Inflation, recession and recovery: a nominal income analysis of the process of global disinflation
BIS Economic Papers No 11
1. The economic and policy background
The world economy has recently been through a period of severe, policy-induced disinflation. It is probably no exaggeration to say that the experience has been without precedent. For never before has the attempt been made - on a global scale and in a fairly concerted manner - to end such a prolonged period of inflation as the world has experienced since the mid-1960s.
The immediate trigger to a new degree of policy resolve was, of course, the second oil shock. The first oil price rise in 1973-74 had already apparently demonstrated how a pre-existing inflationary situation could be seriously inflamed by such an external event. Consequently, as the magnitude of the second oil price rise became apparent in 1979, policy-makers in most of the major industrial countries resolved together on a much less accommodative policy position than some of them had adopted in 1974-75. It may be noted at once, however, that while the oil shock made such a policy stance the more necessary, it also increased - at least to many minds - the likely short-run costs of disinflation in terms of low output growth and high unemployment. That is to say, policy was forced to confront both an ongoing, underlying, inflation as well as the direct and indirect effects of a further sharp rise in the cost of energy. But, short-run costs or no, policy-makers and others had become increasingly aware of the possible costs of inflation itself, especially those of an inflation apparently becoming more and more unstable. In this sense, the range of policy choices was felt by many to be narrower than it might have appeared at first sight.
Speaking broadly, the policy approach adopted gave primacy to monetary policy in the form of targets for one or more measures of the stock of money. For the purpose of the present paper, however, it will be more convenient to view the approach as one involving the setting of limits to the growth of nominal demand for output, that is, of money national income. Indeed, given that fairly stable relationships had been assumed between money stock changes and nominal income, the two amounted, in qualitative intent at least, to the same thing.
The important point about this policy setting was the "cap" which it placed on possible economic developments in nominal terms. Beyond the general intention that the secondary price effects of the oil shock should be firmly resisted, and that inflation should be brought down appreciably, no quantitative targets were set either for inflation itself or for real output and employment. Thus, the actual behaviour of prices and of output would depend on how the so-called price quantity "split" developed. In other words, the more that any slowdown in nominal income growth was reflected in a slowdown in inflation, the less would the "real" economy be adversely affected by disinflation, and vice-versa. Or, again, even if eventually the whole of the decline in nominal income growth appeared in the price component, the extent of any lag in the reaction would determine the severity of the temporary loss of output and employment. The nature and development of the price/quantity split is thus crucial to the working-out of the process of disinflation.
Aware of this fact, policy-makers could be expected to wish to influence the split as much as possible in a favourable direction. However, one possible line of attack - namely incomes policy - was in many important cases judged either to be impracticable or to carry with it unacceptable side effects. Hence the major hope for exerting influence on the price/output division was seen to lie in any possible "announcement effects" associated with the promulgation of rather tight monetary (and, implicitly, nominal income) targets. In the language of economists, the effect on "expectations" could be potentially very important in minimising the real costs of disinflation. Indeed, for some economists of the "rational expectations" school of monetarism there need be hardly any cost at all. The announcement of a credible policy approach would itself help to make the necessary price adjustment instantaneous.
We now know, of course, that this account of the disinflationary process was over-sanguine - at least on this occasion. Output growth in the industrial world over the period 1980-82 was the slowest for any three-year period since the Second World War and unemployment has reached record levels. On the other hand, we also know that there has been - albeit after some lag - a most spectacular reduction in inflation, at least taking the industrial world as a whole. The weighted average increase in OECD consumer prices over the year to September 1983, at around 5%, implies that world inflation is now the lowest it has been for ten years, and well below the peaks of 13-14% seen in 1974 and again in 1980.
Nevertheless, the years since 1979 would seem to demonstrate that disinflation - especially at a time of external price shock - is not a simple one-stage affair, but rather the more drawn-out process suggested in the title of this paper. The world economy has passed from inflation into recession and falling inflation and now, apparently, to recovery. It is, of course, this last stage that has recently been of concern, the questions being, firstly, whether sustained non-inflationary recovery would come fairly automatically or whether, if not, macro-economic policy would be able to do anything to help. On this occasion, this concern has been particularly acute for two reasons. First of all, unemployment in the industrial world is not only very high, but in some cases it is still rising. Secondly, and more novel, there is the international debt crisis for which growing markets in the industrial countries would be a significant alleviation. Without such alleviation the adjustment pressures on already very poor populations could become even more severe.
In these circumstances the relevance of the nominal income type of analysis is twofold. There is first the question of whether, after a period of disinflation, the price/output split is likely to improve automatically, thus facilitating some recovery of real output even with a continuing, and appropriately moderate, rate of nominal demand growth. Secondly, there is the rather different question of whether - especially at the moment - nominal income might be both a more appropriate indicator of policy stance and a better target for policy-makers than the more usual macro-economic magnitudes. Put more directly, some observers feel that, at a time when income velocity in some major countries has clearly become very unstable for a variety of reasons, any or all measures of the money supply may be seriously biased indicators. In particular, a fall in velocity - an increase in the demand to hold money - could, under rigid money supply targeting, lead to an unnecessarily tight stance of monetary policy. A similar argument would apply - mutatis mutandis - in the case of an acceleration in income velocity. Perhaps more controversially, the argument could also apply to fiscal indicators to the extent that these have also become inadequate measures of the demand and price effects of fiscal policy.
The targeting of nominal GNP would thus represent a "middle way" between the former tendency to concentrate on relatively short-term "real" objectives - which proved inflationary - and exclusive concern with a price level targeted via intermediate variables which may from time to time give misleading signals. That said, however, it is considerably nearer in spirit to the policy approach adopted by the world's policy-makers in 1979 and 1980, in the sense that the "cap" to economic developments in nominal terms would be kept in place. Assuming that an adequate nominal growth rate can be assured, attention can then be shifted to other possible means by which non-inflationary recovery could be encouraged - that is, the "split" improved.
Indeed, one important rationale for the pursuit of a rigorous disinflationary policy in the first place was no doubt that such a policy might itself automatically improve the underlying situation in some way or other. The route could be through some "changing expectations" mechanism. Or, it could be based on new perceptions of the possible consequences - both for labour and for firms - of taking the risk of "pricing oneself out of the market". At all events, the possibility of changes in behaviour of this kind should obviously not be ignored. To the extent that they occur, they would of course imply that empirical analyses - such as that presented in this paper - based purely on history, would produce overly pessimistic results. The recovery might, in other words, be stronger on the side of real growth and weaker on that of prices than history alone would suggest. In fact, without the hope of such an outcome some countries might well not have been able to summon up the degree of policy resolve which they did. For others, however, this was one of the main points of the whole exercise, namely to eliminate inflation permanently as at least one of the indispensable means to securing sustainable growth in the longer term.
2. Résumé and plan of the paper
Despite its intimate connection with these current and important policy themes, this paper sets itself a relatively modest and limited task. Its basic concern is with a fairly simple empirical investigation of the behaviour of nominal income in the industrial countries over the past thirty years, and, in particular, during the two oil crises of the 1970s. This with a view to shedding some light, however faint, on the likely course of the process of worldwide disinflation. The analysis is conducted at a highly aggregated level, namely, the group of industrial countries comprising the OECD. It is thus important to bear in mind that individual country experience of disinflation may well differ from the average, particularly as the policy has been applied in various forms and with varying degrees of firmness and perceived resolve.
The starting point of the empirical investigation is the simplest possible model of the inflation process, a model in which it is assumed that - apart from the effects of lags - the only determinant of inflation is the excess of nominal income growth over the underlying trend growth of real output. The latter is determined by other - non-monetary - factors. This model, though simple in the extreme and probably espoused by no school of economic thought in such a pure form, is nevertheless a convenient abstraction with which to begin the analysis. It embodies - albeit in extreme form - one of the basic ideas which now informs the background thinking of many people - policy-makers and others alike.
Perhaps not surprisingly - all other factors being omitted - empirical estimates of this type of model do indeed suggest some rather strong conclusions for the industrial countries taken as a whole. In the long run, all excess nominal growth apparently feeds through into prices. Macro-economic policy is thus unable to aim for some high target level of output and employment except at the expense of ever-accelerating inflation. Equilibrium, or normal, output is determined by other factors. In addition, however, - a point not often reported - when simulation experiments are carried out with them, models of this type also suggest a very pessimistic conclusion to any process of disinflation. Although the output growth rate may well recover to its trend after inflation is eliminated. the fall in actual output levels below earlier trends is large. Implicitly, in other words, the analysis apparently suggests a very high figure for the so-called NAIRU, the non-accelerating inflation rate of unemployment.
However, when this simple model is tested against the data of the 1970s alone - the period of large external price shocks - its performance is very poor indeed. And, in turn, when the rigorous character of the model is accordingly relaxed to permit exogenous influences on prices (that is, influences independent of movements in nominal income) the picture is somewhat different. In particular, if import price changes are allowed for (though their complete exogeneity must be in doubt), or if at least the two oil shocks are assumed to have involved some independent element, then the conclusions of the analysis are modified appreciably. Similarly, the "split" is also found to vary significantly with the degree of slack in the world economy, implying that from low levels of capacity utilisation output recovery may be faster and less inflationary than otherwise - so long as policy is able to generate the requisite nominal demand. Even so, taken literally, the results suggest only limited room for manoeuvre.
One question, however, is how literally such results should be taken. A one-equation model of the industrial world's economy seems a rather weak reed on which to base serious policy analysis. And indeed, as the results themselves also demonstrate, the implications of the econometric estimates vary importantly with what the researcher is prepared to assume a priori about the nature of the inflation process. If he begins with the "classical dichotomy" - that is the independence of the real and monetary sectors of the economy - then he is likely to assume that, apart from the effect of lags, only money, or nominal income growth, can affect the rate of inflation. If, on the other hand, the nominal demand split is assumed to vary from time to time with other factors, then, apparently, a somewhat different answer is obtained.
Nevertheless, despite a fairly high degree of "first sight plausibility", acceptance of the purer nominal income explanation of inflation carries some rather strange implications when applied to the 1970s. This fact is illustrated in the final empirical section of the paper, which computes the hypothetical development of trend, or equilibrium, output which would be required in order to account for the actual inflation performance of the post-oil shock period without invoking the intervention of exogenous factors. The results suggest an underlying output path in which it is very difficult to believe wholeheartedly, involving, as it does, a highly erratic profile.
Around the empirical sections just summarised, the body of the paper is organised as follows. First a fairly straightforward version of a nominal income theory of inflation is sketched out and its implications noted. Next the basic data for the OECD countries combined are examined in simple tabular and graphical form, noting especially that for virtually the whole of the period under review, nominal income growth has clearly been above any likely long-term trend of real growth, not only on average but on a year-by-year basis, too. Then the various regression analyses already mentioned are reported. The extreme case is contrasted with more eclectic formulations in which import prices, oil shocks and the utilisation of capacity are allowed to play a role. And finally, the analysis is reversed to calculate the kind of underlying output path which has to be assumed for the post-1973 period in order to make a pure nominal income model able to account for the observed variation in prices. The results of this and the earlier analyses are then reviewed and some of their implications drawn out in a final, concluding section.