RBA Annual Conference – 1990 Discussion

1. Doug McTaggart[1]

This paper is a comprehensive coverage of the issues, presenting as many of the facts as are available. As such it achieves its goal in providing a very useful basis for discussion. However, the issues and facts are such that one can read the paper and arrive at entirely different conclusions than those presented by the author. This is a cause for considerable concern when it involves an issue as important and pervasive as inflation in Australia today.

The most disturbing aspect of the paper is Carmichael's much too sanguine attitude concerning the costs of inflation and, perhaps because of this, the very casual way in which he attributes the high level of inflation in Australia to the sanguine attitude adopted by the monetary authorities in pursuing their policy objectives, or lack thereof. I find this a damning criticism of monetary policy in Australia and one which urgently needs to be redressed.

My specific comments on the paper begin with the conclusions and work backwards through the analysis.

(a) The Conclusions

As presented in the paper, these are:

  1. Australia's performance in controlling inflation, relative to the rest of the world (or at least Australia's trading partners), was poor over the 1980s.
  2. This was the result of the conscious choice of the Government.
  3. These actions were justified because:
    1. the costs of inflation are low
    2. the costs of reducing inflation are high.

Carmichael has, implicitly, conducted a cost-benefit analysis and concluded that, even though inflation might be costly, the costs of attempting to reduce it are such that efforts in that direction are simply not worth it. However, as with all cost-benefit studies, one can reasonably ask whether or not all the relevant costs and benefits have been included, measured and appropriately discounted. The difficulty here, which opens the door to controversy, is that we have very few, if any, actual measures of the true costs and benefits of inflation. Certainly a casual reading of the press and various business journals would give the impression that others have performed the same calculus as Carmichael and arrived at very different conclusions. Are we thus justified in making essential policy decisions with respect to money and inflation in such an atmosphere of uncertainty and controversy? Would a different reading of the material presented in the paper lead us to act more cautiously and conservatively?

(b) The Costs of Reducing; Inflation

The penultimate section of the paper deals primarily with the costs of reducing inflation, or the benefits of allowing inflation to continue. Here Carmichael discusses two methods: (i) estimating Phillips curves; and (ii) conducting cross-country comparisons. The former approach, briefly considered, concludes that a one percentage point reduction in inflation would increase unemployment by one third of a percentage point. The author quite rightly qualifies these results with consideration of problems discussed at length elsewhere. In particular, the issue of exogeneity emerges. However, to my mind, the most damaging criticism of these studies follows from the rational expectations literature. To wit, irrespective of the causal links generating Phillips curves, can they be exploited for policy purposes? The evidence of the 1970s suggests that, for reasons we do not yet understand, attempts to exploit Phillips curves for policy purposes failed. The curves failed to exhibit stability when it was needed most. Why should we expect otherwise now?

As for cross-country comparisons, in the absence of clear indicators of the structural and policy differences between countries, what can we really learn from their experiences? Do their central banks have more or less credibility than Australia's? Are their labour markets as constrained as Australia's? Have countries like New Zealand really followed an anti-inflationary policy path without straying? These are all very important issues, all raised in the paper, but for which we do not have good answers.

Thus on the basis of my reading of the analysis in the paper, I do not feel justified in concluding that the costs of reducing inflation are intolerably high.

(c) The Costs of Continuing Inflation

With respect to the costs of allowing inflation to continue at present rates, Carmichael is quite clear:

“… it is difficult to mount a case that the costs are overwhelming” (Section 5, p. 333)

But the paper is less clear on what data this assertion is based. Carmichael deals at length with a wide assortment of costs, enumerated below simply to underscore just how pervasive are the effects of inflation, none of which are good. In order of presentation in the paper are:

  • the reduction in after-tax real rates of return, reducing incentives to save;
  • the direct costs of inflation – shoe-leather costs;
  • portfolio distortions induced because taxes and regulations differentially affect rates of return on assets in an inflationary environment;
  • inflation and taxation interactions as they affect investment decisions;
  • investment distortions induced as the user cost of capital falls when nominal interest rates do not fully reflect inflation;
  • bracket drift effects in the labour market;
  • menu costs;
  • other aggregate implications in the saving and investment markets – Tobin effects;
  • the rise in general uncertainty that is associated with periods of high inflation;
  • relative price distortions;
  • general redistributive effects associated with the widespread existence of fixed nominal contracts, or the absence of wide-scale indexing.

By any method of scoring, this is an impressive list of costs associated with high, possibly constant, inflation. And there are others Carmichael missed. Primary among these are the distortions induced by the interaction between inflation and accounting systems. The paper mentions in passing the interaction between inflation and accounting methods as they pertain to depreciation allowances. Similar problems arise with inventory valuation in inflationary periods. More general is the effect on reported earnings from firms during periods of inflation when asset write-ups are included in current earnings. Apparently there is no convention whereby firms either report normal earnings or earnings as they include asset sales and revaluations. How then are we to interpret reported earnings per share under current systems in inflationary periods? What is to be made of inter-firm comparisons of earnings or earnings per share figures when asset prices are rising in line with inflation and firms choose to write-up the value of their assets at different times?

In view of this long list of costs attributable to inflation, how, you might ask, does the author feel justified in asserting that the costs are, on balance, low? It would seem that conclusions follow in Carmichael's analysis because various distortions have been redressed in recent years, thus reducing some of the more outrageous inequities and distortions. But it does not follow from the fact the costs of inflation may have been reduced, that they are also low.

Moreover, my assessment of some of the costs is different to that taken in the paper. To take one example, consider the costs associated with the uncertainty generated in inflationary periods. Table 2 in the paper suggests that inflation was not too variable if one considers a reasonable confidence interval for the CPI to be 6.5–12.5 per cent per year, with a mean rate of 9.5 per cent per year over the period spanning the 1970s and 1980s. The same confidence interval for fuel and electricity prices is 3.5–15.5 per cent per year and for imports 0.0–18.0 per cent per year. These are probably the most relevant prices from industry's point of view. So it seems likely that uncertainty is much higher in the commercial sector where significant resources allocations decisions are made, than for the household sector generally. How much of this variability in imported goods and fuels prices is due to exchange rate variability which itself might be the result of high inflation?

Even high but constant inflation, such that the level of uncertainty attributable to relative price variability is diminished, may contribute to increased overall levels of uncertainty. This would occur to the extent that inflation is perceived as a problem, irrespective of whether or not it in fact is, and thus be seen to warrant some form of remedial policy action. Speculation would ensue as to the form, the timing and the effects of any prospective policy.

Carmichael rests his case concerning the low costs of inflation on the proposition that deregulation and other redistributive measures taken over recent years have fostered a de facto indexed economy. This is by no means apparent to me. A particularly important instance where indexation, de facto or otherwise, breaks down, is the incessant intervention in foreign exchange markets by the monetary authorities. Attempts to insulate domestic prices, hence wages, from incipient exchange rate depreciations, or attempts to protect exporters from incipient appreciations, mean that even if all other nominal prices are rising at the rate of inflation so that relative prices are undisturbed, the relative price of domestic output in terms of foreign output will be changing. This, of course, means that Australia's competitive position vis a vis its trading partners is negatively correlated with the difference between Australia's inflation rate and that of the rest of the world.

All things considered, I do not believe that we are justified in asserting that the costs of inflation in Australia are low.

(d) The Cost-Benefit Analysis

Finally, with respect to the cost-benefit study conducted by Carmichael, is the issue of the present value of the streams of costs and benefits, however they are evaluated at any point in time. The costs of reducing inflation can reasonably be considered to be finite. Some output will be lost, and output foregone once is output lost forever. But there is no reason to suppose that the anti-inflationary measures will permanently reduce the long-term rate of growth of the economy. On the contrary, the costs associated with continuing high inflation do just that. Moreover, the costs of inflation will continue indefinitely unless corrective action is taken. Thus, it is possible that, even if the short term costs of reducing inflation are higher that the short-term gains of allowing it to continue (a position by no means established by Carmichael and certainly not accepted by this author), the present discounted value of the costs (over an infinite horizon) is almost certainly greater than the present discounted value of the finite costs incurred when reducing inflation.

In short, I would hotly dispute the conclusion that a cost-benefit study of inflation, and the costs of reducing it, inextricably lead to a passive acceptance of the current policies and outcomes.

(e) Australia's Performance On Inflation

Carmichael's paper begins with a discussion of the causes of inflation, summarising both the theoretical literature and the literature as it applies to recent Australian experience. With respect to the former, Carmichael appropriately avoids a long discussion on whether money causes prices and income, or vice versa. He is clear, however, that in either scenario, the money supply must increase if prices are ultimately and permanently going to rise.

The real question that has to be answered then is: do monetary authorities willfully and capriciously pump up monetary aggregates such that inflation ensues, or have they locked themselves into a set of operating procedures that forces them to accommodate the smorgasbord of exogenous disturbances that incessantly impinge on national economies?

In Australia's case, Carmichael is quite clear that the latter outcome has, willingly, been adopted over the 1980s (and, in my opinion, over a much longer period). With monetary policy accommodating the Accord process in particular, Australia has endured average rates of inflation which are high and in excess of its trading partners. No relief appears likely in the short to medium run. And, since the long run is composed of a series of short runs, high inflation will characterise the Australian economy for some time to come. This can only be an acceptable policy if one believes in the cost-benefit conclusions presented by Carmichael. In policy circles this is apparently the case as the Australian Government has recently reaffirmed its commitment to sustaining the Accord and explicitly disavowed any attempts to tackle inflation head-on.

I strongly disagree with Carmichael's assessment of the costs and benefits of continuing inflation and urge a restructuring of the operating procedures of the Reserve Bank such that it can reasonably get on with the job of controlling both the money supply and the rate of inflation. At the same time, we need to devote considerably more effort to quantifying the costs and benefits of inflation, and of attempts to control it, so that we can begin to make informed policy choices about such important issues. In the meantime, before such intelligence becomes available would not a prudent assessment of the situation dictate that one err on the side of caution, taking a much less sanguine approach to continued high levels of inflation?


*Bond University. [1]

2. John Nevile[1]

There is much that I agree with in Jeff Carmichael's paper; but I will not spend time either saying how good I thought the paper was or listing points made, of which we should all take note. After a comment on the role of the Accord I will concentrate on the costs of inflation and the costs of anti-inflationary policy. On his retirement from the position of Governor of the Reserve Bank, Bob Johnston is reported to have said:

“To deal with inflation in a permanent way is to accept a fairly great deal of pain. I don't know whether we are ready to pay that price, I wish we were”. (Sydney Morning Herald, 19/6/88)

It is appropriate for the Reserve Bank to be our conscience in the anti-inflationary cause, but it will be a more convincing advocate, if it produces hard evidence on the costs of inflation, and on the costs of inflation at a rate in the neighbourhood of what we are now experiencing. Also, it is usual in economics to weigh benefits against costs. If calls for a strong anti-inflationary policy are to be based on economics not ideology, there has to be some sort of benefit-cost analysis in this area as well. Carmichael's paper is to be commended in giving a significant amount of space both to the costs of inflation and the costs of disinflation. I do not want to argue about anything he has said, but to take further some of his points, and also point to what I think is an important gap in his analysis.

My preliminary point about the Accord also fits into this concern about the costs of inflation and the costs of policies to reduce inflation. It is worth getting our understanding of the Accord right, precisely because the Accord has proved a relatively low cost method of restraining inflation. In this context “relative” means relative to the costs of restraining inflation largely by tight aggregate demand policy, which opponents of the Accord advocate. Two categories of costs flowing from the Accord have been mentioned at this conference; equity costs and efficiency costs. Putting the whole cost of restraining inflation on those people who are made unemployed is more inequitable than anything in the Accord. Output lost through restrictive aggregate demand policy is likely to be greater than that lost through Accord-induced inefficiencies.

In his paper Carmichael states that:

“wages under the Accord could be viewed as the primary determinant of base inflation, with monetary policy determining the path of real wages by the extent to which it accommodated the Accord”.

The Accord has certainly played a central and successful role in anti-inflationary policy since 1983. If we want this success to continue, it is important to understand the aims of the architects of the Accord, both to know what can be expected of it, and because, if it is to continue to be viable, it will have to be successful, at least to some extent, in the eyes of those who created it and who sustain it.

The initial aims of the Accord can be found in the original Accord document itself. This document saw the Accord as a policy:

“enabling Australia to experience prolonged higher rates of economic and employment growth, and accompanying growth in living standards, without incurring the circumscribing penalty of higher inflation, by providing for resolution of conflicting income claims at lower levels of inflation than would otherwise be the case”.[2]

But the Accord also was concerned with real wages, arguing for a “system of full cost of living adjustments” (ibid., p.412), although this was knowingly fudged by what has become known as the “Medibank fiddle”.

After 1986 there was a major explicit change in the Accord's aims with respect to real wages. To quote one thoughtful and prominent union leader, Les Fallick of the Public Sector Union:

“The central focus has remained an attempt to moderate money wage growth, and since the collapse of the terms of trade in 1986, to ensure an ordered reduction in real wages” .[3]

Carmichael is correct in stating that, as a policy instrument, the Accord is the primary determinant of base inflation through its determination of money wage rates. But this does not mean, as he implies, that those responsible for implementing monetary policy are free to determine real wages, by the extent to which the monetary policy accommodates the Accord. The Accord was and is concerned with real wages just as much as with money wages. If movements in real wages depart too far from these envisaged by the Accord partners, the whole Accord could well collapse. Carmichael's comments may describe the mechanisms for determining money and real wages. They do not do justice to the fact that the Accord effectively determines the limits for real wage movements as well as those for money wage movements, nor to the fact that for much of the eighties monetary policy stance appears to have been heavily influenced by an exchange rate target determined in turn by the need to keep inflation at a rate which would not force real wage movements outside the limits considered tolerable.

Carmichael has a comprehensive catalogue of the costs of inflation. At the risk of oversimplification these can be classified as effects on the savings ratio, effects on the distribution of income and wealth, and a large number of disparate factors which might be expected to reduce the rate of productivity growth. Let me make two brief comments and one longer one. The first brief comment is about the saving ratio. There is no doubt that inflation tends to reduce the share of profits in income[4] and this, together with the interaction of inflation and a taxation system based on historical cost accounting, in turn reduces corporate saving as a proportion of GDP. If the private sector saving ratio is unaffected by inflation, as Edey and Britten-Jones suggest, there must be offsetting movements in the household saving ratio. It is not difficult to produce reasons, ranging from superannuation arrangements to short-run money illusion, to explain a positive effect of inflation on the household savings ratio, so there is no difficulty in accepting the argument that inflation has little effect on private saving, at least at the rates of inflation experienced in Australia over the last 30 years.

Secondly, Carmichael mentions the effect of saving on the distribution of income and wealth. At a micro-economic level inflation must have some effect. So too must increased unemployment caused by anti-inflationary policy – not to mention the workings of the stock market and a hundred other factors in any capitalist or mixed economy. The point is that inflation is alleged to have a general effect of changing the overall distribution of income in an undesirable way. Work in the first half of the eighties by Gruen[5] and Nevile and Warren[6] has laid that ghost to rest.

It is true that this work mainly looked at income distribution and not wealth distribution, because of the paucity of data about the latter. As far as wealth distribution is concerned the main losers from inflation are those who invest in assets with values fixed in nominal terms. Except for the very rich, in Australia at least, these are the same group as those who mostly gain from inflation – home owners. Only 20 per cent of all net household assets in Australia are in the form of financial assets and, except for the very rich, owner occupiers have far more financial assets than renters – four times the level among low income earners. In any case the effects of inflation on wealth distribution were not big enough in Australia to change the pattern of income distribution.

Carmichael listed a number of effects resulting from inflation which will reduce productive efficiency in an economy and/or lower the rate of economic growth by reducing investment. I have no doubt that these effects exist, the question is how important are they? One way to get a handle on this is to look at the rate of growth of labour productivity in an industry such as manufacturing which is a prime candidate as an industry where one would expect productivity to be harmed by inflation. Labour productivity is appropriate rather than total factor productivity, because part of the argument is that inflation distorts investment decisions reducing the proportion going to “productive” sectors like manufacturing. Recently I happened to come across an article which gave rates of growth of output per hour in manufacturing for 12 OECD countries for three separate time periods.[7] I regressed these on the corresponding rates of inflation in a cross-section analysis with results as shown in the Table. This is certainly not a profound piece of research, but the results give pause to easy acceptance of assertions that inflation at rates now current in Australia has large adverse effects on efficiency.

Table 1 Regressions of growth in output per hour in manufacturing (PG)
on growth of the implicit GDP deflator (INF)
1960–73 PG
1973–79 PG

1979–85 PG


In the first time period the coefficient on inflation had the wrong sign. This positive sign is not just because Japan had both a high rate of inflation and a high rate of productivity growth. If one excludes Japan there is still a positive coefficient which is significant at the 10 per cent level, and all the countries with inflation noticeably above the average had very high rates of productivity growth (ranging from 6.4 per cent to 10.3 per cent).

In the second and third periods the coefficient on inflation has the right sign but is not significant – not even at the 20 per cent level. In fact if one looks at the scatter diagrams there is no pattern at all discernible to the eye.

It is true that in the period 1960 to 1973 the average rate of inflation across countries was lower than in the subsequent two periods and the average rate of productivity growth was much higher. The conditions before and after 1973 are so different that it is impossible to deduce anything from this. The average rate of productivity growth was actually slightly higher in the period 1973–1979 when inflation averaged 9.4 per cent than in the period 1979–1985 when inflation averaged 7.0 percent.

My next point concerns the analysis Carmichael makes in the last section of his paper on the costs of disinflation. Unless I have misunderstood what he has done, he has got rough estimates of the slope of the short to medium run Phillips curve by comparing Australia first with New Zealand and then with the U.S. and the U.K. What he has not done is look at the question of how long one has to endure these costs if the fall in inflation is to be permanent, and how long the effects of the lower level of economic activity last after policy is relaxed and the level of economic activity is back to what it was before the reduction of the inflation rate.

As far as the first question is concerned, it is ominous that in the case of New Zealand employment growth declined fairly steadily in the whole of the five year period, starting in 1985, in which New Zealand reduced its rate of inflation. There is no evidence there that the period of high unemployment can be restricted to even three or four years.

In the case of the U.S., while there was an initial sharp fall in the rate of employment growth and rise in unemployment, within three years output and employment were growing again at respectable rates and unemployment continued to decline while inflation was kept at a lower rate.

The U.K. falls between the other two cases and personally I find it hard to draw any conclusions from the U.K./Australia comparison. So unless you think that the Australian economy is more like that of the U.S. than it is like the New Zealand economy, there is not much joy in the comparisons.

But the question of whether it is necessary to keep employment and output growth at low or falling levels for only two or three years or for five or more is only part of the picture of the costs of disinflation. The low level of economic activity may have costs that last long after the economy is growing rapidly again. This will depend on the extent to which investment, rather than consumption, declines in periods of low economic activity, and the importance of hysteresis in the labour market. Casual empiricism suggests that both these factors are unfavourable in the case of Australia.

A final very brief point. I am inclined to think that one reason why many applied economists give a very high priority to reducing the rate of inflation is that they are rightly impressed with the costs of high rates of inflation and associate them, without much actual quantitative analysis, to rates of inflation between, say 2 and 10 per cent. Carmichael's paper is to be commended, among other reasons, as an effective antidote to that sort of thinking.


*University of New South Wales. [1]

National Economic Summit Conference, Documents and Proceedings, Vol. 1, Australian Government Publishing Service, Canberra, 1983. [2]

Fallick, L., “The Accord: An Assessment”, The Economic and Labour Relations Review June 1990. [3]

See Nevile, J.W., “Inflation, Company Profits and Investment”, Australian Economic Review, No. 4, 1975. [4]

Gruen, F.H., ‘Australian Inflation and the Distribution of Income and Wealth – A Preliminary View’ in Pagan, A.R. and Trivedi, P.K., (eds.), The Effects of Inflation.: Theoretical Issues and Australian Evidence, Centre for Economic Policy Research, Australian National University, Canberra, 1983. [5]

Nevile, J.W. and Warren, N.A., “Inflation and Personal Income Distribution in Australia”, Australian Economic Review, No. 1, 1984. [6]

Neef, A., quoted in Volcker, P.A., Education and Economic Development, The Inaugural Frederich H. Schultz Distinguished Lecture, Florida Institute of Education and the Florida National Bank, Jacksonville, Florida, 1987. [7]

(c) General Discussion

On the cost of inflation, several points were made:

  • Relative price variability rises with inflation, but there is at least some evidence that this is not dramatic.
  • The intertemporal effects of inflation are insidious and large. They come through misallocated investment – both in terms of the type of asset, and in the lives of asset (higher inflation tends to push businesses towards assets with a shorter life and discourage longer-term investment). Asset price inflation is also relevant – as businesses try to hold on to assets which are perceived as inflation hedges, rather than those which produce an income stream.
  • It was suggested that the taxation of the entire nominal interest receipt, and the deductibility of the entire nominal interest costs, leads to distortion. Higher inflation will mean that saving and investment decisions are distorted unless the nominal pre-tax rate rises sufficiently to equalise the after-tax rate of return and the cost of capital. Even if this does happen, it is possible that in a world where capital markets are integrated, the exchange rate may become misaligned. Flexibility of interest rates does not eliminate the distortion, it shifts it elsewhere.
  • It was also suggested that the real problem may not be so much inflation, as the tax system itself, and that if inflation cannot be fixed or if the costs of doing so are too big, we should focus on fixing the tax system.

Some questioned whether the costs of reducing inflation are as high as is commonly thought. Other commentators noted that, given the central role of wages, an important question was whether there was any constituency for lower inflation in the community generally and in the trade union movement in particular. The answer to this seems to be “no”. This being the case, there was some doubt as to whether attempts to lower inflation purely by using the wages mechanism would be effective. Why has it not been possible to use the Accord to wind down inflation more quickly so that, for example, wages are rising at 4 per cent and prices at 3 per cent? Some wage/prices nexus was foreseen as the Accord had originally been styled a wage/price accord. However, the micro framework did not ensure sufficient competition to enforce an immediate price response if wage increases were more dramatically constrained.

The discussion covered the role of monetary policy in inflation. Some disagreed with McTaggart's interpretation of the relationship between money and prices on the grounds that the money supply is endogenous. McTaggart argued that, while it is endogenous under present operating procedures, policy should be altered so as to make it exogenous, and therefore able to exert more of a controlling influence on inflation. Even without adopting a purely monetarist approach, most participants agreed that monetary policy does have a big role to play in controlling inflation. But some doubts were expressed as to whether the Accord process could be turned around and used as a supplement to monetary policy in reducing inflation, as opposed to monetary policy being used (at least in some views) as a supplement to the Accord.

Other points were:

  • The assertion in the Carmichael paper that 7–8 per cent inflation was accepted by the government through the 1980s may not be strictly correct. For example, it was pointed out that twice in that period, a forecast for inflation of around 4 per cent was articulated and formed an important part of the overall expectations on which wage plans were made.
  • Should central banks look specifically at asset prices, and look at them earlier in the economic cycle than may have recently been the case? There was a difference of opinion here: Carmichael was strongly against the idea of asset prices, which represent an intertemporal price, being made part of any official index of prices for policy considerations.
  • Other OECD countries had made their “breakthrough” to lower levels of inflation in the mid 1980s, greatly helped by commodity price movements (particularly oil). These same commodity price movements had adversely affected Australia and, through the depreciation of the exchange rate, made our inflation worse in this period.
  • There is some evidence that inflation expectations are slow to adjust in Australia; (for example, the Melbourne Institute's survey question shows this). This would have a bearing on the cost of reducing inflation.
  • The efficiency of using macro-economic policies to reduce inflation is greatly hampered by the micro-economic structure. Oligopolistic markets tended to give much more power to firms in setting prices, which would reduce the effectiveness of demand management policies in getting inflation down (or alternatively increase their costs).