stzs,
UB&j&ms
^Tt<^
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/hysteresiseuropeOOblan
working paper
department
of economics
"Hysteresis and the European Unemployment Probl em
by
Olivier
Blanchard
and
Lawrence H. Summers
Number 427
J.,
April 1986
massachusetts
institute of
technology
50 memorial drive
Cambridge, mass. 02139
"Hysteresis and the European Unemployment Problem
by
Blanchard
and
Lawrence H. Summers
Olivier
Number 427
J.,
April 1986
.mm
w
R€CEIV£D
-
^W "
..
-
.,.\ j;-j
April
19B6
Hysteresis and the European unemployment problem
Olivier
*
HIT and NBER,
J.
Blanchard and Lawrence
H.
Summers
*
and Harvard and NBER respectively.
We have benefited from the hospitality of the Center tor Labour Economics at
the LSE. Richard Layard has been especially generous in helping us in a variety of
ways. We thank David Grubb, John Martin and Andrew. Newel 1 for providing us with data,
and Michael Burda, Robert Waldman, Changyong Rhee and Fernando Ramos for research
assistance. We thank Stanley Fischer, David Metcalf, Steven Nickell, James
Poterba and Andre Shleifer and participants in the NBER macro conference at which
this paper was presented for useful discussions and comments.
Abstract
European unemployment has been steadily increasing for the last
expected to remain very high
this -fact
many years to come.
•for
In
this paper,
15
years and
is
we argue that
implies that shocks have much more persistent effects on unemployment than
standard theories can possibly explain.
persistence,
We
develop
a
theory which can explain such
and which is based on the distinction between insiders and outsiders
wage bargaining.
We
insiders and firms,
argue that
H
in
wages are largely set by bargaining between
shocks which affect actual
unemployment tend also to affect
equilibrium unemployment. We then confront the theory to both the detailed -facts of
the European
situation as well as to earlier periods
such as the Great Depression
in
the US.
of
high persistent unemployment,
After 20 years
negligible unemployment, most
of
since the early 70's
protracted period
a
unemployment peaked
United Kingdom
has risen almost
high and rising unemp oynent
1
3.3 percent
at
Western Europe has suffered
1945-1970
over the
continuously since 1970, and now stands at over
Common Market nations as
1970 and
of
of
19B0 and has
a
whole,
12
In
.
the
period, but
percent.
For
the
the unemployment rate more than doubled between
again doubled since then.
decline in unemployment over the next several
Few forecasts call
years,
and
none call
for
for
a
significant
its return
to
levels close to those that prevailed in the 1950's and 1960's.
These events are not easily accounted for by conventional
macroeconomi
c
theories.
classical
or
Keynesian
Rigidities associated with fixed length contracts, or the
costs of adjusting prices or quantities are unlikely to be large enough to account
for rising unemployment over
periods of
decade or more.
a
And
substitution in labor supply is surely not an important aspect
downturn.
The sustained upturn
in
intertemporal
of
such
a
protracted
European unemployment challenges the premise
most macroeconomic theories that there exists some
"natural"
or
of
"non-accelerating
inflation" rate of unemployment towards which the economy tends to gravitate and at
which the level
consideration
of
of
inflation remains constant.
alternative theories
of
"hysteresis" which contemplate the
possibility that increases in unemployment have
of
The European experience compels
a
direct impact on the "natural" rate
unemployment.
This paper explores theoretically and empirically the idea of macroeconomi
hysteresi s--the substantial
shocks on unemployment.
We
Our
persistence
of
c
unemployment and the protracted effects
of
particular motivation is the current European situation.
seek explanations for the pattern of high and rising unemployment that has
prevailed
in
Europe for the past decade and for the very different performance
of
the
labor market
in
the United States and Europe,
and reach some tentative conclusions
about the extent to which European unemployment problems can be solved by
expansionary demand policies.
The central
hypothesis we put forward
is
that
hysteresis resulting from membership considerations plays an important role in
explaining the current European depression in particular and persistent high
general.
The essential
unemployment
in
assymetry
the wage setting process between
in
who are want
A
insiders who are employed and outsiders
view to
a
Shocks which lead to reduced employment change the
oi
is
Section
the equilibrium unemployment rate
to •follow the actual
organized as follows!
1
documents the dimensions oi the current European depression.
United Kingdom over the past century,
It
Section
of
2
It
then considers
unemployment could
the United States and
high unemployment is in tact often quite
be
a
number
mechanisms through which high
of
generated.
explores what we find the most promising
for generating hysteresis.
shocks can have
that
in
a
It
presents
a
formal model
permanent effect on the level
are set by employers who bargain with insiders.
of
of
the possible mechanisms
illustrating how temporary
employment in contexts where wages
Persistence results in this setting
because shocks change employment and membership in the group of insiders,
influencing its subsequent bargaining strategy.
and whether
It
reviews standard explanations of the current European situation and
finds them lacking.
persistence
unemployment rate.
evidence consistent with our hypothesis are adduced.
documents, by looking at the movements in unemployment
persistent.
giving
hembership considerations can therefore explain the general
number of types of empirical
The paper
fundamental
insiders and thereby change the subsequent equilibrium wage rate,
rise to hysteresis.
tendency
a
Outsiders tre disenfranchised and wages are set with
jobs.
insuring the jobs of insiders,
number of
point is that there is
We
thus
then discuss the role of
such effects can arise in non union settings.
unions
Section
hysteresis.
3
It
examines the behavior
d
f
pott
war
presents direct evidence on the role
wages and employment and on the composition of
experience quite consistent with our model.
much more so than the US.
are well
Europe
High unemployment
unions,
unemployment.
on
We
theory
of
the behavior
of
our
of
find the European
Europe appears to have high hysteresis,
in
explained both by different sequences
and by different
of
light
in
Europe end low unemployment
of
especially
shocks,
in
in
the
the US
19B0's,
propagation mechanisms, with Europe exhibiting more persistence than
the US.
Section
4
returns to an issue which is of fundamental
Granting that Europe has more hysteresis than the US,
is
is
importance for policy.
it
really due to unions or
hysteresis itself endogenous, being triggered by bad times
?
In
an
attempt to
answer this question, the section compares Europe now to Europe earlier when
unemployment was low, and compares the current European depression to the US Great
depression.
This last comparison is especially important,
to drastically decrease
demand
unemployment in 1939 and 1940,
given the ability of the US
mostly through aggregate
.
The conclusion summarizes our beliefs and doubts,
our analysis for
policy.
and draws the
implications
of
h
1
.
We
e
T
Reco r d
of
Persistent Unemployment
start this section by documenting the dimensions of the current European
depression.
We
then demonstrate that persistently high unemployment like that
experienced
in
Europe at present
century suggest
a
is
not historically unusual.
surprisingly high degree
United States and United Kingdom.
We
of
Data for the past
persistence in unemployment in both the
argue that such persistence is not easily
explained by standard natural rate theories and conclude that theories which allow
for
hysteresis, by which we mean
past unemployment',
1
are required
to explain
such persistence.
The European Depression
.1
Table
1
presents some information on the evolution
major European countries as well
unemployment rates
States,
very high dependence of current unemployment on
a
the
in
1
unemployment rates
unemployment rates.
960
in
'
s
as
of
unemployment in three
the US over the past 25 years.
While European
were substantially lower than those in the United
Europe today are substantially greater than current US
The unemployment rate in the United States has fluctuated
considerably, rising from 4.B to B.3 percent in the 1973-1975 recession then
declining to 5.B percent
to around 7.0 percent
inexorably since 1973.
today.
In
then rising to 9.7 percent in 19E2 before declining
1979,
in
In
contrast, unemployment in Europe has risen seemingly
France,
the unemployment rate has increased
in
every
single year since 1973, while it has declined only twice in Germany and the United
Kingdom.
The differences between the European countries and the United States are
most pronounced after
19B0.
While the US unemployment rate is at roughly its 1980
A
8
A
7
Table
1
European and U.S. Unemployment
1961-1986
United Kingdom
United States
France
West Germany
1961-1970
A.
1.9
.9
.8
1971-1975
6.1
2.8
2.6
1.8
1976-1980
6.7
5.2
5.3
3.7
1980
7.1
6.0
6.
3.
1981
7.6
9.2
7.7
A.
1982
9.7
10.6
8.7
6.9
1983
9.6
11.6
8.8
8.
1984
7.5
11.8
9.9
8.4
1985
7.3
12.0
10.7
8.
1986*
7.2
11.7
10.9
8.0
Source
.
Annual Economic Review
Forecast.
,
Commission of the European Communities, 1986,
level,
the unemployment
The rapid decline
countries.
continuing increase
forecasts
rate has approximately doubled in the three European
in
in
US unemployment
unemployment
in
Europe.
after
The
lest
Longer run forecasts are very similar
expected change.
European Commission put unemployment for the EEC as
10.
BV.
in
a
the table
line of
unemployment by the European Commission for 1966
of
compared to
19B2 contraiti iharply with the
i
gives
they show little
i
baseline projections by the
whole at
10.4'/.
in
1990,
19B5.
Differences in unemployment rates actually understate the differences in the
performance
American and European labor markets over the past decade.
of
suffered the concomitants
of
Between 1975 and
Europe has
high unempl oyment--r educed labor force participation and
involuntarily reductions in hours-- to
States.
-
19B3,
a
much greater extent than has the United
the labor force participation rate of
men in the
United States remained constant, while the corresponding rate in DECD Europe declined
by six
percent.
Average annual
States between 1975 and
percent in England.
performances
of
19B5 employment
hours worked declined by 2.7 percent in the United
19E2 compared with declines of
Perhaps the most striking contrast
7.5 percent
of
in
France and 6.1
the labor market
Europe and the United States is the observation that between 1975 and
increased by 25 percent,
or
about 25 million jobs in the United
States while declining in absolute terms in Europe.
1.2.
Unemployment Rates in the UK and the US over the last 100 veers
European unemployment has steadily increased and, pending an unexpected change
in
policy,
How unusual
is
is
expected to remain at this new higher level for the foreseable future.
such high and persistent unemployment?
now examine the behavior of
the US.
To answer
unemployment over the last 100 years
this question,
in
we
both the UK and
Figures
1
and
plot
2
unemployment
f
or
each
1B90-19B5 for the UK, and 1B92-19B5 for the US.
Estimation
UK
»
u
i
process
AR(1)
an
of
•for
)
+
e
j
cr.
*
2.1
.90 u(-l)
+
e
|
a.
«
2.0
u (-1
.93
of
the two countries,
for
the period
1
the whole
sample for each country gives
i
7.
(.04)
UB
=
u
:
'/.
(.04)
In
the degree of
both cases,
-first
order serial
Unemployment is indeed surprisingly persistent.
to return to
of
is
exhibits at best
high.
a
weak tendency
its mean.
Examination
evolution
It
correlation
the two figures --as well
of
the unemployment rate over
statistical
as
100 years is
the past
captured by any simple linear aut Dregr essi ve representation.
persistence as captured by the degree
of
wor k--suggests
1940,
it
it
In
The degree of
first order serial correlation reported
the UK,
around
when unemployment goes up from 1920 to
shows little tendency during that period to return to its pre 1920 level
then returns to
a
low level
during WWII,
current episode, both past and forecast,
after having sharply increased,
the
however not well
above arises in large part from relatively infrequent changes in the level
which unemployment fluctuates.
that
is
to stay there until
a
stabilizes at
the
1960's.
;
The
second instance in which unemployment,
a
new,
high level.
The US experienced
sustained increase in unemployment from 1929 to 1939, only to see unemployment drop
sharply during and after the war to
a
new,
much lower,
level.
persistence in unemployment is estimated separately for periods
average unemployment,
periods
of
there is some weak evidence
high average unemployment.
of
When the degree of
of
high and low
greater persistence within
a
cz
70
cz
CZ
ro
en
CD
CD
O
en
o
ro
C/l
LO
CD
O
o
P3
CD
ro
o
CD
o
CD
CD
O
CD
CD
O
s
.
The time series studied
the changes
in
They could be exogenous or
unemployment.
with
itself,
level
a
few years
unemployment,
of
level.
in
that
we
do not
which account
level,
the mean
isolation give little indication it to the cause
in
In
a
for
much of
the pertiitence
in
instead be triggered by unemployment
high unemployment triggering an increase in the mean
of
few years of
the absence of
a
low unemployment triggering
in
turn
a
decrease
tight specification of how this triggering occurs
believe that the data can easily distinguish between these two
possibilities
and we shall
not
attempt to do so at this stage.
Our finding that unemployment exhibits
a
very high degree of persistence over
century parallels the findings of Nelson and Plosser C19B23, Campbell
the past
of
Hankiw C19B6] and others that
a
other non-stationary processes.
variety
In
of
economic variables follow random walks or
many cases such findings can be easily
rationalized by recognizing that the level
of
technology is likely to be non-
stationary and that other variables like the level
But the failure of
and
unemployment to display more
of
output depend on productivity.
of
mean reverting tendency is
a
unlikely that non stationarity in productivity can account for the
troubling.
It
is
persistence
of
unemployment since the secular increase in productivity has not been
associated with any trend
3
3.
Pi
agnosi no Unempl
What sort
and
the
or
of
p
upwards drift in unemployment.
yment Pro bl em
theories can account for persistent high unemployment in general
current European experience in particular?
We
highlight the general
difficulties one encounters in explaining persistent unemployment by focusing on the
problem
of
explaining the current European situation.
its persistence.
The central
While it is easy to point to substantial,
puzzle it poses is
adverse supply and demand
4
e
shocks over the last
15
years, we argue that
our
standard theories do not easily
explain how they have had such enduring effects on the level
of
unemployment
3
.
Aooreoate demand
There is little question that Europe has been affected by large adverse demand
shocks,
especially since 19B0
(see
for
example Dornbusch
a
engaging in
a
major and prolonged fiscal
for the UK,
Germany and France
i
seal
.
19B3).
In
the
19B0's,
contraction
see Buiter
|
19B5 for
(see
Blanchard and Summers
more detailed study of
a
1
9 B
the UK
pol icy).
But to the extent that
natural
al
large extent matched tight US monetary policy while it the same time
Europe has to
f
et
aggregate demand shocks do not affect the equilibrium or
unemployment, one would expect sustained high unemployment to be
rate of
associated with rapid declines in the rate
inflation.
of
More generally,
standard
models of the effects of aggregate demand shocks would not predict that previous
estimates
of
the relationship between
There is substantial
there has been
a
unemployment
is
and
in
2
But
inflation than would have been predicted by
which gives the rates of
unemployment, there is no sign
experiencing
e
estimates
the rate of
of
in
the basic point that previous relations have broken down
19B5 for the United Kingdom,
small
unemployment would break down.
Below we examine the relation between wage inflation and
detail.
evidenced in Table
and
evidence however that this relation has broken down and that
much smaller decline
past relationships.
inflation
inflation and unemployment in 1964
France and Sermany.
of
Despite the high rates
of
disinflation, with the United Kingdom and Germany
increase in inflation and France
a
small
decrease.
Econometric
unemployment consistent with stable inflation show rapid
increases over the past decade.
Layard et
al
(
1
9B4
)
,
using crude time trends in
a
A
Table
2
Inflation and Unemployment in the U.K., France and Germany
1984-1985
United Kingdom
tt
France
U
71
Ge rmanv
U
IT
U
1984
A.
11.8
7.0
9.9
1 .9
8 .A
1985
5.5
12.0
5.7
10.7
2 ,1
8 .A
tt
=
U =
Rate of change of GDP deflator.
Unemployment
Source
.
Annual Economic Review, Commission of the European Communities, 1986.
Phillips curve relation, find the unemployment rate consistent with iteady inflation
to have
risen from 2.4
in
1967-70 to 9.2 in 19B1-19B3 in Britain, from 1.3 to 6.2
Germany and from 2.2 to 6.9 in France.
Coe
and
Gagliardi
framework of the Phillips curve but using instead
determinants
potential
of
explaining the increase
tine trend
a
also within the
a
battery
of
equilibrium unemployment as right hand side variables,
obtain roughly similar results.
in
of
(19B5),
in
Aggregate demand shocks have clearly played
European unemployment
in
|
they cannot
but
be
a
role
the whole
story given the increase in the rate of unemployment consistent with steady
i
nf
1
ati on.
Aooregate supply
Aggregate supply explanations appear more promising
increase in equilibrium unemployment.
the recent
research.
the goal
is
to
explain an
This is indeed the approach followed by much of
Sachs [1979,19B3D and Bruno and Sachs C19B5D have argued that
unemployment in Europe
shocks and real
if
is
in
large part the result of
wage rigidity.
The argument
is
a
combination
that real
of
adverse supply
wages do not adjust to
clear the labor market so that adverse supply shocks which reduce the demand for
labor at
a
given real wage create unemployment.
wage rigidity and the occurence
of
This argument has two parts,
real
adverse supply shocks. We start by reviewing the
evidence on the second.
Table
a
3
presents some information on the behavior
of
potential bearing on unemployment in the UK since I960".
various supply factors with
7
.
Table
3
Supply Factors and U.K. Unemployment
Year
Replacement
Rate (%)
Unemployment
Rate (%)
1960
2.3
42
1965
2.3
48
1970
3.1
1975
Mismatch
Index (%)
Productivity
Growth (%)
Change in
Tax Wedge (2)
1.9
.0
41
2.8
1.0
51
38
3.2
1.0
A.
49
43
2.7
.8
1976
6.0
50
38
1.5
2.8
1977
6.4
51
35
1.7
1.9
1978
6.1
50
35
1.4
-.9
1979
5.6
46
35
2.1
1.3
1980
6.9
45
37
1.5
1.3
1981
10.6
50
41
1.4
2.6
1982
12.8
54
37
1.1
1.0
1983
13.1
54
_
.5
-1.8
Notes
a) Standardized unemployment rate; source OECD.
b) Weighted average of replacement rates relevant to families of different sizes.
Source: Layard and Nickell (1985).
Index constructed as Z
u.-v.
where u. and v. are the proportions of unemplovment
i
i
and vacancies in occupation i respectively.
Source: Layard, Nickell and Jackman
(1984).
d) Rate of change of total factor productivity growth, derived by assuming" labor
augmenting technical change.
The first four numbers refer to the change in the
rate (at annual rate) over the previous five vears. Source: Layard and Nickell
(1985).
e) The tax wedge is the sum of the employment tax rate levied on employers and of
direct and indirect tax rates levied on employees.
The first four numbers refer
Source:
to the change in the rate (at annual rates) over the previous five years.
Layard and Nickell (1985).
c)
I
ii
I
10
first candidate is une mployme nt benefit
A
unemployment
if
Unemployment
s.
causes workers to search longer
it
less
or
reducing the pressure that unemployment puts on wages.
gives the average replacement ratio,
that
for
the average ratio of
is
!
little or no change in the replacement ratio.
since
1970.
shows an increase in real
Indeed,
Attempts to estimate the effect
unemployment have not been very successful
further discussion)
and one
is
shows no
it
wages,
higher unemployment but
of
roughly 30 1
a
in
eligibility rules rather than
the
(19E2)
and
Nickell
(1934)
for
increase in unemployment
large portion of the increase in
.
large scale reallocation of
increase unemployment.
increased the rate
labor
change
.
The argument
is
that the
associated with structural change tends to
Often it is suggested that the energy shocks of the
1970's
structural change and so led to higher unemployment.
The
adjustment to structural changes may be complicated by real wage rigidity.
The
of
fourth column of Table
and
role
another way of reading the
(see Minford
second candidate explanation is structural
Nickell
a
unemployment benefits on
of
led to conclude that
benefits probably does not account for
need for
|
Furthermore, it has been argued that the principle changes in
benefit levels.
A
3
after tax
workers
unemployment benefits
unemployment insurance have occurred through changes
unempl oyment
table
of
one can easily envision mechanisms through which
increases in unemployment benefits lead to higher real
it
of
column
This is not necessarily conclusive evidence against
over time.
unemployment benefits
column is that
intensively for jobs,
The second
unemployment benefits to earnings for different categories
clear movement
insurance may raise
Jackman
C
1
3
934
presents the index
D
.
The index
of
"mismatch" developed by Layard,
tries to represent the degree of structural
change in the economy by examining the extent to which unemployment and vacancies
occur
in
the same sectors.
The results in the table look at occupational
mismatch,
11
but
results are largely similar when industrial
There is little evidence
increase in the rate
an
of
and regional
measures are used =
structural
of
.
change since the
1960's when the unemployment rate was consistently low.
Perhaps the most common supply based explanations for persistent high
unemployment involve factors which reduce labor productivity and or drive
labor to firms and the wage workers receive.
between the cost
of
fifth columns of
the table
the change
the oil
30?.
reduction in the rate
of
total
Over the years the total
by
107.
since
wage consistent with full
it
.
had
in
1970.
of
wedge has also risen substantially,
by
tax
While it is still
of
a
factor productivity growth in the wake
true that the real
employment has risen fairly steadily,
the first half
and
factor productivity growth and
clear from the table that there has been
1960,
slowly than
The fourth
is
shocks.
since
wedge *
wedge
It
the tax
in
substantial
give time series for total
a
it
has
after-tax
increased more
the post war period.
The problem with aggregate supply explanations
now documented the presence of
We have
what might have been expected in the early
long lasting effect
If
and when
wages,
the
not
Dn
long run.
1970's.
But
unemployment.
to have
a
Individual
supply shocks are then reflected in real
labor supply is surely largely inelastic in
with aggregate demand explanations,
As
for these shocks
unemployment, there must be long lasting real wage rigidity.
labor supply becomes inelastic,
in
adverse supply developments relative to
we
face the problem of
explaining the mechanism that causes shocks to have long lived effects.
Recent models of union behavior
this problem by showing that
firms,
if
the result may be real
There is however
a
fundamental
(notably McDonald Solow 19B1) have addressed
wages are the result of bargaining between unions and
wage rigidity, with shocks affecting employment only.
difficulty with this line
of
argument.
To take the
f
12
developed by McDonald and Solow,
model
determined by the interaction
of
if
real
wages were truly rigid at
union preferences and firms'
rate
a
production technology,
employment would steadily increase and unemployment steadily decrease through time.
Annual
productivity improvements due to technical change are equivalent to favorable
supply shocks.
As
long as productivity increments and capital
accuulation led to the
demand curve for labor shifting outwards faster than the population grew,
unemployment would decline.
the cumulative
impact of
This appears count er actual
7
.
Even over the last decade,
productivity growth has almost certainly more than
counterbalanced the adverse supply shocks that occurred.
To rescue
this line of
thought,
it
must be argued that real
along some "norm", which may increase over time.
first is that the dynamic effects
way the norm adjusts to actual
and more
as
of
this has two implications.
The
supply shocks on employment then depend on the
productivity and this
left unexplained.
is
The second
important one here is that adverse supply shocks have an effect only as long
the norm has not adjusted to actual
catches up with actual productivity,
permanently.
can all
But
wages are rigid
be
It
productivity.
Thus,
unless the norm never
adverse supply shocks cannot affect unemployment
seems implausible that the current persistence
attributed to lags in learning about productivity.
of
high unemployment
Both the United
Kingdom and the United States have experienced enormous productivity gains without
evident reduction in unemployment over the last .century.
cannot be blamed simply on poor productivity performance.
to
surprises
in
productivity performance.
But
then
it
protracted unemployment from lower productivity growth.
is
High unemployment therefore
It
hard
can only be attributed
to see
how to explain
13
Where does this leave us
adverse shocks, be
in
it
lower
the tax wedge
of
oil
or
in
the
19B0's.
in
have argued that there is plenty of
than expected productivity growth,
in
the
differently,
Put
that
it
is
increases in the price
a
sustained effect on
difficult to account for the apparent increase
unemployment --or equivalently
the equilibrium rate of
of
standard theories do not provide us with
how these shocks can have such
of
evidence
1970's or contractionary aggregate demand policies
But we have also argued
convincing explanations
unemployment.
We
?
in
the unemployment rate
consistent with stable inflation-- by pointing to these shocks.
Borrowing from the
business cycle terminology,
of
I
it
is
not
difficult to find evidence
the difficulty is in explaining the propagation mechanism.
for mechanisms that can explain the propagation of
over
long periods of
time.
negative impulses
This leads us to look
adverse supply or demand shocks
These include the possibility that current unemployment
depends directly and strongly on past unempl oyment e
.
We
now consider various channels
through which this may happen.
1.4. Theories
Three types
"physical
of
of
capital",
Hysteresis
explanation which loosely speaking might be referred to as the
"human capital", and
"insider-outsider' stories can be adduced to
explain why shocks which cause unemployment in
a
single period might have long term
effects.
The physical
capital
story simply holds that reductions in the capital
stock
associated with the reduced employment that accompanies adverse shocks reduce the
subsequent demand for labor and so cause protracted unemployment.
frequently made in the current European context where
the very substantial
it
is
This argument
is
emphasized that, despite
increase in the unemployment rate that has occurred,
capacity
14
utilization
is
at
levels.
fairly normal
has shown no trend
over
with 76 percent in
1975,
the
B3
the EEC as
For
last decade.
stock
capacity utilization
whole,
currently stands at
It
percent in 1979,
argued that the existing capital
a
and 76 percent
in
percent compared
Bl
19B3.
It
is
then
simply inadequate to employ the current
is
labor <orce,
skeptical
are somewhat
We
of
the argument
account for high unemployment for two reasons.
possibilities for substitution
on
labor
it
is
First,
capital
for
stock affect the demand for labor
capital
above,
of
just
accumulation effects can
that capital
as
long
ex-post,
as
there are some
reductions
in
the
like adverse supply shocks.
As
unlikely that an anticipated supply shock would have an important effect
the unemployment
rate.
Second,
discuss
we
as
in
Section
A
below,
disinvestment during the 1930's did not preclude the rapid recovery
associated with rearmarment
substantial
noted
reduction
in
in
a
other countries.
number of
the civilian capital
the size of
substantial
of
employment
Nor did the very
stock that occurred during
many countries*.
the War prevent the attainment of
full
The argument that reduced capital
accumulation has an important effect on the level
of
employment after the War
unemployment is difficult to support with historical
A
in
examples.
second and perhaps more isportant mechanism works through "human capital"
broadly defined.
Persuasive statements
the potentially important effects of
of
unemployment on human capital accumulation and subsequent labor supply may be found
in
Phelps [19723 and Hargraves-Heap
be found
in
Clark and Summers
[ 1 ,9
B2
[
3
.
1
96
3
J
°
.
some suggestive empirical
Essentially,
the human capital
evidence may
argument holds
that workers who are unemployed lose the opportunity to maintain and update their'
skills by working.
r.ay
find
Particularly for the long term unemployed, the atrophy
of
skills
combine with disaffection from the labor force associated with the inability to
a
job,
to reduce the
effective supply of labor.
Early retirement may for
D
15
example be
middle aged workers to find new jobs.
unemployment environment,
signal
if
reasons employers prefer workers with long horizons,
capital
for
More generally,
semi-irreversible decision.
a
it
will
be
A
final
point
or
human
may be very difficult
it
that
in
high
a
difficult for reliable and able workers to
quality by holding jobs and being promoted.
their
is
incentive
for
inefficiencies in sorting workers may reduce the overall
The resulting
demand for labor.
Beyond the adverse effects on labor supply generated by high unemployment,
benefits
of
high pressure economy are foregone,
a
Clark and Summers
demonstrate that in the United States at least World War
in
raising female labor force participation.
labor force participation of
all
II
had
a
C 1
9
the
B2
long lasting effect
Despite the baby boom,
in
1950 the
female cohorts that were old enough to have worked
during the War was significantly greater than would have been predicted on the basis
of
pre-War trends.
trends.
role of participation during the War is evidenced by
the participation of
the fact that
the War
The causal
was actually lower
very young women who could not have worked during
than would have been predicted on the basis of
Similarly, research by Ellwood
[ 1
leave some "permanent scars" on subsequent
9E
1 3
suggests that teenage unemployment may
labor market perf or
through which this may occur is family composition.
performance
of
noted
One channel
(fiance.
The superior
labor market
married men with children has been noted many times.
Great Depression on fertility rates,
earlier
both in the US and in Europe,
The effect
has
of
the
often been
.
Gauging the quantitative importance of human capital
hysteresis is very difficult.
suggest that
mechanisms generating
Some of the arguments, early retirement for example,
labor force participation should decline rather than that unemployment
should increase in the aftermath of adverse shocks.
Perhaps
a
inore
fundamental
problem is that to the extent that there is some irreversibility associated with
16
unemployment shoe
response to
becomes more difficult to explain why temporary shocks have
early retirement is forever, why should
temporary downturn
that they are sufficient
doubtful
i
a
If
mechanisms can explain some
capital
pers
it
,
large short run effects.
such
in
k s
?
Overall,
of
the protracted response to shocks,
while
it
seems
taken
be
it
likely that human
it
is
completely for the observed degree
to account
of
st
ence
A
third mechanism that can generate persistence and that we regard as the most
.
promising relies on the distinction between "insider" and "outsider" workers,
developed in
for
of
series
a
of
contributions by Lindbeck
and used in an
example)
by bargaining
To
take an extreme case,
their jobs, not insuring the employment of
the absence of
in
importantly:
walk;
suppose that all
wages are set
Insiders are concerned with maintaining
shocks, any level
outsiders.
of
This has two implications.
employment
insiders is self-
of
insiders just set the wage so as to remain employed.
sustaining;
[19B5D
between employed workers--the "insiders "--and firms, with outsiders
playing no role in the bargaining process.
First,
and Snower
important paper by Gregory C19B5] to explain the behavior
Australian economy.
the
(see Lindbeck
in
the presence of
shocks,
employment follows
after an adverse shock for example,
of
process akin to
which reduces employment,
their insider status and the new smaller group of
maintain this new lower level
a
Second and more
employment.
a
random
some workers lose
insiders sets the wage so as to
Employment and unemployment show no
tendency to return to their pre shock value, but are instead determined by the
history
that,
shocks.
of
if
This example is
wage bargaining is
a
extreme but nevertheless suggestive.
prevalent feature
interactions between employment and the size
substantial
in
detail
in
of
of
the
the group of
employment and unemployment persistence.
the next
section.
labor aarket,
It
suggests
the dynamic
insiders may generate
This is the argument we explore
17
2.
Theory
A
of
Unemployment Persistence
This section develops
theory
*
of
unemployment persistence bated on the
distinction between insiders and outsiders.
As
the example sketched at
previous section makes clear, the key assumption
such
of
relation between employment status and insider status.
assumption as an assumption about membershi
p
rules
a
theory is that
We can think
of
of
the
of
this
the
key
the rules which govern the
,
relation between employment status and membership in the group
possibility
the end of
of
insiders.
The
persistent fluctuations in employment arises because changes in
employment may change the group's membership and thereby alter its objective
function 1
J
.
the first
In
bargaining between
part of
a
this section,
group of
we
insiders and
develop
a
a
partial equilibrium model
representative firm and characterize
employment dynamics under alternative membership rules
rather than the more natural
(We
use the term "group"
"union" to avoid prejudging the issue of whether the
membership considerations we stress are important only in settings
unions are present).
of
The second part
of
wh-ere
formal
the section extends the analysis to
a
general
equilibrium setting and shows how both nominal and real shocks can have permanent
effects on unemployment.
issues
of
j
the first
is
In
the remaining part of
that of the endogeneity of
the section,
we
consider nainly two
membership rules. The second is that
whether our analysis is indeed relevant only or mostly in explicit union
setttings.
2.1.
A
Model
of
Membership Rules and Employment Dynamics.
is
focus on the
To
of
d yn *
rr i
effects
c
insiders, the "group" for short,
presenting
aembership rules on the decision
of
formalize the firm is entirely passive,
we
labor demand on which the group chooses
a
start by characterizing employment and wages in
model,
it
is
membership
initial
preferred outcome
its
one period model.
a
In
a
1
*-
is
We
.
period
one
given and membership rules are obviously irrelevant.
is
But
intermediate step, which will allow us to contrast our later results
useful
a
the group
of
ones which treat membership as exogenous.
with traditional
attempt at generality and use convenient functional
retain analytical
Throughout, we make no
forms and some approximations to
simplicity.
The One Period Model
The group has initial
membership
unless otherwise nentioned).
follows,
-
(2. 1)
n
«
where
n
is
cw +
logarithms,
n
(in
It
faces
a
labor demand function given by:
employment,
is
w
the real
wage and
[Ee-a,
e
is
before
it
knows the realisation of
e.
Given
a
Ee+a].
the degree of uncertainty associated with labor demand.
w
w
and
random technological
The coefficient
outsiders are hired,
probability
of
If
n
less than no,
is
being laid off
is
the realisation of
Before specifying the objective function
w
and
n
,
the probability of
insider is equal
to one
if
being employed
n>n
.
no-n insiders are
the same for all
For n<no,
.
of
a
shock,
captures
The group must decide on
firm then chooses labor according to the labor demand function.
n
variables in what
are all
e
with mean Ee, uniformly distributed between
wage
as
If
n
e,
the
exceeds no,
laid off.
a
n-
The
insiders.
the group,
can derive,
we
The probability of
we approximate the
for
given
being employed for an
probability
(which is
19
not
be
in
logarithm)
of
good as long is
probability
(2.2)
n
being employed for an insider by l-n +n.
it
too much smaller thin
not
being employed
p
of
p
*
1
*
1
-
(l/4*)(no
+
is
given by
cw - Ee +
even under the worst outcome
If
the'
of
initial
decreasing function
of
the
probability
for
no
+
cw
>
Ee
-
a
for
n
+
ch
<
Ee
-
a
e»Ee-a and thus n"-cw+Ee-a --
is
-probability is an increasing function
decreasing function
membership no, and
degree of uncertainty
being employed in bad times,
of
mat
i
will
on
Under these assumptions,
then the probability of employment is clearly equal
larger than no,
Otherwise,
— which
i
the
derivations are in the appendix)
(all
a) 2
r,o.
This appr ok
of
a
|
of
is
n
to one.
expected productivity Ee,
the wage w.
the larger
It
a,
i
is
also
a
a
the lower the
while the probability remains equal
to
one in good times.
The second step is to derive the choice of
utility function
of
w
The group maximises the utility function of the
the group.
representative group member, which we specify as
U
*
p
+
is
not the most
The first reason is that,
specification
i
bw
Utility is linear in the probability
specification
This requires specifying the
.
of
as
natural but
will
be
of
employment and the wage.
it
is
however attractive,
seen below,
it
implies,
This
for two reasons,
together with the
probabilities given above, that the group exhibits the stochastic
equivalent of inelastic labor supply
i
an
increase in Ee
increase in real wages and leaves the probability
argued in the previous section that this is
determination given the absence
of
a
of
is
entirely reflected in an
employment unchanged.
desirable feature
of
We
any model
have
of
major trends in unemployment rates over long
wage
)
20
Note however that our
periods of time*?.
labor supply is the opposite of
a
assumption
of
stochastically inelastic
that used by McDonald and Solow.
Where they postulate
wage so that the labor supply curve is perfectly elastic,
rigid real
perfectly inelastic labor supply.
The second reason
is
that
it
li
we
postulate
analytically
convenient.
Replacing
w*
by
p
(1/c) (-no
-
value from
its
Ee
+
a (2
+
(b/c
(2.2)
)
-1
and
solving for the optimal
wage
givesi
w
)
Replacing in labor demand gives
n
no
«=
-
a(2(b/c)-l)
Replacing w*
in
p-
2
*
1-a (b/c)
+
equation
(e-Ee)
(2.2)
and rearranging gives the optimal
membership.
Thus the wage depends negatively on initial
0,
1)
As
by definition
whether expected employment exceeds membership depends on the sign
thus on whether b/c
is
less than
1/2 or
not.
The lower b,
the firm.
workers set
Note,
a
depends neither on the initial
Until
membership,
now,
a
of
membership.
a
(
b / c
)
-
If
c,
b/c
is
the
less
being employed
wage.
Given the initial
This wage and the realisation of
determine employment. But when we go from this one period model to
be
2
membership nor on expected productivity**.
the analysis has been rather conventional!
insiders choose
there may well
(
E(e-Ee)
wage low enough to imply expected net hirings of outsiders by
mentioned above that the optimal probability
as
a
the higher
smaller the wage reduction required to increase expected employment.
1/2,
of
i
the sore importance
workers attach to employment protection as opposed to the wage|
than
probability
a
a
disturbance
dynamic one,
relation between employment this period and next period's
This relation will
depend on the form
how this affects employment dynamics.
of
membership rules.
We
now examine
)
21
first define member shi
We
indexed by
being
first
periods lose member shi
m
is
the case where
The second
changes.
We
.
various membership rules as
can think of
Those workers who have been working in the fir* for the last
m.
periods belong to the group,
than
rules
p
current employment.
is
The
is
m
p
4
Workers who have been laid off for more
insiders.
are
^i
become outsiders.
infinity,
to
equal
m
the case where m«l
There are two extreme casesi
that
so
the initial
the
membe r ship never
that membership always coincides with
so
extreme cases highlight the effects
of
alternative
membership rules so we consider them before turning to the more difficult
i
ntermedi ate case.
'
The case of
a
constant membership
Let us denote by
employment in period
each period,
probability
if
of
ni
n_«
i.
beginning
In
(m*
nf
period
of
the present
exceeds no,
i
case,
i
i
membership, and by
members work
all
t y
;
if
n
s
is
being employed is given for each member by
and that the workers'
n
4
membership is equal to
assume that the one period utility function of
bwi)
ni
a
less than
to 6.
n_o
forever.
n_o,
as
So,
the
(approximately)
worker is given,
discount factor is equal
realised
l-n_o + ni.
above,
by
Thus the utility of
(p
s
We
+
a
member as of time zero is given by!
tpi + bKi3 where 6 is less than one
i«0
Assume for the moment that the shocks affecting
Eo E 6 1
Uo
over time,
or
more precisely that
e
t
iid,
is
uniform on [-&,+*].
below to the case of serially correlated shocks).
previous section, the probability
of
given by
=
(using the fact that Ee
s
labor demand are uncorrected
i
shall
return
Then by the analysis of the
being employed in period
0)
(We
i,
conditional on
w
s
is
)
.
22
pi-1
=
1
forn_o+cwi>-a
-
(
1
/ 4
1
)
(
4
no
w,
c
+
a
2
for
)
Given that employment outcomes do not
assumption that shocks Art white noise,
n_ D
ch,
+
<
-a
iHtct future membership,
the problem faced by
and
given the
members is the same
every period, and thus its solution is the same as that derived abovei
(2.3)
In
W,
=
n,
=
(l/c)(-
n_o
a
(
n_o
2
(
a(2(b/c)-l)>
+
b/c
)
-
1
)
+
e
response to white noise shocks,
and
,
employment will
smaller
has
It
to hold
larger than 1/2.
choose
they will
•firm
or
a
is
cushion
on
average,
outsiders who are laid
employment exceeds membership and the
adverse shocks.
first
in
case
easy to see that the result that employment
is
white noise will
o-f
regardless
of
of
e
the properties of
-f
e
is
e.
As
or
continue
shown above,
stochastically inelastic.
Changes
our
in
the
wages but do not affect the level
of
employment.
from its expected value affects the level
of
employment.
affect real
Only the deviation of
o-f
the stochastic process followed by
assumptions insure that labor supply
expected value
(b/c)
the insiders want strong employment protection,
If
wage so that,
a
Whether
membership depends on whether
employment is on average larger or smaller than
is
also be white noise.
rational expectations,
the unexpected component of
serially uncorr el ated
The case where membership eouals eitDlovment
(
m*
1
e
must be
By
)
(
23
We
now go to the opposite extreme,
employment.
i-1
i
this case membership at time
In
rw-i.
n_i
applied it to
(2.3')
Thus,
which membership comet and goes with
in
n_i
If
group kept the same decision rule 11 in equation
the
rather than to
n,_,
n,
simply given by employment at time
is
i
-
equation
n_o,
a(2(b/c)-l)
employment would follow
+
with drift.
under the assumption that membership equals beginning
given by
not
(2.3').
Optimal
current members should
memberships to wage policies.
changes fundamentally the character of the maximization problem.
membership, when taking wage decisions today,
next
period by
a
membership which will
This implies in particular that
laid off
The group
be
different from that
did not affect his future chances of
being hired
1
simplifying assumptions we have made so far, the problem
today.
j
this
where being
is
Even with the
intractable unless we
further simplify by linearizing the group's intertemporal objective function.
be
taken
be
*.
solution to this problem is treated in the appendix.
The formal
of
keeping employment with the firm
lower wage than in the previous case,
a
This
knows that wage decisions will
general
in
thus considerably decreases his chances of
presumably leads him to choose
The
insider is laid off, he becomes an outsider and
an
if
wage behavior
depend on its then current membership.
the group will
of
i
period employment is however
of
(2.3')
Unlike the behavior implied by
recognize their inability to commit future
subsequent policies
but
e,
random walk,
a
would become
(2.3)
(2.3)
Let w'
the wage around which the objective function is linearized and let the shocks to
labor demand be white noise.
(2.4)
W»
-
(l/c)(-
nt
»
n,-i
=
n_i_,
a (2
The solution to the maximization problem is then
+
(b/c)
a(2(b/c>
(
1
/
(
1
+
(
bSw
The probability of employment for
i+b6w
1/
'
)
a
>
-1
)
+
'
)
>-l
:
>
e
nember is
a
constant and is given by
i
24
«
p*i
Thus,
For
a
1
-
at (b/c) (1/
(l+bGw
'
)
under this membership rule,
2
employment follows
i^tct current employment,
expected employment.
)
drift
The
is
In
such
a
case,
random walk with drift.
Uncorr
and through employment,
positive
if
(b/c)
workers care sufficiently about the probability
wage.
t
there is unemployment hysteresis.
given labor •force,
labor demand
)
of
is
e
1
a t
shocks to
ed
membership and future
less than
(
1
+b©
w
'
)
/ 2
employment as compared to the
although they do not care about the unemployed, they will
the wage each period so as to have the firm hire on average new employees.
the wage
given membership,
the probability of
if
,
is
always set lower than
employment is set higher;
loss of membership and imposes
a
in
For
the m«infinity case
set
a
and thus
this is because being laid off
implies
a
much larger cost than before.
This analysis can again easily be extended to the case where labor demand shocks
The results remain the same;
are serially correlated.
a
random walk.
This is
a
consequence
of
our
maintained assumption that expected
changes in labor demand have no effect on the level
The
Intermediate Case
(m
between
1
and
employment continues to follow
of
employment.
infinity)
The intermediate case where workers remain insiders for some time after
losing
their jobs and where newly hired workers eventually but not immediately become
insiders raises an additional conceptual problem.
among insiders.
There will
no
longer be unanimity
Those who have already experienced some unempl oyaent
have been working in the firm for
a
short period of time,
cautious wage setting policies than those who have not.
face of conflict between members is beyond our grasp**,
for
A
A
,
or
those who
example will favor more
theory
of
behavior
plausible conjecture
in
is
the
25
that allowing for
values
of
to
1
case but
me »
less cautious than in the
More
between
m
• leads
and
to wage
setting policies that trt
importantly, rules corresponding to
m
unexpected shocks
employment.
In
the
of
lose insider status
|
in
level
of
namely
and 2,
1
Short sequences
unemployment.
of
sign have little effect on membership and thus on mean
same
the case of
the
in
case.
between one and infinity are likely
generate unemployment behavior such that shown in figures
infrequent but sustained changes
!
more cautious than in the
adverse shocks,
the case of
enough, to acquire membership.
But
insiders are not laid off
long enough to
favorable shocks, outsiders do not stay long
long --and
infrequent-- sequences
of
shocks
of
the
same sign have large effect on membership and may lead to large effects on the mean
level
of
employment. The length
employment depends
on
the
of
the shock necessary to cause
membership rules.
rules have to be symmetric.
The length of
general
permanent change
in
there is no reason why these
time after which an unemployed worker
becomes an outsider need not equal the length
insider.
In
a
of
time until
a
new worker becomes an
Hence favorable and unfavorable shocks may persist to differing extents.
The results of this section have been derived under very specific assumptions,
from fixed membership rules to the assumption that the firm was passive and that
outsiders played no role, direct or indirect,
return to these assumptions.
in
the negotiation process.
We
must
Before we do so however, we first show how the model of
this section can be used to generate permanent effects on aggregate employment of
both nominal
and real
shocks.
2.2 Persistent Effects of
Nominal
and Real
Disturbances on Unemployment.
26
We
now assume that there ire many firms in the economy,
insider workers.
own group of
so
that nominal
of
nominal
many firms
imperfect substitute for all
facing firm
is
j
gi ven
-k (p j-p)
yj
then characterize the effects
We
+
by
(m-p)
indexed by
others,
but
simplicity.
j
,
m
product which
being otherwise identical.
The demand
and pj
and
p
constants are ignored for notational
denote the output and the nominal
The restriction on
is
k
price charged
denote nominal money and the price level.
the firm's output depends on the relative price
balances.
a
k>l
The variables yj
respectively,
each selling
j,
i
variables are in logarithms and all
All
by firm
terms,
disturbances on employment and real wages.
The economy is composed of
an
nominal
in
its
derived demand for labor facing each group
The
is
further assume that wages trt set
disturbances can affect employment.
real
and
We
each dealing with
as
well
as
on
Demand for
aggregate real
money
needed to obtain an interior maximum for profit
maximisation.
Each firm operates under constant returns to scale
and employment
is
given by yj
«
nj.
If
Wj
constant returns and constant elasticity
are
given by
common to all
pj
Wj
firms
1
-
*.
e,
where
e
is
a
is
of
the
;
the relation between output
wage that firm
the demand
for goods
j
pays its workers,
imply that prices
random technological shock, which is assumed
i
27
Each
•firm
which chooses
faces
j
relation between
the demand
(2.5)
f
and wj,
pj
iniidert with the tame objective function at above,
of
wage and lets the firm determine employment.
nominal
a
group
t
can think of
we
each group
j
as
Given the
choosing wj subject to
unct on
i
nj
-
k(wj-e-p)
The choice of
(m-p)
+
and employment
the wage
decisions
We now characterize the
each group
j
at
time zero
moment we do not introduce the time index explicitly).
We
assume each group to
of
operate under the membership rule m«l,
given by nj(-l).
The group now chooses
its expectations of
the technological
shown earlier,
the price
shock,
Ee,
given such
a
level,
+
assumption
all
i
of
,
rational
we
level,
«
n(-l).
so that
real
wage,
of
employment
(Em -
Ep)
and the expected value of
to its
this implies
function
of
it
As
we
have
chooses
membership plus
a
a
!
nj(-l), Em,
must solve for the value of Ep.
expectations.
is
nj(-l)
«
a
equal
is
j
based on
enter the derived demand for labor.
As
all
We
Ep
and Ee.
do so under
the
firms and groups are the same,
affected by the same aggregate nominal shock, all
nj(-l)
a
demand function and its objective function,
which defines implicitly wj as
To solve for Wj
aembership in group
time zero,
nominal money, Em,
Ep,
Ignoring again the constant,
k(wj-Ee-Ep)
(2.6)
at
nominal rather than
a
which all
wage so that the expected level
constant term,
so that
(and for the
and are
groups have the same membership
Furthermore all nominal prices are the same and equal to the price
the first term in
(2.6)
is
equal
to zero.
Thus,
from
(2.6)
2B
Ep
w
*
j
Em
-
n(-l)
Ee
+
Em
and
n(-l)
-
depends on expected nominal
The expected price level
the
wage
The nominal
membership.
expected technological
values in
by their
dynamic behavior
or,
if
(2.7)
of
shock,
and
and
negatively
aggregating over
(m-Em)
+
=
ni-j
+
Replacing Wj and Ep
membership.
gives the equation characterizing
j
the
(e-Ee)
+
(mi-Em,)
on
money and
i
reintroduce the time index
we
hi
turn depends positively on expected nominal
in
aggregate employment
n(-l)
*
n
(2.5)
money and negatively on
+
i,
(e,-Ee,)
Shocks, employment and wages.
From
(2.7)
only unexpected shocks affect employment.
In
the case
of
real
shocks,
inelastic labor supply, which imply that
this comes as before -from the assumption of
each group sets wages so as to leave employment unaffected by anticipated real
shocks.
In
the case of
contract aodels
nominal
level
(Fischer
wage which,
employment.
of
nominal
in
1977)
shocks,
the
and the
intuition is straightforward.
result
is
the same
view of expected aggregate demand,
as
will
in
other nominal
Workers set
a
maintain last period's
Firms simply mark up over this nominal wage.
Unexpectedly low
aggregate demand leads to unexpected decreases in output and employment, with no
changes in nominal wages
returns
l
)
(by
assumption)
and
prices
in
(because of constant
*.
These unexpected nominal
and real
shocks,
however permanent effects on employment.
membership rules.
unlike other contract nodels, have
This is the result of our assumptions about
Once employment has decreased,
it
remains,
in
the absence
of
other
29
shocks, permanently *t the lower
level.
sequence
A
of
unexpected contractions in
aggregate demand increases equilibrium unemployment permanently.
m,
the membership
was
rule,
greater than one,
While the implications for employment
would again obtain the result that
are straightforward,
there is no simple relation between employment and real
particular the effects of nominal shocks.
scale and constant elasticity of demand,
Equi val ent
This lack of
assumptions
of
1
y
,
the role of
assumption
the model
wages.
of
implies
Consider in
constant returns to
they leave the mark up of prices over wages
wage unaffected.
Thus,
a
sequence
of
a
simple relation between real
wages and employment comes from our
monopolistic competition and constant returns, not from our
membership rules,
But
By our
they leave the real
assumptions about insiders and outsiders.
further.
it
we
is
real
an
do not
As
our
focus is on the dynamic effects of
explore the relation between real
wages and employment
important caveat to the line of research which has focused on
wages in "explaining" high European unemployment.
In
the model
constructed here,
it
is
quite possible to have sustained high unemployment without
high real
It
is
also possible for expansionary policies to raise ecployment
wages.
without altering real
2.3.
a
disturbances will decrease employment, with no effect on the real
adverse nominal
wage.
assumed that
such shocks would increase equilibrium unemployment permanently.
long sequence of
unaffected.
we
adverse shocks had no effect on equilibrium unemployment,
while short sequences of
that
we
If
wages.
The endogeneity of
membership rules
30
In
the rest
extensions.
We
this lection,
of
Here,
employment,
return to the original
focus on the determination of
we
have shown that
the nature of
we
these rules.
substantial
To the
extent
persistence
is
that
examine various
employment depends critically on
insider status is closely linked with
likely to result.
membership does not
If
change or changes relatively little when employment changes,
be
and
the membership rules.
time series evolution of
the
model
employment
is
likely to
much less persistent.
It
optimal
is
clear that at any point
to commit
the group
ignoring the welfare
the rule m=l
to
in
time the currently employed would find it
maximizing their interests indefinitely, while
those currently laid off.
of
this period and m»» hereafter.
But
That
they would like to apply
is,
this means that
if
the currently
employed are those who decide about membership, the only time consistent rule is
which is always the best current period rule for the currently employed.
therefore whether the group can precommit itself,
or
currently employed can commit the group to take care
whether or not they are still
be
The issue is
more accurately whether the
of
their interest in the future
employed by the firm.
Achieving the m*» solution is probably not feasible.
that the group will
i»=l,
able to commit
itself
to at
But
it
seems plausible
least some extent.
The factor
limiting the commitment will be the degree of divergence between the original
membership and the group
of
employed workers in some subsequent period.
divergence is too great, current employees will wrest control
controlling
it
in
the interests
groups can commit themselves
that level
period.
and composition of
is
of
some group of past workers.
of
Where the
the group from those
The extent to which
probably greatest where demand shocks are small so
employment change relatively little from period to
31
This suggest* that
variance,
have large
m
m
sequence
A
the
shocks.
If
shocks
Dr
m
may instead be
a
function
of
the realization of
the
large positive or negative unexpected shocks may lead to the
the group by the
takeover of
original
of
of
may have to be close to one to avoid large differences between
membership and the employed.
shocks.
depend on the distribution
Hill
then current
labor force is on layoff,
employees.
When
a
large fraction of
incentive for the workers still
the
employed to
ignore them and thus not take the pay cut required to get them back Bay be strong.
This is much
less likely in the face of
small
shocks.
Changes in the value
of
n
associated with major shocks provide another possible explanation for coincidence of
persistent and high unemployment.
Our model
thus suggests two alternative explanations for the empirical
observation that unemployment remains
First,
for
a
given fixed value of
sequence of adverse shocks will
the level
of
m
high levels for long periods of time.
at
greater than one but less than infinity,
lead to
employment permanently,
a
change in membership and therefore alter
Second,
in
bad times
currently employed workers
are more likely to take over and disenfranchise the unemployed,
value of
it
and
increasing persistence.
The two differ
process for unemployment at high levels.
process for unemployment will
be higher
2.4.
In
the first,
In
their implications for the
after the level
the second case,
but exhibit more persistence.
Li a
in
thus reducing the
have higher mean but the same degree of
around the new nean as it had before.
a
itations and Extensions of the Model
change, the
persistence
unemployment will
not only
.
32
developing our
In
regarding functional
ana
1
yi
i
,
have made
we
forms and the structure of
The question arises of
clear that
is
simplifying atiumptioni
of
bargaining between workers and firms.
how senstitive our results are to these assumptions.
also carefully avoided using the term
it
number
a
"union" to refer to the group
of
have
We
insiders.
But
"union" would often have sounded more appropriate and the issue
arises of whether our analysis is actually relevant
in
non union contexts.
now
We
discuss these issues informally.
Other bargaining structures
is
It
a
one period model,
the firm choose employment unilaterally given
to let
(1965')
known that even in
well
)
In
our multi-period model,
it
it
the
wage
in
general
(tee
for
inefficient
example Oswald
the assumption that the firm chooses employment
according to its short run profit maximizing labor demand it even more quetti onabl
Even
i-f
bargaining takes the form
to control
the level
(unless «=»)
the union
employment, firms will
of
run labor demand curves.
membership
of
,
setting
a
eaximi
z
ati on.
a
We
level
of
wage and allowing the firm
not choose to operate on their
short
Through its employment decision, the firm can affect future
By employing more workers this period,
membership next period and thus lower the expected cost
firm to choose
e
of
it
labor.
can
increase
This will
lead
the
employment higher than that implied by short run profit
suspect that taking account
substantially alter our analysis
of
of
this consideration would not
employment dynamics.
Rather,
it
would simply
shift each period's labor demand curve outwards.
Another important possibility would be for the firm to introduce two tier
systems, where newly hired workers get lower wages than those hired previously.
such systems,
insiders should have no reluctance to let firms hire more workers,
Under
and
33
employment should increase until new hirees ire paid their reservation
reluctance
general
suggests that
lower wages.
of
the
a
of
unions to accept such arrangements, especially in Europe,
central
issue is that
what
of
happens over time to those hired at
Unions do not encourage two tier arrangements at least in part because
workers will
that second tier
•fear
The
ixige,
Indeed the rarity of two tier
the membership
come to control
arrangements
is
considerations stressed here.
the wage setting process.
strong evidence
•for
Without some such consideration,
difficult to see why unions do not always favor such systems as
the rents that
the relevance of
a
way of
it
is
maximizing
they can capture.
Going back to the setting of the wage,
if
allow the wage not to be set
we
unilaterally by the insiders but to be determined by bargaining between insiders and
wages will
the firm,
depend both on the utility of insiders and on the value
profits.
firm,
the present discounted value of
wage.
Thus, the larger is the weight of
wage,
and thus the higher the average
Profit is
a
decreasing function
the firm in bargaining,
the
of
of
the
the lower is the
level
of
employment.
employment persistence depend on the weight
of
the firm in bargaining when the wage
is
far from the reservation
level
of
workers.
If
the firm is relatively sore powerful
when the wage is much above the reservation wage,
when
it
is
high, employment will
then the wage will
tend to return to
this happens depends on the structure of
The implications for
a
higher level.
tend to decrease
Whether or not
bargaining between insiders and the firm.
The specific utility function we have used for insiders is also important for
our
results.
Its main
implication, which we have argued is
the probability of
employment chosen by the group
group of
or
insiders,
to
the level
of
is
productivity.
of
desirable one,
is
that
invariant to the size of the
If
instead an increase in
membership given productivity led the group to choose both
probability
a
a
lower wage and
employment -which we can think as the stochastic equivalent
a
of
lower
elastic
34
supply-,
labor
components
of
employment would depend on both the anticipated and unanticipated
productivity and may show less persistence. Even under the rule m«l,
unanticipated increase
temporary,
in
lead to the choice of
the •following period,
of
D
same effects would also arise
if
as
group chose
the
employment
level
of
unemployment became
a
higher probability
relevant when workers are not formally organized in
when for example wages are simply set unilaterally by the firm
The work of
950
lower probability of
a
?
our analysis still
unions,
C 1
lower wage and
leading to an expected increase in employment over time
Grouos or Unions
Is
increase in productivity was
the
and being unemployed became more costly,
employment,
formal
a
if
implying an expected return to the initial
The
employment over time.
larger
employment would,
in
an
Lindbeck and Snower C19B5D suggests that even
unions current workers have some leverage vis
provides confirming empirical
in
vis firms.
a
?
the absence of
And
Slichter
evidence suggesting that even before unions were
economically important, wages tended to be high
industries with relatively
in
inelastic labor demand
current incumbent workers and prospective workers
many non-union settings,
In
cannot be regarded symmetrically.
the col
1
ecti ve
The requirement of
cooperation among workers and
knowl edge possessed by incumbent workers make their position very
different from that
of
This leads us to suspect that the
prospective new workers.
membership considerations we have stressed ire at least somewhat applicable even in
non-union contexts.
The potential
applicability
settings may be argued informally as follows.
insider workers.
The firm must
choose
a
of
analysis to non-union
our
Imagine
a
firm facing
wage and an employment
a
level.
collection
It
cannot
of
35
credibly threaten to lay o<< *11
because
cost,
of
its
expertise
the speciali:ed
workers because the remainder
the
of
Under these conditions,
Workers will
unorganized.
long
as
labor •force.
its
If
general
in
firms make an
be
"
force will
labor
On
the other
hand,
not tolerate the hiring of
wages and expected employment will
way to divide the surplus resulting from
firms.
of
except at very high
credibly threaten to replace workers individually with lower wage
the •firm cannot
"scabs".
workers and replace them,
set
in
some
continued relationship between workers and
a
able to extract
some surplus even when they are
inadequate" wage offer,
they have some specific capital,
be
will
it
be
they can refuse to work.
As
preferable for management to make
another higher offer rather than lay the worker off.
If
agreements are renegotiated only periodically and firms are permitted to vary
enployment in the interim, shocks will
Even without
a
formal model
of
seems reasonable to expect that
lead to the setting of
persistence
even
in
in
of
influence the level of employment.
the bargaining process between workers and firas,
a
reduction in the number
higher wage and
a
lower level
of
of
it
incumbent workers will
expected employment.
Thus
employment, though not necessarily as much as with unions, may result
that case.
union sector
a
general
in
Note that this also may help explain what
goes on in the non-
economies with large unions.
This argument is clearly tentative.
But
we
conclude from
it
that,
while the
effects we have described are more likely to be present when there are explicit
unions,
i
they may also arise in settings in which insider-outsider considerations are
mportant
.
The presence of
a
non union sector
i
36
We
finally consider how our conclusion! must be modified
market is neither unionized nor subject to
The
simplest analysis
of
empl oyment Ac
competitive sector would hold that
a
Wages in the non-union sector would fall
those workers ejected
reluctant to lower wages because
suspect.
is
the fear
of
alienated their current labor force.
of
Second,
are available.
In
In
being unionized after they have
unemployment benefits nay be
in
the non-union
sector is below some
another sense the unemployment is involuntary since the unemployed
in
the union sector.
that when there are wage differentials across jobs,
The general
the concept
of
consideration
involuntary
unemployment becomes elusive (see Bulow and Bummers 19B5 for an elaboration
Third,
unemployed
a
is
union job.
alternatively
unemployment
in
unemployment may occur even with
some sense useful
This may occur
if
is
accepting
a
to prevent
a
a
competitive sector
--or thought to be useful
substantial
if
low quality job
search effort
sends
a
or
if
by workers--
of
that
remaining
in
getting
queuing is required or
bad signal
employers.
to
This
related to that of Harris and Todaro [1970] where workers oust
oigrate to urban areas to have
There is
a
one sense their unemployment is voluntary since jobs
nay envy workers with the same skill
theme).
of
competitive firms may be
First,
sufficiently high that the market clearing wage
workers' reservation wage.
to
from the union sector could find
There are at least three reasons why even granting the existence
.
competitive sector, this analysis
is
labor
the
considerations.
der -out si der
ns
setting with
a
there was no involuntary unemployment.
the point where all
1
part of
if
a
more fundamental
unemployment.
chance at high wage urban jobs.
point regarding the inability of
As Weitzman
[1962] persuasively argues,
a
non-union sector
there irt strong
reasons to believe that most economic activity involves fixed costs and monopolistic
competition.
Imagine
a
conopolistically competitive economy with fixed costs
production and constant marginal
costs where there is initially no involuntary
of
37
Suppose that an adverse demand shock reduces the demand for goods in
unemployment.
this economy but
that
nominal
employment and output will
wages remain constant in all
-fall
as
will
existing fines.
Then
existing firms.
the profitability of
low wages?
pay new firms to enter the market and hire the unemployed at
It
Will
it
may not
because unlike incumbent firms, new firms must cover fixed as well as variable costs.
Particularly
sales,
in
settings where labor costs do not represent
entry may not be able to insure the employment
considerations may enhance the power
start up new non-union firms.
of
of
a
large fraction
the unemployed*-*.
of
These
unions because they reduce the incentive to
3£
3.
Evidence on Hysteresis Theories
Empirical
Having developed
is
a
•formal
theory
of
consistent with the observed patterns
Europe and whether
it
can
hysteresis,
of
persistently increasing unemployment
illuminate the very different behavior
Europe and the United States in the recent past.
institutional
now examine whether the model
we
evidence on the strength
of
unions
We
unemployment
We
then
estimate wage and
employment equations implied by our model, for both the Europe and the US.
examine patterns of labor market turnover,
The role of
3.1.
the UK and
We
finally
the US.
unions in Europe
the union sector
The size of
suggests that,
Dur aodel
in
in
start by giving direct,
Europe,
in
of
in
even if
hysteresis may arise in non union contexts,
it
is
probably more likely to arise the stronger and the larger the union sector. Thus,
we
start by reviewing the role of unions in Europe
|
we
limit,
as
before,
our
investigation to the UK, France and Germany. 22
Membership figures indicate
in
France and
367.
the strength of
proportion
coverage is
workers.
of
of
for Germany.
unions.
A
a
But
union density
approximately
of
4 5X
workers covered by some form of collective bargaining.
70'/.
For France and Germany,
coverage numbers are misleading.
institutional background.
20
these figures give very limited information as to
better indicator is union coverage, that is
approximately
the UK,
for
for
aanual
workers,
the proportion of
To understand
why,
and of
all
55X
of
For
the
the UK,
for non nanual
workers exceeds BOX.
one must be given tome
But
even
'A
39
On
the three countries appear
surface,
the
are
three main national
UK,
there is
maze
a
unions,
In
Germany,
In
France there
there irt only industry unions.
craft and industry unions.
of
very different.
to be
the structure of
But
in
fact quite similar and can be described as follows
of
the formal
bargaining is done at the industry level.
But,
bargaining
three countries,
all
in
i
the
In
in
all
is
most
three countries,
wages are determined much more at the company or plant level!
In
the UK,
industry bargaining sets rates,
wages
little effect on actual
extension,
sector.
i.e.
.
the Employment
Until
provisions to extend the terms
for
Act of
of
plant level,
increase
the amount
in
I960,
there was scope for
the agreement
These provisions have been eliminated in I960.
has been an
the
In
to
the whole
between shop stewards and employers.
of
Given that
bargaining,
it
is
there
last 20 years,
bargaining, both formal and informal,
of
bargaining is the really important level
pi
197B,
the number
of
industries and
workers covered by at least
ant /company
relevant to look at how
of
47.7'/.
company agreement was
a
for manufacturing.
the
at
workers are covered by both industry and plant/company level bargaining.
r.any
all
which are usually floors that have
Given the importance
of
of
In
for
33'/.
informal
bargaining, these figures understate the importance of unions in setting wages.
In
France,
the
"Conventions collectives" which are usually but not always
industry level form most
between
a
not
unions sign
all
of
the formal
"representative" union and
it
a
bargaining.
of
Labor.
As
"representative" employer and apply even
(which is frequently the case).
in
the UK however,
the
respect to wages should not be exaggerated.
appear,
either directly or indirectly,
the UK,
a
growing portion of
if
Subject to some sinor
importance
of
of
the
industry agreements with
They usually set floors, which do not
to have
the bargaining
the
These agreements are signed
conditions, they can be extended to all firms in the industry, by decision
Minister
at
a
large effect on actual
takes place
at
wages.
the company level,
As
in
40
although often in haphazard fashion.
unilaterally by firms, or
plant,
in
Until
19B2,
response to complaints
with little bargaining or even consultation
standard instrument used by unions to achieve
been
a
change in the law
company level
number of
In
level.
wages were largely determined
pay and
on
most
|
strikes Here however
local
better deal.
(Lois Auroux)
which requires annual
other matters.
The result has been
Since
a
there has
19B2,
bargaining
at
the formal
of
bargaining again takes place
Agreements can be extended -to either firms
if
half
(1)
the employees of
of
signed the agreement and
have signed 2 '.
But,
as
the
drastic increase in the
in
the
the
at
industry
same industry or to non
union worders in firms which sign the agreements -by the state or federal
Labour
a
agreements.
company level
Germany,
a
union representatives in the
of
(2)
in
place at the company level
Minister
of
the sector are employed by firms which have
extension is approved by both unions and employers who
the other two countries,
and there is general
bargaining is increasingly taking
agreement that pay is very largely
determined at the company level.
To
conclude,
these countries.
it
is
To the
the company level,
difficult to give an exact estimate
determined as
a
union coverage numbers,
Also,
if
we
"union sector"
of
in
which are based on both company and
of
workers for whom the wage
result of bargaining between unions and employers.
adjustment, the size
US.
the
extent that much bargaining over wages in fact takes place at
industry level bargaining probably overstate the number
is
of
the union sector
still
remains high, much higher than in the
believe that the more disaggregated the level
likely it is to take into account the interests
Even with this
of
of
bargaining the less
the unemployed as
a
these countries are good candidates for hysteresis in the union sector.
whole,
t*t
then
41
alternative approach it to ask the question
An
firm become non union
In
?
the UK,
There is nothing
non union.
of
are very
few examples
of
both countries.
In
yes
firms going non union ".
-
France,
on
firms in
There are tome well
US
companies
(Kodak).
France and Germany,
In
given sector,
a
firm be non union
?
can
a
firm can be or can become
a
i
most often subsidiaries of
agreements put some constraints
in
it
a
prevents it.
the law which
in
non union firms,
examples
the antwer
can
i
known
There
extension
There are non union firms
these are nearly exclusively small
firms.
In
France
furthermore, various requirements are imposed on firms with more than 50 employees.
In
particular they must allow for the presence
union representatives within the firm.
represented.
syndicale
d
Since 197B,
'entrepr se"
i
,
All
of
"delegues du personnel" who are
national
unions have
right to be
a
firms must also allow for the presence of
for the presence of
"section
a
Together,
the union inside the firm.
these facts suggest that it is difficult to be or go non union in these countries.
Finally,
union sector.
there is the question of how different the non union lector it from the
A
study by Kaufman C19B43
of
the competitive sector
the UK finds
in
relatively little difference in wage behavior across the two sectors.
the arguments given in the previous section,
formal
shall
union sector nay not be
a
Together with
this suggests that the size of the
major determinant of
the extent
of
hysteresis. We
return to this question in the next section.
Membershio rules
Membership rules determining who the union represents at each point in time,
play an important role in our analysis.
rules is fairly clear
i
The empirical
evidence on actual
workers have the right to join unions
if
membership
they want to.
42
Workers who are laid off can remain in the union although they
to
vote
|
this may happen either because of
takes place inside the plant.
But
interest the union actually acts,
[Barker et
al
1
9B4
D
do
for
not,
the
the most
for
idea that
3.2.
this tells us
study
A
part,
It
little about the question
and
finds that, while laid off
see reasons
to
stay in the union".
in
of
the unions
the union and have their union
in
lose the right
restrictions, or because voting
the unemployed
of
gives some information.
officially encouraged to remain
formal
oUen
in
whose
the
UK
workers are
fees waived,
they
This provides support
union cares mostly about the currently employed.
the
Wage and employment eouations
Theory
We
now derive,
and then estimate
associated with an expanded version
First we allow for
two extensions.
the model
of
a
the wage and employment equations
later,
We
n
=
s
n-i
-
(l-s)b (w-p)
Following the analysis
of
Second,
"
En
«=
(l-a)n*
+
an-i
labor demand
;
the
specify explicitly an
+
|
e
the previous
section,
expected employment according to the relationi
(3.2)
we
of
There are
hysteresis.
thus specify labor demand as
(3.1)
set
of
the previous section.
dynamic specification
reason for introducing it will be clear below.
alternative hypothesis to that
of
we
assume that the union acts to
3
,
43
The case where *»1
and there
corresponds to the
policy is independent
of
where m»l
the preceding section
in
a«0 corresponds to the case where the union's
(the case where
hysteresis
is
cue
history and so there is no hysteresis.
Clearly,
intermediate outcomes are also possible**.
Finally,
the wage which satisfies
let
wage to be given by
assume the actual
w
=
w*
+
i
is
yields
Combining this assumption with equations
(l/b(l-s)
w
*
Ep
(3.4)
n
=
(l-a)n*
+
+
a
n-
)
[-(l-a)n*
s
When the union is larger,
ambiguous.
is
Ce-Ee
+
(s-a)n-i
+
+
1
(
-s
)
b
A
is
it
it
The degree of
Unexpected movements
in
+
u
is
a
)
the union
decreasing function
raises the size
of
the group
(3.3)
i
(3.4)
can be
To do
process followed by
e
it
in
(3.4)
of
at
all
a
on
assume that
yields
i
so requires
follows
a
s.
and
estimated by OLS. This is
expected productivity
derive the reduced form wage equation.
substituting
not
a,
is
correlated with past productivity and thus with past employment.
we
of
an
likely to be
Therefore we now
assumption about the
random walk-*^.
n-
whose interest
implies that employment follows
persistence depends only on
Equation
however not the case for equation
s
in
The impact
employment are due to price and productivity surprises,
wages from target.
e
(3.1)
also increases labor demand.
first order process.
of
Ee]
more cautious in setting wages.
larger value of n-i
maximizing but
+
(p-Ep-u
The ecployment equation on the other hand
deviations
We
equations
wage and an employment
a
(3.3)
the union
denoted by w*.
be
!
The wage equation holds that the wage of
n*.
(3.2)
assumed to be white, uncorrelated with w* and
is
u
reflects factors outside the model.
(3.2)
and
u
where the disturbance term
and
(3.1)
Lagging
(3.1)
and
44
w-w-,
(3.5)
where
-
£
k
(
=
1
/b
(Ep-p-j)
+
k
1
<
-s
>
)
(
(1/b (1-s)
+
l-a> n.
of
expected price inflation,
worth examining
it
[
(
l*s-a)n_,
and
It
lagged employment only.
if
n-
3
3
+
u
wage
inflation as
employment lagged once and twice.
a
is
It
further.
this case the relation gives
employment but
s
gives the rate of
Consider first the case where there are no costs
In
-
.
This equation can be estimated by DLS.
function
)
*<1,
If
leads to lower employment,
relation between expected real wage growth and
then expected wage growth does not depend on
»»1,
does
it
a
adjustment in labor demand.
of
after an unexpected decline in productivity,
i
workers accept
the remaining
a
cut
in
real
which
wages only to
the extent that they care about the workers who have been laid off.
If
there are costs of adjustment to employment,
depends on employment lagged both once and twice.
If
then expected real
a *
,
then
the ratio of
employment lagged twice to employment lagged once cannot exceed 1/2
value).
:
But
as
a
increases, the ratio tends to one.
If
a
1,
wage growth
(in
absolute
the ratio equals unity
expected real wage growth depends on the change rather than on the level
empl oyment
.
Note that we cannot
equation.
But
than
Furthermore,
1/2.
of
a
identify
a
and
s
separately from estimation
of
the wage
must be positive if we find the ratio described above to be larger
a
can be
directly obtained from the employment equation.
45
While we have derived the wage equation
union behavior,
can be motivated
it
monopolistic competitive model
logic
the
of
parallelling
side.
is
Much more generally,
the presence of
rate of
a
Following the logic
of
the
of
section just as we have followed the
equation
(3.5)
very close to
is
change effect,
a
Lipsey.
la
a
standard
The only real
employment rather than unemployment on the right hand
now turn to estimation of
We
rather specific theory
<
gives rise to an equation for wage inflation
competitive model
(3.5).
•from
other ways.
the preceding
in
Phillips curve which allows for
difference
in
(3.5)
the wage and employment equations.
Results
The results of estimation of the wage equations for the UK,
the US,
In
for each
for the period
19B4 are reported
1953 to
in
tables
4
and
France,
Germany and
5.
table 4, four alternative specifications of the wage equation are estimated
Because the appropriate timing is unclear with annual
country.
data,
we
estimate the equations using alternatively contemporaneous and once lagged
enployment, and once and twice lagged employment 36
with and without
Phillips curve by
history
of
time trend
a
time trend,
a
unemployment and
specification when
specifications.
|
a
it
that
is
by
an
also estimate each equation
increase over time unrelated to the
interesting to see what happens to our
is
This gives us the four alternative
Finally, we use for expected inflation the forecast of
of
an
AR
(
1 )
inflation
process for inflation over the sample period and
constrain the coefficient on expected inflation
a
We
many researchers have captured the shift of the
time trend is allowed.
obtained from estimation
constant plus
.
scalar tires lagged inflation)
(which
to equal
is
therefore equal to
unity.
a
1
'
i
i
-a
in
C3s
n
o
.—
CM
os
r~
O
>^
rn
-a
-
o
-a
vo
O
O
iH
E
*—%
O
csi
.—
in
.
i
o
O
H
-a
csi
O
CJS
o
OS
o
rsj
m
—
rs
—
in
r
o
X
m
mo
cm
•
-I
I
cu
m
oo
O
o
rH
in cm
O
VD
O
in
o
in
csi
•
o o
CSI
•
CSI
I
O
03
O
-rK
a*
u
o
3
-O
£=2
—
C
B
.H
or
en
Ln
ON
C3
L-.
CO
o
est
co
•
•
»a
I
i
r-»
o
csi
in
-csi
i
r-~
•
I
I
.
in *c co
•
•
co
.csi
OS •^
CO
I
I
CSI
W<
csi
,-v
is
I
I
CJ
CJ
C5
3
NOHCCN^sTH
os
•
o
•
r*s
•
co
'csi
-a
s-
—
6J
CO
•
r-^
-co
•
I
I
o
o
O
^~v
>-^
•
i-n
•
csi
•
r->
|
|
i
mmoNSHOH
i
I
(
•
vo
os
OS
•
en
^co
rs
OS OS
CSI
CO
CO
CO
O
rH
h < -c
n
CSI
o
o
/-s
*
*
*
*
Ln
r~
Ln
rs
rs
r-~
m
m
CO
e
o
C
>s
U
3
U
>s
^»
«
U
CJ
o
•
CN
I
"3
CO
U
CSI
I
•
CN
.—
O
I
OS ^^.
/^ CO
— oocstocsi.—im
o
co
i
—
!1
<
li
cr.
-a
CS)
PS
3=
n
o
cni
-3-
cm «a
(Nl
•
csi
oo
O
cr.
cr.
/-s -a
•
cni
CO
lo
n
^
r*-
•o
o
•H
°
•
c
E
o
O
^N
O
O O
H
X
N-/
oj
r->
On
^H
1
•
•
1
•H
JJ
OJ
u
^H
OJ
a.
l-l
e
•H
l-i
l-l
re
CO
v_^
u
OJ
OJ
(NJ
I
r-
JJ
o
—1
JJ
~c'
CJ
t-H
JO
O
H
re
3
cr
w
~3 *~N
•
CO
Cr, •B
iH
1
w
(-1
CJ
cr.
m
CO
o o
o
o o o
c
CO
H
JJ
re
re
r-l
•
i-i
OJ
>
3
u
o
XJ
CO
e
•H
>
O
E
3
u
CS
OJ
u 3
OJ
u
rt
jj
d
3
y
re
CO
re
3
fd
cd
O
.
•
I
O
•
•
CN
•
—
•
<r
•
I
co
:
r-i
CO
0J
-3
jj
o
CO
u
3
l-l
o
o
L-l
OJ
OJ
CO
l-l
JJ
•
3
JO
JO
3
o
o
JO
u
re
OJ
_*
l-l
o
•
a
>.
U
r^cor--cou"iooori
JO
JO
l-l
"3
I
JJ
cr.
>.
u
l-l
O
u
X
JJ
re
OJ
00
a
-^
-a
CO
/
QJ
u o
V
en
a.
-w
o
1
OJ
o
JJ
r-i
C
o
l-l
i-i
JJ
o
CO
i—
l-l
—
/^n
U-i
*-{
-3
O
m
cr.
1—1
OJ
re
o
o
J^
l-l
o
a.
JJ
u
3
JJ
CO
-3-
o
CNl
•
o
co
CM
cr.
•
CI
CO
re
u
H
OJ
3
E
3
05
C
u
0)
E
O
OJ
0)
c
JJ
o
CJ
CO
to
OJ
cj
OJ
cp
CO
re
re
"u
"a
Hc
l-l
l-i
O
o
OJ
OJ
3
jj
U
re
—
CO
CO
CJ
re
S
JJ
OJ
u
re
C
CI
OJ
J
KSi
l-l
U
o o
E
o
l-i
l-i
3 Ui Ux
H
CO
to
CJ
C
co
re
JO
l-l
—13
E
re
CJ
3
•H
V—
<—
CO
-J
I
i-i
re
re
•3
CO
3
o o
OJ
x-t
JJ
re
OJ
3
O
CO
3
S
1J
eo
-3
-3
re
OJ
*
re
-
s
E
u
l-l
JJ
>
re
3
C
3
OJ
l-l
CJ
re
-3
^
—
U < <
OJ
o
u
o
Table 4b
Wage Equation Residuals
1953-1984
Year
Germany
1980
-1.91
1981
-
1982
-
France
United States
1.7
1.6
-1.2
.32
-4.1
1.4
-
.8
.75
3.9
.0
-
.1
.57
-2.7
.1
-
.9
.44
1.1
-1.5
.3
0=1.87
o=3.2
o=3.9
o=1.5
1983
1984
United Kin .gdom
-
Residuals from equations
3,
5,
-
11 and 15 in table 4.
I
'
I
O
ON
in
~3-
2
ON
o o
s
i
s
va-
in
in
n
On
ON
O
o
CNI
\£>
cn co
m
r-i
*j-
^j-
<T
o
•
O
CO —I
c
CNI
m
cn
co
o
^N
r-
CNI
n
o
*>
(nj
"
•
N_*
1
rsi
CNI
"
1
—
o m O
,
•
r-l
r-l
8
in
H
-
X
I
I
r-~
O
•—»
o m
cn /—n
-j-
i—
•—
•
I
i
r-~
o ^^
m
T-(
•
CO ~T
in
(NJ
•
I
I
ON
on
co
co
f—
•
i
cni
r-i
O
cni
^N
CNI
I
0]
c
o
CNI
T-l
CO
m
3
tr-
to
«-i
^
rt
H
o
00
C
C5
r~
U
a
c
CO
P>
3
•
CNI
CO
CNI
•
<—
cn
-—-
•
CNI
O
oo
m
-3•
\o
«j
m
•
o o
^-n
i-i
on
/N
H
o>
r-v
z>
cn co
cn
r-l
in
en
,-n
CM
<r
l
cni
«—
CNI
N_'
^-n
r-l
CNI
cn
O ui
— O
CNI
N_/
on
n
m
on
1
•H
1—
to
n
CNI
O
1
AJ
J-l
O
I
4J
co
=>
•
CNI
1
r->
i—
•
<r
<
1
-^
•
CNI
1
O CO
n oO
^n
m
ON
a
!
i—i
•
a
r-t
CNI
CNI
s_^
1
1
.
I
S^
*cr
CNI
j
O ^—
oo h
ON /—.
cn
|
CNI
I
-if
I
rH
I
O
v-'
•
•
o
r-
r-l
CNI
1
I
I
-K
m
-fc
m
-K
in
r-~
r~-
r^
.r^
3
C3
E
O
CJ
rj>
3
J-i
n^
|
I
r-.
;co
^n
co
—
P1HC
r~.
m
CNI
•
CNI
r~ /-n
on on on
r-^
CNI
I
I
in
>
C
I
^.n /n
o
o sr o
CN
o
>N
.
t-i
1
B
U
co
•
-3
J-l
^-N
-W
I
o
CNI
o
•
v_^
|
•
I
r-.
°
CNI
•K
o
00 /-N
. -r-»
cn
fc
CO
CO
CO
CO
I
46
table 5,
In
perform the same set
we
rather than employment as
unemployment
theories
of
a
the variable
is
of
estimations,
right hand side variable.
used
but using unemployment
We
this because
do
standard Phillips curve specifications.
in
Some
hysteresis such as the idea that the long term unemployed exert less
dH
pressure on wages than those recently laid
also suggest
that
unemployment
is
more
appropriate than employment in the Phillips curve.
Tables
and
6
give the results of
7
estimation
processes for each country for the period
1953 to
has
implications only for employment, we think
for
unemployment as well.
it
the
of
1
9B4
is
employment and unemployment
Here again,
.
useful
to
while our theory
also report results
The results are fairly clear cut and indicate that there ire substantial
differences between the European countries and the United States.
wage equations,
(1)
tables
4
Let
conclusions
one can draw the following
Starting with the
i
Virtually all specifications for Germany, France and the United Kingdom
and
us
5
suggest
denote by
a
R
substantial
degree
the absolute value
hysteresis.
of
of
the ratio of
the coefficient
employment -unemployment- to the coefficient on contemporaneous employment
unemployment-
(or
of
the
(a=l)
unity for nearly all
trend,
or
across countries
:
R
this ratio should be equal
specifications
by the use of
is
lagged
on
-
coefficient on employment -unemployment- lagged twice to the
coefficient on employment lagged once as the case may be).
strict hysteresis
in
j
it
is
not
to
unity.
in
the UK,
average to .B5 for Germany and France 3 *.
R
is
under
have seen,
indeed close to
affected by the inclusion of
employment versus unemployment.
higher
we
As
a
time
There is little difference
sometimes exceeding unity.
It
is
closer on
'v
<
'
CNI
CM
v£3
es
On
On
g|
o m
—
-c-
H
1^ VD
IN
•
-r
•
<r
•
E
o
O
H
X
oj
m
c
o
re
"1
3
^
.2
-3-
C7
1
t:
CJ
6£
n
LI
1
a
C
i—
CO
On
.3
i—
u
/—
3
o
o
4-1
I
O
•
1
N
Csl
«a-
<NI
CNJ
O
°
vo
I
CM
I
I
i.
—
c
O
CJ
•—
-H
CJ
M
3
>—
•
CO
On *U
T— •
i
0)
I
I
4J
•
O
CN
•H
On
On
1
4J
re
•
I
•
*—'
•
£3
cm
•
o m
Oo
o
cm
•
•
•
cn
co
M
•
in
4J
/—N
3
*c
CJ
CO
ON
-H
E
>N
o
^**
r-l
c
E
CJ
a
j-i
C3
4-1
2
3
o
OJ
.-(
•o
•o
-o
N
H
re
-H
-3
O
a
-o
to
c
to
4-1
a
3
O
o
o
l-i
CO
4-1
-o
a
4J
Ul
re
Q
U
u
o
re
>N
1-
3
•
c
05
.3
-Q
•
CO
05
a
T-l
r*
4J
5
2
••
o
13
a
a
o
Ul
3
to
Table
6
Employment Processes
1953-1984
a
xlOO
Country
Germany
.76
(22.3)
.86
(26.7)
1.00
-
(5.3)
.78
(3.9)
-
-1.9x10
.96
97
(.0)
United Kingdom
1.07
(23.3)
.54
(2.6)
.41
(2.0)
.95
(16.3)
96
-.20
94
(-3.8)
France
.81
(3.0)
.48
(2.5)
.94
(19.5)
1.08
(19.5)
,94
-.13
,94
(--4.0)
United States
.82
(7.5)
.34
(1.5)
Note
.
Results of estimation of
logE = plogE(-l) +
E
:
.72
.07
(.3)
.46
(1.6)
ex
:
(TIME) +
manufacturing employment.
c
+ c9(-l)
.40
(2.5)
.77
Table
7
Unemployment Processes
1953-1984
Q
Country
P
g
xlOO
R
2
Germany
.92
(14.8)
.94
.65
(3.4)
-
.91
.39
.06
(5.0)
.93
-
.95
(1.9)
(17.5)
United Kingdom
1.02
(20.9)
.77
(3.9)
.82
(3.9)
.81
(9.9)
-
.09
(3.5)
.96
-.06
-
.97
(-.3)
-
France
1.12
(32.7)
1.04
(18.2)
-.22
(-1.1)
.02
(1.4)
.97
-
.58
.07
.63
United States
.06
(.2)
.31
(.9)
.72
(4.5)
.36
(1.4)
Kote
.
Results of estimation of
:
U =
pU(-l) + a (TIME)
U
standardized unemployment
:
+
e
+ 0e(-l)
(1.9)
47
The
time trend itself
contributes little.
increase in the natural
due to an autonomous
France when unemployment
used
is
is
significant or marginally significant.
small.
the case
In
in
France in table 5),
consistent with
which
given level
a
rate over time,
used,
is
its quantitative
increase
1.5'/.
expected real
of
coefficient on the
the
and
the
in
contribution is
(equation
the unemployment
in
for
12
rate
wage growth over the sample period.
Further evidence that the apparent increase in the natural rate through time is
rising unemployment and not autonomous comes from the absence
consequence
of
substantial
serial
correlation
the constant term would manifest
A
final
piece
of
our
in
estimated Phillips curves.
itself
in
the form of
each country,
for
1
9
BO
to
1984.
prediction errors in recent years.
wage equations which do not
(2)
In
of
upwards drift in
correlation.
serial
including
not
a
time trend,
significant
of
This is in sharp contrast to the performance
of
allow lagged employment to enter.
contrast to the results for Europe,
the results for
There is evidence
either lagged employment or lagged unemployment.
specification using employment,
being in most cases around
4,
There is little evidence
provide evidence of much less hysteresis.
of
An
a
evidence is given in table 4b which reports the residuals
associated with the best fitting equations from table
for
UK
time trend positive and either
the
Even then,
a
unemployment was
in
largest and most significant
is
it
only explains
it
the increise
Only in the UK when employment
time trend should be positive.
and
If
.5.
the value
of
R
But,
of
the United States
a
significant effect
with the exception of
for the US is smaller
There is also no evidence in favor of
one
than for Europe,
a
time trend
in
the wage equation
(3)
A
comparison
of
the results
of
estimation
in
tables
4
and
5
does
not give
clear answer as to whether employment or unemployment belongs in the wage equation.
Using
R 2 's
gives
a
draw,
with employment doing better
for France,
unemployment doing
a
46
better
UK.
also run regressions including current and lagged values
We have
both unemployment and employment -or equi val ent
of
.
the
•for
They give the same ambiguous answer,
UK,
but
not
France
in
Germany.
or
We
1
y
,
employment and the labor force-
with the labor force being significant
see the UK results however
in
presenting
as
the
a
problem for our model.
The employment
and
unemployment equations reported in table
large extent the conclusions from the wage equations,
whether or not
a
The data however
is
a
particular, the process
three European countries,
The US process
is
needed tD fit the employment and
unemployment processes
of
all
four countries.
period
in
the
model
a
confirm to
strongly suggest that an ARMA {1,1), rather than
process implied by our theory,
the length of
In
tine trend is included in the regressions.
instead stationary.
the AR(1)
all
in
7
Both unemployment and
employment are more persistent in Europe than in the US.
generating unemployment appears. non stationary
and
6
and annual
This may reflect
frequency
of
difference between
a
observation used in the
estimation.
3.3 Patterns of
A
central
Labor Market Turnover
element
in
our
workers for the unemployed.
the outstanding
nodel
theory
It
is
of
hysteresis is the lack
the fear of
job
of
enough to insure the jobs
firms to hire new workers.
of
of
employed
loss for current workers and not
labor market pool that restrains wage demands.
explains why firms hire at all
concern
Indeed the formal
only by assuming that wages which are set low
current workers will
sometimes sake
it
profitable for
While this is clearly an oversimplification, the point
remains that insider-outsider or union models
of
the type we have
considered are
49
really theories of why the unemployed ire not hired,
place.
This suggests the utility of
finding
of
looking at
not
data
on
high turnover with many workers having short
theories
labor
market
spells
the relevance
formulations, while
flow into and out
finding that the rate
of
relatively low but that the unemployed remained out
of
turnover.
take
A
unemployment and
of
then being rehired would tend to cast doubt on
a
why layoffs
of
insider-outsider
of
of
work for
employment was
very long time
a
would tend to support these theories.
Table
B
presents some evidence
on
the rate of
flow into unemployment
United States and United Kingdom over the past decade.
The flow
number of persons becoming unemployed over
a
States,
this is estimated as the number of
unemployed reporting durations
than
weeks.
14
For Britain
it
is
the number
of
the
measured as the
is
three month period.
in
For
the United
of
unemployment registants over
less
three
a
month period.
Two conclusions emerge clearly from the table.
rate of unemployment in the United Kingdom than
flow into unemployment is actually lower there.
unemployment problem
is
insufficient rate
hiring of the unemployed.
of
not
one of
an
little as unemployment has soared.
unemployment
in
The implication
a
s
the United
between
1979 and
rate rose by 44 percent.
19E2,
the rate
of
that the
loss but of
an
striking feature of the
increased surprisingly
the rate of
of
flow into
This pattern of
rising unemployment
inflow appears more pronounced in
British
States, the inflow rate has accounted for
significant part of the increase in unemployment
example,
is
Between 1970 and 1984 when the rate
modest increase in the rate of
In
job
Britain has
in
Britain rose more than 300 percent,
than American labor aarkets.
of
The second
unemployment has risen by only about 75 percent.
with only
despite the much higher
the United States,
in
excessive rate
data is that the rate of flow into unemployment
First,
during recession periods.
For
unemployment increased by hi percent and the inflow
I
i
—
1
—
C
CJ
E
>.
O
tH
D.
E
W
U-i
o
c
H
fr-f
—
i-i
CO
u
en
ca
rt
«
i_i
re
0)
u
o
n o o
ro
ctn
cnj
r-i
r->
(ni
r-\
CO
cr\
cs)
in
3
Ho
U-l
c
u
01
u
u
a
3
C
c
0)
E
>i
o
o
o
.—I
a
E
3c
CO
o
3
-3
O
0)
o
s
to
H
a)
a
3
o
CO
as
O O
-3-
sD
-c
CM
On
CO
o->
r-»
o
CO
CI
o
a
E
>>
ED
05
a
8-S
t-H
C3
-3
a
a:
>>
CO
c
—
05
cj:
E
a
r-
3
s-
O
01
3
<r
iH
o
.c
c:
E
«
xj
g
o
CJ
O
3
E
>.
—
c.
c
H
-3-
CO
LTl
~T-3'-3
-
CM
>3'-comLnLniriinor^OLri
a
>
u
^
C!
o
U
a
3
a*
CJ
rf
CNI
r-~
ri
r>-
1
ON
as
o>
*c
L-l
r~
en
o
r-^
en
r^
r~0">
CO
r^
as
CI
1
OS
o
CO
ON
u
>-i
CO
as
CN
CO
Oi
(-)
*?
QJ
CO
CO
as
<
en
>
50
The OECD
(
1
9 B5
summarizes the fragmentary information available on labor market
)
turnover for other European nations,
Britain
— suggesting
starting from
a
The data
relatively modest increases
very low base.
parallel
general
in
in
the rate of
our
How
bindings
f
or
into unemployment
They do suggest however that the composition
the
of
newly unemployed has changed over time as the unemployment rate has increased.
Layoff rates have increased while quit rates have declined.
Given the magnitude
relatively small
increases
in
European unemployment rates and the
increases in flow rates,
it
is
of
the
durations have increased substantially.
increasing importance
the average duration
9
long term unemployment
of
of
Table
unemployment,
inevitable that unemployment
presents some information on the
in
Along with information on
Europe.
presents estimates
it
of
the fraction of
all
unemployment due to persons whose complete spells will exceed various threshold
lengths'34-.
unemployment
The table demonstrates that
is
at
the same level
much more important in Europe.
In
19B0,
of
unemployment,
when the American
unemployment rate was 7.2 percent, only an estimated 15 percent
was due to persons
were
74
percent,
out
of
work for more than
59 percent
and 75 percent
even though the unemployment rates were
a
in
lower.
year.
proportional
Summary
Indeed,
to the
the increase
in
the United Kingdom,
all
unemployment
Germany and France
The table also demonstrates that
duration
increase in unemployment.
of
The corresponding percentages
term unemployment has increased in importance as overall
risen in Europe.
long term
of
unemployment rates have
unemployment is almost
long
<
!
aCO
On
—1
O
o
<—
1
I
<T
o
•
•
<T
C5>
<T
VJ3
CO
>-(
CJ
o
c
ro
u
fcu,
O
vn
CO
On
•
•
rsi
v£>
CM
o>
vD
CO
ST
jj
c
01
>><r
E
c
CO
re
ex*
.-1
E
.-<
1-
E
ti
c
O
(Nl
CM
CO
CO
CO
a
0)
c
o
=)
O
CO
CO
E
BS C^
U< ^H
m
CM
<u
H
.
DO
c
o
-J
JO
O
H
o>
o
c
C3
-J
u
o
c
E
-3"
CO
o\
m
CO
•
•
r--
LI
CM
o
o
ON
O
co
cr\
^H
C
O
JJ
-H
>.
U-l
01
JJ
re
c
o
i-i
JJ
jj
c
0)
c
>1
O
»—
C
E
01
^-
5
re
l-l
u 3
o O
u
u
JJ 3
g U
CJ
3
.3
-W
-3
3 E 3 01
•o ^ CJ >>
E o
OJ
OS
—
c
y-i
jj
£ —
u o 3
CJ
c -3
> 3 C5
re
re
-3
JJ
1—
c
U
01
t-l
jj
re
c E
o
CJ
.-I
JJ
3
3
*3
£
a
2
3
U
a
JJ
CJ
CJ
c
i—i
01
O
>N
o
JJ 'H
3 3-
E
E 0)
>. 3
O 3
CJ
.—
C
i
01
E K
01
o
O 3 —
re
JJ
3
jj
1
JJ
03
re
0)
H
JJ
re
m
CO
CO
CO
j3
J=
j3
JJ
JJ
JJ
^
C
o
E
c
o
E
C
o
E
c
o
E
o
CM
CO
<T
.-i
rsi
^
f-
I1
—
1
1
'
00
u<
—
i
1
en
o
i
o
00
sD
CO
00
as
m
rsi
-JT
>> <r
c CO
to
E
o^
—
CM
C^
LTl
rsl
in
CO
o-\
CO
CO
LTl
M
CJ
O
uj
C
0)
E
>,
o
CO
o
u rH
ci
00
rn
CM
o
.-I
C
E
01
c
E
0)
H
E
o
CO
00|
C
o
e
J2
-3-
r~
co
OS
O
CM
O-v
o
CO
-3
re
H
CJ
JJ
CJ
1-1
a
r"
c
=>
o
CM
co
as
CM
-3-
—
I
m
re
;
1
S_
CJ
O
CO
C^
U1
CO
r^
m
CM
o
r-i
u
O
—
H
O
in
I
05
o
u
o
i—
OJ
JJ
C3
M
JJ
f-
3
E
>>
O
^c
E
0)
C jj
O c
3
**
>,
H
-J
c
o
H
JJ
a
3
re
t-i
o
Ui
J=
T-i
0)
i-i
i-l
JJ
E
3
P
—
JJ
JJ
re
r-
c
E
re
"
>,
U
*o
r-i
3
c
5
E
0)
CO
c;
-C3
re
jj
oi
C
~
CJ
s
> O
re
>-,
o
-H
jj
—
3
tj -3
=S
0)
re
C-
~
CJ
£
^
o
u
jj
01
jj
~3
0)
re
CJ
>,
>-.
c^.
5
jj
=
01
Ci,
f-
~
01
CD
01
V
o
u o c J3
C
-o
—
5 ^
E a
— —o
2
r;
OJ
G
"B
_>
I
JJ
CO
re
0)
—
jj
re
m
W
w
j3
J]^
J^
re
jj
JJ
JJ
sa
c
o
e
C
o
c
O
E
c
CM
—I
CO
CO
CJ
-J
.-1
=
o
CO
51
In
that
this section,
the behavior
hysteresis.
It
is
European problem.
we
European unemployment
of
is
But
the
of
they just create
a
unfavorable shocks at least some
may equally well
be
said to be the cause of
channel
First,
for persistence,
even
of
of
the
if
which implies
both have long lasting effects.
The
which are the consequence of policy
persistent high unemployment.
is
as
yet unclear
of
of
adverse shocks which has caused high unemployment.
the next section.
Europe and
consistent with our hypothesis about
temptation must be strongly resisted.
both favorable and adverse shocks will
sequence
in
obviously tempting to conclude that unions are at the root
unions create hysteresis,
that
have shown that unions play an important role
Second,
it
whether the cause Df hysteresis in Europe is unions or the sequence
We
consider this issue
in
Ti
Eurosclerosis Really the Problem'
Is
4.
The previous section has shown that
explain important aspects
behavior
European and American labor markets.
of
leaves open
crucial
a
unemployment
question however.
consequence
a
European labor markets?
of
full
employment
first question,
compelling
if
in
Alternatively,
The case
in
hysteresis
European
in
unioniied character
hysteresis the result
is
that major structural
the second question can be given
or
and
of
while the case for expansionary macr Deconomi
of
a
of
sequence
of
reforms are needed
if
of
i
c
policies is more
positive answer.
hysteresis lies ultimately in European
of
adverse shocks that have buffetted European
the sequence of
economies requires comparisons
one of
the presence
Is
presented so far
Europe is to be restored depends on an affirmative answer to the
Resolving whether the source
institutions
The evidence
the heavily regulated
adverse shocks to employment?
persistent unemployment may
of
current European depression and the very different
the
of
model
our
the
current situation with situations where only
Comparison with the United States
these elements is present.
at
present
cannot resolve the issue because the American economy lacks institutions like those
in
Europe and has not suffered
a
like those experienced by Europe
sequence
able to sake two
But
we
are
comparisons which can shed some light on the sources
of
hysteresis.
comparison
of
the behavior
of
in
the
contractionary aggregate demand shocks
of
1
9B
'
s
The first
is
European labor markets in the recent period with their
behavior over the 1953-1963 period.
Broadly speaking,
labor market
institutions were
similar in the two periods but the pattern of shocks was very different 33
.
The
second comparison is between the current European depression and the US depression
the 1930's
a
labor market
At
the time of
the US depression,
regulations were
a
a
small
factor,
unions were weak,
and there were few
social
if
of
programs and
any important
s
33
The US Depression may also shed
adverse supply shocks.
expansionary policies
in
light
on
the role
alleviating persistent high unemployment.
of
consider these
We
compar isons in turn.
4.1
European labor markets before the current depression
The previous
behavior
of
section has examined the persistence
wages in Europe over the past 35 years.
current depression period and the period
and
We
1960's.
of
unparal
1
el
1
ed
examine the extent to which hysteresis is
presents estimates
of
of
prosperity
a
product
the
in
.
The degree of
the
of
two periods.
persistence
1
9
50
'
bad times by
of
Table 10
separately for
unemployment in
in
latter period when unemployment was high.
the
contains the
interval
the stochastic process followed by unemployment
1952-196B and 1969-19B4 periods 34
Europe is much higher
unemployment and the
This long
considering labor market behavior separately over each
the
of
Similar but
somewhat less dramatic results are obtained using employment rather than
unemployment.
For
persistent
the United States than
in
the earlier
tend to suggest that hysteresis
of
the structure of
Table
in
Table
5,
11
is
a
in
the
United Kingdom
or
These results
France.
feature of bad times rather than
a
consequence
European labor markets.
presents estimates
but
unemployment appears to be more
sample period,
now for the
of
wage change equations paralleling those reported
1953-1967 period.
Taken together the results suggest
somewhat less hysteresis in the 1953-1967 period than is present over the whole
sample period, with the difference being pronounced in the United Kingdom where the
ratio
R,
which was close to one for the full
sample is now close to
.5.
However,
results for the 1953-1967 period like those for the entire period suggest
degree of hysteresis in Europe than in the United States.
The
fact
a
the
greater
that persistence
Table 10
The Persistence of Unemployrnen t in Good and Bad Times
SE Regression
Country
France
1952-1968
.41
(1.1)
1968-1984
1.11
.81
(1.8)
.3
-.48
.4
(5.0)
(1.4)
.86
(12.3)
1.07
(5.1)
.22
.5
(.9)
.51
(1.4)
.8
.97
.5
1.0
(27.6)
(2.5)
.99
(3.8)
.9
.75
(1.6)
.59
(1.7)
-.37
1.0
(-.7)
.50
(1.1)
1.1
Germany
1952-1968
1968-1984
United Kingdom
1952-1968
1968-1984
.01
(.0)
United States
1952-1968
1968-1984
Kote
The results represent estimates of ARMA (1,1) process for the
unemployment rate.
.
'
I
o
o
U-l
s
r-.
(SI
in
WO
o
CO
cr>
CM
CM
CM
<
,-^
vj y—.
>J -1 sj
H
^-s
•^ — m
O
O O .H
i
•
•
^-*\
r~-
co
-J
.
O
0)
O
m
C
vO
CO
o
u-l
o
^^
o>
m
•
r~-
cr.
s?
O
I
I
o
HE O
!
.—<
rH
I
I
vd in
O
\o
.-(
•
•
•
I
-
-a
I
l
I
o
o
in
co
•
in
r~•
n
•»->
m
in
Oi
co
cs
•
cm
-c •—
•
•
•
•
n
m
co
•
r-l
•
•
CM
•
CM
i-H
W
N Oi CO
OS
n m n
CM
•
•
•
.
i
•
«T
o
m
o o n
co r^ in
• CN
•C
^
o<
r~
CM
•
i
I
•
<3"
I
^
o
|
H
o
I
m
•*:
o
-—'
~9 >-'
co
m
CM
en
•
i—
•
nn
r-»
m
m
co
cm
m
.
—
w w
I
vO
«
I
|
m
*
r^
E
O
-o
DO
c
>s
u
iJ
e
3
>1
c
C3
E
l-i
o
6)
CJ
CJ
a
CJ
r^ r^ cm
•
•
in
cm
•
m
f
m o
cm
<
•
-•
—
r-»
I
54
is
present in the early period in Europe to
States but that
it
balance,
greater degree than in the United
becomes increasingly important as the unemployment rate increases
makes it difficult to draw any
On
«
-firm
conclusion about its causes.
evidence on the changing behavior
that bad times as well
unions account for bindings
as
evidence is not sufficiently powerful to permit
i
European labor markets suggests
of
of
hysteresis.
But
this
judgement about their relative
a
mportance.
4.2
A
Tale of Two Deoressions
Salient features
of
many discussions of the current European depression include
pessimistic forecasts that unemployment will never return to earlier levels,
that reduced investment and lower capital
the entire labor
force,
stocks have made
concern
impossible to employ
it
and fears that expansionary policies will
lead directly into
inflation with little or no favorable impact on output or employment.
These
pessimistic views are premised on the conviction that structural problems are central
to high
unemployment
in
Europe,
sequence
of
adverse shocks.
go beyond
a
was ended by the expansion
in
and that
the causes of
Yet
persistent high unemployment
the American depression of
aggregate demand associated with rearmarment
Unemployment recovered to pre-Depression levels.
Are this experience and the current European experience
permit the inference that hysteresis arises from
1
930
'
s
.
Recovery was not inhibited by an
insufficient capital stock or by the overly rapid adjustment
than from structural
the
a
problems in the labor market?
sufficiently
sequence
Or
wages and prices.
of
of
comparable to
adverse shocks rather
do major
differences in the
character of the American and European depressions render the American experience
irrelevant for thinking about current European problems?
55
We
19
4
5
period.
began
the
1925-
economic statistic! tre presented in Table
12.
The
the period
of
is
of
course the dramatic upsurge in unemployment
of
Unemployment rose from levels comparable to those experienced
1929.
in
basic
number of
A
outstanding feature
that
the American economy over
begin by briefly reviewing the record
in
Europe in the late 1960's and early 1970's to 25 percent in 1933 and remained above
14
percent until
year
1940.
period despite
declaration
War
of
As
rapidly increasing population.
a
stimulated the economy.
Any
in
November
increased by
variety
The benefits of
1940 and 2.1
6
1941
Beginning in late 1939 with the
While there were only B22 thousand men in the
the economy.
percent or
million
year later,
a
non-agricultural
million persons between 1939 and 1941.
Mitchell
(1947)
Production
full
of
a
reports that between
industrial production rose by 20 percent.
Overall
These rapid improvements in economic performance were unexpected.
the
employment
automobile sales rose by 35 percent, refrigerators by 69 percent and
washing machines by 63 percent.
wake of
10
increased defense spending spilled over
non-defense goods increased rapidly.
of
1939 and
16
of
of
a
unemployment began to decline rapidly as rearmarment
Europe,
in
widely into the rest
Europe today employment actually declined over
in
1937 recession many observers had
employment.
Paul
Samuelson noted
there were noticeable signs of
in
despaired
1944 that
"in
of
Indeed in the
any eventual
just prior to 1939
the years
dwindling interest in the problem
return to
of
unemployment
which took the form of ostrich-like attempts to think away the very fact of
unemployment by recourse to bad arithmetic and doubtful
even among economists there was
income to 1929 levels."
even
in
"it
is
increased emphasis on the recovery
techniques.
of
And
production and
Such pessimism was pervasive even among those charged with
alleviating the situation.
1937 that
statistical
Harry Hopkins
reasonable to expect
future prosperity periods."
a
a
liberal confidante
probable minimum
(Leucht enber g
(1963)
of
p.
4
of
Roosevelt wrote in
to 5 Billion
263).
unemployed
Similar sentiments
2
Table 1?
The American Economy
1925-1945
w
(all workers)
p(CPI)
Index of
Productivity
Non-Residential
Capital (1958$)
Year
U
1925
3.2
.9
4.0
92.6
211.0
1926
1.8
1.5
0.0
95.0
218.7
1927
3.3
3.2
-6.0
95.4
223.9
1928
4.2
.3
-1.0
96.1
-229.3
1929
3.2
3.5
-1.0
100.0
236.6
1930
8.9
-0.6
-3.0
97.0
238.8
1931
16.3
-5.0
-8.3
98.5
233.5
1932
24.1
-8.9
-9.0
95.4
222.8
1933
25.2
-5.8
-5.0
93.2
212.2
1934
22.0
12.0
2.6
103.3
203.9
1935
20.3
2.3
2.6
106.7
198.3
1936
17.0
1.9
1.2
111.3
197.0
1937
14.3
5.9
3.7
110.4
198.4
1938
19.1
1.8
-2.4
113.5
194.5
1939
17.2
1.2
-1.2
117.6
192
1940
14.6
2.4
1.2
122.2
193.6
1941
9.9
9.7
4.9
124.2
198.3
1942
4.7
26.9
10.5
123.3
193.5
1943
1.9
10.6
6.3
124.6
186.5
1944
1.2
7.8
2.0
134.4
183.0
1945
1.9
9.0
1.9
142.0
185.5
Source.
Baily (1983) and Historical Statistics.
.
56
were echoed by others including LaGuardia who concluded that the situation had passed
from being an emergency to being the new norm,
Similar pessimism is often expressed in Europe today.
the view that
unemployment
The pessimism reflects
unlike the US depression's persistent unemployment,
in
adverse shocks.
Europe is caused by structural
H.
problems not merely the residue
Giersch has coined and popularized the word
denote these structural
problems.
persistent
"eurosclerosis" to
there some important difference between the two
Is
situations which suggests that rapid expansionary policies would fail
even where they succeeded so spectacularly in the United States in
surprisingly many similarities between the two experiences.
and real
Indeed,
problems are structural
In
the latter half
of
is
it
and that
1940?
Europe today
There are
of
inflation
current
the
this observation that drives conclusions that
equilibrium rate
the
the Depression,
a
of
unemployment has increased.
similar pattern appears in the United States.
Between 1936 and 1940 unemployment fluctuated around
essentially no deceleration
in
close to the normal
productivity growth.
rate of
in
The failure of
wages to recede nore rapidly is an often noticed aspect
European experience.
of
inflation and real
very high mean but there was
a
wages rose by about
Previous to the
1
930
10
'
s
percent,
periods
of
steady inflation had had much lower average unemployment rates.
Oust as unemployment
in
Europe is highly persistent today,
persistent during the American Depression.
.874 in the United States over the
the
1920-1941 period.
The autocorrelation of
1919-1941
hysteresis during the Depression, Table
13
period.
To
because
When only contemporaneous employment or unemployment
in
unemployment was
of
presents some estimated wage equations for
The War years are omitted
insignificant, but the change
appeared highly
further examine the issue
of
the
The results dramatically suggest hysteresis parallelling that
is
it
is
influence of controls.
found in Europe today.
entered into the equation,
employment or unemployment
is
it
strongly associated
Table 13
Wage Equations and the American Depression
U
U
t-1
IorE.
lo & E
t-i
-.06
(1)
(.2)
(2)
(3)
-1.13
1.26
(2.9)
(3.2)
.67
.22
(.8)
0.0
1.71
.50
(2.1)
.29
2.13
.24
.03
1.75
2.71
-2.72
(.74)
(3.2)
.36
1.99
.38
(1.7)
Note
The dependent variable is the rate of wage inflation.
Historical Statistics of the United States.
.
DW
(.9)
(.50)
(A)
"t-1
Data drawn from
57
with changes in the rate of
of
These results are robust to
This may be taken as evidence that hysteresis is
with bad times rather
than with particular
labor
market
importance
of
long term unemployment
presents some
of
the
phenomenon associated
institutions.
There is little evidence
of
an
Again,
As
in
the results parallel
Europe today the duration
Philadelphia survey which found that 61.7 percent
era saw the emergence of
a
^o drawn from Woytinsky
Europe today.
year.
of
More generally,
unemployment
a
a
1937
unemployed adult men had been
he
concludes that the Depression
new group of hard core unemployed.
market turnover do not appear to provide
of
Woytinsky reports evidence from
appears to have increased substantially.
a
e
increase in the flow rate into unemployment, though
quits decline and layoffs increase.
work for more than
Tabl
if
limited evidence available on patterns of labor market
turnover during the American depression.
of
a
emphasizing that turnover rates were
anything lower in Europe than in the United States.
out
variety
considering contemporary European labor markets, we laid considerable stress
In
(1942)
a
these results differ from our results using American data for the Post-
war period.
the
.
While parallelling our results for present
treating expected inflation.
ways of
day Europe,
on
wage inflation 35
Patterns in labor
basis for distinguishing European labor
markets and American labor markets during the Depression.
Hysteresis appears to be an important feature
in
the paper,
we have
physical
capital
the real
value of
and
of
suggested three possible sources
American Depression.
of
hysteresis.
accumulation appears an unlikely culprit.
the non-residential
capital
stock actually declined between 192?
during or after the War when demand for goods was strong.
However,
of
it
claims that insufficient capital
should be noted that Mitchell
these
Table 12 demonstrates,
This reduction did not represent an important bar to full
1939.
skeptical
As
Of
Earlier
is
(1947)
holding up
employment
This aakes us somewhat
a
European recovery.
claims that capacity utilization
U
Table
Labor Market Turnover and the American Depression
Extent or La»or Turnover riou 1919 to 1929
(Median monthly rale* per 100 workrn)
Vm
1
Sep«r»uoru
Acc««ionj
Ten'
Quit?
1
L.yofl.
E>l*cUAr-*e
1
1919
1920
1921
1922
10.1
10.1
2.7
7.5
10.3
g.O
9.0
5.3
7.5
6
2
3.3
5.2
3.8
2
7
4
4.6
3.3
3.7
4.4
39
3.1
2.9
2.1
1923
1924
1925
1926
1927
1928
1929
i~rtr UcmiU, Lmtor
5
2.2
4.2
4
4
3.3
3.1
3.8
f.
1.1
1.1
8
8 4
2
2
2.7
Jn*r
im.
Year
1
r.ntes
0.6
0.8
4
1.8
0.7
1.0
0.5
O.S
0.5
0.5
0.4
0.5
0.4
0.3
0.6
0.4
O.S
0.7
0.5
0.6
PP M. *J; Frbrnmry 1«J1.
Extent of Labor Turnover from
(Average monthly
.
p.
I
1930 to 1940, by
OS.
Years
per 100 workers)
auons
Di'rharc-
Accession*
Tola!
Median
Lavcfi<
rales
j
1929
1930
4.4
1.6
3.6
2.4
1930
3.1
3.1
3.4
5.4
4.7
4.2
4.3
3.5
3.S
4.1
3 4
5.0
i
2.7
j
1.1
V 'lighted
1931
1932
1933
1934
1935
1936
1937
193S
1939
1940
.6
1.0
0.7
4.3
3.S
4.1
3.6
3.4
4.4
0.9
9
0.9
1.1
1.2
4.1
3.1
!
3.35
!
6
1.2
average rates
1
4.1
0.5
0.2
i
|
0.6
o.s
1.0
0.4
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.1
0.1
0.15
3.0
2
9
3.4
2.7
3.0
2.5
2.1
3.0
3 4
2.2
2
2
Including miscellaneous reparations because of death, retirement on pension, etc.,
reported KparaLely uccc January 1940.
•
Sovlc: Monthly L&bor
1931 to 1939.
Source.
Woytinsky (1942)
Rmrv.
»«
ibid.,
19X> u> 1941. For a
lummiry
September 19+0. pp. 690-7CW.
of labor
turnover from
56
were very low prior
There
ic
labor
force participation rate of
1940 censuses 3t
men over
1930,
it
fell
could have much effect on unemployment.
unemp oyment
1
only
by
It
1
percent,
and
it
remained
seems unlikely however that this
Indeed to the extent that marginal
labor force,
workers
bad times might have reduced subsequent
.
This leaves our
importance
the
of
dropped from 54 to 42 percent between
65
essentially constant between 1940 and 1950.
were induced to drop out
The
This is considerably more rapid than its trend rate of
.
Between 1920 and
decline.
hysteresis in labor force participation,
human capital
some evidence of
1930 and
the
the
to
This it not true in Europe today.
1939 expansion.
rites,
of
i
nsi der-Dut si der
story
of
wage setting.
Beyond documenting the
hysteresis, and confirming its implications for wage equations,
difficult to test the story directly.
perhaps revealing,
But
the
"By Roosevelt's second term,
judgement
as
it
of
Leuchtenberg
ii
it
(1963)
is
seemed the country night never
the unemployed had become too exhausting
wholly recover, the burden
of
economic weight to carry.
Those who drew income from other sources could hardly help
but feel
unsaved.
that
the Depression had been
Increasingly,
the
a
a
r,
oral
judgement which divided the saved from the
jobless seemed net merely worthless mendicants but
nenacing Lurepenproletariat."
and
a
While Leuchtenberg is referring primarily to public
attitudes towards the unemployed, similar private attitudes are the driving force
behind the hysteresis mechanism we have stressed.
The finding of
so many parallels
between the current European depression and the
American depression suggests to us that hysteresis in Europe may be more the result
of
a
long sequence of
adverse shocks than the result
of
structural problems.
Perhaps most telling is the observation that the apparent natural rate
of
unemployment drifted upwards following the actual unemployment rate during the
American depression just as
it
has in Europe.
Given the absence
of
structural
59
explanations for this drift,
the
unemployment seems compelling.
unemployment may be the result
of
adverse shocks,
policies may well
As
we
inference that
So
of
too,
it
the high
resulted <ron high past
apparent European natural
hysteresis arising
discuss below,
this
in
the
aftermath
of
a
rate
of
sequence
implies that expansionary macroeconomic
work in reducing unemployment
in
Europe,
fcO
5.
Conclusions
Periods
persistently high unemployment are not uncommon events
of
historical
context.
accounting
•for
theories
standard macroeconomi
Yet
them.
have argued that
We
c
theories have
argued that membership effects may well
important sources
be
effects appear to be an important source
difficult time
a
they can only be understood in terms of
hysteresis that make long run equilibrium depend
of
broad
in
persistence
of
on
history.
of
hysteresis.
And we have
Such
unemployment in Europe
in
today.
High unemployment
however always persistent.
not
is
A
crucial
issue is
identifying the circumstances under which persistence is likely to arise.
whether hysteresis is the result
issue is that of
structures,
of
the presence of
unions
outsider dynamics we have discussed
historical
record,
is
on
fully satisfactory theory
theory permits
surprises in the form
of
increases in tax rates.
a
of
the decrease
the
1
9B
1
itself
is
the
'
s
,
We
'
s
from the
bad times and
in
are
have not provided however
a
membership effects in non-union settings.
In
rising oil
increase
in
unemployment in
1570's European economies were hit with
the
the productivity slowdown,
prices,
With wages rigid
Because
the
of
in
the short run each of
the membership
the equilibrium level
of
and
rapid
these types of
considerations stressed here,
employment was validated by higher wage demands.
in
97
unions.
broad brush account of
shocks created unemployment.
the
it
Our tentative conclusion,
paper.
the
in
the presence of
Europe over the past 15 years.
In
particular, or whether
in
that membership effects become important
not crucially dependent
end of
specific labor market
adverse shocks, which by increasing unemployment, trigger the insider-
result of
Dur
of
The main
As
a
result by the
unemployment had increased substantially.
the European economies unlike the US economy experienced
a
series
of
adverse aggregate demand shocks as European monetary policies followed US policies,
61
liver
but
aggregate demand shocks
se
the power
the past
This led to further unemployment which
validated by wage demands by those who remained employed.
unemployment will remain high even
Our
European monetary policies followed US policies,
policies turned contractionary.
fiscal
was then
as
of
employment.
this point,
there are no more adverse shocks,
if
because
of
insider workers to set wages.
argument
15
At
is
that Europe has experienced
years each
of
which had
a
adverse shocks during
sequence of
fairly permanent effects on the level
a
Current high unemployment can equally be blamed on
mechanism which leads the adverse shocks
the shocks themselves.
of
the
to have
propagation
a
lasting impact,
a
on
or
Unlike simple Keynes an explanations for the European
i
depression which stress only aggregate demand,
apparent natural
past
of
rate of unemployment.
our
theory explains increases in the
Unlike some classical
explanations for
European unemployment which deny any role for demand management policies, our theory
explains how aggregate demand can have protracted effects even in the absence
long lasting nominal
of
any
rigidities.
This view of the European unemployment problem
policy implications.
of
A
first policy implication of
our
has
a
number
analysis
using measures to "enfranchise" as many workers as possible.
is
If
of
the
fairly direct
desirability
worksharing
programs cause more workers to be employed and therefore represented in wage setting
decisions,
they may lead to reduced wage demands and increased employment.
sharing plans such as those proposed by Weitzman
(1955)
Profit
may also raise employment by
Raking it possible for employers to reduce the cost of labor by increasing hiring.
On
the other hand
they would increase unions'
eight thereby increase membership problems.
measures to reduce the power
of
larger impact on wage bargains,
resistance to hiring new workers and
An
obvious alternative policy is
unions and thereby allow outsider workers to have
Our
findings regarding the US depression where
i
62
unions were probably not
efficacy
of
the power
Certainly
such measures.
of
importance
great
of
it
lead
uc
somewhat skeptical
to be
of
does not yet appear that efforts to reduce
unions in the United Kingdom have borne -macroeconomic- fruit.
Our model
suggests that shocks,
employment changes,
positive
or
negative are
in
wage setting practices adopt
a
sense self
validating.
If
employment.
This means that positive shocks contrived through demand management
policies can reduce unemployment regardless
it.
Even
if
the source of
of
to
the new
level
of
the shocks which caused
unemployment initiallly originated from adverse productivity shocks,
expansionary policies,
permanent benefits.
if
they succeed in raising the level
Symmetrically,
the 19B0's is due to demand,
permanently.
it
the
At
the
even if
the large decrease in
same time the model
which are in some sense surprises will
difficult to increase employment
question becomes the length
To whatever
wage setters.
most of
of
a
be
the
of
employment will
increase in unemployment in
the price of
oil
may well
decrease
suggests that only policies or shocks
This means that
efficacious.
great deal
may be
it
with expansionary policies.
time over which expansionary policies can
extent they can,
yield
The crucial
"surprise"
very long lasting benefits will
be
derived.
Do the many parallels
a
between the American and European depressions
major expansion in aggregate demand would create the same miracles in Europe as it
did
in
lead to
the United States?
a
Unfortunately comparison
very definite answer.
increase in the natural rate
of
of
While it does dispose
the two depressions cannot
of
the
idea that the apparent
unemployment means that demand expansion cannot
possibly succeed, and the idea that real wage growth must be restrained
is
imply that
to take place,
surprise for
a
an
important problem remains.
The likelihood of
a
decade including
a
expansion
achieving
protracted period through inflationary policies cay well
much greater in the United States after
if
a
have been
major deflation than it
63
is
in
Europe today titer
political
i
nf eas
i
b
i
1
i
t
y
a
of
decade
of
expansion
stagflation.
in
On
the other
hand,
the
very
Europe suggests its possible efficacy.
Certainly the protracted high unemployment caused by the deflationary policies
recent past stands as
a
testament to the potent effects
of
of
macroeconomic policies.
the
.
bA
Footnotes
Formally,
1
one eigenvalue
equal
Thus,
we
zero
to
(unity,
the system will
the steady state of
system.
system is said to exhibit hysteresis
dynamic
a
if
specified
in
discrete time).
depend on the history
of
has
it
if
such
In
case,
shocks affecting the
the
past
Dn
values with coefficients summing to one.
shall
We
instead use "hysteresis" more loosely to refer to the case where the degree
on
a
should say that unemployment exhibits hysteresis when current
unemployment depends
dependence
least
at
the past
very high,
is
of
where the sum of coefficients is close but not
necessarily equal to one.
For
2
the United
calculated by Romer
(19B6)
for
of
the revised unemployment
European countries
Economica,
1
9S 6
The
.
work presented for
the Chelwood Gate Conference on Unemployment,
at
rates
1690-1929 period.
the
This part relies heavily on the empirical
3
in
States we made use
reader is refered to individual
individual
to be
published
country papers for further
evidence.
4
We focus
the UK because detailed data are more easily available.
on
Available data for France and Germany tell
very similar story.
index
by
industry goes up however in 3961 and 1982 --which are
last two years for which
it
has been computed--.
5
the
a
6
wedge.
capital
7
The mismatch
Let
be
a
the
Then the rate
of
rate of
growth
growth
of
the
Df
productivity and
after tax real
labor ratio is approximately given by
When
a
estimated above,
time trend
is
is
both small
be
the change
wage consistent with
in
a
the
tax
given
a - $
added to the AR(1)
its coefficient
6
specification
of
and insignificant,
unemployment
for
both countries.
65
This is also the direction of
B
research recently •followed by Sachs
I19B5)
to
explain European unemployment.
Unemployment remained high --around
9
other factors are thought to be
but
Drazen
10
at
constructs
(1979)
which also generates hysteresis,
work
10'/.
--in
Italy until
1960 approximately
that case.
in
related model, based on learning by doing,
a
Hall
explores the possibility that
(1976)
unemployment has long lasting effects on productivity, and its implications for
economic policy.
The issue of
11
section
for
6)
dynamic implications
a
of
survey.
literature has not however focused on the
This
as
passive allows to concentrate on the effects
alternative membership rules on the decisions
ex
of
the group of
insiders.
post by insiders introduces additional
of
Allowing for
insiders as well as for some control
wage bargaining between the firm and
employment
See Farber
membership rules.
Formalizing the firm
12
clearly closely related to
is
union size and union membership in the union literature.
the issue of
(1984,
membership and membership rules
issues which we shall
of
discuss
later,
The assumption of
13
not realistic for
as
Because we use
negative.
values
of
We
membership,
mpl
It
i
cati ons
a
log
is
2
not
best
to think
m2
later.
is
as
of
the firm under
consideration
other firms.
approximation to define
only acceptable for
p
p,
p*
as
defined can
close to one,
that is for
too large.
may also think of
and
linear
the approximation
But
a(b/c)
15
i
single firm.
representative firm, facing the same shocks
a
14
be
a
stochastically inelastic labor supply maintained here is
assymetric rules where
periods to lose it.
We
shall
it
takes
su
periods to acquire
briefly return to their likely
66
There is mother
16
high real
that
This effect
wages in the future.
opposite direction.
thus
Farber
(
1
9B4
)
a
lower membership and higher
however turns out to be dominated by
reviews the research on union behavior when members have
different seniority status,
we
Thus,
IB
and
thus conflicting
interests.
assume implicitly that the technological
the relation between
output and employment.
output is produced with two inputs,
Leontieff technology,
and the the
labor and
a
technological
relative price in the non labor input.
A
Allowing the technological
the model
is
non labor
and
straightforward but introduces ambiguities
the
to our
staggering
the union wage differential
supply
the effects of
in
argument.
changes which have decreased the legal
si2e of
a
between prices and wages,
of
wage decisions across
shocks.
There is some evidence that this has actually occurred
Despite the legal
according to
shock reflects changes in the
unions would lead to effects of even anticipated nominal
decade,
input,
shock to affect the relation betwen output and employment
Like in other contracting models,
20
but
change in productivity growth would instead
shocks on employment which are not central
19
shock affects costs,
This is the case for example if
affect both the relation between output and employment,
in
Choosing
emphasized in the text.
17
not
the
in
leads to lower expected employment,
wage
expected real
effect which works
power of
See Taylor
in
[1579],
Britain.
unions in the last
appears to have risen sharply in
recent years.
21
Consider
simple example.
a
Suppose restaurant wages were rigid, and
a
big
decline in the demand for restaurant meals took place so there were unemployed chefs.
Would it pay to open
costs were high.
a
new restaurant
with
a
low paid chef?
Probably not
if
fixed
These considerations may have something to do with why in bad times
employment growth may be concentrated in small establishments.
67
Given that our paper
22
unions in the US
the role of
Europe,
in
unions in the US play
Actual
23
In
forward here.
much more
will
As
clear
be
There are
a
number
important differences
as
of
extension is considered to be
of
these agreements,
labor market
persistent unemployment put
similarities between Japaneese and European
of
of
company level
particularly
well,
of
limited role than in Europe.
view to evaluating the theories of
a
from our description
would be valuable to study Japanese
it
institutions including the importance
be
American audience, we do not review
firms respect the content
future research,
institutions with
any detail.
an
extensions are rare but the threat
very effective in making all
24
a
written for
is
in
the
bargaining.
attitude
of
There may however
Japanese unions
towards outsiders.
Two recent cases have been
25
has gone non unon union for
some of
Murdoch who has in effect gone to
appears to be due
that
of
British Petroleum which
shipping operations,
and
that
Robert
of
more accomodating union.
desire to increase membership figures,
part to their
in
Allowing labor demand to depend on current and expected real
should under costs
adjustment,
of
and through
union movement.
their role in the national
27
up
its
the news,
The reason why unions encourage the unemployed to remain in the union
26
these,
a
in
would complicate our task here.
wages,
as
it
John Kennan takes
this issue in his comments on our paper.
Note that
28
infinity.
linear,
29
As
between
and
1
does not correspond exactly to
argued before,
m
between
a
we have
1
and » leads to
a
n
between
more complex,
1
and
non
specification.
This is
a
plausible and convenient assumption.
that productivity was the sum of
a
Suppose we assumed instead
linear function of observable variables and
stationary or borderline stationary process,
say an AR(l)
a
process with coefficient
p.
.
66
would then differ <ron that
The wage equation
would be the presence
the presence
the
of
productivity.
first
differences
results were obtained using total
-first
second would be
observable variables affecting
the
in
specifications empirically for the
the results reported
the argument
•following
(when available)
-1
value
The
the previous
of
current and lagged values of the capi tal
productivity growth
use manufacturing
we
Very similar
here.
employment.
these findings kre quite robust.
All
31
The
The
simple wage equation not to be misleading.
employment as the employment variable
•for
two wiyi,
Because our wage data refers to manufacturing wages,
30
same if,
of
in
with coefficient p-1.
wages,
lagged real
have explored these more general
We
found our
UK and
p
of
the text
in
footnote,
abor ratio,
are added
are also robust to changes in the coefficient
on
to
of
the
is
R
substantively the
lagged real
the price of
oil,
the regressions.
lagged inflation,
wage,
and
a
proxy
The results
say within
.2
of
the values used in the table.
The motivation for calculations of
32
Summers
(1979).
In
from unemployment
performing the calculations,
is
it
to
up
even more clearly.
33
•from
Some of
It
unemployment
35
duration dependent.
A
is
in
rigidities
the institutional
policies introduced
social
34
not
the estimated concentration of
decline,
this type is laid out
in
the
If
1960's and
more realistically,
in
long spells would
1
97
'
and such
a
drastic increase in
estimation cannot be very precise.
similar finding is emphasized by Gordon and Wilcox
emphasizes the importance
of
show
s
(1979)
provide evidence that it holds for Europe during the Depression period.
(19S3)
allowed for
we
European labor markets date however
clear that with such short samples,
the second subsample,
Clark and
have assumed that the exit rate
we
unemployment
of
in
the rate of
who also
Gordon
change effect in the Phillips curve
69
during the Depression period in both the United States and UK but finds the level
effect to be dominant outside
36
This dropoff
Security to some extent.
of
this
may reflect
interval.
the
effects
The program was
unlikely to be the whole story.
Moreover,
something to do with the fact
the
of
of
the
introduction
sufficiently small
the
timing
Depression.
of
its
in
of
1940,
Social
that
this
is
introduction surely had
1
70
Appendix to Section
II
l)Derivation
probability
For
of
the
given realisation
a
empl Dyment
is
gi
ven by
of
thus for
e,
i
given n«- cw
+
1
n
>
no
,
or
equivalently for
e
>
no*
cw
,
then
p
If
n
<
n
,
or
equivalently for
e
<
n o+
cw
,
then
p
implies that,
and support
f(e),
for
te~e~3,
arbitrary distribution
an
the probability is
no+cw
p
=
-cw + e)f (e)de
+
/
e"
If,
assumed in the text,
as
p
(l/2a){
1
-
(
e,
N/No
*
1
-n o+n
with density function
!
[
(
1
is
+
of
2
the solution
uniform on [Ee-a,Ee+a3,
p
becomes
•:
cw
l-no-cw e/2) e]
Ee-a
+
/Aa) (no + cw-Ee + a)
2)Derivation
We
given by
of
=
f(e)de
no+cw
e
n
=
the probability of
e*
(l-n
;
e,
i
If
This
time,
being employed.
Df
+
(Ee
+a-no-cw))
for
no+cw
<
Ee-a
for
no + cw
>
Ee-a
in
the case when m=l
first derive the objective function maximised by the union at any point in
71
We
assume that,
dH,
laid
if
we
text,
the
equal
to
zero.
equal
to
zero. .Let
be,
as
in
{or
member of the union
at
time
a
As
in
p,
depends on employment
in
time zero to still
a
the utility of
Uo
or,
Uo
=
po
+
bw
+
i.
Then,
+
as
(pi+bw,)
ep
poEo(6Ui
+
e
r
it.
it
is
being employed at time
the probability for
i
is
2
p
i
given the membership rule that membership
period
of
f
being unemployed
of
the probability of
the text,
union member in period
recursive form by
in
assume that the utility
the previous period,
the union member
(po + bwo)
Eo(
=
be
being rehired by the
the probability of
p
.
is
given by Eo
given by
(pu+bHi)
+
union member at
a
p
(p
>
.
•
.
p» -
1
)
.
Thus,
:
...)
:
)
Even under the assumption thit the shocks are independently distributed through
the random variables within
time,
the expectation operator
independent, making the maximisation problem intractable.
above are not in general
Thus,
solve instead the
we
problem associated with the objective function linearized around some p',
linearized objective function is given by the following recursion
V
=
dpo
bwo
+
A
*
-6p'(p'+bw)'/(l-6p')
d
e
1
=
(i
where
)
+
Slp'+bw'Ml
+
+
be*')/ d-ep
probability
of
This
is
i
+
6p'
and
+
6
: p' 2
+
...)
')
The weight put on the probability of
the previous case.
This
ep'EoVi
(A
+
w',
because
pQ
being employed,
po
,
is
now higher than
affects not only today's outcome but the
union membership and employment
in
the future.
in
i
i
:
.
72
now derive the solution to the maximisation using the linearized objective
We
Under the assumption that shocks to labor demand are independent
•function.
uniformly distributed on
is
[
-a
,
+ a
3
the solution to the
,
linearized maximisation problem
derived as 4011 ows
We
first guess that
Vo
=
given
and
o
Vo
If
the maximised
Vo "
p
e-pn-i, then EoVi
=
(A +
ep'o)
4
(b +
to
=
Replacing
«=
(
1
/ c
in
p c
)
C
-
n_o
(
a2
-a
4
This in turn gives
1-
=
a
(
)
p
=
the form
is
(b/c)
given by
/
(1-Gp
n
=
o+pcw
')
.
w
Replacing in the recursive form
dp D
+
(a2)
:
2
for
optimal
wo
gives
:
(a3)
:
2
(a4)
.
o
and
gives the values of « and
p
and
p Q
and p.
and po as functions of
w
)
then solve for optimal
We
2a(b + 6p'pc)/dc3
now solve for the values of
value of
a
B-pE
and solving
(b46p 'pc)/dc)
This gives us
(al)
of
:
6p 'pc) wo
l-(l/4a)(n_ 4cwo + a)
po =
determined.
be
solve for
finally
and
p,
and
a
The probability po is given by
po
is
V
(al
which characterize* Uo gives
wo
value
c-pn-
with coefficients
and
and
:
p.
p,
structural
Replacing
w
The value of
parameters and
and po in
a
is
of
(a2)
no
of
a
and
and comparing
interest here.
p.
(a2)
The
We
References
Barker, A.,
P
tt
.
flcCann,
British Journal
the Unemployed,"
Blancherd,
Lewis and
.
and
0.
L.
Bummers,
9
8
4,
B4
,
Basil
,
November
Bulow,
to
J
.
Sachs,
19 65,
3,
3 91 - A
04
Interest
273-334.
2,
"The Economics
of
O'f
Worldwide Stagflation," Oxford,
Blackwell
Buiter,
1
and
M.
Relations,
"Perspectives on High World Real
Rates," Brookings Papers on Economic Activity,
Bruno,
"Trades Unions and the Organisation
Industrial
oi
1
19
W.
o-f
J,
and
"A
Guide to Public Sector Debt and Deficits," Economic Policy,
L.
Summers,
!.
9B&
,
"A
Theory
oi
Dual
Labor Markets with Application
forthcoming
Policy, Discrimination and Keynesian Unemployment,"
Labor
Campbell,
Harvard
19E5,
13-60
,
Industrial
Journal
,
J.
Economics
and
G.
ttankiw,
19E5,
"Are
Output Fluctuations Transitory?",
mimeo
,
Clark,
K
.
and
Reconsideration,"
Clark,
K.
and
L,
Summers,
1979,
"
Labor
Market Dynamics and Unemployment,
Brookings Papers on Economic Activity,
L.
Persistence," Review
Summers,
of
19E2,
1,
13-60.
"Labour Force Participation!
Economic Studies 49
(159),
E25-344,
Timing and
A
74
Coe,
D.
Countries",
and
Community",
Wage Determination in ten DECD
"Nominal
1965,
Economics and Statistics Department Working Paper 19, March
D£ CD
Dornbusch,
Gagliardi,
F.
et
R.
Center
al,
for
19B3,
"
Macr oeconomi
Prospects and Policies
c
European Policy Studies Discussion Paper
reprinted in "Restoring Europe's Prosperity",
eds,
Cambridge MIT Press,
Drazen,
1979,
A.,
"On
Blemishes?"
D
.
Permanent Effects
H
3
,
of
R.
,
1
.
,
.
984
,
e
1
1
e s
Dornbusch and
D.
Wise eds,
"The Analysis of
Transitory Phenomena in
Permanent Scars
"Teenage Unemployment!
Freeman and
R.
x
,
R.
Layard
a
Simple
25-30
"The Youth Labor Market Problem
in
Consequences",
Farber,
19B2,
,
Blanchard,
Br u
1-32
1966,
Growth Model", Economics Letters
Ellwood,
D.
1,
the European
of
:
its
Nature,
or
Temporary
Causes and
Chicago Press-N2ER
Union Behavior",
MIT working paper 355,
November
Fischer,
S
1977,
"Long Term Contracts,
Money Supply Rule," Journal
cordon,
R,
,
1983,
"A
of
Political
Economy,
J,
Expectations, and the Optimal
E5-1,
February,
191-205.
century of Evidence on Wage and Price Stickiness in the
United States, the United Kingdom and Japan",
Quantities",
Rational
in
Tobin ed, Brookings Institution,
"Macroeconomics, Prices and
85-120
,
75
Gordon,
R
and
.
Wilcox,
J.
Evaluation and
"Monetarist
19B1,
Critique",
Depression,
An
Brunner ed,
Martinus Nijhoff, 49-207
Gregory,
R.,
1
9B5
,
a
D.
"The OECD Data Set,"
19B4,
,
"The Great
"Wages Policy and Unemployment
Chelwood Gate Conference on Unemployment,
Grubb,
in
Interpretations
in
of
the Great
Depression Revisited",
Australia," presented
K.
at
the
Chelwood Gate
L5E
Center
•for
Labour Economics working
paper 615
Hall,
R
"The Phillips Curve and Macroeconomic Policy",
1976,
,
.
Conference on Economic Policy, Vol
Margraves Heap, S.P.,
1980,
1
,
Carnegie Rochester
127—146
"Choosing the Wrong Natural
Rate,
Accelerating
Inflation or Decelerating Unemployment and Growth," Economic Journal,
90,
September,
611-620
Harris,
J
.
and
M
.
Todaro,
1970,
"Migration,
Sector Analysis," American Economic Review,
Kaufman,
R.
,
1934,
"On
Unemployment and Development:
March,
60-1,
126-143,
Britain's Competitive Sector" British Journal
Industrial Relations," March
of
A
Two
76
Layard,
Nickel]
S.
ft.,
and
Jackman,
R.
Labour Economics Discussion paper
Layard,
and
R.
Nickel
S.
1,
1984,
i
or
August
165,
"Unemployment
19B5,
Vacancies," LSE Center
"On
in
Britain," presented at the
Chelwood Bate Conference on Unemployment,
Layard,
et
R.
a
1
1
,
9B4
,
"Europe
the Case
i
<or
Unsustainable Growth", Center for
European Policy Studies, B/9, Hay, reprinted in "Restoring Europe's Prosperity",
Blanchard,
Dornbusch and
R.
Leuchtenburg,
W
.
1963,
,
R.
Layard eds,
"Franklin
D
.
KIT Press,
19B6,
D.
33-94
Roosevelt and the new deal," Harper
&
Row,
New York
Lindbeck
A.
and
D.
Snower,
S9B5,
"Wage setting,
Relations" Institute tot International
McDonald,
and
I.
R.
Solow,
1
Economic Review, December, 71-5,
Min-ford,
P.,
Mitchell,
Holt,
Rinehart
E.
1952,
,
1947,
9B
1
,
Unemployment and
1
nsi ber-Dut si der
Economic Studies, Stockholm, WP 344, December
"Wage Bargaining and Unemployment," American
B96-90B.
"Unemployment
:
Cause and Cure",
"Depression Decade:
Oxford,
Martin Robertson
From New Era Through New Deal," New York,
77
Nelson,
Series:
R
and
,
C.
P
1
o
s s e r
Some Evidence and
1
,
9B
1
,
"Trends and Random Walks in Economic Time
Implications",
Journal
of
Monetary Economics, 10,
September,
139-162
Nickel
B.
1,
"A
19 64,
,
Review
Minford with David Davies, Michael
Centre for Labour
DECD,
A.,
19E5,
Scandinavian Journal
Phelps,
Romer,
of
E.
C
Political
Sachs,
.
,
1972,
,
1
9E6
,
1979,
and
of
'
of
J.,
19S3,
Economi cs
,
J.,
19B5,
Trade Unions
:
An
Introductory Survey,"
"forthcoming
"Spurious Volatility in Historical
February
"Wages,
"Real
1E5,
Economics
"Inflation Policy and Unemployment Theory," New York,
,
Unemployment Data", Journal
Profits and Macroeconomic Adjustment:
1,
2,
in
A
Comparative
269-319
in
the OECD Countries,"
255-2S9
"High unemployment
Norton
1-37
Wages and Unemployment
Papers on Economic Activity,
Sachs,
Cause and Cure" by Patrick
1984
Study," Brookings Papers on Economic Activity
Sachs,
:
Alison Prague", Discussion Paper
"The Economic Theory of
Economy, 94-1,
J.,
Peel
London School
Economics,
Employment Outlook,
Dswald,
"Unemployment
of
Europe",
mimeo Harvard
Brookings
78
Samuelson,
II.
,
P.,
1966,
"The
Collected Scientific Papers
Paul
A.
Samuelson," Vol
Wages," Review
of
Economics and
of
Cambridge, MIT Press
Slichter,
"Notes on the Structure
1950,
5.,
of
Statistics, 31, 2B3-2BB
Taylor,
Review,
May,
69-2,
Weitzman,
Weitzman,
s
1
1
y
OB-
Martin,
1
in
a
Macro Model,"
American Economic
13
"Increasing Returns and the Foundations
1962,
of
Unemployment
19S5,
M.,
"The Share Economy",
Cambridge Massachusetts, Harvard
Press
Woytinsky,
Social
1
"Staggered Wage Setting
Economic Journal, 92, December, 7E7-B04
Theory",
Uni ver
1979,
J.,
W
.
S
.
,
1942,
"Three aspects
Science Research Council
4953 080
of
Labor Dynamics," Washington,
D.C.:
Date Due
w$
yun 5
mo
my 06
^
5
may
y
t
Lib-26-67
Mil
3
i
I8RARIE
TDfiO ODM
S
53^
j
23