According to our union, the “cumulative loss to your pay since 2009 when compared to rises in RPI inflation is 14.5%”. Unsurprisingly, our employers would rather measure our pay in comparison with the more modest rises in CPI. They favour this price index in spite of the fact that the basket of goods whose prices CPI tracks is less well tailored than RPI’s basket to the expenses of a typical wage earner.
There is, however, one huge problem with RPI as a measure of inflation: for about 30% of the items whose prices it tracks, RPI uses a Carli formula, which (to quote from a document of the UK Statistics Authority):
takes the rate of change in each price, and then takes the arithmetic average of those changes…. The use of Carli is effectively prohibited by a legally binding European regulation … because it can be shown in certain circumstances that the use of Carli combined with chain linking of in-year indices introduces an upward bias known as ‘chain drift’.
It’s easy to demonstrate this upward bias. If an item doubles in price and then drops by a half, it returns to its original price, and therefore there is no price inflation, when we compare prices at the beginning and the end of the relevant period. But the Carli formula generates price inflation of (100% — 50%)/2 = 25% !
Because of this problem, RPI has been “found not to meet the required standard for designation as National Statistics”.
There is, however, a variant of RPI, called RPIJ, which strips out the unsound Carli formula, and replaces it with a sound Jevons formula (hence the ‘J’ in ‘RPIJ’), which takes the geometric rather than the arithmetic mean. RPIJ has National Statistics designation.
By the measure of RPIJ, there has been a smaller, 10.5%, erosion in pay by inflation since 2009.
Here are graphs that show changes in the pay of academics as measured by RPI (first graph) as well as RPIJ (second graph).
These graphs both reveal that there was a sharp increase in pay in real terms in 2008. This was due to the combination of (1) an unexpectedly generous 5% pay award in October 2008 (because the award was pegged by an agreement made years in advance to the September 2008 year-on-year rise in RPI, which happened to be unusually high) and (2) a period of deflation in the autumn and winter of 08–09, as the economy went into meltdown as a result of the financial crisis.
Given the unusual nature of the events of 2008, it’s not obvious that the peak in pay of 2009 should serve as the benchmark against which today’s pay is measured. It might be more reasonable to compare current pay with the long pre-financial-crisis period from 2002 to September 2008, when pay settlements kept pace with RPI and RPIJ inflation.
By that benchmark, pay has almost been restored to what it was in real terms in September 2008, by the sound measure of RPIJ as opposed to the unsound measure of RPI.
(Here is a link to a technical appendix to this post.)