Авторы: 147 А Б В Г Д Е З И Й К Л М Н О П Р С Т У Ф Х Ц Ч Ш Щ Э Ю Я

Книги:  180 А Б В Г Д Е З И Й К Л М Н О П Р С Т У Ф Х Ц Ч Ш Щ Э Ю Я


7.9 Hysteresis and the NAIRU

Since the early 1970s the natural rate of unemployment ‘seems to have taken

a wild ride’ in OECD countries. For OECD countries in general, unemployment

in the 1980s and 90s was higher than during the ‘Golden Age’ of the

1950–73 period. The steadily rising unemployment rates appear to have their

origins in the two OPEC oil price shocks in 1973 and 1979 respectively

(Phelps and Zoega, 1998) and in the case of the European OECD countries,

unemployment that averaged 1.7 per cent in the early 1960s rose to 11 per

cent by the mid-1990s. This high average also hides the large dispersion of

unemployment rates across the European countries (Blanchard and Wolfers,

2000). Gordon’s (1997, 1998) estimates show that the US natural rate of

unemployment has also varied during this same period although the long-run

unemployment repercussions of the 1980s recessions appear to have been

much more persistent in Europe than in the USA. Figure 7.10 shows the

standardized unemployment rates for the USA and OECD Europe for the

period 1972–98. While unemployment in OECD Europe was less than US

unemployment until the early 1980s, since then European unemployment has

remained stubbornly high while it has fallen in the USA.

While the problem of inflation was a major policy concern during the

1970s and early 1980s, by the mid-1980s economists were once again turning

their attention to the problem of unemployment, in particular the rise in the

estimated NAIRU (see Bean et al., 1986; Fitoussi and Phelps, 1988; Summers,

1990; Layard et al., 1991, 1994; Bean, 1994; Cross, 1995; Nickell,

1997, 1998; Siebert, 1997; Katz and Krueger, 1999; Blanchard and Wolfers,

2000; Fitoussi et al., 2000; Hall, 2003).

While estimates of the NAIRU are obviously subject to uncertainty given

the broad range of determinants, recent OECD estimates shown in Table 7.1

indicate the much superior performance of the US economy compared to the

euro area and G7 countries.

During the late 1960s, Friedman and Phelps independently put forward

expectations-augmented models of the Phillips curve. In Friedman’s model

the market-clearing rate of unemployment is called the natural rate of unemployment

and is associated with a stable rate of inflation. As we noted in

Chapter 4, many economists (especially those sympathetic to Keynesianism)

prefer to use the ‘NAIRU’ concept (non-accelerating inflation rate of unemployment),

rather than ‘natural rate’ when discussing long-run unemployment.

The NAIRU terminology was first introduced by Modigliani and Papademos

Source: ‘Labour market performance and the OECD jobs strategy’, OECD, June 1999,


Figure 7.10 Standardized unemployment rates for North America (USA

and Canada) and OECD Europe, 1972–98

% 12













1972 74 76 78 80 82 84 86 88 90 92 94 96 98

North America

OECD Europe

Table 7.1 NAIRU estimates for the G7 countries and the euro area

Country/area 1980 1985 1990 1995 1999

Canada 8.9 10.1 9.0 8.8 7.7

France 5.8 6.5 9.3 10.3 9.5

Germany 3.3 4.4 5.3 6.7 6.9

Italy 6.8 7.8 9.1 10.0 10.4

Japan 1.9 2.7 2.2 2.9 4.0

UK 4.4 8.1 8.6 6.9 7.0

USA 6.1 5.6 5.4 5.3 5.2

Euro area 5.5 7.1 8.8 9.2 8.8

Source: ‘Revised OECD Measures of Structural Unemployment’, OECD, December 2000.

(1975) as ‘NIRU’ (non-inflationary rate of unemployment), defined as ‘a rate

such that, as long as unemployment is above it, inflation can be expected to

decline’. The NAIRU acronym was introduced by James Tobin (1980c) and

has since been used to describe estimates of the natural rate of unemployment

(see Cross et al., 1993; Cross, 1995). However, according to King (1999):

the natural rate of unemployment and the NAIRU are quite different concepts. The

former describes a real equilibrium determined by the structural characteristics of

the labour and product markets – the grinding out of Friedman’s Walrasian general

equilibrium system (modified, if necessary, by non-Walrasian features of

labour markets such as imperfect competition, search behaviour and efficiency

wages). It exists independently of the inflation rate. In contrast, the latter, as well

as being affected by these structural characteristics, is also affected by the gradual

adjustment of the economy to past economic shocks that determine the path of

inflation. Because it is defined as the unemployment rate at which there is no

immediate pressure for a change in the inflation rate, it is a reduced form – not a

structural – variable.

Therefore, the NAIRU concept takes into account inertia in the system which

allows a protracted response of the economy to various economic shocks.

Another way to distinguish between these concepts relates to their

microfoundations. Friedman’s natural rate is a market-clearing concept,

whereas the NAIRU is that rate of unemployment which generates consistency

between the target real wage of workers and the feasible real wage

determined by labour productivity and the size of a firm’s mark-up. Since the

NAIRU is determined by the balance of power between workers and firms,

the microfoundations of the NAIRU relate to theories of imperfect competition

in the labour and product markets (see Carlin and Soskice, 1990; Layard

et al., 1991). However, while recognizing these differences between the concepts

of the natural rate and the NAIRU, Ball and Mankiw (2002) argue that

NAIRU is ‘approximately a synonym for the natural rate of unemployment’.

Therefore, in the discussion that follows we will assume that the two concepts

can be used interchangeably.

According to Friedman’s natural rate hypothesis, fluctuations of aggregate

demand cannot exercise any influence over the natural rate of unemployment,

which is determined by real supply-side influences. The conventional natural

rate view allows monetary and other demand shocks to shift aggregate demand,

thereby influencing the actual rate of unemployment in the short run.

But, as inflationary expectations adjust, unemployment returns to its long-run

equilibrium (natural) value. In new classical models, if the change in aggregate

demand is unanticipated, the combined effect of perfectly flexible prices

and rational expectations ensures that unemployment will quickly return to

its natural rate.

This conventional view is illustrated in Figure 7.11, where the natural rate

of unemployment (UN) is given by point A. Any decline in aggregate demand

will increase the actual rate of unemployment temporarily to point B, while

an expansion of aggregate demand will lower actual unemployment and will

Figure 7.11 The ‘natural rate’ view of the relationship between actual

unemployment and equilibrium unemployment


Expansionary demand shock O Contractionary demand shock




Rising rate

of inflation

Falling rate

of inflation


move the economy temporarily to point C. However, in the long run, unemployment

returns to the natural rate of unemployment at point A.

The dramatic rise in unemployment rates, particularly in Europe during the

1980s, suggested that this conventional view of the natural rate of unemployment

(or NAIRU) must be wrong. It seems that the NAIRU must have risen,

and estimates made by econometricians, such as those presented in Table 7.1,

confirm this view. Several explanations have been put forward to explain

these higher levels of unemployment. One view explains it as a result of

specific policy changes that have reduced the flexibility of the labour market;

more powerful trade unions, higher unemployment compensation and longer

duration of benefits, minimum wage laws, excessive regulations, employment

protection, and higher taxation are, or have been, favourite candidates (see

Minford, 1991; Nickell, 1997; Siebert, 1997; Ljungquist and Sargent, 1998;

Fitoussi et al., 2000; Roed and Zhang, 2003). However, while some of these

factors may account for rising unemployment in the 1970s, many economists

do not believe that they offer a complete explanation of the unemployment

experienced in the 1980s and 1990s (union power, for example, has been

significantly reduced in the UK and has never been a major factor in the US


The simultaneous rise in the actual and equilibrium rates of unemployment

has led some new Keynesian economists to explore a second explanation

which allows aggregate demand to influence the natural rate (or NAIRU).

Models which embody the idea that the natural rate depends on the history of

the equilibrium rate are called ‘hysteresis’ theories. It was Phelps (1972) who

first suggested that the natural rate equilibrium will be partly influenced by

the path taken to reach equilibrium. Phelps called this path dependence

‘hysteresis’, a term borrowed from physics, where it is used to describe the

lagging of magnetic induction behind the source of magnetism (see Cross,


In hysteresis models the natural rate of unemployment will increase if the

actual rate of unemployment in the previous period exceeds the former time

period’s natural rate (Hargreaves-Heap, 1980). This can be expressed as


UNt = UNt−1 + a(Ut−1 −UNt−1) + bt (7.14)

In equation (7.14) UNt is the natural rate of unemployment at time t, UNt–1 is

the previous period’s natural rate of unemployment, Ut–1 is the previous

period’s actual rate of unemployment and bt captures other influences on the

natural rate such as unemployment compensation. If we assume bt = 0, then

equation (7.14) can be rearranged as (7.15):

UNt −UNt−1 = a(Ut−1 −UNt−1) (7.15)

From equation (7.15) it can be seen that UNt > UNt–1 if Ut–1 > UNt–1. In other

words, the shifting actual rate of unemployment acts like a magnet, pulling

the natural rate of unemployment in the same direction. Thus while it may be

reasonable to argue that aggregate demand does not affect UN in the short run,

it is likely that prolonged periods of abnormally high or low economic activity

will shift the natural rate of unemployment.

The impact of hysteresis is illustrated in Figure 7.12. The initial equilibrium

unemployment rate is represented by point A. If the economy is subject

to a negative aggregate demand shock, output falls and unemployment rises

to point B. When the economy recovers from recession, the unemployment

rate does not return to point A. Instead, because of hysteresis effects, the new

NAIRU is at point C. If the economy is now subject to a positive aggregate

demand shock, unemployment falls to point D. When the economy returns to

equilibrium the NAIRU has now fallen to point E. A further recession traces

out a path for this economy through points F to G. In other words, the

NAIRU is influenced by the actual rate of unemployment which itself is

determined mainly by aggregate demand.

Figure 7.12 The hysteresis view of a ‘time-varying’ NAIRU


Expansionary demand shock O Contractionary demand shock



Rising rate

of inflation

Falling rate

of inflation






Theories of hysteresis fall into two main categories, namely duration theories

and insider–outsider theories. Duration theories point out that, when Ut >

UNt, the problem of structural unemployment is exacerbated because the unemployed

suffer a depreciation of their human capital (skills) and as a result

become increasingly unemployable. A high rate of unemployment also tends to

generate an increasing number of long-term unemployed who exercise little

influence on wage bargaining, which also raises the NAIRU. Insider–outsider

theories emphasize the power of insiders which prevents the downward adjustment

of wages in the face of high unemployment. As a result, outsiders are

unable to price their way back into jobs following a rise in unemployment (see

Blanchard and Summers, 1986, 1988). If hysteresis effects are important, the

sacrifice ratio associated with disinflation and recessions is much greater than

is suggested by the original natural rate hypothesis, since high unemployment

will tend to persist (for an extended discussion of the issues raised in this

section, the reader is referred to Cross, 1988; Cross et al., 1993, Layard et al.,

1991; Blanchard and Katz, 1997; Gordon, 2003).

Another distinctive approach to explaining movements in the equilibrium

rate of unemployment over time has been developed by Edmund Phelps and

his co-researchers. In a series of books and papers Phelps has sought to

construct an endogenous theory of the natural rate of unemployment where

‘the equilibrium path of unemployment is driven by the natural rate that is a

variable of the system rather than a constant or a forcing function of time …

hence a moving-natural-rate theory holds the solution to the mystery of what

is behind the shifts and long swings of the unemployment rate’ (Phelps, 1994;

see also Fitoussi and Phelps, 1988; Phelps and Zoega, 1998; Phelps, 2000).

While in Friedman’s natural rate model equilibrium unemployment can change

due to supply-side influences, in Phelps’s dynamic intertemporal non-monetary

equilibrium model it is real demand shocks that are ‘the great movers

and shakers of the economy’s equilibrium path’, although real supply (energy)

shocks also play an important role. Phelps (1990, 1994) classifies his

approach to explaining unemployment as both ‘modern’ and ‘structuralist’,

although it does contain both neoclassical (the role of real interest rates

determined in the capital market), Austrian (the effect of the rate of interest

on the supply of output) and new Keynesian elements (asymmetric information

and efficiency wages). Phelps highlights the impact on the path of the

equilibrium unemployment rate of real influences such as technology, preferences,

social values and institutions. As Phelps (1994) recalls, by the 1980s

he had decided that any chance of accounting for the major swings in economic

activity since the war would require:

abandoning the simplification of a natural rate unemployment rate invariant to

non-monetary (not just monetary) macro shocks in favour of models making the

equilibrium rate an endogenous variable determined by a variety of non-monetary

forces … the longer booms and slumps … must be explained largely as

displacements of the equilibrium path of unemployment itself, not as deviations of

unemployment around an impervious equilibrium path.

In looking for the causes of what Fitoussi et al. (2000) call ‘the great slump’,

that is, the upward shift of equilibrium unemployment rates in the 1980s, the

chief suspects identified are five OECD-wide real shocks to business profitability

and worker’s incentives (see Phelps, 1994), namely:

1. reduced expectations of productivity growth, hence increased effective

cost of capital;

2. an increase in the expected real interest rate which also raises the effective

cost of capital;

3. an increase in services from workers private assets (see Phelps, 2000);

4. an increase in social entitlements relative to after-tax real wages resulting

from the 1970s productivity slowdown and expansion of the welfare


5. the two OPEC oil price shocks in 1973 and 1979.

In the Phelps (1994) model the main driving force behind the rise of the

NAIRU is the increase in real interest rates that occurred across the OECD

countries after the mid-1970s and on into the 1980s (Blanchard and Wolfers,

2000). The rise in world real interest rates, to a large extent induced by US

fiscal expansion in the early 1980s, lowered incentives to accumulate capital,

and, for a given real wage, led to a reduction of labour demand. The high real

interest rate which induced an appreciation of the US dollar (real depreciation

of European currencies) during this period also led to an increase in

European price mark-ups (firms do not lower their export prices in proportion

to the depreciation) and, in consequence, this led to a reduction in labour

demand and a rise in the equilibrium rate of unemployment. For example,

Phelps and Zoega (1998, p. 788) find a very strong correlation between the

world real rate of interest and UK unemployment for the period 1975–95.

Note that in contrast to real business cycle models, where changes in the real

interest rate influence the supply of labour through the intertemporal labour

substitution hypothesis, in Phelps’s model changes in the real interest affect

the demand for labour (for a critique see Madsen, 1998).

While the impact of real shocks as an explanation of increases in the broad

evolution of European unemployment is persuasive, Blanchard and Wolfers

(2000) argue that ‘there is insufficient heterogeneity in these shocks to explain

cross-country differences … Adverse shocks can potentially explain the

general increase in unemployment. Differences in institutions can potentially

explain differences in outcomes across countries.’ Therefore, a more convincing

story of the evolution of the NAIRU in Europe must involve the interaction

of observable real shocks combined with a recognition of the institutional

diversity present across European countries (see Nickell, 1997; and Layard

and Nickell, 1998).