What Really Matters when Forecasting?
Given its pace of development, India is becoming a magnet for transport professionals. Many planners are keen to apply in India what they have learnt from overseas — international best practices and state-of-the-art techniques. However, it may be necessary to re-question what is most fit for the purpose. Optimistic forecast assumptions sometimes need tempering with a touch of cynicism, to address possible risks, even if this goes against the consensus view.
Practitioners may feel encouraged to use state-of-the-art techniques as these are often equated to best practice. However, such techniques are typically developed in western environments where conditions are usually quite stable and a wealth of data are available, going back many years. It is the combination of relative stability and data (combined of course with research budgets) which enable innovations. State-of-the-art techniques are often embellishments to pre-existing methods. That is not to say that such techniques do not have a role; but their limitations should be considered.
Despite practitioners? best efforts, forecasts all too often remain embarrassingly inaccurate as the demand is often substantially over-forecast. There may be over-emphasis on the state-of-the-art rather than concentrating more fundamentally on appropriate, fit-for-purpose methods. This is also true in the West but such dangers multiply in rapidly developing environments.
Nature and Challenges of Rapid Development
In an urban context, rapid development typically comprises rapid economic growth accompanied by population growth (e.g. rapid urbanisation), quite possibly together with new urban development areas (population, commercial and/or industrial centres). All of these can change the previously existing consumer ? and transport ? behaviour. The development of new transport modes can also drastically change habits. Hence, I am also including this in a broad definition of rapid development; e.g. the implementation of metro systems for the first time in a city, or an area of a city.
It is this destabilisation or transformation of behaviour which poses particular risks in rapidly developing environments. Data (often relatively sparse to begin with) can become rapidly out-of-date; and data based on previous conditions can be misleading.
The pace of change poses challenges to institutions: can the governments ensure that the planning codes, etc., are up-to-date and enforceable? Do they have sufficient staff with sufficient skills and sufficient time to keep up?
Can transit operators cope with the evolving and changing demands or are they stuck operating the legacy networks attuned to how cities were, not how they are? A further problem is one of transition: as networks evolve from the ones which were viable (or at least were manageable with subsidy levels) reflecting old demands towards new networks reflecting new demands, the management and financing of the changes can be a particular challenge.
Moreover, rapid development entails far greater uncertainty. All too often forecast assumptions of economic or population growth are based upon trends in the last few years, extrapolated perhaps 30 years into the future. Given the effects of compounding, a small variation per year can translate into a very large difference thirty years hence. Furthermore, rapid development is rarely linear: different parts or segments of a city and its population will change in different ways and at different rates.
Different Baseline Conditions
India has quite different baseline conditions from western cities and more than simple quantitative differences in income levels (e.g., labour market structures, income distribution, urbanisation rates including growth thereof) and urban structures. And differences occur between Indian cities also. Its vehicle mix is also very different. In the west, a vast majority of transport is motorised, with some people now making a lifestyle choice to switch to cycling. Conditions in India are quite different. Such differences must be borne in mind when determining how to analyse the transport patterns to assess possible new transport infrastructure and /or policies.
But How Good Is Western Practice Anyway?
It is perhaps also worth reviewing how successful transport planners and modellers have been in the West.
Focussing simply on toll road studies where there is typically a single road as the focus of investigation, one might presume that forecasts ought not to be too unreliable. Yet experience quite strongly suggests otherwise: with average initial year traffic on toll roads being just 70% of forecast levels, (Bain, R. & Wilkins, M. ?Credit Implications of Traffic Risk in Start-Up Toll Facilities?, Standard & Poor?s, September 2002) and rather than being attributable simply to ramp-up, forecast errors are often quite consistent over the first five years of operation (Bain, R. & Polakovic, L, ?Traffic Forecasting Risk Study Update 2005: Through Ramp-Up and Beyond?, Standard & Poor?s, August 2005).