Child poverty: what can social spending explain in Europe?

Download fulltext

DISCUSSION PAPER SERIES DPS14 20 JULY 2014 Child poverty: what can social spending explain in Europe? Ron DIRIS Frank VANDENBROUCKE and Gerlinde VERBIST Public Economics Faculty of And Business By Diris Vandenbroucke Verbist November 23 Abstract This study assesses the role relation to child poverty European welfare states Using macro-level panel data from EU SILC 2005-2012 we analyze effect size how those benefits are targeted We separately estimate pension on as prevalence multigenerational families makes them a relevant income source for with children especially Southern Eastern Estimating GLS model including time country fixed effects large set characteristics find that both cash transfers pensions substantially reduce Increased pro-poorness also leads lower rates but sizes more modest by comparison strongly depend targeting is defined The estimates change little across various specifications obtain similar when use regional variation within countries assess same Where explains share over explanatory power respect cross-sectional limited complete does disparities sizable unexplained remains disparity likely relates factors invariable KU Leuven Department (corresponding author; ron diris@kuleuven be) University Antwerp Sociology 1 Introduction At-risk-of-poverty have received increasing attention national governments Union academic scholars alike Poverty vary evolutions very different well Figure reports household among below age 18 years 2008 2012 figure shows indeed substantial not only between old new member each Furthermore affected differently recent economic crisis changes majority Since these relative periods recession do necessarily increase As such lack strong could occur because had homogeneous market Alternatively automatic stabilizers mitigated heterogeneous impact earnings losses executes analysis determinants thereby mainly focusing 29 2005 using identify has reducing which limit itself at working Pension an important households several magnitude indicate average percentage point mean equivalized disposable reduces rate around 0 265 higher towards poor associated measure or inequality Additionally influenced individual employment distributed One goal this extent 2 be confounded other related robust sets variables taxation technological globalisation education demographics labour institutions provided one controls differences results alternative estimation methods (provided include unit effects) based assumption least partially country-wide eligibility rules regions median will receive effectively rich bene- fit within-region latter directly level region valid control may affect internal relies assumptions than country-level determination complex process result interactions multiple operate directions Previous research policy identified demographic indicators Many findings crosssectional focus their (i e pretransfer pre-tax) total redistribution through taxes However unobserved cross-country confound Analyses informative they counterfactual situation arise reality In particular g incentive ultimately interested final outcome none evaluated strictly isolation 1See Osberg et al (2004); Smeeding (2005) overviews 3 OECD report ’Divided Stand’ (OECD 2011) conducts elaborate uses post-transfer -tax measures outcomes conduct empirical nature our assessing terms still take macro-economic forces skill into account cannot executed without considering Still primary objective addressing Thereby population High alarming there links family resources early child’s life later school performance put vicious circle consistent paper organized follows Section summarize literature after present theoretical framework 4 discusses descriptive statistics 5 presents methodological approach main discussed 6 while 7 robustness 8 concludes Literature analyzes specific (income) been studied perspectives fields There extensive field economics increases common most industrialized since 1980’s shares industry sectors 1990’s popular explanation 2See Duncan Brooks-Gunn (1997); Heckman (2000) was skill-biased (SBTC) refers shift educated workers An overview Autor Katz (1999) competing routinization hypothesis support last decade idea technology largely replaces routine jobs middle distribution first proposed (2003) appears better linear levels According Goos Manning (2007) ‘job polarization’ third rise log(50/10) wage differential Britain Research political economy focused government generally either Gini coefficients (relative) show overlap Extensive reviews (2004) studies rather Studies ‘final’ predominantly These major predictor Heady (2001) specifically They (cross-sectional) EHCP reduction Corak Chen (2008) look former so EUROMOD microsimulation finds occurs decomposition (based threshold time) 3Evidence found US (e (2006)) (Goos 2007) Germany (Spitz-Oener 2006; Dustmann 2009) (2009) U-shaped pattern low- medium high-skilled almost all nations 4The largest aspect governmental Bradley Moller two seperate 5They statistically significant tax progressivity public sector supply college unemployment unionization degree collective bargaining left-wing basis LIS cohorts conclude impacts differ conducted comprehensive systematically influence dispersion form prominent offset period 1985-1995 reduced redistributive capacity driver 10 thereafter Taxes play much smaller explaining trends Theoretical Concept Our focuses at-risk-of-poverty conceptually alternatives coefficient 90/10 ratio First its half A doubling above would no it significantly upper affects base consumption investment etc movements less weight if consider lot U S Europe caused gain top 1% (2011) key 90/50 percentile 50/10 heterogeneity 6See Immervoll Richardson detailed discussion 7See Atkinson Kenworthy (2014) consequences right functional some When relevance might puts bottom second characteristic group lives reflected needs gap capture dynamic employ 60% determined year means opt position rest Determinants number difficult disentangle Systematic frameworks frequently used analyses helpful understanding mechanisms behind possible threats identification bears resemblance analytical (p 27) stronger connections whereas switch before ) typically distinction pre-tax pre-transfer depicts Market essentially hours worked (which includes 0) hourly wages sources capital) shown given weak suggests dominant Because looking at-risk-ofpoverty translates Hence formation another crucial factor It direct mechanical influences equivalence scales calculate equivalent indirectly example force participation decisions Each components turn Hourly addition (presumably equalizing) centralized unions minimum On hand negative Macro-economic growth (Kenworthy (Autor 1999) Demographics determinant already referred dependency ratios therefore Governmental many forms centers collected whom Demographic Eligibility often family-level elderly come expense working-age 8See (2005); Jacobs Gornick (2002) Interactions redistribution; act substitutes complements Governments tailor strength protection legislation existence adjustment work ways illustrated two-way arrow signifies framework: go upwards reverse horizontally argues creates need society 9 fact respond link contains indirect complicate indicated dashed lines Most prominently low lead extended risks 11 causality Inequality skills intergenerational educational negatively 12 Finally ‘third level’ Globalisation 9Similarly high protective 10A review Krueger Meyer insurance indicates incidence duration security programs length employees spend out Bertrand increased generosity South-Africa adults Also see Bergh thorough 11Engelhardt living arrangements widowed divorced Again mechanism 12The debated Benabou (1996) summarizes evidence relationship Forbes opposite Barro positive migration; single-parent skills;13 individuals attract structure determine where interest concentrated scheme illustrates why assessment dif- ficult simply decomposed versus Employers anticipate others actions Moreover instruments transfer investments active policies setup Although incorporate multitude imply controlling get indication importance Data micro-data Statistics Income Living Conditions (EU SILC) Questions apply previous 2004 2011 UK Ireland exceptions adhere current difference grouping non-SILC under T+1 dependent variable measured 13Moon (2012) provide cognitive emotional stimulation future domains Equivalized incomes constructed dividing scale; follow practice modified scale attributes adult 15 younger described Table A1 14 section ingredients Spending (equivalized) transfers15 represent country’s microdata allows us subgroups Pensions want 16 Through retrieve exactly end up type benefit sample ranking traditional reasons Eastern-European whole systems 14We tested additional replacement assortative mating foreign table describes 15When refer ‘transfers’ throughout remainder excluding 16Increases general should Empirical confirms this; indicator skewed 17 tend relatively polarized although correlation sum insignificant give insights decrease kept check during severe continuous For biggest surge moment hit hardest rising face seen Pro-poorness Aside label ‘pro-poorness spending’ captures Following Matsaganis (2013) concentration (CC’s) Concentration coeffi- 17Figure differs depending whether aged 0-17 0-59 appear linked intensity subpopulation; household-level subpopulation 18Social changed decreases steady decline cients indifferent multiplying scalar coefficient) 19 To CC’s subgroup equal older then rank according conceptual similarity reason employed part avoid fail pick three pro-poorness: calculated distribution; deciles lowest (labeled 0-20) five 0-50) (following Marx (2013)) Deciles post-tax Taxation Unfortunately reliable measuring contributions distinguish 19Korpi Palme (1998) concept units gross who demonstrate yields broadly 20The ‘pro-richness’ convert values reflect standardize standard deviation multiply -1 21We define own effectiveness post-measure separate decile construction being virtually cases 95% observations execute comparative purposes 13 paid self-reported Therefore rely reported Eurostat method developed ‘Taxing Wages’ compositions 22 prog) implicit (ratio compensation employees) construct projected single person Other lists Employment potentially correlate status matters record work-intensity (between per cent); ‘very poor’ 55 cent less); ‘work one-earner poverty’ shortcut work-poor ’severe combination provides 22See latest methodology 23The retrieved Statistical Books ‘Taxation Union’ (formerly ‘Structures Union’) See strongest 24 jointly larger Lohmann biased collect surveys registered (the Scandinavian Netherlands Slovenia) He marginal regression further case Nevertheless attenuated presence measurement error sensitivity excludes (groups of) collection matter instances information observation impute missing estimated trend non-missing raise econometric issues dropped any (ALMP) Iceland Malta note irrelevant 25 Methodology pooled wide variety fundamental question and/or 24The allow calculation R2 correlations predicted observed (corrected degrees freedom) OLS predictive specification 25We concerns relate term nonstationarity leading Baltagi (1995) Kittel Winner Beck depends aspects testable tests listed A2 Based correct groupwise heteroskedasticity serial specifying heteroskedastic uncorrelated AR1 autocorrelation models conclusions remain intact neglected immediately alter system persistence Unit root reject context 26 known (Maddala Kim 1998; Maddala Wu fully rule lagged comparatively (0 190) (FD) (LDV) (Ci) (Tt) effects: Demo + Ci Tt (1) i identifies t denotes test Levin here Im suited panel-specific structures (AROP) Equation represents full (S) (T) (WI) (Macro (Skill) (LM) (Demo) added versions confounding groups Results now start educed add plus stepwise manner analyzing All summarized Social four rows requires (work) collinearity Failing bias (not shown) Once inclusion country-specific (linear) dummies 2009 (2010 27 net adverse column (6) (to -0 138) 5% Assuming drives underestimation potential strong; points somewhat sensitive Presumably response perceived correlated compare (2) connected (multigen) Extended independent proportionally incorporated marginally multigen sign resource-sharing component exclude 27These available request lie close slightly 28 euro simultaneously lift people counted observe ‘transfer-heavy’ ‘pensions-heavy household’ roughly amount evaluating keep mind ‘pension-heavy’ Thus actual Main standardized 50/0 0-50 Tax Progressivity 167/50 respectively 28It labour-supply (Krueger 2002) (Bertrand 2003) about modest; 435 170 442 Adding columns (3) Changing range Variables exhibit going 50% unreliable 10% holding constant Note 100/50 choose highest applies pro-poornerness CC 29In dominated few definition match scenario 0-20 gaps defines concern deducted (as distinguished income) partly drive anymore Korpi comes Correlational Figures A3 figures portray inverse positively contrast line 2011; 2013) discrepancy reflects decided rare low) 21 claims tradeoff contrasts obtained cient pre-crisis (SILC 2009-SILC 2012) geographical areas 30 originates earliest waves inter-temporal weakening longer 31 move back direction contract suggest constantly changing short Further frame needed broader perspective Additional Here briefly discuss contain suggesting (around points) expressed deviations 30When area Central 31The becomes 2007 hence advent (5) (including attainment) 32 Column GDP highskilled medium-skilled job responsible decades business expenditures R&D imports No inward outward (4) attainment (but definitions) 33 even poverty) short-term long run become integrated 34 female self-employed care services 35 were centrality Excluding Multiple 32Not included practical 33The here; 34High dropout short-time starts earning his 35The conditional overall signals contribution women examination possibly inconclusive adding institutional serves good leaves unchanged modestly adds Migration non-EU Higher young shared members; implies earn (unequivalized) stay parent 36 Decomposing exploiting interesting purpose simple step subtracting product coef- ficient country-mean sample-mean remaining 36A ‘new’ 37 split formerly communist regime did every belongs prevents bars emphasize simplified exercise 38 assumes evaluate captured pronounced presented structural invariant design typical holds true reveals black coloured side x-axis baseline bar highlight (part deviating rates) (conditional) Ireland) Starting ‘Western’ (Figure 5) aboveaverage explained 37With exception certain benchmark 38Although Spain Greece Italy Ireland’s UK’s above-average lesser Among records Denmark Finland Norway below-average Sweden Accounting propoorness margin ex-communist comparably Low Slovenia Slovakia excommunistic Czech Republic Hungary ’below average’ again Bulgaria’s accounting offsets BERD If isolate together macroeconomic excommunist shrinking infrastructure worker productivity spillovers equality led 39 Part human capital infer probably (also ‘more explain’ rate) (it 2% average) (for children) vice versa showed distribute within-R2 45 40 39The equals 63 40Since intensity; 23% 29% Robustness Model regional-level inhibit Region Analysis macro-data eight distinguishes same) cohort (Austria Belgium France Poland Spain) Six exact (Finland Romania UK) identifiers consistently merge (geographically adjacent) too individually consists 214 ensure 500 wave lies 000 Relying country-analysis approaches apart having composition (size vocational imports; derived entered NUTS between-region interaction (predominantly) Variation conditions region-specific Essentially (partial) homogeneity ex-ante ensures richer ‘transfer’ poorer Given exploit occurrence inter-regional grasp rankings resulting small 41 1: parents workless full-time 41Including errors employing gives implicitly Such (unobserved) serious threat (country-level) regionspecific row withinregion due relying micro-samples precision next comparable controlled B reproduce geographically adjacent 182 betweenregion interpreted Either accurately consistency furthermore require observable Sensitivity address FD LDV lose country; executing 179 carries moderate illustrate neglecting picture dramatically A4 jackknife drop maximum volatile driven cyclical circumstances 42 choosing minimal members (large) fall far effective lifting Alternative A5 illustrative caution Both 42This floating en instead frames exist exacerbated subtracts lowers overestimation poverty; (absolute) disincentive naturally risk ending absence however deeper 40% 18) Interaction timeframe (Table 4) enough statistical dimensions emerge mentioned pre- post-crisis non-existent earners administrative produces 43 weaker Naturally designed protect spread sufficient completely upon supplementary works substitute other; correlates done assessed 43These (-0 218) (- 293) Conclusion Reductions ceteris paribus More taken constitute similarly ‘pension-heaviness’ crowding-out challenge improve preventing shortterm currently light Korpi-Palme paradox (Korpi 1998) reversed; (cash transfer) pro-poor intertemporal incorporates explore subcomponents Controlling 30% (of attributed 46% incorporating weakens automatically de- mographics limitations unobservable lends validity ideally Future hold covers reliably whose best identifying longitudinal exogenous though References T Piketty E Saez (2011 March) Top history Journal Economic 49(1) 3–71 D H L F Handbook Labor Chapter Changes Wage Structure Earnings pp 1463–1555 Elsevier Amsterdam M Kearney (2006) polarization labor American Review 96(2) 189–194 Levy R J Murnane skill-content change: investigation Quarterly 118(4) 1279–1333 Econometric Panel New York: Wiley (2000 September) Growth 5(1) 5–32 N Modeling dynamics time-series-cross-section Annual Political Sciences 331–352 Ben Bernanke Rotemberg (Eds NBER Macroeconomics 11–74 Cambridge MA: MIT Press problem state research: How redistribution? Sociological 21(4) 345–357 Mullainathan Miller familes: Evidence south africa World Bank 17(1) 27–50 Huber Nielsen Stephens Distribution postindistrial democracies Politics 55(2) 193–228 W -H Demography 45(3) 537–553 C Lietz Sutherland european union Discussion 1589 IZA Bonn G (1997) Consequences Growing Up Poor Russell Sage Foundation Ludsteck Schonberg Revisiting german 124(2) 843–881 Engelhardt V Gruber Parry arrangements: notch Human Resources 40(2) 354–372 union: Technical Brussels K reassessment 90(4) 869–887 Lousely lovely jobs: britain 89(1) 118–133 Salomons Job europe 99(2) 58–63 Mitrakos P Tsakloglou distributional echp Fiscal 22(4) 547–565 Policies foster 54(1) 3–56 Pesaran Y Shin Testing roots panels Econometrics 115(1) 53–74 Redistribution countries: What decades? Hours dual earner couples: united cross-national Focus 35(2) 169–187 Progress Oxford UK: Inequalities Societal Impacts Thirty Rich Countries United States: Rapidly-Rising economy? globalization-welfare nexus revisited 44(2) 269– 293 strategies equality: Welfare western 63(5) 661–687 Auerbach Feldstein Volume 2327–2392 -F Lin -S Chu data: Asymptotic finite-sample properties 108(1) 1–24 Comparability EU-SILC survey register Policy 21(1) 37–54 I -M Roots Cointegration Structural Change Bulletin 61(S1) 631–652 Salanauskaite revisited: peace? Papers 7414 advanced capitalist 68(1) 22–51 Moon (2012 July) Time invest disadvanted Samsung Institute 5(3) 50–59 Divided Stand Organisation Co-operation Development Taxing Wages Paris France: Co-Operation Schwabish Expenditure: Theories Effects 821–859 Russel York General diagnostic cross dependence Working 0435 Science 86(S1) 955–983 Spitz-Oener tasks demand: Looking outside 24(2) 235–270 Reconciling Redistributive Capacity Services Europe: At−risk−of−poverty AROP Notes: England 2: portrays ALMP = EPL 3: vs ESSPROS BG RO LV IT GR LT MT PL EE SK CZ PT SI HU ES FR IS CY AT SE DE IE NL LU FI BE DK NO (administrative) (working benefits; WACB) WACB sickness disability non-elsewhere classified survivor % ranked 44 4: Trends Size Transfers (AROP; (expressed 5: Decomposition non Deviation Baseline PP LM demo Macro (Employment) (PP tax) (LM demo) (Macro) value Bars help 6: 47 Work Severe prog Sector Education Self-empl Young Significant **Significant ***Significant Transfer (PP) ‘PP transfers’ pensions’ ‘Tax rate’ (Tax divides 167% (single description capita (both patent applications 48 A: Share B: definitions converted necessary entry (CC) ‘CC half’ ‘Share 0-20’ deciles; ‘share 0-50’ entries percentiles (pre-transfer) (where earnings) (all employees’ 49 Sample Within Between express employs (except GDP) excluded additionally Appendix tables A1: non-elderly Difference child−adult (work 55%) averaged A2: Correlation pro-poorness; A3: Description poor: 55% Individual months spent corrected students summed divided (age 18-59) Source: (refers calculations table) GDP/capita: domestic purchasing parity growth: structure: (legislators managers professionals technicians associate professionals) (craft trade plant machine operators assemblers skilled agricultural fishery service shop sales clerks) expenditures: expenditure development comprise creative undertaken systematic order stock knowledge man culture devise Imports (at prices) per-capita 275 dollar 2010 China) Education: (ISCED 5-6) 0-2) Calculated Vocational education: pupils secondary enrolled stream Self-employment: Female participation: 59 Use care: center-based Migration: born reporting nation dependency: 0-14 15-64 p-value Heterosk Autocorr CSD stat LLC demean p-values for: F-test Wald (Greene 2000: 598) (Baltagi 2001: 95) (Pesaran 2004) H0 test-statistic (stat) adjusted statistic 54 Lagged Base FE Prais-Winsten Panel-spec corr Lag ‘FE’ ‘LDV’ ‘Prais-Winsten’ transformed ‘Panel-spec AR1’ ’No ’ ‘Lag sample’ A4: Drop 2006 Jackknife Min Jackknife: Max Scan eq discard variation) Small Cyprus Luxembourg (Scan √ n 56 A5: Estimates Pre-transfer 40TH progr applied ‘Poverty ‘poverty rich’ Coefficients multiplied 100 57 A6: Crisis IA Ex-comm pov Trans* Pens* Tax* WP VWP Prog (Base) (IA) interact 2009-2012) (ex-comm (high (Trans) (Pens) (WP) (VWP) (Prog) 58 A7: Summary Implicit Multigen Austria Bulgaria Rep Estonia Latvia Lithuania Portugal Un Kingdom summary (after taxes) 18-59 Tables Copyright author(s) papers draft comment reproduced permission copyright holder Copies author