Download fulltext
261

DISCUSSION PAPER SERIES DPS16 27 NOVEMBER 2016 How does early deprivation relate to later-life outcomes? A longitudinal analysis Ron DIRIS and Frank VANDENBROUCKE Public Economics Faculty of Business Diris* Vandenbroucke† Abstract Measures material are increasingly used as alternatives traditional poverty indicators While there exists extensive literature focusing on the impact that growing up in a (financially) poor household has future success little is known about how relates long-run outcomes This study uses data from 1970 British Cohort Study assess relationship between adult life We control for an set observable characteristics further employ valueadded generalized sensitivity nature this find diverse outcome variables but magnitude conditional relationships generally small Immaterial family quality show relatively stronger ties especially with respect non-cognitive skills Keywords: disadvantage JEL Classification: I32 J13 J62 *Department Maastricht University 6200 MD Netherlands r diris@maastrichtuniversity nl (corresponding author) †University Amsterdam KU Leuven Antwerp would like thank Erwin Ooghe Brian Nolan Geranda Notten Kristof de Witte participants APPAM conference London PE seminar their helpful comments 1 Introduction Classifications or social exclusion have traditionally relied measures individual income Material (MD) alternative indicator refer list ’basic necessities’ households different domains The increasing use these reflects perception captures more than lack Although depends what perceived basket necessities at given point time it essentially absolute measure contrasts commonly at-risk-of-poverty rate which relative positions within country become popular international intertemporal comparisons ‘Severe deprivation’ included target Europe 2020 strategy European Union (European Commission 2010) However contrast specifically related important analyzes confronted (BCS) follows total 17000 individuals born Britain first week April BCS reports information child its parents birth contains follow-ups multiple ages both childhood until age 42 It provides possessions circumstances well vast range several progress extensively extent raw correlations driven by associations other determinants Moreover we value-added developed Imbens (2003) address selection bias establish whether likely causal remains Using factor six 1The character should be interpreted nuance Poverty can also least context one when threshold anchored based arbitrary choice base year (in defined) principle possible construct countries Roelen (2012) constructing basis hazardous exercise 2 estimated two strands literature: studies analyzing (or complementary) background former group mainly focuses explaining mismatch being income-poor materially deprived2 encompassing items 3 Advocates ofMDemphasize benefits over strictly conceptual view (income neglects preferences risk factors) terms measurement (yearly volatile across prone error extremes distribution) shows into matters greatly There strong children later example educational attainment 4 Evidence adoption indicates variation families not solely due genes therefore ‘family quality’ crucial importance 5 still unclear specific aspects capture Studies composite socio-economic status (SES) typically combine parental education occupation home and/or linked those 6 Brooks-Gunn Duncan (1997) provide overview focus relation conclude (preschool school years) most strongly difficult empirically disentangle occu- 2See e g Perry (2002); Whelan et al (2004) 3Different methods elicit single such prevalence weighting principal component item response theory structural equation modeling no consensus For examples each approaches see Cappellari Jenkins (2006); Maˆıtre (2005); Tomlinson (2008) An provided (2010) 4See Corak (2013) intergenerational transmission OECD (2015) 5See Bj¨orklund Sacerdote (2008); Beckett (2006) 6See Bradley Corwyn (2002) pation neighbourhood rearing behavior etc Recent aimed uncover direct links Many role credit constraints type research finds short-term becomes limited best once factors achievement concludes permanent markedly liquidity (Heckman 2000; Carneiro Heckman 2003; Dearden 2004; Chevalier 2013) Still (permanent) Several exploited exogenous directly Frijters (2005) using sibling fixed effects combination event German reunification identifies low health while Løken Norwegian oil boom shock Other identify comparatively larger estimates substantially below simple suggest; Blanden Gregg (partially same (British) paper) Akee These results call question provision will lead substantial improvements prospects (2009) through evaluation EMA program students weekly cash transfers attendance staying clear alleviation because reduce opportunity costs Overall findings tend suggest correlation large part led researchers argue largely immaterial (see (2008)) explanation advocates often only imperfectly restrictions opportunities face As meaningful analyze either substitute complement Establishing existing emphasis put policy policies targeted reducing Identifying relations towards improve evaluations In general few key Filmer Pritchett (1999) exception conducting macro-level they link differences wealth (measured presence basic facilities drinking water electricity) Relying rich micro-level current various measured Additionally add addressing potentially confounding likelihood providing comparison sample paper organized introduce theoretical considerations Section describes methodological issues discussed presents empirical discusses robustness analyses 7 Theory Defining section discuss concept arise measuring constructs definition states “material refers inability afford consumption goods activities typical certain society irrespective people’s items” (OECD 2007) words concerns able ‘typical’ major broad characterization considerable exact construction Virtually all incorporate housing conditions More elaborate include access healthy lifestyle Since aim broadest sense since unexplored ultimately want affects developmental process child’s learning development (outside formal processes extra-curricular programs) additional domain define deprivation: possessional already suggested ‘material’ aspect always adhered make distinction ‘immaterial’ discussion reflected subdivision possession ambiguous contain tangible tools intangible support thereby divide sub-domain separately might alternatively thought cultural ‘capital’ thereof) 7One crime believe conceptually tied household’s living arrangements (which evidently captured domain) categorize under types goes beyond (often depending subjective interpretations) them things everyone ‘should have’ Ermisch makes similar his parenting inequality labels ‘what buy’ versus do’ consider light bigger matter simultaneously recognize aware comparing interpreting Measuring Another variable Data availability inevitably determines some any application criteria employed First clearly services case 1970s 1980s concern ‘enrichment’ available share population connotation leaves room interpretation specify constraint half could seen rather loose limit higher result affordability reason comprise questions distinguish having personal preference looks who bound own predominantly ranks value necessity never completely ‘irrespective preferences’ technically requires 8 made odd situated high-crime reasons do our main conduct stage where order final potential assessing considered natural consequence difference ‘concise’ rely very dataset Hence although lower bounds fact inexhaustible relevant viewed upper explain adulthood (we label ‘cohort members’) baseline 17196 10 16 26 30 34 38 9 suffer amount observations drops out fairly (86% 73% 26) waves 0 8The dependent adaptive feelings shame recognized Fusco (2011) addressed adapt Item Response approach correct differential reporting propensities 9The wave around administered tests following (i cohort members) school-level teachers principals taken verbal wide focal somewhat weighted specification supplementary Each Parental categorical (seven categories eleven 16) follow McKnight assigning midpoint band including obtained qualifications mental body mass index satisfaction gross net structure four variables: reading highest qualification income10 self-reported If missing impute next recent observation apply increases rapidly observe avoid leads 11 non-missing years established trends then calculate average express rank 100 10To confusion member’s serves serve ‘adult income’ ‘parental 11The mean values stable test scores intelligence math Questionnaires carried sets allow self-esteem locus person feels life) Rutter behavioral problems reported members latter Locus Estimation Measurement mentioned separate health/nutritional subdivided 50% inputs take affected (intermediate) dummy aspiration levels desired level particular performs Similarly exclude number friends visited same-aged peers assume choices carry ‘ambiguous’ determine fit relevance uniqueness measures) choose method explanatory power weight assigned inverse sample) cases includes Being deprived TV sources overlap automatically ensures much receive excluded altogether All standardized zero standard deviation 12 assessment purpose report completion presented Table A1 13 model estimate OLS model: Yi = 0+ 1Possi+ 2Housei+ 3Neighi+ 4Healthi+ 5EduMi+ 6EduIi+ 7Soci+X0 i+i (1) vector X0 employment 12The Cronbach’s alpha are: 801 700 640 554 545 eduational 447 13The defined unlikely (e appliances) chosen priors expected ex ante affect indirectly spend child-rearing style complete Appendix A2 inclusion ofX0 account outside emphasize controls effect operate spending tutoring classes) among When father’s without impacts parameter  Model represents classical term represent mechanisms 14 array cognition socio-emotional play mediating 15 Imputation To ensure enough observed 14We channels shaped grows 15See Almlund cognitive imputation W¨oßmann ‘fundamental’ (labeled F) fundamental virtually birth; gestational mother’s ethnicity wedlock gender was hospital-born M (Mk) (Mj ) regress Mj F coefficients regression Mk Further dummies indicate imputed Results correlational regresses isolation step jointly subsequently signals chances grow obtaining favourable (including domains) informative evidence holding constant reflect unobservable issue Sections portrayed graphically Figure figure (Model 1) finally additionally specifications Tables A4 A6 detail Main estimation Reading shown left quadrant A3 Not surprisingly strongest increase reduction 29 remain statistically significant coefficient suggests per With last rows marginally reduces appear mechanism Educational mem- ber (different of) distinguishing significantly associated decrease corresponds 25 deviation) Including severely longer full (mainly income) -0 22 drive previous exist high Achievement appears (both noncognitive cutoff degree attributed comparable Among GCSE A-C connections Dummy end distribution weaker Adult during right A5 smaller here margin (‘separate’) ranking percentiles added decreases percentile compared initial Controlling Part operates mimics 42) incomes highly consistent Health Interestingly dominate Social 136 five-point scale (and initially association hardly (observed) change (once occur questionnaires predate physical prominent selective controlling class) Similar mediate Non-linearity assumed now linear worthwhile explore extreme need reach before nonlinearities subsection estimating polynomials 17 non-linearity apparent Comparing inhibits non-linear tendencies quadratic positive indicating negative diminishing possibly skewed implies fits hous- 16See Kessler (2007) 17These request ing interesting (especially) severe non-linearities certainly involve sign reversal surprising begin attenuated Finally interaction complementarity neighbourhoods vice versa assessed noteworthy summarized A7 Coefficients (the size well-being) (immaterial) Estimates Furthermore acts rooms house highlights persistence proper Body Mass Index (BMI) BMI allows positively negatively 18 appliances microwaves) less diet lag adolescence Value-added Because points lagged earlier achieved X0) growth lags contrasting remarkable electronics bedroom 19 opposite line revealed score exhibit unobserved investments identified biased influence (GSA) extended Harada continuous unobservables required 18We speculate underlying One watching complementing subjects (Borzekowski Robinson 2005; Sharif ways 19See Cunha (2010); 20 insignificant plausibility parameters partial R2’s needed render combined plausibly On hand if away plots plausible cases: straightforward so (unobservable) weak GSA produces emphasized variance explained plotted graph conservative addition 20One 21 implausible even though (very slightly) X adding classroom peer R2 01 017 representative curve area Explanatory valuable joint marginal reveals uniquely explains (extensive) student 21Additionally statistical significance (at 10% level) majority conclusion condition lines above minor means (especially self-esteem) compare ‘gross’ ‘net’ look reversed) naturally shares figures confirms adverse better essence argued bands subject Keeping mind portray restrict overallMDare Correlations slightly perform (an imperfect eliciting at-risk 23 modest third fourth column split pattern indicator; subset predictive second except again dominance (strictly) (likely attenuated) ‘traditional’ weakest characteristics) worth noting lose multidimensional advantages unique seven con- 22One focused bottom distinguishes (its close normal long tails side) topcoded 24 siderably Differences constructed too robustly age-effects Robustness assumptions relax Bad problem styles divorces downwardly 10; 5; ‘bad controls’ Deprivation incorporating Nonetheless panel B contribute overall downward influenced Conversely excludes controlled Affordability count executed belong (these unconditional Only looking simply owning connection (possessional) Different specified restriction cannot 25% 15% tighter remove 20% 40% respectively fall changes 25-50% 15-50% restrictive almost Sensitivity thresholds gradually three models Endogenous items? present misleading way visits museum library plays musical instrument partially interest (partially) parent-child member club Especially expect however lies A9 proportional ex- 23The restricted subdomains: neighbours financial responsible 28 ercise interpret care truly resulting (also) personality independently state representing (perceived) environment surrounding Attrition heterogeneity disappear attrition non-random Those differ Most prominently male (58 4% vs 49 4%) non-native (16 8% 6%) (12 2% fully period external validity may turns boys out-of-wedlock None similarly moderate loss representativeness described employing conventional identifying attenuation interactions corresponding (thereby allowing intercept slope respective variable) A8 approaches: applied sum binary remarkably (commonly used) relating sizes Judging Financial hardship incorporated lacking namely ability unexpected expenses arrears bills ask were troubled past Conditional percentage Conclusion experienced reveal Plausible adds (mental) analyzed ‘deprivation’ form (a capital (conditional) 31 diminish Previous huge Our isolated contributor disconnect ‘material state’ causally supportive McLanahan Bianchi educated fathers mothers contributors Research low-income improved changing guidance beliefs relief (Kautz 2014) ‘immaterial (imperfectly measured) At invalidate alto- 32 gether Basic (although groups improves identifier limitations automatic taking life-time perspective changed inevitable linking achievements causality likely) exploitation elements help segments Future precisely greatest challenge exactly why obtain widely References R K W E Copeland G Keeler Angold J Costello Parents children’s outcomes: quasi-experiment American economic journal Applied economics 2(1) 86–115 L Duckworth T D Kautz Personality psychology Handbook Education Volume pp 1–181 Amsterdam: Elsevier 33 C Maughan Castle Colvert Groothues Kreppner S Stevens O’Connor Sonuga-Barke Do persist adolescence? Findings English Romanian adoptees Child Development 77(3) 696–711 P Milkie Changing Rhythms Of Family Life New York NY: Russell Sage Foundation Lindahl Plug (2006 August) origins associations: Lessons swedish Quarterly Journal 121(3) 999–1028 attainment: review Oxford Review Economic Policy 20(2) 245–263 Borzekowski N remote mouse pencil: media academic grade Archives Pediatrics & Adolescent Medicine 159(7) 607–613 H Socioeconomic Annual 53(1) 371–399 55–71 Summarizing ISERWorking Paper Series 2006-40 Human Krueger Friedman (Eds Inequality America: What Role Capital Policies? 77–239 Cambridge MA: MIT Press Harmon V O’Sullivan I Walker schooling IZA Labor 1–22 Income equality mobility Perspectives 79–102 Schennach Estimating technology skill formation Econometrica 78(3) 883–931 Emmerson Frayne Meghir dropout rates Resources 44(4) 827–857 McGranahan Sianesi Credit Constraints Choices: NCDS BCS70 CEE DP 48 ERIC Origins immobility inequality: National Institute 205(1) 62–71 2020: smart sustainable inclusive 35 Population 25(1) 85–120 Haisken-DeNew Shields health: german 24(5) 997–1017 -C Guio Marlier Technical CEPS/INSTEAD Generalized NYU working (2000 March) Policies foster human 54(1) 3–56 (2008 July) Schools synapses Inquiry 46(3) 289–324 (2003 May) exogeneity 93(2) 126–132 Diris Ter Weel Borghans (2014) Fostering skills: Improving promote lifetime Bureau Amminger Aguilar-Gaxiola Alonso Lee Ustun Age onset disorders: Current opinion psychiatry 20(4) 359–364 education: norwegian experiment Labour 17(1) 118–129 Downward hoarding glass floor Diverging destinies: faring demographic transition Demography 41(4) 607–627 non-monetary exclusion: Europe? Analysis Management 29(2) 305–325 new tool monitoring (child) poverty: cumulative Indicators 5(2) 335–355 Society Glance: 2006 Edition Organisation Co-operation Together: Why Less Benefits Paris France: Publishing Zealand 101–127 Nature nurture learned twins adoptees? North- Holland 36 Wills Sargent Effect visual performance: prospective 46(1) 52–61 Williams 1991 2003 37(04) 597–620 Layte Understanding dynamic comparative Sociological 287– 302 Vulnerability perspectives europe: latent class Societies 7(3) 423–450 equal opportunities? US CESifo Working 1162 37 1: Possess House Neighbour Educ (M) (I) − att −3 −2 −1 05 Raw C+M Notes: portrays (‘Raw’) (‘C’) (‘C+M’) ‘C+M’ horizontal bars 95% confidence intervals equally spaced averaged expressed 2: Partial EV −−> HE HO (partial R2) (‘EV’; deprivation) (D) (X0) Figures (C (I)) (all row) (P) (HE) (HO) 39 3: Math Self−esteem Mental Controls Domains (vector regressing 40 Possession Housing Neigh Panel A: VA Reading16 035* 003 001 030*** 050*** 107*** (0 021) 018) 010) 011) 012) LC 051*** 016 006 018* 029** 054*** 002 019) 016) 009) Reading10 020** 026** 138*** 008 014) 031** 019* 021** 110*** 005 013) Math16 066*** 014 059*** 049*** 093*** 024 025) 020) 015) 040* 019 010 038*** 027** 030* 022* 017) Rutter16 007 056*** 028*** 032*** 046*** 065*** 047*** 025*** 035*** 024** 028** 044*** Rutter10 036** 019** 070*** 020 009 048*** 008) Locus16 042** 027 041*** 023) 038** 015 023 037*** Self-esteem16 030 026 033*** 047** 161*** 029 046** 157*** B: Reading5 077*** 034*** 004 111*** 011 Math10 057*** 022** 018** 042*** 121*** Rutter5 053*** 026*** 027*** Locus10 023** 045*** Self-esteem10 025** table ‘LC’ (taken outcomes) regressions ‘Rutter’ ‘Locus’ internal 41 No Y (all) 801*** 818*** 434*** 594*** 179*** 295*** 127*** 249*** 035) 039) 046) 043) 042) 038) 29*** -7 26*** -5 20*** -3 30*** 66*** -2 51*** 999** 34*** 396) 399) 436) 440) 484) 450) 439) 382) 154*** 113*** 087*** 055*** 242*** 218*** 172*** 042* 024) 022) 301*** 335*** 176*** 244*** 131*** 124*** 279*** 307*** 213*** 079*** 149*** 100*** 082*** 029) 027) 026) 130*** 197*** 123*** 119*** 026* 072*** 129*** 258*** 114*** 231*** 041 219*** 184*** 898*** -1 25*** 468*** 27*** 291 790*** 240 937*** 138) 157) 162) 161) 193) 194) 184) 169) (Y) Effects estimations Possession16 Possession10 Possession5 Housing16 Housing10 Housing5 036 035 032 033 028 036) 031) 192 342 951*** 872*** 783*** 158 311) 395) 329) 245) 246) 253) 012 039* 034* 025* Age-specific 4: Exclusion 209*** 418*** 13*** 312) lim 09*** 288) 137*** 82*** 028) 317) 288*** 730*** 57*** 136*** 301) 064*** 297) 208*** 43*** 306) 264*** 513*** -4 03*** 112*** 310) 463*** 76*** 098*** 279) 043* 062** 23*** 305) 203*** 65*** 030) 341) 205*** 50*** (taking together) 43 5: Lack (A) (B) 426*** 423*** 0024 025 49*** 31*** 022 48*** 729*** 018 286) 307) 275) 0093 089*** 0098) 0097) Add 210*** 054** 104 196*** 075*** 101 77*** 574** 072 63*** 118 069 303) 283) 273) 017* 0065 040** 00011 0095) 059** 235 034 434 740*** 225 405 188 224 295) 265) 298) 251) 0027 013 054 0058 053 0055 248 0098 00018 0091) compares afforded 44 6: maximally allowed (main) 083*** 031 088*** 061** 231 790** 922*** 537** 630*** 310 718*** 010*** 324 333) 272) 228) 232) 249) 271) 047 103*** 237 067** 085*** 060*** 226*** 229 416 826*** 507** 433* 303 674*** 615** 321 313) 229) 233) 239) 262) 250) 036*** 236 038 181*** 226 771** 648** 120 399* 762*** 178*** 257 319 319) 269) 237) 231) 260) 256) 016* 031*** 045 099*** 232 top 45 A1: variance: (reversed) 46 Items • Possessions: (does own:) refrigerator washing machine dryer car phone video recorder camera stereo radio PC sewing vacuum cleaner microwave 16); freezer holiday 10); 5); child: cassette player bicycle Health: eats meat fish times breakfast lunch iron vitamins milk fibre carbons sugar intake 10) Neighbourhood: noisy graffiti youth loitering streets drunks rubbish street victim beak-in unsafe night (age 5) Housing: bathroom indoor toilet hot garden kitchen bed difficulties heating moisture untidy furniture (material): played books studying newspapers calculator constructional toys (immaterial): read Social: participate activity (excursions charities concerts) talk rarely participates misses money organization 47 A2: Control Birth controls: abnormalities hospital head circumference mother married father Household income: eligibility free class: employment: works (averaged hours worked work experience structure: people older siblings younger Parenting style: attitude toward independence authoritarian world smoking (ever) pregnancy heavy A3: Separate 128*** 169*** 289*** – 044** 096*** 193*** 115 000003 039*** 181 000 020* 167*** 021* 132 081*** 170*** 144 Parent empl 041** 029*** 091*** 198*** 119 031* 188*** 141 090*** 165*** 133 Non-cog (NC) 249 (WC) row regressed isolation) Rows Row See contained A4: 560*** 449*** 260*** 451*** 707*** 433*** 271*** 140*** 162*** 486*** 106 261*** 171*** 456*** 198 074** 056** 057** 135*** 390*** 101*** 147 153*** 055** 126*** 399*** 159 239*** 147*** 462*** 104*** 092*** 148*** 160*** 425*** 142 187*** 071*** 132*** 394*** 137 058*** 192*** 080*** 243 073*** 045* 301 078*** 141*** 429*** 128 194*** 242 037* 50 A5: 17*** 12*** 81*** 04*** 89*** 299) 282) 254) 266) 278) 268) 700*** 540*** 392*** 158*** 576*** 454 071 353) 255) 264) 308) 583*** 971*** 956*** 851*** 238*** 276 321) 284) 281) 651*** 097*** 611*** 383*** 153 354) 309) 261) 901** 889*** 798*** 088 099 371) 314) 276) 552*** 300*** 929*** 462 090 358) 277) 285) 381*** 592*** 537*** 419 364) 315) 287) 562*** 570*** 527*** 374*** 368 360) 052*** 469*** 376*** 255*** 016*** 964*** 326 084 357) 684*** 314*** 838*** 544* 116 129 348) 304) 639* 726*** 516** 270 077 600** 976*** 363 326) 221) 242) 243) 572*** 612*** 371*** 182*** 402*** 052 083 352) 292) 792** 857*** 401* 476** 222 526*** 549** 332 332) 270) 650** 710*** 204 113 563** 699*** 366 description 51 A6: 108*** 125*** 0099) 0092) 0093) 040*** 030** 068*** 037 069*** 044 048 059 061 Subjective 52 A7: Arrests34 007) 006) health42 300** 117 066 262*** 105 709*** 147) 121) 095) 098) 101) 122) 103) satis34 043** 150*** 120) Finances42 024* BMI42 276*** 190*** 345*** 057 091) 073) 059) 063) 064) 078) 065) 278** 221*** 042 115) 099) 060) 076) 071) BMI16 040 046 048) 032) 053) 060* 060 067* 275 047) BMI10 151*** BMI42) ‘BMI’ ‘Arrests34’ arrests ‘Mental health42’ Warwick Edinburgh well-being ‘Finances42’ situation 53 A8: Comparison (educational attainment) Factor PW 332*** 191*** 357*** 122*** 093 Sum 292*** 517*** 118*** 086 Binary 633*** 129** 982*** 199*** 057) 056) 058) 095*** 049** 224*** 138** 476*** 130** 065 228 061) 055) 054) (baseline) ‘Factor’ ‘PW’ ‘Sum’ sums applies ‘Binary’ 0/1 75th labeled deprived) 54 A9: Excluding Exclude 062*** 166*** 227 000*** 573** 714*** 286 320 334) 230) 248) 234) 021 partly member) exclusively 55 Copyright © @ author(s) Discussion papers draft distributed purposes comment reproduced permission copyright holder Copies author