You've come back from holiday, you've eaten a bit too much and you decide it's time to lose a few pounds. To gauge your progress, you invest in a set of digital bathroom scales. You then have to choose whether to measure your weight in kilograms or pounds. Once you've made your choice, you can then check from day-to-day or week-to-week how your weight-loss programme is progressing.
Your scales then develop a rather peculiar fault. Measured in kilos, you have steadily been losing weight. Measured in pounds, you discover to your horror that you have gained weight. Which measure should you believe? Other than looking at your body in the mirror (sadly, not the body of an Adonis) you have no idea: yes, you've cut down on the cream cakes, but perhaps you shouldn't have had quite so many chocolates.
Of course, faced with this dilemma, your best option is simply to head back to the shop to get a new set of scales which, presumably, you then take home with some trepidation.
In economics, you get only one set of faulty scales. And you can't take them back. Initial read-outs are subject to revision. Different yardsticks provide competing versions of the truth. At the aggregate level, perhaps there isn't an overall truth. The man who spends all his money on computers might think that inflation has been persistently heading lower, but the woman driving a gas-guzzling car might have a very different perception. Aggregation hides a multitude of individual experiences.
The problems don't end there. Different aggregations produce different outcomes. The UK retail prices index, for example, includes mortgage interest payments. As a result, when interest rates go up, so do prices. From a policy-making point of view, this is a silly result, because an increase in interest rates designed to lower inflation does exactly the opposite. When the UK moved to an inflation-targeting regime in the 1990s, mortgage interest payments were sensibly removed from the retail prices index, at least for the purposes of setting the inflation target.
If, though, a simple change like this alters the near-term path for inflation, can we be sure that our measures of inflation really contain any universal truths? And if we're not sure, what does this say about the effectiveness of inflation-targeting regimes?
At HSBC, we've recently taken a close look at American inflationary developments. The US is a good place to start: it has, after all, seen a much bigger rise in inflation than elsewhere, so if anyone should be worrying, it's the Americans and, in particular, Ben Bernanke, the chairman of the Federal Reserve.
We looked at five different measures of inflation. Three were so-called "headline" measures, including the volatile food and energy components. The others were "core" measures, which strip these supposedly distorting influences.
The five measures are shown in the table. Two are from the regular monthly consumer price index release. Two - the deflators - are the measures that Ben Bernanke prefers to focus on (his favourite is the "core" personal consumers' expenditure deflator). The final measure is a bit of an oddity: it is a "harmonised" measure of US inflation calculated by statisticians at the US Bureau of Labor Statistics based on the methodology used to construct consumer price inflation in the eurozone and the UK.
So which measure provides the truth? The answer, of course, is they all do. They all provide ways of looking at the truth through different prisms. Yet the results vary rather a lot. For example, the overall increase in prices recorded since 1997 seems to have been a lot bigger based on the headline CPI and harmonised measures than on the other measures. The measure offering the lowest increase in prices is, oddly enough, Bernanke's favourite: the core consumers' expenditure deflator.
These longer-term differences are relatively easy to explain. The headline measures have a big energy component and, hence, are bound to have risen more during this latest period of higher oil prices. Meanwhile, the deflators include services for consumers that are not directly paid for by the consumers (medical services funded by companies or governments are a good example).
These variations need not matter for inflation targeting so long as the biases are systematically in one or another direction: if you know that your chosen measure of inflation gives an unusually high or low reading relative to the alternatives, then it's easy enough to choose a higher or lower inflation target to match.
But what happens if the competing measures of inflation head off in different directions, just like my faulty scales? You'd hope this sort of thing wouldn't happen but, unfortunately, it does. Inflation using Mr Bernanke's favoured yardstick was, in 2004, remarkably low. The European-style harmonised measure, though, was rising rapidly.
At the time, the Federal Reserve was rather relaxed, choosing to raise its key interest rate by just a few basis points. Imagine, though, what would have happened if the European Central Bank had been in charge of US monetary policy. The ECB's view of price stability is an inflation rate of less than 2 per cent. Yet US inflation in 2004, measured on European metrics, was up to 3.7 per cent, sufficiently high, one might think, to fill Jean-Claude Trichet's expansive office with a large number of metaphorical kittens.
The Federal Reserve would doubtless argue that the harmonised measure is inappropriate for the US, having too big an energy component: differing tax regimes imply that US energy prices tend to be a lot more up and down than those in Europe. Nevertheless, my example demonstrates that there is no one truth about inflation. Our experiences vary according to the basket of goods and services that we choose to buy and any aggregation of those experiences is, by its very nature, a partially-subjective exercise. As the Manic Street Preachers might say, "This is my truth, tell me yours."
Aggregation difficulties create real problems for policymakers. Which measure of inflation has the biggest impact on inflation expectations? Which measure of inflation sets the going rate for wage negotiations? Should inflation measures include changes in house prices, or taxes, or various public sector charges? In the absence of precise answers - and there are none - we simply have to accept our central banks are forced to operate with the economic equivalent of my faulty bathroom scales. And because they are, it would be foolish to believe that inflation-targeting regimes, as successful as they have been, have abolished the ups and downs of the business cycle.
Stephen King is managing director of economics at HSBCReuse content