Conference – 1992 Discussion

1. John Pitchford[*]

Goodhart's overview of the prospects for central banking in the nineties is wide ranging and I shall not try to cover it all. In any case, I tend to agree with much that is in the sections I shall not touch on. My comments will be largely confined to questioning the basis for the consensus he finds on inflation as the sacrosanct objective of monetary policy for countries which do not target an exchange rate. Before coming to that I should like to comment briefly on his analysis of the policy situation of those countries which are members of the Exchange Rate Mechanism (ERM).

There are, of course, at least two basically different types of central bank in Western economies. The move toward closer economic union in Europe has meant that many countries find it convenient to tie their exchange rates closely to the mark. Being a member of the ERM requires a sacrifice, for the countries involved, of sovereignty over their monetary policy. In Austria I was told that there is a joke about this to the effect that the country does not need a central bank, only a telephone to the Bundesbank. While I feel that there are several reasons why this is not a fair comment, it does make the point that the situation of such central banks is so distinctive that it warrants separate and special treatment in any essay on the likely future of central banking. Goodhart has done this by including ‘an exchange rate’ as one of three possible intermediate targets for monetary policy.[1] He covers many of the questions this raises in section 1(b), though with two notable exceptions.

One of the classic issues of international monetary economics is the appreciation of sterling in 1924, and the effect this had of raising unemployment in the exportables sector for the rest of the decade. Hence I was surprised that this issue, of the level at which exchange rates were pegged by the United Kingdom and other European economies when joining the ERM, was not mentioned as a potential source of difficult adjustment for these countries. As well as a discussion of this point, I had expected a general assessment of the debate on the merits and demerits of joining the ERM and of establishing a common currency. Membership remains an issue for some countries, and the principles are of general interest. The creation of a common currency is yet to come. Moreover, is it inconceivable that there might be circumstances in which sovereign members might wish to withdraw?

It is almost impossible to discuss monetary policy without explicitly or implicitly treating fiscal policy. In what follows I am assuming that while it is possible to use fiscal policy for stabilisation, it is not nearly as flexible and perhaps not as effective an instrument of macro management as monetary policy.

For countries which float their exchange rate, the message of the paper, firmly and consistently repeated, is that ‘there is now a consensus that absolute primacy must be given to the achievement of domestic price stability’.[2] This position is reached by the following reasoning:

There has been no serious challenge to the claim that the medium- and longer-term Phillips curve is vertical, and hence that monetary policy should focus, primarily, if not solely, on controlling the level of some intermediate nominal variable, so as to anchor the rate of inflation at zero, or some very small positive number.

As Goodhart notes, there is a delicate issue of deciding what is operationally meant by ‘an absence of inflation’, but for simplicity I shall refer to and treat the consensus target as ‘zero inflation’.

I should like it to be clear that I agree with the proposition that the long-term Phillips curve is probably vertical.[3] Whether or not one is persuaded by the econometric evidence, the notion makes good theoretical sense. But if one believes, as I do, that the choice of targets for macro and monetary policy should be based on a reasonable assessment of the benefits and costs to the private sector of that targeting, for two reasons it does not necessarily follow that zero inflation should be the one and (almost) only target of the monetary authorities.

(a) If the Natural Rate were Steady

Suppose first that there is a unique unchanging natural rate of unemployment. If inflation is larger than the target level it is no guide to policy to specify that inflation should be reduced to zero. Aside from the unlikely world of new classical macro-economics, transitional costs of manipulating inflation are unlikely to be zero. A good example is Australia's present position which appears to involve an inflation rate between 0 and 4 per cent, and an unemployment rate over 10 per cent. Many hope this is a transitional phase, but would anyone seriously ask the Reserve Bank now (July 1992) to follow a monetary policy whose sole aim was to push the inflation rate closer to zero? Or again, the Australian inflation rate has, in the last few years, gone down by 4 or 5 percentage points, while unemployment has risen from 6 per cent to 10.5 per cent. If we could go back to 1988/89 would we really do the exercise the same way?[4]

To be able to infer from a vertical longer-run Phillips curve that central banks should target only inflation, it would seem to be necessary to observe that unemployment was near its constant natural rate, at least in some average sense over not too great a time period. For one thing this would follow if the economy showed signs that it converged rapidly to the natural rate. Alternatively, if, say, this quarter unemployment was 4 per cent and next quarter 12 per cent and the one after 6 per cent it might be possible to infer that irreducible random deviations about the natural rate were at work and that, the economy being as close to the natural rate as possible, the best that policy could do was to target inflation. But unemployment (and growth) do not behave this way, staying above or below average levels for considerable periods of time. That is, it is the transition phases of the Phillips curve which dominate experience. When in the last twenty years could it be said that the Australian economy (or almost any other Western economy) was near to stable equilibrium at steady natural rates for real variables?

There are several reasons which could explain why transitional outcomes seem to dominate equilibrium outcomes. One such is that central banks, in their pursuit of a variety of targets (including zero inflation), actually create (or exacerbate) cycles. It is this view of the world which inspired Milton Friedman's monetary rule, which I believe arose not just from a desire for lower inflation, but also for an absence of (publicly-induced) real fluctuations. Probably also it is the reason for the popularity of the inflation first and foremost rule that Goodhart's paper espouses. In part, it was behind my suggestion that the Reserve Bank of Australia should attempt to moderate the amplitude of the cycle in real interest rates.[5] But it does not appear to be a complete explanation of cycles. In any case, it only needs the central banks of one or two large countries to create cycles for the rest to experience these as exogenous fluctuations. The other main class of explanations is that adjustment, rather than being fast and monotonic, is slow and possibly cyclical (either of an endogenous nature or imported).[6] If this is the case, central bank policy is needed to offset fluctuations from domestic sources or abroad and/or to speed up adjustment. Such policy would need to take account of the performance of both real variables and inflation.[7]

(b) Hysteresis and the Natural Rate

Inflation could be justified as the sole target of monetary policy if the natural rate were steady, and the actual rate converged rapidly to the natural rate. But if the natural rate has any systematic relationship with other economic variables the case for the primacy of the inflation target must be slim indeed. It is hardly possible to interpret the unemployment experience of the last twenty years without conceding that the natural rate has moved upwards. In addition, if one accepts the (hysteresis) view that the natural rate of unemployment at each point in time depends on the path actual unemployment has taken, then the basis for concentrating monetary policy solely on the inflation target is entirely gone. Figure 1 shows the unemployment experiences of Switzerland and the United Kingdom, but almost all OECD countries follow a similar pattern. Sharp rises in the unemployment rate are associated with deep recessions and succeeding booms only gradually reduce those increases. The presumption that the natural rate has also risen is based on the observation that if it had not followed the actual rate upward, Phillips curve theory predicts that inflation through the eighties should have been close to zero, or even negative.

Figure 1: Unemployment Rate

Now, if the view is taken that the weight given to any target of economic policy should reflect the cost to the private sector of not achieving that target, a major objective of policy must be to reduce the natural rate of unemployment. There is no doubt that this should involve micro-economic measures to free up labour markets. However, consideration of the route by which the present levels of unemployment were reached strongly suggest that a long period of growth, uninterrupted by significant recessions, is the major requirement if unemployment rates are to be prevented from remaining high and eventually restored to earlier levels. Putting it another way, for Australia, the 1990–92 recession demonstrates how easy it is substantially to interrupt a process of growth.

This does not mean that inflation should be ignored. But it should be borne in mind that attempts to bring inflation down rapidly may be inconsistent with long-term steady growth, so that from time to time modest inflation would have to be tolerated. This was the policy stance of the Australian authorities for much of the eighties, and unemployment and inflation rates both came down from above 10tonear6 per cent between the 1982–83 and the 1990–92 recessions.[8] However, the further (in my view misplaced) objective of current account balance was accorded policy treatment inimical to steady growth, and the authorities underestimated the strength of their monetary tightening so that the Australian economy went into recession earlier and more deeply than it might have.[9] When hysteresis is significant in the determination of unemployment rates, monetary policy which attempts to bring inflation down rapidly will contribute to sustained high and possibly rising natural and actual rates of unemployment.

2. General Discussion

The general discussion of Goodhart's paper focused on three main areas:

  1. whether or not asset prices should be targeted by monetary policy;
  2. the Phillips curve trade-off and the time consistency problem; and
  3. the inter-relationships between monetary policy and the other arms of policy.

There was considerable discussion as to whether it is appropriate to target asset prices, and whether or not asset prices should be included in the price index used by the central bank. One argument was that people borrow to purchase assets, so that asset price inflation is relevant for interpreting the perceived level of real interest rates, implying that there might be some case for the central bank targeting them. However, it was generally felt that this would not be appropriate. Asset prices are too volatile to include in a price index, and changes in the relative values of assets are important in the adjustment of economies to changes in both real fundamentals and monetary policy. Thus, for example, if improved fundamentals led to higher profit expectations, asset prices should rise. This is not inflation, and it would be inappropriate to tighten monetary policy. Only the cost of the consumption stream from an asset should be included in the price index, not the price of the asset itself. Asset prices should not, however, be ignored in assessing inflationary pressure.

A number of participants focused on Goodhart's preference for a low-inflation objective, with relatively little weight being accorded to activity. Some argued that high unemployment was relatively inefficient for reducing inflation because of hysteresis problems. Long-term unemployed essentially leave the labour market, and therefore place no downward pressure on wages. Thus, attempting to achieve a large fall in inflation (even if there is considerable credibility) will necessarily involve a considerable cost in terms of unemployment. Some speakers thought that these costs dominated the costs of steady-state inflation at moderate levels. Others, however, pointed out that there was little international evidence favouring hysteresis, with the possible exception of the United Kingdom. Moreover, if policy became too focused on the unemployment costs of disinflation, then inflationary expectations would begin to rise again. Agents would fear that the authorities will reneg on the commitment to low inflation, so that time-inconsistency problems would emerge.

Finally, there was some discussion of whether prudential policy instruments (such as capital adequacy ratios) had some role as monetary policy instruments, or at least in easing the activity effects of tight monetary policy. Most speakers on this issue argued that supervisory instruments should not be used as a tool of macroeconomic management. While it may make sense from a supervisory point of view to vary prudential controls over the cycle, this issue should be decided independently of monetary policy.


I am indebted to Fred Gruen for discussion on a number of issues covered in these comments. [*]

The other two are ‘a monetary aggregate’ and ‘the direct targeting of inflation’. I would have preferred an even sharper distinction between the cases. [1]

Perhaps the flavour of the policy approach of the paper can best be gauged from the fact that it does not appear to contain the word ‘unemployment’. [2]

Goodhart also refers to but does not define a ‘medium-term’ Phillips curve. [3]

While these two examples are couched in terms of unemployment, they could just as readily be specified in terms of the consequences for real output growth or the deviation of real output from potential output. [4]

See Pitchford (1992), the second reference in footnote 9, below. [5]

Indeed, Goodhart appears to hold this view of the operation of the economy. See Goodhart, C. (1989), ‘The Conduct of Monetary Policy’, Economic Journal, 99, pp. 293–346 (especially pp. 334–336). [6]

This explanation of fluctuations appears to be implicit in Goodhart's exposition which has inflation apparently worsening of itself and refers to ‘inflationary/deflationary spirals’. [7]

A 6 per cent inflation figure in 1989 is obtained by the reasonable adjustment of excluding mortgage interest charges from the CPI data. See Tuck, M. (1992), Economic & Fixed Interest Outlook, Potter Warburg, Melbourne. [8]

See Pitchford, J.D. (1990), Australia's Foreign Debt: Myths and Realities, Allen and Unwin, Sydney, and Pitchford, J.D. (1992), ‘Macroeconomic Policy: Issues for the 1990s’, National Focus, 4 February. [9]