giving the holder the choiceof using it to buy a car or selling it. Although in principle transferable COEshave a premium over non-transferable COEs, this is not necessarily always thecase. The transferability premium depends on how restrictive the quota is relativeto demand; a restrictive quota gives a positive transferability premium but anot-so-restrictive quota may give a negative transferability premium.

However, the public blamed rapidly raising quota premiums on the secondaryCOE market, encouraged by reports in local newspapers of the high profits beingmade by some speculators (the biggest group of speculators were the used-cartraders). The government, having initially maintained that transferability wasdesirable, bowed to public pressure and in October 1991 made Categories 1–4and Category 6 COEs non-transferable. At the same time the period of validitywas increased from three months to six months. Category 5 and 7 COEs remainedtransferable and valid for three months.

Transferability is unlikely to have been the cause of escalating quota premiums.As Fig. 1 shows, COE prices rose even more dramatically after the introduction ofnon-transferability, though this could be due to demand shifts such as rising incomeand increased highway capacity. Koh and Lee (1993) regress the quota premiumon a dummy variable and other variables such as the range of bids and the ratio ofsubmitted bids to successful bids. They find that the change to non-transferabilityled to lower quota premiums for Category 1 and higher quota premiums forCategories 3 and 4, and had no significant effect on premiums for Category 2.

Tan (2001) uses as regression variables the quota premiums for Categories1–4 relative to Category 5, which was not made non-transferable. The regressionalso includes a dummy variable, the quota level relative to Category 5 and thenumber of bids relative to Category 5. The paper finds that non-transferabilitylowered the quota premium for Categories 1–4. However, transferability providesfor flexibility in the final allocation of licences, and therefore carried economicbenefits for auction participants. Most of the participants in the COE auctionshave always typically been either car buyers who need a COE or car dealers whoare already committed to a sales contract, and neither of these would have anyintention of reselling their COEs. When transferability was still possible, theymight have changed their mind occasionally if prices in the secondary COE market

222 GEORGINA SANTOS ET AL.

were attractive, but this was unlikely as they would have had to bid again in thefollowing auction.

The main reasons for the escalating prices since the COEs were introducedhave been rising incomes and speculation. A major speculator could easily cornerthe market, since in each auction there were only 100–400 quotas for Categories1 and 3 and less than 100 for Category 4 (these are pre-1994 figures). Speculatorsthen had an incentive to submit higher bids to keep pushing the quota premiumup. Since the transferable COEs were valid for three months, the speculators wereguaranteed a profit as long as the quota premium kept rising (Koh & Lee, 1994).

When non-transferability was imposed in October 1991, car dealers and traderswere able to continue trading COEs, although in a round-about manner, by usingthe COE to purchase a vehicle (specified by the buyer) and immediately selling thevehicle with its COE. This was known as “double transfer.” A double transfer wassubject to a levy of 2% of the vehicle’s OMV. In April 1995, transfers of ownershipof a vehicle registered with a non-transferable Categories 1–4 COE were bannedin the first three months after purchase, and an additional levy was imposed ontransfers of ownership made within four to six months (Phang et al., 1996). Phanget al. (1996) find that the measures introduced in 1995 were effective in reducingthe quota premium. However, Fig. 1 shows that by late-1997 the quota premiumshad returned to the peak levels of 1995.

Many bidders were risk-averse or had an urgent need for a COE. These includedpeople who had de-registered their vehicles and needed a new car, dealers whohad to honour sales contracts, dealers who wanted to increase sales, and peoplewho simply did not want to risk being unsuccessful and having to bid in thefollowing auction, incurring interest, transaction costs and frustration (Koh &Lee, 1994). With the sealed-bid uniform auction format, these people submittedhigher bids than their true valuation in order to secure their COE, knowing that theclearing price would be lower. Competition was intense between car dealerships.Once a customer gave a deposit to purchase a car to be delivered by a certain date,the dealer had to obtain a COE on their behalf. Luxury car dealers, with their higherprofit margins, could afford to put in higher bids, which forced dealers of cheapcars to follow suit. Failure to obtain a COE not only meant the loss of an order, thedealer also incurred storage and capital costs for the unsold cars (Koh & Lee, 1994).

Furthermore, dealers of cheap cars could not turn to the secondary market intransferable open category COEs since they were typically priced at Category 4quota premium levels. Many car dealerships went out of business after the VQSwas introduced. In some cases the luxury car dealerships replaced the cheap cardealerships. Within a dealership, the more expensive cars were promoted morevigorously due to their higher profit margins. These factors pushed the quotapremium up and helped produce the shift in the vehicle mix towards luxury cars(Koh & Lee, 1994).

Transport Policies in Singapore 223

3.5. Auction Format

When it was initially implemented, the VQS was a uniform-price sealed-bidauction. Milgrom and Weber (1982) show that in a uniform-price auction wheresuccessful bidders pay the highest losing bid,15 bidders bid according to theirtrue valuation. This is because a bidder that bids higher than his true valuationmay end up having to pay a price higher than his valuation, whilst bidding lowerthan his valuation reduces the probability of winning but does not increase hisconsumer surplus when he wins.

In July 2001, the auction format was changed to open online bidding format.Potential bidders can now see the current market-clearing quota premium beforedeciding to whether participate in the auction or revise their bids. A bidder maychange his bid as often as desired once he is in the auction. Bids can be madethrough the Internet, phone or Automatic Teller Machines. Once submitted, abid cannot be withdrawn, and a bidder may only adjust his bid upwards, whichreduces volatility in the clearing price.

When there is a common-value element in the valuations of quota licences,an open auction improves transparency and offers an efficient price-discoveryprocess. Milgrom and Weber (1982) show that an open auction format is indeedpreferable when bidders’ valuations are affected by the information held by otherbidders. An open ascending auction provides a reliable process of price discovery,so that the allocation of licences goes to the bidders that value the COEs the most.However, more information may also facilitate collusive outcomes given the smallnumber of quotas in each category in each auction. Table 2 shows that the changeto open bidding in July 2001 was accompanied by lower average quota premiumsand volatility in all categories.

Table 2. Quota Premiums Before and After Change of Auction Format in July2001.

Cat A Cat B Cat C Cat D Cat E

Sealed biddingMean (S$) 36,685 38,345 22,140 1,084 38,738Standard deviation 9,208 10,939 7,321 361 8,305

Open biddingMean (S$) 29,876 30,434 18,263 324 30,650Standard deviation 2,432 3,787 5,241 296 3,786

Note: To give a fair comparison, only the period after the combining of categories in May 1999 isconsidered. Sealed bidding covers the period May 1999 to June 2001; open bidding covers theperiod July 2001 to June 2003.

Source:Land Transport Authority. www.lta.gov.sg.

224 GEORGINA SANTOS ET AL.

Although the introduction of open bidding has improved the transparency ofthe auction process, the system continues to suffer from the problems caused bysub-categorisation, the open category and non-transferability. In the first auctionof June 2003 for example, the price of a Category A COE was S$31,799 and theprice of a Category B COE was S$31,501. As a percentage of the car price, aCategory B licence continues to cost less than a Category A licence.

3.6. Summary of the Shortcomings

The need for sub-categorisation depends on the ultimate aim of the VQS. If theVQS is intended to ultimately reduce road usage, then since all cars occupy roughlythe same amount of road space and parking, an auction system that placed asimilar cost of owning a car on all cars, regardless of size, would be reasonable.Since cars of different engine sizes generate different amounts of emissions, froman environmental point of view, there is an argument for sub-categorisation, inaddition to the argument of social equity.

However, sub-categorisation to bring about social equity (so that buyers of smallcars do not pay the same premium as buyers of large cars) clearly conflicts withopen category to bring about flexibility. Sub-categorisation is aimed at segregatingbuyers, so that the quota premiums for smaller cars are lower than the premiums forlarger cars. On the other hand, the open category provides flexibility in the vehiclefleet composition and has the effect of helping to equalise the premiums of all thecategories. The relatively small quota of the open category has mostly been usedto purchase larger cars (Categories 3 and 4), leading to an increase in the supply ofquotas for larger cars over time and lowering their quota prices. The outcome hasbeen a highly regressive system. Buyers of cheap cars have had to pay more fortheir quotas relative to the car’s price, and pay more in absolute terms too at times.

There have been calls to scrap the open category and adjust the size of quotasfor each category in order to make the quota premiums for Category A COEscheaper than Category B COEs. It is doubtful however that equity will be easilyrestored even if the open category disappears. The government would need todetermine the demand for vehicles in each category and the volatility in quotapremiums would probably not be eliminated.

Alternatively, the government could consider raising the ARF and the annualroad tax rate. As the ARF is proportional to the vehicle’s OMV and the road tax isfully proportional to its engine capacity,16 raising both the road tax and the ARFwould go some way to restoring the equity in the system. As this would promptthe COE premiums to fall, it would also help to reduce the distortionary effectsof the VQS.

Transport Policies in Singapore 225

Koh and Lee (1994) propose an ad valorembidding system for the VQS,whereby bidders bid a percentage duty on the OMV of their desired vehicle,rather than nominal amounts. In a VQS ad valoremauction, only three categoriesfor cars, motorcycles and “goods vehicles and buses” would be required, andCOEs could be made transferable again. This was considered by the VQSreview committee but was rejected on the basis that it would make the system“unnecessarily complex” (Tan, 2001). However, ad valorembidding shouldsimplify the system and correct the problems created by sub-categorisation, theopen category and transferability. The current system is regressive with buyersof cheap cars paying a higher percentage of their car price for their COE thanbuyers of expensive cars. Under ad valorembidding, wealthy people buying amore expensive vehicle would pay more for their COE, and the quota licenceas a percentage of each successful bidder’s vehicle’s OMV would be the samefor everyone. Scrapping the open category would reduce the bias towards buyingmore expensive cars, and transferable COEs would improve efficiency given theuncertainty in the future value of a successful bidder’s COE.

4. ELECTRONIC ROAD PRICING

Despite the crudeness of the system and the labour-intensive manual enforcement,the ALS introduced in 1975 was effective and succeeded in reducing traffic con-gestion in the central area of Singapore, which contained the CBD, during themorning and evening peak hours.

As Chin (2002) explains, the main problems however were precisely linked tothe lack of any automation. Being manual the system was prone to error. Therewere different vehicle types, plus the option of part-day licences, leaving room formistakes. Enforcement was made at the entry points to the RZ and extending theALS would have meant hiring more personnel to enforce the system. It was alsodifficult to fine-tune it, as introducing shoulder-peak charges would have lead toeven more options and an even higher probability of mistakes by the enforcementofficers. Thus, there was always a rush to enter just before or after the restrictedhours that could not be smoothed by a shoulder-peak charge. Also, since a licenceoffered a vehicle an unlimited number of passages through the ALS control points17

there must have been some temptation for drivers to transfer their licences to otherdrivers, even though this was not legal. Introducing a manually controlled charge-per-pass would have delayed traffic as queues would have built at toll-booths.

There was also a perception that a paper based ALS scheme was becomingout-of-place in a city-state that was becoming high-tech and aspired to be regardedas such.

226 GEORGINA SANTOS ET AL.

Thus, in 1989 the government approved the use of ERP for Singapore roads(Chin, 2002). The main contract to implement the ERP system was a “Design andBuild” project awarded to a Consortium in October 1995 at about S$197 million(S$208 million at 2003 prices and US$116 million at 2003 exchange rates).18

From those S$197 million, about S$100 million covered the initial supply ofIUs given to motorists for free during a ten-month grace period that was givenat the time. As of 2003, motorists who wish to install an IU have to purchaseit at S$150 (US$84) inclusive of the installation cost. The IU has several partsincluding a one-inch liquid crystal display, a sealed battery, a card acceptor,and electronic components that can operate reliably in the Singaporean tropicalheat and humidity, with a five-year warranty. Foreign visiting vehicles can eitherrent IUs (at S$5 per day) and pay the variable ERP charges or buy an Autopasscard (at S$10, comprising a pre-loaded cash value of S$4 and a non-refundablecard cost of S$6) and pay a fixed ERP fee of S$10 for each day they drive onpriced roads.19

The remaining amount of S$97 million was for the design, development, supply,installation and one-year warranty of the ERP equipment, including the gantriesand central computer system. This computer system includes a powerful back-endsystem with three computer servers (with backups), monitoring systems and amaster-clock to ensure that the timing at all the ERP gantries are synchronised.All the transactions and violation images are processed there. The system isdesigned to support up to a maximum of 100 ERP points. However, as of 2003,there are only 44 gantries on the road.

A separate five-year maintenance contract of about S$39 million (S$41 at 2003prices and US$23 million at 2003 exchange rates) was also awarded together withthe main contract for the maintenance of the system.

The ERP system is managed by the Authority at an annual operating cost of aboutS$8 million (US$4.5 million). This however varies from year to year and comprisesmainly the salary of the staff and administrative expenses to manage the system.

As explained above, the ERP system in Singapore charges motorists each timethey pass through a gantry when the system is in operation. The ERP chargesare published on the Land Transport Authority (LTA) website.20 The currentscheme is a rather complex system with different charges according to vehicletype, time of the day, and location of the gantry. Police cars, ambulances andfire engines are all exempt. Charges for passenger cars, taxis and light goodsvehicles vary between S$0.50 and S$3, charges for motorcycles vary betweenS$0.25 and S$1.50, charges for heavy goods vehicles and light buses varybetween S$0.75 and S$4.50, and charges for very heavy goods vehicles and bigbuses vary between S$1 and S$6. In February 2003 a graduated ERP rate wasintroduced in the first five minutes of the time slot with a higher charge in order to

Transport Policies in Singapore 227

discourage motorists from speeding up or slowing down to avoid higher charges.For example, where the charge for passenger cars was S$2 between 8 and8.30 AM and S$3 between 8.30 and 9 AM, it is now S$2 between 8.05 and8.30 AM, S$2.50 between 8.30 and 8.35 AM, and S$3 between 8.35 and 8.55 AM,when it changes to S$2.

The charging area is divided into central business districts, where chargingapplies from 7.30 AM to 7 PM, and expressways/outer ring roads, where chargingapplies from 7.30 AM to 9.30 AM.

In principle, the ERP is able to charge the difference between marginal socialand marginal private costs of road use, or at least, approximate the differencemore accurately than a flat charge. The per-entry charge of the ERP has enabledthe use of lower charges in comparison to the per-day charges under the ALS.Even though the ERP charges were lower than ALS charges, Menon (2000) findsthat a year after its introduction, traffic volumes in the restricted areas had fallenby 15% during the whole day and by 16% during the morning peak. There wereonly minor increases in traffic at the times just outside the evening peak restrictedperiod. He attributes this reduction to the sharp fall in multiple trips, as couldbe expected with the change to per-entry charging. Traffic speeds in the CBDremained in the optimum of 20–30 km per hour (12.4–18.6 mi per hour) range.

5. INTEGRATED TRANSPORT POLICY

Apart from discouraging vehicle ownership and usage, Singaporean transport pol-icy has also included the provision of a wide range of attractive alternatives todriving.

The development of the MRT rail system was a strategic decision, linked tothe objective of building a world-class urban infrastructure to make Singaporea destination city to invest, work and live in. Phang (2003) argues that in othercities, the increase in land use density and land value along the rail corridor areusually accompanied by decreases in land use density and land value elsewhere.In Singapore, things are different. It is a small island where land is scarce andthe government effectively controls its use. The railway network expansion inSingapore has increased the intensity of land use not only in the CBD but alsoaround many suburban railway stations.

The Singaporean government provided the financing for the construction of theMRT rail system. During the 1980s and early 1990s, the government paid for thelong-term infrastructure and initial operating assets of the MRT project, includingrails, tunnels, stations, trains and signalling systems (Phang, 2003). The operatorwas required to generate enough revenue to cover operating and maintenance

228 GEORGINA SANTOS ET AL.

costs, including the replacement cost of operating assets (Phang, 2003). In 1996with the publication of the LTA (1996) White Paper, which recognised that thecurrent rules would require the imposition of higher fares, the government decidedto bear the burden of asset inflation. Existing and future railway projects wereonly required to generate enough revenues to cover operating and maintenancecosts, depreciation, and the historical replacement cost of operating assets(Phang, 2003). Hence in 2017, of the S$6.9 billion projected cost of replacingthe current set of operating assets, the operator will only pay S$1.6 billion(cost of the assets in 1987) and the government will pay S$5.3 billion. In 2047when the second set of operating assets are due for replacement, the operatorwill pay S$6.9 billion and the government will pay the asset inflation costs(Phang, 2003). This change paved the way for the subsequent approval of sevennew MRT/LRT projects, more than doubling the existing rail network length(Phang, 2003).

In other areas, the government has also been proactive to make public transport apalatable alternative to driving. For instance, in 1996, the government encouragedall three taxi operators to introduce Global Positioning System (GPS) technologyfor booking and dispatching taxis (LTA, 1996). Fares are allowed to vary accordingto time of day and demand, and can be set much higher during peak periods.

Similarly, in 1997, Singapore Bus Services introduced an “on time” GPSsystem to provide customers with accurate information about bus arrival times.In 2001 it launched a pilot project to install electronic display panels at majorbus stops to inform waiting passengers the expected arrival time of the next bus.Commuters could also telephone to find out the expected arrival time of the busat their stop (Goh, 2002). However, the project was soon halted in light of the softmacroeconomic conditions, and lukewarm response.

Two other developments in the 1990s deserve some mention. First introducedin 1975 but abandoned several months later, a Park-and-Ride scheme wasre-introduced in 1990 to encourage drivers to park their vehicles at suburban MRTstations and take the MRT to the central business district. After changes in 1992and 1995, the revived scheme is still successful (Foo, 1997b), in contrast withthe poor response it had in 1975. Then, it was a bus-based system, where driversparked their cars at designated car parks in the fringe of the CBD and then tookthe bus to work. Not only did it not solve the congestion problem, but also it wasinconvenient and cumbersome.

The government has also encouraged the formation of car cooperatives. Carsharing using a car cooperative scheme allows an individual to have access to acar and use it without having to own it. The first car cooperative in Singaporestarted in 1997. The idea of a cooperative is to operate and maintain a smallfleet of cooperative cars, which are parked at designated sites for the cooperative

Transport Policies in Singapore 229

members, who pay an annual membership in exchange for the right to use thesecars, plus the cost of their trips. Similar schemes for the U.K. and the U.S. arementioned in Chapters 11 and 12 of this volume respectively. Foo (2000) findsthat a majority of car cooperative members commute to work by public transportand use the cooperative car for other purposes such as leisure and social trips.Advantages of car cooperatives over renting cars include ease of access and theoption of hiring the car for only a few hours rather than per day.

Finally, in an effort to create smooth-flowing traffic, the government alsoinstalled a Green Link Determining (GLIDE) system in 1996, which coordinatestraffic signals to give more “green” time where traffic flow is heavy (LTA, 1996).The Expressway Monitoring and Advisory System was installed in 1998. Thesystem uses cameras to detect accidents and traffic conditions. The control centrealerts drivers of problems ahead through the television, radio and electronicsignboards placed along and at the approaches to expressways (LTA, 2002).

6. POLICY DEVELOPMENTS

Willoughby (2000) finds that since the 1960s, the revenues collected by the govern-ment from road transport have systematically exceeded government expenditureon road transport by between three and six times. In 2002 for example, only 19.8%of road transport revenues were spent on road transport. Revenues from roadtransport were 7.6% of total government revenues and expenditure on road trans-port was 2% of total expenditures.21 Even when government investment in MRTis included, transport revenues have typically exceeded expenditure on transport.

In 1995, 11% of Singapore’s land area was occupied by roads (LTA, 1996).At the time the government considered this to be an optimal share, since furtherincreases would have been at the expense of housing, offices, factories, etc.Although many Asian cities have a similar share, the share for most Europeancities is twice as high, and the figure for many North American cities is threetimes or more higher (Willoughby, 2001).

Hau (1992) states that road space should be added whenever the economicrent earned on a congested stretch of road exceeds the economic rent of havingalternative uses for the adjoining land. A policy that limits roads to 11% ofland area means that the decision to add new roads was not based on economicefficiency but rather on policy judgment and strategic intent.

The extremely high VQS quota premiums seen in the mid-1990s is evidencethat with increasing affluence and disposable incomes, Singaporeans wanted toown a car, even if they only made limited use of it. The ERP is able to chargeefficiently for the use of individual roads by setting appropriate fees for varying

230 GEORGINA SANTOS ET AL.

levels of congestion over the day. It has also brought about a rationalisation ofvehicle ownership and usage taxes: in 1998 the ARF was lowered and the annualgrowth rate of VQS quotas was increased by 9–12% (Toh & Phang, 1997). TheERP is now the centerpiece of the government’s transport policies, and is shiftingthe burden of taxation away from ownership taxes, with an aim to lowering the3:1 ratio of upfront and annual fixed costs against usage costs (LTA, 1996) and thehigh annual mileage of the average car. The government introduced these changesgradually to prevent a sharp decline in motor vehicle prices and a destabilisationin banks’ consumer lending, much of which used to be used for car purchases(Willoughby, 2000).

Following the LTA’s recommendation, the government has also allowed anincrease in the vehicle population, with the goal of increasing car ownership fromone car per ten people in 1995 to one per seven people by 2010 (LTA, 1996). TheERP charges have remained stable, but the ERP zones have been continuouslyexpanded.

Between 1996 and 2001, the road network increased by 415-lane-km (SingaporeMinistry of Transport, 2001), a figure much higher than the 225-lane-km targetedby the end of 2001, as stated in LTA (1996). Although this may suggest that thegovernment has relaxed the 11% land use policy, it should also be borne in mindthat Singapore land area is growing due to land reclamation from the sea.22 In2002, ARF charges were lowered, higher PARF discounts were given, and importduties were reduced from 31 to 20%. These recent developments indicate thatthe Singaporean government is sensitive to the people’s aspirations to own a car,and accordingly, has adjusted the transport policies to meet these aspirations.By shifting the cost of car ownership towards a usage-based charge under theexpanding ERP scheme, and with further investment in the road network, thegovernment has been able to cater for the expected increase in demand for roadusage without compromising its goal to control urban congestion.

7. CONCLUSIONS

The ERP scheme, the first of its kind in the world, has proven to be a flexibleinstrument that can price road usage, according to congestion conditions, time ofthe day, vehicle type and road. Since it was implemented in 1998, the charges havebeen fine-tuned to achieve the objective of maintaining traffic speeds within anacceptable range.

By controlling the number of new vehicle registrations each year, the VQS hasslowed the growth of the vehicle population. The introduction of the open-biddingformat has made the VQS more transparent compared with the previous sealed

Transport Policies in Singapore 231

format. Sub-categorisation however has prevented the system from achievingequity or efficiency. Luxury car buyers continue to pay a lower COE premiumrelative to the car price, the incentives are distorted, and this favours the purchaseof expensive cars. Non-transferability has also removed some flexibility in thesystem to adjust allocation in response to changing demand conditions. Furtherimprovements can be made to the VQS.

One solution proposed in this chapter is to raise the annual road tax. Thiswould weaken the impact of the VQS by reducing its quota prices, but it wouldretain the present auction format. Although eliminating the open category wouldprobably reduce the bias towards buying more expensive cars (as open categoryCOEs effectively supplement Category B COEs at present), it would not resolvethe problem of equity. Perhaps a simpler way to address the equity problemwould be to change to an ad valorembidding system, as suggested by Koh andLee (1994). This would simplify the system and correct the distortions. Under advalorembidding, buyers of expensive cars would pay more for their COE, but thequota licence as a percentage of each successful bidder’s vehicle’s OMV wouldbe the same for everyone. Although licence transferability has led to speculationin the past, some degree of transferability may be desirable in order to inject someflexibility in the final allocation, and improve efficiency.

Singapore’s integrated transport policy has been largely successful in bothcontrolling urban congestion and meeting the people’s aspirations to own a car.Most people day-to-day commute by public transport, and if they have a car, itis used mainly for leisure and social trips. By investing in public transport (MRT,taxis and buses) and encouraging its use, the provision of public transport hasbecome profitable. This has produced a virtuous cycle as further improvementsin infrastructure are being made to enhance the quality of services. Without thehands-on involvement of the government in driving the various transport policies,there would have been a vicious cycle with increased car usage, deteriorating qual-ity of bus and rail services, worsening congestion and a sharp drop in the quality ofpublic transport.

The 10-yearly Household Expenditure Survey in both 1983 and 1993 findsthat private vehicle ownership in Singapore has been a privilege of the rich,with the richest 10% of households accounting for 30% of total expenditure oncars (Willoughby, 2000). Although this is a reflection of the high cost of vehicleownership, the government has ensured that the less wealthy households haveaccess to convenient, frequent, reliable and safe public transport.

In many cities in both the developed and developing world, traffic congestionthreatens to seriously hinder the functioning of businesses. The success of roadpricing in Singapore serves as model for other countries. Even less sophisticatedsystems, such as the one implemented in London in February 2003, reduce traffic

232 GEORGINA SANTOS ET AL.

congestion and increase economic efficiency. A lesson for London and othercities around the world that might introduce congestion charging is that the ERPin Singapore has shown that a per-entry charge is more effective at reducingcongestion than a per-day charge.

To conclude, Singapore’s experience with vehicle quotas and electronic roadpricing has been successful and this gives hope to other towns and cities around theworld. Most importantly, the Singaporean government has managed to convincethe motoring public that controlling congestion is essential for an economyto function efficiently. Though many suggest that the one-party democracy inSingapore is uniquely able to carry out its policies decisively, Singaporeans ingeneral understand the rationale and support the government’s transport policies,perhaps because these constitute an integrated package that increases the costs ofdriving but at the same time makes the option of public transport attractive (Chu& Goh, 1997). If the policies had only been aimed at raising the costs of owningand using cars they would have probably failed. This is where the Singaporeanexample can claim its major victory: the support from the public thanks to itsintegrated approach to transport policy.

NOTES

1. This was enhanced at the beginning of 1991 with the introduction of an automaticticketing system, which consisted of a stored value card that could be used to pay for theMRT and for public buses. In April 2002 a smart-card ticketing system replaced the storedvalue one.

2. LRT is less costly and takes half the time to build (3–4 years) compared to the MRT.3. The OMV is essentially similar to Cost plus Insurance and Freight (CIF). It includes

purchase price, freight, insurance and all other charges incidental to the sale and deliveryof the car to Singapore.

4. Land Transport Authority (LTA) website: http://www.lta.gov.sg.5. It also recommended usage restraint measures and upgrading and expansion of the

road infrastructure.6. The introduction of the VQS was accompanied by a reduction of the ARF rate, from

175% in 1990 to 150% in 1991. The ARF has however remained high and as of 2003 itstands at 130%.

7. Initially the period of validity was six months but this was changed to three monthsfrom the second auction in August 1990. This was changed back to six months forCategories 1–4 and 6 in October 1991, when the transferability of the COEs was removed.

8. As originally intended, a Category 7 COE could also be used to purchase a motorcycleat one third of the quota premium, but given the wide disparity in the quota premiums forCategory 7 and Category 6 (motorcycles and scooters), this flexibility was seldom, if atall, utilised.

9. Apart from paying tolls under the ERP system, the smart cards can also be used to payfor parking. Many privately run car parks are equipped with automatic sensors so that when

Transport Policies in Singapore 233

vehicles equipped with IUs and smart cards enter the car park the sensor detects the smartcard, and when they leave a charge is automatically deducted according to the time thecar was parked. Other uses include payment of National Library Board fees and purchasesfrom retail shops that support cash card payments. These cards can store up to S$500.

10. These estimates were produced by the Centre for Transportation Studies at NanyangTechnological Institute, Singapore (1991) and are still used by the government whensetting the level of fees (Chin, 2002).

11. Here as in all other instances where the figures given are not referenced, the sourceis the LTA website, http://www.lta.gov.sg.

12. This has always been the case for the VQS. Theoretically, a large quota for the opencategory would, by arbitrage, equate the quota premiums of all the categories and defeatthe point of sub-categorisation.

13. This time period was chosen for the analysis here because in May 1999, Categories1 and 2 were combined and Categories 3 and 4 were also combined.

14. At the same time, Categories 5, 6 and 7 were simply renamed Categories C, D and E.15. In the VQS, the successful bidders pay the lowest successful bid. With many

bidders in the market so that bids are continuously distributed, this is analyticallyequivalent to paying the highest rejected bid. All bidders would be expected to bid theirvaluation and everyone gets a surplus except the lowest winning bidder who gets zerosurplus.

16. In Singapore the road tax structure is split into categories by engine size but withineach category there is a sliding scale. Source: LTA website: http://www.lta.gov.sg.

17. In the case of the RPS on the expressways, the licences also offered an unlimitednumber of passages through the RPS manually controlled points.

18. All the data on costs in this section were kindly provided by the Land TransportAuthority.

19. This fixed S$10 fee was computed by the Land Transport Authority by taking intoaccount the cost of renting a temporary IU and the average ERP charges incurred by aforeign motorist each day.

20. http://www.lta.gov.sg/.21. Government expenditure on road transport in Singapore in 2002 was S$379 million,

out of total expenditures of S$19,344 million; government revenues from road transportwere S$1,559 million from vehicle taxation plus S$359.3 million from vehicle importduties, out of total revenues of S$25,401 million (Customs and Excise, 2003; Departmentof Statistics, 2003).

22. The area increased from 633 km2 in 1990 to 683 km2 in 2003 (equivalent to agrowth of 8.2%).

ACKNOWLEDGMENTS

The authors are indebted to Kian Keong Chin, Eddie Lim Sing Loong, GeraldineChan and the Land Transport Authority in Singapore, for provision of useful infor-mation, and to David Newbery, Martin Richards and Piotr Olszewski for helpfulcomments on an earlier draft. Any remaining errors are the authors’ own. Supportfrom the British Academy for Georgina Santos is gratefully acknowledged.

234 GEORGINA SANTOS ET AL.

REFERENCES

Bhagwati, J., & Srinivasan, T. (1969). Optimal intervention to achieve noneconomic objectives.Review of Economic Studies, 36, 27–38.

Centre for Transportation Studies, Nanyang Technological Institute, Singapore (1991). Speed-flowmodel for traffic within the restricted zone.Nanyang Technological Institute.

Chin, K. K. (2002). Road pricing: Singapore’s experience. Essay prepared for the third seminarof the IMPRINT-EUROPE Thematic Network: Implementing reform on transport pric-ing: Constraints and solutions: Learning from best practice. Brussels, October 23–24.http://www.imprint-eu.org/public/Papers/IMPRINT3 chin.pdf.

Chu, S., & Goh, M. (1997). The price of car ownership in Singapore: An empirical assessmentof the COE scheme. International Journal of Transport Economics, 24, 457–472. Cited inWilloughby (2001).

Customs and Excise, Singapore (2003). Revenue collected by Singapore customs department.http://www.gov.sg/customs/trade/trde12 3.html.

Department of Statistics, Singapore (2003). Economic survey of Singapore: First quarter, 2003.http://www.singstat.gov.sg/.

Falvey, R. E. (1979). The composition of trade within import-restricted product categories. Journal ofPolitical Economy, 87, 1105–1114.

Foo, T. S. (1997a). An effective demand management instrument in urban transport: The area licensingscheme in Singapore. Cities, 14, 155–164.

Foo, T. S. (1997b). Experiences from Singapore’s park-and-ride scheme (1975–1996). HabitatInternational, 21, 427–443.

Foo, T. S. (2000). Vehicle ownership constraints and car sharing in Singapore. Habitat International,24, 75–90.

Goh, M. (2002). Congestion management and electronic road pricing in Singapore. Journal ofTransport Geography, 10, 29–38.

Hau, T. D. (1992). Economic fundamentals of road pricing: A diagrammatic analysis. World BankPolicy Research Working Paper Series 1070, The World Bank, Washington DC.

Holland, E. P., & Watson, P. L. (1978). Traffic restraint in Singapore. Traffic Engineering and Control,19, 14–22.

Koh, W., & Lee, D. (1993). Auctions for transferable objects: Theory and evidence from the vehiclequota system in Singapore. Asia Pacific Journal of Management, 10, 177–194.

Koh, W., & Lee, D. (1994). The vehicle quota system in Singapore: An assessment. TransportationResearch A, 28, 31–47.

Koh, W. (2003). Control of vehicle ownership and market competition: Theory and Singapore’sexperience with the vehicle quota system. Transportation Research A, 37, 749–770.

Krishna, K., & Tan, L. H. (1997). A note on India’s MFA quota allocation system: The effect ofsub-categorisation. Annales d’Economie et Statistique, 47, 39–50.

Krishna, K., & Tan, L. H. (1999). Transferable licences vs. non-transferable licences: What is thedifference? International Economic Review, 40, 785–800.

Land Transport Authority (1996, January). A world class land transport system. White Paper presentedto the Singaporean Parliament. Singapore.

Land Transport Authority (2002). Annual report 2001–2002: Riding the next wave. Singapore.http://www.lta.gov.sg/AbtLTA/pdf/annual LTA.pdf.

Li, M. Z. F. (1999). Estimating congestion toll by using traffic count data – Singapore’s area licensingscheme. Transportation Research E, 35, 1–10.

Transport Policies in Singapore 235

McCarthy, P., & Tay, R. (1993). Economic efficiency vs. traffic restraint: A note on Singapore’s arealicence scheme. Journal of Urban Economics, 34, 96–100.

Menon, A. (2000). ERP in Singapore – A perspective one year on. Traffic Engineering and Control,41, 40–45.

Milgrom, P., & Weber, R. (1982). A theory of auctions and competitive bidding. Econometrica, 50,1089–1122.

Mohring, H. (1983). The Singapore MRT: What price a large central area work force? Suara Ekonomi,20, 9–15. Cited in Phang (2003).

Olszewski, P., & Turner, D. (1993). New methods for controlling vehicle ownership and usage inSingapore. Transportation, 20, 355–371.

Phang, S. Y. (2003). Strategic development of airport and rail infrastructure: The case of Singapore.Transport Policy, 10, 27–33.

Phang, S. Y., & Chin, A. (1990). An evaluation of car ownership and car usage policies in Singapore.Report of the Select Committee on Land Transportation Policy, 105–117. Cited in Toh andPhang (1997).

Phang, S. Y., & Toh, R. S. (1997). From manual to electronic road congestion pricing: The Singaporeexperience and experiment. Transportation Research E, 33, 97–106.

Phang, S. Y., Wong, W. K., & Chia, N. C. (1996). Singapore’s experience with car quotas: Issues andpolicy processes. Transport Policy, 3, 145–153.

Singapore Ministry of Transport (2001). Parliament questions & answers: World class transport system.http://www.mot.gov.sg/newsroom nav/parliament matters/pq worldclasstransport.html.

Tan, L. H. (2001). Rationing rules and outcomes: The experience of Singapore’s vehicle quotasystem. IMF Working Paper 136, Washington DC. www.imf.org/external/pubs/ft/wp/2001/wp01136.pdf.

Toh, R. S. (1992). Experimental measures to curb road congestion in Singapore: Pricing and quotas.Logistics and Transportation Review, 28, 289–317.

Toh, R. S., & Phang, S. Y. (1997). Curbing urban traffic congestion in Singapore: A comprehensivereview. Transportation Journal, 37, 24–33.

Wilbur Smith & Associates (1974). Singapore mass transit study, phase 1. Final report. Wilbur Smithand Associates, Singapore. Cited in Foo (2000).

Willoughby, C. (2000). Singapore’s experience in managing motorization, and its relevance toother countries.World Bank Transport Division TWU-43. www.worldbank.org/transport/publicat/twu 43.pdf.

Willoughby, C. (2001). Singapore’s motorization policies 1960–2000. Transport Policy, 8, 125–139.