Download Examining Unemployment in European and Nordic Labor Markets and more Lecture notes Swedish in PDF only on Docsity!
SWEDISH ECONOMIC POI..ICY REVIEW 1 ( 1 994) 27-
Nordic Unemployment
in a Europem krspective
III In contrast to the high and persistent unemployment which has
plagued most European countries for two decades, the Nordic countries until recently maintained a resemblance of full employ-
ment. Since the early 1990s, however, unemployment has increased
dramatically in the Nordic countries. Against that background, this paper reviews a number of potential explanations for the different unemployment experiences in the past in order to assess their rele-
vance and the implications for future unemployment trends in the
Nordic countries. Higher past unemployment in other European countries probably contained both an element reflecting a rise in the natural rate of unemployment and an element reflecting slow adjustment back to the natural rate following adverse employment shocks. There are some indications that Nordic labour markets
have been exposed to smaller shocks, inter alia because real wages
have been relatively flexible, labour forces have been pro-cyclical and public sector employment has at times offset some of the weaknesses in private labour markets. With many features common to both Nordic and other European labour markets, there is a risk that Nordic unemployment could persist at a high level as has hap- pened elsewhere in Europe. H
- TIP nut17or ij Principnl Adnzinijtrntor in the Ecoizornic Prospects Diuision of the Economics Depnrtment, OECD. Witl~inthe Orgnniration, he previously worked on the country desk re- sponsible for Denmnrk, Finland nnd Sweden as well as in the Resource Allocation Divition, l~oldingrrsponsibiliiyfor sz(rveil/anceof OECD labozir markets.
SWEDISH ECONOMIC POLICY REVIEW 1 ( 1 9 9 4 ) 27-
Nordic Unemployment
in a European Perspective
Jergen Ekneskov*
A few years ago most of the Nordic region resembled an island of low un-
employment in a sea of high European joblessness. Since the early 1990s this picture has changed dramatically and Nordic unemployment is now
at almost the same level as that of the EC countries (Figure 1). This has
naturally prompted fears that the Nordic countries are about to repeat the
EC experience of unemployment persisting at a high level and has also
increased the interest in finding explanations for the previous, superior performance. Against that background, this paper reviews and compares a number of labour market features and institutions which may have had an impact on past developments and potentially also a bearing on wheth- er the Nordic countries are about to go down the same road of persistent-
ly high unemployment as the EC countries.
An initial word of caution may be appropriate, however. As will be seen subsequently, it is very difficult to distil either a European or a Nor- dic unemployment story. Indeed, variation across countries in both their experience with unemployment and their institutional set-up makes it
hazardous to group them in regions like the EC and the Nordic coun-
tries. An obvious example is that Denmark has experienced much the
same aggregate unemployment rates as the other EC countries but has in-
stitutional arrangements in the labour market which in many ways re- semble those of the (other) Nordic countries. Throughout this paper,
*The author wishes to thank Sveinbjom Rlondal, Lars Calwzfirs, Michael P Feiner,
Constantino Lluch, Sigurd Naess-Schmidt and Anders Vredinfor comments on a first draj qf this paper. Thanks also go to Catherine Chaptlis-Grabine~,Isabelle Duong and Isabelle Wanner who provided statistical assistance and to Rosemary Chahed, Lyn Louichaoui and Laura Schaub for technical assistance. The responsibiligfor all errors and omissions rests with the author The views expressed in the paper are those of the author and shot~ldnot be attri- buted to the OECD or thegovernments of its member countries.
N O R D I C UNEMPLOYIMENT IN A EUROPEAhr PERSPECTIVE, Jnrgen Elmeskov
Figure 2. A conceptud frmework for studying unemployment
1 Panel B I
Employment Employment
Real wage
Employment
Note: LD is a labour demand schedule, WS is a wage setting schedule, and LS is a labour supply schedule. Subscripts L and S refer, respectively, to long and short term arid superscripts 0 and 1 refer, re- spectively, to original and new position.
The labour supply schedule ( L S ) shows the amount of labour people are
willing to supply at different real wages when acting individually. Both
LS and the other curves in Figure 2 are drawn with labour relative to the
total population on the horizontal axis. For simplicity, the LS curve is
drawn as being vertical, i.e., labour supply is assumed to be independent of the real wage.
The location of the labour demand schedule (LD), which may alterna-
tively be thought of as a price setting schedule, depends, inter alia, on la-
bour efficiency, on the extent of product market competition, on cost com- ponents other than wages and on the size of public sector employment. Its downward slope may reflect either pro-cyclical price mark-ups or the pre- sence of a fixed factor, entailing a diminishing marginal product of labour. In the short term, thus, when the capital stock is fixed, the labour demand curve may consequently be relatively steep, whereas it may become flatter over the longer term when factor proportions are more flexible and labour demand therefore responds more strongly to real wages.
The wage setting schedule ( W S ) reflects the moderating effect of unem-
ployment on real wages. Its position relative to that of labour supply illu-
N O R D I C C N E M P L O Y M E N T I N A E U R O P E A N PERSI'ECTIVE, Jorgen Elrrlesliov
strates how much unemployment is "necessary" for the wage bargaining process to deliver a certain real wage. This depends, inter alia, on social at- titudes, the level of unemployment benefits and other factors which affect the relative bargaining strengths of labour market parties. Moreover, the wage bargaining process may be thought of as "internalising" firms' labour demand behaviour, implying that factors which affect labour demand may also affect wage setting. The wage setting curve is likely to be steeply sloped over the long run, with the hypothesis of a constant natural rate of unem- ployment represented by a vertical long-run wage setting curve. In the shorter run the wage-setting schedule is probably flatter since wages can be assumed to react only slowly to changes in unemployment. This framework can be used to illustrate various hypotheses concern- ing the rise in European unemployment. The simplest one is that various factors have weakened incentives for wage restraint. This can be illustra-
ted by a leftward shift of the wage setting schedule (Figure 2, Panel B)
implying that higher unemployment is associated with each level of real wages. I'ossible reasons are increased generosity of unemployment bene- fits, increased public acceptance of unemployment and increased mis- match in the labour marker. An alternative view of European unemployment is that it reflects very slow adjustment back to a lower level of equilibrium unemployment.
Panel C in Figure 2 illustrates the situation following a downward shock
to labour demand due, for example, to higher oil prices or higher real interest rates. T h e long-run equilibrium is portrayed as having shifted
from A to B. Due to aggregate demand effects, the short-run labour de-
mand curve shifts strongly to the left. As a result, the immediate effect of the shock is sharply lower employment but, because of slow wage adjust- ment, only a small reduction in real wages (point C). Over time, as high- er unemployment exerts downward pressure on the short-term wage set- ting schedule, real wages continue to fall, but the fall may be slow, as those out of job gradually see a reduction in their influence on wage for- mation. In the short term, lower real wages have only a limited impact on employment because the short-term labour demand schedule is relatively steep and only gradually shifts rightwards as a result of lower real wages. The upshot is that adjustment to a new equilibrium can be a very drawn-out process, which on the way may involve a complicated process of convergence with risks of overshooting. This evidently also goes for a disturbance originating on the wage setting side of the labour market.
A final view on European unemployment is that, following a shock,
somehow the process of adjustment comes to a halt before joblessness has
N O R D I C UNEMI'LOYMEN?' IN A EUKOPEAK I'ERSI'ECTIVE, J a r g e n Elmeskov
Fipre 3. Actkmd and trend unemp~oymenb
United States NAWRU indicator Actual and trend unemployment 2 10.0 10. 8 9.5 9. 5 9.0 9.
4. 4.
3.0 Unemployment rate 2.5 NAWRU^ indicator
Year-to-year change in wage inflation, percentage points
NAWRU indicator Actual and trend unemployment 3.0 I I
1 -^ -^ .-^ -^ .-^ NAWKU^ indicator
Year-to-year change in wage inflation, percentage points
NORDIC.UNE.MPLOYMENT IN A EUROPEAN PERSPECTIVE,J s r g e n E,lmeskov
Figure 3 (continued)
ECa NAWRU indicator Actual and trend unemployment 5 12 12 :: 11 11 "0 10 V E 9 9
0 0 -4 -3 -2 -1 0 1 2 1960 63 6669 72757881 84879093 Year-to-year change in wage inflation, percentage points
Nordic Countriesb
Year-to-year change in wage inflation, percentage points
NA WRU indicator Actual and end unemployment 3 12 12
a Germany, France, United Kingdom, Belgium and Denmark. b Finland, Norway and Sweden.
U 5 11 11
a ,- 10
u^10 U^99 8 8 E 0 (^) 7 7 d a 5 6 5 5 5
3 2 1 0 0 " " " ' " " " ' " " ' " " " ' " " ' " -4-3 -2-1 0 1 2 3 4 5 6 1960 6 3 6 6 6 9 7 2 7 5 7881 8 4 8 7 9 0 9 3
C L A & $ '
5^ C^ g 2. cE^ )^ ~^? G ~2 y , b K L g. q c
. g F $ 2 5
E + 'E cd c + -
" g g g ,* W U U C 5 6 % % % Z $ Y ) "
-^ .-^ 9 .E .c G^ .r: G 5-c^ "^ "^ Z ,so^2 c 2 2 ;.2.=:$ c 3 c c. 5 .;: .g 'z 2
r=^ .?.$ % g z c^ E^ K^ z
3 2 = E m^ g^ a
E.2, 4
5 2. 2 U - d KL m
.g .? ,d,
? - $ ; C c a,, 52 9
2 x. 5 Yes
u. z * ,. 9 Q~ '" C 'E 2 2 g g g c 0 c-2 G ." , C W $ 0 B cd 2 2 $ % %
+ " z g^ u^ O - L Q
4 $ 3 a,.= L4^ c.,^ g^2 a s % ; G s 5 ;z 2 2 ,;. .: 2
N O R D I C L ' N E M P L. O Y M E N T I N A EUKOI'EAN I'EKSI'ECTIVE, J a r g e r ~Elrneskov
Figure 4. Wage shaes in the business sector
United States Japan
Nordic Counrriesd
a Calculated by imputing a wage compensation to self-employed equal to average corn- pensation for wage earners. b Horizontal lines show fluctation interval over the period up to arid including 1980. They have been calculated as the mean wage share +I- two times its standard deviation.
' Belgium and the Netherlands are excluded.
d Finland, Norway and Sweden.
N O R D I C UNEMI'LOYMEN?' I N A E U R O P E A N PERSPECTIVE, Jorgen Elrneskob
Figure 5. Cyclicality and trend rise of unemployment
Percentage points
Standard deviation of yearly changes in unemployment rate, 1960--
N O R D I C UNEMPL.OYMENT I N A EUROPEAN PE,RSPECTIVE, Jclrgen Elmeskov
Table 2. Time-series properties of unemployment
Probability of unit roota against hypothesis of stationarity with: Constancy Drift Drift and trend United States Japan Germany France Italy United Kir~gdom Canada Australia Aurrria Belgium Denmark Finland Greece Iceland Ireland Netherlands New Zealand Norway Portugal Spain Sweden Switzerland Turkey a Based on augmented Dickey-Fuller rest. For the seven major countries, semi-annual data covering the period 1961:1 to 1991:2 were used, while annual data covering the same period were used for the small countries. T h e hypotheses being rested concern the noise process, Z, of the unemployment series, U, which also contains a deterministic trend, D, i.e., U = D + Z. Three different specifications of the deterministic trend are con- sidered, D = c (first column), D = cl + c 2. T I M E (second column), and D = cl + c2.TIME
- c 3 , T I M E 2 (third column). In all cases, the noise component is assumed to be an AR(1) process and the test concerns the parameter a in Z = a.Z(-1) + e, where e is white noise. The null hypothesis of a unit root is represented by a = 1 while stationarity implies that a < l. The table reports probabilities of accepting the null hypothesis. For additional infor- mation, see Campbell and Perron (1991).
countries or the absence of an adjustment process as in the extreme hyster- esis hypothesis. The power of these tests to distinguish between slow rever- sion and a unit root is, however, limited. Moreover, other specifications of the stationarity hypothesis might conceivably have led to rejection of the
unit root hypothesis also for most EC countries.
If the circumstantial evidence suggests that there is an important ele- ment of slow adjustment in European unemployment, the question still
N O R D I C UNEMPI.OYME,NT I N A EUROPEAN PERSPECTIVE, J ~ r g e nElmeskov
2. Comparing EC and Nordic countries
The rather abrupt rise in unemployment and the many signs of excess sup- ply in Nordic labour markets suggest that slow adjustment back to lower equilibrium unemployment is an important potential threat in the Nordic countries and that, therefore, learning from European experience in chis field is useful. In addition, it is important to identify factors which in the past may have accounted for equilibrium unemployment rising more strongly in E C than in Nordic countries. This section goes about these tasks by examining across countries a number of features relating to the three curves in the basic conceptual framework outlined in Section 1.1. First, labour supply is discussed; secondly, labour demand is taken up; thirdly, the wage setting relationship is considered; and, finally, the role of government intervention is examined within the overall framework.
2. 1 .Labour supply: trends and cycles
Trends. Labour forces in Sweden and Finland grew in line with most
other Northern and Central European countries over the two decades
from 1970 while in Norway, and particularly in Iceland, growth was
markedly higher (Figure 6). The composition of labour force growth dif-
fered considerably from most other European countries, with compara- tively slow growth in the population at working age but marked rises in participation rates (even though these were already relatively high at the beginning of the period).
Relatively stable participation rates in most EC and Central European
countries were the combined result of strongly falling male participation and some rise in female participation, whereas in the Nordic countries male participation fell much less and female participation rose more strongly. Despite strongly rising female participation rates and the particularly high frequency of female part-time work in the Nordic countries, the in- dications are that average working hours declined less in most Nordic countries than in the rest of Europe. This is explained, not least, by the fact that female part-time frequencies rose strongly elsewhere in Europe while they declined slightly, albeit from a high level, in most Nordic countries, Norway being an exception. Overall, thus, labour supply meas- ured in hours has in the past developed more strongly in the Nordic countries than in most other European countries. It cannot, therefore, ac- count for contrasting unemployment trends until a few years ago, even if
NORDIC U N E M I ' L O Y M E N T I N A EUKOI'EAN II'EKSPECTIVE, Jorgen Elrrleskov
Figure 6. Decomposition of labour force growth, 1970-
Contributions to average annual growth, percentage points
e Sweden
@ Norway
e Iceland
Canada porrugal United States e e Netherlands @ @ New^ Zealand Denrnark
Conrributiorl from population at working age Source: OECD, Labour Force Statistics
labour supply growth were thought to have a lasting impact on unem- ployment.
Cycles. Countries differ strongly in the extent to which a negative shock
to output shows up in higher unemployment. O n e of the determinants of the unemployment sensitivity to output variations is the degree of pro-cyclicality of the labour force, which also varies strongly across coun- tries. This pro-cyclicality, which in most cases seems to reflect causality predominantly running from employment to labour force (Elmeskov and
Pichelmann, 1993), has historically been relatively high in some of the
Nordic countries. In Sweden this may perhaps partly reflect that partici- pants in some active labour market programmes are classified as being
N O R D I C UNEMI'LOYMENT I N A E U R O P E A N I'ERSPECTIVE, Jargen Elmeskov
Figure 7. Unemployment and participation rates,
cross-country trends and levels
A. Levels 1990 (per cent)
85
80
75
70
6 5
60
5 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Unemployment rate
B. Changes between average 1966-70 and average 1986-90 (percentage points)
-2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Changes in unemployment rate a Labour force as a share of total population 15-64 years b The line is based on the estimated equation:
Participation rate = 80.4 - 1.38. unemployment rate; R2 = 0.
absolute t-values (31.5) (4.05)) The line is based on the estimated equation: changes in participation rate = 7.47 - 0.74 changes in unemployment rate; R2 = 0. absolute t-values (4.43) (2.68)
N O R D I C UNEMP1.OYMENT IN A EUROPEAK PERSPECTIVE, J e r g e n Elmeskov
Higher real interest rates in the 1980s entailed an increase in the costs of capital and tended to shift the labour demand curve downwards. T h e size of the shift evidently depended on the size of the increase in real interest rates as well as on the capital intensity and the substitution pos- sibilities in production. Evidence on the two latter points is not very ro- bust. However, there is some indication that production is relatively capi- tal intensive in the Nordic countries compared to other European coun- tries (Englander and Gurney, 1994). Real interest rates are generally unobservable and, in addition, the Nordic countries moved from having highly regulated financial markets in the 1970s to gradually more liberalised markets in the 1980s. Compa- risons across countries are therefore difficult. However, results based on a number of strong assumptions do not suggest that long-term real interest rates rose either more or less in the Nordic countries than was the case in most other European countries, though their average level in the 1980s
may have been comparatively low by European standards (Tease et a/.,
1991). In recent years, capital costs in the Nordic countries were evident- ly boosted by exchange rate pressures as well as by troubles in financial sectors which are likely to have contributed to a downward shift of the la- bour demand curve at the time when unemployment rose. Another common factor lowering labour demand (or limiting its rate of increase over time) is the slowdown in growth of total factor produc-
tivity (TFP) at the beginning of the 1970s. To the extent this slowdown
was not fully internalised in wage setting behaviour, the labour demand curve shifted down relative to the wage setting schedule and unemploy-
ment increased. There is some indication that the slowdown in TFP
growth may have been slightly less pronounced in the Nordic countries than in EC countries. Comparing the pre-1973 period with the 1980-
period, a simple average of annual TFP growth rates over the 12 EC
countries fell from 3.1 per cent to 1.2 per cent. The simple average for Finland, Norway and Sweden fell from 2.6 to 1.3 per cent between the same two periods (Englander and Gurney, 1994). One factor which has clearly tended to raise labour demand in the Nordic countries compared to other European countries is rapidly grow- ing public sector employment (Table 3). The evidence on the net effects over the long run is not clear, though if the wage setting and labour sup-
? Whether real interest rates (or energy prices) have any impact on equilibrium unem- ployment depends, intev aha, on the extent to which wage-setting behaviour internalises the effect of higher real interest rates on labour demand. Some arguments and evidence concerning such long-run effects are surveyed in Elmeskov (1993).
