One of the things we most like about writing for FT Alphaville is that the readers’ comments often make us look at the topic we’ve written about from an entirely different angle.
With that in mind, we want to focus on the reaction to a post last week about the yawning gap between what central banks think inflation is and what the public do. The point some readers raised was that, when measuring prices, statisticians look at the cost of technological progress in a different way to you or I.
While Joe Public might see the extra money he pays for this year’s iPhone compared with last year’s model as pure inflation, statisticians see part of that rise as a reflection of innovation — a better camera, say — and discount for it accordingly.
This can often lead to higher prices for new, but better quality, products being recorded in the index as deflationary — despite them costing more.
It is to our shame that we were not aware of this.
We also considered it worthy of further investigation, and so spent a chunk of Friday looking through the UK Office for National Statistics’ technical manual with respect to its inflation methodology to find out exactly what’s going on.
We should start with what the ONS thinks it ought to measure, via the technical manual: (our emphasis)
As a measure of price change alone, measures of consumer price inflation should reflect the cost of buying a fixed basket of goods and services of constant quality.
We’ve emphasised constant quality because we think that’s key to understanding what’s happening here.
Consider the advancement of the quality of televisions over the past decade. Ten years ago, the average set in the average home was probably smaller, with much worse sound, and far fewer — if any — smartTV features. Clearly these innovations make a difference to the quality of the experience. However, if the price has risen by a few hundred pounds, how much of that price rise should be discounted as a result?
The ONS looks at this in terms of a “direct quality adjustment”, which it describes as an attempt “made to place a quantitative value on the quality or specification difference”. Their statisticians use two methods to assess the value of the direct quality adjustment: hedonic regression and option costing. For technological innovations, hedonic regression is usually used. Here’s the ONS’ example for a PC:
Along with the formula used to calculate the “predicted price” of the old and new PCs, that is the price adjusted for the quality of this model:
The results of which are then used to calculate a new base price that reflects the difference in quality of the two machines.
What does this mean for inflation?
The first image shows the cost of the new PC was £550 compared with £475 for the old machine. However, once the technological improvements are taken into account, the base price — the representative cost of the old machine for the purposes of measuring inflation — becomes £562.40. Meaning the introduction of the newer model is deflationary. To the tune of 2.2 per cent to be precise. In contrast, without the regression, the inflation attributed to the new PC would have been 15.8 per cent.
A massive difference, then.
We can see the logic of what the statisticians have done here. But we think it’s another challenge to the myth that monetary policymakers can somehow control the public’s view of what inflation is through setting inflation targets.
Hardly anyone will see buying a new computer that costs them £75 more than the old one as deflationary. And yet that is what the measure the Bank of England targets (and uses to anchor public expectations) says it is.
We don’t see Apple stopping making new models of iPhones any time soon — even when the innovations introduced become ever more spurious and network-trapping in nature. So how can the central banks in the UK and elsewhere possibly hope to narrow the inflation expectations gap?
Related links
Is inflation higher than central banks think? FT Alphaville
