There was an interesting speech by George Osborne at the centennial of the Harvard Business School on April 8th. In it, he considers two radical departures in monetary policy terms from the “consensus” of the past fifteen years, which has favoured targeting retail consumer price inflation using interest rates. These two departures are:
· Including within the policy target some indicator of asset prices (or perhaps of credit — he exhibits studied ambiguity on this point)
· Using, in addition to interest rates, a system-wide capital requirement (as the BBC’s Robert Peston puts it “It would work by giving monetary authorities, such as the Bank of England, the power to oblige banks to hold more capital in their balance sheets relative to their loans or assets when the economy is growing strongly and less capital when — like now — it would help if the banks could be encouraged to lend a bit more...So, for example, in the good years the banks could perhaps lend £13 for every £1 of tier one [i.e. the more reliable] capital they hold – and that could rise to £17 in less benign times.”)
Though Osborne is careful not to over-commit and to say that these are ideas-in-progress, it seems pretty clear that he thinks recent events merit a possible significant shake-up in monetary policy.
I want to explain why inflation targeting has a potential problem with the kind of asset price developments we have seen in recent years, and why price-level targeting (which I have advocated for many years — including here, here, here, and here) does not suffer from the same problem.
First let us rehearse the difference between price-level targeting and inflation targeting. Under inflation targeting the monetary authority (e.g. the Bank of England) targets an annual percentage change in the price-level (currently, in the UK the price-level is defined as the level of the Consumer Prices Index (CPI) and the annual inflation target for 2008 is 2%). In contrast, under price-level targeting the target is a trend change in the price level. So, instead of there being a target of 2% inflation, then if the Consumer Prices Index were 100 at the start of 2008, the target would be for it to be 102 at the end of 2008.
So far, no practical difference. But now consider later years. Suppose that inflation were actually only 1% in 2008, so that at the end of the year the price-level were 101. Then under inflation targeting a 2% inflation target for 2009 would imply that the price-level targeted at the end of the year would be 103.02, whilst under price-level targeting the target would remain at a 2% rise on the 102 end-2008 target — so, 104.04. Under inflation targeting, the price-level is “reset” in the event of policy misses — bygones are left as bygones — whilst under price-level targeting there is an attempt to remedy past failures.
This means that the long-term price level (and hence the long-term inflation rate) is more certain under price-level targeting than under inflation targeting (other things being equal). Because of “base slippage” caused by the effective price-level target being updated each year, under inflation targeting the price level wanders about randomly, and after a few years the difference can become quite significant. This is a crucial point to grasp: under inflation targeting the price-level implied by the target for more than a few years ahead is subject to wide uncertainty, whilst for price-level targeting it is completely certain.
Now consider the following. Suppose that there is, at the same time, very high credibility to the inflation targeting regime but also a considerable change in the stock of money — for simplicity, let us assume a rise. Because of the high credibility of the inflation target, under certain (not regular, but also not extraordinarily unlikely) circumstances, it could be attractive to place that money into assets (say, the stock market, or houses) instead of buying goods. Now, the rise in the value of these assets (e.g. rising house prices) may tend to make people feel wealthier, and thence spend more, and thence create some inflationary pressure, but if inflationary expectations are extremely well anchored, this inflationary effect may be considerably dampened in the short term — or just appear as above-trend growth. Now, one day, off in the future, this extra wealth must either imply greater spending and hence greater prices, or the value of these assets must depreciate — for dynamic equilibrium, if the stock of money is to stay large indefinitely then prices must rise (so that the value of these assets falls in real, though not nominal terms), or, alternatively, the implied stock of money must fall at some point (implying something like a credit crunch or a period of very high interest rates).
But if credibility is sufficiently high, it might happen that the timescales here were rather long. So, if we take the case where matters are resolved through rising inflation, it could potentially be a number of years before rising asset values (driven by monetary expansion) turned into rising inflation — essentially, this would only happen once people ceased to believe that the Bank of England could plausibly raise interest rates high enough for long enough to at some point have a sufficiently sustained period of very low monetary growth (or even monetary contraction).
Of course, when this happened, that would mean rising inflation, which would imply a policy response from the Bank of England — much higher interest rates. So one might think that, earlier on in the process, the Bank would be able to look ahead and understand that if it did not act early enough, then it would eventually lose credibility and have to raise rates very high. Unfortunately, under an inflation targeting regime this isn’t quite so. Remember, under inflation targeting the implied price-level that the Bank would be targeting off far into the future is highly uncertain. Hence, early on in the process, when that money is turning into rising asset prices, the Bank doesn’t know whether it should care! Shocks in the meantime might offset the effects created by these rising asset prices, so that at the end of the day it did not prove necessary to raise rates very high in the end. Of course, such shocks would be possible under any regime (including price-level targeting), but because of the bygones-are-bygones nature of inflation targeting, the correct thing to do is only to care about policy over a relatively short timescale (two or three years). Off further into the future than that, matters are so uncertain (not merely about the effects of policy — concerning which matters are uncertain under all regimes — but also about the average objective of policy) — matters are so uncertain that their implications for current policy are very limited.
So, to put all of the above very bluntly, if asset price “bubbles”[1] aren’t going to turn into inflation (or, if they burst, into deflation) over a timescale of a few years, an inflation targeting central bank should not care about them.
The same is not true of a price-level targeting central bank. The reason is that a price-level targeter cares about the entire future pathway of prices — off forever into the future. The policy objective is well-anchored into the future, as well as in the present. Consequently, any current development that might affect the future path of prices will enter into consideration under the rule.
This means that if we had a price-level target, it would be simply unnecessary to have additional explicit targeting of asset prices. Of course there would be uncertainties over the implications of asset price developments for future price levels, and hence over how interest rates should respond, but these would be precisely the uncertainties over what is the appropriate target for asset prices. By using a price-level target, we would embody the things we really cared about — the effects of current asset price developments on future inflation and output — into one simple rule, rather than using a complicated proliferation of rules.
Given that price-level targeting has many other independent merits (explored in the links I gave above), and given that it represents an evolutionary, well-understood development of the current policy framework, I would urge that we give serious consideration to a switch to price-level targeting ahead of more complicated asset-price targeting rules, which would be hard to devise, would probably result in policy-makers second-guessing financial innovations and forecasting “appropriate” stock market movements (creating serious price endogeneity problems), and would represent a large change from current regimes. Inflation targeting isn’t perfect, and I (obviously) believe it can be improved upon, but we should avoid casting out the whole of what has been a very successful system if we can find a sufficient organic development of it. I suggest that price-level targeting is just that.
Update 12:15
I'm just back from Osborne's speech in which he attacked Brown's management of stability, prudence and productivity. One failing of Gordon Brown that he didn't mention, but perhaps should have done, was the euro. By the late 1990s it was already known that there were ways to build on inflation targeting - at that stage the principal UK advocates of price-level targeting were myself, Patrick Minford and Mervyn King. But price-level targeting (as with other credit-managing frameworks) is a long-term monetary policy. But New Labour's policy was (and indeed still is, in principle) that in the long-term the UK should not have its own monetary policy - we were to join the euro! So New Labour could not possibly commit to the long-term frameworks that would have been required to deal with asset price problems.
I would say that this is a good example of the consequences of dithering, and the perils of sticking too long with holding formulae along the line of "when the time is right". A less for us, there, I think...
[1] I don’t like the word bubbles, or the underlying concept, but since everyone else uses it, then when expressing myself bluntly I guess I’ll have to...
