January 10, 2013 7:00 pm

Lies, damned lies and odd statistics

  • Share
  • Print
  • Clip
  • Gift Article
  • Comments
UK national statisticians are wrong not to change RPI

When the Office for National Statistics decided in 2009 to improve the way it collects clothing prices, the national statisticians thought this change would just make the life of price collectors easier. But the tweak also magnified the existing difference between the retail price index and the consumer price index. Inflation calculated through these two methods has diverged significantly, so much so that the former is now almost a full percentage point above the latter.

The CPI is the better measure because, as opposed to the RPI, it does not fail some simple statistical tests. Since there are flaws in the way the RPI is calculated, the national statisticians have asked themselves whether they should correct them. This question would have a seemingly obvious answer: if an index is broke, fix it. But this was not what the officials decided. The ONS will continue to calculate the RPI as if nothing happened.

This curious decision would be comical, except that it has big distributional consequences. The RPI underpins £294bn in index-linked government bonds. Pensioners and investors who have bought linkers only to defend themselves from inflation will continue to benefit unduly from a statistical quirk. Meanwhile, the taxpayer will foot the bill, estimated at about £3bn per year. The decision moved the market: index-linked gilt yields fell about 25 basis points in response.

The ONS – and the UK Statistical Authority that rubber-stamped its recommendation – argue that there is a “significant value to users” in maintaining the series. They may also have worried about the possibility of legal challenges from disgruntled investors.

Yet such matters are for the politicians to decide. Had the ONS and UKSA recommended to amend RPI, their decision would have required the approval of George Osborne, the chancellor. He could have vetoed the change, fearing a judicial review. But the statisticians should have made their decision on purely scientific grounds.

The ONS should reconsider its recommendation. But even if it does not, the Treasury should adopt the CPI as the standard index for its future issuances of linkers. It is regrettable that the Debt Management Office has stated that it will not change its old habits. Were RPI to diverge further from reality, the government could pay a price. The decision to insist on using an unchanged RPI harms the credibility of national statistics. The DMO should at the very least limit the damage this odd choice will have on the government’s finances.

Copyright The Financial Times Limited 2016. You may share using our article tools.
Please don't cut articles from FT.com and redistribute by email or post to the web.

  • Share
  • Print
  • Clip
  • Gift Article
  • Comments
SHARE THIS QUOTE