|
The METS model is quite complicated inside, but the basic ideas are simple enough. METS is a large computer program which represents London's transport system as a series of inter-related equations. There is an equation, for example, that describes the demand for bus trips as a function of the cost of the journey and the cost of alternatives, such as cars or the tube,
and similar equations for the tube, overground trains, cars and taxis. The relations between cost and demand are expressed using the price elasticities you are all familiar with. Here is a table of some of the elasticities used:
Table 2 METS cost elasticities.
| |
Car |
Bus |
Underground |
| Car |
-0.30 |
0.09 |
0.057 |
| Bus |
0.17 |
-0.64 |
0.13 |
| Underground |
0.056 |
0.20 |
-0.50 |
Source: Grayling and Glaister p35.
Each row here tells us how demand for that form of transport changes as costs change. [fn: these are not quite price elasticites: as we will see below, the costs here are more than just the price of the ticket]. Look at the top row. The first number tells us that the own-price elasticity of demand for car journeys is -0.3, so a 10% rise in car costs will cause a 3% fall
in car use. The second number in the first row (0.09) is the cross-price elasticity of demand for car use with respect to bus costs: a 10% increase in bus costs would cause a 0.9% increase in car use, as people switch to using their cars as they become relatively cheaper. The third number (0.057) is the cross-price elasticity of car use with respect to Underground costs: car
users seem less responsive to changes in tube costs than bus costs.
From the second row, second column, you can see that buses are rather more responsive to own-cost changes (an own-price elasticity of -0.64, so a 10% cost increase causes a 6% fall in use), and from the third row that the Underground elasticity, at -0.5, is somewhere in-between cars and buses.
Note that all the own-cost elasticities are absolutely less that -1, which implies that total revenues should rise if fares go up (can you see why?).
Where do these elasticities come from? Chiefly, they are estimated using statistical techniques (on data collected from the National Travel Survey, an annual survey of transport use). These techniques are just clever versions of the linear regression techniques that those of you who have studied statistics will probably have come across. Ironically, the Fares Fare policy
changes described above, with their huge swings in costs and use over quite a short period, provided an excellent natural experiment to get started. (As an exercise, see if you can work out the elasticities implied in table 1; are they the same as in table 2? Why might this be?).
Much of the complexity in METS comes from the need to accurately measure costs. The costs of making a journey are not just the price of the bus ticket or of your car's petrol. Your time is worth something, too. In METS, as in most cost-benefit analysis, the cost of a hour of someone's time is measured by their hourly wage rate. Average hourly wage rates for people
using different modes of transport can be estimated using the National Travel Survey. Table 3 shows the values used in METS.
Table 3
Value of time per person for different modes of transport (£s per hour)
| Car |
Bus |
Tube |
Overground Rail |
Taxi |
| 6.0 |
3.42 |
5.52 |
6.0 |
9.69 |
Source: Grayling and Glaister p35
This shows that people who use the Tube are estimated to have higher wages on average than people who use buses, but taxi users have higher wages than either. So, for example, keeping a tube train with 100 people in it stuck in a tunnel for an hour (as happened to one of the authors recently) imposes a time cost in METS of £552 (£5.52x100), whereas keeping a bus with 50
people stuck for an hour has a time cost of £171 (£3.42x50).
In METS, the length of a journey depends on waiting and boarding times for public transport (how long you have to wait before a non-full bus or train turns up, and then how long it takes you to get on) and average traffic speeds. Fares, waiting times and traffic speeds are inter-related in quite complex ways. For example, consider a bus fare cut. This will increase the
demand for buses and reduce the demand for alternative modes of transport, which should increase average speeds, since there will be fewer cars on the road (and so roads are less congested). Consequently, you may get to your destination faster, but in some circumstances you might have longer boarding and waiting times, since the buses will be fuller and you might not be able
to get on the next one.
METS models the effect of all such relationships simultaneously. Try just changing bus fares, for example: this will affect almost every variable in the output. You will see changes not merely to bus use, but to the use of all other modes of transport, to average speeds, and to costs (both time costs and money costs).
Although METS was developed almost 20 years ago, it has been updated several times since, most recently in 2000, when it was used to produce a report (Grayling and Glaister (2000) on the policy options available to the new Mayor of London. This is the version of the model used here.
Cost-benefit analysis often takes other forms of cost into account, too, such as pollution and the costs of accidents. However, these are not calculated in METS, mainly because of the lack of suitable data.
|
