By -

The all new 2007 figures from the Department for Transport for Vehicle Excise Duty (aka Road Tax) evasion have just been released.

You can tell something is going to be different, when you see statements like this:

Substantial improvements in the way that the roadside survey data are collected mean that evasion estimates for 2007 are not directly comparable with those from previous years.
Analyses of this year’s survey data also suggest that misread registration marks do not have a neutral effect on estimates as previously thought and, instead, tend to inflate estimates of evasion.

This all sums up to a breathtaking conclusion – the evasion estimates reported last year for motorcycle were probably overestimated by staggering 300% (or thereabouts – effectively the stats were done in such a different way that it is impossible to do a direct comparison. Note also that the figures for cars were also overestimated by a similar percentage – but with less dramatic effect or tabloid outrage.)

Put another way, the headline 38% evasion figure reported last year, and repeated last month with some anti-biker vitriol by MP Edward Leigh, were roughly 4 times higher than they should have been.

At least.

In fact there are still some problems with the reported figure of 9.8% evasion for bikers.

First the sample size is still very small – that makes the error margin over 50%, so (even taking nothing else into account) the figures for bikers could be as low as 4.7%.

Second, the change to the survey methodology that had the biggest impact was the switch to using Automatic NumberPlate Recognition (ANPR) cameras. Using these they were able to check misread plates for the first time, and found they were incorrectly matching to vehicles removed from the roads much more often than they had expected.

BUT, for collecting the data on motorcycles they did not use ANPR, but instead relied on contractors stood by the side of the road with a clipboard. It seems inevitable that a guy with a clipboard by a motorway trying to jot the number of a moving bike travelling at 70mph (let’s assume bikers don’t ever break the speed limit) is going to write down the wrong number more often than an ANPR computer which takes a still photo of the same vehicle and then uses Optical Character Recognition software to match up the letters, for the simple reason that, the computer doesn’t have to deal with the effects of a high speed movement.

Someone might have picked up on this, had the DfT not glibly stated in the previous years report, that they had computed the effect as a ‘slight upward bias.’ The admission that they got this so badly wrong will be little comfor

The DfT also notes they made a number of other changes to the statistical methodology, in line with the Southampton university report into their previous methods and assumptions.

It is therefore my opinion that the figures for tax evasion by motorcyclists, although markedly reduced and only a quarter of what was previously being claimed, is still a considerable overestimate.

If next year they manage to use ANPR to record motorcycles as well as cars, and also collect some hard data about relative mileages traveled by taxed vs untaxed motorcycles (which currently they only have for trucks), my bet is that the numbers will dramatically fall again.

But in light of this publication, where are the apologies.

Miscalculations of this magnitude represent some serious bungling by the ‘top statisticians’ we pay our taxes to employ. I think, at the very least, bikers are owed some major apologies from Edward Leigh, the House of Commons Public Accounts Committee, the Department for Transport and National Statistics. The DVLA probably deserve an apology too – they were castigated for their poor performance in managing tax evasion, even though their own figures suggested they were collecting more tax than ever.