This column has argued in the past that it is a good idea to take some of the political heat out of this debate by looking at formal rules for setting optimal monetary policy. The rule that has been discussed here before is the so-called Taylor rule which has has been proposed by senior central bankers at the US Federal Reserve as a good way of setting "optimal" monetary policy.
As a reminder, the Taylor rule automatically increases real interest rates when inflation rises relative to target, and reduces real rates when output is below normal capacity. When this rule is implemented in the UK, it provides a reasonably good fit with the past behaviour of the authorities, and currently suggests that base rates are about right. Thus, while it would not support the Chancellor if he were to cut rates again, it does not suggest that he has already taken untoward risks with monetary policy.
However, an important criticism of the Taylor rule is that it has not always in the past given the right signals when crucial mistakes were being made in the setting of monetary policy. In particular, as the first graph shows, the rule suggests base rates should have been lower than they actually were throughout the period from 1986-88, whereas all economists would now agree rates should have been much higher. Since we would hope these rules would at the very minimum help us to "coarse-tune" the economy - in other words to avoid egregious policy errors - this is not especially encouraging.
Partly for this reason, the Bank of England has examined other automatic policy rules as well as the Taylor rule, especially one suggested by Bennett McCallum. (See an article by Alison Stuart in the August Quarterly Bulletin, and the Bank's Working Paper No. 45 by Andrew Haldane, Chris Salmon and Mr McCallum himself.) The McCallum rule has a lengthy pedigree, since its inventor has previously applied it successfully to both the US and Japan, so it is a welcome newcomer in the UK. It has every claim to be taken seriously.
Here is how it works. Instead of providing direct advice about the appropriate level of interest rates, the rule focuses on the growth of base money (M0). The rate of growth of M0 is set to be equal to the target growth of nominal GDP (say about 4.5 per cent per annum), subject to two qualifications. The first is that allowance needs to be made for the trend rise in the velocity of circulation of base money, which means that M0 growth on average needs to be kept systematically below that of nominal GDP. The second adjustment, a feedback mechanism based on nominal income growth, represents the guts of the rule itself. M0 growth is reduced by 0.5 per cent for every 1 per cent by which nominal GDP exceeds its target level, and vice versa.
Thus, the rule enforces a tightening in monetary conditions whenever nominal GDP is higher than its desired level, with the intention of bringing national income back towards its target with the usual lags.
At first sight, the McCallum rule looks very different from the Taylor rule, since the former sets an optimal level for M0 rather than interest rates, and since it has a feedback mechanism based on nominal GDP instead of the output gap and inflation. But in fact, the difference between the two rules is not as large as it seems. First, the optimal level of M0 can be translated into an implied level for interest rates fairly easily; in fact this needs to be done to make the rule operational because the Bank of England fixes interest rates directly, and not M0.
To do this, we use an estimated relationship that links the demand for M0 to the level of base rates and other variables, and then simply pick that level of base rates which produces the desired outcome for M0. Second, the nominal GDP target can be broken down into two component parts - real GDP relative to trend, and the price level relative to its target - which happen to be the two variables which form the feedback mechanism in the Taylor rule.
So there is considerable overlap between the two formulae. In fact, it is almost correct to view the McCallum rule as an augmented Taylor rule, where the augmentation comes from adding into the formula the extra information about the economy which is contained in the growth of M0. This is probably a good idea, since there is a vast amount of evidence in the UK which suggests that M0 is a very valuable leading indicator for inflation. (Incidentally, there is no doubt whatever, from a large number of econometric studies, that it is a much better leading indicator than M4, should any of the warring monetarist clans be listening.)
The second graph shows the Bank of England's estimates for optimal M0 growth under the McCallum rule, and compares these with the actual growth in M0 over the past decade. Basically, a comparison between the two series shows that M0 growth should have been kept much lower than it actually was between 1985 and 1990, implying that optimal interest rates under the rule should have been much higher than the government chose. Thus the McCallum rule, unlike the Taylor rule, gives broadly the right advice during the period of intense policy errors in the 1980s (and most significantly during the worst years of 1987-88). Importantly, the rule says that the current level of base rates should be increased slightly, which is more hawkish than the advice which emerges from Taylor.
However, before we jump to the conclusion, as some analysts have, that the McCallum rule is always superior to Taylor, consider the period from 1992-95. Taylor quite rightly advises sharply reduced interest rates over this period, while McCallum wrongly suggests that base rates should have been higher than the Chancellor actually set. So neither rule is perfect.
It would be useful to see a rigorous econometric horse race being conducted between these two rules so that we could establish which is superior over lengthy past periods. But in the absence of this, I would hope that our monetary policy-makers will examine both of the rules on a continuous basis, and think about their implications when setting policy.
One thing is for sure. I never met an econometric formula that cut base rates for electoral purposes. And right now the two rules taken together imply base rates should not be cut. The Chancellor should pay attention.Reuse content