11 March 2019

Why we’re getting inflation wrong: The paradox of the Amazon Echo


Today we learn that Amazon’s Echo speaker and similar smart speakers have been added to the basket of goods the Office for National Statistics uses to calculate inflation. It’s another useful reminder of how we tend to overstate inflation and thus understate economic and real income growth.

It’s a problem that has also been getting worse for the past few decades and it’s large enough to entirely skew what we think we know about living standards.

The basic problem is that not only do the relative prices of things change, but that people consume different things over time. Vesta curry kits were all the rage in the 1970s and who would bother with them these days? Landlines are becoming increasingly obsolete thanks to mobile-only packages, so why would we include the former in the inflation statistics? These kind of changes need to be reflected in how inflation statistics are measured over time.

To do this, statisticians come up with a “basket” of consumer goods (the ONS provides a good and deep explanation here). That basket measures price changes in some thousands of goods – a lot, certainly but nowhere near the tens of millions actually on the market at any given time.

To try to make the basket representative of consumer activity, it’s weighted across what most of us tend to buy and entirely ignores things which are rare. Any system of statistical sampling is going to have to do something like this. So, eight per cent of the basket is food – an interesting example of how rich we are as a society, for it’s only very recently that household food bills have taken up so small a portion of total income. A hefty 30 per cent of the basket is rent and housing expenses, with various other items making up the remainder.

As new things are invented a decision has to be made on if and when to add them to the basket. The obvious thing to do is to add them when most people are spending money on them. Mobile phones didn’t exist in 1975, today they’re a commonplace. At some point they’ve got to go into said basket. This is where a big methodological problem comes in.

These new technologies always start out as rich men’s toys. In the early 80s mobile phones cost as much as a decent second hand car and air time per minute was around the hourly wage. But the inflation – or deflation – rate of rich men’s toys is not much use to the ONS, which tries to measure the average experience. So the phone, and its declining price, only comes into our consumer basket when it’s actually a usual consumer purchase. That is, when much of the deflation has already happened.

The Amazon Echo is a perfect example. Today’s price is around £70, according to one survey. The inflation rate will therefore record any deflation in that price over time (and deflation it will almost certainly be, given electronics prices). But the speaker only appears in the consumer basket after a lot of deflation has already happened. This second generation of speaker started out at £89.99, compared to £150 or so when the first generation was released. Leave aside the problem of hedonic adjustment – how much tech improves over time – and just think of the price.The way official inflation is calculated ignores fully 50 per cent of the deflation that has already happened.

There is no real cure for this either. There is little sense having goods in the calculation basket until enough of us are buying the thing to make it representative. At the same time, if an item is not making it into the basket until 50 per cent or more has come off its price, clearly inflation is not being measured accurately.

This conundrum has knock-on effects for the rate of real economic growth, that is growth in GDP minus the inflation rate. If inflation is being over-estimated, then it follows that real income growth is being understated. The way inflation is put together therefore systematically records less growth in our standard of living than is actually happening.

This is not just some tiny statistical quibble. Entirely reasonable estimates in the US say that this is a 1 per cent a year problem, perhaps even 1.5 per cent a year. When compounded over the decades that leads to wildly wrong estimations of the rise in living standards. It means that over the decades since neoliberal globalisation started in the 1980s, real income growth may have been underestimated by 80 per cent or so, which is quite a lot really.

ONS tells us that this is a smaller problem here in the UK, which they would say, wouldn’t they? But note that even they acknowledge the problem and also that there’s no cute cure for it either.

To add something which is obviously true but not so generally widely trumpeted. The faster technological development the worse this problem is. If we were in an entirely static society in terms of new products then it wouldn’t exist at all. If every year saw a new rich man’s toy become an everyday purchase, then the problem would be very large indeed.

This is a particularly important point when it comes to the changes of the last few decades. The usual complaint from leftwingers has been that neoliberalism has not translated to improved lifestyles for the man or woman on the street.  “Just look at real wages!” they aver.

At the same time it is abundantly clear that our lives, even for the least well off, have been transformed by new technologies from the internet to mobile phones and all sorts of other miraculous inventions. The obvious way to reconcile these two apparent irreconcilables is that inflation has been consistently overstated during this period of rapid technological change, thanks to an inherent flaw in the methodology. And since we’ve been overstating inflation, the improvements in real incomes have also been underplayed.

All of which is rather darkly amusing. If there’s one thing globalisation or neoliberalism sells itself on, it’s advancing the spread of new technology and the corresponding improvement in living standards. But the very fact that the rise in living standards is coming from technological advance means that we undermeasure it — and it’s a problem to which there is no easy or apparent solution.

CapX depends on the generosity of its readers. If you value what we do, please consider making a donation.

Tim Worstall works for the Continental Telegraph and the Adam Smith Institute.