In the debate about whether to change the inflation target, there’s one point that both sides agree upon – that inflation expectations are important.
Supporters of a change hope that expectations of higher inflation or higher nominal GDP growth would encourage people to spend more now, partly because higher expected inflation would mean even lower expected returns on cash.
Opponents of a change, such as Martin Weale, fear that if expectations of higher inflation get built into people’s behaviour, then it will be harder for the Bank of England to reduce inflation when it needs to.
I’m not sure either argument is right.
The Bank of England has been measuring the public’s inflation expectations since 1999. Every three months, it has asked people what they think inflation has been in the last 12 months, and what they expect it to be in the next 12. The correlation coefficient between the two answers has been a whopping 0.87 since 1999. This tells us that inflation expectations are shaped by experience of actual inflation, not by the policy target; if the latter were the case, inflation expectations would be stable and they are not*.
Granted, inflation expectations tend to rise and fall less than actual inflation. But I suspect this is because people rightly see that there’s a big transitory element in inflation – thanks to fluctuating oil prices or VAT rates – rather than because they trust the Bank to hit the inflation target.
It’s quite possible, then, that policy targets matter less for the public’s inflation expectations than the textbooks assume. This is not surprising, as standard textbook models still tend to ignore cognitive biases, not least of which is limited attention; people just don’t pay much attention to monetary policy-makers.
Calls for a credible money GDP target, then, are question-begging: they beg the question of how to achieve credibility when the public pays little attention to policy targets. Equally, it’s wrong to worry that a change in the policy target will raise inflation expectations, as it might well not.
There’s another problem. The correlation between inflation expectations and the savings ratio is slightly positive – 0.16 since 1999. Higher inflation expectations are, if anything, associated with higher savings. This is not just true in recent years; when inflation (and presumably inflation expectations too) rose in the late 60s and 1970s, the savings ratio also rose.
This suggests that, even if a shift in policy target does raise inflation expectations, it might actually depress economic activity rather than raise it. There’s a simple reason for this. If people expect negative real interest rates, they will save more in an effort to maintain their future real wealth. This income effect offsets the substitution effect, whereby negative real interest rates encourage spending rather than saving.
This implies that, even if policy-makers could engineer a fall in the public’s expectations of real interest rates – either by raising inflation expectations or by pledging to hold down nominal rates – they might not boost economic activity, and could even depress it.
My point here is that policy has less impact upon expectations than thought, and that the link between inflation expectations and activity is ambiguous. I fear that the emphasis placed upon expectations in the monetary policy debate is an example of what Simon Wren-Lewis has called “zero lower bound denial” – a pretence that monetary policy remains powerful even at zero interest rates.
In this sense, whilst I think there’s some merit in shifting to a money GDP target, this merit is easily overstated. Instead, the question should be: how much power does monetary policy still have – be it QE, guidance about future interest rates, helicopter drops or whatever? If you’ve run out of ammunition, it doesn’t much matter what your target is.
* Well, stableish. The inflation target was changed in 2003 from 2.5 per cent RPI to 2 per cent CPI. There’s no evidence in the data that this affected inflation expectations.
MORE FROM CHRIS DILLOW....
Chris blogs at http://stumblingandmumbling.typepad.com