The half life of a radioactive element is defined as the time it takes for half of that substance to decay into another element. For example, half of any quantity of Uranium 238 will decay into Thorium 234 in 4,468,000,000 years. Clearly, the half life of Uranium 238 is very long; but some radioactive substances have very short half lives measured in hours, or even minutes. Using this as a metaphor, we can say that the life of an average British Government budget is roughly a year before about half of it decays into policies which need adjustment. This is why we have a budget every year.
About twice a year, the Chancellor of the Exchequer emerges from the dusty but serene peace of Number 11 Downing Street and delivers a budget or a Treasury spending review. The normal public response to this event, at least in previous governments, is that the nation grumbles briefly as their petrol, fags and beer are taxed a little more. Then they go back to work to pay for it. During the Brown/Balls era of Labour budgets, journalists and economists learnt to scour the dense pages of Treasury documentation to read the small print behind whatever it was that Gordon Brown had announced in his budget. Sometimes, it would take days or weeks to delve and analyse before the really bad news would slowly emerge into the public domain.
George Osborne has broken with this tradition of burying bad news and faults in the deeper detail of budget documents. He has bravely taken the decision to expose the bad news, the badly thought out policies and the naked politicking to instant public scrutiny. In his (apparently) heroic pursuit of transparency, Osborne has adopted the policy of inserting a dead rat into the body of his budgets. The dead rat is is discovered and discussed in the media – it might be a pasty tax, or VAT on tampons. Whilst the media are diverted with the taxable pasty, tampon or whatever, George hopes that no-one will notice the other problems with his budget. Unfortunately, each time a dead rat is discovered, it means that the policy has to be revised or dumped (George calls this “listening”) and George himself loses a little credibility each time.
Credibility decays into ridicule.
Yesterday, 18th April 2016, George Osborne released a Treasury analysis “The long term economic impact of EU membership and the alternatives”. The details were not actually published until late in the morning. This Spectator podcast was timed by Fraser Nelson at 10.30 am and the document had not yet appeared. However, by lunchtime, it was being pored over by many on both sides of the Brexit debate. After Fraser Nelson’s headlines describing it as ‘deception’ came John Redwood describing it as ‘absurd’. Later in the day, after a series of tweets came the penetrating observations of Andrew Lilico. Lord Lamont gave us this:
Meanwhile, amongst a series of tweets on the subject, Ruth Lea tweeted this:
There are many problems with this Treasury document, but the principle ones seem to fall into two categories:
- The extrapolation of the available data beyond the limits of its usefulness.
- Deliberate selection of model assumptions which exclude any benefits that might accrue from Brexit.
The sages of statistical modelling warn repeatedly that extrapolation of data should not be carried out beyond a certain point. This point is often a matter of judgement, because the confidence intervals of the original data are widened exponentially beyond the limit of that data. This means that within a very short time, any predictions based upon those extrapolations very quickly become uncertain – to the point where the prediction is meaningless. An example of the way in which predictions become uncertain, and then have to be revised is contained within this tweet from Ruth Lea:
That is: within 3 months forecasts for GDP in 2020 have been revised downwards by £72 Billion. 2020 is only four years away. The forecast given by the offending Treasury analysis is for 2030, i.e. fourteen years away. There is no way the data can be tortured to extend to this point and retain any level of credibility.
The second category is explained nicely by Andrew Lilico in his blog which has been linked above. The point of statistical models is that they are only ever as good as their assumptions. If the model assumptions are rubbish, or biased, then so will be the predictions obtained from them.
From somewhere near the bottom of the social heap, I have watched many budgets and political events with interest over the years. I have seen policies described as unfair, or swingeing or having unintended consequences. But I have never seen the basic methodology of reaching those policies destroyed so completely and in such a short time.
Yesterday’s Treasury analysis was completely demolished within six hours.
John Redwood described this document as “absurd”. Ruth Lea said it was “risible”. Boris Johnson summed it up as “bollocks”.
Whatever the result of the referendum, George Osborne’s position as Chancellor of the Exchequer will be untenable from 24th June 2016.