working poor in western europe: what is the influence of welfare state provisions and labour market...
TRANSCRIPT
Working poor in Western Europe: What is the influence of welfare state provisions
and labour market institutions?
Henning LohmannUniversity of Cologne
“Conference of the EuroPanel Users Network (EPUNet-2006)” 8-9 May 2006, Barcelona (Universitat Pompeu Fabra), Spain
1. Introduction
• welfare state reform: one aim is to bring the workless poor into the labour market: activation, strengthening of work incentives, workfare re-commodification (see e.g. Haveman 1997, Lødemel/Trickey 2000)
• against this background increasing interest in working poor in Europe
• work in comparative perspective: Marx/Verbist 1998, Strengmann-Kuhn 2003, Peña-Casas/Latta 2004, Bardone/Guio 2005
• studies in single European countries: especially France, Switzerland, Germany, Ireland
• however: not much work on the explanation of country differences in in-work-poverty
1
1. Introduction
Welfare states and in-work-poverty:
• availability and level of transfers direct reduction of poverty and implicit minimum wages (de-commodification; Esping-Andersen 1990/99, Scruggs/Allan 2006)
• dual-earner policies higher number of earners per household (de-familisation/female autonomy; Lister 1994, Orloff 1993, O’Connor 1993)
• family solidarity care and protection for family members, but negative consequences for needs of households of workers (de-familisation/young or old-age dependence; McLaughlin/Glendinning 1994, Millar/Waxman 1996, Hantrais 1999) 2
1. Introduction
Labour market institutions and in-work-poverty:
• bargaining centralisation and coordination influence wage distribution and extent of low-wage work (Lucifora 2000, Blau/Kahn 1996, Wallerstein 1999)
• minimum wages secure a lower wage bond: influence on poverty depending on level of minimum wage (Lucifora 2000)
• however: personal earnings are only one source of income, earnings of other household members and transfers are also relevant
3
1. Introduction
Hypotheses (I):
• higher level of de-commodification lower level of in-work-poverty
• higher level of de-familisation lower level of in-work-poverty
but:
• higher needs of households of workers
• protection of (younger) low-wage workers via family solidarity
• centralisation of wage bargaining lower in-work-poverty
4
1. Introduction
Hypotheses (II):
• higher level of de-familisation (female autonomy) lower in-work-poverty of families with small children
• lower level of de-familisation (young age dependence) lower in-work-poverty of younger workers
• centralisation of wage bargaining lower in-work-poverty of low-skilled workers
5
1. Introduction
2. Data and indicators
micro:
• European Community Household Panel (includes all EU countries before Eastern enlargement, Sweden had to be dropped) 14 countries
• observation period: 1994-2001, 5-8 years of observation for each country
macro:
• various sources (OECD, Golden-Lange-Wallerstein 2006, European Industrial Relations Observatory, Künzler et al. 1999)
• observation period: around 2000
6
2. Data and indicators
• poverty rate:
• relative income poverty
• 60 percent of median equivalised monthly net household income (using non-modified OECD scale)
• based on current household income
• sample:
• working population in working age (17-64)
• ‘working’ defined on the basis of current employment status
7
2. Data and indicators
• independent variables (macro)
• de-commodification: net unemployment benefit replacement rate (in % of wage of average production worker)
• de-familisation I: children under 3 years in childcare per 100 children
• de-familisation II: number of household members 16+ years (country average)
• wage bargaining: centralisation (plant-level national level) and number of union members as a % of all employees (trade union density)
8
2. Data and indicators
3. Descriptive results
3. Descriptive results
9
0
5
10
15
20
%
DK A B FIN IRL UK D ES PT LUX F I NL GRcountry
working living in working hh total
Figure 1: Poverty rate by country (working age population, 17-64 years)
Source: ECHP 2001, own calculations.
4. Multivariate analysis
4. Multivariate analysis
10
• modelling strategy
I. panel regression models with fixed country effects (random effects, 1994-2001)
• micro perspective: needs, resources and restrictions
II. multilevel models with macro variables (two levels, random intercept, 2001)
• micro + macro perspective: de-commodification, de-familisation and wage bargaining
4. Multivariate analysis (I)
4. Multivariate analysis (I)
11
M1 M2 country (ref.: DK) NL 0.840 *** 0.437 *** BE 0.288 *** -0.186 * FR 1.296 *** 0.505 *** IRL 0.920 *** -1.013 *** IT 1.374 *** 0.060 GR 1.934 *** 0.234 ** ES 0.947 *** -0.845 *** PT 1.853 *** 0.369 *** AT 0.616 *** -0.403 *** FIN 0.696 *** 0.057 DE 0.655 *** 0.235 *** LUX 0.727 *** -0.173 * UK 0.817 *** 0.317 *** age in years -0.096 *** in years squared 0.001 *** gender (ref.: male) female -0.034 education (ref.: ISCED 0-2) ISCED 3 -0.489 *** ISCED 5-7 -0.899 *** no of children/persons in hh 0-2 years 0.812 *** 3-5 years 0.788 *** 6-14 years 0.856 *** 15+ years 0.979 *** marital status (ref.: married, never married, widowed) divorced/seperated 0.612 ***
Table 2: Coefficients (log-odds) from random-effects logit models on probability of being poor(working population, 17-64 years)
(continued)
4. Multivariate analysis (I)
12
working time < 15h 0.615 *** 15-29h 0.728 *** no of workers in household (by working time) < 15h -0.336 *** 15-29h -0.910 *** 30+ h -1.439 *** employment status (ref.: employee, non-low wage) self-employed/family member 1.233 *** low-wage worker 1.099 *** occupation (ref.: professional) legislators, senior officials, managers 0.613 *** technicians, associate professionals 0.293 *** clerks 0.652 *** service workers, shop/market sales workers 1.078 *** skilled agricultural/fishery workers 2.456 *** craft workers, plant/mach. operators, assemblers 1.256 *** elementary occupations 1.627 *** year (ref.: 1994) 1995 0.040 1996 -0.047 1997 0.000 1998 0.087 ** 1999 0.115 *** 2000 0.140 *** 2001 0.215 *** intercept -4.297 *** rho 0.573 0.476 N (observations) 440313 N (persons) 104885
Table 2 (continued)
Source: ECHP 1994-2001, own calculations. Notes: significant at p < 0.1 (***), < 1 (**), < 5 (*).
4. Multivariate analysis (I)
13
poverty risks (summary of model 2):
• young age / old age
• low education
• living in larger households
• divorce / separation
• part-time work
• self-employment / low-wage work
• living in single earner household (second earner working part-time already reduces poverty risk)
• agricultural occupations
4. Multivariate analysis (I)
14
M3 socio-demographic
age, age squared, gender, marital status, education
M4 household-composition
number of persons in household by age group (0-2, 3-5 6-14, 15+ years)
M5 household-composition and employment
number of persons in household by age group (0-2, 3-5 6-14, 15+ years), number of employed persons in household by working time, working time of individual
M6 self-employment/low-wage work
employment status + low wage work
M7 occupation 8 occupational categories
Models with subsets of variables of model 2
4. Multivariate analysis (I)
15
-2
-1.5
-1
-.5
0
log
od
ds
ES GR IT PT AT BE DE FR IRL UK FIN NLLUX
variables: full model [M2]
-1.2
-.6
0
log
od
ds
ES GR IT PT AT BE DE FR IRL UK FIN NLLUX
variables: socio-demographic [M3]
-1.2
-.6
0
log
od
ds
ES GR IT PT AT BE DE FR IRL UK FIN NLLUX
variables: hh-composition [M4]
-1.2
-.6
0
log
od
ds
ES GR IT PT AT BE DE FR IRL UK FIN NLLUX
variables: hh-comp.+employment [M5]
-1.2
-.6
0
log
od
ds
ES GR IT PT AT BE DE FR IRL UK FIN NLLUX
variables: self-empl.+low-wage [M6]
-1.2
-.6
0
log
od
ds
ES GR IT PT AT BE DE FR IRL UK FIN NLLUX
variables: occupation [M7]
Figure 5: Absolute change in country coefficients in comparison to model 1
Notes: Information on full model (M2) see table 2. Results from other models not reported.
4. Multivariate analysis (II)
4. Multivariate analysis (II)
16
Table 3: Coefficients (log-odds) from multilevel logit models on prob. of being poor
Source: ECHP 1994-2001, macro indicators see appendix, own calculations. Notes: significant at p <0.1 (***), <1 (**), <5 (*), <10(+).MXa: controlling for age/age-squared, sex, no of children in hh (under 3/6 years), marital status, year.MXb/c: additionally controlling for education, no of children (under 15 years), no of persons 15+ years,working time, working time of household members (instead of no of additional workers in hh), occupation.
M1 M2decommodification/defamilisation
replacement rate (unempl.) 0.001persons 16+ y. per hh (country mean) -0.803 ***childcare (places per 100 children) -0.016 ***
wage settingbargaining level 0.030union density -0.627 ***
cross-level interactionspersons 16+ y. per hh * agechildcare * child 0-2y. in hhbargaining level * low-skill occupations
n (obs.) 51063 51063n (groups) 14 14change of deviance (vs. micro model) 14.116 *** 8.768 *
4. Multivariate analysis (II)
16
Table 3: Coefficients (log-odds) from multilevel logit models on prob. of being poor
Source: ECHP 1994-2001, macro indicators see appendix, own calculations. Notes: significant at p <0.1 (***), <1 (**), <5 (*), <10(+).MXa: controlling for age/age-squared, sex, no of children in hh (under 3/6 years), marital status, year.MXb/c: additionally controlling for education, no of children (under 15 years), no of persons 15+ years,working time, working time of household members (instead of no of additional workers in hh), occupation.
M3 M4 M5decommodification/defamilisation
replacement rate (unempl.) -0.003 * 0.000persons 16+ y. per hh (country mean) -0.963 ***childcare (places per 100 children) 0.002
wage settingbargaining level 0.058 *union density -0.623 ***
cross-level interactionspersons 16+ y. per hh * age 0.011 *childcare * child 0-2y. in hh 0.006 ***bargaining level * low-skill occupations -0.046
n (obs.) 51063 51063 51063n (groups) 14 14 14change of deviance (vs. micro model) 7.170 6.346 10.818 *
5. Conclusion
• There is large variation of in-work-poverty in Europe.
• These differences are reduced by controlling for country differences in individual and household-related characteristics. However, differences in composition are not exogenous.
• The incidence of low-wage work has an influence on the extent of in-work-poverty but there are other, partly more important factors.
• At the macro level such differences are a bit hard to grasp, but there is evidence that de-commodification, de-familisation and wage-setting play a role.
• Changes in welfare state and labour market institutions changes in in-work-poverty?
18
5. Conclusion
Working poor in Western Europe: What is the influence of welfare state provisions
and labour market institutions?
Henning LohmannUniversity of Cologne
“Conference of the EuroPanel Users Network (EPUNet-2006)” 8-9 May 2006, Barcelona (Universitat Pompeu Fabra), Spain