Four out of five people believe Cadbury’s creme eggs are getting smaller. That’s the result of an exclusive survey of my colleagues. Cadburys deny this. The eggs haven’t got smaller, they say – we’ve got bigger.
Whatever. This raises a neglected issue about measuring inflation.
The standard view is that reported RPI numbers overstate “true” inflation because they fail to fully capture the tendency for some goods to improve in quality over time; the RPI wrongly counts some quality improvements as a rise in prices.
Creme eggs show that this ain’t necessarily so. The quality of some goods can fall over time. If so, the RPI might understate inflation rather than overstate it. As Bart Hobijn points out, without discussing creme eggs, quality biases can go in either direction.
What’s more, the standard way of adjusting for quality doesn’t solve this problem. Hedonic pricing uses regression equations to estimate prices of older goods, based on measurable qualities, such as processor speed or memory size for computers. The price change is then measured from this imputed price, rather than actual prices of older, worse, goods.
However, if the fall in quality is subjective, as Cadbury’s allege, hedonic pricing won’t capture it. The RPI will then understate inflation*.
If this problem were confined to creme eggs, it would be small for practical purposes. But it’s not. The curly-wurly has been enshrunkened too. The quality of many football teams has declined since the 70s; think Leeds, Forest, Derby and Leicester. And then there’s the demise of spangles and furry Parkas.
All this has important implications.
1. The inflation target might be too high. There are good reasons to think the optimum inflation rate is negative or zero. But the target rate is positive, partly to accommodate the belief that inflation is lower than recorded figures say. But one of the upward biases in the RPI might not exist.
2. Official inflation figures are not pure facts. They’re built upon numerous theories and judgments; there are several theories of index numbers. In economics, fact and theory intertwine. The only thing the RPI tells us about with 100% accuracy is the RPI itself – not some “true” inflation rate.
3. For this reason, macroeconomists and journalists should not make a fetish of statistical precision. This much was well appreciated by economists in the 1930s, when interest in macroeconomic numbers was growing. One wrote:
The well-known, but unavoidable, element of vagueness which admittedly attends the concept of the general price-level makes this term very unsatisfactory for the purposes of a causal analysis…Our precision will be a mock-precision if we try to use such partly vague and non-quantitative concepts as the basis of a quantitative analysis.
That wasn’t one of the Austrian economists who have always been dismissive of macroeconomics. It was Keynes (ch3 of the General Theory).
* There are two objections to this which are wrong. One is that kids today see creme eggs the same way we did when we were kids, when they were the size of rugby balls. Consider, though, the utility of parents. In the 70s, they saw the creme egg as a nice novelty – it was launched in 1971. But parents today see it as a puny inadequate thing.
The other objection is that the fall in quality will lead to a fall in demand, so the weight of creme eggs in the RPI will fall. The problem here is that the basket is, necessarily, always one year out-of-date.
Spangle wrappers were so handy for fixing the windscreen wipers on Morris Minors. Bet the index doesn't allow for that.
Posted by: dearieme | January 19, 2006 at 10:45 PM
You don't have to go to anything quite as subjective as the eggs.
Does hedonic pricing track the pretty-much universally agreed upon *substantial* decrease in customer service over the last 15-25 years? (This has nothing to do with complaints about Indian call centers --- they have never bugged me. It has to do with ever more insulting phone menus "for your *convenience* we have ...", ever longer wait times, ever higher likelihood of being disconnected while on hold.)
Does it likewise track fallout from these matters like the *substantial* increase in identity theft (which is surely a commercial matter, given that the financial institutions could do something about it if they made the effort).
Is there information on whether hedonic pricing *ever* is used in a context where it would reflect a decrease rather than an increase in quality?
Posted by: Maynard Handley | January 19, 2006 at 10:54 PM
Put yourself in Brown or Greenspan's shoes. You desperately need high inflation to bring down national debt and deficits, and you desperately need low inflation to minimise annual index linked cost increases such as pension.
Enter hedonic RPI analysis. The answer to a con man's prayer.
Posted by: John East | January 19, 2006 at 11:34 PM
The other distinguished economic thinker called Maynard, John: quite right, boys.
Posted by: dearieme | January 20, 2006 at 01:36 AM
Real interest rates on government bonds of about 1% allow Brown to be relaxed about debt, surely?
And I recll Hamish McRae saying that a component of various indices where the quality was declining visibly was clothing. Not sure if that's still - in the last 3 years - true, but I know that shoes of the 60s lasted longer & wore better.
Posted by: dave heasman | January 23, 2006 at 04:56 PM
We don't 'need' inflation at all.
Read Mises 'Human Action' or look at historical periods of hard money. They were characterised by 'good deflation', the kind we dont complain about when paying half the price for a better spec'd 42" plasma this year compared to last year's model. The deflation we fear is a direct consequence of the deliberate debasement of the unit of currency, e.g. currently bogeyman being house price falls but think about 1929, or dotcom crash etc. Preceded by extreme credit/money creation and hence asset price inflation/speculation.
Posted by: Jonathan | January 24, 2006 at 08:50 AM