The Expectations-augmented Phillips Curve Analysis
The second stage in the development of orthodox monetarism came with a
more precise analysis of the way the effects of changes in the rate of monetary
expansion are divided between real and nominal magnitudes. This
analysis involved the independent contributions made by Friedman (1968a)
and Phelps (1967, 1968) to the Phillips curve literature (see Chapter 3,
section 3.6). The notion of a stable relationship between inflation and unemployment
was challenged by Friedman and Phelps, who both denied the
existence of a permanent (long-run) trade-off between inflation and unemployment
(Phelps’s analysis originated from a non-monetarist perspective;
see Cross, 1995). The problem with the original specification of the Phillips
curve is that the rate of change of money wages is determined quite independently
of the rate of inflation. This in turn implies that workers are
irrational and suffer from complete money illusion, in that they base their
labour supply decisions on the level of money wages quite independently of
what is happening to prices. In what follows we focus on the highly influential
arguments put forward by Friedman (1968a) in his 1967 Presidential
Address to the American Economic Association. Before doing so we should
recognize just how important Friedman’s paper proved to be for the development
of macroeconomics after 1968. While A Monetary History has
undoubtedly been Friedman’s most influential book in the macroeconomics
sphere, his 1967 Presidential Address published as ‘The Role of Monetary
Policy’ has certainly been his most influential article. In 1981 Robert Gordon
described this paper as probably the most influential article written in macroeconomics
in the previous 20 years. James Tobin (1995), one of Friedman’s
most eloquent, effective and long-standing critics, went even further, describing
the 1968 paper as ‘very likely the most influential article ever published
in an economics journal’ (emphasis added). Paul Krugman (1994a) describes
Friedman’s paper as ‘one of the decisive intellectual achievements of postwar
economics’ and both Mark Blaug (1997) and Robert Skideksky (1996b)
view it as ‘easily the most influential paper on macroeconomics published in
the post-war era’. Between 1968 and 1997 Friedman’s paper has approximately
924 citation counts recorded by the Social Sciences Citation Index
and it continues to be one of the most heavily cited papers in economics (see
Snowdon and Vane, 1998). Friedman’s utilization of Wicksell’s concept of
the ‘natural rate’ in the context of unemployment was in rhetorical terms a
‘masterpiece of marketing’ (see Dixon, 1995), just as the application of the
term ‘rational’ to the expectations hypothesis turned out to be in the rise of
new classical economics during the 1970s. The impact of Professor Friedman’s
work forced Keynesians to restate and remake their case for policy
activism even before that case was further undermined by the penetrating
theoretical critiques of Professor Lucas and other leading new classical economists.
4.3.1 The expectations-augmented Phillips curve
The prevailing Keynesian view of the Phillips curve was overturned by new
ideas hatched during the 1960s and events in the 1970s (Mankiw, 1990). A
central component of the new thinking involved Friedman’s critique of the
trade-off interpretation of the Phillips curve. This was first provided by Friedman
(1966) in his debate with Solow (1966) over wage and price guideposts
and had even been outlined much earlier in conversation with Richard Lipsey
in 1960 (Leeson, 1997a). However, the argument was developed more fully
in his famous 1967 Presidential Address. According to Friedman, the original
Phillips curve which related the rate of change of money wages to unemployment
was misspecified. Although money wages are set in negotiations, both
employers and employees are interested in real, not money, wages. Since
wage bargains are negotiated for discrete time periods, what affects the
anticipated real wage is the rate of inflation expected to exist throughout the
period of the contract. Friedman argued that the Phillips curve should be set
in terms of the rate of change of real wages. He therefore augmented the
basic Phillips curve with the anticipated or expected rate of inflation as an
additional variable determining the rate of change of money wages. The
expectations-augmented Phillips curve can be expressed mathematically by
the equation:
W˙ = f (U) + P˙ e (4.2)
Equation (4.2) shows that the rate of money wage increase is equal to a
component determined by the state of excess demand (as proxied by the level
of unemployment) plus the expected rate of inflation.
Introducing the expected rate of inflation as an additional variable to excess
demand which determines the rate of change of money wages implies
that, instead of one unique Phillips curve, there will be a family of Phillips
curves, each associated with a different expected rate of inflation. Two such
curves are illustrated in Figure 4.4. Suppose the economy is initially in
equilibrium at point A along the short-run Phillips curve (SRPC1) with unemployment
at UN, its natural level (see below) and with a zero rate of increase
of money wages. For simplification purposes in this, and subsequent, analysis
we assume a zero growth in productivity so that with a zero rate of money
wage increase the price level would also be constant and the expected rate of
inflation would be zero; that is, W˙ = P˙ = P˙ e = 0 per cent. Now imagine the
authorities reduce unemployment from UN to U1 by expanding aggregate
demand through monetary expansion. Excess demand in goods and labour
markets would result in upward pressure on prices and money wages, with
commodity prices typically adjusting more rapidly than wages. Having recently
experienced a period of price stability (P˙ e = 0), workers would
misinterpret their money wage increases as real wage increases and supply
more labour; that is, they would suffer from temporary money illusion. Real
wages would, however, actually fall and, as firms demanded more labour,
unemployment would fall, with money wages rising at a rate of W˙1, that is,
point B on the short-run Phillips curve (SRPC1). As workers started slowly to
adapt their inflation expectations in the light of the actual rate of inflation
experienced (P˙ = W˙1), they would realize that, although their money wages
had risen, their real wages had fallen, and they would press for increased
money wages, shifting the short-run Phillips curve upwards from SRPC1 to
SRPC2. Money wages would rise at a rate of ˙W1 plus the expected rate of
inflation. Firms would lay off workers as real wages rose and unemployment
would increase until, at point C, real wages were restored to their original level,
with unemployment at its natural level. This means that, once the actual rate of
inflation is completely anticipated (P˙ P˙ e )
1 = in wage bargains (W˙ P˙ , e
1 = that is
to say there is no money illusion), there will be no long-run trade-off between
unemployment and wage inflation. It follows that if there is no excess demand
(that is, the economy is operating at the natural rate of unemployment),
then the rate of increase of money wages will equal the expected rate of
inflation and only in the special case where the expected rate of inflation is
zero will wage inflation be zero, that is, at point A in Figure 4.4. By joining
points such as A and C together, a long-run vertical Phillips curve is obtained
at the natural rate of unemployment (UN). At UN the rate of increase in money
wages is exactly equal to the rate of increase in prices, so that the real wage is
constant. In consequence there will be no disturbance in the labour market.
At the natural rate the labour market is in a state of equilibrium and the actual
and expected rates of inflation are equal; that is, inflation is fully anticipated.
Friedman’s analysis helped reconcile the classical proposition with respect to
the long-run neutrality of money (see Chapter 2, section 2.5), while still
allowing money to have real effects in the short run.
Following Friedman’s attack on the Phillips curve numerous empirical
studies of the expectations-augmented Phillips curve were undertaken using
the type of equation:
W˙ = f (U) + βP˙ e (4.3)
Estimated values for β of unity imply no long-run trade-off. Conversely
estimates of β of less than unity, but greater than zero, imply a long-run
trade-off but one which is less favourable than in the short run. This can be
demonstrated algebraically in the following manner. Assuming a zero growth
in productivity so that W˙ = P˙, equation (4.3) can be written as:
P˙ = f (U) + βP˙ e (4.4)
Rearranging equation (4.4) we obtain:
P˙ − βP˙ e = f (U) (4.5)
Starting from a position of equilibrium where unemployment equals U* (see
Figure 4.5) and the actual and expected rates of inflation are both equal to
zero (that is, P˙ = P˙ e ), equation (4.5) can be factorized and written as:
P˙(1− β) = f (U) (4.6)
Finally, dividing both sides of equation (4.6) by 1 – β, we obtain
˙ ( )
P
f U =
1− β
(4.7)
Now imagine the authorities initially reduce unemployment below U* (see
Figure 4.5) by expanding aggregate demand through monetary expansion.
From equation (4.7) we can see that, as illustrated in Figure 4.5, (i) estimated
values for β of zero imply both a stable short- and long-run trade-off between
inflation and unemployment in line with the original Phillips curve; (ii)
estimates of β of unity imply no long-run trade-off; and (iii) estimates of β of
less than unity, but greater than zero, imply a long-run trade-off but one
which is less favourable than in the short run. Early evidence from a wide
range of studies that sought to test whether the coefficient (β) on the inflation
expectations term is equal to one proved far from clear-cut. In consequence,
during the early 1970s, the subject of the possible existence of a long-run
vertical Phillips curve became a controversial issue in the monetarist–
Keynesian debate. While there was a body of evidence that monetarists could
draw on to justify their belief that β equals unity, so that there would be no
trade-off between unemployment and inflation in the long run, there was
insufficient evidence to convince all the sceptics. However, according to one
prominent American Keynesian economist, ‘by 1972 the “vertical-in-thelong-
run” view of the Phillips curve had won the day’ (Blinder, 1992a). The
reader is referred to Santomero and Seater (1978) for a very readable review
of the vast literature on the Phillips curve up to 1978. By the mid- to late
1970s, the majority of mainstream Keynesians (especially in the USA) had
come to accept that the long-run Phillips curve is vertical. There is, however,
still considerable controversy on the time it takes for the economy to return to
the long-run solution following a disturbance.
Before turning to discuss the policy implications of the expectations-augmented
Phillips curve, it is worth mentioning that in his Nobel Memorial
Lecture Friedman (1977) offered an explanation for the existence of a positively
sloped Phillips curve for a period of several years, which is compatible
with a vertical long-run Phillips curve at the natural rate of unemployment.
Friedman noted that inflation rates tend to become increasingly volatile at
higher rates of inflation. Increased volatility of inflation results in greater
uncertainty, and unemployment may rise as market efficiency is reduced and
the price system becomes less efficient as a coordinating/communication mechanism
(see Hayek, 1948). Increased uncertainty may also cause a fall in investment
and result in an increase in unemployment. Friedman further argued that, as
inflation rates increase and become increasingly volatile, governments tend to
intervene more in the price-setting process by imposing wage and price controls,
which reduces the efficiency of the price system and results in an increase
in unemployment. The positive relationship between inflation and unemployment
then results from an unanticipated increase in the rate and volatility of
inflation. While the period of transition could be quite long, extending over
decades, once the economy had adjusted to high and volatile inflation, in
Friedman’s view, it would return to the natural rate of unemployment.
4.3.2 The policy implications of the expectations-augmented Phillips
curve
The scope for short-run output–employment gains The monetarist belief in
a long-run vertical Phillips curve implies that an increased rate of monetary
expansion can reduce unemployment below the natural rate only because the
resulting inflation is unexpected. As we have discussed, as soon as inflation is
fully anticipated it will be incorporated into wage bargains and unemployment
will return to the natural rate. The assumption underlying orthodox
monetarist analysis is that expected inflation adjusts to actual inflation only
gradually, in line with the so-called ‘adaptive’ or error-learning expectations
hypothesis. Interestingly, it seems that Friedman was profoundly influenced
by ‘Phillips’s adaptive inflationary expectations formula’ (Leeson, 1999). The
adaptive expectations equation implicit in Friedman’s analysis of the Phillips
curve, and used in Studies in the Quantity Theory of Money (1956), appears
to have been developed by Friedman in conjunction with Philip Cagan following
a discussion he had with Phillips which took place on a park bench
somewhere in London in May 1952 (Leeson, 1994b, 1997a). In fact Friedman
was so impressed with Phillips as an economist that he twice (in 1955
and 1960) tried to persuade him to move to the University of Chicago
(Hammond, 1996).
The main idea behind the adaptive expectations hypothesis is that economic
agents adapt their inflation expectations in the light of past inflation
rates and that they learn from their errors. Workers are assumed to adjust their
inflation expectations by a fraction of the last error made: that is, the difference
between the actual rate of inflation and the expected rate of inflation.
This can be expressed by the equation:
P˙ P˙ (P˙ P˙ ) t
e
t
e
t t
− = − e −1 α −1 (4.8)
where α is a constant fraction. By repeated back substitution expected inflation
can be shown to be a geometrically weighted average of past actual inflation
rates with greater importance attached to more recent experience of inflation:
P˙ P˙ ( )P˙ ( ) P˙ t
e
t t
n
= α + α 1− α −1…α 1− α t−n (4.9)
In this ‘backward-looking’ model, expectations of inflation are based solely
on past actual inflation rates. The existence of a gap in time between an
increase in the actual rate of inflation and an increase in the expected rate
permits a temporary reduction in unemployment below the natural rate. Once
inflation is fully anticipated, the economy returns to its natural rate of unemployment
but with a higher equilibrium rate of wage and price inflation equal
to the rate of monetary growth. As we will discuss in Chapter 5, section 5.5.1,
if expectations are formed according to the rational expectations hypothesis
and economic agents have access to the same information as the authorities,
then the expected rate of inflation will rise immediately in response to an
increased rate of monetary expansion. In the case where there was no lag
between an increase in the actual and expected rate of inflation the authorities
would be powerless to influence output and employment even in the short
run.
The accelerationist hypothesis A second important policy implication of
the belief in a vertical long-run Phillips curve concerns the so-called
‘accelerationist’ hypothesis. This hypothesis implies that any attempt to maintain
unemployment permanently below the natural rate would result in
accelerating inflation and require the authorities to increase continuously the
rate of monetary expansion. Reference to Figure 4.4 reveals that, if unemployment
were held permanently at U1 (that is, below the natural rate UN), the
continued existence of excess demand in the labour market would lead to a
higher actual rate of inflation than expected. As the actual rate of inflation
increased, people would revise their inflation expectations upwards (that is,
shifting the short-run Phillips curve upwards), which would in turn lead to a
higher actual rate of inflation and so on, leading to hyperinflation. In other
words, in order to maintain unemployment below the natural rate, real wages
would have to be kept below their equilibrium level. For this to happen actual
prices would have to rise at a faster rate than money wages. In such a
situation employees would revise their expectations of inflation upwards and
press for higher money wage increases, which would in turn lead to a higher
actual rate of inflation. The end result would be accelerating inflation which
would necessitate continuous increases in the rate of monetary expansion to
validate the continuously rising rate of inflation. Conversely, if unemployment
is held permanently above the natural rate, accelerating deflation will
occur. Where unemployment is held permanently above the natural rate, the
continued existence of excess supply in the labour market will lead to a lower
actual rate of inflation than expected. In this situation people will revise their
inflation expectations downwards (that is, the short-run Phillips curve will
shift downwards), which will in turn lead to a lower actual rate of inflation
and so on. It follows from this analysis that the natural rate is the only level of
unemployment at which a constant rate of inflation may be maintained. In
other words, in long-run equilibrium with the economy at the natural rate of
unemployment, the rate of monetary expansion will determine the rate of
inflation (assuming a constant growth of output and velocity) in line with the
quantity theory of money approach to macroeconomic analysis.
Undoubtedly the influence of Friedman’s (1968a) paper was greatly enhanced
because he anticipated the acceleration of inflation that occurred during
the 1970s as a consequence of the repeated use of expansionary monetary
policy geared to an over-optimistic employment target. The failure of inflation
to slow down in both the US and UK economies in 1970–71, despite rising
unemployment and the subsequent simultaneous existence of high unemployment
and high inflation (so-called stagflation) in many countries, following the
first adverse OPEC oil price (supply) shock in 1973–4, destroyed the idea that
there might be a permanent long-run trade-off between inflation and unemployment.
Lucas (1981b) regards the Friedman–Phelps model and the verification
of its predictions as providing ‘as clear cut an experimental distinction as
macroeconomics is ever likely to see’. In the philosophy of science literature
Imre Lakatos (1978) makes the prediction of novel facts the sole criterion by
which theories should be judged, a view shared by Friedman (1953a). While
Blaug (1991b, 1992) has argued that the principal novel fact of the General
Theory was the prediction that the size of the instantaneous multiplier is greater
than one, he also argues that the prediction of novel facts emanating from
Friedman’s 1968 paper were enough to make Mark I monetarism a progressive
research programme during the 1960s and early 1970s. As Backhouse (1995)
notes, ‘the novel facts predicted by Phelps and Friedman were dramatically
corroborated by the events of the early 1970s’.
The output–employment costs of reducing inflation Friedman (1970c) has
suggested that ‘inflation is always and everywhere a monetary phenomenon
in the sense that it can be produced only by a more rapid increase in the
quantity of money than in output’. Given the orthodox monetarist belief that
inflation is essentially a monetary phenomenon propagated by excessive monetary
growth, monetarists argue that inflation can only be reduced by slowing
down the rate of growth of the money supply. Reducing the rate of monetary
expansion results in an increase in the level of unemployment. The policy
dilemma the authorities face is that, the more rapidly they seek to reduce
Figure 4.6 The output–employment costs of reducing inflation
inflation through monetary contraction, the higher will be the costs in terms
of unemployment. Recognition of this fact has led some orthodox monetarists
(such as David Laidler) to advocate a gradual adjustment process whereby
the rate of monetary expansion is slowly brought down to its desired level in
order to minimize the output–employment costs of reducing inflation. The
costs of the alternative policy options of gradualism versus cold turkey are
illustrated in Figure 4.6.
In Figure 4.6 we assume the economy is initially operating at point A, the
intersection of the short-run Phillips curve (SRPC1) and the long-run vertical
Phillips curve (LRPC). The initial starting position is then both a short- and
long-run equilibrium situation where the economy is experiencing a constant
rate of wage and price inflation which is fully anticipated (that is, W˙ P˙ P˙ e
1= = )
and unemployment is at the natural rate (UN). Now suppose that this rate of
inflation is too high for the authorities’ liking and that they wish to reduce the
rate of inflation by lowering the rate of monetary expansion and move to
position D on the long-run vertical Phillips curve. Consider two alternative
policy options open to the authorities to move to their preferred position at
point D. One (cold turkey) option would be to reduce dramatically the rate of
monetary expansion and raise unemployment to UB, so that wage and price
inflation quickly fell to W˙3; that is, an initial movement along SRPC1 from
point A to B. The initial cost of this option would be a relatively large increase
in unemployment, from UN to UB. As the actual rate of inflation fell below the
expected rate, expectations of future rates of inflation would be revised in a
downward direction. The short-run Phillips curve would shift downwards and a
184 Modern macroeconomics
new short- and long-run equilibrium would eventually be achieved at point D,
the intersection of SRPC3 and LRPC where W˙ P˙ P˙ e
3= = with unemployment
at UN. Another (gradual) policy option open to the authorities would be to begin
with a much smaller reduction in the rate of monetary expansion and initially
increase unemployment to, say, UC so that wage and price inflation fell
to W˙2 , that is, an initial movement along SRPC1 from point A to C. Compared
to the cold turkey option, this gradual option would involve a much smaller
initial increase in unemployment, from UN to UC. As the actual rate of inflation
fell below the expected rate (but to a much lesser extent than in the first option),
expectations would be revised downwards. The short-run Phillips curve would
move downwards as the economy adjusted to a new lower rate of inflation. The
short-run Phillips curve (SRPC2) would be associated with an expected rate of
inflation of W˙2 . A further reduction in the rate of monetary expansion would
further reduce the rate of inflation until the inflation target of ˙W3 was achieved.
The transition to point D on the LRPC would, however, take a much longer
time span than under the first policy option. Such a policy entails living with
inflation for quite long periods of time and has led some economists to advocate
supplementary policy measures to accompany the gradual adjustment
process to a lower rate of inflation. Before we consider the potential scope for
such supplementary measures as indexation and prices and incomes policy, we
should stress the importance of the credibility of any anti-inflation strategy (this
issue is discussed more fully in Chapter 5, section 5.5.3). If the public believes
that the authorities are committed to contractionary monetary policies to reduce
inflation, economic agents will adjust their inflation expectations downwards
more quickly, thereby reducing the output–employment costs associated with
the adjustment process.
Some monetarists (for example Friedman, 1974) have suggested that some
form of indexation would be a useful supplementary policy measure to accompany
the gradual adjustment process to a lower rate of inflation. It is
claimed that indexation would reduce not only the cost of unanticipated
inflation incurred through arbitrary redistribution of income and wealth, but
also the output–employment costs that are associated with a reduction in the
rate of monetary expansion. With indexation, money wage increases would
automatically decline as inflation decreased, thereby removing the danger
that employers would be committed, under existing contracts, to excessive
money wage increases when inflation fell. In other words, with indexation
wage increases would be less rapid and unemployment would therefore rise
by a smaller amount. Further some economists (for example Tobin, 1977,
1981; Trevithick and Stevenson, 1977) have suggested that a prices and
incomes policy could have a role to play, as a temporary and supplementary
policy measure to monetary contraction, to assist the transition to a lower rate
of inflation by reducing inflationary expectations. In terms of Figure 4.6, to
the extent that a prices and incomes policy succeeded in reducing inflationary
expectations, the short-run Phillips curves would shift downwards more
quickly. This in turn would enable adjustment to a lower rate of inflation to
be achieved both more quickly and at a lower cost in terms of the extent and
duration of unemployment that accompanies monetary contraction. However,
one of the problems of using prices and incomes policy is that, even if the
policy initially succeeds in reducing inflationary expectations, once the policy
begins to break down or is ended, inflationary expectations may be revised
upwards. As a result the short-run Phillips curve will shift upwards, thereby
offsetting the initial benefit of the policy in terms of lower unemployment
and wage inflation. For example, Henry and Ormerod (1978) concluded that:
Whilst some incomes policies have reduced the rate of wage inflation during the
period in which they operated, this reduction has only been temporary. Wage
increases in the period immediately following the ending of policies were higher
than they would otherwise have been, and these increases match losses incurred
during the operation of the incomes policy.
In summary, within the orthodox monetarist approach the output–employment
costs associated with monetary contraction depend upon three main
factors: first, whether the authorities pursue a rapid or gradual reduction in
the rate of monetary expansion; second, the extent of institutional adaptations
– for example, whether or not wage contracts are indexed; and third, the
speed with which economic agents adjust their inflationary expectations downwards.
The monetarist view that inflation can only be reduced by slowing down
the rate of growth of the money supply had an important bearing on the
course of macroeconomic policy pursued both in the USA (see Brimmer,
1983) and in the UK during the early 1980s. For example, in the UK the
Conservative government elected into office in 1979 sought, as part of its
medium-term financial strategy, to reduce progressively the rate of monetary
growth (with pre-announced target ranges for four years ahead) in order to
achieve its overriding economic policy objective of permanently reducing the
rate of inflation. Furthermore, the orthodox monetarist contention that inflation
cannot be reduced without output–employment costs appears to have
been borne out by the recessions experienced in the US and UK economies in
1981–2 and 1980–81, respectively (see Chapter 5, section 5.5.2). For wellwritten
and highly accessible accounts of the background to, and execution
and effects of what the media dubbed ‘Thatcher’s monetarist experiment’, the
interested reader is referred to Keegan (1984) and Smith (1987).
The role and conduct of monetary policy The belief in a long-run vertical
Phillips curve and that aggregate-demand management policies can only
affect the level of output and employment in the short run has important
implications for the role and conduct of monetary policy. Before discussing
the rationale for Friedman’s policy prescription for a fixed monetary growth
rule, it is important to stress that, even if the long-run Phillips curve is
vertical, arguments justifying discretionary monetary intervention to stabilize
the economy in the short run can be made on the grounds of either the
potential to identify and respond to economic disturbances or the length of
time required for the economy to return to the natural rate following a
disturbance. Friedman’s policy prescription for a fixed rate of monetary growth
(combined with a floating exchange rate), in line with the trend/long-run
growth rate of the economy, is based on a number of arguments. These
arguments include the beliefs that: (i) if the authorities expand the money
supply at a steady rate over time the economy will tend to settle down at the
natural rate of unemployment with a steady rate of inflation, that is, at a point
along the long-run vertical Phillips curve; (ii) the adoption of a monetary rule
would remove the greatest source of instability in the economy; that is, unless
disturbed by erratic monetary growth, advanced capitalist economies are
inherently stable around the natural rate of unemployment; (iii) in the present
state of economic knowledge, discretionary monetary policy could turn out to
be destabilizing and make matters worse rather than better, owing to the long
and variable lags associated with monetary policy; and (iv) because of ignorance
of the natural rate itself (which may change over time), the government
should not aim at a target unemployment rate for fear of the consequences
noted earlier, most notably accelerating inflation.
We finally consider the implication of the belief in a natural rate of unemployment
for employment policy.
The natural rate of unemployment and supply-side policies As we have
discussed earlier, the natural rate of unemployment is associated with equilibrium
in the labour market and hence in the structure of real wage rates.
Friedman (1968a) has defined the natural rate as:
the level that would be ground out by the Walrasian system of general equilibrium
equations provided there is embedded in them the actual structural characteristics
of the labor and commodity markets, including market imperfections, stochastic
variability in demands and supplies, the cost of gathering information about job
vacancies and labor availabilities, the costs of mobility and so on.
What this approach implies is that, if governments wish to reduce the natural
rate of unemployment in order to achieve higher output and employment
levels, they should pursue supply-management policies that are designed to
improve the structure and functioning of the labour market and industry,
rather than demand-management policies. Examples of the wide range of
(often highly controversial) supply-side policies which were pursued over the
1980s both in the UK (see for example Vane, 1992) and elsewhere include
measures designed to increase: (i) the incentive to work, for example through
reductions in marginal income tax rates and reductions in unemployment and
social security benefits; (ii) the flexibility of wages and working practices, for
example by curtailing trade union power; (iii) the occupational and geographical
mobility of labour, for example in the former case through greater
provision of government retraining schemes; and (iv) the efficiency of markets
for goods and services, for example by privatization.
Following the Friedman–Phelps papers the concept of the natural rate of
unemployment has remained controversial (see Tobin, 1972a, 1995; Cross,
1995). It has also been defined in a large variety of ways. As Rogerson (1997)
shows, the natural rate has been equated with ‘long run = frictional = average =
equilibrium = normal = full employment = steady state = lowest sustainable =
efficient = Hodrick–Prescott trend = natural’. Such definitional problems have
led sceptics such as Solow (1998) to describe the ‘doctrine’ of the natural rate
to be ‘as soft as a grape’. When discussing the relationship between unemployment
and inflation many economists prefer to use the ‘NAIRU’ concept
(non-accelerating inflation rate of unemployment), a term first introduced by
Modigliani and Papademos (1975) as ‘NIRU’ (non-inflationary rate of unemployment).
While the majority of economists would probably admit that it is
‘hard to think about macroeconomic policy without the concept of NAIRU’
(Stiglitz, 1997), others remain unconvinced that the natural rate concept is
helpful (J. Galbraith, 1997; Arestis and Sawyer, 1998; Akerlof, 2002).
The second stage in the development of orthodox monetarism came with a
more precise analysis of the way the effects of changes in the rate of monetary
expansion are divided between real and nominal magnitudes. This
analysis involved the independent contributions made by Friedman (1968a)
and Phelps (1967, 1968) to the Phillips curve literature (see Chapter 3,
section 3.6). The notion of a stable relationship between inflation and unemployment
was challenged by Friedman and Phelps, who both denied the
existence of a permanent (long-run) trade-off between inflation and unemployment
(Phelps’s analysis originated from a non-monetarist perspective;
see Cross, 1995). The problem with the original specification of the Phillips
curve is that the rate of change of money wages is determined quite independently
of the rate of inflation. This in turn implies that workers are
irrational and suffer from complete money illusion, in that they base their
labour supply decisions on the level of money wages quite independently of
what is happening to prices. In what follows we focus on the highly influential
arguments put forward by Friedman (1968a) in his 1967 Presidential
Address to the American Economic Association. Before doing so we should
recognize just how important Friedman’s paper proved to be for the development
of macroeconomics after 1968. While A Monetary History has
undoubtedly been Friedman’s most influential book in the macroeconomics
sphere, his 1967 Presidential Address published as ‘The Role of Monetary
Policy’ has certainly been his most influential article. In 1981 Robert Gordon
described this paper as probably the most influential article written in macroeconomics
in the previous 20 years. James Tobin (1995), one of Friedman’s
most eloquent, effective and long-standing critics, went even further, describing
the 1968 paper as ‘very likely the most influential article ever published
in an economics journal’ (emphasis added). Paul Krugman (1994a) describes
Friedman’s paper as ‘one of the decisive intellectual achievements of postwar
economics’ and both Mark Blaug (1997) and Robert Skideksky (1996b)
view it as ‘easily the most influential paper on macroeconomics published in
the post-war era’. Between 1968 and 1997 Friedman’s paper has approximately
924 citation counts recorded by the Social Sciences Citation Index
and it continues to be one of the most heavily cited papers in economics (see
Snowdon and Vane, 1998). Friedman’s utilization of Wicksell’s concept of
the ‘natural rate’ in the context of unemployment was in rhetorical terms a
‘masterpiece of marketing’ (see Dixon, 1995), just as the application of the
term ‘rational’ to the expectations hypothesis turned out to be in the rise of
new classical economics during the 1970s. The impact of Professor Friedman’s
work forced Keynesians to restate and remake their case for policy
activism even before that case was further undermined by the penetrating
theoretical critiques of Professor Lucas and other leading new classical economists.
4.3.1 The expectations-augmented Phillips curve
The prevailing Keynesian view of the Phillips curve was overturned by new
ideas hatched during the 1960s and events in the 1970s (Mankiw, 1990). A
central component of the new thinking involved Friedman’s critique of the
trade-off interpretation of the Phillips curve. This was first provided by Friedman
(1966) in his debate with Solow (1966) over wage and price guideposts
and had even been outlined much earlier in conversation with Richard Lipsey
in 1960 (Leeson, 1997a). However, the argument was developed more fully
in his famous 1967 Presidential Address. According to Friedman, the original
Phillips curve which related the rate of change of money wages to unemployment
was misspecified. Although money wages are set in negotiations, both
employers and employees are interested in real, not money, wages. Since
wage bargains are negotiated for discrete time periods, what affects the
anticipated real wage is the rate of inflation expected to exist throughout the
period of the contract. Friedman argued that the Phillips curve should be set
in terms of the rate of change of real wages. He therefore augmented the
basic Phillips curve with the anticipated or expected rate of inflation as an
additional variable determining the rate of change of money wages. The
expectations-augmented Phillips curve can be expressed mathematically by
the equation:
W˙ = f (U) + P˙ e (4.2)
Equation (4.2) shows that the rate of money wage increase is equal to a
component determined by the state of excess demand (as proxied by the level
of unemployment) plus the expected rate of inflation.
Introducing the expected rate of inflation as an additional variable to excess
demand which determines the rate of change of money wages implies
that, instead of one unique Phillips curve, there will be a family of Phillips
curves, each associated with a different expected rate of inflation. Two such
curves are illustrated in Figure 4.4. Suppose the economy is initially in
equilibrium at point A along the short-run Phillips curve (SRPC1) with unemployment
at UN, its natural level (see below) and with a zero rate of increase
of money wages. For simplification purposes in this, and subsequent, analysis
we assume a zero growth in productivity so that with a zero rate of money
wage increase the price level would also be constant and the expected rate of
inflation would be zero; that is, W˙ = P˙ = P˙ e = 0 per cent. Now imagine the
authorities reduce unemployment from UN to U1 by expanding aggregate
demand through monetary expansion. Excess demand in goods and labour
markets would result in upward pressure on prices and money wages, with
commodity prices typically adjusting more rapidly than wages. Having recently
experienced a period of price stability (P˙ e = 0), workers would
misinterpret their money wage increases as real wage increases and supply
more labour; that is, they would suffer from temporary money illusion. Real
wages would, however, actually fall and, as firms demanded more labour,
unemployment would fall, with money wages rising at a rate of W˙1, that is,
point B on the short-run Phillips curve (SRPC1). As workers started slowly to
adapt their inflation expectations in the light of the actual rate of inflation
experienced (P˙ = W˙1), they would realize that, although their money wages
had risen, their real wages had fallen, and they would press for increased
money wages, shifting the short-run Phillips curve upwards from SRPC1 to
SRPC2. Money wages would rise at a rate of ˙W1 plus the expected rate of
inflation. Firms would lay off workers as real wages rose and unemployment
would increase until, at point C, real wages were restored to their original level,
with unemployment at its natural level. This means that, once the actual rate of
inflation is completely anticipated (P˙ P˙ e )
1 = in wage bargains (W˙ P˙ , e
1 = that is
to say there is no money illusion), there will be no long-run trade-off between
unemployment and wage inflation. It follows that if there is no excess demand
(that is, the economy is operating at the natural rate of unemployment),
then the rate of increase of money wages will equal the expected rate of
inflation and only in the special case where the expected rate of inflation is
zero will wage inflation be zero, that is, at point A in Figure 4.4. By joining
points such as A and C together, a long-run vertical Phillips curve is obtained
at the natural rate of unemployment (UN). At UN the rate of increase in money
wages is exactly equal to the rate of increase in prices, so that the real wage is
constant. In consequence there will be no disturbance in the labour market.
At the natural rate the labour market is in a state of equilibrium and the actual
and expected rates of inflation are equal; that is, inflation is fully anticipated.
Friedman’s analysis helped reconcile the classical proposition with respect to
the long-run neutrality of money (see Chapter 2, section 2.5), while still
allowing money to have real effects in the short run.
Following Friedman’s attack on the Phillips curve numerous empirical
studies of the expectations-augmented Phillips curve were undertaken using
the type of equation:
W˙ = f (U) + βP˙ e (4.3)
Estimated values for β of unity imply no long-run trade-off. Conversely
estimates of β of less than unity, but greater than zero, imply a long-run
trade-off but one which is less favourable than in the short run. This can be
demonstrated algebraically in the following manner. Assuming a zero growth
in productivity so that W˙ = P˙, equation (4.3) can be written as:
P˙ = f (U) + βP˙ e (4.4)
Rearranging equation (4.4) we obtain:
P˙ − βP˙ e = f (U) (4.5)
Starting from a position of equilibrium where unemployment equals U* (see
Figure 4.5) and the actual and expected rates of inflation are both equal to
zero (that is, P˙ = P˙ e ), equation (4.5) can be factorized and written as:
P˙(1− β) = f (U) (4.6)
Finally, dividing both sides of equation (4.6) by 1 – β, we obtain
˙ ( )
P
f U =
1− β
(4.7)
Now imagine the authorities initially reduce unemployment below U* (see
Figure 4.5) by expanding aggregate demand through monetary expansion.
From equation (4.7) we can see that, as illustrated in Figure 4.5, (i) estimated
values for β of zero imply both a stable short- and long-run trade-off between
inflation and unemployment in line with the original Phillips curve; (ii)
estimates of β of unity imply no long-run trade-off; and (iii) estimates of β of
less than unity, but greater than zero, imply a long-run trade-off but one
which is less favourable than in the short run. Early evidence from a wide
range of studies that sought to test whether the coefficient (β) on the inflation
expectations term is equal to one proved far from clear-cut. In consequence,
during the early 1970s, the subject of the possible existence of a long-run
vertical Phillips curve became a controversial issue in the monetarist–
Keynesian debate. While there was a body of evidence that monetarists could
draw on to justify their belief that β equals unity, so that there would be no
trade-off between unemployment and inflation in the long run, there was
insufficient evidence to convince all the sceptics. However, according to one
prominent American Keynesian economist, ‘by 1972 the “vertical-in-thelong-
run” view of the Phillips curve had won the day’ (Blinder, 1992a). The
reader is referred to Santomero and Seater (1978) for a very readable review
of the vast literature on the Phillips curve up to 1978. By the mid- to late
1970s, the majority of mainstream Keynesians (especially in the USA) had
come to accept that the long-run Phillips curve is vertical. There is, however,
still considerable controversy on the time it takes for the economy to return to
the long-run solution following a disturbance.
Before turning to discuss the policy implications of the expectations-augmented
Phillips curve, it is worth mentioning that in his Nobel Memorial
Lecture Friedman (1977) offered an explanation for the existence of a positively
sloped Phillips curve for a period of several years, which is compatible
with a vertical long-run Phillips curve at the natural rate of unemployment.
Friedman noted that inflation rates tend to become increasingly volatile at
higher rates of inflation. Increased volatility of inflation results in greater
uncertainty, and unemployment may rise as market efficiency is reduced and
the price system becomes less efficient as a coordinating/communication mechanism
(see Hayek, 1948). Increased uncertainty may also cause a fall in investment
and result in an increase in unemployment. Friedman further argued that, as
inflation rates increase and become increasingly volatile, governments tend to
intervene more in the price-setting process by imposing wage and price controls,
which reduces the efficiency of the price system and results in an increase
in unemployment. The positive relationship between inflation and unemployment
then results from an unanticipated increase in the rate and volatility of
inflation. While the period of transition could be quite long, extending over
decades, once the economy had adjusted to high and volatile inflation, in
Friedman’s view, it would return to the natural rate of unemployment.
4.3.2 The policy implications of the expectations-augmented Phillips
curve
The scope for short-run output–employment gains The monetarist belief in
a long-run vertical Phillips curve implies that an increased rate of monetary
expansion can reduce unemployment below the natural rate only because the
resulting inflation is unexpected. As we have discussed, as soon as inflation is
fully anticipated it will be incorporated into wage bargains and unemployment
will return to the natural rate. The assumption underlying orthodox
monetarist analysis is that expected inflation adjusts to actual inflation only
gradually, in line with the so-called ‘adaptive’ or error-learning expectations
hypothesis. Interestingly, it seems that Friedman was profoundly influenced
by ‘Phillips’s adaptive inflationary expectations formula’ (Leeson, 1999). The
adaptive expectations equation implicit in Friedman’s analysis of the Phillips
curve, and used in Studies in the Quantity Theory of Money (1956), appears
to have been developed by Friedman in conjunction with Philip Cagan following
a discussion he had with Phillips which took place on a park bench
somewhere in London in May 1952 (Leeson, 1994b, 1997a). In fact Friedman
was so impressed with Phillips as an economist that he twice (in 1955
and 1960) tried to persuade him to move to the University of Chicago
(Hammond, 1996).
The main idea behind the adaptive expectations hypothesis is that economic
agents adapt their inflation expectations in the light of past inflation
rates and that they learn from their errors. Workers are assumed to adjust their
inflation expectations by a fraction of the last error made: that is, the difference
between the actual rate of inflation and the expected rate of inflation.
This can be expressed by the equation:
P˙ P˙ (P˙ P˙ ) t
e
t
e
t t
− = − e −1 α −1 (4.8)
where α is a constant fraction. By repeated back substitution expected inflation
can be shown to be a geometrically weighted average of past actual inflation
rates with greater importance attached to more recent experience of inflation:
P˙ P˙ ( )P˙ ( ) P˙ t
e
t t
n
= α + α 1− α −1…α 1− α t−n (4.9)
In this ‘backward-looking’ model, expectations of inflation are based solely
on past actual inflation rates. The existence of a gap in time between an
increase in the actual rate of inflation and an increase in the expected rate
permits a temporary reduction in unemployment below the natural rate. Once
inflation is fully anticipated, the economy returns to its natural rate of unemployment
but with a higher equilibrium rate of wage and price inflation equal
to the rate of monetary growth. As we will discuss in Chapter 5, section 5.5.1,
if expectations are formed according to the rational expectations hypothesis
and economic agents have access to the same information as the authorities,
then the expected rate of inflation will rise immediately in response to an
increased rate of monetary expansion. In the case where there was no lag
between an increase in the actual and expected rate of inflation the authorities
would be powerless to influence output and employment even in the short
run.
The accelerationist hypothesis A second important policy implication of
the belief in a vertical long-run Phillips curve concerns the so-called
‘accelerationist’ hypothesis. This hypothesis implies that any attempt to maintain
unemployment permanently below the natural rate would result in
accelerating inflation and require the authorities to increase continuously the
rate of monetary expansion. Reference to Figure 4.4 reveals that, if unemployment
were held permanently at U1 (that is, below the natural rate UN), the
continued existence of excess demand in the labour market would lead to a
higher actual rate of inflation than expected. As the actual rate of inflation
increased, people would revise their inflation expectations upwards (that is,
shifting the short-run Phillips curve upwards), which would in turn lead to a
higher actual rate of inflation and so on, leading to hyperinflation. In other
words, in order to maintain unemployment below the natural rate, real wages
would have to be kept below their equilibrium level. For this to happen actual
prices would have to rise at a faster rate than money wages. In such a
situation employees would revise their expectations of inflation upwards and
press for higher money wage increases, which would in turn lead to a higher
actual rate of inflation. The end result would be accelerating inflation which
would necessitate continuous increases in the rate of monetary expansion to
validate the continuously rising rate of inflation. Conversely, if unemployment
is held permanently above the natural rate, accelerating deflation will
occur. Where unemployment is held permanently above the natural rate, the
continued existence of excess supply in the labour market will lead to a lower
actual rate of inflation than expected. In this situation people will revise their
inflation expectations downwards (that is, the short-run Phillips curve will
shift downwards), which will in turn lead to a lower actual rate of inflation
and so on. It follows from this analysis that the natural rate is the only level of
unemployment at which a constant rate of inflation may be maintained. In
other words, in long-run equilibrium with the economy at the natural rate of
unemployment, the rate of monetary expansion will determine the rate of
inflation (assuming a constant growth of output and velocity) in line with the
quantity theory of money approach to macroeconomic analysis.
Undoubtedly the influence of Friedman’s (1968a) paper was greatly enhanced
because he anticipated the acceleration of inflation that occurred during
the 1970s as a consequence of the repeated use of expansionary monetary
policy geared to an over-optimistic employment target. The failure of inflation
to slow down in both the US and UK economies in 1970–71, despite rising
unemployment and the subsequent simultaneous existence of high unemployment
and high inflation (so-called stagflation) in many countries, following the
first adverse OPEC oil price (supply) shock in 1973–4, destroyed the idea that
there might be a permanent long-run trade-off between inflation and unemployment.
Lucas (1981b) regards the Friedman–Phelps model and the verification
of its predictions as providing ‘as clear cut an experimental distinction as
macroeconomics is ever likely to see’. In the philosophy of science literature
Imre Lakatos (1978) makes the prediction of novel facts the sole criterion by
which theories should be judged, a view shared by Friedman (1953a). While
Blaug (1991b, 1992) has argued that the principal novel fact of the General
Theory was the prediction that the size of the instantaneous multiplier is greater
than one, he also argues that the prediction of novel facts emanating from
Friedman’s 1968 paper were enough to make Mark I monetarism a progressive
research programme during the 1960s and early 1970s. As Backhouse (1995)
notes, ‘the novel facts predicted by Phelps and Friedman were dramatically
corroborated by the events of the early 1970s’.
The output–employment costs of reducing inflation Friedman (1970c) has
suggested that ‘inflation is always and everywhere a monetary phenomenon
in the sense that it can be produced only by a more rapid increase in the
quantity of money than in output’. Given the orthodox monetarist belief that
inflation is essentially a monetary phenomenon propagated by excessive monetary
growth, monetarists argue that inflation can only be reduced by slowing
down the rate of growth of the money supply. Reducing the rate of monetary
expansion results in an increase in the level of unemployment. The policy
dilemma the authorities face is that, the more rapidly they seek to reduce
Figure 4.6 The output–employment costs of reducing inflation
inflation through monetary contraction, the higher will be the costs in terms
of unemployment. Recognition of this fact has led some orthodox monetarists
(such as David Laidler) to advocate a gradual adjustment process whereby
the rate of monetary expansion is slowly brought down to its desired level in
order to minimize the output–employment costs of reducing inflation. The
costs of the alternative policy options of gradualism versus cold turkey are
illustrated in Figure 4.6.
In Figure 4.6 we assume the economy is initially operating at point A, the
intersection of the short-run Phillips curve (SRPC1) and the long-run vertical
Phillips curve (LRPC). The initial starting position is then both a short- and
long-run equilibrium situation where the economy is experiencing a constant
rate of wage and price inflation which is fully anticipated (that is, W˙ P˙ P˙ e
1= = )
and unemployment is at the natural rate (UN). Now suppose that this rate of
inflation is too high for the authorities’ liking and that they wish to reduce the
rate of inflation by lowering the rate of monetary expansion and move to
position D on the long-run vertical Phillips curve. Consider two alternative
policy options open to the authorities to move to their preferred position at
point D. One (cold turkey) option would be to reduce dramatically the rate of
monetary expansion and raise unemployment to UB, so that wage and price
inflation quickly fell to W˙3; that is, an initial movement along SRPC1 from
point A to B. The initial cost of this option would be a relatively large increase
in unemployment, from UN to UB. As the actual rate of inflation fell below the
expected rate, expectations of future rates of inflation would be revised in a
downward direction. The short-run Phillips curve would shift downwards and a
184 Modern macroeconomics
new short- and long-run equilibrium would eventually be achieved at point D,
the intersection of SRPC3 and LRPC where W˙ P˙ P˙ e
3= = with unemployment
at UN. Another (gradual) policy option open to the authorities would be to begin
with a much smaller reduction in the rate of monetary expansion and initially
increase unemployment to, say, UC so that wage and price inflation fell
to W˙2 , that is, an initial movement along SRPC1 from point A to C. Compared
to the cold turkey option, this gradual option would involve a much smaller
initial increase in unemployment, from UN to UC. As the actual rate of inflation
fell below the expected rate (but to a much lesser extent than in the first option),
expectations would be revised downwards. The short-run Phillips curve would
move downwards as the economy adjusted to a new lower rate of inflation. The
short-run Phillips curve (SRPC2) would be associated with an expected rate of
inflation of W˙2 . A further reduction in the rate of monetary expansion would
further reduce the rate of inflation until the inflation target of ˙W3 was achieved.
The transition to point D on the LRPC would, however, take a much longer
time span than under the first policy option. Such a policy entails living with
inflation for quite long periods of time and has led some economists to advocate
supplementary policy measures to accompany the gradual adjustment
process to a lower rate of inflation. Before we consider the potential scope for
such supplementary measures as indexation and prices and incomes policy, we
should stress the importance of the credibility of any anti-inflation strategy (this
issue is discussed more fully in Chapter 5, section 5.5.3). If the public believes
that the authorities are committed to contractionary monetary policies to reduce
inflation, economic agents will adjust their inflation expectations downwards
more quickly, thereby reducing the output–employment costs associated with
the adjustment process.
Some monetarists (for example Friedman, 1974) have suggested that some
form of indexation would be a useful supplementary policy measure to accompany
the gradual adjustment process to a lower rate of inflation. It is
claimed that indexation would reduce not only the cost of unanticipated
inflation incurred through arbitrary redistribution of income and wealth, but
also the output–employment costs that are associated with a reduction in the
rate of monetary expansion. With indexation, money wage increases would
automatically decline as inflation decreased, thereby removing the danger
that employers would be committed, under existing contracts, to excessive
money wage increases when inflation fell. In other words, with indexation
wage increases would be less rapid and unemployment would therefore rise
by a smaller amount. Further some economists (for example Tobin, 1977,
1981; Trevithick and Stevenson, 1977) have suggested that a prices and
incomes policy could have a role to play, as a temporary and supplementary
policy measure to monetary contraction, to assist the transition to a lower rate
of inflation by reducing inflationary expectations. In terms of Figure 4.6, to
the extent that a prices and incomes policy succeeded in reducing inflationary
expectations, the short-run Phillips curves would shift downwards more
quickly. This in turn would enable adjustment to a lower rate of inflation to
be achieved both more quickly and at a lower cost in terms of the extent and
duration of unemployment that accompanies monetary contraction. However,
one of the problems of using prices and incomes policy is that, even if the
policy initially succeeds in reducing inflationary expectations, once the policy
begins to break down or is ended, inflationary expectations may be revised
upwards. As a result the short-run Phillips curve will shift upwards, thereby
offsetting the initial benefit of the policy in terms of lower unemployment
and wage inflation. For example, Henry and Ormerod (1978) concluded that:
Whilst some incomes policies have reduced the rate of wage inflation during the
period in which they operated, this reduction has only been temporary. Wage
increases in the period immediately following the ending of policies were higher
than they would otherwise have been, and these increases match losses incurred
during the operation of the incomes policy.
In summary, within the orthodox monetarist approach the output–employment
costs associated with monetary contraction depend upon three main
factors: first, whether the authorities pursue a rapid or gradual reduction in
the rate of monetary expansion; second, the extent of institutional adaptations
– for example, whether or not wage contracts are indexed; and third, the
speed with which economic agents adjust their inflationary expectations downwards.
The monetarist view that inflation can only be reduced by slowing down
the rate of growth of the money supply had an important bearing on the
course of macroeconomic policy pursued both in the USA (see Brimmer,
1983) and in the UK during the early 1980s. For example, in the UK the
Conservative government elected into office in 1979 sought, as part of its
medium-term financial strategy, to reduce progressively the rate of monetary
growth (with pre-announced target ranges for four years ahead) in order to
achieve its overriding economic policy objective of permanently reducing the
rate of inflation. Furthermore, the orthodox monetarist contention that inflation
cannot be reduced without output–employment costs appears to have
been borne out by the recessions experienced in the US and UK economies in
1981–2 and 1980–81, respectively (see Chapter 5, section 5.5.2). For wellwritten
and highly accessible accounts of the background to, and execution
and effects of what the media dubbed ‘Thatcher’s monetarist experiment’, the
interested reader is referred to Keegan (1984) and Smith (1987).
The role and conduct of monetary policy The belief in a long-run vertical
Phillips curve and that aggregate-demand management policies can only
affect the level of output and employment in the short run has important
implications for the role and conduct of monetary policy. Before discussing
the rationale for Friedman’s policy prescription for a fixed monetary growth
rule, it is important to stress that, even if the long-run Phillips curve is
vertical, arguments justifying discretionary monetary intervention to stabilize
the economy in the short run can be made on the grounds of either the
potential to identify and respond to economic disturbances or the length of
time required for the economy to return to the natural rate following a
disturbance. Friedman’s policy prescription for a fixed rate of monetary growth
(combined with a floating exchange rate), in line with the trend/long-run
growth rate of the economy, is based on a number of arguments. These
arguments include the beliefs that: (i) if the authorities expand the money
supply at a steady rate over time the economy will tend to settle down at the
natural rate of unemployment with a steady rate of inflation, that is, at a point
along the long-run vertical Phillips curve; (ii) the adoption of a monetary rule
would remove the greatest source of instability in the economy; that is, unless
disturbed by erratic monetary growth, advanced capitalist economies are
inherently stable around the natural rate of unemployment; (iii) in the present
state of economic knowledge, discretionary monetary policy could turn out to
be destabilizing and make matters worse rather than better, owing to the long
and variable lags associated with monetary policy; and (iv) because of ignorance
of the natural rate itself (which may change over time), the government
should not aim at a target unemployment rate for fear of the consequences
noted earlier, most notably accelerating inflation.
We finally consider the implication of the belief in a natural rate of unemployment
for employment policy.
The natural rate of unemployment and supply-side policies As we have
discussed earlier, the natural rate of unemployment is associated with equilibrium
in the labour market and hence in the structure of real wage rates.
Friedman (1968a) has defined the natural rate as:
the level that would be ground out by the Walrasian system of general equilibrium
equations provided there is embedded in them the actual structural characteristics
of the labor and commodity markets, including market imperfections, stochastic
variability in demands and supplies, the cost of gathering information about job
vacancies and labor availabilities, the costs of mobility and so on.
What this approach implies is that, if governments wish to reduce the natural
rate of unemployment in order to achieve higher output and employment
levels, they should pursue supply-management policies that are designed to
improve the structure and functioning of the labour market and industry,
rather than demand-management policies. Examples of the wide range of
(often highly controversial) supply-side policies which were pursued over the
1980s both in the UK (see for example Vane, 1992) and elsewhere include
measures designed to increase: (i) the incentive to work, for example through
reductions in marginal income tax rates and reductions in unemployment and
social security benefits; (ii) the flexibility of wages and working practices, for
example by curtailing trade union power; (iii) the occupational and geographical
mobility of labour, for example in the former case through greater
provision of government retraining schemes; and (iv) the efficiency of markets
for goods and services, for example by privatization.
Following the Friedman–Phelps papers the concept of the natural rate of
unemployment has remained controversial (see Tobin, 1972a, 1995; Cross,
1995). It has also been defined in a large variety of ways. As Rogerson (1997)
shows, the natural rate has been equated with ‘long run = frictional = average =
equilibrium = normal = full employment = steady state = lowest sustainable =
efficient = Hodrick–Prescott trend = natural’. Such definitional problems have
led sceptics such as Solow (1998) to describe the ‘doctrine’ of the natural rate
to be ‘as soft as a grape’. When discussing the relationship between unemployment
and inflation many economists prefer to use the ‘NAIRU’ concept
(non-accelerating inflation rate of unemployment), a term first introduced by
Modigliani and Papademos (1975) as ‘NIRU’ (non-inflationary rate of unemployment).
While the majority of economists would probably admit that it is
‘hard to think about macroeconomic policy without the concept of NAIRU’
(Stiglitz, 1997), others remain unconvinced that the natural rate concept is
helpful (J. Galbraith, 1997; Arestis and Sawyer, 1998; Akerlof, 2002).
No comments:
Post a Comment