Regional Transport in Canterbury Health Impact Analysis Dynamic Simulation Model Final Report for Environment Canterbury David Rees and Adrian Field 11 June 2010
CONTENTS 1. Background 2. Model Structure and Assumptions 3 4 Key Scenarios 3.1 Scenario 1: Baseline No Change Scenario 3. 6 6 3.2 Scenario 2: Car Culture 7 3.3 Scenario 3: The Decline of the Car 9 3.4 Scenario 4: Continuation of Current Trends 12 4. Modifying Key Assumptions 4.1 Impact of Decline of the Car Under a Range 14 of Assumptions 14 5. Different Populations 16 6. Conclusions 18
1. BACKGROUND The model was developed to support the Health Impact Assessment (HIA) being undertaken by Synergia as part of the development of the Canterbury Regional Land Transport Strategy (CRLTS). Its purpose was to explore the potential impacts of changing mode choice upon key areas of health impact being considered within the HIA. The transport strategy simulation model was built according to the principles of System Dynamics (SD), a simulation-based methodology for improving understanding of how systems evolve over time, and for analysing the short-term and long-term effects of interventions in such systems. HIAs seek to influence policy by making predictions about the consequences for health from decisions made outside of the health field; in this case the consequences of transport decisions upon health indices such as safety, physical activity and pollution. The purpose of the model within this HIA was to provide a greater understanding of the direction and size of impact of various changes being proposed within the CRLTS. Specifically, it was developed to provide a quantitative estimate of the health impacts that could be expected and the magnitude of change that would be required to bring it about. A SD model is a causal model which tries to tease out the causal links between key variables and the behavioural consequences that arise from these links. The model consists of a causally linked set of factors that provide a visual description of a dynamic situation. The visual maps developed in a SD model consist of key accumulations, or stocks, the inflows and outflows to those stocks, and the factors, activities, and decisions determining those flows. For example, within the HIA model key stocks included the population stocks, total kilometres travelled and people with adequate physical activity. Examples of key flows include change in fuel usage, change in kilometres per driver and change in percentage of population walking. To illustrate, the following extract from the model shows the calculation for cumulative fuel savings in millions of litres. In this case it is a function of additional driving km per year millions and a fuel saving frac, which is based on work undertaken by the School of Population Health at Auckland University. As kilometres travelled by car goes down so does fuel usage and as kilometres travelled by care goes up fuel usage increases. The model is built up from a linked set of these stock/flow structures. Within the model safety is explored through data on cycle injuries, physical activity is explored through the use of the Ministry of Health guidelines on adequate physical activity and pollution is explored through looking at changes in CO 2 emissions. COPYRIGHT SYNERGIA 2010 PAGE 3
Data for the model was obtained from the New Zealand Travel Survey, the New Zealand Health Survey, data extracts provided by staff at Environment Canterbury and members of the HIA review group. Some of the key assumptions were tested with Dr. Graeme Lindsay at Auckland University s School of Population Health. In addition, Dr. Lindsay and Dr. Alex Macmillan provided advice on research on the health and environmental impacts of different transport modes. 2. MODEL STRUCTURE AND ASSUMPTIONS The structure of the model is built around total kilometres travelled per year which forms the basis of the data within the New Zealand Transport Survey. The model focuses on the population of Greater Christchurch and runs for 15 years through to 2025; section 3 explores the impacts of different transport use scenarios on transport and health outcomes in this geographic area. Section 4 looks at the impact of changing some key assumptions within the model and section 5 explores the different impact of transport patterns among Greater Christchurch and other Canterbury populations on some key outcomes. Use of data on the kilometres travelled per year within each of the different travel modes, combined with knowing the average kilometres travelled per trip, the number of people using each of the mode choices can be calculated. The following extract from the model shows the basic structure that drives the usage of public transport (PT in the model). The same structure is replicated for walking, cycling and driving. COPYRIGHT SYNERGIA 2010 PAGE 4
Underpinning this structure however are some assumptions that influence the final outcomes shown in the model outputs. The most significant assumption in the model is that not all extra kilometres travelled are undertaken by additional people. Some of these additional kilometres come from people making greater use of their existing modal choice, such as drivers making more and longer trips. As a result, doubling the amount of kilometres travelled does not double the number of people using that transport mode. The baseline assumption is that at the beginning of the simulation each additional 100 kilometres travelled brings with it 60 people shifting to that mode choice, and by the end of the simulation this reduces to 30% i.e. as usage of any mode increases the change is brought about more by existing users travelling more than it is by new people shifting to that mode. In putting this model together we found no data that could help inform the validity of this assumption. It is instead based on conversations with the team involved in leading the RLTS HIA and representatives from the health, local government, regional government and other sectors. Because it is an important assumption section 3 shows model outputs under a range of alternative assumptions. A second important assumption concerns the impact of mode change upon population health. It is clear that walking, cycling and to a lesser extent use of public transport have positive benefits for the individual in terms of physical activity. However at a population level the situation is not so clear. If, for example, you manage to shift 100 people to make use of cycling as a regular transport mode it does not mean that the community now has an additional 100 people taking up adequate physical activity. At one extreme it is entirely plausible that all of those 100 people who now cycle already had active lives and the overall health of the community has not increased at all. The New Zealand Health Survey estimates that 51% of the Canterbury population currently have adequate physical activity. The mode uses this estimate and assumes that 50% of all those who shift to more active modes of transport will have inadequate physical activity and therefore contribute, by the change to improving overall community health. This highlights two important points. Firstly the model has had to make estimates in the absence of good data, and therefore model outputs should be viewed as indicative and the relative differences between each scenario is more important, and more robust, that the actual quantum. Secondly it is important to understand these assumptions underlie real policy choices. If one is promoting a shift to more active transport modes as a means of improving the health of the community one needs to think very carefully about any projections one makes about the actual impact at a population level. Despite these uncertainties however the model does provide a sense of the relative impacts across all four transport modes. COPYRIGHT SYNERGIA 2010 PAGE 5
A third assumption is that the new percentages shown within each scenario take 15 years to eventuate. This is simply to acknowledge that behaviour changes do not occur overnight and any major change in transport use will take many years to eventuate. The model is designed so that that shifts occur gradually over the 15 year period between 2010 and 2025. 3. KEY SCENARIOS 3.1 SCENARIO 1: BASELINE NO CHANGE SCENARIO The following outputs from the model represent the baseline scenario. The baseline scenario assumes that current mode choice remains unchanged. The baseline assumptions for mode choice (based on vehicle kilometres travelled) within greater Christchurch are: Public transport 5.4% Cycling 1.8% Cars 88.9% Walking 3.9% These figures are taken from the New Zealand Transport survey and represent the percentage of total kilometres travelled within each mode choice. Thus a shift, for example of public transport from 5.4% to 10.8% represents a doubling on the kilometres travelled by public transport, not a doubling of the number of people travelling by public transport. The baseline scenario is useful as it enables alternative scenarios to be tested against what would happen if no changes were made. For example, the following graph shows the changes in number of cyclists that is additional cyclists above the baseline - within Greater Christchurch if a policy was enacted to double the number of cycling kilometres travelled over a five year period i.e. shifted the percentage to 3.6%. As would be expected there is a rise in the number of cyclists as the new policy is enacted, and assuming no change in the avg km travelled per cyclist, this would result in an additional 5,500 cyclists riding throughout greater Christchurch (scenario 2). Scenario 1 marked 1 on the graph is horizontal as it is the base scenario in which no changes have been made to mode choice. The result of this is an increase the number of cyclists from the baseline of 26,000 to just under 32,000 by 2025. This is shown on the following graph: COPYRIGHT SYNERGIA 2010 PAGE 6
The following sections show the model outputs resulting from a number of scenarios that were explored as part of the HIA. For each scenario the outcomes are noted for changes in the number of people using each transport mode. In addition, the outputs also describe the impact upon cycle injuries, people with adequate physical activity, as measured by Ministry of Health guidelines, and CO 2 emissions. 3.2 SCENARIO 2: CAR CULTURE This scenario increases the dominance of cars as the transport mode of choice. As the use of cars continues to grow other transport modes decline, leading to the following mix of transport modes: Public transport 4.0%: a decline of 1.4% compared to the baseline Cycling 1.5%: a decline of 0.3 % compared to the baseline Walking 3.0%: a decline of 0.3% compared to the baseline Cars 91.5%: a rise of 2.6% compared to the baseline 3.2.1 Change in users of different transport modes As to be expected within the scenario the number of people using public transport declines against the baseline. By 2030 the number of people using public transport has declined to just over 43,000. Within this scenario, there is no significant change; the total number of cyclists remaining around 26,000. COPYRIGHT SYNERGIA 2010 PAGE 7
The number of people walking also declines in this scenario. The model indicates a drop of around 13,000 people over the next 15 years, with the number of walkers dropping from around 184,000 to 171,000. The drop in number of users in other transport modes is balanced by the rise in private vehicle usage. Compared to the baseline scenario the number of additional drivers is just over 2000 increasing the number of drivers from the current figure of 241,000 to just over 243,000. 3.2.2 Impact upon cycle injuries The model shows no change in either the number of cycle injuries or the relative risk of cycle injuries within this scenario. This is probably due to the small number of additional cyclists that this scenario produces, resulting in no discernable change in relative risk of an accident. 3.2.3 People with adequate physical activity The Ministry of Health survey indicates that around 50% of people with Canterbury undertake enough exercise to be considered to have adequate physical activity. The car culture scenario has a negative impact upon this, decreasing the COPYRIGHT SYNERGIA 2010 PAGE 8
number of people within greater Christchurch considered to have adequate physical activity by around 4,000 over the next 15 years. 3.2.4 CO 2 emissions As to be expected, the car culture scenario shows a significant increase in CO 2 emissions of nearly 20,000 tonnes per year above the baseline by 2025. The model also explores the impact upon the level of air pollution as measured by PM10 and fuel usage. However, as they are all calculated as a fraction of kilometres travelled the shape of the graph is identical. 3.3 SCENARIO 3: THE DECLINE OF THE CAR This scenario decreases the dominance of cars as the transport mode of choice. As the use of cars declines other transport modes increase, leading to the following mix of transport modes: Public transport 15.0%: a rise of 9.6% compared to the baseline Cycling 7.0%: a rise of 5.2 % compared to the baseline Walking 8.0%: a rise of 4.1% compared to the baseline Cars 70.0%: a decline of 18.9% compared to the baseline 3.3.1 Change in users of different transport modes This scenario increases the use of public transport with over 2000 additional people using public transport. This scenario has a significant impact upon the number of cyclists with a far greater increase in numbers, due to the much smaller average kilometres per trip, than public transport. COPYRIGHT SYNERGIA 2010 PAGE 9
The number of people walking also rises in this scenario. As the number of kilometres walked nearly doubles, the number of people walking increases by 40,000, rising from a current estimate of 184,000 to 224,000 by 2025. The increase in the number of users in other transport modes is balanced by a decline in private vehicle usage. Compared to the baseline scenario the number of drivers decreases by nearly 20,000, reducing the number of drivers from the current figure of 241,000 to just over 223,000. 3.3.2 Impact upon cycle injuries As would be expected the rise in the number of cyclists also brings with it a rise in cycle injuries. Of significant however is the significant drop in the relative risk of cycling with the risk per cyclists being approximately 50% of the baseline by 2030. This is because of what is termed safety in numbers ; the greater the number of cyclists on the road the less risk for each individual cyclist. This is hypothesised to occur due to the increased visibility that an increase in numbers would bring. Because there are more cyclists on the road drivers notice them more and therefore are less likely to have an accident. COPYRIGHT SYNERGIA 2010 PAGE 10
3.3.3 People with adequate physical activity The greater use of walking and cycling brings with it a significant increase in the number of people with adequate physical activity. This scenario indicates that the health gain could be around an extra 17,000 people within greater Christchurch have adequate physical activity. 3.3.4 CO 2 emissions As to be expected, this scenario brings with it significant decrease in CO 2 emissions of over 125,000 tonnes per year. COPYRIGHT SYNERGIA 2010 PAGE 11
3.4 SCENARIO 4: CONTINUATION OF CURRENT TRENDS This scenario continues current trends resulting in minimal shifts in current baseline percentages. While there is a slight increase in car usage, public transport and cycling, this is offset by a decline in walking Public transport 5.5%: a rise of 0.1% compared to the baseline Cycling 2.3%: a rise of 0.5 % compared to the baseline Walking 3.1%: a decline of 0.8% compared to the baseline Cars 89.1%: a rise of 1.0% compared to the baseline 3.4.1 Change in users of different transport modes This scenario has no discernable impact upon the number of people using public transport with the total number of public transport users in greater Christchurch remaining around 44,000. This scenario does however increase the number of cyclists who increase from the current figure of just over 26,000 to just under 29,000. The number of people walking declines considerable in this scenario. Walkers drop by nearly 20,000 so that by 2025 the number of people walking declines from 18,500 to 16,700. Car usage increase slightly in this scenario. As a result the number of drivers increase from the current figure of 241,500 to 241,800. COPYRIGHT SYNERGIA 2010 PAGE 12
3.4.2 Impact upon cycle injuries This scenario generates a small increase in the number of cycle injuries, while the risk per cyclist decrease to around 87% of the baseline risk. 3.4.3 People with adequate physical activity The increase in the number of cyclists within this scenario is more than offset by the decrease in the numbers of walkers. As a result the number of people with adequate physical activity decreases. 3.4.4 CO 2 emissions The small increase in the percentage of kilometers travelled by car, results in a relatively small increase in CO 2 emissions. COPYRIGHT SYNERGIA 2010 PAGE 13
4. MODIFYING KEY ASSUMPTIONS As pointed out in section 2, the outputs shown in section 3 are influenced by the assumption made about uptake of new transport modes as opposed to people increasing their use of their current mode choice. This section explores the scenario, decline of the car under different assumptions. In the following graphs, run number 1 is based on the assumption made in section 2 i.e. each additional 100 trips brings with it an additional 60 users, reducing over the time of the simulation to 30 additional users for every 100 trips. Run number 2 assumes each additional 100 trips brings an additional 100 users i.e. all the additional travelled in any given mode is brought about by additional users. Run number 3 halves this, assuming each additional 100 trips brings with it 50 additional users to that transport mode. 4.1 IMPACT OF DECLINE OF THE CAR UNDER A RANGE OF ASSUMPTIONS The significance of the assumptions is shown clearly in the first graph additional PT transport users. If we use the baseline assumption (run 1) - each additional 100 trips brings with it an additional 60 users, reducing over the time of the simulation to 30 additional users for every 100 trips there are approximately an additional 3,500 users of public transport by 2025. If on the other we assume each additional trip brings with it an additional user of public transport (run 2) then this figure rises to just under 7,000. Thus the plausible range is quite considerable. Increasing the kilometres travelled to 15% of all trips could increase patronage on pubic transport by anywhere between 3,500 to 7,000 people, depending on the assumptions you make about the balance between new people travelling by public transport and existing people using it more. COPYRIGHT SYNERGIA 2010 PAGE 14
For cyclists, increasing this mode of transport to a level where it is used for 7.0% of total kilometers travelled results in a plausible increase between 10,000 and 16,000 cyclists. A similar picture emerges for walkers with the plausible range being somewhere between 45,000 and 80,000 additional walkers depending on the assumptions you make. For drivers the plausible decrease is somewhere between 17,000 and 70,000 drivers. Because of the range of plausibility the base assumptions are conservative. As a result there is confidence that the changes described in section 3 are at the lower end. One can safely assume that the scenarios described will bring about a change AT LEAST as big as shown in the model output graphs. COPYRIGHT SYNERGIA 2010 PAGE 15
5. DIFFERENT POPULATIONS This section shows some of the model outputs for different populations. The populations for which data was incorporated into the model were: 1. Canterbury Region 2. Canterbury rural 3. Greater Christchurch 4. Christchurch City 5. Christchurch city 65+ 6. Christchurch low socio-economic population The following graphs show the outputs for these populations using the baseline scenario. The numbers shown on the graphs match the numbers shown above for each population i.e. run 1 is the regional population, while run 4 is the population of Christchurch city. That is, the run numbers match the numbers alongside each population group shown above. The following graphs show the number of users of public transport, cyclists, walkers and drivers for each of the above populations. As these graphs show Christchurch City (run 4) make up just over 63% of the people using each of the transport modes. Greater Christchurch (run 3) makes up just under 78%. Because of the dominance of Christchurch City and Greater Christchurch it is important that any policy changes explore, very carefully their impact upon these two population groups as they will have the most significant impact upon any health outcomes within the region. COPYRIGHT SYNERGIA 2010 PAGE 16
The additional CO 2 emissions shows a slightly different profile in which the longer trips made by rural drivers means that they contribute to 45% of that increase. COPYRIGHT SYNERGIA 2010 PAGE 17
6. CONCLUSIONS There are a number of conclusions that can be drawn from the simulations shown with the four scenarios. Some of these are clear and unambiguous. For example, unless efforts are made to reduce the use of private motor vehicles then Christchurch is faced with an ongoing increase in emissions with their resulting impact upon many respiratory diseases. Furthermore, continued increases in the use of private vehicles brings with it less use of more active transport modes such as walking and cycling and with that a decreasing percentage of people who will meet the Ministry of Health s guidelines for physical activity. However, this also works the other way, and efforts to reduce the use of private motor vehicles will have positive health impacts, arising from better air quality and higher levels of physical activity through use of more active transport modes. There are also some other conclusions that are not so clear. One is that it is important to distinguish improvements for the individual from improvement for the population. For example, in calculating the health benefits of increasing the percentage of people using cycling we had to make assumptions about the percentage of those new cyclists who did not meet the Ministry of Health s activity guidelines. The base model assumes 50%, which is the approximate percentage of the people within Christchurch who do not meet those guidelines. There is no doubt that cycling increases activity levels, and for an individual that invariably has a positive impact. However, from a population perspective if, by increasing the kilometres travelled by bicycle, you only encourage existing cyclists to cycle more, rather than getting new people to cycle, you are doing very little for population health. So, if the aim is to improve population health it is important to consider who would respond to any signals that are made about shifting mode choice. Simply increasing kilometres travelled will not necessarily translate into health gains. The model highlights the need to focus efforts so that the shifts in transport modes increase the numbers of new people shifting from less to more active modes. A further area that needs to be considered is the time involved. Shifting peoples patterns of behaviour takes time, and capturing the health benefits from that shift in behaviour will take even longer. The model brings in the mode shifts described in each scenario over a 20 year period. Those expecting gains to be made within normal planning or political cycles will be disappointed. However, if actions are committed to over the long term, then the modelling indicates that improved population health and environmental health outcomes can be achieved. Small gains in the early years do build into significant gains over the long-term. Section 5 highlighted that Greater Christchurch has a significant impact upon the region as a whole. With 65% of the kilometres travelled being by people from Christchurch City and a total of 78% within Greater Christchurch it is important that the CRLTS and the CTP work together to ensure some consistency in approach to the changes underlined in the model. COPYRIGHT SYNERGIA 2010 PAGE 18
A final point is that the model can be seen as a reflection of the best we know at this point in time. Physical scientists continue to work with and develop models over time, so what begins as crude and tentative becomes a basis for rigorous analysis and prediction. We need to take the same attitude to models of social behaviour. Models are simply reflections of our own knowledge and provide a mechanism to capture new information and new data as our understanding increases. Continual development of the model can support ongoing decisionmaking and future learning by providing: an exploratory tool to quantify impact an alternative source of input into economic modelling an opportunity to raise questions of how to drive change towards healthier outcomes a range of scenarios, beyond those described in this report a place to capture our increasing understanding and additional data as it emerges The model, should therefore been seen as a useful initial step, capturing some of the information we have about transport modes and the health impacts of changing them. It is also a repository of knowledge so that as we learn more the model can be refined and developed further to capture that knowledge and thereby provide an increasingly sophisticated tool over time. COPYRIGHT SYNERGIA 2010 PAGE 19