RBA Annual Conference – 1992 Discussion

1. Jeffrey Carmichael

Let me begin by congratulating John on a fine survey paper to commence the Conference. Survey papers, however, are normally difficult to comment on because they are rarely contentious. John's is no exception, and so I will leave most of it as read and unobjectionable and focus my comments on three questions, only one of which plays a major role in his paper. My three questions are the following:

  1. Can monetary policy really control inflation?
  2. If so, why has theory not spelled out better the mechanics of the process?
  3. Why haven't central banks focused more vigorously on eliminating inflation over the past three decades?

You will recognise the third question as the focus of the first section of John's paper and, indeed, of several other papers at the Conference.

(a) Can Monetary Policy Really Control Inflation?

The answer to this question may seem self-evident. It struck me as interesting that John and several other authors at this Conference assert it as though it were hardly worthy of consideration.

John cites Friedman, Bomhoff and Fry as providing evidence that money and prices are correlated. But he also notes that money demand functions have become notoriously unstable. I should also point to Glenn Stevens' paper which offers little joy for the money/price relationship in Australia over the past decade.

My point, however, is not that the money/price relationship has changed, but the more basic one that money is not monetary policy. Even for those central banks that have retained some pretence of monetary targeting, monetary policy has often fallen well wide of hitting those targets.

Monetary policy in most countries is a complex process of intervention in a combination of the money and foreign exchange markets with a view to influencing the price of short-term finance or the price of the currency. This point is noted by John, and by others, but what is not emphasized is that there are, to put it back in Friedman's words, ‘long and variable lags’ between the policy and the impact. My guess, from the Australian experience of the past decade, is that the lags are up to two years or more and that the accuracy of the connection between the policy actions and inflation is extraordinarily imprecise. We might know the direction, but I suspect we don't know a whole lot more. Palle Schelde-Andersen, in a later paper, picks up on this point to some extent when he surveys the wide range of sacrifice ratios.

(b) Why Has Theory Not Spelled Out Better the Mechanics of the Process?

Following on from this point I find it disappointing that, as a profession, we still have not made very great advances in terms of modelling these linkages.

Despite the widely acknowledged focus on interest rates by central banks in their operations, most theoretical models still work with money supply as the monetary policy variable or, if they use an interest rate instrument, it is operationalised by simply inverting an assumed stable monetary demand equation.

It is time that our theoretical models took better account of the way in which monetary policy is implemented. In particular, there is a need to have at least a rudimentary term structure of interest rates in which the central bank controls the short rate while the long rate is influenced by a range of factors. Importantly, we need some fuzziness in the process of transmission from the monetary policy action out through the term structure and eventually into prices.

(c) Why Haven't Central Banks Focused More Vigorously on Eliminating Inflation Over the Past Three Decades?

My third question comes back more to John's paper. I liked the way he posed the question himself. Having concluded that monetary policy can control inflation, he asks:

Why do governments increase the quantity of money too rapidly? Why do they produce inflation when they understand its potential for harm?

Part of my answer would be that maybe they are less effective in stopping inflation than we often feel they should be – because of those imprecise links.

John suggests three other explanations:

  1. Time inconsistency;
  2. Mistaken analysis of the trade-offs; and
  3. Deliberate use of the inflation tax.

Like John, I am less than enthusiastic about the first and third explanations. The second explanation clearly holds some truth in that policy was, for a time, focused very much on trade-offs. In many ways it still is. Like John, I think our perceptions about those trade-offs are what have changed most in the past three decades.

But I think there is also another explanation. One important shift in thinking over the past decade has been a growing recognition of supply-side factors as a source of both shocks to the economy and interference with the transmission of monetary policy to inflation. John mentions demographics – a very important factor and one much underrated by economists. In Australia, we have had the Accord, major changes in the composition of the workforce and terms of trade shocks, all of which have contributed economic shocks of one kind or another.

The result is that price movements are influenced by a host of both demand and supply-side shocks. To counter these, and maintain a consistent anti-inflationary policy, requires both a lot of skill on the part of central bankers and a lot of confidence in the mechanisms through which policy can offset these shocks.

In practice, it is often easier, and I would argue in many cases more efficient, to focus policy on some nominal objective such as a short-term interest rate or the exchange rate, even though this may have the effect of short-term accommodation of an inflationary shock. This, of course, falls well short of actively seeking to exploit a short-term trade-off between inflation and unemployment, but it may still give the appearance of drifting with the tide.

Whether or not this is an efficient approach will, I am sure, re-surface in the discussions of the next day and a half.

2. General Discussion

In the discussion of Taylor's paper there was general agreement that for most countries the short-run Phillips curve was negatively sloped, and that the long-run curve was either vertical or positively sloped. However, for a small number of countries with severe labour market rigidities, it was conceded that hysteresis problems could result in increased difficulty in defining inflation/unemployment trade-offs.

With regard to longer-run inflation trends, some participants felt that monetary aggregates continue to provide useful information about inflation, and that inflation is essentially a monetary phenomenon. The evidence of the stability of M2 in the United States was cited in this respect. Supporters of this view felt that the money supply mainly affected underlying inflation trends, and might be more important for high inflation rates than for low rates. However, a number of participants suggested that, unlike the United States, the evidence from many countries was that money and prices have no long-run statistical relationship with each other. In all of these cases, innovations in the banking industry had undermined the stability of money demand. Because of this, money did not explain the recent evolution of inflation.

Turning to more short-run cycles in inflation, there was broad acceptance of the view that the trade-off with activity was important. There was also wide agreement that monetary policy had to operate through variations in short-term interest rates to influence activity first and inflation with some lag. The debate here focused on how to minimise the short-run trade-offs, and frequent reference was made to Figure 8 in Taylor's paper, which shows that policy rules in small macro models generate permanent trade-offs between the variability of inflation and the variability of output. A number of participants were struck by the notion that achieving a low standard deviation for inflation might involve accepting a high standard deviation for output. There was considerable discussion about the relative benefits of output and price stability, and how to improve the trade-off by shifting the curve shown in Figure 8 towards the origin.

Some participants thought that taking account of the source of shocks to activity was important when setting monetary policy, if the trade-off between output and inflation variability was to be improved. Here the persistence of shocks as well as their volatility was generally agreed to be an important influence. In this respect there was some consensus that it would be inappropriate to offset price pressures arising from a terms of trade shock in Australia. Others suggested that the trade-off could be improved through policies aimed at facilitating job search in labour markets, and by increasing the flexibility of markets more generally. Still others felt that the trade-off was not independent of the public's view of the policy reaction function of the authorities. Greater commitment to price stability should improve the trade-off between output and inflation variability.

In general, there was some consensus that monetary policy and inflation should be thought about in a systematic way. While the unique features of each business cycle should be taken into account, some participants felt there were advantages in the adoption of some sort of low-inflation emphasis for monetary policy. Others felt that optimising rules that might be relevant in economic models were less relevant in practice, because relationships affecting the trade-off between inflation and activity changed over time.