House of Lords Debate: Government Statistics
Robert Skidelsky
Hansard: column 84-6 | Monday, June 01, 2009

My Lords, I should like to take the opportunity provided by the question of the noble Lord, Lord Hamilton, to raise two topics, one general and one relating to the particular issue of inflation statistics.
 
My general comment is that I am appalled by the degree of statistical illiteracy abroad. Almost every time I read a newspaper I am aware that the journalists writing it have no knowledge of statistics. They simply pluck out things to create stories, as the noble Lord, Lord Lipsey, said, and therefore there is constant statistical abuse. Of course, it is very hard to be against more information but sometimes I think we would be better off with less. A good example of that kind of statistical abuse is the debate on climate change.
 
I do not know what the answer is. You could say that, as part of their training, journalists ought to have a compulsory course in statistics and should not be licensed to write anything unless they do. That, of course, is not feasible. However, a more sensible suggestion is that health warnings should come with official statistical information. I know that ONS guidelines require that statistics come with health warnings, but the kind of health warnings that they come with are totally incomprehensible to anyone but those who write them. For example:
 
“The variance of the IoP is fairly insensitive to the assumptions made about the variance of the EPD. This continues to be the case at 4-digit level. Thus the assumption made about the variance of the EPD when deriving formula (4) should be suitable”.
 
You can do better than that.
 
If you want any of these official statistics to have any impact on the public, then alongside the necessary technical blurb you must provide much more user-friendly health warnings. One of the most useful that you could provide is a list of a few unlikely but possible events which would render the forecast invalid, such as the collapse by 25 per cent of US house prices between 2006 and 2008. A list of those kinds of unlikely “black swans”, as they have been called recently, would be useful to have.
 
My second topic involves the battle of the indexes—that is, the pros and cons of the RPI index and the CPI index. The change in the index for the purposes of inflation forecasting was made in 2003. I remember it very well because I was a member of the Lords Select Committee on Economic Affairs. The assumption was—we were told so by the Chancellor and other witnesses—that the two indexes would converge. That reflected the efficient market hypothesis: after all, you cannot have two indexes which measure roughly the same based on roughly the same things that may diverge in the long term. But, of course, diverge is exactly what they have done.
 
This raises the question of what the purpose of the change was. As I understood it at the time, the purpose was to lower the headline rate of inflation in order to present the inflation record of the Government in a better light and therefore to decrease wage pressure. What has happened though is that the two indexes have diverged considerably. Which index inspires more confidence as a measure of the rate of inflation? In terms of confidence in the economy, the consumer prices index is better since it regularly grinds out lower rates of inflation than the retail prices index. In terms of confidence in the statistics, however, the RPI might be better because, it seems obvious to me, any credible inflation index should include mortgage interest payments, especially in a country such as Britain where housing is such a huge economic component.
 
RPI, then, is a better index than CPI, but mortgage payments are not an accurate measure of the flow of consumption in the economy, especially if asset prices are going up. I wish we could find a way of incorporating the increasing prices of housing stock in the retail prices index so that it more accurately reflected the trend of transactions in the economy. I hope that we can work towards that, and that this may be one of the lessons we learn from the present wreck of inflation targeting.