Polarisation of employment over households revisited: the Belgian case

Download fulltext

Polarisation of employment over households revisited: the Belgian case Vincent Corluy (KULeuven-CES & UA-CSB) and Frank Vandenbroucke Abstract Over last 30 years share individuals in working-age population without (‘individual joblessness’) has fallen continuously while with no member (‘household remained fairly stable In this paper we examine why individual joblessness household diverge The growing gap between both measures reflects changes composition distribution We describe latter phenomenon by constructing a measure ‘polarisation households’ which is based on difference observed ‘expected’ On basis formation one would have expected Belgium to decrease However increasing polarisation i e an increasingly unequal jobs counteracted evolution Singles constitute that are most vulnerable but shift towards such more offers only small explanation time observe rising levels single adult couples Within these groups similar among important factor overall increase personal characteristics associated (gender education age region origin) explain significant part level 2012 almost half can be explained through typical singles third related marital selection also means substantial cannot A comparison three low-polarisation countries (Germany France Netherlands) two high-polarisation (Ireland United Kingdom) suggests high when compared Germany Netherlands for following reason First than those weak socio-economic profile terms skills origin: total they often low-skilled older born outside European Union Second strong profiles larger (whether it concerns or not) reinforced regional differences within Third educational homogamy JEL keywords Employment jobless poverty Corresponding author: frank vandenbroucke@kuleuven 2 Content 1 Introduction 5 Background 3 Purpose structure 7 4 Data definitions 8 Trends 9 6 Realities behind non-employment 11 concept 13 Decomposition 15 Explaining polarisation: conditional randomization risks 17 10 Cross-country 20 relative severity work-poverty 24 12 Conclusions questions further research 27 Bibliography 56 List figures Figure 1: Individual (expected observed) (non-)employment rates 1983-2012 Belgium; LFS 2: region; 31 3: Household size 32 4: 1983-2012; 33 5: 34 6: Evolution 1983 – 2012; 35 7: Age adults 36 8: Share at least child as percentage all living 37 9: Education 1992-2012 region: 38 10: Country birth 1995-2012 39 11: Levels 1983-2012: 40 12: 41 13: weighted contribution 42 14: 43 15: its regions; 44 16: unconditional 1993- (1992-)2012; 45 17: EU24 (+ regions); 47 18: 48 19: 49 20: other 50 21: Polarisation: (random distribution) 53 22: indicators: ratio (ILO-definition); polarisation; work (based intensity) 2011; SILC 55 tables Table Marginal effects probability (after probit estimations) regions Ireland UK 51 Socio-economic working Standard deviation across subgroups; 52 Educational Kingdom Correlation matrix indicators measured typically proportion who These aggregate data may not fully capture economic social impact Exclusion person from labour market affects him her directly members household; increases insecurity entire limited collective mechanisms protect against shocks (in particular new states Southern countries) perspective particularly At-risk-of-poverty income inequality Hence policy analysis should consider emphasis intensity Europe 2020 strategy rationale integrated dimension into exclusion appropriate remains (2013) discuss alternative concepts underlying intensity; indicator apply instance class Our focus observation individual-based household-based offer conflicting signals about performance fell since early 1980s Such confined discrepancy been documented comprehensively number Gregg Wadsworth (2008 2012) understanding relationship far obvious By exploring highlights caveats difficulties possible pitfalls our patterns When indices well-being Berloffa Modena (2011) take account (un)employed They propose ‘inactive-unemployed dependency rate’ defined average inactive depend employed ones Similar unemployment different because demographic change processes affect inactiveunemployed rate Even if does one-earner parent types Marital instability vulnerability Indicators include persons (aged 18-59) (households where work) below 60 years) very low less percent their potential) See (1996) OECD (1998) Dawkins Scutella (2002) (2008) (2012) Brandolini Viviano Ozdemir Ward Duyver Gradin et al (2014) framework measuring deprivation Their takes incidence (how many touched lack any members) being non-deprived) concentrated households) aim ascertain extent incorporation trends general economically active affected large contrast middle-range illustrate country This literature focuses construction ranking countries; explore reasons divergences (2010 analyze factors contributing measures: order construct ‘polarisation’ Increasing mainly observable two-adult explaining divergence five up mid-2000s Meanwhile researchers interested started link records welfare characterize Shifting attention natural move hence rather correlation mature prima facie come surprise solid association identified higher lower at-risk-of-poverty micro-level much weaker macro-level Paul de Beer (2007) analyzes failure Lisbon Strategy regard goal contrasting long-term Outlook 2001 (2001 59-61) pointed absence correlations hand finding cited inspired authors Dickens Ellwood Nickell (2004) Graaf-Zijl Nolan analyses during ‘good years’ collected 2000s volume Cantillon emphasize As (2014 see note p 54 113) results contradict findings OECD; however relate sample (ECHP Canada USA) database (ECHP) obtain For recent examination cross-country Diris Verbist decomposition inter-temporal Vinck All show Moreover rankings substantially severe examined (even job was available household) fact under review registers highest deprived indicating burden strongly certain rank along will use constructed quantify sources elaborating upon highlight first caveat interpretation what decreased considerably points ‘full-employment employment) increased spectacularly 40% 65% Now spectacular mathematical corollary rise rates: expect (a of) distributed randomly relevant question portion mathematics random distributions calls sociological cultural explanations region- country-specific underscores relevance proposed his co-authors used allows subgroups drive second realities notion ‘household joblessness’; index counterfactual (or ‘expected’) occur were given specific actual rate: there fewer) national technique refinements decomposed It appears (as driven shares single-adult conceptual possibly pitfall What label words decline parameter α indicator) drives disappointing truism black box contains shifts markets societies well-known lone parents refined taking traits influence gender live calculate ‘conditional polarisation’ complexity emerges yields emphases: whilst today’s crosscountry concentrates interregional Subsequently shown comparing structures relatively bad position Finally importantly complex binary distinction ‘jobless’ ‘non-jobless’ probably too crude meaningful section correlates ‘relative poverty’ observations (continuous) establish fine-grained underscore circumspection applied wrap concluding point out Labour Force Survey conducted Bureau Statistics6 detailed information Around 000 surveyed yearly Throughout ‘adult’ short cut ‘Working age’ refers bracket 20-59 excluding full-time students reference 59 elderly try minimize participation retirement adults’ decision participating force7 At cluster present Doing so distribute sizes singleadult footnote ADSEI: Algemene Directie Statistiek en Economische Informatie exclude count (to classify ‘not jobless’) define Because respondents lacking survey ‘full-time students’ old ILO status ‘inactive’ According he she hour week (excluding 59) primarily Adults either unemployed force (rate)’ refer same categories) ‘Individual joblessness’ denominator elementary do impose scale example part-time versus nor effective hours months achieved calendar year decade emerged each demarcations populations employment; joblessness8 definition great advantages decomposing evaluation correlate compare ILO-based (household index) finegrained EU-SILC (survey 2011) Individual-based documents using [Figure here] lines contrary 25 65 ‘mixed-employment (with mixture non-working adults) 26 replaced ‘mixedemployment If situation median belongs household’; 1991 onwards household’ rapid male breadwinner model dual earner appendix Next pattern dotted red line full-employment households; yellow mixed-employment hypothesis; green vein Consider simple understand nature counterfactuals: organized singles; fraction δ risk equal (smaller) jobless; conversely 1-δ (1-δ)2 finally (2δ-2δ2 ) expectations prevalent households: allow pooling individual) smaller 0 decreases prevalence mixed (frozen) value fullemployment Simultaneously counterfactuals evolved frozen theoretical previous paragraph Given evolving (the line) 19 points; still considerable diminished instead slightly purely concomitant contain partly explains grown even faster standstill could view interesting (e g singles) (expected) lines? how (observed) repeats figure disparities ranged Brussels Capital Region) Wallonia Flanders Looking remarkable altered (one forget rapidly) although 14 Flanders) Regional diverged next reflected 16 dropped While bottom Just joblessness: ranges respectively look Wallonia: respective growth Ceteris paribus having one; lead things approach ‘expected function decades trend occurred shows doubled; bigger every pace (see 5) (33 percent) followed (24 turn (16 type (by couple diverging 67 61 lives did period image stability hides opposite evolutions 1987 sharply 6) categories younger differ especially cohort dependent children gradually well group around per cent (and Brussels) persistent changed past thirty beginning lived man woman wasn’t During industrial restructuring (between 1981 1987) sizeable men lost jobs9 Due sole suddenly few women downturn immediate effect Gauged took digest wave exits participated stayed longer held differs 50-59 exit plays role presumption workers via pre-retirement circumstances partner become cohorts significantly Flemish Both reverses crisis alone precarious pronounced house (around 70 child) Between start series) unchanged slightly) corroborate emphasized relation find ‘severe (an characterizing below) And people disparity diploma secondary hardly Its host tertiary educated background rapidly medium high-skilled socio-demographic (at partly) origin non-EU15 doubled (initially) capital foreign-born non-native 18 proportions non-EU stronger Brussels10 introduced presented Thus hypothesis subsequent made concept; ‘observed’ ‘counterfactual’ difficult work-intensity Formally written: (2015) divergent foreignborn native non-European natives immigrants But despite (very) female current follows traditional negative t expression carry normative meaning us benchmark ideal context opportunities ‘positive might seen kind ‘Matthew effect’: concentration additional advantage (say someone already employed) some (compared partners jobless); ‘negative appreciated form solidarity fair scarce suggest maximally ‘randomly jobs’ serves neither sustainable desirable argument comes cost: need supported transfers cost avoidable state sense suboptimal equilibrium whole Despite place subgroup points) currently confronted similar: diminishes jobs: implied -by definition- -2 Since 1990s expectation segment changed: approached zero then became positive starting mid-nineties consistent added worker theories Cullen Gruber 2000) presence reduce supply occupants Indeed lagged equals combined resulted -1 900 estimates 206 500 extra Strong variation prevails (14 double (6 times (3 9) Changes outspoken twice strongest Wallonia) paragraphs size) separately together integrate them Figures display (weighted structure) 68 due within-household ends stems three-adult constitutes 81 merely Observed consists One nonemployment joblessness) evaluate stem evolve distinguish ‘within-household change’ structure; itself ‘between-household four terms: (i) (for structure); (ii) joblessness; (iii) between-household degree levels; (iv) n = k (number K maximal presents lower; light blue bar (contribution counteracts decreasing reduces dark Additionally ‘change within’ between’: shifted (due ‘within’ ‘between’ change) net diamond 15) now diamonds diminishing (light bars) prominent (dark close quarter induced offset improvement Here quick exerted upward pressure (about Most skewed cardinal Instead known combination Lower translate common speak assortative mating amplified variety vary of: (three groups: 20-29 30-49 50-59) education11 primary education) (Brussels Walloon region) predict trait interaction characteristics; reality displays (singles couples) attained ISCED EU-LFS gathered 1992 confine better prediction closer assess characteristic ‘unconditional only) various ‘explanatory power’ (conditional characteristic) expressed interpreting bear mind absolute Therefore explanatory power changing tend gender-specific found Thirty earlier amounted 100 little (positive) implies (i convergence women) contributes declining 1993 poorly freeze stay fixed mean likely (among emerge conditioning (from identical Keeping constant decreased12 global homogeneous non-educated declined ‘mixed educational’ residence definition: always Although high: estimate importance (13 se elsewhere) combining variations sharing within-group component unexplained rose supra) yet To (rather stable) hide young (20-29) struggling (over profile’ Assortative independent overrepresentation expense (5 time) keep Among had 76 58 period) Allowing introducing variances comparative countries13 straightforward bulk (on EU24) (Figure 17) Countries system late transition adulthood family ties (like Member States) Romania roof (28 compares With performs (1 (Lithuania Spain Greece Ireland) perform worse succeed keeping indeed Lithuania Bulgaria above Malta Italy Luxemburg Southern-European States (Malta Estonia) division values (-7 Italy) Scandinavian excluded 21 SouthernEuropean EU23 clusters EspingAndersen initially decommodification; later (1999: 72) brought play ‘the ways production allocated varies rising) parents’ home New continental belongs) correspondence exists geography regimes (Hank 2007) contribute ceteris contributory sometimes direction (which result structures) mitigated addition illustrates exception Esping-Andersen’s ‘conservative’ (AT NL DE FR) Bulgaria) scores thus comparatively poor remainder international six neighboring logistic regression causal factors; complements return end aims identify thereby irrespective run make 22 variance ‘good’ ‘bad’ develops leads again fourth [Table determine regions) after controlling marginal variables control greater non-singles ‘being single’ characteristics) lowest largest though driving notably ‘weak There UK; singles’ Thirdly whether subset here indicates standard deviation14 nowhere big spread origin; mutually variable root deviations 23 comparably creates homogamy15 magnitude somewhat Netherland identifies must done way calculated left correcting except Germany16 add (third bar) (fourth fifth exercise 90 occurs controlled seems deviating rest exclusively (BE IE) seem divide individuals’ call 4) German ‘country birth’ linked and/or Conditional Income basis: at-risk-of counts standardized disposable threshold unambiguous systematic belong (Vandenbroucke 2014 pp 16-17; Commission Chapter 2) ‘Household intensity’ Eurostat defines worked theory worked17 reported equivalents computed habitually interview crucial comparisons logical Contrary 2015) non-elderly pretransfer Intuitively follows: compensate transfers; committed successful doing Understanding requires subtle bivariate metric ‘joblessness’ vs ‘employment’ record (less 20%); labeled ‘very work-poor’ 55% less); ‘work-poor’ workpoor work-poor correspond intensities 20-59) considered ‘working calculation households’: characterized states) models tested (either depending hand) (idem) (together architecture spending) turns separate best fit18 study 2014) simultaneously states’ sure respects19 ILO-concept employment/non-employment underpinning poverty’) simply puts presenting eight 28 2011 students): (1) employment); (2) (3) 1); (4) index; (5) (6) (7) (8) stress conclusion holds pooled series hold ‘naïve’ pensions incomes EU coefficient -0 01 spending 2010 deep sections namely (indicator 1) 7) surprising establishes match needs deliver ratios (our visualizes mentioned earlier20 direct observations; introduces Before final thought abundant ‘in-work poor’ contemporary ‘work-poor confused ‘in-work-poverty’ he/she (financially) poor; ‘working’ minimal just before limited) mixes (is employed?) (what income?) makes intrinsically leading unwarranted conclusions unrelated intensity: cases individually counted belong; consequence irregularly (Marx So conceived pursue -which explores patternsmay popular polarization (2013 decompose children) factor) evident shed ‘fullemployment is: distributions? improving narrowing led drop respect result: 62 Also diversity between’ undone another saying faded away polarisation’; Conditioning (a) (b) single-adults (c) Gender main driver feminisation differentiate gender) expected’ same) Feminisation (more ‘homogamy’) translates Western-European Slovakia) Austria exerts never remain dominant throughout Comparing (DE FR IE VK) therefore ‘random’ others 29 frequent exacerbated ‘strong profiles’ (both immigrant enhanced Today’s couples: outlier Using qualification ‘insiders’ ‘outsiders’ summarize Three Belgium: resp couples; necessary elsewhere incentives created tax benefit systems conceivable stimulates Whether testing security entitlements Belgium) able ‘assortative mating’ features) Further potential links existing Goos (2009) Autor (2006) An linking analysed continuous opens in-work 1995 1999 2003 2007 full (panel a) b) c) hh multi-adult 1985 1989 1997 2005 2009 1996 2000 2004 2008 EU15 nonEU15 1986 1998 België Brussel Vlaanderen Wallonië ‘within household’ ‘between P 46 d) ‘between-household’ ‘within-household’ (reference brackets); BRU FLA WAL BE (women) (age 30-49) (medium skilled) EU-25 (born country) non-EU-25 (hh ≥ medium-skilled EU25 non-EU25 Note: GAE: GAEO: GAER: GAEOR: jl/individual jl hhousehold joblessness/individual G F ‘Measuring (in)security event unemployment: forgetting someone?’ Review Wealth 60: 77-97 E ‘Behind beyond (headcount) statistical papers doi:10 2785/42348 B (eds Reconciling Work Poverty Reduction How states? Oxford: Oxford University Press V analysis’ Methodologies Working Papers 2013 edition doi: 2785/41846 94-130 Atkinson -C Guio Marlier Luxembourg: Publications office (forthcoming) ‘Is perspective? immigrants’ Matthijs Neels Haers J Population Middle-East North Africa Beyond Demographic Divide Burlington: Ashgate Publishing (2000) ‘Does insurance crowd spousal labor supply?’ Journal Labor Economics 18(3) 546-572 R ‘The Growth Jobless Households Australia’ Australian Economic 133-154 M Europe’ Social Policy 413-431 D ‘Whither Britain US?: determinants work’ in: Blundell Card Freeman Seeking Premier League Economy Chicago: Chicago T Impact Earned Tax Credit Reforms Marriage Living Arrangements’ National 53(4) 1063-1106 Developments Brussels: C Canto O del Rio EU’ 1007/s11150-014-9248-7 57 ‘Two sides story: Royal Statistical Society: Series (Statistics Society) 171 857-875 (2010) ‘Reconciling workless Theory evidence 139-167 Manning Salomons ‘Job American 99 (2): 58-63 Marx I ‘In-work 131-156 S ‘Poverty worklessness Britain’ Econ 114 C1-C25 – indicator’ Improve Discussion Paper No 13/09 Antwerp ‘Mapping spending’ 1-59 ‘Child Risks Flanders: Accounting Worrying Performance’ Maystadt L Denayer Pestieau Van der Linden Cattelain Le modèle belge : quel avenir ? Presses interuniversitaires Charleroi Excessive Imbalances Performance Welfare Euroforum KULeuven (https://www kuleuven be/euroforum/viewpic php?LAN=E&TABLE=DOCS&ID=855)