Inflation

A Theoretical Survey and Synthesis

John Hudson

• Publisher: Routledge
• Year: 2016
• ISBN:
  9781317225331
• Edition: 1
• Topic: Economics
• Subtopic: Economic Theory
• Index: Economics

Originally published in 1982, this book begins with a wide-ranging and critical review of both first and second generation theories of inflation (and the related problem of unemployment), including the classical approach to macroeconomics. The author systematically integrates search, implicit contract, expectations and wage-bargaining theeoriees to outline a new and original synthesis. This synthesis and switching regimes model is then rigorously examined to see how well it can explain inflation the US and the UK.

Chapter 1

Theories of Inflation

Within economics there have grown up two main approaches to inflation. First, the mainstream view regards it as, for want of a better phrase, an excess demand phenomenon. The alternative is to see it in terms of a bargaining situation between bilateral monopolists – employers on the one side and trade unions on the other. In this case, although the forces of supply and demand may influence the outcome, they may not be the most important factors with, for example, the degree of union militancy taking a principal role. We shall begin by examining the development of the mainstream approach and then turn to bargaining theories.

Excess Demand Theories of Inflation

The possible existence of a relationship between inflation and excess demand has long been realised; for example, Joseph Lowe (1822) assigned a causal role to both supply and demand factors when discussing the inflation associated with the Napoleonic wars. In particular he listed the extra demand for men for government service as causing an increase in wages and salaries. In the twentieth century attention has focused on the specific form of this relationship which links unemployment with inflation. One of the earliest works which explicitly postulated such a relationship was Fisher’s 1926 paper. Fisher took the rate of change of prices as the independent variable; in other words, the causation runs from price changes to unemployment. In his model an increase in prices preceded an increase in such contractual costs as wages, thereby stimulating employment, for a short time at least.

In more recent times interest in such a relationship stems, not from Fisher, but from Phillips’s classic paper (1958), where, to quote Friedman (1975), he ‘rediscovered’ this relationship. This rediscovery consisted of fitting a curve through observations in the unemployment-wage inflation plane for the period 1861 to 1913; when he came to compare more recent observations with this curve he found a ‘stunning correspondence’. He also found that actual observations tended to loop around this curve in, usually, an anticlockwise direction. Thus if unemployment was falling, the rate of wage inflation would be higher than that indicated by the curve and vice versa if unemployment was rising. Phillips’s rationale for these loops was not made entirely clear, although Lipsey (1960) thought that he might have had some expectational mechanism in mind, whereby employers might vary the strength of their bidding not merely in response to present needs, but because of expected future needs. However, as Lipsey also noted, there are certain difficulties with this and other possible explanations, and finding a rationale for these loops became a favoured pastime for economists for several years.

Finally, Phillips postulated a restricted role for price increases, which operates with a threshold effect. It is only when the cost of living rises more rapidly than money wages that these become operative. He argued that when this was not the case, that is, when money wages are rising more rapidly than the cost of living, then employers will merely be giving under the name of cost of living adjustments part of the increases they would in any case have given as a result of their competitive bidding for labour. This hypothesis has not received much attention, nor been further developed, and Trevithick and Mulvey (1975), for example, find it ‘not particularly convincing’. However, it seems to me a hypothesis worth further consideration, particularly in how it links up with more recent developments.

One important implication of Phillips’s work is that it indicates the existence of a trade-off between unemployment and inflation. For example, the curve showed that at a rate of unemployment of about 2.5 per cent wages would rise at about 2 per cent p.a., which is consistent with price stability if productivity is also rising at 2 per cent. A lower level of unemployment could then only be achieved at the cost of inflation. The existence of this trade-off generated a great deal of literature, particularly on the optimal combination of unemployment and inflation.

Although it is clear that Phillips had in mind the hypothesis that wage inflation is a function of excess demand in the labour market, there was little in the way of theoretical justification for this. This had to wait until Lipsey’s paper, which, basically, for the case of a single micro-labour market, postulated a wage reaction function dependent upon the ratio of excess demand for labour to total supply. He then linked the rate of unemployment with this ratio, and upon combining the two relationships obtained an additional one between wage inflation and unemployment. Lipsey’s explanation of the loops was simply that in the upswing some labour markets might lag behind others, pushing the Phillips curve to the right, while in the downswing the lag disappears and hence the macrocurve coincides with the micro-curves. Ingenious as it is this explanation suffers from a number of flaws. First, relatively little in the way of justification is given for the operation of the lags in this manner. Secondly, it assumes identical micro-reaction curves, an assumption which has been called into question by Bowers et al. (1970) and Sargan (1971).

An alternative explanation for the loops, which can be made compatible with the rest of Lipsey’s theory, was put forward by Hines (1971). He proposed that job vacancies and unemployment are not related in any simple linear way, but that when excess demand is rising vacancies will rise more rapidly than unemployment falls, and vice versa when excess demand is falling. Thus unemployment will understate the true level of excess demand when it is rising and overstate it when it is falling. Hines argued that a valid proxy for excess demand is provided by the level of unemployment together with its rate of change. This then would seem to offer an explanation for the loops which suffers from none of the drawbacks of either Phillips’s or Lipsey’s.

In the USA the earliest contribution to this literature came from Samuelson and Solow (1960). They found the American Phillips curve lay slightly to the right of that of the UK, with a rate of unemployment of 5 per cent stabilising the wage rate in the UK, but leading to 2.5 per cent inflation in the USA. They also drew attention to a Phillips curve with prices and not wages on the vertical axis. This was derived from the wage-defined Phillips curve, by assuming that wage increases are fully passed on and a rate of productivity growth of 2.5 per cent p.a., and as such was compatible with Phillips’s original approach. It did however set something of a precedent for future work done on the US economy, where many economists have estimated directly the relationship between price inflation and unemployment, and contrasts sharply with the UK, where wages have been the subject of most attention. This has some implications for empirical work, for if it is assumed that excess demand affects prices only through wage changes, then an exact and immediate relationship between price changes and unemployment will only emerge if wage changes are immediately passed on to prices in full. If they are not, or if there is a lag between the two, perhaps varying with economic conditions, then a price-defined Phillips curve becomes more difficult to identify.

However, this is only a relatively minor point, and it is clear that the notion of a relationship between unemployment and inflation was quickly absorbed by the economics profession on both sides of the Atlantic. As an empirical concept it seemed beyond dispute. In addition, Lipsey’s analysis seemed to have provided the basis for a satisfactory theoretical explanation of the relationship. Thus, by the second half of the 1960s it was probably the opinion of the majority of the profession that the only remaining questions were essentially peripheral ones, such as those surrounding the trade-off. Unfortunately this feeling of satisfaction was rudely shattered by actual events. In the UK the Phillips curve appeared to shift substantially and unpredictably to the right in 1966–7 and again in 1969–70, this latter shift being replicated in the USA as well as many other developed countries. The years 1969–70, in particular, mark a watershed in our thinking about inflation. In both countries deflationary policies were being followed in an attempt to reduce inflation. Yet, as we shall see later, they met with little success. Unemployment increased, but with relatively little effect on inflation. In the face of these failures many economists attempted to reconstruct the Phillips curve in a way which could account for these events. This reconstruction took place on two planes, the first revolved around attempts to improve upon unemployment as a measure of excess demand. The second was more fundamental and involved a reconstruction of the theoretical framework proposed by Lipsey, which has led to what has become known as the expectations augmented Phillips curve.

In the UK, those economists who argued that unemployment was a less than adequate measure of excess demand noted that there was also an apparent shift in the relationship between unemployment and vacancies (Bowers et al., 1970). Possible reasons for this shift, which implied a shift between unemployment and excess demand and hence in the Phillips curve, include the introduction of earnings related benefits in October 1966, which had the effect of almost doubling unemployment benefit payable to a man who had previously been receiving average earnings, the introduction of statutory redundancy payments in December 1965 and various labour shake-out hypotheses. Because of this it has been argued that vacancies give a more accurate measure of excess demand than unemployment. Thus Trevithick and Mulvey (1975) report that the vacancy rate performs considerably more satisfactorily than the unemployment rate as an explanatory variable in the wage equation for the years 1966–9; however, in 1970–1 this too breaks down.

In a similar vein Simler and Tella (1968) use a labour reserves’ variable for the USA, which corrected for variations in labour force participation rates. Taylor, in a series of papers (1970, 1972 and Godfrey and Taylor, 1973), has used a measure of unemployment which includes estimates of hoarded labour. The results of such exercises are somewhat contradictory. Perry (1970) and Taylor (1970) found that including such variables in the equations for the USA did not improve the results, whereas Simler and Tella found that they did. For the UK, Taylor (1972) and Godfrey and Taylor found hoarded labour to be a significant factor, though this only seems to apply to the rate of change of earnings corrected for overtime and not to the wage rate change equation.

In recent years, interest in the USA has centred on the effects of demographic and legislative factors on unemployment. Amongst the former are the growth of the youth proportion of the labour force due to the maturing of the post-Second World War baby boom, the increase in female participation rates and the growth of multi-worker families, while on the legislative side, increases in the coverage and value of unemployment insurance have, as in the UK, been the subject of considerable analysis. All these factors will be discussed in more detail when analysing the US economy. Meanwhile we may note that although these factors may well have led to some increase in unemployment, it is unlikely that they can by themselves provide an explanation for the changes in the Phillips curve, especially the sluggish response of inflation to unemployment in recent years. This is especially the case as many of the above changes occurred gradually over time, for example the growth in female participation rates, whereas the change in the Phillips curve was relatively sudden. Thus although these factors might account for a gradual shift in the Phillips curve over time, they cannot explain what actually happened. In addition there has almost certainly been some selection bias in this whole area. Economists are aware that a level of unemployment which twenty years ago would have signalled a recession can no longer be relied upon to reduce inflation. Therefore the search has been on for factors which have tended to increase the level of unemployment. There has not been the same incentive to search for other factors which might have had the reverse effect. Some of the results of this search have made valid contributions to the literature, but others have not. The case of labour hoarding seems particularly weak. This occurs when employers keep labour in excess of their immediate requirements. They do not therefore represent a source of excess supply of labour, because their services have been bid for, and in excess demand terms it matters little as to the possible motives of the bidder, although in another context it could be argued that, as these workers are not actively engaged in productive work, the employer might not be so keen to bid for more workers as he might otherwise be. However, in this case hoarded labour would be acting as a proxy for employers’ keenness to bid for labour, and it does not seem valid to place it in the equation on the same grounds as registered unemployment, which is there to represent excess demand. If hoarded labour is to be entered in the equation as a proxy for employers’ keenness, then it should properly be entered as affecting the speed with which the market responds to excess demand, not as a component of excess demand itself.

The Expectations Augmented Phillips Curve

The second line of approach, aimed at rehabilitating the Phillips curve, involved a much more fundamental reappraisal of the theoretical framework proposed by Lipsey. Once again within this general reappraisal there appear to have been two fairly distinct approaches, both of which had their beginnings prior to the 1970 period which saw the shift in the Phillips curve. The first was developed by Friedman (1968), and involved a more rigorous application of the commodity market approach which lies at the root of present monetarist theories of inflation. The second saw light of day in a remarkable group of essays by, among others, Phelps (1968), Mortenson (1970) and Holt (1970). These provided the major impetus to search theories which have had a significant impact on other areas of economics besides inflation. This approach is, in some respects, more revolutionary than Friedman’s, as it abandons the perfectly competitive labour market assumption made by both him and Lipsey. In addition, it is, as developed by Mortenson, for example, a dynamic theory, in the sense that decision makers are not restricted to considerations of the present alone. However, a more detailed analysis of this approach will have to wait until we have examined the contributions made by Friedman and others who have been influenced by him.

Friedman argues that the relevant wage rate in Phillips’s and Lipsey’s theories should be the real one and not the money one. In addition, as both potential employers and employees envisage the employment contract covering a fairly long period, it will be the anticipated real wage, not the current one, which is relevant. The Phillips curve should therefore be written as

or

where

and U_t denote the rate of wage inflation and the level of unemployment in period t, respectively, and

denotes the expected rate of price inflation over some future period related to average contract duration. This equation is known as the price expectations augmented Phillips curve. Friedman also argued that for short periods price inflation expectations might lag behind actual inflation, but that, given a constant inflation rate, expectations would eventually catch up with inflation. If the assumption is then made that in the long run the rate of price inflation equals the rate of wage inflation plus a constant, k, which may be negative, and represents the rate of productivity growth plus any other long-term effects on the inflation rate, then we can rewrite (1.2) as

It then follows that in the long run, if the coefficient on expectations equals unity (which is implicit in the way the equations have been presented), any long-run trade-off between unemployment and inflation disappears. In this case there is only one possible sustainable level of unemployment, which is known as the ‘natural rate of unemployment’, and is given by the following formula:

If the employment rate is below this level then it will lead to a rate of wage inflation higher than the growth in productivity and hence to price inflation, this in turn will generate price inflation expectations which will further increase the rate of wage inflation and so on. This hypothesis has come to be called the accelerationist, or natural rate hypothesis, as a policy of trying to hold unemployment below the natural rate must lead to ever accelerating inflation. Similarly a rate of unemployment above the natural rate would lead to ever accelerating deflation. The natural rate itself will be determined by two factors, the rate of productivity growth, and the nature of the functional relationship between unemployment and inflation. This, in turn, will depend upon such factors as the efficiency with which the labour market allocates workers to job vacancies, the degree of structural unemployment, etc.

Friedman’s hypothesis was originally greeted with considerable scepticism, but considerable effort was directed at testing it, with particular attention to the coefficient on the expectations variable. The chief problem in all this work has been to find a suitable measure of expectations of inflation. An early attempt was made by Solow (1969); he assumed expectations to be formed by an adaptive expectations mechanism and experimented with different parameters, choosing the most satisfactory in terms of statistical significance. He estimated the coefficient on price expectations to be about 0.4, which was significantly less than 1, and therefore concluded that there was a long-run trade-off between unemployment and inflation. Other studies followed by, among others, Lucas and Rapping (1969), Saunders and Nobay (1972), Turnovsky (1972), Parkin, Sumner and Ward (...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Original Copyright Page
• Dedication
• Table of Contents
• Preface
• Chapter 1 Theories of Inflation
• Chapter 2 A Synthesis
• Chapter 3 Expectations of Inflation
• Chapter 4 Inflation in the United Kingdom and the United States
• Bibliography
• Index