LinksGlossaryMessagesSitemapHelp


Home

Policy Instruments

Select
Search
Filter

Public transport services


SummaryTaxonomy and descriptionFirst principles assesmentEvidence on performancePolicy contributionComplementary instrumentsReferences

Evidence on performance

Case study 1: Mass Transportation Demonstration Projects in Massachusetts (1962-64)

Context

Between 1962 and 1964 the Mass Transportation Commission of the Commonwealth of Massachusetts (MTC) conducted a number of mass transit service improvement and fare reduction experiments. The experiments centred around Boston and its inner suburbs, and involved bus operators, other than the Massachusetts Transit Authority (MTA), throughout the states and commuter railroads that served Boston.

The bus experiments mostly involved increasing the service frequencies of a number of local bus services, these are reported in the table below. The rail experiments were carried out on the Boston & Maine Railroad (B&M), the New Haven Railroad (NH), with the New York Central Railroad (NYCR) used as a control. The B&M experiment incorporated three phases,
  1. 77% equivalent increase in all services (92% increase in weekday services) & a 28% equivalent reduction in fares (ranging from 12% to 72%).
  2. Retention of phase 1 service improvements and virtual elimination of fares reduction (except for an off-peak reduction).
  3. Service levels adjusted and fare levels remain the same.
The NH experiment had 2 phases,
  1. Total average service level increased by 42% and fares reduced by 10% on average.
  2. Service levels and fares returned to pre-experiment levels, with off-peak fare incentives retained.

Impacts on demand

Bus Company Experiments

The results in terms of revenue and ridership changes are shown in the table below "Massachusetts Bus Headway Changes and Ridership/Revene Results". In all but one case they demonstrated positive increases in revenue and ridership.

Massachusetts Bus Headway Changes and Ridership/Revenue Results

Route

Service Area Population

New Headway

Results & Comments

Implied Service Elasticity

Milford to Downtown Boston

22,000 (suburban area only)

1 hour all day (78% service increase)

12 month revenue up 22% (18% first 3 months; 27% in the last 3 months)

+0.28

Uxbridge to Worcester (pop. 187,000)

28,000 (suburban area only)

Similar to above

9 month revenue up 5% (none in first 3 months, 16% in the last 3 months)

+0.06

Amesbury –Newburyport 25,000 Half hourly in the peak; hourly in the base (67% service increase) 8 month revenue up 19% (route through depressed industrial areas) +0.28
Adams – Williamstown 40,000 Better that hourly frequency (100% service increase) 3 month ridership up 48% +0.48
Pittsfield 74,000 (SMSA*) Service increased to 8 round trips (16% service increase) 3 month ridership up 87% (3 mile long radial route) +5.44
Pittsfield 74,000 (SMSA) Service increased to 15 round trips (50% service increase) 3 month ridership up 30% (3 mile long radical route) +0.6
Fitchburg –
Leominster
72,000 (SMSA) 1:40pm to 6.00pm bus trips doubled to give 10 min. headway all day; minor route extension 8 month revenue up 8% (high density service area; fare increase from 20 cents to 25 cents in 9th month) n/a

Fall River

124,000 (SMSA)

Service increase of 20%

Halted but did not reverse ridership decline (high unemployment and disruptive construction)

n/a

* SMSA - US Census Standard Metropolitan Statistical Area (1960); n/a - not applicable

Source: Adapted from TRB (2003)

The implied service elasticities have a wide range, however two thirds of the elasticities appear in the +0.28 to +0.6 range.

Rail Experiments

The results in terms of revenue and ridership changes are shown in the table below,"Masachusetts Rail Headway Changes and Ridership Results". The key figures to note in terms of service elasticities are those presented for phase 2 when fares were increased. The increase in patronage was taken by MTC to infer that improvements in service levels were more effective at increasing ridership than were fare reductions. Overall the additional revenues covered the full incremental cost of the experiment.

Massachusetts Rail Headway Changes and Ridership Results

Rail System Phase 1 Phase 2 Phase 3
B&M +27% increase in ridership +37.5% increase in ridership +44% increase in ridership
NH +10% increase in ridership +11.5% increase in ridership n/a
NYCR -5.9% decrease in ridership -5.9% decrease in ridership -5.9% decrease in ridership

Source: Adapted from TRB (2003)

Surveys indicated that the majority of passengers on the commuting trains used to travel by the following modes, own car (63.6%), carpool member (16.9%), and bus (19.5%).

Impacts on Supply

No cost figures were reported in the TRB publication (2003), however the increase in service levels is likely to have resulted in additional costs from the purchasing/leasing of additional vehicles and hiring of additional operating staff.

Contribution to Objectives

Objective

Contribution to objective

Comment

Efficiency

Evidence indicated that a reduction in car use is likely to have contributed to an efficiency improvement by reducing road congestion.

Liveable streets

The reductions in car use are likely to have contributed to a liveability improvement.

Protection of the environment

The reductions in car use will have contributed to a reduction in environmental impacts.

Equity and social inclusion

Whilst no direct evidence was presented the increase in services is assumed to have had a sizeable impact upon equity and social inclusion.

Safety

The reduction in car use will have contributed to a reduction in accidents.

Economic growth

0

The generalised cost of travel by public transport will be reduced directly by the improved service level. Furthermore, mode switch from car may reduce congestion levels so leading to further reductions in travel time. These two impacts may increase productivity. On the other hand increased subsidy was necessary and the requisite increase in local taxes may stifle economic growth.

Finance

No information given for the bus elements of the experiment but the low elasticities suggest that costs will not have been covered. The higher elasticities for the rail services meant that increased revenue covered the cost of the increased services.

1 = Weakest possible positive contribution, 5 = strongest possible positive contribution
-1 = Weakest possible negative contribution -5 = strongest possible negative contribution
0 = No contribution

Top of the page



Case Study 2: Frequency and Service Hours Enhancements in Santa Clarita, California

Context

Santa Clarita is an outlying suburb of Los Angeles in California (U.S.) with a population of around 150,000 that is served by a metrolink commuter rail service and local bus coverage. Between 1992 and 1998 there were significant extensions to public transport service hours and service frequencies. These are outlined below:
  • 1992 - Saturday services expanded by 3 hours, i.e. larger operating period.
  • 1992 - Weekday service hours expanded by 2 hours, i.e. larger operating period.
  • 1994/5 - New express commuter bus services added.
  • 1995 - Weekday services expanded on three routes.
  • 1995/8 - 30 minute all day headways introduced on 4 routes (including 2 on a weekend) and 15 minute peak headways on two routes.
In addition a 90 minute pass was introduced in 1992, fares raised by 33% in 1993, youth passes rose to $15 from $10 in 1996, and Sunday services introduced on about two thirds of local routes.

Impacts on demand

The increase in both bus miles and bus hours over the five years in question has seen a greater than proportionate rise in ridership. The lack of statistical smoothing of short run anomalies however, means that not much weight can be placed upon the yearly elasticities. More reliance can be placed on the long run elasticities which are +1.14 for bus miles and +1.55 for bus hours.

Santa Clarita, CA Local Fixed Route Performance and Log Arc Service Elasticities

Local Fixed
Routes-Year

City
Population

Annual
Bus Hours

Annual
Bus Miles

Annual Bus
Rides

Bus Hours
Elasticity

Bus Miles
Elasticity

FY 1992-93
1993-94
1994-95
1995-96
1996-97
1997-98
123,400
124,000
124,300
124,800
n/a
n/a
48,778
53,391
60,028
62,750
66,947
81,216
787,807
1,018,021
1,163,607
1,179,140
1,389,082
1,569,891
769,137
915,869
1,107,587
1,366,537
1,527,253
1,693,173
-
+1.93
+1.62
+4.74
+1.72
+0.53
-
+0.68
+1.42
+15.84
+0.68
+0.84
5 fiscal years +2% (4 yrs*) +66% +99% +120% +1.55 +1.14
    FY - Financial Year: * - Calendar years 1992 (122,949 population) through to 1996 (125,153 population)

Source: TRB (2003)

Impacts on Supply

No cost figures were reported, however the increase in service levels is likely to have resulted in additional costs from the purchasing/leasing of additional vehicles and hiring of additional operating staff.

Contribution to Objectives

Objective

Contribution to objective

Comment

Efficiency

No direct evidence was provided, however it is likely that some modal shift has occurred, reducing congestion costs and improving efficiency. The generalised cost of bus travel will have also declined.

Liveable streets

No direct evidence was provided, however it is likely that some modal shift has occurred and that this has led to an improvement in liveable streets.

Protection of the environment

No direct evidence was provided, however it is likely that some modal shift has occurred, leading to a reduction in environmental externalities.

Equity and social inclusion

The extension of service will have opened up a wider range of services, goods and opportunities to those on low incomes.

Safety

No direct evidence was provided, however it is likely that some modal shift has occurred, which is likely to have lead to a reduction in accident rates.

Economic growth

No direct evidence was provided but it is likely that the improved service levels with no increased subsidy requirement will have been beneficial for the local economy.

Finance

The high service level elasticities suggest that increased fare revenue will have more than covered the cost of increased service provision.

1 = Weakest possible positive contribution, 5 = strongest possible positive contribution
-1 = Weakest possible negative contribution -5 = strongest possible negative contribution
0 = No contribution


Top of the page



Further Evidence on service level Elasticities

Public Transport Elasticities: Time For A Re-Think (Preston, 1998)

Preston (1998) used evidence taken from an English Metropolitan area to estimate short and long run bus service elasticities over various time periods. The data had been collected by the area Passenger Transport Executive (PTE) using continuous on-vehicle surveys and consisted of 4 weekly bus usage data by ticket type and time of day (3,822 observations) covering the periods 1987/88 to 1992/93.

The estimated service elasticities are presented in the table below “Elasticity Estimates for Adult Bus Users in a Metropolitan Area".

Elasticity Estimates for Adult Bus Users in a Metropolitan Area

Time Period Service Short Run Service Long Run
Early morning +0.38 +0.56
Peak am +0.36 +0.58
Inter-peak +0.17 +0.30
Peak pm +0.32 +0.42
Late +0.35 +1.95
Saturday +0.52 +0.67
Sunday +1.05 +1.67
Source: Preston (1998)

The service elasticities appear to be intuitively correct for the peaks (inelastic), and the inter-peak, which is more elastic. Early morning trips are also very inelastic, which probably reflects the dominance of work trips at that time of day. The service elasticities on Sundays and late at night are very high in the long run, however, the confidence interval around these estimates is also very large, suggesting they are unreliable.

Further evidence on service level elasticities (The Demand for Public Transport: a practical guide, 2004)

The service level elasticities presented in the two tables below have been derived from a number of studies conducted throughout the UK for bus services.

Service elasticities, with the range and standard deviation according to average values – bus (TRL 2004)

Time period Elasticity Range Standard deviation Number of measurements
Short run 0.38 0.10 to 0.74 0.135 27
Long run 0.66 0.22 to 1.04 0.275 23

Service elasticities, with the range and standard deviation according to average values – rail (TRL 2004)

Time period Elasticity Range Standard deviation Number of measurements
Time period not stated 0.49 0.33 to 0.65 0.23 2
Short run 0.75 0.65 to 0.90 0.13 3

The table below presents price and service level elasticities for rail demand in Spanish cities. It is clear that customers are more responsive to service level changes than price changes.

Price and service level elasticities for Spanish cities - bus (Arsenio 2000) (taken from TRL 2004)

Elsaticity Large cities Small cities
Price -0.3 -0.32
Service quantity (train km) 0.78 0.39

The table below presents elasticities with respect to wait time (a function of headway) for bus based on analysis of data in 23 UK towns.

Elasticities with respect to wait time – bus

Dependent variable Time period/destination Elasticity with respect to wait time
Total trips   -0.64
Adult trips   -0.74
Adult trips Peak/town centre -0.65
Adult trips Off-peak/town centre -0.85
Adult trips Peak/other -0.39
Adult trips Peak/town centre -1.17
Total trips Peak/town centre -0.64
Total trips Off-peak/town centre -0.64
Total trips Peak/other -0.50
Total trips Peak/town centre -1.05

The Transportation Research board interim handbook (Pratt et al, 2000) found the following fare and service elasticities in the period 1985 to 87 when fare and service changes were introduced.

Fare and service elasticities – Dallas

  Fare elasticity Service elasticity
Urban bus -0.35 0.32
Suburban express bus -0.26 0.38
Suburban local -0.25 0.36

The isotope research study (European commission, 1997) reported Service elasticity is for bus in a number of European cities by city sides (small: population less than 500,000)

Service elasticities for bus in European cities

  Small city Large city
Service elasticity 0.33 0.49

Gaps and Weaknesses in the Evidence

Changes in service levels are likely to impact upon motorists and other travel and even location decisions in the very long-term. The difficulty is that the greater the length of time period that I studied them the greater the number and magnitude of confounding factors. Long-term impacts are therefore very difficult to derive despite the fact that they may well be very much greater than short to medium-term impacts.

It was difficult to find completed studies looking at the quality aspect of public transport services. There was information readily available about various schemes such as quality bus partnerships; however results on the impacts proved to be limited.

Top of the page


Text edited at the Institute for Transport Studies, University of Leeds, Leeds LS2 9JT