Transcript 2011 (III)

Development of the Finnish labour
market
Annual Meeting of the International Forecasting Network
Helsinki
May 9/10 2011
Ilkka Nio
Finland has got rid of the global recession
• In 2009 the GDP decreased in 2009 by 7.2 % but quite rapid growth started from
the second half of 2010. Growth for the year 2010 was 3.1 per cent. All demand
items are contributing positively to growth, with exports as the strongest driver.
• Earnings increased rapidly in 2009 relative to the cyclical environment keeping
up the positive expectations and private consumption, which started grow during
the last quarter of 2009. In 2010 it grew by 2 %. However, many facts indicate
only modest growth of consumption. The new wage settlements will likely be
rather moderate after the recession. Purchasing power will be eroded by rising
interest rates and the acceleration of inflation.
• Negative risks are associated with the public sector indebtedness and financial
imbalances and the slow pace of employment growth. Finland's strong public
finances deteriorated sharply during the economic crisis, although the deficit is
still under the 3 % threshold.
• Economic growth forecast for 2011-2013 is not high enough that the deficit in
state finances would turn into a surplus during the economic recovery, but
without additional measures, they will continue to remain in deficit in 2015.
• All in all, the Finnish economy is expected to grow by 4 % in 2011. Without
Nokia´s production problem the GDP would increase even more. The growth is
expected to continue moderately ( 2-3 %) during the next years.
19
9
19 7 -1
97
-1
19 0
98
19 -7
99
20 -4
0
20 0 -1
00
-1
20 0
01
20 -7
02
20 -4
0
20 3 -1
03
-1
20 0
04
20 -7
05
20 -4
0
20 6 -1
06
20 10
07
20 -7
08
20 -4
0
20 9 -1
09
-1
20 0
10
-7
GDP growth in 1997-2011 (I), monthly figures
15
10
5
0
-5
-10
-15
:1
99
Q 1/1
:1
99
Q 2/1
:1
9
Q 93/
:1 1
99
Q 4/1
:1
99
Q 5/1
:1
99
Q 6/1
:1
99
Q 7/1
:1
99
Q 8/1
:1
99
Q 9/1
:2
0
Q 00/
:2 1
00
Q 1/1
:2
00
Q 2/1
:2
00
Q 3/1
:2
00
Q 4/1
:2
00
Q 5/1
:2
0
Q 06/
:2 1
00
Q 7/1
:2
00
Q 8/1
:2
00
Q 9/1
:2
01
0/
1
Q
Change in demand 1991-2010 (IV), quarterly figures
30
20
10
0
-10
-20
-30
Export
Consumtion
Investments
Supply and demand of labour
• Alongside the growth of the economy the demand for labour has started to
strengthen slowly.
• Young peolple were mainly influenced by the impacts of the recession, while
the aged labour force kept their jobs and stayed in the labour market. It
seems that working careers are still being prolonged.
• Even then the changing population age structure will begin to affect the
supply of labour. The ageing of the population will soon begin to constrain
the labour market.
• The number of hours worked by persons employed fell during the recession
twice as much as the number of people in employment: the number of hours
worked dropped by 6 % and the number of persons employed only by 3 %.
• Employment is expected to grow slowly. In particular the financial problems
of the public sector will limit employment opportunities. The productivity
will be raised by rationalizing the use of current labour force. Employment
will start to grow faster after the labour input and production are
rebalanced.
Labour force and employed persons in 1988-2011 (III)
Thousand persons
2900
2800
Labour force
2700
2600
2500
2400
2300
Employed
2200
2100
2000
1900
'88
'90
'92
'94
'96
'98
'00
'02
'04
'06
'08
'10
Rapidly changing age structure
Age
2010
2020
Change
Persons
%
15-24
659 800
605 800
-54 100
-8.2
25-49
1 727 100
1 730 700
3 700
0.2
50-64
1 160 600
1 072 000
-88 600
-7.6
65-74
506 700
719 000
212 300
41.9
4 054 200
4 127 500
73 300
1.8
Total
Expectation of time spent in the labour market at age 50 in 1990 - 2010
Expection, years
11
10
9
8
In labour force
In employment
7
6
5
'90
'92
'94
'96
'98
'00
'02
'04
'06
'08
'10
Lengthening of working careers by two years untill 2020
100
Change 2010-2020
Age
60
40
20
Persons
%
15-24
-24 000
-7.3
25-49
2 000
0.1
50-64
33 000
4.1
65-74
49 000
121.9
Total
60 000
2.2
74
70
72
66
68
62
64
58
60
54
56
0
50
52
Per cent
80
Age
* participation rates for ages 15-49
In labour force 2010
In labour force 2020
on the level of 2010
Unemployment situation
• The deterioration of labour market situation during the crisis was not as
bad as predicted. Companies were keen to retain their skilled staff,
offering them shorter working hours and layoffs over redundancies.
(Labour hoarding)
• Part of the reason behind the recent decreasing unemployment is the
considerable increase in the use of active labour market policy measures.
• The structural composition of unemployment has been aggravated. The
long-term unemployment and structural unemployment has increased,
slowing down the decrease of total unemployment.
• The unemployment rate ( 8.4 % in 2010) will decrease annually by not
more than one percentage point > to 7,5 per cent in 2011 and on to 6,5
per cent in 2012.
Unemployment rates according to the LFS and JSR in 1989 - 2011 (III)
%
25
20
15
10
5
0
'89 '90 '91 '92 '93 '94 '95 '96 '97 '98 '99 '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12
LFS
JSR
Youth unemployment according to the jobseekers register and Labour
Force Survey 1991 – 2011 (III)
200
Thousand persons
180
160
140
120
100
80
60
40
20
0
'91
'93
'95
'97
'99
'01
JSR
'03
LFS
'05
'07
'09
'11
Hysteresis or persistence in Unemployment ?
• There is considerable evidence ( micro and macro ) that hysteresis ( or
persistence) of unemployment is an important factor in the Finnish labour
market.
• Strong negative duration dependence means rapidly declining chances of
becoming employed when the duration of unemployment spell lengthens.
• Persistent high unemployment can lead to an increase in the long-term natural
rate of unemployment ( NAIRU) determined in the labour market. Different
country can follow different time path in achieving the equilibrium.
• The effects of shocks are difficult to assess and forecast. Hysteresis is a lagging
variable among explanatory variables of unemployment.
• Overcoming the problem of hysteresis has been the major policy issue in Finland.
For this purpose a specific indicator was created in 2003 to measure the core of
structural unemployment. In March it accounted for 144 000 ( 5,5 % ).
• Considerable increase in the use of ALMP measure ( in March 4,1 %)
Unemployed persons seeking work (1) and unfilled vacancies at the employment
service (2) . Original monthly figures and seasonally adjusted figures (S)
Thousand persons
550
500
450
400
350
(1)
300
250
(S)
200
150
100
(2)
50
0
'61 '63 '65 '67 '69 '71 '73 '75 '77 '79 '81 '83 '85 '87 '89 '91 '93 '95 '97 '99 '01 '03 '05 '07 '09 '11
Average duration of incomplete unemployment ( stock) is a
lagging cyclical indicator
Average incomplete duration ( stock)
30
60
25
50
20
40
Weeks
Weeks
Average terminated duration ( flow)
2008 VI
15
2011 II
10
2008 VI
2009 IX
30
20
1991 I
1991 I
10
5
0
0
0
0
5
10
15
20
Unemployment rate % ( JSR)
5
10
15
25
Unemployment rate (JSR)
20
25
Structural unemployment in 2004 – 2011 (III)
200 000
Recurrent participation labour policy measures
Unemployed after participation in active labour policy measures
180 000
Recurrent unemployment
160 000
Long-term unemployment
140 000
120 000
100 000
80 000
60 000
40 000
20 000
0
2004
2005
2006
2007
2008
2009
2010
2011
Unemployed and persons placed on ALMP-measures, per cent of labour
force 1988-2011( III)
25
6
20
5
4
15
3
10
2
5
1
0
0
-1 9-2 0-3 1-4 2-5 3-6 4-7 5-8 6-9 -10 -11 -12 1-1 2-2 3-3 4-4 5-5 6-6 7-7 8-8 9-9 -10
8
8 8 9 9 9 9 9 9 9 7 8 9 0 0 0 0 0 0 0 0 0 0
19 19 19 19 19 19 19 19 19 199 199 199 20 20 20 20 20 20 20 20 20 201
Unemployment rate (JSR)
In ALMP-measure, per cent of labour force
Both quantitative and qualitative methods are used in order to get
maximum information from the available data
Big macroeconomic models do not provide robust information to forecast the turning points
and cyclical changes in employment. >> Need for short term forecasts.
Qualitative approach
• Collecting qualitative information from regions
• Comparing the forecasts of different authors
• Herd instinct among forecasters > The forecasts do not often differ much from each others
Time Series Analysis ( STAMP, ARIMA)
• Monthly follow up of time series from the economy and the labour market
• Specification of cyclical, seasonal and irregular variations in order to search turning points
Experimental quantitative approach
• Often simple models are useful to help understanding the relationships between the
economy and the labour market.
• Searching leading indicators
• It is difficult to interprete the variables regarding expectations
Unemployment rate by regions
2010
Expectations, one
year ahead by
regions Spring 2011
Whole country 8.4 %
11,0 % and more
9,0 % - 10,9 %
Lappi
Lappi
11,3
Little decline
7,0 % - 8,9 %
Very little decline
Less than 7,0 %
PohjoisPohjanmaa
Source: Statistics Finland
10,2
Kainuu
9,0
PohjoisSavo
PohjoisEteläKarjala
Pohjanmaa Pohjanmaa
10,0
Keski12,5
6,7
8,2
Suomi
Etelä-Savo
Pirkan- 9,9
7,9
Satakunta maa
9,7 Häme
8,8
Kaakkois-Suomi
9,0
10,6
Varsinais-Suomi
Uusimaa
8,1
6,4
No change
PohjoisPohjanmaa
Kainuu
Source: MEE
PohjoisPohjoisEteläSavo
Pohjanmaa
Karjala
Pohjanmaa
KeskiSuomi
Etelä-Savo
PirkanSatakunta maa
Häme
Varsinais-Suomi
Uusimaa
Kaakkois-Suomi
STAMP Structural Time Series Analyser, Modeller and Predictor
•
Follow up of estimated local trends every month in order to understand the behaviour of the time series
and the fit of the forecasts.
•
Labour market figures are dominated by stochastic trend component
 without explanatory variables forecasting possible only in the very short run
•
The aim of structural time series modelling is the specification of various components: cyclical, seasonal
and irregular. The components are considered as stochastic unobserved components which are assessed
by looking at the behaviour of the series throughout the whole sample rather than at the end of the
period.
•
Comparing the structural time series models with ARIMA models, the essential difference is that the
trend or the unit root component is modelled together with the stationary part of time series. There are
two parts to the trend: Stochastic level which is the actual value of the trend and stochastic slope.
•
Kalman filter forecasts the continuation of the series and also gives, as a side product, the estimates for
the unobserved components. The forecast function of STAMP estimation is a straight line with upward or
downward slope.
•
Including explanatory variables in a structural time series model - as stochastic variables - results in a
mixture of time series and classical regression.
•
The combination of unobserved stochastic components with explanatory variables opens in principle
many possibilities for dynamic modelling. However, in practice the choice of potential explaining
variables is rather limited.
•
The predictive power of forecasts can be strengthened by means of leading economic indicators such as
business expectations or business assessment using data on order books, various confidence indexes,
stock market index, etc
JSR
LFS
2010-3
2009-1
2007-11
2006-9
2005-7
2004-5
2003-3
2002-1
2000-11
1999-9
1998-7
1997-5
1996-3
1995-1
1993-11
1992-9
1991-7
1990-5
1989-3
1988-1
Slope of unemployment rate 1988-2011(III), STAMP
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
-0,1
-0,2
-0,3
Consumers´ confidence ( saldo) and change in employment ( slope), 1000
persons, monthly figures ( 2011 III)
25
20
15
10
5
0
-5
-10
-15
-20
Consumers´ confidence
Slope of unempl. job seekers
M:2010/11
M:2010/04
M:2009/09
M:2009/02
M:2008/07
M:2007/12
M:2007/05
M:2006/10
M:2006/03
M:2005/08
M:2005/01
M:2004/06
M:2003/11
M:2003/04
M:2002/09
M:2002/02
M:2001/07
M:2000/12
M:2000/05
M:1999/04
M:1998/09
M:1998/02
M:1997/07
M:1996/12
M:1996/05
M:1995/10
Consumers´ confidence ( saldo) and change in unemployment (
slope 1000 persons), monthly figures until 2011 (III)
10
8
6
4
2
0
-2
-4
-6
Cross-correlation of consumers´ confidence and slope
of unemployment, monthly figures
Quantitative experimental approach
• Purpose is to find handy and simple models which could be precise, convincing and clear
enough, so as to be interesting for decision-makers
• Starting with simple models which contain just a small number of equations and variables
formulated on a priori considerations, in order to combine the theoretical considerations
with the empirical observations.
• Different calculations are carried out by adding explanatory variables in various
combinations and taking advantage of background economic forecasts for output and
various demand items, which have a lead over the employment data. Changing their values,
the path of the forecasts can be examined under different scenarios.
• Due to intercorrelated variables, we need to compromise with the scientific needs regarding
explanatory ability and accuracy of the estimates ( when certain that the same pattern of
multicollinearity of dependent variables will continue ).
• However, there is no standard business cycle and thus the stability of the estimates and their
sensitivity to the changes in the sample period can vary.
GDP growth and change in employment 1977 - 2010
Thousand persons
100
50
1994 - 2010
0
2010
-50
1977 - 1993
2009
-100
-150
GDP (t)=0,5*((GDP (t)+GDP(t-1))
-200
-6
-4
-2
0
2
4
6
8
Model: Dependent variable: Quarterly change in employment
Explanatory variables: change in demand items, lag 2 quarters
Method: Least squares, R-squared 0,69
Demand item
Unstandardized
Coefficients
B
Std. Error
(Constant)
-9,743
5,206
Export(-2)
1,445
,313
Consumption(-2)
6,956
Investments(-2)
1,554
Standardized
Coefficients
Beta
Collinearity Statistics
t
Sig.
Tolerance
VIF
-1,871
,066
,365
4,616
,000
,783
1,276
2,131
,321
3,264
,002
,508
1,970
,495
,331
3,140
,003
,442
2,262
:1
99
Q 4/1
:1
99
Q 5/1
:1
99
Q 6/1
:1
99
Q 7/1
:1
99
Q 8/1
:1
99
Q 9/1
:2
00
Q 0/1
:2
00
Q 1/1
:2
00
Q 2/1
:2
00
Q 3/1
:2
00
Q 4/1
:2
00
Q 5/1
:2
00
Q 6/1
:2
00
Q 7/1
:2
00
Q 8/1
:2
00
Q 9/1
:2
01
Q 0/1
:2
01
1/
1
Q
Fit of employment model, 1000 persons, demand items as
explanatory variables
2600
2500
2400
2300
2200
2100
2000
1900
Actual
Fitted
Residual of the employment model