We're getting a new measure of inflation. From next Tuesday, the ONS's headline measure will shift from the CPI to CPIH, a measure that includes owner-occupiers' housing costs: last month, the latter was 0.2 percentage points higher than CPI inflation.
Although nothing much else will change - inflation-linked products will be linked to whichever inflation rate they are currently linked to and the Bank of England will continue to target CPI inflation - the shift highlights something important: that precise inflation is not an objective fact, but a mix of fact and theory unrelated to individuals' lived experience.
The case for the ONS's shift seems reasonable. Consumer prices should measure the cost of consuming stuff. Owner-occupiers consume housing services, so the cost of these should be part of the inflation measure.
But how do we measure them? The ONS has chosen a simple method: rents. They measure the cost of housing services by asking how much we'd pay to rent a similar home.
They do this because the alternatives aren't attractive. In theory, they could measure costs of home ownership such as repairs. They chose not to do this because it's difficult to get reliable consistent data on the price of repairs: anybody who's employed a builder will sympathise.
Or they could include house prices directly. Doing so, however, raises all sorts of problems. Let's take just one. Common sense says inflation is a measure of our purchasing power: if prices rise, we can buy less and if they fall we can buy more. But what if house prices rise? If you own a big home and are planning on trading down, your purchasing power has risen. But if you were hoping to trade up, it has fallen. So, does a rise in house prices mean that inflation has risen or fallen? To avoid problems like these, it's best to keep house prices out of standard inflation measures. Of course, there's a case for policy-makers to be concerned if house prices rise a lot, but this case is independent of whether they are part of a CPI.
So, has the ONS made the right decision? Maybe not. There's something odd here. If, say, petrol prices rise, I'm out of pocket: I must hand over more money to BP. But if rents rise, then as an owner-occupier I don't need to spend more. So why include them in an inflation rate at all?
The point here is that the ONS has made a choice on practical and theoretical grounds, and its choice, as it says "is not a clear one". Inflation isn't a fact that's out there, but something that's constructed on theoretical grounds.
This is not the only example of what I mean. One difference between the CPI and RPI is that the former uses geometric means whereas RPI uses arithmetic ones. Geometric means are lower than arithmetic means. In effect, they give more weight to prices that fall relative to others. This makes sense if consumers shift spending a lot in response to prices, but not if they don't.
And then there's the matter of the weight items have in the CPI. These reflect our average spending. But nobody is average: who spends 2.6 per cent of his monthly income on new cars, which is their weight in the CPI? This is why some people disbelieve the inflation numbers. If you spend more than the average person on cigarettes, car insurance and petrol, inflation is higher for you than the official figures suggest.
Anand Menon, a professor at King's College London, tells the story of when he was arguing against Brexit in Newcastle. He claimed it would cause GDP to fall. A heckler replied: "that's your bloody GDP, not ours". This is true of any macroeconomic aggregate: these are theory-laden constructs which aren't necessarily representative of individuals' actual experience.