From the TUC

Controversies around inflation measurement: Have annual real wages fallen by £2,100 or £1,200 or £800?

22 Mar 2017, by in Economics

Whatever way you look at it, this week’s inflation figures illustrate the threat to living standards that result from the fall in sterling after the referendum.

Headline inflation in February rose to 2.3%. Last week’s average earnings data for January (regular pay) was 2.3%. Real earnings growth is therefore zero. If as is likely inflation continues to rise and earnings growth continues to fall, once more pay growth will fall behind rises in the costs of living. ‘Once more’, because of course ever since the financial crisis this has been the norm.

But the story is complicated by the ONS leading on a new measure of ‘CPIH’ inflation rather than CPI inflation.

As the ONS put it: “This measure extends the Consumer Prices Index (CPI) to include a measure of the costs associated with owning, maintaining and living in one’s own home, known as owner occupiers’ housing costs (OOH), along with Council Tax.”

As it happens, this month both CPI and CPIH are 2.3%. But this is the exception rather than the rule – though the figures are not vastly different. Over the short period from 2005 for which we have CPIH data, CPIH is a little lower than CPI – average annual growth of 2.2% against 2.3% on CPI. Both are greatly lower than the older and more familiar RPI measure, which averages 2.9% pa (and 3.2% for February 2017).

But even small differences in inflation measures can lead to quite big differences over a number of years.

The TUC regularly derive cash figures for how far earnings have fallen behind prices since the onset of the financial crisis. When making these calculations choices have to be made about which measure of earnings and which measure of prices to use. For earnings we have tended to use average weekly earnings (as a timely monthly statistic) including bonuses (as the OBR forecast these figures).

We now have a choice of three measures of prices: RPI, CPIH and CPI. The chart below shows the consequent real earnings figures . The cash amounts are the differences between current earnings (in 2016 prices) and the pre-recession peak on each measure. On RPI we are £40 down; on CPI, £23 and on CPIH, £15. In the headline to this post I have turned these figures into the annual rather than weekly salary impact (multiplying the weekly figure by 52). These differences matter.

Real wages on different inflation measures, £ per week 2016 prices

Source: ONS and TUC calculations

We have tended in our commentary to use the CPI-based figure, under the justification that it is the government’s preferred measure of inflation. But in doing so we remain conscious (and are often reminded) that the government’s preference for this measure is contestable, especially when thinking in terms of the cost of living (as in this sort of analysis).

The introduction of CPIH is therefore another controversial move in a series of longer-standing controversies around inflation measurement. The main focus of the controversies this week is how to measure housing costs – and despite its name, this hasn’t necessarily been solved by the introduction of CPIH.  Below, we set out some of the main continuing areas of disagreement.


It seems pretty clear that changes in housing costs should be a key part of how we measure living standards. But measuring the cost of owner-occupiers’ housing – and how this is changing over time – is a massive challenge (and obviously it is impossible to do justice to the issue in a single paragraph – the ONS briefing is here). Many consider that house prices don’t fit the bill, as houses are normally regarded as an asset. The obvious alternative is to use mortgage interest payments (MIPS) as a proxy, which is the RPI approach. CPIH uses rental equivalence, which regards houses as providing a flow of services, and that flow of services can be ‘proxied’ by the rental value of an equivalent property. For me MIPS seems more intuitive and the most relevant to cost of living calculations, as well as corresponding to an actual flow of money. But the majority view in the price statistician community is for rental equivalence, though this is far from unanimity.  Andrew Sentence’s considered commentary gives a fuller account of the case against: here.

In fact the statistics watchdog (see discussion below) reports:

This degree of user scepticism and disagreement is, in our experience, unusual for an official statistic.

They also stress that the ONS face an uphill battle convincing the public that a measure with a bigger role for rental data would have a lower rate of inflation than the measure it replaces.

The CPIH measure also includes council tax, which seems sensible. Not least given bills are likely to begin rising, with councils pushed to breaking point by austerity.


Outside housing, the other key difference is our old friend the formula effect. This technical issue (on how individual prices are averaged together) has meant that RPI inflation is about one percentage point higher than CPI inflation. The ONS view is that the RPI approach is wrong and CPI is right. But others contend that the case is not so clear cut, and while the RPI approach is likely to overstate inflation, the CPI approach might understate. The CPIH approach is exactly the same as CPI, and so the same applies. Since we last reported on these discussions (here) the ONS have given work on the formula effect a higher priority, implicitly (as far as I am concerned) recognising that matters remain unresolved.

A household inflation index

One result of all of these tensions is that the ONS have agreed to produce a ‘household costs index’. This is thanks to pressure from the Royal Statistical Society for a measure that better reflects the cost of living. The detailed approach is now under development by the ONS, and experimental data (including some back data) will be published at the end of the year.

The RSS see the measure as a potential replacement for the RPI. There have also been encouraging noises from the National Statistician about addressing the formula effect in this context, which would be an important (necessary) step.


We share reservations expressed by various commentators and users, not least the Royal Statistical Society, that the ONS are in too much of a rush.

This appears vindicated by the curious decision on the part of ONS to move ahead with CPIH when the official statistics watchdog less than two weeks ago refused to give the measure a clean bill of health. They explain:

This is because although considerable progress has been made, ONS has not yet fully addressed some of the Requirements in the Assessment Report, particularly related to comparisons with other sources, explanations of the methods of quality assurance and description of the weights used in the calculation of CPIH.

(full material is here)

These matters have proved intractable since at least 2010 when George Osborne started selectively to ditch the RPI in favour of the CPI to support the public finances.

Rightly or wrongly the subsequent trashing of the RPI originally looked like an after the event justification for these actions.

Cynicism is not well served by the holders of index-linked gilts being guaranteed RPI-based returns on an indefinite basis. Nor is it well served by the Treasury continuing to deploy the RPI in taxes and duties when to its own advantage – most recently with the uprating of air passenger duty.

As a result, it would be foolish and unjust to expect wage negotiators to use anything other than RPI inflation in wage negotiations.

The RPI may indeed have flaws. But the ONS have been unable to find a way forward that commands public trust – let alone agreement among experts. It is of great important to resolve the issue, but it is even more important that it is resolved in a satisfactory manner.