Join our community of smart investors

Expecting inflation

It's difficult to tell whether the public's inflation expectations are rational or not. Luckily, though, it doesn't much matter
May 31, 2018

Is the economic theory taught to generations of students actually correct? Some Bank of England figures next week might shed light on this.

Every three months, the Bank asks people what they expect future inflation to be. This matters because such expectations should in principle be partly self-fulfilling. If people expect prices to rise a lot, they’ll ask for pay rises to compensate and companies, expecting their rivals to raise prices, will do so themselves.

But how are expectations formed? Conventional theory says they should be rational. That is, they should embody all relevant information. If so, they’ll be correct on average. Of course, this doesn’t mean each individual is a great economic forecaster. It just means that their errors should cancel out on average.

Rational expectations are often contrasted to adaptive expectations, whereby our view of the future is shaped by what’s happened in the recent past.

Bank data allow us to see which of these theories is right. And there’s evidence that expectations are merely adaptive. Since the survey began in 1999 there’s been a huge correlation between what people believe inflation has been in the past 12 months (which isn’t the same as the official CPI inflation measure) and what they expect it to be in the next 12. It has been 0.88. The coefficient on past inflation has been less than one, however. This is consistent with inflation expectations being set by a rule of thumb: if inflation has been high recently, expect it to fall, and if it has been low expect it to rise.

But we’ve also seen evidence for rational expectations. A regression of actual perceived inflation to the rate expected 12 months previously shows that the coefficient on expected inflation is almost exactly one – which is what you’d expect if expectations were rational. Granted, there’s some evidence that people have under-forecast inflation, but the gap isn’t statistically significant; it’s due largely to the surprise rise in oil prices in 2007. And we can’t exclude the possibility that the forecast errors are indeed random, as rational expectations predicts (although testing for randomness in small samples is tricky).

The facts, then, seem consistent with both theories. There might be a reason for this. If the Bank of England was successfully targeting inflation, inflation would be around its target (2 per cent on the CPI measure) give or take random shocks such as moves in commodity prices or sterling. If this were the case, however, both adaptive and rational expectations would work. An adaptive expectation would be for inflation to be what it was, assuming that any shock would disappear. And a rational expectation would be for inflation to be around target, give or take a shock. It would be very difficult to distinguish between the two. And this is just what’s happened.

Of course, the Bank hasn’t wholly successfully targeted inflation in part because it hasn’t tried – in 2008-09 for example it preferred (rightly I think) to try to moderate the recession. But the point holds: with a more or less successful inflation target, it’ll be hard to tell how aggregate inflation expectations are formed.

The good news, though, is that it won’t much matter. The target should help to stabilise both actual and expected inflation. And we’re likely to see this next week: I doubt that expected inflation will have moved much. For those of us who remember the huge lurches in inflation in the 1970s and 1980s, this stability represents huge progress – something critics of the Bank of England and economics profession wrongly under-rate.