5. Employment Effects

advertisement
How Do Employment Effects of Job
Creation Schemes Differ with Respect to the
Foregoing Unemployment Duration?
Stephan L. Thomsen
ZEW, Mannheim
Reinhard Hujer
University Frankfurt/M.
3rd Conference on Evaluation Research, Mannheim
Contents
1.
Motivation
2.
Job Creation Schemes in Germany
3.
Evaluation Approach
4.
Data Set
5.
Employment Effects
6.
Conclusion
1. Motivation
•
Job Creation Schemes (JCS) are part of German ALMP since
1969:
 Subsidised employment
 For unemployed persons facing barriers to employment
•
Purpose:
 Re-integration of participants into regular employment
 Provision of stable foundation and relevant qualifications
•
Between 1997 and 2003:
 About 1.6 million participants
 Corresponding spending: about 23 billion Euros
1. Motivation (2)
•
Scepticism about effectiveness
 Lack of components that improve human capital,
stigmatisation
•
Recent empirical literature (e.g., Sianesi, 2004, Abbring/van
den Berg, 2003):
 Timing of treatment in the unemployment spell is
important for evaluating effects
•
Overall finding of previous empirical studies:
 JCS do not improve the employment rate of the
participants
1. Motivation (3)
•
Previous empirical studies evaluating JCS do not account for
the timing of the programme explicitly, but
 Timing seems to be important for JCS
 Differences should be analysed
•
Data used for the analysis: merged data from administrative
sources of Federal Employment Agency (FEA)
•
Evaluation Approach:
 Propensity Score Matching in Dynamic Setting
2. Job Creation Schemes
•
Provide jobs at public and non-commercial institutions for
unemployed persons facing barriers to employment
 Long-term unemployed
 Older unemployed
 Young unemployed without professional training/
apprenticeship
•
Financial assistance (paid to the employer)
 Wage subsidy of 30 to 75% (until 2003)
 Lump sum payment (since 2002 optional/ 2004
mandatory)
•
Duration
 Normally 1 year, but for two and up to three years
2. Job Creation Schemes – Pre-Conditions
•
For jobs to be promoted
–
–
–
•
Additional in nature
For the collective good
Appropriate to the problems of the regional labour market
For participants
–
–
(long-term) unemployment
Eligible for unemployment insurance benefits
2. Expected/Possible Effects of Job Creation Schemes
Pros
Cons
Microeconomic Dimension
• Adjustment (prevention) of human
capital (loss)
• Bridge to regular employment/
retirement
• ‘Soft‘ human capital effects/ Improve
Motivation
• Discreation of human capital
• Negative Signal to potential
employers
• Reduce one‘s own initiative
• Locking-in effects
• Discourage People
Macroeconomic Dimension
• Relief of labour market
• Investment in infrastructure
• Misallocation of Resources
• Competition with private production
• Displacement and substitution effects
3. Evaluation Approach (1)
•
Standard framework: model of potential outcomes (Y1, Y0)
 Designed for the case where programme is exposed once and
at one specific point of time
•
Purpose: Estimation of causal effect, e.g., average effect of
treatment on the treated (ATT):
- E (  | D  1) E (Y 1 | D  1)  E (Y 0 | D  1)
- E(Y0|D=1) has to be estimated
•
In comprehensive ALMP systems, unemployed persons
 face a number of different programs,
 could start at different points of time
3. Evaluation Approach (2)
•
Definition of non-participation is not straightforward
 All persons are potential non-participants as long as they do not
join a programme or leave the labour market for work
 Time until start of programme contains important information for
the effects and has to be considered
•
ATT with respect to the starting point of programmes:
(t, )  E (Y1 | Dt  1,U  t  1, D1 
E (Y0 | Dt  1,U  t  1, D1 
 Dt 1  0)
 Dt 1  0)
3. Evaluation Approach (3)
•
Parameter answers the following question:
“What is the impact at time  of participation in JCS for an
individual that joined the programme in time t of the
unemployment spell?”
•
Descriptive comparison of the estimated programme effects
for the single points t
 no causal interpretation of differences!
3. Evaluation Approach – Matching (4)
•
Idea: Conditioning on all relevant characteristics, X, to make
both groups comparable
 X must be observable!
•
Rosenbaum/Rubin (1983): not the single X, but
 a scalar function, p(X), propensity score
•
Identifying Assumption (Mean Conditional Independence
Assumption for ATT):
E(Y 0 )  D | p( X )  E[Y 0 | D  1, p( X )]  E[Y 0 | D  0, p( X )]
•
Dynamic Setting (see Fitzenberger/Speckesser, 2005):
E(Y )  Dt | p( Xt ),U  t  1,D1  ...  Dt 1  0
0
4. Data Set (1)
•
6 Samples of part. and non-part. (1:20):
Jul, Sep, Nov 00, Jan, Mar, May 01
•
Main sources:
 Programme Participants’ Master Data Set
 Job-Seekers Data Base
 Employment Statistics Register
•
Available information (objective and subjective):





socio-demographics (e.g., age, gender)
qualification/career variables (e.g., schooling, occupation)
labour market history (e.g., duration of last job),
information on regional labour market
Outcome variable: Regular employment (until Dec 03)
4. Data Set (2)
•
Data of the six samples are pooled (32,641 participants/
1,104,664 non-participants)
 consideration of time individuals spent in unemployment
•
Persons younger than 25 years or older than 55 are excluded
 better homogeneity of group in analysis
•
Employment effects are analysed separately for East and
West Germany and gender
•
Berlin is excluded from analysis
5. Employment Effects - Implementation
•
Unemployment is discretised into quarters (u=1,…,12=Umax)
 Programme effects of JCS are analysed for programmes
starting during the first three years of unemployment
•
Programme effects are estimated until =30
•
Estimation of four series of 12 probit models
•
Only the first programme in the current unemployment spell
is analysed, subsequent programmes are viewed as an
outcome of the first
5. Employment Effects for Men (t=1, t=5, and t=9)
West
(1,30)= -6.3
(5,30)= 7.5
(9,30)= 5.8
0.25
0.20
0.20
0.20
0.15
0.15
0.15
0.10
0.10
0.10
0.05
-0.00
-0.05
-0.10
ATT (Employment)
0.25
ATT (Employment)
0.25
0.05
-0.00
-0.05
-0.10
0.05
-0.00
-0.05
-0.10
-0.15
-0.15
-0.15
-0.20
-0.20
-0.20
-0.25
-0.25
-0.25
-0.30
-0.30
1
5
10
15
20
Month after programme start
Employment Effect
25
30
-0.30
1
5
95% Conf. Interval
10
15
20
Month after programme start
Employment Effect
25
30
1
95% Conf. Interval
0.20
0.20
0.20
0.15
0.15
0.15
0.10
0.10
0.10
-0.00
-0.05
-0.10
ATT (Employment)
0.25
ATT (Employment)
0.25
0.05
-0.00
-0.05
-0.10
-0.05
-0.10
-0.15
-0.20
-0.20
-0.20
-0.25
-0.25
-0.25
-0.30
-0.30
10
15
20
Month after programme start
Employment Effect
25
95% Conf. Interval
30
95% Conf. Interval
0.05
-0.15
5
30
-0.00
-0.15
1
25
(9,30)= -.8 n.s.
0.25
0.05
10
15
20
Month after programme start
Employment Effect
East
(5,30)= 1.0 n.s.
(1,30)= -5.5
5
-0.30
1
5
10
15
20
Month after programme start
Employment Effect
25
95% Conf. Interval
30
1
5
10
15
20
Month after programme start
Employment Effect
25
95% Conf. Interval
30
5. Employment Effects for Women (t=1, t=5 and t=9)
West
(1,30)= -2.6 n.s.
(5,30)= 11.9
(9,30)= 13.3
0.25
0.20
0.20
0.20
0.15
0.15
0.15
0.10
0.10
0.10
0.05
-0.00
-0.05
-0.10
ATT (Employment)
0.25
ATT (Employment)
0.25
0.05
-0.00
-0.05
-0.10
0.05
-0.00
-0.05
-0.10
-0.15
-0.15
-0.15
-0.20
-0.20
-0.20
-0.25
-0.25
-0.25
-0.30
-0.30
1
5
10
15
20
Month after programme start
Employment Effect
25
30
-0.30
1
5
95% Conf. Interval
10
15
20
Month after programme start
Employment Effect
25
1
30
0.20
0.20
0.20
0.15
0.15
0.15
0.10
0.10
0.10
-0.00
-0.05
-0.10
ATT (Employment)
0.25
ATT (Employment)
0.25
0.05
-0.00
-0.05
-0.10
-0.05
-0.10
-0.15
-0.20
-0.20
-0.20
-0.25
-0.25
-0.25
-0.30
-0.30
10
15
20
Month after programme start
Employment Effect
25
95% Conf. Interval
30
95% Conf. Interval
0.05
-0.15
5
30
-0.00
-0.15
1
25
(9,30)= -1.6 n.s.
0.25
0.05
10
15
20
Month after programme start
Employment Effect
95% Conf. Interval
East
(5,30)= 2.3 n.s.
(1,30)= -3.4
5
-0.30
1
5
10
15
20
Month after programme start
Employment Effect
25
95% Conf. Interval
30
1
5
10
15
20
Month after programme start
Employment Effect
25
95% Conf. Interval
30
5. Employment Effects (3)
•
West Germany
 Negative employment effects when starting early
 Positive when starting after one or two years, but:
result could not be established for all groups
•
East Germany
 Negative effects when starting early
 No positive effects for any of the groups in analysis
6. Conclusion
•
Overall
 Effects differ by t (descriptive comparison)
 Participation is associated with strong locking-in effects
 Persons who join after a short period of unemployment
are worse off
 Results tend to be better for long-term unemployed people
 Programmes do not improve the re-employment chances
of the participants compared to non-participation (in
adequate time after start of programmes)
 Results indicate a low target-oriented allocation of
unemployed persons into programmes
Download