The Rise in the Disability Rolls and the Decline in Unemployment* David H
Total Page:16
File Type:pdf, Size:1020Kb
THERISE INTHEDISABILITY ROLLS AND THE DECLINE INUNEMPLOYMENT* DAVID H. AUTOR AND MARK G. DUGGAN Between1984 and 2001, the share of nonelderly adults receiving Social Security DisabilityInsurance income (DI) rose by 60 percent to 5.3 million bene- ciaries.Rapid program growth despiteimproving aggregate health appears to be explainedby reduced screening stringency, declining demand for less skilled workers, andan unforeseen increase in the earnings replacement rate. We esti- matethat the sum of these forces doubled the labor force exit propensity of displacedhigh school dropouts after 1984, lowering measured U. S.unemploy- mentby one-halfa percentagepoint. Steady state calculations augur afurther 40 percentincrease in the rate of DIreceipt. Thefederal Disability Insurance(DI) programis thelargest incomereplacement program in theUnited States directedto- ward nonelderlyadults, with annual cash transfersexceeding $54 billion in 2001. Oncebene ts areawarded, recipientsreceive incomereplacement and healthinsurance through Medicare until returnto work, medical recovery, death, or retirement at age65, at whichpoint theyobtain equivalentbene ts from theSocial Security Administration’ s Old Ageand Survivors Insuranceprogram. Over the past twodecades, two important changeshave impacted eligibility and generosityfor this pro- gram.The rst,occurring in 1984, liberalized thedisability de- terminationprocess, reversing a dramatic reductionin disability rollsunderway. The second occurred more gradually. Becausethe DIbenets formulais progressiveand indexed tothe mean wage in theeconomy, the widening dispersionof earnings during the 1980s and 1990s substantially raised theratio of disability in- *Thisresearch was supported in part bytheCenter for Retirement Research atBoston College pursuant to a grant fromthe U. S.SocialSecurity Administra- tionfunded as part ofthe Retirement Research Consortium. The opinions and conclusionsare solely those of theauthors and should not be construedas repre- sentingthe opinions or policyof the Social Security Administrationor anyagency ofthe Federal Government or the Center for Retirement Research at Boston College.We are indebted to Daron Acemoglu, Joshua Angrist, Orley Ashenfelter, JohnBound, Esther Du o, DanielHamermesh, James Heckman, Caroline Hoxby, JuanJimeno, Lawrence Katz, JeffreyKling, Alan Krueger, StevenLevitt, Sendhil Mullainathan,Casey Mulligan, Derek Neal,David Stapleton, Paul Van de Water, numerousseminar participants, three astute referees, and the editors of this Journal fornumerous insightful suggestions. We thank Hung-Lin Chen, Radha Iyengar,and Liz March foroutstanding research assistance, and Alan Krueger, BrooksPierce, Kalman Rupp, Charles Scott, andAlan Shafer for generous assis- tancewith data sources. Authors may be contacted at [email protected], and [email protected]. © 2003by thePresident and Fellowsof Harvard Collegeand theMassachusetts Institute of Technology. The Quarterly Journal ofEconomics, February2003 157 158 QUARTERLY JOURNALOF ECONOMICS cometo prior earnings (the “replacementrate ”)forlow-skilled workers,a trendaugmented by therising real value of medical benets provided.Contemporaneously, the share of nonelderly adults receivingDI incomerose by morethan 60 percentfrom its 1984 troughto 3.7 percentof the adult population ages25 to64 (5.3 millionbene ciaries).Relative to prior DI cohorts,new entrants wereyounger and substantially morelikely to suffer fromlow mortalityimpairments, particularly pain and mentaldisorders. Theobjective of this paper is toassess how these program- maticchanges —reducedscreening stringency and arisingre- placementrate —interactedwith decliningdemand forless- skilledworkers to impact laborforce participation ofthe low skilledduring theperiod of 1978 to1998. Thequestion is intrin- sically difcult.DI isforthe most part auniformlyadministered federalprogram, making for limited treatment-control compari- sons.Additionally, whileaggregate health improved during the past twodecades, reduced DI screeningstringency coincided with aperiodof declining labor market opportunities for less-skilled workers,making it likelythat demand for benets (as aformof incomereplacement) rose while bene ts supply was shifting out- ward.1 Finally,the administrative de nitionof disability —the inability to “engagein asubstantial gainful activity ”—ensures that holding healthconstant, disability bene ts areeasier to obtain whenjob opportunities are scarcer. Given these confounds, distinguishing theimpacts ofincreasing supply fromincreasing demand forbene ts is amajortheme of our paper. To(attempt to)surmount these issues, we developtwo sets of instrumentalvariables toproxy demand and supply conditions. Toidentify exogenousvariation in thesupply ofdisability bene- ts,we exploit the progressivity of theDI bene ts formula.This formuladoes not account for variation in regionalwage levels, and hencepotential replacement rates are substantially higherin lowwage states. This variation yields differential cross-state benetsupply shifts during programcontractions and expansions (both ofwhichoccurred in oursample frame) in high versuslow replacementrate states. To identify exogenousvariation in the demand forbene ts,we follow Bartik [1991] in constructinga measureof state level labor demand shifts at lowand medium frequency.This measure, a weightedsum of national industry 1.See Juhn, Murphy, andTopel [1991], Juhn [1992], and Katz andAutor [1999]on adversedemand shifts faced by low-skilledworkers inrecent decades. DISABILITYAND UNEMPLOYMENT 159 employmentchanges (excluding ownstate employment) projected ontostate industry composition,performs well in predicting the owof DIapplications by state,indicating that adversedemand shocksraise demand fordisability bene ts.Exploiting this com- bination ofinstruments, we study theimpact ofthe supply and demand forDI bene ts onthe labor force behavior of low-skilled workersduring theperiod of 1978 –1998. Ourcontribution builds onan in uentialliterature exploring theimpacts ofdisability bene ts onthe level of labor force par- ticipation in theUnited States and abroad. 2 Distinct fromthis literature,our paper analyzes theimpact ofthesupply ofdisabil- ity benets onthe responses of low-skilled workersto adverse labormarket shocks. The rationale for this focusis that nonem- ploymentis ade factoprecondition for disability application, makingthe opportunity cost of seeking DI bene ts higherfor employedworkers than forjob losers and nonparticipants. Many potentialbene ciarieswill thereforeseek bene tsin theevent of jobloss but nototherwise. We call thesepotential bene ciaries “conditionalapplicants. ” As ourconceptual model demonstrates, thesize of this conditionalgroup is likelyto be relatively elastic toprogram and labormarket conditions. By contrast,direct em- ploymentto disability transitionswill berelatively inelastic. Our frameworktherefore implies that reducedDI screeningstrin- gency,rising replacement rates, and slackeningdemand forlow- skilledworkers should haveraised thepropensity of workers facing an adversedemand shockto exit the labor force to seek disability bene ts. Thisprediction structures our empirical analysis. After es- tablishing that cross-stateshifts in thesupply ofbene ts raised thelabor force participation ofless-skilled workers during the 1979 –1984 disability retrenchmentand loweredthem thereafter, 2.Key papersin thisliterature include Parsons [1980], Bound [1989], Have- man,De Jong, and Wolfe [1991], Bound and Waidman [1992], Gruber andKubik [1997],Aarts, Burkhauser, and De Jong [1996], and Gruber [2000],who study the impactof disability bene tson laborsupply in Canada,the Netherlands, and the UnitedStates. Lewin-VHI, Inc. [1995], Rupp andStapleton [1995], and Stapleton etal. [1998] analyze the importance of the economic climate to disability appli- cationand receipt. Bound and Burkhauser [1999] provide a comprehensiveover- viewof thelabor market impacts of disabilityprograms. The work closestin spirit tothe approach taken here is theexcellent study byBlack, Daniel,and Sanders [2002],who measure the impact of shocks to coal prices on DIand SSI incomein miningintensive counties. The Black, Daniel,and Sanders study doesnot explore labormarket implications. 160 QUARTERLY JOURNALOF ECONOMICS weturn to an examinationof the interaction of the disability programwith adversedemand shocks.We nd that DIapplica- tionrates for given demand shocksrose secularly after the dis- ability reformsof 1984, reachingtwo to three times their prere- formlevels by thelate 1990s. Parallelingthis change,we estimatea 60 percentincrease in thepropensity of displaced high schooldropouts to exit the labor force. Simple calculations sug- gestthat increaseddisability application propensitycan account forthis behavioralchange. As alternativeexplanations forthe observedlabor force exit of less-skilled workers, we explore the importanceof changing mortality rates, rising immigration and incarcerationrates, uctuationsin UIbene ts,and, logically, falling wages.Although many of these factors appear relevantto cross-statepatterns oflow-skilled laborforce participation, none substantially altersthe nding that theincreasing supply ofDI benets induced substantial laborforce exit of low-skilled work- ersduring 1984 –1998. Togaugethe importance of changes in theDI programfor the aggregatelabor market, we calculate the demand contraction experiencedby high schooldropouts over the post-disability-re- formyears to form a counterfactuallabor force participation