State vocational rehabilitation programs and federal disability insurance: An analysis of Virginia's vocational rehabilitation program

Dean, Da id; Peppe , John V.; Schmid , Robe M.; S e n, S e en A icle S a e oca ional ehabili a ion p og ams and ede al disabili y insu ance: An analysis o Vi ginia's oca ional ehabili a ion p og am IZA Jou nal o Labo Policy P o ided in Coope a ion wi h: IZA – Ins i u e o Labo Economics Sugges ed Ci a ion: Dean, Da id; Peppe , John V.; Schmid , Robe M.; S e n, S e en (2014) : S a e oca ional ehabili a ion p og ams and ede al disabili y insu ance: An analysis o Vi ginia's oca ional ehabili a ion p og am, IZA Jou nal o Labo Policy, ISSN 2193-9004, Sp inge , Heidelbe g, Vol. 3, pp. 1-19, h ps://doi.o g/10.1186/2193-9004-3-7 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/154681 S anda d-Nu zungsbedingungen: Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen Zwecken und zum P i a geb auch gespeiche und kopie we den. Sie dü en die Dokumen e nich ü ö en liche ode komme zielle Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich machen, e eiben ode ande wei ig nu zen. So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen (insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en, gel en abweichend on diesen Nu zungsbedingungen die in de do genann en Lizenz gewäh en Nu zungs ech e. Te ms o use: Documen s in EconS o may be sa ed and copied o you pe sonal and schola ly pu poses. You a e no o copy documen s o public o comme cial pu poses, o exhibi he documen s publicly, o make hem publicly a ailable on he in e ne , o o dis ibu e o o he wise use he documen s in public. I he documen s ha e been made a ailable unde an Open Con en Licence (especially C ea i e Commons Licences), you may exe cise u he usage igh s as speci ied in he indica ed licence. h p://c ea i ecommons.o g/licenses/by/2.0/ ORIGINAL ARTICLE Open Access S a e oca ional ehabili a ion p og ams and ede al disabili y insu ance: an analysis o Vi ginia’s oca ional ehabili a ion p og am Da id Dean 1 ˆ , John V Peppe 2* , Robe M Schmid 1 and S e en S e n 2 * Co espondence: [email p o ec ed] ˆ Deceased 2 Uni e si y o Vi ginia, Cha lo es ille, VA, USA Full lis o au ho in o ma ion is a ailable a he end o he a icle Abs ac We examine he associa ion be ween he eceip o oca ional ehabili a ion (VR) se ices and Fede al Disabili y Insu ance using a unique panel da a sou ce on pe sons who applied o assis ance om Vi ginia’s VR p og am in 2000. Th ee cen al indings eme ge: i s , VR se ices a e associa ed wi h lowe a es o pa icipa ion in disabili y insu ance p og ams-a nea ly 2 poin d op in SSDI eceip and 1 poin d op in SSI eceip . Second, VR se ice eceip is associa ed wi h lowe ake-up a es o SSDI/SSI. Finally, among VR applican s on SSDI/SSI, hose who ecei e subs an i e VR se ices a e mo e likely o be employed. JEL codes: H51, I13, J24 Keywo ds: Voca ional ehabili a ion; Disabili y insu ance I. In oduc ion As he en ollmen and cos s o he Social Secu i y Adminis a ion’s Disabili y Insu - ance (SSDI) and Supplemen al Secu i y Income (SSI) p og ams ha e g own o e he pas wo decades, he e has been g owing in e es in whe he oca ional ehabili a ion (VR) p og ams migh se e o educe he numbe o pe sons ecei ing ede al disabili y insu ance bene i s (e.g., Au o and Duggan 2010; S aple on and Ma in 2012). Disabili y insu ance p og ams p o ide cash assis ance o people wi h de e minable and signi ican wo k disabili ies. By con as , he public-sec o Voca ional Rehabili a ion (VR) p og am, a $3 billion ede al-s a e pa ne ship, is designed o p o ide employmen - ela ed assis - ance o pe sons wi h a b oad spec um o disabling condi ions. As he na ion’s p ima y employmen suppo p og am o people wi h disabili ies, s a e VR agencies seem well- posi ioned o assis people wi h disabili ies in secu ing gain ul employmen a he han ecei ing SSDI/SSI (S aple on and Ma in 2012). Ye , while ound o play an impo an ole in helping pe sons wi h disabili ies o engage in he labo ma ke (Lop es 2007; Dean e al. 2013a, 2013b and 2013c), e y li le is known abou whe he VR se ices e- duce pa icipa ion in Fede al Disabili y Insu ance p og ams 1 . I VR se ices imp o e labo ma ke ou comes o po en ial SSDI/SSI bene icia ies, some clien s may choose o ully pa icipa e in he labo ma ke a he han ake up SSDI/SSI. Ye , o hose wi h lim- i ed abili ies o in e es , VR p og ams may ins ead se e o help clien s unde s and © Dean e al.; licensee Sp inge . This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.o g/licenses/by/2.0), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly c edi ed. Dean e al. IZA Jou nal o Labo Policy 2014 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 Fede al Disabili y Insu ance p og ams and ules, and consequen ly lead o an inc ease in ake-up (S aple on and Ma in 2012; Dean e al. 2013a). Using a unique panel da a sou ce on all pe sons who applied o VR se ices in Vi ginia in S a e Fiscal Yea (SFY) 2000, we s udy he ela ionship be ween VR se ice eceip and SSDI/SSI eceip . Fo each applican , we obse e qua e ly SSDI/SSI, employmen and ea nings da a, and VR se ice eceip da a om 1995 o 2010. This ich panel da a enable us o p o ide impo an insigh s in o he long- un impac o VR se ices on eceip o Fede al Disabili y Insu ance. Gi en ha he labo ma ke impac s o VR se ices a e known o di e by he ype o limi a ion (Dean and Dolan 1991; Baldwin 1999; Dean e al. 1999; Ma co e e al. 2000; Dean e al. 2013a, 2013b and 2013c), we analyze he ull popu- la ion as well as subpopula ions o clien s wi h a physical impai men , men al illness, o cogni i e impai men 2 . We also examine SSDI and SSI eceip sepa a ely: he wo g oups o ecipien s ha e e y di e en cha ac e is ics, ecei e di e en bene i amoun s, and ace di e en ea nings incen i es. In pa icula , he non-medical eligibili y c i e ia o SSDI e- qui es a signi ican wo k his o y,SSDI bene i s a y wi h wo k his o y, and SSDI bene i s a e e mina ed when a ecipien s ea nings exceed some h eshold. In con as , SSI is means- es ed and SSI bene i s a e de e mined using a ai ly s anda d nega i e income ax o mula wi h a basic gua an ee and 0.50 ea nings ax a e a small ea nings dis ega d. A e desc ibing he da a in Sec ion 2, we p oceed in wo pa s. In Sec ion 3, we con- side he associa ion be ween VR se ice eceip and pa icipa ion in SSDI and/o SSI, speci ically examining ake-up and exi s om he Fede al Disabili y Insu ance p o- g ams. In Sec ion 4, we e alua e he associa ion be ween VR eceip and labo ma ke ou comes, ocusing on clien s who ecei ed SSDI/SSI p io o applying o VR se ices. Finally, in Sec ion 5, we d aw conclusions. Th ee cen al indings eme ge. Fi s , in gen- e al, VR se ices a e associa ed wi h lowe a es o pa icipa ion in disabili y insu ance p og ams –a nea ly 2 poin d op in SSDI eceip and 1 poin d op in SSI eceip –bu he e is subs an ial a ia ion ac oss impai men g oups. Mos no ably, o he cogni i ely impai ed subg oup, VR se ices a e associa ed wi h g ea e SSDI/SSI eceip . Second, VR eceip is associa ed wi h lowe a es o en y in o SSDI/SSI bu also lowe a es o exi om he SSDI oles. Finally, among VR applican s on SSDI/SSI, hose who ecei e sub- s an i e se ices a e mo e likely o be employed bu , on a e age, ea n less pe qua e . While hese esul s may sugges ha VR p og ams can e ec i ely educe SSDI/SSI oles o ce ain subg oups, we cau ion agains d awing his ype o causal conclusion om he e idence p esen ed in his pape . Ra he , his analysis is la gely desc ip i e, and does no o mally add ess he undamen al me hodological p oblem in ol ed in d awing such in e ences om obse a ional da a. Speci ically, he decision o p o ide VR se ices is no likely o be andom, o exogenous, wi h espec o SSDI/SSI eceip . Se- lec ion has been ound o be a cen al p oblem in add essing he impac o job aining p og ams in gene al (LaLonde 1995; F iedlande e al. 1997 and Imbens and Woold idge 2009), and VR p og am in pa icula (Aak ik, Heckman, and Vy lacil, 2005; S aple on and Ma in 2012 and Dean e al. 2013a, 2013b and 2013c). Hence, he obse ed associa ions e alua ed in his pape a e likely o e lec , in pa , unobse ed ac o s such as he se e i y o disabling limi a ion and gene al heal h s a us which a e likely o be join ly associa ed wi h VR se ice and SSDI/SSI eceip . Likewise, he Social Secu i y Adminis a ion’scos - sha ing a angemen wi h VR agencies o se ices p o ided o ce ain SSDI/SSI bene i- cia ies u he con ounds causal in e ence (see S aple on and Ma in 2012). Dean e al. IZA Jou nal o Labo Policy Page 2 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 II. Da a The da a used o his analysis come om h ee di e en adminis a i e sou ces. Ou s a ing poin is he adminis a i e eco ds o he Vi ginia Depa men o Aging and Rehabili a i e Se ices (DARS) on indi iduals who applied o VR se ices o he i s ime in SFY 2000 (July 1, 1999 - June 30, 2000). We limi he sample o 6,732 i s - ime applican s be ween he ages o 21 and 55 a he qua e o applica ion 3 . Fo each e- sponden , we obse e an indica o o whe he hey ecei ed subs an i e VR se ices, which we e e o as “ ea ed”applican s 4 . In addi ion o obse ing whe he o no he clien ecei ed VR se ices, we obse e a ange o demog aphic a iables as well as in- dica o s o whe he he clien had a physical (PI), men al (MI), o cogni i e impai - men (CI) 5 . Va iables om he DARS eco ds a e measu ed a he ime o a clien s mos ecen applica ion o VR se ices, and hus do no a y o e he es ima ion ime ame. We hen me ge hese DARS eco ds wi h qua e ly adminis a i e eco ds on labo ma ke ac i i y om he Vi ginia Employmen Commission (VEC) 6 and mon hly adminis a i e eco ds o he Social Secu i y Adminis a ion (SSA) on Social Secu i y Disabili y Insu ance (SSDI) and Supplemen al Secu i y Income (SSI) bene i eceip om 1995 (qua e 2) o 2009 (qua e 4) 7 . Thus, o each DARS applican , we obse e qua e ly employmen , ea nings, and disabili y insu ance eceip o many yea s be o e and a e he applica ion qua e in SFY 2000. Summa y s a is ics Table 1 displays he means and s anda d de ia ions o he a iables o he ull sample, o he ea men g oups, and o h ee impai men g oups. O he 6,732 i s - ime DARS applican s, 61% a e p o ided VR se ices. The ea men a e is simila o cli- en s in he PI and MI subsamples bu much la ge o he CI subsample whe e jus o e h ee- ou hs o clien s wi h cogni i e impai men s a e p o ided wi h subs an ial VR se ices. Fo each applican , we obse e indica o s o disabili y insu ance and labo ma ke ou comes o e he i y-eigh qua e pe iod om 1995 h ough 2009. O e his pe iod, applican s ecei ed disabili y insu ance in 27% o he qua e s, we e employed in 38% o he qua e s, and, on a e age, ea ned $1,371 pe qua e . The e is in e es ing a i- a ion in hese a e ages ac oss he ea men and impai men g oups. Fo example, he a e o disabili y insu ance eceip a ies ac oss he h ee impai men subg oups, espe- cially SSI eceip which is 9% o he PI, 16% o he MI, and 35% o he CI. P esum- ably, his e lec s he ac ha SSI is means- es ed and, as obse ed in Table 1, employmen a es and a e age ea nings a e highes o clien s wi h physical impai - men s and lowes o hose wi h cogni i e impai men s. Finally, ea ed clien s ha e highe employmen a es and a e age ea nings han un ea ed clien s. Table 1 also p o ides summa y s a is ics o he demog aphic a iables which include measu es o gende , ace, educa ion, age, ma i al s a us, and numbe o dependen s, as well as measu es o anspo a ion a ailabili y and pa icipa ion in Medicaid 8 . Rela i e o a andom sample o US adul s, DARS clien s a e mo e likely o be black (33%), less likely o g adua e om high school (42%), less likely o be ma ied (18%), and less likely o ha e access o anspo a ion. The e a e only mino di e ences in he a e age cha - ac e is ics o ea ed and un ea ed clien s 9 . As migh be expec ed, he e a e la ge Dean e al. IZA Jou nal o Labo Policy Page 3 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 di e ences in he obse ed cha ac e is ics ac oss he h ee impai men g oups. Fo ex- ample, he high school g adua ion a e o he cogni i ely impai ed subg oup is 17%, while he a es o he physically and men ally impai ed subg oups a e a ound 60%. Va ia ion in SSDI/SSI eceip o e ime Figu es 1A-C display he qua e ly disabili y insu ance eceip a es measu ed ela i e o he VR applica ion da e. Thus, pe iod 0 is he qua e o applica ion in SFY 2000, pe iod −4 is one yea p io o applica ion, and pe iod 4 is one yea pos -applica ion. Figu e 1A displays he a es o eceip o SSDI/SSI o all 6,732 cases. The a e o SSDI eceip mono onically inc eases o e his 15-yea pe iod om a low o a ound 6% o a high o 26%. In con as , he SSI eceip a e slowly inc eases ( om a ound 10% o 17%) p io o he applica ion qua e and hen emains s able a a ound 17% he ea e . Table 1 Means and s anda d de ia ions by ea men s a us and impai men Full Sample Impai men a All T ea ed Un ea ed PI MI CI Va iable b Mean S d De Mean S d De Mean S d De Mean S d De Mean S d De Mean S d De T ea 0.61 0.49 1.00 0.00 0.00 0.00 0.60 0.49 0.61 0.49 0.76 0.43 SSDI o SSI b 0.27 0.45 0.28 0.45 0.27 0.44 0.28 0.45 0.37 0.48 0.48 0.50 SSI 0.14 0.35 0.14 0.35 0.14 0.34 0.09 0.29 0.16 0.37 0.35 0.48 SSDI 0.17 0.38 0.17 0.38 0.17 0.37 0.22 0.41 0.26 0.44 0.21 0.41 Employed 0.38 0.49 0.40 0.49 0.35 0.48 0.40 0.49 0.35 0.48 0.32 0.47 Ea nings 1371 3001 1436 2712 1268 3411 1608 3728 1219 2586 770 1786 Age 31 12 31 12 31 12 39 9 37 9 25 9 Male 0.52 0.50 0.50 0.50 0.53 0.50 0.47 0.50 0.40 0.49 0.51 0.50 Whi e 0.66 0.47 0.65 0.48 0.67 0.47 0.65 0.48 0.71 0.45 0.55 0.50 Black 0.33 0.47 0.34 0.47 0.31 0.46 0.33 0.47 0.28 0.45 0.42 0.49 Ma ied 0.18 0.39 0.17 0.37 0.21 0.41 0.37 0.48 0.18 0.38 0.04 0.19 High school 0.42 0.49 0.49 0.50 0.30 0.46 0.57 0.50 0.61 0.49 0.17 0.38 Educ. missing 0.15 0.36 0.00 0.00 0.39 0.49 0.17 0.38 0.14 0.35 0.05 0.22 Special educ 0.09 0.28 0.12 0.32 0.04 0.20 0.02 0.12 0.02 0.14 0.33 0.47 Family size 2.8 1.6 2.8 1.6 2.7 1.6 2.4 1.4 2.1 1.5 3.2 1.8 # Dependen s 0.66 1.12 0.61 1.08 0.74 1.18 1.08 1.27 0.84 1.19 0.37 0.93 T anspo a ion 0.71 0.45 0.71 0.46 0.73 0.44 0.82 0.38 0.74 0.44 0.46 0.50 D i e ’s license 0.60 0.49 0.59 0.49 0.61 0.49 0.79 0.41 0.68 0.46 0.17 0.38 Medicaid 0.17 0.37 0.16 0.37 0.18 0.38 0.14 0.35 0.18 0.39 0.29 0.45 PI 0.33 0.47 0.32 0.47 0.33 0.47 1.00 0.00 0.19 0.39 0.06 0.23 MI 0.22 0.41 0.22 0.41 0.22 0.41 0.13 0.33 1.00 0.00 0.11 0.31 CI 0.15 0.35 0.18 0.39 0.09 0.29 0.03 0.16 0.07 0.26 1.00 0.00 N 6732 4139 2593 2192 1479 992 No e: a. The impai men g oups a e no mu ually exclusi e o exhaus i e. Fo example, he e is no sepa a e g oup o clien s wi h lea ning disabili ies. Also, o he subg oup analysis, he PI and MI subsamples a e es ic ed o clien s age 21 and olde . The CI sample includes clien s 17 and olde . b. Momen s o he disabili y insu ance eceip and labo ma ke a iables a e based on all pe son-qua e s in samples. Momen s o he demog aphic and impai men a iables a e based on in o ma ion a ailable in he applica ion qua e . Dean e al. IZA Jou nal o Labo Policy Page 4 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 Figu e 1B displays he SSDI eceip a es by impai men g oup, and Figu e 1C p o ides he analogous in o ma ion o SSI eceip . While he a e o SSDI eceip mono onically inc eases o e ime o he h ee impai men g oups examined in his s udy, he pa e ns o change di e . Fo hose wi h cogni i e impai men s, he SSDI eceip a e ba ely changes p io o he VR applica ion, a e which he a e inc eases a he quickly. Fo hose wi h men al and physical impai men s, he a es o SSDI eceip inc ease s eadily o e ime excep o a sha p disc e e jump in he SSDI u iliza ion a e o he PI sub- sample beginning in qua e ze o, he VR applica ion qua e . The sha p jump in he SSDI a e among he physically impai ed subg oup is consis en wi h he sudden and unex- pec ed onse associa ed wi h some physical impai men s 10 . Wo ke s who suddenly su e se ious physical impai men s, o example, may become join ly eligible o VR and SSDI. The imese iespa e nso SSI eceip a edi e en (seeFigu e1C)inpa e lec ing he di e ences in he nonmedical eligibili y c i e ia be ween he wo p og ams; o he mos pa , SSDI equi es signi ican wo k his o y, while SSI is means- es ed. Fo he MI and PI Figu e 1 Fede al Disabili y Insu ance Receip Ra es by Qua e . (A) SSDI/SSI Receip Ra es. (B) SSDI Receip Ra es by Impai men . (C) SSI Receip Ra e by Impai men . Dean e al. IZA Jou nal o Labo Policy Page 5 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 subg oups, he a es o SSI eceip inc ease sligh ly un il he applica ion pe iod and basically emain ixed he ea e . No ably, he e is no disc e e jump in pa icipa ion a es in he pe iod o applica ion. Fo he cogni i ely impai ed g oup, he a es inc ease om abou 30% in qua e −16 o 40% in qua e 8, and hen s eadily all o a ound 30% in qua e 40. The cog- ni i ely impai ed a e much mo e likely o be on SSI han o he g oups o applican s. Ea nings and SSDI/SSI eceip In his sec ion, we p o ide basic desc ip i e s a is ics on he associa ion be ween ea n- ings and SSDI/SSI eceip . Eligibili y o SSDI/SSI bene i s equi es a pe son mus be unable o engage in subs an ial gain ul ac i i y (SGA), de ined as ea nings a leas $700 pe mon h in 2000 and a leas $980 in 2009. SSDI bene icia ies lose hei bene i s i hey engage in SGA o ex ended pe iods 11 . Table 2 displays he dis ibu ion o he pe son-qua e s by ea nings and SSDI/SSI eceip s a us 12 . Pe haps wha is mos s iking is he la ge ac ion o obse ed pe son-qua e s wi h no ea nings; 62% ha e ze o qua e ly ea nings and, o hose ecei ing disabili y insu - ance (ei he SSDI o SSI), his “unemploymen ” a e is 84%. The as majo i y o SSDI/SSI ecipien s who apply o VR se ices a e no employed. Table 2 also e eals s iking di e - ences in he ea nings pa e ns by SSDI/SSI pa icipa ion s a us. As should be he case, hose ecei ing SSDI/SSI a e e y unlikely o ea n in excess o he SGA h eshold. By con- as , a ela i ely la ge ac ion o hose no ecei ing disabili y insu ance ea n in excess o his h eshold. Finally, hese pa e ns a e common ac oss he impai men g oups. To u he explo e he ela ionship be ween ea nings and SSDI/SSI eceip , Figu e 2 dis- plays he dis ibu ion o pe son-qua e s by he di e ence in qua e ly ea nings and he app op ia e qua e ly SGA le el. To ocus on he key ea nings le els, we es ic he sam- ple pe son wi h posi i e qua e ly ea nings (>$100) ha lie be ween [−$2000, +$2000] o he qua e ly SGA h eshold. Figu e 2A displays he esul s by SSDI eceip s a us and Figu e 2B by SSI eceip s a us. Two dis inc pa e ns a e e ealed in hese igu es. Fi s , among non-SSDI/SSI ecipien s, ea nings a e e enly dis ibu ed ela i e o he SGA le el. Tha is, ea nings a e no sensi i e o he SGA h eshold. Second, among SSDI/SSI ecipi- en s, he ac ion o pe son-qua e s dec eases wi h ea nings 13 . Table 2 Ea nings ela i e o SGA by SSDI/SSI eceip Impai men Subg oups ALL PI MI CI Pe son q s % o o al Pe son q s % o o al Pe son q s % o o al Pe son q s % o o al No Ea nings No DI 152309 0.390 45690 0.359 28853 0.336 17366 0.302 DI 89429 0.229 31143 0.245 26513 0.309 21873 0.380 Ea nings ≤SGA No DI 46778 0.120 12409 0.098 8823 0.103 5821 0.101 DI 14198 0.036 3113 0.024 4376 0.051 5282 0.092 Ea nings > SGA No DI 84537 0.217 33754 0.266 16121 0.188 6521 0.113 DI 3205 0.008 1007 0.008 1096 0.013 673 0.012 To al 390456 1.000 127116 1.000 85782 1.000 57536 1.000 No e: SGA is he qua e ly subs an ial gain ul ac i i y le el de e mined by he SSA. Dean e al. IZA Jou nal o Labo Policy Page 6 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 Some ha e sugges ed ha many SSDI ecipien s migh s a egically ea n jus up o he SGA h eshold in o de o maximize ea nings wi hou losing bene i s (e.g., GAO Gene al Accoun abili y O ice 2002 and Schimmel e al. 2011), a beha io al phenomena e e ed o as “pa king.”I ue, one migh expec he dis ibu ion o SSDI ecipien s o spike upwa ds when ea nings app oach he SGA h esholds. Fo VR applican s, how- e e , we ind no e idence ha SSDI ecipien s a e pa king. In ac , he e is li le e i- dence o subs an ial pa king in he gene al popula ion o SSDI/SSI ecipien s (GAO Gene al Accoun abili y O ice 2002; Schimmel e al. 2011). III. VR se ices and SSDI/SSI eceip The e is ongoing in e es in whe he VR p og ams migh educe he numbe o pe sons e- cei ing disabili y bene i s (e.g., see Hennessey and Mulle 1995; Au o and Duggan 2010; S aple on and Ma in 2012; Sosulski e al. 2012; Dean e al. 2013a). In his sec ion, we examine he ela ionship be ween VR and SSDI/SSI eceip . We begin by displaying he ime se ies pa e ns o SSDI/SSI eceip by VR ea men s a us and hen epo esul s om a numbe o di e en linea eg essions ha allow us o o mally accoun o p e- applica ion ac o s as well as he obse ed co a ia es lis ed in Table 1. Finally, we ex end he analysis o ocus on whe he VR se ices migh be associa ed wi h SSDI/SSI en y o exi . VR ea men and SSDI/SSI eceip Figu es 3, 4 and 5 display he ime se ies pa e ns o qua e ly SSDI/SSI eceip a es by VR ea men s a us whe e qua e s a e measu ed ela i e o he applica ion da e in SFY 2000. Figu e 3 displays he ime se ies o SSDI o SSI eceip , Figu e 4 ocuses on Figu e 2 Dis ibu ion o VR Applican s (pe son-qua e s) by Ea nings –SGA. (A) SSDI Recipien s and Non-Recipien s. (B) SSI Recipien s and Non-Recipien s. Dean e al. IZA Jou nal o Labo Policy Page 7 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 SSDI eceip alone, and Figu e 5 ocuses on SSI eceip . Each igu e has ou panels: one o he ull sample and one o each o he h ee impai men g oups iden i ied in his analysis (PI, MI, and CI). A ew gene al pa e ns a e obse ed in hese igu es. Fi s , o he ull sample as well as he MI and PI subsamples, disabili y insu ance pa icipa ion a es o he ea men g oup inc ease a a slowe a e han in he con ol g oup. Conside , o example, he e- sul s o he physically impai ed subsample displayed in Figu e 3B. One yea p io o he applica ion qua e , he SSDI/SSI eceip a es equal 15% o bo h he ea ed and un ea ed. One yea a e he applica ion, he SSDI/SSI a e is 26% o he ea ed g oup and 32% o un ea ed. This six poin di e ence be ween he ea ed and un ea ed las s o nea ly en yea s. Thus, in his case, VR se ices appea o be nega i ely associa ed wi h SSDI/SSI eceip . Second, o he cogni i ely impai ed subg oup, hese pa e ns a e e e sed; SSDI/SSI eceip inc eases a a sligh ly as e a e o he ea ed, Figu e 3 F ac ion o VR Applican s Recei ing SSDI/SSI by Qua e . (A) Full Sample. (B) PI Subsample. (C) MI Subsample. (D) CI Subsample. Figu e 4 F ac ion o VR Applican s Recei ing SSDI by Qua e . (A) Full Sample. (B) PI Subsample. (C) MI Subsample. (D) CI Subsample. Dean e al. IZA Jou nal o Labo Policy Page 8 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 labo ma ke ime se ies o clien s ecei ing SSDI/SSI bene i s in all ou qua e s p io o he applica ion da e. Panel B displays he ime se ies o clien s who did no ecei e SSDI/SSI bene i s in he yea be o e applying o VR se ices. Pe haps he mos s iking inding is ha , p io o he applica ion qua e , employ- men a es o he ea ed and un ea ed a e nea ly iden ical, while, jus a e wa ds, he ea ed expe ience a p onounced inc ease in employmen a es. This spike is especially s iking o clien s who ecei ed SSDI/SSI in he qua e s p io o applying o VR se - ices. Fo example, among applican s ecei ing SSDI/SSI, he employmen a es a e 0.15 o bo h he un ea ed and ea ed one yea p io o he applica ion qua e . One yea a e he applica ion, he analogous employmen a es a e 0.19 o he un ea ed and 0.32 o he ea ed. Abou one yea a e he applica ion, he employmen a es o bo h he ea ed and un ea ed s a o decline bu a gap be ween he ea ed and un- ea ed emains nea ly ixed o a decade. A e en yea s, he employmen a es a e 0.12 o he un ea ed and 0.19 o he ea ed. Figu e 9 displays he ime se ies in ea nings among he employed. This igu e shows ha ends in ea nings among he employed a e almos iden ical o bo h he ea ed and he un ea ed h oughou . The basic esul s shown in Figu es 8 and 9 a e u he explo ed using a ixed-e ec s eg ession model o labo ma ke ou comes (i.e., employmen o ea nings) on he VR ea men indica o . Table 5 displays he coe icien es ima es and s anda d e o s o he VR ea men o a eg ession on he ull sample as well as on he h ee impai men subsamples. Fo clien s ecei ing SSDI/SSI p io o applying o se ices, VR is associ- a ed wi h a 11.7% inc ease in he employmen a e 16 . Fo clien s no ecei ing SSDI/ SSI, employmen a es a e 9.0 poin s highe o VR bene icia ies. While employmen a es a e posi i ely associa ed wi h VR se ices, ea nings among he employed a e no - ably lowe (abou $650 pe qua e ) among SSDI/SSI ecipien s. In summa y, we ind ha , o hose ecei ing SSDI/SSI, VR se ices a e associa ed wi h a sha p, subs an ial, and sus ained inc ease in employmen bu a d op in mean ea nings among he employed. Table 5 Indi idual ixed-e ec eg ession es ima es o VR on employmen and ea nings (>0) by P e-VR applica ion SSDI/SSI eceip s a us c Employmen Ea nings (>0) P e SSDI/SSI P e SSDI/SSI Yes a No b Yes a No b Coe SE Coe SE Coe SE Coe SE All 0.117* 0.005 0.090* 0.003 −652* 106 −646 PI 0.117* 0.011 0.095* 0.005 −444* 213 −128 105 MI 0.107* 0.010 0.091* 0.007 −1129* 159 55 71 CI 0.052* 0.012 0.058* 0.010 −31 121 22 95 No es: The coe icien es ima es a e o he associa ion be ween VR se ices and employmen o ea nings among he employed. The eg essions a e s a i ied by whe he he applican ecei ed SSDI/SSI in he yea p io o applying o VR se ices. a. Responden s ecei ing SSDI in all ou qua e s p io o he VR applica ion (N = 1,266). b. Responden s no ecei ing SSDI in any o he ou qua e s p io o he VR applica ion (N = 5,163). c. The eg essions include he co a ia es lis ed in Table 1. * = s a is ically signi ican a he 5% le el. No e: These coe icien s a e es ima ed using he indi idual ixed-e ec s eg ession model. Dean e al. IZA Jou nal o Labo Policy Page 15 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 V. Conclusion In his pape , we use a unique panel da a se o SFY 2000 applican s o he Vi ginia VR p og am o explo e he ela ionship be ween VR se ices and pa icipa ion in he wo p ima y Fede al Disabili y Insu ance p og ams, SSDI and SSI. Fo he mos pa , we ind compelling and consis en e idence ha VR eceip is nega i ely associa ed wi h SSDI and SSI eceip , a esul ha holds o he ull popula ion o VR clien s as well as subg oups o people wi h physical and men al impai men s. Fo hose wi h cogni i e impai men s, howe e , VR is posi i ely associa ed wi h SSDI/SSI eceip . In e es ingly, we ind ha VR is nega i ely associa ed wi h bo h en y and exi om SSDI. Thus, o clien s al eady ecei ing disabili y insu ance, VR is associa ed wi h a educed likelihood o lea ing he insu ance oles, while o clien s no ecei ing SSDI/SSI, VR appea s o educe he chances o aking up SSDI bene i s. Fo SSI, VR appea s o be associa ed wi h inc eased exi s and educed en y. Finally, we ind ha VR se ices a e associa ed wi h highe employmen a es bu somewha lowe a e age ea nings among he employed. As no ed abo e, we cau ion agains using hese esul s o d aw causal conclusions abou he impac o VR on SSDI/SSI pa icipa ion. Al hough we use some basic eg es- sion echniques o con ol o obse able co a ia es and ixed e ec s, hese esul s should be in e p e ed as desc ip i e in na u e. A mo e sophis ica ed modeling o he se- lec ion p oblem comp ises on-going and u u e esea ch (see Dean e al. 2013a, 2013b, and 2013c). S ill, he esul s in his pape p o ide an impo an i s s ep in analyzing he da a. The es ima ed associa ions sugges ha VR may play a ole in educing SSDI/SSI e- ceip . A he same ime, we see ha he unde lying ela ionships a e likely o be nu- anced and complex, a ying ac oss di e en disabili y insu ance p og ams, ypes o VR se ices p o ided, and cha ac e is ics o he VR applican s. Endno es 1 The e a e a hand ul o s udies assessing he co ela ion be ween VR se ices and SSDI/SSI eceip (e.g., Hennessey and Sco Mulle 1995; T emblay e al. 2006; Roge s e al. 2005; S aple on and Ma in 2012 and S aple on and E ickson 2004). These s udies, which o he mos pa do no add ess he selec ion p oblem, end o ind ha SSDI/SSI eceip is nega i ely associa ed wi h VR. Mo e ecen ly, Dean e al. (2013a) examined SSDI/SSI eceip in a s uc u al model o VR se ice p o ision, employmen p obabili y, log ea nings i employed, and SSI/DI eceip o indi iduals wi h men al illness. They ind ha VR se ice p o ision mode a ely inc eased SSDI/SSI eceip . 2 Fo ou analysis, he cogni i e impai men subsample only includes clien s wi h di- agnosed in ellec ual disabili ies. Those wi h a lea ning disabili y o au ism spec um diso de s a e no included in he subsample unless hey a e also classi ied as ha ing an in ellec ual disabili y. 3 In o al, he e we e 10,322 applican s in SFY2000. To a oid biases associa ed wi h le censo ing (e.g., Heckman and Singe 1984) we d op obse a ions whe e he indi- idual’s i s se ice spell was p io o SFY 2000 (2782 obse a ions). We also ex- clude applican s younge han 17 yea s (433 obse a ions) o olde han 55 yea s Dean e al. IZA Jou nal o Labo Policy Page 16 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 (375 obse a ions) a ime o applica ion as well as hose who died while in he p og am. 4 The e a e numbe o easons applican s a e labeled by he DARS as un ea ed. Some wi hd aw hei applica ion; o he s a e ound o be ineligible o VR se ices because hei disabili ies a e oo se e e o no se e e enough; and some eligible applican s d op ou o he p og am be o e subs an ial se ices a e p o ided. 5 The limi a ions g oupings a e based on he p ima y o seconda y diagnosis lis ed in he DARS adminis a i e eco ds when he indi idual has an open case; his may be he i s case in 2000, o i may be a subsequen case. These h ee ca ego ies a e no exclu- si e o exhaus i e. The PI and MI subsamples a e es ic ed o clien s age 21 o o e a he ime o applica ion. Addi ional de ails on hese subsamples a e p o ided by (Dean e al. 2013a, 2013b and 2013c). 6 VEC eco ds do no co e all employmen , o example, i does no include pe sons who we e sel -employed, wo ked ou o s a e, we e ede al employe s, o wo ked a con ingen - ype jobs ha do no p o ide bene i s. A compa ison by he au ho s o e- po ed ea nings o VEC e sus he SSA indica es ha abou wel e pe cen o indi id- uals in his coho show ea nings in SSA bu no VEC in 2001. (see Dean e al. 2013a, o u he de ails). 7 A Memo andum o Ag eemen be ween he SSA and he Vi ginia DARS allowed us o ob ain mon hly SSI and SSDI paymen s o indi iduals in ou sample h ough a p o- cedu e ha p e en s us om iden i ying indi iduals in he sample. 8 Educa ion in o ma ion is missing o 15.0% o he sample. Ra he han exclude such obse a ions, we code he high school g adua ion and special educa ion indica o s as ze o and included a dummy a iable o when educa ion in o ma ion. 9 The la ge di e ences in he a es o high school g adua ion and eceip o special educa ion a e highly in luenced by he ea men o he cases wi h missing educa ion in o ma ion. This will be accoun ed o in he eg ession models by using he indica o o whe he educa ion is missing. 10 Dean e al. 2013c, discuss and illus a e he impo ance o modeling his ea u e o physical impai men s in models o VR ake-up. 11 Some SSDI bene icia ies a e allowed o engage in SGA o limi ed pe iods wi hou bene i loss (see S aple on and Ma in (2012) o de ails). SSI bene i s a e based on a basic nega i e income ax o mula whe e, a e a modes ea nings dis ega d, each add- i ional dolla in ea nings educes bene i s by i y cen s. 12 The quali a i e indings a e simila when SSDI and SSI eceip a e disagg ega ed. 13 These gene al pa e ns a e eplica ed ac oss he h ee impai men subg oups as well as o o he subg oups o VR applican s (e.g., new SSDI bene icia ies). 14 In pa icula , we es ima e he ollowing mean eg ession model: EYT;X¼αiþθTþ qua e ;αq     whe e Y is an indica o o SSDI/SSI eceip , α i is he unknown ec o o ixed indi id- ual e ec s, T is an indica o o whe he he esponden ecei ed subs an ial VR se - ices, and (qua e , α q ) allows he eceip o SSDI/SSI o a y o e ime (we use a quad a ic ime end wi h an indica o o qua e s pos VR applica ion). We also es i- ma ed simila di e ence-in-di e ence models. The esul s a e almos iden ical o he ixed e ec s model es ima es and hus a e no epo ed in his pape . Dean e al. IZA Jou nal o Labo Policy Page 17 o 19 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 15 Gi en ha we only obse e VR eceip in a single SFY, we canno e alua e whe he he associa ion be ween VR and SSDI/SSI o a ies o e ime. 16 While we classi y an applican as employed i hey ha e any qua e ly ea nings, he esul s a e simila o o he ea nings h esholds. In pa icula , when using an indica o o qua e ly ea nings in excess o $1500, he esul ing coe icien s es ima es we e somewha smalle in magni ude bu had he same quali a i e implica ions. Fo example, he es ima e associa ed wi h he PI coho ecei ing SSDI/SSI p io o VR applica ion alls om 0.117 o 0.075, bo h signi ican a he 5% le el. Compe ing in e es The IZA Jou nal o Labo Policy is commi ed o he IZA Guiding P inciples o Resea ch In eg i y. The au ho s decla e ha hey ha e obse ed hese p inciples. Acknowledgemen s We would like o hank Joe Ashley, John Phelps, Ki s en Rowe, Ann S an ield, Vlad Medniko , and Jim Ro h ock om Vi ginia Depa men o Aging and Rehabili a ion Se ices (DARS), Paul O.Lea y o he Social Secu i y Adminis a ion, Da id S aple on, o he membe s o he Na ional Ins i u e o Disabili y and Rehabili a ion Resea ch (NIDRR) Ad iso y G oup, and Rachel Fowley o excellen esea ch assis ance. The DARS, he Vi ginia Depa men o Medical Assis i e Se ices, he NIDRR, and he Uni e si y o Vi ginia Banka d Fund o Poli ical Economy p o ided gene ous inancial suppo . All e o s a e ou s. This pape was submi ed o he IZA Jou nal o Labo Policy's call o pape s on “Social Secu i y Disabili y Bene i s: Finding Al e na i es o Bene i Receip ”. Two special edi o s, Da id Wi enbu g and Gina Li e mo e, we e sponso ed by he Uni e si y o New Hampshi e’s Rehabili a ion, Resea ch, and T aining Cen e on Employmen Policy and Measu emen , unded by he U.S. Depa men o Educa ion (ED), Na ional Ins i u e on Disabili y and Rehabili a ion Resea ch (coope a i e ag eemen no. H133B100030). Thei commen s do no necessa ily ep esen he policies o ED o any o he ede al agency (Edga , 75.620 (b)). The au ho s a e solely esponsible o all iews exp essed. Responsible edi o : Da id Wi enbu g Au ho de ails 1 Uni e si y o Richmond, Richmond, VA, USA. 2 Uni e si y o Vi ginia, Cha lo es ille, VA, USA. Recei ed: 14 Oc obe 2013 Accep ed: 10 Feb ua y 2014 Published: Re e ences Aak ik A, Heckman J, Vy lacil E (2005) Es ima ing ea men e ec s o disc e e ou comes when esponses o ea men a y: an applica ion o No wegian Voca ional Rehabili a ion P og ams. J Ecome 125:15–51 Au o D, Duggan M (2010) Suppo ing Wo k: A P oposal o Mode nizing he U.S. Disabili y Insu ance Sys em. Hamil on P ojec Baldwin M (1999) The E ec s o Impai men s o Employmen and Wages: Es ima es om he 1984 and 1990 SIPP. Beha Sci Law 17:7–27 Dean D, Dolan R (1991) Fixed-e ec s es ima es o ea nings impac s o he oca ional ehabili a ion p og am. J Hum Resou 26(2):380–391 Dean D, Dolan R, Schmid R (1999) E alua ing he oca ional ehabili a ion p og am using longi udinal da a: e idence o a quasi-expe imen al esea ch design. E al Re 23(2):162–189 Dean D, Peppe J, Schmid R, S e n S (2013a) The E ec s o Voca ional Rehabili a ion o People wi h Men al Illness., Unpublished manusc ip Dean D, Peppe J, Schmid R, S e n S (2013b) The E ec s o Voca ional Rehabili a ion o People wi h Cogni i e Impai men s. Unpublished manusc ip Dean D, Peppe J, Schmid R, S e n S (2013c) The E ec s o Voca ional Rehabili a ion o People wi h Physical Disabili ies. Unpublished manusc ip F iedlande D, G eenbe g D, Robins P (1997) E alua ing go e nmen aining p og ams o he economically disad an aged. J Econ Li 35(4):1809–1855 (GAO) Gene al Accoun abili y O ice (2002) SGA Le els Appea o A ec he Wo k Beha io o Rela i ely Few Bene icia ies, bu Mo e Da a Needed, Repo o he Commi ee on Finance, US Sena e and he Commi ee on Ways and Means, House o Rep esen a i es, Repo s 02–224 Heckman J, Singe B (1984) Econome ic du a ion analysis. J Econ 24:63–132 Hennessey J, Sco Mulle L (1995) The e ec o oca ional ehabili a ion and wo k incen i es on helping he disabled- wo ke bene icia y back o wo k. Soc Secu Bull 58(1):15–28 Imbens G, Woold idge J (2009) Recen de elopmen s in he econome ics o p og am e alua ion. J Econ Li 47(1):5.86 LaLonde R (1995) The p omise o public sec o -sponso ed aining p og ams. J Econ Pe spec 9(2):149–168 Lop es P (2007) S a egic Assessmen o he S a e o he Science in Resea ch on Employmen o Indi iduals wi h Disabili ies, Na ional Ins i u e o Disabili y and Rehabili a ion Resea ch. Con ac GS23F8198H Ma co e D, Wilxox-Gok V, Redmond D (2000) The Labo Ma ke E ec s o Men al Illness: The Case o A ec i e Diso de s. In: Salke e D, So kin A (eds) The Economics o Disabili y. JAI P ess, G eenwich, CT Dean e al. IZA Jou nal o Labo Policy Page 18 o 19 08 Ap 2014 2014, 3:7 h p://www.izajolp.com/con en /3/1/7 Roge s J, Bishop M, C ys al M (2005) P edic ing ehabili a ion ou come o supplemen al secu i y income and social secu i y disabili y insu ance ecipien s: implica ions o conside a ion wi h he icke o wo k p og am. J Rehabil 71 (3):5–10 Schimmel J, S aple on D, Song J (2011) How common is ‘Pa king’among social secu i y insu ance bene icia ies? e idence om he 1999 change in he ea nings le el o subs an ial gain ul ac i i y. Soc Secu Bull 71(4):77–92 Sosulski M, Donnell C, Kim WJ (2012) Disabili y and employee bene i s eceip : e idence om he U.S. Voca ional ehabili a ion se ices p og am. J Soc Wo k Disabil Rehabil 11(1):33–54 S aple on D, E ickson W (2004) Cha ac e is ics o Incen i es: Why Do Employmen Ou comes o he SSA Bene icia y Clien s o VR Agencies Di e , on A e age, om Those o O he Clien s? Co nell Uni e si y Ins i u e o Policy Resea ch, Washing on DC S aple on D, Ma in F (2012) Voca ional Rehabili a ion on he Road o Social Secu i y Disabili y: Longi udinal S a is ics om Ma ched Adminis a i e Da a. Unpublished manusc ip T emblay T, Smi h J, Xie H, D ake R (2006) E ec o bene i s counseling se ices on employmen ou comes o people wi h psychia ic disabili ies. Psychia Se 57(6):816–821 Ci e his a icle as: Dean e al.:S a e oca ional ehabili a ion p og ams and ede al disabili y insu ance: an analysis o Vi ginia’s oca ional ehabili a ion p og am. IZA Jou nal o Labo Policy Submi you manusc ip o a jou nal and benefi om: 7 Con enien online submission 7 Rigo ous pee e iew 7 Immedia e publica ion on accep ance 7 Open access: a icles eely a ailable online 7 High isibili y wi hin he fi eld 7 Re aining he copy igh o you a icle Submi you nex manusc ip a 7 sp inge open.com Dean e al. IZA Jou nal o Labo Policy Page 19 o 19 10.1186/2193-9004-3-7 2014, 3:7 2014, 3:7 h p://www.izajolp.com/con en /3/1/7