PowerPoint-Präsentation

Download Report

Transcript PowerPoint-Präsentation

The Transition from Welfare to Work and the
Role of Potential Labor Income
Hilmar Schneider (IZA, DIW Berlin)
Arne Uhlendorff (DIW Berlin, IZA)
1/17
The Transition from Welfare to Work and the
Role of Potential Labour Income
– Introduction
– Literature
– Social assistance in Germany
– Data & Methodology
– Results
– Conclusion
2/17
Introduction
Figure 1: Development of the expenditures for social assistance in Germany
Bill. €
3/17
Introduction
Aim of the Study:
Estimating the Effect of the Ratio between the potential labor
income and the amount of social assistance on the
probability of a transition from welfare to work
Data: GSOEP 1992-2000
Methodology:
Discrete time Hazard Rate Models with competing risks
4/17
Literature
 Income variables usually not considered (Gangl 1998
or Voges and Rohwer 1992)
 Wilde (2003): Difference between social benefits and
the average income for unskilled employees
 Riphahn (1999): No effect of a predicted real net income
variable
 North American studies usually find negative effects of
the amount of benefits (Hoynes and Macurdy 1994 or
Fortin et al. 2004)
5/17
Social Assistance in Germany
 Means-tested transfer program
 In principle, everybody in need is eligible
 The amount is related to a basic minimum income
concept depending on household composition
 fills the gap between own income and the maximum
benefit for the household
 Labour income up to 25% of the basic allowance is
not taken into account, additional income is
deducted at an implicit marginal tax rate from 85 –
100%
6/17
Social Assistance in Germany
 Static labor supply theory: Participation probability
increases with the amount of benefits
 Dynamic Job-Search model: reservation wage depends
positively on the amount of benefits
 Hypothesis: Higher ratio between expected wage
and benefits  Higher transition probability
Derives a higher transition probability from a higher
acceptance probability or from a higher job offer
arrival rate?
7/17
Data
 GSOEP waves 1992-2000
 579 welfare spells between January 1991 - December
1999; 455 households
 386 uncensored, 193 right-censored cases
 199 transitions to work, 187 alternative transitions
 Transition to work: at least one adult household member
working fulltime, both working part-time, single
household: One person working part-time
8/17
Potential Net-Income and Social Assistance
1. Estimation of gross market wages of all heads of the
household and their partner
2.Potential net income: Highest gross income
accounting for income taxes, social security contributions and child allowance
3.Amount of social assistance: calculate the maximum
of social assistance
4.Calculation of the ratio between potential labour
income and social assistance
9/17
Estimation of the hourly gross wage
Wage equation
Selection equation
log wi  X 1i  1   1i
z i*  X 2i  2   2i
wi  wi* , z i  1 if z *ii  0
wi not observed, z i  0 if z *ii  0
Pooled sample using the GSOEP
waves 1991-1999
10/17
0
5
Percent
10
15
Distribution of the income ratio
0
1
2
ratio
3
4
Ratio 1: takes on the ratio value if the ratio is below 1
(25%)
Ratio 2: takes on the ratio value if the 1 < ratio < 1.5
(45%)
Ratio 3: takes on the ratio value if the ratio is above 1.5
(29%)
11/17
Model Specification
 Monthly data  discrete hazard rate models
 Assumption of an underlying continous time
proportional hazard rate
 Interval constant covariates and baseline transition
rates
 Competing risks: Employment and Alternative
transitions
 Unobserved heterogeneity: bivariate normal
distribution
 Random Effects Piecewise Exponential Model
12/17
Random Effects Piecewise Exponential Model
 (t x)  1 (t x1 )  2 (t x2 )
Hazard rate:
r (t xr )  0r (t )exp( xr r  r )
Risk-specific Transition rate:
 2 j
S ( j )  exp   exp( xrk  r   rk  r )
 r 1 k 1
Survivor Function:
Transition Probability:
fr ( j) 
exp( xrj  r   rj  r )
2
 exp( x
q 1

qj
 q   qj  q )
 S ( j  1)  S ( j ) 
Destination-specific components have to be
maximized jointly.
13/17
Random effects piecewise exponential model
Transitions to Work
Variable
December dummy
Coefficient
2.33***
Alternative Transitions
t-value
15.15
Coef.
3.08***
t-value
15.99
January dummy
-1.44**
-2.01
-0.52
-0.87
East Germany
0.78**
2.28
0.79**
1.97
Local unemployment rate
-0.05
-1.38
-0.05
-1.22
At least one adult household member with
vocational training
At least one adult household member with
school graduation
No partner household (female)
0.20
1.03
-0.17
-0.85
0.17
0.44
0.29
0.84
-0.68***
-3.67
0.18
0.80
-0.61*
-1.71
0.15
0.40
-0.84***
-2.85
-0.49
-1.60
-0.27
-1.56
-0.15
-0.73
0.46**
2.49
0.08
0.36
Non German adult household member
-0.23
-1.22
0.24
0.96
Handicapped adult household member
-0.16
-0.58
-0.04
-0.14
Income Ratio less or equal 1
0.91
1.59
0.50
0.42
Income Ratio between 1 and 1.5
0.93**
2.20
0.30
0.51
Income Ratio greater equal 1.5
0.64**
2.29
0.42
0.18
-4.29***
-5.29
-5.11***
-5.12
--4.33
-
-0,49
-
No partner household (male)
Adult household member aged > 50
Children aged 6 and younger
Children aged between 6 and 18
Constant
2
Ln(σ )
COV(1, 2)
Log Likelihood
-0.18
-1414.24
14/17
Results
 Ratio between potential labour income and welfare
level: positive effect for ratios above 1
 Interpretation of coefficients:
1< income ratio<1.5 : + 0.1  + 10% trans. prob.
1.5 income ratio:
+ 0.1  + 7% trans. prob.
 No influence on the probability of alternative transitions
15/17
Summary and Conlusions
 In contrast to previous studies we identify an effect of
the income ratio on the probability of transition from
welfare to work
 This „new“ result derives from a consideration of both
sources of income and from a differentiation between
transitions to work and alternative transitions.
 Incentive effects seem to be of prior relevance
16/17
Future Research
- What are the alternative transitions besides from
welfare to work?
- Which income situations do we observe following a
transition to work?
17/17