So far we have discussed modelling from a general systems analysis point of view and briefly described some examples of modelling tools, from large-scale integrated land use/transport models to simple pedagogic devices. We touched upon the modelling of dynamics in general, which is a part of coding the real world system into a formal one.
Let us now go into more detail on what we consider to be the important features of a land use and transport model and point out some shortcomings.
Ideally, we would use the indicators from Chapter 3 as a checklist of output from the model. Similarly a list of available policy instruments could be used as a wish list of input variables. But indicators and instruments are represented in different ways in different models. Instead we define two general categories of model capability:
The representation of the supply effects which result from the implementation of transport instruments. These effects are of two types. Firstly there are those effects which result automatically from the implementation of an instrument (without any behavioural response occurring). Secondly, there are those changes in supply that occur once such behavioural responses have taken place. Both types of effect can be subdivided into system internal supply effects (such as changes in capacity and direct user costs) and social and environmental effects (such as accidents and pollution). The latter may also be called system external effects.
The implementation of transport or land use instruments lead to supply effects which in turn trigger behavioural responses by the various actors in the transport / land use system. These can be further subdivided into: responses by system users (either individuals or organisations); responses by suppliers; and public opinion responses.
6.8.1 System internal supply effects
System internal supply effects can be defined as those supply effects that lead directly to user responses in the land use / transport system.14 Five basic classes of internal supply effects are considered:
•
•
•
•
•
Capacity/congestion Direct user costs
Reliability of journey time Quality of journey
Information provision
Capacity concerns the capacity of the whole transport system and results from the aggregation of the capacities of individual elements of the system, such as the capacity of a road. Congestion concerns the interaction between capacity and demand, and in particular how the level of system service deteriorates as demand increases.
Direct user costs are those costs which the land use/transport system user experiences subjectively. Typically, such costs include expected journey time and money costs and are aggregated to form a generalised cost function. It is usually argued that these costs are the most important to take into account when modelling behavioural responses. However, other user costs (for example those considered immediately below) can also be considered.
Reliability, quality and information provision are here understood to be objective characteristics of the land use/transport system which might be automatically altered by the implementation of an instrument. In order to represent such a change, it is firstly necessary to be able to measure (in some quantitative way) the overall level of reliability, quality or information in the system. It is important not to confuse such objective measures with the contribution that reliability, quality and information pro- vision might make towards a user’s subjectively experienced direct costs. Although it is likely that there would be a correlation between objective characteristics and subjective costs, they are essentially different elements of the land use / transport sys- tem.
Congestion and other user costs are always modelled in land use and transport models, whereas reliability, quality provision and information provision are often not. This is obviously a shortcoming. In particular, one might wish to incorporate the effects of intelligent transport systems (ITS), e-commerce and other new technology developments, which will currently have to be assessed as part of the description of the scenario (Section 2.5).
6.8.2 Social and environmental effects
Social and environmental effects are defined as those effects which occur outside the land use and transport systems being studied, and such effects are not assumed to
14 In some models, social and environmental effects may also trigger behavioural responses (as they will in the real world). This is ignored for the purpose of a simple classification of effects.
change the behaviour or affect the choices of the users of the system. A broad range of effects may be considered under this heading, including environmental effects, accidents, implications for government budgets, and wider economic and social impacts. Some of these effects (especially global and regional environmental effects and some of the wider economic impacts) will be felt by residents living outside the studied area and even living far into the future.
Effects that are experienced by residents in the studied land use/transport system will also be classified here provided they do not affect them in their capacity as travellers or influence their location choices. Thus a policy might affect them as taxpayers, but since their response to tax increases is not a part of the land use/transport system as we define it in most cases, this will be an external supply effect in most cases. Also, travellers in the transport system are currently not assumed to change their decisions based on, say, the changes in the accident rates of different modes or the levels of local pollution experienced on a trip. This is why we can regard such effects as lying outside the studied system.
The concept of social and environmental (system external) effects must therefore be defined relative to the studied system and the purpose of the study (strategic, tactical) being made. In an integrated land use/transport context, local pollution will perhaps be a borderline case. For the travellers in the transport system, local pollution is an external effect, since the level of local pollution does not affect their trip behaviour.
But the same individuals are also residents in the location system. If changes in local pollution levels in the zones affect their location choices, and if the link between traffic volumes and zonal levels of pollution is established in our model of the system, local pollution can clearly not be seen as a wholly external effect any longer. In fact, in the DELTA modelling system used in PROSPECTS local air pollution (along with noise pollution) from transport is considered to be a factor in the residential location choice model15. Hence, in the DELTA system, local pollution (both air and noise) from transport is a system internal effect. Unfortunately, though, this approach is not common in land use modelling systems, and so local pollution is treated as a system external effect here.16
We consider the following system external effects, which can be seen to tie in closely with the indicators defined in Section 3.3:
•
•
•
Environmental effects Traffic accidents Health effects
15 However, a distinction needs to be made here between zonal levels of pollution (the subject of this discussion) and the amount of emissions from traffic generated in each zone. The latter does not take account of pollution from non-transport sources or the dispersion effects of pollution due to, for example, weather conditions, and hence it is the former concept that more closely fits with standard perceptions of air quality. In the DELTA modelling package there is a link between traffic emissions generated in a zone and location choice. However, this link should be seen as a proxy for the link between zonal levels of pollution and location choice and is clearly a modelling simplification.
16 Even if agents in the model do respond to a supply effect, they might not respond properly from an economic efficiency point of view, that is, there might still be uninternalised external costs in the sense of economics. Thus the concepts of system external and system internal effects do not coincide with externalities and the internalisation of externalities in economics.
The only difference between a system internal and a system external effect in our terminology is that the latter does not influence the behaviour of the users of the system as we conceive or model it.
Liveable streets and neighbourhoods
•
•
•
•
•
•
•
•
Implications for government budgets Equity and social inclusion
Economic growth
Per definition of these effects, they can be modelled by models appended to the modelling system. Data to compute these effects comes from either the land use/transport model input or from its output. More or less sophisticated processing of the data might be required to establish the effects. Examples of such post-modelling include equity analysis, accident analysis and air pollution modelling. Usually, current models apply very crude post-models of these types, and there is scope for large improvements.
6.8.3 User responses
Demand/behavioural responses by system users can be separated into four categories:
Location responses
Strategic transport responses
Responses to expected daily traffic conditions Responses to unexpected conditions
The meaning of location responses is probably clear without further explanation, except that users in this case need not be equivalent to a person. It can also be companies or other organisations.
However, it is useful to explain further the categories of transport response, and such discussion will hopefully help to distinguish between them. Two types of strategic transport responses are considered. On the one hand, there are discrete long-term decisions which are likely to have a heavy consequential influence on transport behaviour. For example, buying a car, motorcycle or public transport season ticket are such events. The other type of strategic transport response simply concerns the overall quantity of travel carried out, without disaggregating between purpose, mode or other factors.
The responses to expected daily traffic conditions include the choices of destination, mode, time-of-day and route, and ways in which daily activities are combined and trips are chained. Destination, mode and route choice are included in most models, whereas time-of-day and trip chaining are still considered “advanced” by many.
Responses to unexpected conditions include reconsideration of the usual or habitual choices in the light of information about unexpected events, or in the light of other information issued or traffic control measures taken on that particular day. Such responses are usually modelled by microsimulation models, which are not a part of the ordinary modelling system. But even so, events are frequent enough and the information and traffic control measures are important enough to be able to influence ordinary behaviour.
Modelling of user responses are at the core of current models, but even so, there are considerable challenges in taking more aspects of behaviour into account and in combining the long-term and short-term responses in a single consistent modelling system.
6.8.4 Supplier responses
The assumption underlying much of the planning approach described in this Guidebook is that there is a responsible “transport/land use authority” who is the main initiator of the instruments given in the rows of the tables. The reality that there might be splits in responsibility between a number of organisations was addressed at the start of this guidebook (in Section 1.1). This split in responsibility leads to the issue of supplier responses.
Where organisations different from the main transport authority have supplier responsibilities, they are termed third party suppliers. They may be private firms (public transport operators, property developers, other businesses), neighbouring local government or others. The actions of third party suppliers, except developers and landlords, are rarely modelled. Such actions may include:
Developing unused land (greenfield, whitefield, brownfield) and deciding on its use (housing, business etc.) and density of use.
•
•
•
•
•
•
•
•
•
•
•
Building of houses, office buildings and other facilities on the available land.
(Some such decisions are included in most LUTI models).
Determining rents (included in most models).
Operating or stopping to operate public transport services, reorganisations of public transport.
Changing rail/bus frequency.
Changing size of trains/buses.
Changing rail/bus fares.
Changing public transport quality.
Changing total car parking capacity.
Reallocating car parking space between “long term” and “short term”.
Changing car park charges.
Furthermore, other land/transport authorities (such as neighbouring authorities or authorities on a higher/lower level) could make supplier responses by implementing any of the land use / transport instruments considered throughout this Guidebook.
Because of the difficulties of predicting these responses they are often modelled as changes in the input assumptions. In the light of the current trend of liberalisation of transport markets and the continuing lack of coordination of land use and transport responsibilities in many cities, this is a shortcoming of current model systems.
6.8.5 Public opinion responses
The term public opinion responses encompasses both the impacts on public opinion of implementing particular instruments as well as the action taken by the public in response to these impacts. In general, the term public opinion includes both the majority opinion of society (as expressed through democratic processes) and the opinion of special interest groups who have the power to affect transport policy.
Examples of the latter are business organisations, the media, the police and environ- mental organisations.
Current land use/transportation models do not generally represent public opinion responses and it is legitimate to question why they should. On one hand, it could be
argued that if planners are attempting to make predictions about the future development of the land use/transport system, they need to take into account all the actions of participants in this system that are liable to change it. These actors include users, suppliers and the public. If actors are missing from the representation of the system, predictions about it are liable to be wrong. On the other hand, no such models exist (except for a similar feature of PLUTO, see section 6.7), and it might be argued that to present the public and decision-makers with a model of how they are likely to act in the future is inappropriate.
6.8.6 Capabilities of already-existing software packages
Our survey of modelling capabilities has identified the following weaknesses and shortcomings of most model systems in use at present:
• Reliability, quality provision and information provision is not modelled (Section 6.8.1).
• Most models usually apply very crude post-models of the social and
environmental effects, and there is scope for large improvements (Section 6.8.2).
• There are considerable challenges in taking more aspects of user behaviour into account and in combining the long-term and short-term responses in a single consistent modelling system (Section 6.8.3).
• In the light of the current trend of liberalisation of transport markets and the continuing lack of coordination of land use and transport responsibilities in many cities, there is a need for modelling the responses of at least some of the major categories of suppliers (Section 6.8.4).
Some modelling systems are definitely better than the average in some of these respects, and we may expect more to happen in the future. But even if these shortcomings are serious, the question must be asked if they are so serious as to invalidate any long-term predictions using the average, state-of-the-art model. We believe that provided the long-term (“strategic”) decisions about car ownership and location have been adequately incorporated in the current state-of-the art modelling system, it may be regarded as a “well-made caricature” of the real world system (Section 6.3), capable of telling us important things about the future development of the city.
Timms and Minken (2002) carried out a review of the computer model packages used in PROSPECTS with regard to whether they include the above-described capabilities.
The results are summarised in Table 6.2, where the number of √s represents the relative degree of model package capability, and where X represents no capability.
Three points must be made when interpreting this table:
There is clearly a great deal of variety between different large-scale models. Table 6.2 can only provide a rough overview.
There is no indication in the table as to how well any particular model package represents an effect or response. The table is simply concerned as to whether the model packages try to make the relevant representation.
The different types of model are all intentionally pitched at different levels of aggregation (and in fact the Policy Explorers do not even represent specific cities).
Thus a simplistic comparison (without trying to understand the needs of the model- users) should not be inferred from the table.
Table 6.2. Capabilities of different types of model package.
Large-scale models
Sketch Planning Models
Policy explorers
System internal supply effects √√√ √√ √
System external effects √√√ √√ √
User responses √√√ √√ √√
Supplier responses √ √ √
Public opinion responses X X √