Transport and the Economy

Further supplementary written evidence from Malcolm Griffiths (Bluespace Thinking Ltd) (TE 07B)

1. Based on our principle finding the Chair of the Transport Select Committee asked how it happens that the current appraisal methods favour major capital projects. Because this question is key to improving UK transportation decisions I would like to amplify for the record the reason we reached these conclusions by evidencing our findings.

2. Our principle finding is that the current methods of analysing and prioritising projects are not sound. The processes are over detailed and the guidance is based on very old data both for forecasting demand and evaluating benefits. The methods and assumptions used have a systematic bias towards long term, long distance, high risk, capital projects. The flaws in the analysis are evident by the application of degree level mathematics/statistics.

Long term –

3. Project specific webtag guidance for demand forecasting is based predominantly on correlation with GDP. Due to population growth and improved efficiency of physical activity (mechanisation, information technology etc.) it is reasonable to expect that GDP growth will, with good government decisions, continue to increase at about 2% -2.5% /year in real terms.

4. However due to the level of travel saturation in the UK it is much more realistic to start with a premise that overall travel growth will increase at about the rate of population increase nominally 0.7% / year and then assess impacts based on past and future predicted trends that will either reduce or increase travel from this current per person norm . For general forecasts (not project specific) the DfT predictions of travel growth are at about 0.8%/year, the Dept of Energy and Climate Change (DECC) and the independent Climate Change Committee (CCC) both use about 0.8-1%/year in their assessments of travel related emissions

5. Using GDP as the base may produce a reasonable result if an elasticity of about 0.3 is used however this would not be sound analysis.

6. Although dealing with a single travel mode a clear example of this can be seen be looking at the correlation over the last 15 years between long distance train travel growth and GDP growth.

7.
The graph and calculation show no statistical correlation, this is obviously a technical detail however assuming a causal correlation when one does not exist produces totally inaccurate forecasts, the errors are exaggerated the longer the time period considered. We would encourage interested economists or mathematician to consider if they can see a structural causal correlation between long distance rail travel and GDP over the last 15 years.

Long distance –

8. The elasticities to GDP specified in webtag refer to the Passenger Demand Forecasting Handbook (PDFH) version 4.1. which, as an example, create elasticities for rail travel originating in London to Glasgow of 2.12 and for Glasgow to London of 3.28.

9. DfT consultants have advised that these and other long distance elasticities are totally unrealistic and the DfT include in their revised guidance the comment "We agree that the PDFH 4.1 recommendations produce unfeasibly large elasticities over long distances".

10. Revised elasticities (which still do not take account of saturation and the lack of any correlation to GDP over the last 15 years ) of 0.9 for London originating journeys to Glasgow and 1.9 for Glasgow to London journeys are recommended in the proposed new version 5 of the PDFH.

11. However although the new version was complete in Aug 2009 issued in Jan 2010 and due to come into use in April 2010 we have been advised that the new guidance can not be used because the Secretary of State has not approved it for use.

12. While it will not be the only project impacted HS2 provides the clearest example of how these sort of errors flow through to project analysis conclusions. HS2 Ltd predict that long distance travel over 50 miles, in the areas under consideration, will triple to over 7million trips /day.

13. This conclusion, from their computer analysis, means that assuming other travel increases at 0.7%/year, in line with population increase, total travel will increase a minimum of 50%-60% by 2033. This is about twice that predicted by the DfT, DECC and the CCC. The over estimate of demand goes on to produce a Net Benefit Ratio (NBR) for the project above 2, when the demand forecasts are corrected the NBR will probably be nearer 1.

14. Because of the nature of long term, long distance infrastructure projects and the uncertainty involved, even if the analysis is totally accurate the projects are inherently high risk.

15. I would agree with Professor Goodwin that if the calculations are carried out correctly the overall approach to analysis probably provides an adequate framework. However as analysis becomes more sophisticated and computer dependant it is crucial that both input data and output conclusions are assessed against the criteria of reasonableness, using simple application of logic to establish reasonable boundary conditions for results.

October 2010