We can expect to hear an awful lot about the closing gap between pay and inflation over the next few months as, inevitably and thankfully, on some measure we close in on a “cross-over point” where wages overtake price rises.
But this poses the question: which measure of inflation and, indeed, which measure of pay?
Confusion abounds on this – and this provides the space for different political parties to choose numbers which stand up the story they want to tell about the recovery and living standards. Get ready for a war of competing statistics.
When it comes to pay, average wages (that is, mean pay across the economy or, more accurately, across employees) regularly get reported as if they relate to the experience of a typical worker in the economy. They don’t – they are skewed by whatever is going on at the top of the distribution. For this reason we need to look at median pay – that of the typical worker. While the period since the financial crisis have been marked by relatively even movements in pay across the earnings distribution, the typical experience in recent decades has been for the mean to significantly outpace the median, reflecting growing wage inequality. No one knows for sure how this will pan out during economic recovery, but few would be surprised if the historic relationship resumes.
The trickier issue is the measure of inflation that should be used to deflate trends in wages. And here there is a bit of disarray. This debate may sound nerdy – indeed, it is quite nerdy – but it matters and we are going to hear a lot about all this, so it’s worth reflecting on.
The Retail Price Index (RPI), introduced after WWII, was traditionally considered the best measure for gauging what was happening to living standards, covering a wider suite of prices (and generally being higher) than the CPI which was introduced in the 1990s to meet the need for international harmonisation. Recently RPI has fallen out of favour. The formula it uses for aggregating prices (the Carli index, if you are into this sort of thing) has been fairly widely criticised and is thought to overstate inflation, leading the ONS to deem that it no longer qualifies as a National Statistic (though that hasn’t stopped the government from continuing to use it in relation to index-linked gilts and bonds).
This has left CPI as the main reported measure for inflation and it is used for uprating benefits, tax credits, pensions and tax thresholds (the government switched from RPI to CPI for uprating benefits from April 2011 and in doing so made a massive saving). But unlike RPI, CPI takes no account of a range of housing costs, such as mortgage interest payments. Arguably, it tells us quite a lot less about living standards.
The controversy about how to measure inflation is such that the UK Statistics Authority has established two reviews including one by the IFS’s Paul Johnson looking specifically at the arguments for using ‘cost of living’ or ‘cost of goods’ concepts in defining inflation. The former concept is likely to have more relevance for households and for the purposes of deflating pay and incomes; the latter is likely to be more useful from a macroeconomic perspective. As things stand, the various measures used in the UK tend to fall somewhere between these two camps.
Just to complicate matters further, two new measures have been already introduced: CPI-H (which adds an owner occupied housing element to CPI) and RPI-J (which maintains the RPI coverage but uses a more reliable formula similar to CPI). But neither of these measures is used by the government in policy formulation so when it comes to official wage projections we are left with the traditional choice between CPI and RPI.
To see how important – and politically relevant – these different measures can be consider this chart.
Source: OBR, Economic and Fiscal Outllook; and Resolution Foundation modelling
The CPI-deflated mean (average) wage projection is taken directly from the OBR’s latest Economic and Fiscal Outlook. It looks pretty rosy in the years ahead – at least compared to the recent past – and has caught the eye of many economic commentators. But it only tells part of the story.
If we want to get a sense of what this might mean for median pay we can adjust the average (assuming, as discussed above, that the relationship between the mean and median over the next few years is the same as that in the decade prior to the financial crisis).
What the chart shows is that if we then adjust this median wage figure for RPI inflation then pay looks set to fall in the years ahead. But if we use CPI it’s set to rise. And if we try and find some middle ground that avoids the narrowness of CPI or the unreliability of RPI, then we could use an imputed projection for RPI-J. (This assumes – imperfectly, but defensibly – that past relationships hold: holding constant the ratio between annual growth in the RPI and RPI-J in the years ahead, reflecting the relative stability of this ratio over the course of the history of the RPI-J). And under this RPI-J measure, pay is set to flat-line. So according to which measure of inflation you use wages are set to rise. Or flat-line. Or fall. Take your pick.
For now, at least, this leaves us in no man’s land. We are heading into a so-called ‘living standards election’ in which different sides will be able to tell whatever story they want about the prospects for wages depending on the measures used (with no official ‘best measure’). Add to this the fact that when it comes to what is happening to household incomes – a far superior measure of living standards – the only accurate data will be more than two years out of date by polling day. Given that some of our key economic measures are misleading and others are out of date, the electorate should stand ready to be bamboozled. Is this really the best we can do?