Welfare stigma allowing for psychological and cultural effects. An Agent-Based simulation study. Dalit Contini Università di Torino Matteo Richiardi Università Politecnica delle Marche LABOratorio Riccardo Revelli
Aim To study the effects of income support on unemployment and welfare dynamics when social stigma is attached to welfare provision. Stigma in the literature Stigma is acknowledged as one of the determinants of a “bad” welfare take-up behaviour (Hernanz et al 2004) and it has been modelled as a cost of entry into welfare (Moffit 1983). By providing a disincentive for welfare participation stigma negatively affects propensity to enter welfare: high cost of entering welfare low take-up rates Moreover: by reducing the work disincentive of welfare a “good” provision stigma should reduce unemployment .
Simulation study with Our job-search model: main features Agent Based Modelling We let stigma affect preferences by representing a cost of entry into welfare, but in addition we assume that with stigma welfare provision can lead to a reduction of search effectiveness , due to: - progressive loss of self-confidence of recipients - unfavourable attitudes of potential employers Moreover, we allow for interaction among individuals : when people live in environments where most people rely on welfare, preferences can change: the perceived cost of stigma is reduced thus benefit is more desirable.
Results in brief With stigma: • welfare take-up rates decrease • welfare spells get longer • unemployment rates are not monotonically related with the strength of stigma As a consequence: we can find ( ceteris paribus ) higher levels of unemployment with stigma than with no stigma
The model.. Environment • all individuals unemployed “at birth” • only full-time jobs C 0 if unemployed without benefit • no savings: consumption=income consumption = B if unemployed with benefit C E if employed • total time T=2 employed time for work h=1, time for leisure L =1 Individuals choose whether to: - search for a job unemployed { } - enter welfare ∈ s 0 , 1 time for job-search { } ∈ time for leisure L 1 , 2 the employed lose their job with probability δ • • benefit unlimited
The model.. Utility function • Cobb-Douglas: α L β U = C • Moffit’s model for stigma: ( ) ( ) = − φ U C , L , A U C , L A Our specification: • ( ) ( ) ( ) A = − φ − U C , L , A , f U C , L 1 f α L β proportion of C welfare recipients among neighbours
The model.. Employment probability (given search) time elapsed in unemployment ( ) ( ) τ τ U A = γ − θ U − θ A p 1 1 t 0 time elapsed in welfare loss of employability loss of employability due to unemployment due to welfare (loss of skills) (loss of self-confidence, negative attitudes of potential employers) When individuals enter work, employment probability goes back to initial value γ 0
The model.. Choice function [ ] [ ] = + + 2 V U E U R E U R 0 0 1 2 [ ] ( ) ( )( ) = + − + + V U s , A , f U s , A , f 1 p U p R 0 0 0 0 1 1 1 0 E 0 [ ] ( )( )( ) ( ) − − + + − 2 U s , A , f 1 p 1 p U p ( 1 p p R 2 2 2 0 1 E 0 0 1 There are 2 6 =64 different combinations of values 0 and 1 for ( s 0 , A 0 , s 1 , A 1 , s 2 , A 2 ). V 0 is evaluated at each combination: the ( s 0 , A 0 ) maximising V 0 is taken as the optimal choice for time t =0 .
preferences Identity stigma ( ) ( ) ( ) A negative characterization = − φ − U C , L , A , f U C , L 1 f of self-identity employability Treatment stigma ( ) ( ) τ τ ( ) ( ) ( ) A U A = − φ − concern about being U A = γ − θ − θ U C , L , A , f U C , L 1 f p 1 1 treated poorly by others t 0 identity/treatment stigma Stuber, Schlesinger (Soc Sci & Med 2006)
INCOME BENEFIT (UNLIMITED) C 0 =1 “charity” B=1.5 “low” C E =4 employed B=2.5 “high” JOB TURNOVER STIGMA initial prob find job γ 0 ={0.25,0.4} NOSTIGMA: θ A =0 , φ =0 δ =0.05 STIGMA1: θ A =0.1, φ =1 prob lose job loss of skills θ U =0.05 STIGMA2: θ A =0.15 , φ =1.5 INDIVIDUAL CHOICES 2 L 0 . 5 U = C UTILITY FUNCTION LIFE LENGTH DISCOUNT FACTOR R=0.98 120 TIME UNITS
Outputs: - welfare take-up rates cross- - % unemployed section - % welfare participants - unemployment spell length longitudinal - welfare spell length
0,35 1,2 0,3 1 θ A =0.1 0,25 0,8 C B =2.5 0,2 γ 0 =0.4 0,6 0,15 0,4 0,1 0,2 0,05 0 0 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 1,2 1,3 1,4 1,5 Φ Costo all’entrata unemployed assisted t-u rate 0.35 1.2 0.3 1 φ =1 0.25 0.8 C B =2.5 0.2 γ 0 =0.4 0.6 0.15 0.4 0.1 0.2 0.05 0 0 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 θ A Loss of employability unemployed assisted t-u rate
BENEFIT BENEFIT BENEFIT LOW LOW HIGH NOSTIGMA STIGMA1 NOSTIGMA 13.3 26.3 14.3 % unemployed 9.0 21.5 10.1 % assisted 100.0 95.2 100.0 take-up rate 99.0 43.4 84.9 % searchers among assisted
BENEFIT BENEFIT BENEFIT LOW LOW HIGH NOSTIGMA STIGMA1 NOSTIGMA unemployment spell length 23639 21092 28743 n° spells 2, 3, 6 2, 4, 8 2, 3, 6 50°, 75°, 90° percentile welfare spell length 13569 10448 16730 n° spells 2, 4, 6 2, 5, 15 2, 4, 6 50°, 75°, 90° percentile
DK SW ITA ESP unemploy. 5.4 5.8 8.7 11.3 OECD, 2005 rate (2003) poverty 4.3 5.3 12.9 11.6 OECD, 2005 rate (2000) % persistent 2.5 - 8 8 EUROSTAT, 2002 poor (ECHP) at least 3 years (1996) take-up n.e n.e n.e n.e Hernanz et al, 2004 rate welfare spell _ 3-4 6 26 Saraceno, 2002 length months months months (ESOPO) median (Gothemborg, (Torino) (Barcelona) (1990’s) Helsimborg)
Limits and further developments • sensitivity analysis • choice function: only 3 periods considered • calibration of parameter • behavioural mechanisms underlying φ and θ A .... Is empirical validation possible?
1,2 1 0,8 unemployed 0,6 assisted search|a=1 0,4 0,2 0 θ A φ =0 0 0,04 0,08 0,12 0,16 0,2 theta_A
Recommend
More recommend
