3. Travel demand impacts — diversion rates

3.1 Introduction

To value the benefits of a TBhC project, it is necessary to have estimates of the impacts that the project is likely to have on travel demand, including impacts on mode shares, average trip lengths, and any changes in the overall amount of travel.

Diversion rates are the (quantitative) estimates of the differences or changes in travel on various modes between the Base Case (without the initiative) and Project Case (with the TBhC initiative). Diversion rates are expressed as changes in mode share, with decreases for some modes (private car) and increases for others (public transport and cycling/walking).

It is also necessary to define the target population to which the diversion rate applies. For example, for a workplace travel plan this might be the total number of people employed at the workplace; for a household/community based initiative, it might be either the total population of the area covered by the initiative or just the households actually contacted or agreeing to participate in the program.

Diversion rates should be based on evidence from similar previous TBhC initiatives that have been implemented and then monitored and evaluated subsequent to implementation. Robust monitoring and evaluation of TBhC initiatives is difficult and expensive. Nevertheless, some studies have now been conducted that provide a fair indication of the likely effects of different types of TBhC initiatives in different situations. Default diversion rates based on this evidence are provided in Section 3.4.

3.2 Diversion rate interpretation issues

A significant issue in relation to diversion rates is the different ways that they are expressed and the different target populations against whom the changes are measured. If diversion rates are derived from relevant local studies, rather than the default rates in this guidance, care is required to identify the approach that has been adopted in each study.

If a diversion rate that was derived as a change in the travel behaviour of the people participating in a TBhC project (which could be quite high) is assumed to apply to the whole population in an area when forecasting the impacts of a TBhC project, the effects are likely to be over-estimated. It is important to be clear what target population the diversion rate applies to, both when collating diversion rates from the literature and when using appraisal procedures to assess potential projects. The different target populations against which diversion rates have been expressed in appraisals and monitoring include:

  • Total population in the suburb/area covered by a TBhC project
  • Total households in the suburb/area covered by a TBhC project
  • Total population or households to be contacted by the project
  • Total number of people actually participating in TBhC project
  • Total roll of a school covered by a school travel plan
  • Total number of students participating in project
  • Total number of employees in a workplace
  • Combined total of employees and visitors for a hospital.

Diversion rates are sometimes expressed as the percentage point change in mode share and sometimes as the percentage change from the initial mode share. As an example, if the car mode share of all trips (total trips by all modes) is 66 per cent without the TBhC project and 62 per cent with the project, this might be expressed as either a 4.0 percentage point reduction in car mode share or as a 6.1 per cent ((66 – 62) / 66) reduction from the initial car mode share.

Diversion rates are also sometimes expressed as the percentage change in a variable such as car trips or car vehicle kilometres travelled (VKT) compared with the situation that would have occurred without the TBhC project.

Care is required to avoid misinterpreting and/or misapplying diversion rates.

3.3 Diversion rate evidence

Most of the evidence on the effects of TBhC initiatives is drawn from a literature review conducted during the development of New Zealand’s travel behaviour change appraisal procedures in 2004. That study collated diversion rates from a range of projects in Australia, New Zealand and the United Kingdom. This has been supplemented with evidence from subsequent Australian monitoring and evaluation studies.

The observed diversion rates were often represented as a percentage change in the number of trips, or percentage point change in mode shares, but some were also in terms of an increase or decrease in vehicle kilometres travelled.

Some of the diversion rate information was taken from ex-ante assessments rather than from monitoring of projects that have been implemented. However, these tended to be based on previous reports and pilot studies and therefore still reflect actual experience.

Substantial investment was made in TBhC projects in Australia in the first half of the 2000s, particularly in Western Australia, South Australia, Victoria and Queensland.

A paper at the Australasian Transport Research Forum (Roth et al, 2003) presented results achieved by Individualised Marketing programs from around the world. Reductions in car driver trips ranging from 6 to 14 per cent were observed in the various programs. Based on these findings, the paper suggested that reductions in car driver trips could range from 5.5 to 13 per cent of whole populations (that is, including non-participants). The paper also noted European evidence suggesting that little to no maintenance is required for five years to maintain public transport patronage increases.

Another paper presented at the same forum (Stopher et al, 2003) reported the findings of a 'critical appraisal' of travel behaviour modification programs in Australia. This indicated that the reduction in car driver trips was in the order of 7 to 9 per cent for those who participate in a project and 5 to 7 per cent for a target population as a whole (allowing for non-respondents). UK evidence summarised in the Smarter Choices report (Cairns et al, 2004) suggested a lower range of 1 to 5 per cent reduction in all car driver trips over 10 years.

The most extensive monitoring and evaluation project in Australia in recent years was conducted by the Institute of Transport and Logistics Studies (ITLS), which is also reported in an ATRF paper (Stopher et al, 2013).

During the four-year period from 2004 to 2007, various TBhC programs were implemented in South Australia, Victoria, Queensland and the ACT using social marketing and community development initiatives. The ITLS study measured the travel patterns of a number of households in these states over a five-year period from 2007 to 2012, following the completion of the TBhC programs. GPS devices were used to measure household members’ travel over a 15 day period in September to November each year. The survey covered roughly 120 households per year, of which approximately 80 had participated in TravelSmart and 40 households had not participated (the control group). Each year, some households declined to continue participating and were replaced with new households to maintain the numbers in each state and in the participant and control groups.

The ITLS study concludes that the aggregate analysis of the six waves of long term data (one before TravelSmart implementation and one each year for five years after implementation) indicates that there has been a continuing decrease in total person kilometres of travel (PKT) by car over the five-year monitoring period for both TravelSmart and non-TravelSmart households. Non-TravelSmart households performed consistently more PKT per day than their TravelSmart counterparts, and the difference between the two remained more or less the same throughout the monitoring period. This suggests that TravelSmart households succeeded in reducing PKT by car during the implementation of this intervention, and then maintained their lower level of driving though the long-term monitoring. There is no evidence of a return by the overall sample to levels of driving that match those prior to the TravelSmart intervention. Presumably other factors have led to a continuing decrease in PKT by car for all households in the areas surveyed in subsequent years.

A limitation of the study report is that it does not show what proportion of the decrease in PKT by car was matched by an increase in travel by other modes and what proportion was simply a reduction in the amount of travel. This is important in determining the net benefits of a TBhC initiative.

The study also observed that the measured reductions in PKT almost certainly include some level of change due to other reasons. TravelSmart non-participants decreased their vehicle kilometres travelled by 15.3 per cent over the five years while participants decreased their vehicle kilometres travelled by 18.2 per cent. Almost half of the reduction for both participants and non-participants occurred in the last two years of the study, well after the TravelSmart programs had been delivered. The difference between these results suggest that TravelSmart household programs may achieve an approximate (non-statistically reliable) one-off reduction in car travel of approximately 2.9 per cent and maintain this over the medium term (five or more years), but that other factors appear to have had a much greater effect on changes in travel over the monitoring period.

Prior to the above five-year monitoring study, the ITLS performed a shorter duration TravelSmart monitoring study, also using GPS data logging, for the South Australian Department of Transport, Energy and Infrastructure (Stopher et al, 2009). This was expanded into the subsequent larger five-year study.

A conclusion from the above evidence is that the long term effect of TBhC household/community initiatives is probably lower than some of the ranges suggested in the early years and is more in line with the ranges noted in the United Kingdom. Another conclusion is that the changes achieved by TBhC programs do appear to persist over the medium to long term.

The UK Smarter Choices report (Cairns et al, 2004) presented a range of diversion rates for 10 different types of TBhC projects surveyed in that report. It also provided diversion rates for the combined effect of overlapping TBhC initiatives recognising the potential to double count impacts if diversion rates are simply added together.

Workplace travel plans in the UK achieved reductions of 5 to 9 per cent in all car trips to/from work in the area and appear to be able to achieve larger reductions due to being tailored to particular workplaces and hence able to influence a greater proportion of the population (the workforce at the workplace in this case).

A significant issue is the range of diversion rates that can apply for the same type of TBhC project in different situations. For example, in the case of workplace travel plans, where there is a considerable amount of experience and evidence, the UK Smarter Choices report found that broadly:

  • 10% of travel plans achieve no change
  • 20% reduce car use by >0 – 10%
  • 35% reduce car use by >10 – 25%
  • 25% reduce car use by >25 – 35%
  • 10% reduce car use by over 35%.

Australian experience for all types of TBhC projects has been similar. For example, Perkins (2001) conducted a statistical analysis of the observed travel behaviour change against a set of commonly measured socio-demographic characteristics. Perkins’ paper analysed the results of the implementation of Travel Blending pilot programs in Adelaide and concluded that the characteristics that explained a significant amount of the variation in the total number of trips made by a household were unable to explain any significant amount of the change in travel behaviour, suggesting no relationship between socio-demographic characteristics and the diversion rates achieved by a project. The paper concluded that either the travel behaviour change is explained by a set of characteristics not currently measured or that the sample size was too small to determine any relationship. However, it was highlighted that those individuals who used cars the most tended to make the greater degree of positive change.

Research in Western Australia in 2005, while not conclusive, indicated at least a partial relationship between TBhC diversion rates (for household programs) and suburb form (land use mix, connectivity, amenity for active transport and public transport options) in Perth.

The work to derive default diversion rates described in the next section attempted to differentiate the factors that might contribute to higher or lower diversion rates and to provide different sets of diversion rates based on these. However, often there were no statistically significant differentiating factors and it needs to be accepted that the effects and hence benefits are always likely to be uncertain. The best results will be achieved by drawing on all available experience, selecting target areas with the greatest potential to achieve the proponent organisation’s objectives and investing effort in good design and implementation of TBhC projects.

3.4 Default diversion rates

Default diversion rates are provided for work travel plans, school travel plans and household/community based projects. These have been collated from the New Zealand and Victorian studies and reviewed in the light of more recent evidence from monitoring studies of TBhC initiatives in Australia.

If TBhC project proponents and analysts have different diversion rates, supported by sound evidence, that they consider would be more applicable for their particular project, these may be used instead of the default rates. Different diversion rates can also be used for sensitivity testing.

Originally it was envisaged that there would be a number of different sets of default diversion rates for each of the TBhC project types, based on characteristics of the target population and the scope of the TBhC programs. However, statistical analysis of the available data on results of TBhC projects did not support greater disaggregation than what has been provided.

Results reported in previous evaluations and papers are often an upper bound because they only report the observed behaviour changes for a certain subset of the population. For example, the reported diversion rate may be, say a 14 per cent reduction in car as driver mode share amongst those who participated in the program. This does not take into account the individuals that did not participate in the program (such individuals can be classified into two categories: those who did not respond/could not be contacted to participate in the program and those who chose not to participate in the program). The 14 per cent diversion is an upper bound because for this same reduction to apply across the entire population requires the assumption that all non-participants would have made the same average change as those who participated. Basing default diversion rates only upon the upper bounds will result in a consistent over-estimation of the likely benefits from a TBhC project.

To make a correction for this, a lower bound was calculated in these cases by adjusting the reported diversion rate for the program participation rate. The assumption for calculating the lower bound is that individuals in the target population who do not participate make no change in their travel behaviour and that they have average mode share prior to the implementation of a TBhC project.

Information about the TBhC program will disseminate throughout a community, resulting in the actual diversion rate for a target population being between the upper and lower bounds. The default diversion rates are based on mid-points between upper and lower bounds.

Default diversion rates are presented as percentage point changes in mode shares, chosen in preference to percentage changes relative to initial (Base Case) mode shares[1]. The use of percentage point changes supports simplified analysis procedures. Using change relative to initial mode share would require the initial mode shares of trips in a community, company or school to be known prior to the analysis. Also, tables of default diversion rates would be more complex, as would the resulting calculations. Fortunately, the evidence from TBhC projects implemented to date is that the percentage point change in mode share does not appear to vary significantly across projects with different initial mode shares.

The use of percentage point changes for the default diversion rate values might be seen as placing a constraint on the analysis because it assumes that total trip numbers are unchanged. In the derivation of default diversion rates, it is necessary that the sum of the diversion rates for the ‘from’ modes equals, in magnitude, the sum of the diversion rates for the ’to’ modes (and they will be opposite in sign). This ensures that total mode share sums to 100 per cent. This constraint can be overcome, if necessary for a particular TBhC project, by including an additional 'to' mode labelled 'no trip' with a diversion rate equal to the percentage reduction in total trips (also see Section 5.5.3).

The benefits for a particular mode changer depend on both the ‘from’ mode and the ‘to’ mode. For example, the benefits associated with cycling will differ from walking, due to the different lengths of car journeys replaced and the different level of physical activity. However, the distinction in benefits between car as passenger and car sharer is not so well defined. Primarily, the difference between the two will depend upon average trip lengths and, since this is likely to be marginal, it is assumed that there will be no difference in trip lengths for car sharer and car passenger trips. Hence, for appraisal purposes the two are the same - so car as passenger incorporates the diversion to car sharer.

3.4.1 Household/community based initiatives

After considerable investigation and statistical analysis, the New Zealand and Victorian studies concluded that household programs cannot be categorised into groups with similar diversion rates based upon socio-economic or other characteristics of the area, or upon the measures to be implemented. Similar TBhC household programs have been implemented in a number of different areas and the diversion rates observed have varied from one area to the other, with little consistent relationship to any of the above criteria.

Given the above findings, a standard set of default diversion rates is provided based on the average of all household/community projects that have been undertaken and monitored. A low set of default rates is also provided, based on the average of the bottom half of diversion rates achieved, to account for any projects that may not implement the full range of initiatives that have become standard in household based programs such as TravelSmart, or where public transport services or cycle/walk facilities are poor: the decision to use the low set is at the discretion of analysts.

The two sets of default rates were estimated by sorting data into ascending order based on the change in car as driver mode share. The standard set of diversion rates for each mode used the average mode shares from the whole sample, while the low set used the average from the half of results with the least change in car as driver. It was also necessary to adjust these values to meet the constraint of mode share summing to 100 per cent.

The two sets of default diversion rates for Household based programs are shown in Table 2.

Table 2: Household programs – default diversion rates

Car as driver

Car as passenger

PT

Cycling

Walking

Low

-1.0%

-0.2%

0.5%

0.3%

0.4%

Standard

-3.1%

-0.5%

1.4%

0.9%

1.3%

These percentage point changes apply to the whole population in the area (such as a suburb or region) targeted by the TBhC program (the target population), not just the households or people in that area who agree to participate in the program[2].

It is recommended that the standard diversion rate profile based on the average of all case studies is used for most household TBhC projects. The low set of default rates, based on the average of the bottom half of diversion rates achieved, may be more appropriate for any projects that may not implement the full range of initiatives that have become standard in household-based programs such as TravelSmart, and where public transport services or cycle/walk facilities are poor.

The WA Department of Transport has accumulated a significant database of results from 16 TBhC household/community projects in Perth (delivered to a target population of 388,733 residents). These results indicate that Perth has achieved higher than average diversion rates of -5% car-as-driver, 0% car-as-passenger, +2% PT, +1% cycling, and +2% walking.

TBhC project proponents may consider using higher diversion rates than the standard rates if they have a suitable depth of evidence to support this and their projects are delivered with matching content and commitment.

3.4.2 Workplace travel plans

A large number of workplace travel plans were analysed for to determine diversion rate impacts. After considerable investigation, there was found to be insufficient data to determine statistically significant differences between the diversion rates that were achieved with different combinations of characteristics. Instead, an approach that assigns one of three default diversion rates to a project based on a scoring system was favoured.

A proposed project is scored against the measures listed below and then classified into a set of default diversion rates based upon the aggregate score:

  • Car parking management strategies
  • Public transport service improvements and/or public transport subsidies
  • Improvements to walking/cycling facilities
  • Promotion of ride sharing.

Evidence in the literature suggests that the most significant factors in achieving lower car as driver mode share are initiatives targeted at the availability of parking, and provision of an adequate substitute for car commuting. Parking management strategies and public transport service improvements or subsidies are the two types of measures that address these barriers most directly and are thus weighted more heavily.

The scoring and classification worksheet for workplace travel plans is shown in Table 3 The various categories and guidance for scoring are described in the paragraphs following the table.

Table 3: Workplace travel plans – scoring and classification

WORKPLACE TRAVEL PLANS SCORING AND CLASSIFICATION

Parking management strategy

Is there a car parking constraint/issue

No strategy

One or more parking strategies in travel plan

Parking Strategies include:

  • Parking cash out
  • Parking charges
  • Reduced supply, etc.

No

0

1

Yes

1

2

Parking management score

__

Are these attributes part of the Travel Plan

Yes/No

score of 1 if Yes, 0 if No

Public transport service improvements

__

Public transport subsidies

__

Ride sharing matching service

__

Improved cycling/walking facilities

__

Total Score (out of 6)

__

Diversion Rate

Score

Low

1 or 2

Medium

3 or 4

High

5 or 6

Separate diversion rates are provided for projects that include public transport measures such as service improvements or fare subsidies

Car parking management strategies

A wide range of measures could fall into the category of a parking management strategy including, but not limited to, the introduction of car parking fees, parking cash-out schemes and restricting/reducing the supply of car parking spaces. It is also important to consider the current parking situation - that is, whether there is currently ample car parking space or whether parking availability is already constrained.

It is suggested that projects be given scores of zero, 1 or 2 for parking. A score of zero would be applicable if current parking arrangements adequately meet demand and a parking management strategy is not implemented as part of the plan. This reflects the fact that without a parking issue, and without the introduction of parking demand management, there are fewer incentives for individuals to change.

If parking arrangements are already constrained in some way (in other words, there is a parking issue), it is more probable that individuals are seeking an alternative mode and that the travel plan will stimulate a change in mode. In this case, a project should receive a score of 1.

A travel plan that introduces a parking management strategy is also likely to deter people from driving to work (for example, by charging for parking, some individuals will consider it more attractive to travel by another mode) and should receive a score of 1.

Implementing a single parking management strategy is probably not as effective as the introduction of a number of measures (such as introducing parking charges and reducing the number of car parks available for staff). However, analysis of the case study data did not show sufficient evidence to justify a higher score for the introduction of a combination of parking management strategies.

A score of 2 should be assigned in situations when there is an existing parking issue/constraint and the workplace travel plan involves the introduction of one or more parking management strategies.

Public transport service improvements

Improvements to public transport services could be through the provision of new bus/train routes or through the introduction of new services along existing routes. The provision of a company shuttle bus could also count as a public transport system improvement.

If a workplace travel plan includes any such improvements to the public transport system (although it is not limited to those listed above), then a score of 1 is appropriate. If no such improvements are included, a score of zero should be given.

Two sets of diversion rates are estimated, one set for when there are no public transport service improvements as part of the travel plan and one set for when service improvements or subsidies are included.

Public transport subsidies

Public transport subsidies could be either in the form of a subsidy to an operator (to improve services and/or reduce fares) or through direct fare subsidies to employees (such as free or subsidised weekly/monthly tickets). A score of 1 should apply if public transport subsidies are a measure to be included in a work travel plan, and zero otherwise.

Ride sharing

Ride sharing can be promoted in a number of ways, through the provision of preferential parking for car sharers or through the introduction of a ride sharing matching service or similar. If the travel plan introduces a ride sharing measure (or a number of measures), then a score of 1 should apply, and a score of zero otherwise.

Improvements to cycling/walking facilities

Improving the conditions for cyclists and pedestrians will encourage the use of these two modes. Two common improvements to cycling and walking facilities are the improvement of onsite facilities (such as lockers, showers and bike storage) and the improvement to external facilities (such as cycling paths/tracks/lanes). A score of 1 should apply if a cycling/walking measure is implemented and zero otherwise.

Select appropriate diversion rate based on score

The scoring system is intended to assign individual projects the correct magnitude of car as driver mode share reduction. Within the case studies reviewed, the distribution of this percentage point change across the ‘to’ modes is influenced by the measures implemented by the travel plan. For plans that do not include any public transport service improvements, it is expected that the diversion to public transport will be similar to diversion to the other ‘to’ modes. For a project that is done in conjunction with public transport service improvements or that includes company provided transport (such as a shuttle from nearest train station),the evidence suggests that a far greater proportion of the mode shift will be to public transport. Hence, separate sets of diversion rates are derived for ‘with public transport measures’ and ‘without’.

The recommended sets of diversion rates for use in the appraisal of workplace travel plans are shown in Table 4.

Table 4: Workplace travel plans – default diversion rates

Score

Set

Car as driver

Car as passenger

PT

Cycling

Walking

WITH PT service improvements

2 or less

Low

0.0%

0.0%

0.0%

0.0%

0.0%

>2 but <5

Medium

-5.0%

1.3%

2.6%

0.3%

0.8%

5 or more

High

-12.9%

3.3%

7.4%

1.0%

1.2%

WITHOUT PT service improvements

2 or less

Low

0.0%

0.0%

0.0%

0.0%

0.0%

>2 but <5

Medium

-5.0%

1.3%

1.3%

0.6%

1.8%

5 or more

High

-12.9%

3.3%

3.7%

2.7%

3.2%

The target population for these diversion rates is the total workforce (number of employees) at the workplace covered by the travel plan.

Back-testing of this scoring system with actual case studies (mainly from UK experience) showed a good match for the ‘from’ mode (car as driver) diversion rates but considerable variation in distributions across the ‘to’ modes. This was found to depend on whether or not projects involved or were accompanied by public transport service improvements.

The adoption of an alternative set of distributions for the ‘to’ modes for workplace travel plans that were not accompanied by public transport service improvements, as shown in the lower half of Table 4, improved the accuracy of the scoring system.

Estimation of ‘without PT’ diversion rates was done by assuming that half the public transport diversions (in the ‘with’ sets) is redistributed between walking and cycling. The amount that either walking or cycling receives is weighted by the relative sizes of the two in the ‘with’ group of diversion rates.

There is increasing interest in the use of travel plans as tools to influence travel demand for large new shopping complexes and expansions of existing centres. As customers generate significantly larger travel demand to these centres than staff, such travel plans are likely to include measures for both groups, not just staff. Adjustments to default diversion rates for workplaces should be made if a travel plan is directed at customers as well as staff. Diversion rate evidence from household/community initiatives might be considered for these adjustments.

3.4.3 School travel plans

Evidence on the effects of school travel plans comes mostly from the UK. This was supplemented with Australian data for tertiary institutions, in particular from Victoria.

The only differentiating factor between primary, secondary and tertiary educational institutions that gave clearly different sets of diversion rates is the level (age group) of the school or institution: that is, whether it is primary, secondary or tertiary. Other factors, such as the socio-economic level of the area, may be considered likely to cause differing effects but any such differences were not discernible from the evidence.

Primary school students typically only have a short journey, are often accompanied by their parents and generally do not use public transport. Secondary school students have longer journeys on average (relative to primary school students) and make substantial use of public transport. Tertiary institutions are different again. Student trip lengths vary significantly, with students living on campus or in close proximity having very short journeys, whereas others commute long distances. Also, the majority of tertiary students hold a driver licence and have access to a car.

The mean percentage point change in car as passenger for primary and secondary schools combined is used as the default ‘from’ diversion rate for both primary and secondary schools.

In the case of primary schools, the ‘to’ diversion rates are considered to be mainly cycling and walking. Evidence suggests that public transport is generally not an important mode for most primary schools, largely due to the relatively short journey distances for most students.

A proportion of children attending private primary schools do use public transport, possibly because such schools have a wider catchment area than public schools. The default diversion rates for secondary schools are more appropriate for such schools.

Default diversion rates for secondary schools were estimated for public transport, cycling and walking. There is limited evidence on proportions of ‘to’ mode shares. Based on experience and judgement, public transport is considered to account for most diversion and cycling is considered to receive the least change.

The default diversion rate values for primary and secondary schools are shown in Table 5. The secondary school diversion rate profile is also appropriate for private primary schools. The target population that these diversion rates apply to is the total school roll.

Table 5: School travel plans – default diversion rates for primary and secondary schools

Car as passenger

PT

Cycling

Walking

Primary schools – public

-9.0%

0.5%

2.0%

6.5%

Secondary schools

-9.0%

5.0%

3.5%

0.5%

Default diversion rates for tertiary institutions were derived from a post-implementation review of a TravelSmart program at Melbourne’s Monash University Clayton Campus. University campuses vary significantly in their location and transport availability. Monash is considered to represent an average of the characteristics of tertiary campuses and hence the default diversion rates will be appropriate for programs covering several tertiary institutions in different locations as well as individual institutions with similar 'middle-ring' locations and transport facilities.

Adjustments to the default diversion rates may be justified for travel plans for individual institutions that are located in more central locations close to a CBD or more distant outer metropolitan locations. Given the size of workforce at many tertiary institutions, it would also be appropriate to cross-check diversion rates against the default rates for workplace travel plans, particularly if both staff and students are targeted by the travel behaviour change program.

In some States the types of measures included in travel plans for tertiary institutions are closer to those in workplace travel plans than school travel plans. However, this does not mean that workplace diversion rates should be applied for tertiary institutions. Adjustments to default diversion rates for tertiary institutions could be made if the travel plan implementation is directed more at staff than students or than students and staff.

The Monash University TravelSmart program led to a 9.2 percentage point reduction in car as driver mode share by first year students compared with the previous year and significant increases in car as passenger and public transport mode shares. These results have been rounded to obtain the default diversion rates.

The default diversion rates for tertiary institutions are shown in Table 6.

Table 6: School travel plans – default diversion rates for tertiary institutions

Car as driver

Car as passenger

PT

Cycling

Walking

Tertiary institutions

-9.0%

3.5%

5.0%

0.5%

0.0%

3.5 Durability of changes

The default diversion rates in the previous section apply to the first year benefits after the implementation of a TBhC project. An economic appraisal also requires an understanding of the likely trend of benefits in future years. There are two interrelated aspects to this that need to be decided:

  • Whether benefits of TBhC projects persist or decay over time, and at what rate
  • Whether travel by all modes grows at the same rate in future years.

An issue for the appraisal procedure is whether benefits of TBhC projects persist or decay after the completion of the program. The following four possibilities have been suggested based on experience to date:

  • Benefits decay over time (in the absence of periodic ‘maintenance’ efforts)
  • Benefits can be maintained by ongoing “maintenance” expenditure
  • Benefits are durable without maintenance
  • Benefits increase over time.

Some early TBhC economic appraisals assumed that benefits would decay over time or that maintenance expenditure would be required. For example, the paper by Tisato and Robinson (1999) included an assumption that benefits of (an individualised marketing program in Adelaide) would decay without maintenance. To maintain the level of benefits, they included cost estimates for annual reinforcement programs and periodic (five-year) repeats of the travel diary. Recent experience in Western Australia also points to a gradual drop-off over time, particularly at workplaces and schools, unless there are ongoing incentives and support.

Other papers give examples of TBhC programs where benefits appeared to have been sustained for up to four years after the program and more recent evidence supports the case that benefits may be self-sustaining without specific maintenance. The most intensive study was the recent ITLS five-year monitoring project (Stopher et al, 2013) that found no drop-off in the effect of TBhC programs in four states over the five years from 2009 to 2013. This was achieved without any systematic follow-up reinforcement TBhC programs.

The UK Smarter Choices report discussed the possibility of ongoing growth in benefits in future years, but there does not appear to be any evidence to support this (without further TBhC investment).

It is intuitively plausible that if TBhC programs provide information to correct misperceptions about alternative travel options and modes of which people were unaware, many of the people making changes will find the new option to be an improvement and will not have an incentive to revert. Reversion is more likely in cases where people were persuaded to change to a less convenient option because this was more environmentally sustainable. People in this situation may be more likely to revert without occasional reinforcement. If TBhC projects are implemented along with infrastructure changes, this may also help to increase the durability of benefits.

For household/community initiatives there appears to be some reversion to previous travel choices over the first nine months following the TBhC project, but people who have not reverted by this time tend to stay with their new travel choice. Experience from Perth over a four- to five-year period in the 2000s indicated stable mode shares at the same proportions as they settled at 9 – 12 months after the TBhC project. This finding, reported in Maunsell et al (2004b), was drawn from a number of papers on the Perth TBhC experience between 1999 and 2004.

Default diversion rates are based on results from before and after surveys. The first 'after' surveys are typically undertaken around 12 months after implementation of a project when the initial reversion described above has already occurred.

Therefore appraisals of household/community projects could generally assume that benefits will be retained in future years with little or no maintenance expenditure, subject to adopting a suitable appraisal period. As a result of consultation during the development of the ATAP Guidelines, a default appraisal period comprising three years of benefits following implementation of a TBhC initiative has been adopted as being appropriate for most projects when there is no follow up expenditure.

Workplace travel plans and particularly school travel plans are more likely to require ongoing maintenance expenditure due to staff and student turnover. In the case of workplace travel plans, some of this will become part of the companies’ cost of business. However, in the case of school travel plans, this may require ongoing expenditure that would need to be estimated and included in the analysis. Embedding the learning of road rules and about active travel in the school curriculum may reduce the need for ongoing expenditure.

A related issue is whether travel by all modes would grow at the same rate in future years. In other words, if underlying growth in total travel is forecast at 2 per cent per annum, is it reasonable to assume the same growth rate for trips by all modes in the absence of any intervention, and is the TBhC project likely to change the relative future year growth rates in addition to the initial shift in mode shares?

Possible assumptions that could be adopted are that trips on all modes grow at the same rate as underlying demand growth or that, after the initial response, there is no further growth in the changed trips — all subsequent growth in each mode is what would have occurred anyway in the absence of the TBhC project.

The preferred option in this case is to assume that benefits of the TBhC project do not grow in line with traffic growth in future years. This is not to say that trips by alternative modes will not grow, but rather that this is underlying growth that would have occurred anyway in the Base Case and is not attributable to the TBhC project. It could only be attributed to the TBhC project if the project was being repeated each year and 'diverting' the usual proportion of the growth in car trips.

3.6 Modelling of travel behaviour change

For smaller TBhC initiatives, which are likely to account for the majority of projects, the most cost-effective approach for estimating benefits is likely to be spreadsheet calculations using diversion rates, trip lengths, benefit unit values and other default data provided in this guideline. However, for major TBhC programs and packages involving significant expenditure, it may be appropriate to consider using a strategic transport network model to assist with estimating the benefits.

The UK Department for Transport’s TAG Unit M5.2 Modelling Smarter Choices provides helpful guidance on the use of four-step multi-modal transport network models for estimating TBhC benefits. The guidance in UK TAG M5.2 is directed primarily at modelling practitioners. However, TAG M5.2 also contains advice that is relevant to the economic assessment of TBhC initiatives more generally.

As noted earlier, TAG M5.2 makes a distinction between 'hard' measures and 'soft' measures in a TBhC package. It notes that there is some evidence about the combined effects of several TBhC measures, but much less evidence about the isolated effects of individual 'soft' measures in a form that can be included in models. The suggested approach to modelling packages of TBhC initiatives is to use a step-by-step approach, where 'hard' measures are modelled specifically using the existing model demand vs generalised cost relationships, and adjustments for 'soft' measures are used in order to achieve the diversion rates suggested by the evidence (such as the default diversion rates). Once adjustments have been made that achieve the expected diversion rates, the model outputs can be used as estimates of the economic benefits.

[1] For example, if the initial mode shares of total trips by all modes are 85% car and 10% PT (and 5% other) and a project changes this to 80% car and 15% PT this is a change of ‑5 percentage points for car mode share and +5 percentage points for PT. The percentage change relative to initial car mode share is ‑5.88% (‑5/85). If initial mode shares were 75% car and 20% PT and a project changed this to 70% car and 25% PT this would still be a ‑5 percentage point change in car mode share and +5 percentage point change for PT. However, in this case the percentage change relative to initial car mode share would be ‑6.67% (‑5/75).

[2] If the target population only comprised the TBhC program participants in the area covered by the program, the diversion rates would be higher but the target population would be lower - so that the resulting numbers of mode changers would be the same.