Determinants of the Growth in the Social Security Administration s Disability Programs-An Overview

Transcription

1 Determinants of the Growth in the Socia Security Administration s Disabiity Programs-An Overview by Kaman Rupp and David Stapeton* This artice examines factors affecting the growth in the Socia Security Administration s disabiity programs. We synthesize recent empirica evidence on factors affecting trends in appications and awards for Disabiity Insurance and Suppementa Security Income (SSI) benefits and duration on the ros. Econometric anayses of pooed time-series, cross-sectiona data for States provide strong evidence of business cyce effects on appications and, to a esser extent, on awards. Substantia effects of cutbacks in State genera assistance programs are aso found, especiay for SSI. Estimated effects of the aging of the baby boomers, growth in the share of women who are disabiity insured, the AIDS epidemic, and changes in famiy structure are aso presented. Indirect evidence suggests the importance of programmatic factors, especiay for awards, and especiay in the menta and muscuoskeeta impairment categories. The decine in the average age of new awardees has substantiay increased duration, particuary for SSI. As a resut, caseoad growth woud be expected to continue even in the absence of future award growth. *The authors are, respectivey, from the Office of Research and Statistics, Socia Security Administration, and Lewin-VHI, Incorporated. Acknowedgments: An earier version of this artice was presented at the Aied Socia Science Associations meetings on January 8, 1995, in Washington, DC, in a session on disabiity caseoad growth organized by the Society of Government Economists. The authors wish to thank Gibert Fisher, Richard Foster, Michae Leonesio, L. Scott Muer, Wende Primus, Steven Sande, Lawrence Thompson, Peggy Trout, and Peter Wheeer for their hepfu comments. We are aso gratefu for anaytic support, suggestions, or data provided by a arge number of individuas, especiay Burt Bamow, John Bound, Richard Burkhauser, Kevin Coeman, Kimbery Dietrich, Ei Donkar, Jeffrey Furman, Gina Livermore, Gibert Lo, Chares Scott, Aan Shafer, and Meinda Upp. This artice soey refects the opinions and interpretations of the authors, and does not necessariy represent the officia positions of SSA or Lewin-VHI, Inc. The Federa Government provides cash benefits for persons with severe disabiities through two Socia Security Administration (SSA) programs-the Socia Security Disabiity Insurance (DI) program under Tite II of the Socia Security Act, and the Suppementa Security Income (SSI) program under Tite XVI. Both programs use the same defmition of disabiity, but other eigibiity criteria differ. In particuar, DI is a socia insurance program with disabed-worker eigibiity based on prior Socia Security covered empoyment. Prior work experience is not required under SSI, but the program is means tested, using income and asset eigibiity criteria. Some persons may be eigibe under both programs and receive DI and SSI benefits concurrenty. DI benefits stop when a DI beneficiary reaches age 65 and he or she is transferred to the Od-Age and Survivors Insurance program (OASI). SSI disabiity recipients may continue to receive benefits past age 65, if they continue to meet the income and asset tests. Chidren with quaifying disabiities are eigibe for SSI payments on their own right subject to income and eigibiity requirements. About 20 years ago a series of econometric studies, primariy using aggregate time series techniques (Land0 1974; Hambor 1975; Thompson and Van de Water 1975), were conducted, focusing on the effects of the business cyce on the growth of the DI program. This heightened interest coincided with a peak in the number of new awards in 1975 (chart 1). SSA actuaries continued to monitor the growth of the DI program on an ongoing basis, focusing on demographic and egisative changes that shape program growth. Academic interest in the DI and SSI programs waned during the ate 1970 s and eary 1980 s, as appications and awards started to decine, but the recent growth in both programs refocused attention on them. The upsurge was particuary notabe in the number of chidren on the SSI disabiity ros. A better understanding of the factors affecting program growth is necessary to improve our abiity to make predictions about future growth both in the short and ong run. The number of appications has a direct effect on SSA s administrative costs and abiity to process appications in Socia Security Buetin Vo. 58, No. 4 Winter I,

2 a timey fashion, whie the number of new awards and ength of stay determine caseoad growth and program cost. SSA prepared a report (Department of Heath and Human Services (DHHS) 1992) summarizing existing knowedge about the growth of the D program and providing a comprehensive ist of various demographic, economic, and programmatic factors hypothesized to affect caseoad growth. After receiving the 1992 report, the Board of Trustees of the Federa OASI and DI Trust Funds recommended that SSA initiate a research effort to es- tabish whether the growth represents a temporary phenomenon or a onger-term trend. In response to this recommendation, and in cooperation with the Office of the Assistant Secretary for Panning and Evauation at DHHS, SSA initiated a series of research projects conducted through a combination of m-house anayses and severa contracts with Lewin- VHI, focusing on appication and award growth to produce an assessment of the reasons for disabiity program growth. This additiona research was needed to assess the causa roe of various factors and to quantify their effects. In particuar, it was important to assess whether various factors are primariy responsibe for short term temporary or cycica changes in caseoad growth or for permanent changes. To improve future projections, it is aso important to identify factors unaccounted for or impropery incuded in previous modes. Improved knowedge about caseoad growth might aso faciitate usefu poicy interventions reated to particuar factors, the identifi- cation of programmatic options to contro future growth, and the improvement of the incentives associated with disabiity programs. This artice summarizes what we currenty know about the factors affecting caseoad growth. Our summary is primariy based on the resuts of quantitative anayses of factors affecting appications, awards, and duration on the ros, but we aso rey on the resuts of other reevant studies. Overa Conceptua Framework Caseoad growth is affected by both economic and noneconomic factors. Factors such as the vaue of potentia cash benefits reative to wages, the vaue of compementary or substitute program benefits, and business conditions are Chart 1.-DI and SSI disabiity awards, (In thousands) years 44 Socia Security Buetin Vo. 58, No. 4 Winter 1995

3 ceary in the domain of economics. Demographic and epidemioogica factors, as we as the criteria for determining disabiity status and their impementation are, at east in a proximate sense, noneconomic factors that might affect caseoad growth, often substantiay. The economic perspective, emphasizing the roe of opportunity costs-individuas making choices comparing various aternatives-is particuary usefu in understanding how economic and noneconomic forces interact in shaping decisions such as appying for and being awarded disabiity benefits, as we as decisions concerning eaving the disabiity ros. From the economic perspective, program participation is an outcome of the interaction of the demand for program benefits by individuas and the suppy of program benefits by the Government. On the demand side, the number of appications (representing the demand for awards) in part depends on: the reative advantages of working and not working; the avaiabiity of substitute forms of pubic assistance (such as Aid to Famiies with Dependent Chidren (AFDC) and genera assistance (GA)); compementary benefits provided to those receiving disabiity benefitsespeciay heath insurance benefits (Medicare for DI beneficiaries and Medicaid for SSI recipients); and various features of the DI and SSI programs-that is, the costs of appying for benefits, the probabiity of receiving an award, and how ong benefits are expected to continue. Once persons with disabiities begin receiving DI and/or SSI benefits, their continued demand for benefits is infuenced by: the ength of time aready on the ros-over time, a beneficiary s potentia earnings decine as continued separation from the abor force erodes working skis; programmatic disincentives to work-with imited exceptions, DI and SSI beneficiaries who engage in substantia gainfu activity (that is, those earning over $500 per month) subsequent to a 9-month tria work period ose a their disabiity benefits;2 changes in their heath and disabiity status; and changes in the abor market. Suppy side factors affecting appication for disabiity benefits incude egisative factors, as we as administrative procedures, judicia ruings, and the resources avaiabe for making award decisions. The suppy side for those on the ros is affected by: the number of Continuing Disabiity Reviews (CDRstto determine if current beneficiaries sti meet the disabiity standard (with benefit termination for those who do not); the number of CDRs conducted depends both on the avaiabiity of administrative resources and the poitica wi to conduct these often unpopuar reviews; changes in rues concerning the effects of work on program eigibiity and benefits, especiay for the SSI program; and the avaiabiity of vocationa rehabiitation programs and incentives and requirements to use them. Whie factors affecting initia awards and ength of stay can be seen as sequentia, changes at the back end of the process have potentia feedback effects as we. For exampe, anticipated reconsideration and administrative aw judge (ALJ) decisions may affect initia eigibiity determination decisions by the State Disabiity Determination Services (DDS), as we as appicant decisions to ask for the reconsideration of unfavorabe decisions and to exercise their appea rights. The perceived strictness of the disabiity determination process might aso affect appications. In addition, perceptions about SSA s tennination and suspension poicies might affect work activities, and therefore continued eigibiity, among beneficiaries. Athough the economic perspective focuses on choices made by individuas given their avaiabe opportunities, it must aso be recognized that third parties often have a significant interest in this choice and may activey try to infuence it. An important exampe is State and oca governments who have an interest in shifting the costs of wefare and heath expenditures to the Federa Government. Other interested third parties incude empoyers, heath care providers, and private insurers. In the next section, we focus on factors affecting appications and awards. In the section foowing it, we address the factors affecting ength of stay, suspensions and terminations, as we as the effect of ength of stay on caseoads. In the fina section, we briefy discuss impications and directions for future research. Appications and Awards This section provides an overview of the factors affecting appications and awards. Appications have a major effect on SSA s administrative costs, whie the number and characteristics of new awardees fundamentay shapes future program cost. First, we discuss popuation factors and trends affecting the target popuations of the DI and SSI programs. Subsequenty, we address the business cyce and economic restructuring, the avaiabiity and vaue of other benefits, and programmatic factors affecting appications and awards. Methodoogy Much of the foowing discussion is based on the findings from a State-eve econometric anaysis of appications and awards for the period and from a foowup anaysis for The dependent variabe of interest in these anayses is the voume of appications and awards at the State eve; the independent variabes were designed to capture reevant factors affecting these Stateeve aggregates. For the first of the econometric anayses, SSA prepared tabuations of appications and awards at the State eve for the period from its new Disabiity Research Fie (DRF), aggregated by Socia Security Buetin Vo. 58, No. 4 Winter

4 program, sex, age, and impairment category. These data became the core of our database, to which we added data for various State-eve expanatory variabes for the same period (Stapeton, Coeman, and Dietrich 1995). For the foowup anaysis, we used previousy existing State-eve tabuations of initia (medica) determinations and initia eowance rates, disaggregated by program ony, aong with expanatory variabe data for the period (Stapeton and Dietrich 1995). For both anayses, we appied a methodoogy to the anaysis of appication and award growth that had not been previousy used for that purpose+ne that poos time-series data for a cross section of individua States. The methodoogy provides estimates of the effects of various factors on appications and awards that are based on the reationship across States between changes in the factors and changes in appications and awards. An important advantage of this approach is that it contros for changes in nationa factors, such as disabiity program rues and poicy, to the extent that they have a common infuence on appications and awards in a States, as we as for permanent differences across States. Conversey, an important imitation of the methodoogy is that it cannot be used to estimate the effects of factors for which changes did not vary across States during this period, or that are not accuratey measured at the State eve. Therefore, we suppement this econometric anaysis with actuaria anayses and other pieces of evidence. For the anaysis, we estimated 40 appication and 40 award equations for each program. The dependent variabe in each equation is the ogarithm of either an appication or incidence rate for a specific age/sex/impairment group (five age categories, two sex categories, and four impairment categories-menta disorders, muscuoskeeta disorders, infectious diseases, and a others). The independent variabes in each equation refer to a singe State-eve measure, and do not refect disaggregated information specific to a given age/sex/impairment group. Beow is an overview of the inde- pendent variabes used and their hypothesized effects: The State unempoyment rate is utiized to represent the business cyce. We hypothesize that increases in the rate of unempoyment have positive effects on the voume of appications and awards, whie decreases have negative effects. The GA program cuts variabe is a proxy for the effects of State budgetary pressures on the generosity of State-funded substitute programs (especiay for SSI). We hypothesize that State cutbacks in State funded substitute programs positivey affect disabiity appications and awards. Conversey, increased generosity and access to State funded substitute programs is expected to have a negative effect on disabiity appications and awards. Specificay, we hypothesize that cutbacks and the eimination of GA programs in seected States increased disabiity program growth, and that this effect is stronger for the means-tested SSI program than for the DI program. An AIDS/HIV incidence variabe accounts for the effects of the AIDS epidemic on the incidence and prevaence of disabiity. We hypothesize that the increasing incidence of AIDS/ HIV has a positive effect on disabiity appications and awards. A variabe measuring the number of immigrants granted ega aien status under the Immigration Reform and Contro Act (IRCA) to represent the effects of the aw. Unike most other immigrants, those granted ega aien status under IRCA were immediatey eigibe for SSI. We hypothesize that the number of immigrants granted ega aien status positivey affects the voume of SSI appications and awards. The percentage of chidren iving in one-parent famiies as a proxy for the effects of the proportion of househods headed by singe parents on appications and awards (particuary for SSI). We hypothesize that the propor- tion of househods headed by singe parents positivey affects SSI appications and awards. A dummy variabe for each year to contro for nationa factors. The operationa definitions and sources of each of the variabes used in these anayses are detaied in Lewin-VHI (1995a). For DI, we aso estimated separate equations for those who appied for just DI (DI-ony) and for those who had appied for SSI and DI concurrenty.4 Individuas who appy for both programs are of specia interest for two important reasons. First, they are the fastest growing group. Second, as wi be discussed in more detai ater, the majority of those who receive SSI awards are reativey short stayers on the SSI ros because they can no onger pass the SSI means test once their 5-month waiting period for DI benefits expires and they begin receiving such benefits. For the anaysis, we estimated a singe initia determination and aowance rate equation for each of three program groups: D&ony, SSI-ony, and concurrent5 The SSI-ony data incudes chid initia determinations, which is probematic because of the exceptionay arge growth in chid appications from 1990 on. The dependent variabe in each equation is the ogarithm of either initia determinations per capita or the initia aowance rate (initia aowances divided by initia determinations). The expanatory variabes that foow incude those from the anaysis that were avaiabe for the fu period, pus severa others that represent overa State-eve vaues: The expected appication rate variabe is based on 1990 nationa appication rates by age group and the age distribution of the State s popuation in the current year-to capture the effect of the aging of the popuation. This variabe is equa to a weighted average of age-specific nationa appication rates for 1990, with the weight for each age group equa to the share of the State s working-age popuation in the age category. We hypothesize that the expected appication rate variabe 46 Socia Security Buetin Vo. 58, No. 4 Winter 1995

5 positivey affects initia determina- tions and aowances. * The abor-force participation rate variabe captures the negative, cycica effect of discouraged workers eaving the abor force during recessions. We hypothesize that decines in aborforce participation are positivey associated with initia determinations and awards. However, for DI, the aborforce participation rate variabe may aso capture the ong-term positive effect of growth in the share of women who are disabiity insured. The share of empoyment in manufacturing variabe is a proxy for the effect of economic restructuring. In the short run we anticipate a positive reationship between changes in the share of manufacturing empoyment and initia determinations and aowances as a resut of job osses associated with decreases in manufacturing empoyment. However, in the onger term an opposite effect can be hypothesized refecting the ong-term shift to service sector jobs that tend to be associated with a ower incidence of work disabiities. The incidence rate of the disabing work injuries variabe captures changes in the nature of work. We hypothesize a positive effect on initia determinations and aowances. The poverty rate variabe captures changes in poverty that are not picked up by other variabes in the mode. We hypothesize a positive reationship between the poverty rate and SSI initia determinations and aowances. * The mean AFDC payment for a twoperson househod reative to mean earnings captures the reative attractiveness of AFDC benefits. We hypothesize a negative effect on SSI initia determinations and aowances. The mean SSI payment incuding State suppement payments, reative to mean earnings, captures the reative vaue of SSI benefits. We hypothesize that the reative vaue of SSI payments positivey affects initia determinations and aowances. When feasibe, we adjusted the expanatory variabes for changes in the age distribution of the popuation. For a more detaied discussion, see Stapeton and Dietrich (1995). For the anaysis we reated current-year changes in expanatory variabes to current changes in appication and incidence rates. For the anaysis, we aso examined the effect of prior year ( agged ) changes in the expanatory variabes on current year initia determinations and aowance rates and found substantia impacts for two variabes: the unempoyment rate and the abor-force participation rate. A technica description of the econometric methodoogy appears in the appendix, aong with seected regression and simuation resuts. In order to better design, interpret, and vaidate the econometric anaysis, we conducted a substantia review of reevant iterature, interviewed a series of government and academic experts on disabiity, and conducted case studies of appication and award growth in five States. The States seected for the case studies incude the four argest States- Caifornia, Forida, New York, and Texas-and Michigan-a State that terminated its genera1 assistance program in We visited each State, during which we interviewed officias in SSA fied offices, State Disabiity Determination Services, State and oca wefare agencies, and other agencies. Foowing the site visits, we anayzed program growth in each State using the econometric modes. In addition, survey data for Michigan GA recipients that had been matched to SSI administrative records were anayzed (Bound, Kossoudji, and Ricart-Moes 199Q6 The next sections summarize the evidence concerning the effects of popuation factors, the business cyce, economic restructuring, other support programs, and features of the disabiity programs themseves. Popuation Factors In the sections that foow we wi examine the areas of popuation growth and target popuations. Popuation growth.-changes in the size and age/gender composition of the popuation provide the simpest and most direct expanation of changes in the number of DI and SSI appications and awards. The size of the working-age SSA area popuation grew steadiy from 1975 to 1992, and is expected to continue growing steadiy in the near future. (The SSA area popuation refers to residents of the 50 States and the District of Coumbia adjusted for net census undercount; civiian residents of Puerto Rico, the Virgin Isands, Guam, and American Samoa; Federa civiian empoyees and persons in the Armed Forces abroad and their dependents; crew members of merchant vesses, and a other U.S. citizens abroad.) The baby boom generation, born between 1946 and 1964, was sti entering the working-age popuation in As it did, the average age of the working-age popuation decined, but this decine was eventuay reversed as the generation aged. Both the growth in the size of the working-age popuation and the aging of the baby boom generation have contributed to recent growth in appications and awards for SSA s disabiity programs. The SSA area popuation between ages 15 and 64 grew at an average annua rate of 1.1 percent from 1975 to 1992, but the growth in recent years has been much sower than in earier years. From 1975 to 1980 the average annua growth rate was 1.5 percent, whie it was ony 0.6 percent from 1988 to During the ater period, however, changes in the age distribution of the working-age popuation substantiay offset the effect of the sowing of popuation growth. There is a strong positive association between age and the incidence of disabing conditions. The entry of the baby boom generation initiay resuted in a decine in the average age of the SSA area popuation, but with the aging of the baby boom generation this decine has reversed. Consequenty, changes in the age distribution had a strong negative effect on appication and award rates during the eary part of the period, thereby moderating the effect of the increasing size of the SSA area popuation, but, for the period, aging of the popuation added Socia Security Buetin Vo. 58, No. 4 Winter

6 to the effect of popuation growth. Based on 1988 age-specific appication rates and popuation growth by age from 1988 to 1992, we estimate7 that popuation growth and aging together account for average annua DI appication growth of 1.3 percentage points and SSI appication growth of 1.2 percentage points8- both more than twice the average annua growth rate of the working-age popuation. Resuts for awards are amost identica. Popuation growth during the period did not expain increased SSI appications and awards for chidren. Actuay, from 1975 to 1988 the number of persons 18 years and younger decined by 2.1 percent. Whie the number of chidren increased by 4.8 percent from 1988 to 1992 (an annua growth rate of 1.2 percent), this growth is dwarfed by the exposion of SSI appications and awards among chidren for this period of time. Target popuations.-the number of DI and SSI appications and awards shoud be infuenced by changes in the size of the popuation eigibe for either or both programs, that is, each program s target popuation. The most important eigibiity factors are the presence of quaifying disabiities and economic eigibiity. The disabiity criteria are identica for the two programs; economic eigibiity is tied to disabiity-insured status for DI and to a means test for SSI. A three of these criteria are infuenced by factors externa to the DI and SSI programs, as we as by egisative, administrative, and judicia factors. Our focus here is on exogenous changes in the size of the target popuations. To be eigibe for DI, a person has to satisfy the insured status and disabiity requirements. SSI eigibiity requires meeting the means test and the disabiity requirement. Persons with quaifying disabiities who are disabiity insured but do not meet the means test are eigibe for DI ony; those who are disabiity insured and meet the SSI means test quaify for both programs (concurrent eigibiity); and those who meet the SSI means test but are not disabiity insured are eigibe for SSI ony. Unfortunatey, based on currenty avaiabe data we cannot observe time series on the three main target popuations directy, and indeed, not even crosssectiona data are avaiabe in which the popuation satisfying the disabiity criteria in the genera popuation is identified. Therefore, we must rey on an item by item examination of evidence on trends in these three target popuations. The share of the DI-insured popuation grew at an average annua rate of 1.2 percent from 1975 to The rate of growth was much higher for women (2.6 percent) than for men (0.2 percent), refecting the increase of femae aborforce participation rates (chart 2). The narrowing of gender differences aso suggests that this source of increased growth is approaching exhaustion. Actuaria anaysis of the contribution of changes in the size and the age/gender composition of the disabiity insured popuation to the growth of DI appications from 1988 to 1992 found an average annua contribution of 2.1 percentage points-o.8 percentage points greater than the estimated impact of popuation growth and aging aone-with amost a of the added contribution due to changes in the disabiity insured status of women (Lewin-VHI 1995a). Resuts for awards were amost identica. Importanty, the growth in the proportion of the disabiity insured popuation suggests an increase in the share of SSI eigibes concurrenty quaifying for DI, thereby depressing the growth of the SSI-ony group, particuary for women. To proxy for trends in the popuation that is economicay eigibe for SSI, we examined changes in the poverty popuation for working-age aduts and chidren between 1975 and During the ate 1960 s and eary 1970 s, Nobe Laureate James Tobin and Robert Lampman optimisticay predicted the eimination of poverty by However, between 1975 and 1992 the size of the workingage poverty popuation after government transfers increased from 11.5 miion to 18.3 miion, increasing the working-age poverty rate to 9.4 percent for maes and 13.9 percent for femaes. Pre-transfer poverty is of more reevance for assessing the impact of poverty on SSI eigibiity, but consistent measures are not avaiabe before From 1979 to 1992, the pretransfer poverty rate for the working-age popuation grew at an average annua rate of 1.6 percent. Growth was highest for persons aged and in the subperiods and (chart 3) both periods of sow economic growth or even decine; in the atter period, the average annua growth rate of the pre-transfer poverty rate was 3.5 percent. Assuming that increases in the poverty rate directy transate into increases in SSI appications on top of the effects of popuation growth and aging, these factors combined account for 4.7 percentage points of the average annua growth in SSI appications over this period, or about 45 percent of the average annua growth of 10.5 percent. The pre-transfer poverty rate for chidren aso grew substantiay during the period, and, as with the adut rate, growth was greatest in the first and ast few years of the period. For the period, the chid poverty rate grew at an average annua rate of 2.9 percent. Whie this growth is substantia, it can account for ony a very sma fraction of the 44 percent average annua growth in chid SSI appications over the period. Athough trends in poverty are particuary important proximate determinants of SSI growth, these trends may refect a variety of underying reasons affecting the size of the financiay eigibe popuation that are of interest in their own right. Ideay, we woud ike to know how various factors that are behind the growth in poverty--changes in the economy and in famiy structure-affect program participation. It shoud aso be kept in mind that poverty rates may be imperfecty correated with the percentage of the popuation that satisfies the SSI means test. We incuded the poverty rate as an expanatory variabe in our D-concurrent and SSI regressions for , but found that it did not have a statisticay significant effect on appications and awards. A simiar finding was obtained in the anaysis of initia determinations. Athough marginay significant, positive coefficients were obtained when 48 Socia Security Buetin Vo. 58, No. 4 Winter 1995

7 data for the subperiod aone were used. The weak findings might be attributabe to substantia measurement errorsin State-eve poverty rate estimates. Another expanation is that a major determinant of the poverty rate, unempoyment, is incuded separatey in a of the anayses, so ony variations in the poverty rate that are not expained by the unempoyment rate (or other expanatory variabes) are being used to identify the impact of poverty. In addition, the incusion of the percent of chidren iving with ony one parent as an expanatory variabe in the anaysis may aso be an effective contro for another important determinant of poverty-growth in the number of femae-headed househods. Future work on the reationship between poverty, SSI means-testing, and disabiity appications and awards coud be further enhanced on the basis of the rich source of information provided by SSA s matched SIPP data fies. The findings with respect to the percent of chidren iving with ony one parent merit further discussion. We incuded this variabe in the regressions as a proxy for changes in famiy structure-decines in marriage rates that have eft many individuas with no source of support other than their own earnings. This variabe accounts for a significant amount of SSI appication growth as we as DI-concurrent appication growth-on the order of 5.0 percent of annua growth over the period. Effects were somewhat arger for women than for men, were arger for younger age groups than for oder age groups, and were con- centrated in the menta disorders category. A negative association between severe menta iness and marriage has been documented in the menta heath iterature; individuas who are mentay i are ess ikey to marry than others, and are more ikey to get divorced if they do marry (Barte and Taubman 1986). Our findings are consistent with the notion that decines in the avaiabiity of economic, physica, and emotiona support from a spouse may be contributing to growth in appications and awards. The data avaiabe to study the prevaence of disabing heath conditions is imited, especiay for anayzing trends. Long-term trends in the prevaence of disabing conditions may be infuencing ong-term growth in appications and awards (in some cases negativey), but Chart 2.-Percent of the working age popuation that is disabiity insured, Percent i 1 -c Men under40 --% Women under t Tota a-.- Tota Year Socia Security Buetin Vo. 58, No. 4 Winter

8 with one exception (AIDS/HIV), we did not fmd convincing evidence of heath trends expaining the recent acceeration of appication and award growth. As wi be discussed ater, however, expansion of the popuation with quaifying disabiities due to egisative and reguatory changes and increased awareness of disabing conditions might have substantiay contributed to appication and award growth. The incidence of AIDS/HIV grew at an annua rate of 9.3 percent from 1988 to Our regression estimates for , aong with counts of the number of appications in the AIDS/HIV impairment category, suggest that AIDS/ HIV accounts for between 0.6 and 0.9 percentage points of both DI and SSI appication growth over this period. SSI appications from ega aiens and those iving in the United States under the coor of aw grew much more rapidy than those from citizens-at an average annua rate of 17.4 percent from 1988 to 1992, versus 9.8 percent for citizensathough the share of a appications from the former group is sti sma (6.8 Chart 3.-Percent of persons in poverty before transfer payments, by age, c 15 IO Percent 5 I-~~ O C Year percent in 1992). We previousy have hypothesized that the Immigration Reform and Contro Act (IRCA) expained the reativey rapid growth among appications from this popuation. Nationa time series data of IRCA egaizations show a striking resembance to nationa time series data for SSI appications from ega aiens (Lewin-VHI 1994). Because IRCA egaizations are concentrated in a reativey few States, we expected that any impact of IRCA egaizations woud be ceary distinguished in the appication and award anaysis for In fact, however, the findings were very weak. To verify the econometric findings, the number of annua SSI awards to IRCA immigrants in a o-percent sampe of a SSI appications was tabuated for the period from 1989 (the first year of IRCA egaizations) to The number identified as IRCA immigrants turned out to be very sma-peaking at an estimated 3,200 of the 88,500 appications from a ega aiens in Thus, the rapid growth in ega aien appications over this period appears to be party due to the same factors that are behind the growth in appications from citizens. Thus, IRCA is apparenty not responsibe for the reativey rapid growth of appications from noncitizens. In the anaysis of the data we examined whether growth in the number of egaized immigrants who have satisfied the 3-year waiting period coud expain this phenomenon, but again found no significant resuts. Evidence from the case studies suggests that the recession had a much arger impact on the immigrant popuation than on citizens, but we have not tested the hypothesis empiricay. It is aso known that middeman fraud has payed a roe in heping immigrants in some areas obtain awards, but the extent of the fraud is unknown.1 An important feature of our findings concerning popuation factors is that they expain why growth in concurrent appications has been greater than growth in appications and awards for either program aone, and especiay why concurrent appication and award growth has greaty exceeded that in the DI-ony category. Femae and young DI appicants are more ikey to meet the SSI means test than are oder mae DI appicants, and growth in the disabiity insured popuation has been greatest for women and for the young to midde age groups. The effects of poverty and changes in famiy structure have roughy equa impacts on concurrent and SSI-ony appications and awards, but at most have sma impacts on DI-ony appications and awards. Finay, the effect of AIDS/ HIV on concurrent appications and awards has been substantiay greater than its effects on those in either the DIony or SSI-ony categories. Our anaysis suggests that the increased size of both the disabiity insured and the poverty popuations contributed to the growth of DI and SSI appications. Because of the substantia increase in the size of the poverty popuation and the dramaticay increasing proportion of women who are DI-insured, the increase in the size of the target popuation satisfying both DI and SSI criteria appears to have grown most rapidy, expaining the preeminence of this group in both appication and award growth. 50 Socia Security Buetin Vo. 58, No. 4 Winter 1995

9 Business Cyces There have been numerous previous econometric studies estimating the effect ofthe business cyce on DI appications, awards, and caseoads. Most of the previous studies used aggregate time-series methods, athough some work has been conducted using State or individua eve cross-sectiona data (chart 4). The point estimates vary across individua studies, but no study has found substantia effects in a direction opposite from the predictions of economic theory. Previous studies have suffered from various specification probems or ow statistica power, or both. One of the key resuts from our work using annua pooed cross-sectionitimeseries data for States reates to our estimates of business cyce effects, because our abiity to contro for permanent differences among the States and to eiminate the confounding effect of nationa changes endemic to time-series studies makes the resuts obtained from our anaysis methodoogicay much stronger and more credibe. Strong resuts were found in both the anaysis of appications and awards and the anaysis of initia determinations and initia aowance rates (chart 4). In genera, we found stronger effects for appications than awards, and for the DI program than for SSI. In the anaysis of initia determinations we found that the impact of a change in unempoyment begins in the year of the change, but is greatest 2 years after the change. Such agged effects are presumaby greater for initia determinations than for appications because of the substantia ag between fiing and the initia determination, but nonetheess they coud be very significant. We did not examine this issue in the appication anaysis. We were abe to extend our DI initia determination anaysis back to 1976, and found remarkaby stabe unempoyment effects for DI in each of three subperiods , and We aso found that unempoyment effects for SSI-ony initia determinations were essentiay as arge as for DI-ony and concurrent initia determinations in the period, but we did not find an unempoyment effect for SSI-ony initia determinations in the period. The atter finding may be reated to the fact that the SSI-ony data incude chidren. In the initia determination anaysis we aso found evidence of a discouragedworker effect; hoding the unempoyment rate constant, a decine in abor-force participation as individuas give up their search for work during a recession is associated with a significant increase in initia determinations. The findings from the five case studies add credibiity to the econometric findings about business cyces, suggesting, if anything, that they are conservative. It is cear from the case studies that subteties of business cyces not captured by the unempoyment rate are reevant to a recession s impact-the industria distribution ofjob osses, the perceived permanence of ayoffs, and key characteristics of workers who ose their jobs, such as age, gender, prior earnings, and skis. We beieve that the econometric estimates of the effect of the unempoyment rate do not fuy account for the effects of the business cyce. First, because of errors in the measurement of the unempoyment rate variabe, the estimates tend to understate the effects of the unempoyment rate. Second, because the unempoyment rate does not fuy capture important aspects of the business cyce that are expected to affect appications (for exampe, discouraged-worker effects), we beieve that our estimates tend to provide a conservative assessment of the proportion of the appication and award growth attributabe to business cyce effects. The case study evidence supports this interpretation. The business cyce findings suggest a need to search for improved programmatic responses to the business cyce. Current program incentives make it highy ikey that marginay quaified appicants drawn into the program by the business cyce wi provide a ong-term burden for the disabiity ros even if business conditions Improve and the severity of the person s disabing condition(s) does not worsen. We know reativey itte about the mechanisms through which business cyces affect program growth. We cannot determine, for instance, the extent to which our resuts refect the effects of State and oca fisca responses to recessions rather than appications by workers with serious disabiities who ose their jobs or whose spouses ose their jobs. The smaer estimated business cyce effects for SSI-ony appications and awards-in comparison to the DI findings-is consistent with the hypothesis that much of the DI effect is due to job osses by workers with disabiities. Findings from the case studies support this interpretation as we, but they aso provide evidence of an important roe for State and oca fisca responses to revenue osses, a subject we wi return to ater. The dynamic aspects of business cyce impacts are aso poory understood. The considerabe agged effects found in the initia determination anaysis suggest that many individuas who are induced to appy during a recession ony do so after an extensive search for other sources of support. Economic Restructuring Many have hypothesized that economic restructuring-the repacement of high paying manufacturing jobs with reativey ow paying service sector jobs-has had an impact on appication and award growth. The short-term effect of economic restructuring is thought to increase appications, because disabed workers who ose their manufacturing jobs may choose to appy for disabiity benefits rather than find new work in the service sector. The ong-term effect may be to decrease appications, however, because service sector workers are ess susceptibe to disabing injuries and inesses (see Loprest, Rupp, and Sande 1995). The ong-term effect may vary by impairment group; for instance, some have suggested that the effect is negative for physica impairments but positive for menta impairments. We previousy specuated that the arge business cyce effects found in the appication anaysis may party refect the short-term, positive impact of Socia Security Buctin Vo. 58. No. 4 Winter

10 Chart 4.-Estimates of the effect of a 1 -percentage point increase in the unempoyment rate on disabiity program growth for aduts Estimated effect of a 1- percentage Studv Data tvne I Period ooint increase in unemnovnicnt Appications (Hambor. 1975) Quartery, nationa % for DI (Lando, 1974) Quartery. nationa % for D (Lando. Coate, and Krauss, Quartery. nationa % for D I. / (Hapern, 1979) Quartery, nationa Negigibe for D (Muer, 1982) Annua, individua. cross-section 1972 Negigibe for D (Levy and Krute, 1983) Annua, individua, cross-section 1978 Ncgi&ibe for D f Hambor. 1992) Annua. nationa Negigibe for DI 4% for DI-ony. (Stapeton. Coeman. and Annua. pooed, cross- 4% for DI-concurrent. Dietrich, 1995) scctionitime-series % for SSI Initia determinations (Stapeton and Dietrich, Annua, pooed cross % in year of change. 1995) sectionitimc-series 3% after I year. (SSI-ony incudes chidren) 5% after 2 years 2% in year of change. 4% after 1 year. 5% after 2 years 0% in year of change 1% after I year 3% after 2 years Awards (Hambor, 1975) I Ouarterv. nationa I % for D (Lando, 1979) Annua, State-eve cross-section 1975! 5-6% for DI \- (Muer. ~~~~~ I 1982) I 1 I Annua. individua. cross-section Negigibe for DI (Levy and Krute. 1983) Annua. individua. cross-section 1978! Negigibe for DI (Hambor, 1992) I Annua. nationa Negigibe - - I 3% for DI-onv. i (Stapeton, Coeman, and Annua, pooed, cross-section/ 2% for DI-concurrent. Deitrich 1995) I time-series I 2% for SSI Initia aowance (Stapeton and Dietrich, Annua. pooed cross D- 0 points in year ) section/time-series ony change, (SSI-ony incudes chidren) -1 point after 1 y ear, -1 noint after 2 vears Con- 0 points year of change. cur- -1 point after I year. rent -2.points after 2 years SSI- 0 points year of change. ony -I point after I year, Beneficiaries -2 Points after 2 y ears 1 (Thompson and Van de Water. 1975) Quartery. nationa % for D I I I (Cromwe ct a ) Quartery, poo~e~i$!$tte-cvc cross Ncgigibc for SSI Medicaid enroees rate 52 Socia Security Buetin Vo. 58, No. 4 Winter 1995

11 economic restructuring (Lewin-VHI 1995b). In the initia determination anaysis we tried to capture this effect using the percent of empoyment in manufacturing as an additiona expanatory variabe. We did find the expected negative effect for the DI-ony category, but it was sma and not repicated for other program categories. We aso deveoped two indices of job-reated injuries and inesses to capture the onger-term impact of economic restructuring, but found no significant resuts. Athough it may be that measurement and other specification errors account for the weak findings, this is aso consistent with the hypothesis that business cyce effects overwhem the effects of economic restructuring in the periods we have examined. Oiher Support Programs Just as economic theory suggests that the reative vaue of disabiity cash benefits to potentia earnings affects the decision to appy, it is reasonabe to expect that the avaiabiity and reative vaue of benefits through other programs shoud aso affect the decision to appy. This is an important topic, particuary in ight of substantia secuar changes in the reative vaue of pubic benefits such as genera assistance (GA-the generic term for wefare programs funded entirey by State and oca governments), Aid to Famiies with Dependent Chidren (AFDC-a State/Federa program that primariy provides support for ow income, singe-parent househods), Medicaid, and Medicare. Other programs can be cassified as either substitutes or compements for DI and/or SSI, in the economic sense of these terms. Substitute programs are those for which an expansion in the vaue of benefits reduces appications and awards for the SSA programs; benefit expansion for compementary programs increases appications and awards. AFDC is a cear exampe of a substitute program for SSI. Individuas cannot receive benefits from both programs, and if eigibe for both AFDC and SSI, must choose which benefits to receive. Increases in the reative vaue of AFDC benefits are expected to decrease the reative attractiveness of SSI, whie decreases in the reative vaue of AFDC benefits shoud have the opposite effect. Medicaid and Medicare are ceary compements of SSI and DI, respectivey; most SSI recipients are automaticay eigibe for Medicaid, whie DI beneficiaries receive Medicare coverage after a 2- year waiting period. Increases in the cash vaue of Medicaid and Medicare benefits increase the reative attractiveness of SSA s disabiity programs, and hence the demand for their benefits. Changes in eigibiity rues for other programs can change the degree to which they are substitutes or compements for the SSA disabiity programs. For exampe, expansion of Medicaid to individuas who are not sufficienty poor to quaify for SSI or universa heath insurance coverage woud reduce or eiminate the compementarity between medica insurance and disabiity programs. In our State-eve anaysis for the period, we found strong evidence of effects for State and oca genera assistance programs both on appications and awards. GA reductions in seven States and the District of Coumbia had highy significant, positive effects on SSI appications and awards for both men and women, and on concurrent appications and awards among men. Estimated effects on appications and awards were neary identica, and the easticities were often arge, particuary for younger men, and for appications and awards in the menta disorders category. We ater found simiar resuts for initia determinations in both the and periods. For the period we aso found evidence that both reductions in AFDC benefits and increases in State SSI suppements increased SSI initia determinations, but these findings were not repicated in the anaysis. The ack of findings for the ater period may simpy refect ack of arge changes in either AFDC or State SSI suppements. These findings are the first cear econometric evidence we are aware of demonstrating that changes in other income and in-kind transfer programs have an impact on SSI appications and awards, but the ack of previous progress in this area may just refect the difficuty of measuring such effects and the virtua absence of serious research efforts in this area to date. One reason for the ack of previous research is a common beief that anyone who is eigibe for SSI as we as either AFDC or GA woud aready have appied for SSI because SSI benefits are greater. As severa wefare administrators and other wefare experts have tod us, this reasoning negects the fact that the SSI appication and appeas process is prohibitivey difficut for many who can much more readiy quaify for GA or AFDC+speciay those with menta disorders. A primary objective of the case studies was to earn more about the impact of changes in State and oca wefare programs on SSI appications and awards. We found that cuts in GA benefits during the period represent ony a fraction of State and oca efforts to shift wefare recipients-primariy GA recipients-onto SSI. The most dramatic exampe of shifting efforts from this period occurred in Michigan, which terminated its GA program and, simutaneousy, aunched a arge, coordinated effort to hep GA recipients obtain SSI awards. Interviews with State and oca officias and others in Michigan as we as foowup empirica anaysis confirmed that the termination of GA, and the coordinated efforts to hep former GA recipients, expain why SSI award growth in Michigan was the second highest in the nation during this period (Bound, Kossoudji, and Ricart-Moes 1995). We aso found new or intensified efforts to shift GA and other wefare recipients to SSI in other case study States, and significant efforts in many other States have been reported as we (see, for instance, Bordeon 1995). The findings from the case studies suggest that the econometric modes understate the impact of the combination of GA cuts and other State and oca shifting efforts. The GA cuts variabe used for the anaysis is a crude proxy for genera State and oca wefare changes and shifting efforts; we beieve that its estimated coefficient probaby understates the impacts of these changes because it fais to capture the effects of shifting efforts that do not invove cuts in GA benefits. 53 Socia Security Buetin Vo. 58. No. 4 Winter 1995

12 Athough the findings from the case studies and econometric anaysis provide much ess support for the hypothesized effect of AFDC benefit changes on SSI, the AFDC findings for , a ongterm decine in the vaue of AFDC benefits reative to SSI benefits (from 1975 to 1992 the eve of median AFDC benefits for a famiy of four decined by 37 percent reative to the vaue of Federa SSI benefits for coupes) and evidence that a substantia share of AFDC mothers have disabiities (see Ader 1993) suggest that AFDC program changes have contributed to ong-term SSI appication and award growth. Proposed future reforms to both AFDC and GA programs coud have a substantia positive impact on SSI caseoads. We aso attempted to estimate the effect of the rising vaue of Medicaid benefits on SSI appications and awards, but were not successfu in identifying an effect. It seems ikey, however, that the absence of a positive finding refects the difficuty of measuring the vaue of the benefits rather than the absence of true effects. Wefare administrators and other experts generay attest to the importance of Medicaid benefits to SSI appicants, and recent research on the reated topics of continuation of coverage mandates (Gruber and Madrian 1993), the effects of Medicaid on AFDC caseoads (Moffitt and Wofe 1992; Congressiona Budget Office 1993; and Yeowitz 1994) confirm the importance of medica benefits to, abor-force and program participation decisions. Yeowitz (1995) expores the effects of Medicaid on SSI. In the case studies, we found that the burden of heath care costs on State and oca governments and heath care providers was a major factor behind efforts to shift GA recipients and others onto SSI. Heath care expenses for indigent patients who are not Medicaid recipients are usuay paid by State and oca governments or, impicity, by providers themseves. When an individua who is not Medicaid eigibe obtains eigibiity via SSI, the Federa Government begins to pay at east haf of these expenses. State and oca savings from reduced heath spending may be greater than savings from reduced cash benefits. The fact that AFDC beneficiaries are aso eigibe for Medicaid heps expain why State and oca efforts have focused on GA recipients. Medicaid reform or genera heath care reform coud have a significant effect on SSI caseoads. Medicaid bock grants, which woud resut in Federa payments to States that are not tied directy to Medicaid enroment, woud significanty reduce the incentives to shift State and oca wefare recipients onto SSI. Cutbacks in Medicaid benefits coud aso have a negative effect. Making Medicaid benefits avaiabe to persons with disabiities independenty of SSI, or otherwise increasing their access to heath insurance, woud aso be ikey to reduce SSI caseoad growth. Features of SSA s Diwbiity Programs und Other Suppy Fcctors (1) there is ony imited variation in the data; (2) most changes that do occur tend to affect the whoe program, so there are no natura comparison groups; andor (3) it is extremey difficut to disentange the effect of programmatic factors from potentia confounding factors. Nevertheess. we review here severa research directions with some promise of a better understanding of the roe of programmatic factors. Previous econometric work has addressed some important programmatic factors. Most importanty, there is a considerabe body of econometric work since the pioneering work of Parsons and Leonard focusing on the effect of wage repacement rates on abor-force and Features of SSA s disabiity programs-such as the rea vaue of benefits, egisative and administrative actions affecting eigibiity determination, work incentive provisions. and SSA outreach activities-might substantiay affect appications and awards. Other suppy factors-such as judicia ruings on appeaed cases-aso pay a roe. Research on the effects of these factors is, unfortunatey, extremey difficut to perform. for three reasons: disabiity program participation. This body of econometric work has been pagued by serious identification probems, and has produced a wide range of estimates. An aternative quasiexperimenta approach using rejected appicants as a comparison group (Bound 1989) raised fundamenta questions about the vaidity of these estimates, but reies on somewhat questionabe assumptions as we. In the future, potentia new insights might be gained in this important area by using information on changes in the benefit fomrua that occurred during the seventies to identify the effects of the reative vaue of DI benefits. Economic theory suggests that the expected probabiity of award and future benefit streams shoud affect appications, and, therefore, changed eigibiity rues and their enforcement might be important directy in determining not ony awards, but aso the poo of appicants. Two pioneering studies that performed State-eve anaysis for the 1970 s that is simiar methodoogicay to our own anaysis for ater years focused on the impact of changes in initia denia rates on DI appications and abor-force participation, taking advantage of what appear to have been exogenous changes in State denia rates. Parsons (1991) estimated that a 1 O-percent increase in denia rates from 1977 to 1978 decreased appications by 4.5 percent from 1978 to Gruber and Kubik (1994) refined Parsons anaysis in some respects, but came to essentiay simiar concusions. One imitation of both studies is that they did not contro for changes in unempoyment at the State eve or growth and aging of the popuation during this period. We repicated Parsons findings and then found that controing for the unempoyment rate and growth and aging of the popuation reduced the size of the estimated effect by about 50 percent. Ceary more work is needed in this important area, and the focus of econometric work shoud be extended to reated areas, such as the effect of other important features of the disabiity determination process (such as processing times) on appication behavior. Such studies woud be particuary timey in ight of current efforts to streamine the disabiity determination process Socia Security Buetin Vo. 58. No. 4 Winter 1995

13 Our econometric anaysis of State data provides no direct evidence of program suppy effects, by design. Despite the ack of any direct evidence, it is possibe to make some inferences concerning the effects of suppy changes indirecty. Recent growth not accounted for by the effects of factors in the modes (such as demographic changes, the business cyce, GA program cuts, and so forth) represents an impicit upper bound on the net effects of suppy changes on appications and awards. Our anaysis of appications, for instance, accounts for 77 percent of DI-ony mae appication growth from 1988 to 1992, eaving ony imited room for the net effect of either suppy factors or other omitted factors.? Our modes account for substantiay smaer shares of DI-concurrent appication growth for men (58 percent), DI appication growth for women (46 percent), SSI appication growth for both men and women (50 and 30 percent, respectivey), growth in appications from those under 50 (46 percent for DI and 39 percent for SSI), and growth in appications in the menta and muscuoskeeta categories (36 percent for DI and 29 percent for SSI). The econometric modes aso account for much ess award growth than appication growth. The award modes account for 23 percent of DI and 7 percent of SSI award growth. The fact that the proportion of award growth accounted for by the same factors that were incuded in the appication modes is ower is consistent with the hypothesis that SSA s disabiity programs screen out marginay quaified appicants drawn into the appication process by the business cyce and other factors, such as GA program cuts. In fact, athough aowance rates increased over the period, the modes predict that they shoud have decined. The finding that recessions have a negative effect on aowance rates was strongy confirmed in our anaysis of initia aowance rates for It is a mistake to attribute a of the appication and award growth not accounted for by the variabes in the modes to suppy factors. These shares do not take into account DI growth due to increases in the share of women who are disabiity insured. The actuaria anaysis shows that this factor by itsef accounts for substantia appication and award growth for women, in younger age groups, and in the menta disorders categories.13 We aso beieve that, if anything, the econometric anaysis is ikey to understate the effect of the recession, GA benefit cuts and associated shifting efforts, and other factors that we were abe to measure ony imperfecty. Nonetheess, there are strong reasons to beieve that a substantia share of unaccounted for appication and, especiay, award growth is due to suppy factors. With respect to awards, it is difficut to conceive of an aternative expanation for the growth in the aowance rate. With respect to appication growth, the fact that amost a of the appication growth in the interna organs category is accounted for by factors in the mode (chart 5), whie the other diagnostic groups show substantia unaccounted for growth, is consistent with the hypothesis that reguatory changes-such as increasing the weight given to pain and other symptoms, increasing reiance on source evidence (for exampe, evidence from the appicant s own heath care provider), and broadening the standards for those with menta impairments-resuted in substantia appication growth during the period.14 A number of cautions are in order in interpreting the anaysis of growth not accounted for in the modes focusing on the voume of appications and awards disaggregated by impairment category. First, appication growth in some categories may simpy refect switching of impairment cassifications toward categories in which it has become easier to obtain an award rather than appications that woud not have been tied in the absence of suppy changes. Second, it is not possibe to sort out the effects of various specific suppy changes from these resuts. Third. there are some competing expanations of why unaccounted for growth is especiay high in some impairment categories. For instance, unaccounted for growth in the menta disorder category may refect the fact that State and oca efforts to shift wefare recipients onto SSI often target popua- tions with a reativey high prevaence of menta disorders. In a simiar vein, comparison of chid and adut SSI disabiity awards demonstrates that the exceptionay high growth of chid awards from I990 to 1992 was due to suppy factors (chart 6). Unti 1989, growth in awards for chidren and aduts appeared to respond to very simiar forces, but chid awards expoded reative to adut awards from 1990 to This evidence provides support for the hypothesis that the recent rise of chid SSI awards is argey attributabe to the February 20, 1990, Suivan v. Zebey Supreme Court decision (commony referred to as the Zebfey decision), and the 1991 revisions to the chidhood menta disorder istings (GAO 1994). The direct estimation of the number of awards affected by reguatory changes, assuming no behaviora response, is another approach that can often provide usefu information. For exampe, Hannsgen and Sande (1995) simuated the effect of SSA s revised 1992 deeming rues by using the od rues to create a counterfactua and concuded that the seemingy minor rues change resuted in a permanent 2-percent increase in the number of chidren awarded SSI benefits. Overa, we concude that ececticism is ikey to continue to be a virtue in this important and compex area. Actuaria and simuation methods are particuary we-suited for estimating the effect of programmatic changes to the extent that behaviora responses can be reasonaby assumed to be nonexistent or negigibe. Economic anaysis is caed for in assessing the effects of changes in the economic environment and to tacke the more compex behaviora effects of programmatic changes. Concuding Comments on Appications and A wards Changes in the size of the disabiity insured popuation and our econometric anaysis of State-eve changes in appications and awards account for a substantia portion of DI appication growth, especiay for men. Our anaysis aso accounts for a substantia share of SSI appication growth, athough not as much Socia Security Buetin Vo. 58. No. 4 Winter

14 as for DI. Popuation factors are the primary reason that concurrent appications are growing faster than those for either program aone. Based on the econometric anaysis, our review of the iterature, interviews with experts, and case studies in five States, we concude that the acceeration of appication and award growth during the period, above ong-term trends, is argey attributabe to three factors: (1) the recession; (2) cuts in genera assistance programs and other efforts by States and ocaities to shift wefare recipients to SSI; and (3) a variety of suppy factors that had the effect of expanding eigibiity. The importance of these three factors varies by program category and is different for awards than for appications. The recession was reativey more important for DI appications than for SSI appications, especiay in the DI-ony category, whie genera assistance cuts and State and oca shifting efforts had their greatest impact on SSI appications. The effects of suppy factors on awards appears to be much greater than on appications. One other factor, the AIDS/HIV epidemic, ceary contributed to the acceeration of appication and award growth during this period, but its roe appears to have been reativey modest in comparison to the importance of the other three factors. The effects of each of these three factors on appication and award growth rates are sef-imiting. That is, uness there is continuing change in the factor itsef we woud not expect the rapid growth it caused initiay to continue in the future. Recessions end, and even if they did not, it seems ikey that their effect on appications and awards woud diminish after a wave of appicants induced by initia job osses passes through the appication process. As is evident in the case of Michigan, new State and oca efforts to shift wefare recipients onto SSI cause a arge surge in appications as the wefare recipients who are most ikey to be eigibe for SSI appy, but, after this surge, appications subside as the poo of potentia appicants who meet eigibiity requirements diminishes. Simiary, suppy changes that expand eigibiity initiay draw many appicants from the poo of newy eigibe persons, but appication growth subsides as the number remaining in the poo diminishes; this was most evident in the surge of adut appications foowing the 1985 revisions to the adut menta impairment istings and in the surge of chid appications foowing Zebey and the new chidhood menta impairment istings. We beieve that the major quaitative impication of these findings for future DI and SSI projections is that we need to discount the importance of recent deviations from ong-term trends in projecting future trends in appication and award growth.15 This is-especiay true of business cyces because economic recovery can be expected to foow a recession and this shoud have the opposite effect on appications. It is aso true of once-andfor-a changes in State and oca efforts to shift wefare recipients and in the suppy of benefits; such changes wi ead to permanenty higher eves of appications and awards, but not permanenty higher growth rates. Length of Stay and Terminations Caseoad growth is affected not ony by appications and awards, but aso by the ength of time new awardees stay on the ros. This is particuary important in the case of programs, ike the DI and SSI disabiity programs, where average program stays are ong and substantiay vary by characteristic, as is ceary demonstrated by the pioneering studies of Hennessey and Dykacz (1989) and Rupp and Scott (1995a). Chart 7 shows cohortbased estimated mean ength of stay on the DI and SSI ros for working-age DI and SSI-ony new awardees and SSI chidhood new awardees. The expected mean tota stays of DI and SSI adut awardees during the preretirement years Chart 5.-Annua appication growth rate, accounted for and not accounted for, by program and impairment, Average growth rate - Menta Musc~~o- nfecious Interna Menta Muscuo- nfecious Interna disorder skeeta diseases organs disorder skeeta diseases organs DI SSI 56 Socia Security Buetin Vo. 58, No. 4 Winter 1995

15 are roughy comparabe (approximatey 10 years).r6 Rupp and Scott (1995a) estimate that based on past program experience chidren, in contrast, are expected to spend, on the average, amost 27 years on the SSI ros prior to reaching their 65th birthday. Mean SSI engths of stay among concurrent awardees is much shorter than among nonconcurrent adut awardees, primariy because many concurrent awardees ose SSI benefits once they start to receive DI benefits. The most important piece of information on concurrent awardees-the combined ength of SSI and DI stays-has not been estimated yet. Expected ength of stay on both the DI and SSI ros varies substantiay by age and diagnostic group. Length of stay is important in two respects: first, by transating new awards into benefit years, ength of stay determines the effects of new awards on utimate program caseoads and therefore substantiay infuences program costs; and second, the expected ength of stay may aso affect appication behavior. Because age, diagnosis, and the severity of disabing conditions are a powerfu predictors of ength of stay, changes in any of these factors coud affect program caseoads and the average ength of stay for those on the ros. Average ength of stay is particuary sensitive to outiers, and therefore to any factors affecting the proportion of ong stayers. Rupp and Scott (1995a) estimate that whie SSI awardees with expected preretirementage disabiity stays of 10 years or more comprise ony about one-third of new awardees, they comprise 83.3 percent of the impied eventua caseoad in SSI. Athough the proportion of ong-stayers is somewhat ower in DI (as a resut of the absence of chidren in the disabedworker program), ong-stayers are much more important in both programs than in AFDC, the program defining the popuar image of wefare dependence. Legisative and reguatory changes, such as the conduct of Continuing Disabiity Reviews (CDRs) and iberaized work incentives under both SSI and DI, might affect caseoads directy through infuencing ength of payment eigibiity spes, and indirecty through the effect of impied changes in expected ifetime benefit streams on appication behavior. Chart 6.-SSI bind and disabed awards as a percent of previous year s awards, Percent Year Socia Security Buetin Vo. 58, No. 4 Winter

16 There is some evidence that change in the mix of new awardees has contributed to a secuar increase in average ength of stay in both DI and SSI. According to Chirikos (1993) changes in the composition of DI awardees, primariy by age and diagnosis, is responsibe for an increase of average expected ength of DI spes from 8.5 years during to 10.7 years during Rupp and Scott ( ) using a different methodoogy, arrive at estimates consistent with the resuts of the Chirikos study. Rupp and Scott estimate that the onger term shift toward younger awardees during the period contributed to an increase from a simuated mean ength of stay of 9.5 years of first DI stays for persons first awarded benefits in 1975, to a simuated mean ength of first DI stays of amost 11 years among those first awarded benefits in Simuated ength of tota stay during the preretirement years in the SSI disabiity program changed even more dramaticay. Rupp and Scott (1995b) estimate that the average tota stay for chidhood and nonconcurrent adut SSI disabiity awardees combined was approximatey 12 years for persons first awarded ben- efits in In contrast, persons first awarded SSI disabiity benefits in 1993 are expected to stay on the SSI disabiity ros prior to reaching their 65th birthday for an average of amost 18 years, assuming that current program rues do not change dramaticay in the future. Much of this estimated increase in average SSI disabiity duration is attributabe to the recent infux of chidhood awardees in the SSI program. For working-age, nonconcurrent aduts the SSI trends are simiar to the DI trends. The trends attributabe to past changes in the age mix of new awardees are expected to create an upward pressure on future caseoads even if the characteristics of new awardee cohorts were to change markedy in the future. The strong negative association between age and expected ifetime stays demonstrated by Hennessey and Dykacz (1989) for DI, and Rupp and Scott (1995a) for SSI, has important impications for the disabiity caseoad effects of the aging of the baby boom generation. Rupp and Scott (1995b) estimate that the increase in the proportion of younger DI insured workers during the ast two decades expains about haf of the increase in the expected average duration of successive cohorts of new DI entrants. However, as the baby boom generation ages during the decades ahead, demographic factors are expected to have an opposite effect on duration. Rupp and Scott (1995b) estimates that changes in the age-composition of DI-insured workers between 1993 and 2006, by themseves, woud ead to an approximatey 1 -year decine in the expected duration of new DI awardees on the disabiity ros. To obtain a compete picture of the effect of demographic changes on caseoad the effect of age both on incidence rates and on expected duration needs to be considered. The two effects tend to work in opposite directions. Chart 8, based on Rupp and Scott Chart 7.-Mean of first spe and expected ifetime disabiity program stays 30 Years SSI chidren (1995b), shows that age is positivey associated with incidence rates in both the D and SSI disabiity programs. I7 In contrast, duration is negativey associated with age at award. Because the effect of age on incidence rates tends to be stronger than its effect on duration, the net effect on caseoads-expressed as benefit years associated with awardee cohort by age-tends to be positive. Thus, we can infer that the caseoad effects of the entry of the baby boom generation has been initiay moderated by the reativey ow disabiity incidence rates at younger ages. As the baby boom generation ages, the upward pressure on new awards arising from the strong positive association between disabiity incidence rates shoud be moderated by the associated anticipated decine in expected ength of stay; the resut is a muted effect on tota benefit years. Because the disabiity incidence rate rises more steepy 58 Socia Security Buetin * Vo. 58, No. 4 Winter 1995

17 with age than ength of stay decines, the (Hennessey and Dykacz 1993) suggests net effect of the two contrasting forces is that the iberaized DI work incentives sti an expected increase in caseoads introduced during the eighties actuay resuting from the aging of the baby might have increased ength of stay and boom generation. might have induced additiona appica- More research is needed to understand tions as we (see Hoynes and Moffitt the impications of other factors, such as 1994). More research is needed on the the business cyce, on the expected ength effect of CDRs and the medica improveof stay of new awardees. Factors affect- ment standards not ony on ength of stay ing the diagnostic mix of new awardees among the directy affected beneficiaries, (such as the revised menta reguations) but aso on appications.18 Current proas we as case severity (assessment of gram design suggests ony a imited roe pain, functiona assessments) might af- for vocationa rehabiitation and return to fect not ony the number of new awards, work (Rupp, Be, and McManus 1994), but aso expected ength of stay. In par- but more radica reforms might induce ticuar, egisative and reguatory changes arger effects on ength of stay in the affecting the proportion of cases with ow future. age-adjusted mortaity risk shoud affect Whie annua termination rates have expected ength of stay. decined over the ong term, such data Length of stay is aso affected by confound the effects of changes in the poicies focusing on peope on the ros. reative size of successive annua entry The cohort-based comparison of DI exit cohorts with changes in exit probabiities. rates during the seventies and eighties Cohort-based studies are better suited for Chart 8.-DI and SSI incidence rates, mean duration, and benefit years, expressed as a percent of the mean for year od DI awardees, by age ICKI 8a Percent understanding caseoad dynamics and the effects of changes in the characteristics of new cohorts of awardees on ong-term caseoad growth. In order to deveop a better understanding of appication and award growth, future studies inking award growth to caseoad dynamics are ikey to be fruitfu. Issues for Future Research Improvements in our understanding of factors affecting caseoad growth are important for improving our abiity to make projections, as we as to address the poicy issues raised by program growth in a proactive manner. Recent work confirmed the importance of demographics and ed to a better understanding of business cyce effects. More work is needed to assess the interaction of SSA s disabiity programs with other programs and ong-term changes in abor markets, as we as the interaction of award growth with caseoad dynamics. Most important, the assessment of the effects of changes in SSA s disabiity programs themseves cas for further improvements in our abiity to mode the effect of such changes, particuary their behaviora effects. Ongoing and future anaytic work using new data sets that provide a rich array of individua-eve information on variabes reevant to modeing behav- iora responses to economic and programmatic factors (such as the panned Disabiity Evauation Study, and the matched SIPP-SSA administrative microdata fies that have been deveoped by SSA s Office of Research and Statistics) wi substantiay enhance our understanding of the behaviora processes affecting disabiity program participation. Athough recent work increased our understanding of caseoad growth, ongoing changes in the economy and pubic poicies, as we as methodoogica cha- enges, suggest the need for a rigorous program of ongoing future research efforts in this important area. 2cQ 0 DI ss ci ss DI s Notes More detais of the findings on appications and awards can be found in five papers that were prepared for the Juy , 1995, conference on growth in the disabiity pro- Socia Security Buetin * Vo. 58, No. 4 * Winter

18 grams sponsored by the SSA and the Office of the Assistant Secretary for Panning and Evauation (Bound, Kossoudji, and Ricart- Moes 1995; Stapeton, Coeman, and Dietrich 1995; Stapeton and Dietrich 1995; Stapeton and Livermore 1995; and Stapeton, Livermore, and Zueschner 1995) and in the fina reports for four Lewin-VHI projects (Lewin-VHI 1995a, b, c, and d). More detais of the duration findings appear in another conference paper (Rupp and Scott 1995a) and in a previous artice in the Socia Security Buetin (Rupp and Scott 1995b). Other papers presented at the recent conference incude Muer and Wheeer (1995) presenting resuts from SSA s survey of fied office managers; Day (1995) focusing on the characteristics of SSI and DI recipients based on the Pane Survey of Income Dynamics data, and Yeowitz (1995) discussing the impact of heath care costs and Medicaid on the SSI program. z The aw provides a 45-month period for disabed beneficiaries to test their abiity to work without osing their entitement for benefits (Committee on Ways and Means 1994). The period consists of (1) a tria work period (TWP) which aows disabed beneficiaries to work for up to 9 months (within a 5-year period) with no effect on their disabiity or (if eigibe) Medicare benefits, and (2) a 36-month extended period of eigibiity. during the ast 33 of which disabiity benefits are suspended for any month in which the individua is engaged in substantia gainfu activity (SGA). Medicare coverage continues so ong as the individua remains entited to disabiity benefits and, depending on when the ast month of SGA occurs, may continue for 3 to 24 months after entitement to disabiity benefits ends. Note that ony one TWP is aowed in any one period of disabiity. The TWP is competed ony if the 9 months are within a 60-month period. By reguation, earnings of more than $200 a month constitute tria work. i A conceptuay simiar approach has been used by James Tobin (1994) in a recent anaysis of the effect of unempoyment on poverty rates. Because DI and SSI caims are not aways fied simutaneousy, this definition of concurrent appicants is not equivaent to SSI appicants who aso fied for DI; hence the abe D-concurrent. s For initia determinations the definition of concurrent is based on the status of caims at the time the determination is made. 6 The GA survey data were coected by Sherrie Kossoudji at the University of Michigan. We are gratefu to Aan Shafer of SSA s Office of Disabiity for faciitating the matching process. See Bound, Kossoudji, and Ricart-Moes (1995) for further detais. For this anaysis we first estimated what 1992 appications for each program woud have been for each age group in each State by appying the 1988 appication rate to the 1992 popuation in the age group and State. We then added across age groups and States to get a projected nationa vaue for We divided the projected vaue by the actua 1988 vaue, took the fourth root, and subtracted 1.O to get the annua growth rate reported. * During the period the average annua growth rate of DI appications was 8.9 percent, whie the corresponding figure for SSI was 10.5 percent. Thus, popuation growth and aging is estimated to account for a 15-percent share of DI appication growth and a 1 -percent share of of SSI appication growth during this period. 9 We note that there are severa ongoing data coection and anaytic efforts that are expected to resut in major improvements in our abiity to address these issues more directy in the future. Most notaby, anaysts at the Office of Research and Statistics at SSA assembed a matched data fie containing survey information from the SIPP and SSA administrative records at the individua eve. This data fie has the potentia for measuring DI-insured status and SSI financia eigibiity at the individua eve. Lahiri, Vaughan, and Wixon (in this issue of the Buetin) deveoped and tested a structura mode of the disabiity determination process using the SIPP/SSA matched data fie. SSA s panned Disabiity Examination Study is expected to provide detaied cross-sectiona information on the most important variabes affecting the disabiity determination process. Rupp and Scott (1995b) report cross-sectiona estimates of the size of the financiay eigibe SSI popuation, derived using the SSI microsimuation mode that was deveoped by Vaughan and Wixon using the 1984 SIPP (1989). o See Genera Accounting Office (1995). A tota of 6,500 fraud cases have been identified in the States of Caifornia and Washington, combined. ii Estimated easticities represent the estimated percent effect of a percent change in the given independent variabe on the dependent variabe (appications or awards). I2 Note that this upper bound refers to net effects of unmeasured factors. It is entirey conceivabe that even if the net residua is sma, there is room for potentiay arger effects that work in opposite directions. Nevertheess, we note that the most prominent suppy-side factors that have been hypothesized to affect appications and awards during this period were presumed to affect appications and awards in one directionpositivey. To the extent suppy-side factors operating in the opposite direction are negigibe, the upper bound woud appy to any singe suppy-side factor hypothesized to account for the unexpained residua. Note, however, that the independent variabes considered in our modes may aso pick up some suppy-side effects. I3 Our actuaria anaysis indicates that this factor aone accounts for 19 percent of femae DI appication growth. Because the actuaria and econometric anayses were performed independenty, however, we can not simpy add this share to the share accounted for by the econometric anaysis to get the share expained by a factors combined; the actua share accounted for by the combined factors together coud be either arger or smaer. i4 New istings for menta impairments in 1985 paced substantiay more weight on functiona assessments. Whie a arge initia impact of this change was evident in 1986, it coud be that the impact of this change was aso fet more recenty as State and oca governments, advocates, and awyers earned how to best take advantage of the changes. is We use the term discount here in a technica sense meaning that the weight given to observations refecting recent deviations from onger-term trends shoud be reativey sma in creating projections of future trends. We do not propose a specific methodoogy to estabish such weights in this artice. Nor do we impy that such deviations are unimportant in a quaitative sense. Indeed, we beieve that short-term deviations from onger-term trends are important for severa additiona reasons-such as in budgeting administrative expenditures, making management decisions concerning the aocation of resources for the processing of appications, and in affecting future caseoads due to the ong expected duration on the ros among persons drawn into SSA s disabiity programs by factors affecting short-term deviations from onger-term trends, such as the business cyce. i6 The DI estimates represent the experience of a cohort of 1972 new DI awardees, whie the SSI estimates are based on a pooed sampe of 1974 through 1982 new SSI awardees. Despite these differences in 60 Socia Security Buetin Vo. 58, No. 4 Winter 1995

19 sampeframes, we beieve that the DI and SSI figures are roughy comparabe. DI incidence rates refect DI awards reative to the size of the DI insured popuation. The SSI incidence rates represent SSI nonconcurrent awards reative to the popuation financiay eigibe for SSI (Rupp and Scott (1995b)) I* Since the tightening of initia determinations (increased denia rates) might coincide with attempts to take peope off the ros, this potentia confounding needs to be taken into account in assessing the vaidity of previous estimates of the effect of denia rates on appications. References Ader, M Disabiity Among Women on AFDC: An Issue Revisited. Paper presented at the Proceedings of the American Statistica Association, Washington, DC: prepared by the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services. Barre, A. and P. Taubman Some Economic and Demographic Conse- quences of Menta Iness. Journa of Lubor Economics, Vo. 4, No. 2, pp Bordeon. B Comments on Lewin- VHI Case Studies On Appication and Award Growth. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (Juy.) Bound, J., S. Kossoudji, and G. Ricart-Moes The Ending of Genera Assistance and SSI Disabiity Growth in Michigan: A Case Study. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (Juy.) Bound, J The Heath and Earnings of Rejected Disabiity Insurance Appicants. American Economic Review, Vo. 79, No. 3, pp Chirikos, T.N The Composition of Disabiity Beneficiary Popuations: Trends and Poicy Impications. Fina report prepared for the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (December.) Day, M.C Characteristics of SSI and DI Recipients in the Years Prior to Receiving Benefits: Evidence from the PSID. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (Juy.) Gruber, J. and J.D. Kubik Disabiity Insurance Rejection Rates and the Labor Suppy of Oder Workers. Cambridge, MA: Nationa Bureau of Economic Research. Working Paper No Gruber, J. and B.C. Madrian Heath Insurance Avaiabiity and the Retirement Decision. Cambridge, MA: Nationa Bureau of Economic Research. Working Paper No Hapern, J.D The Socia Security Disabiity Program: Reasons for Its Growth and Prospects for the Future. New Engand Economic Review, May1 June, pp Hambor, J.C Unempoyment and Disabiity: An Econometric Anaysis with Time Series Data. Washington, DC: Socia Security Administration, Office of Research and Statistics. Staff Paper No The Roe of Economic Factors in the Decine of the DI Trust Fund. U.S. Department of Treasury, unpubished paper. Hannsgen, G. and S.H. Sande Deeming Rues and the Increase in the Number of Chidren with Disabiities Receiving SSI: Evauating the Effects of a Reguatory Change. Washington, DC: Socia Security Administration, unpubished paper. Hennessey, J.C. and J. Dykacz Projected Outcomes and Length of Time in the Disabiity Insurance Program. Socia Security Buetin, Vo. 52, No. 9 (September), pp A Comparison of the Recovery Termination Rates of Disabed- Worker Beneficiaries Entited in 1975 and Socia Security Buetin, Vo. 56, No. 2 (Summer), pp Hoynes, H.W. and R. Moffitt The Effectiveness of Work Incentives in DI and SSI: Lessons fom Other Transfer Programs. University of Caifornia, Berkeey. Paper presented at the Nationa Academy of Socia Insurance and the Nationa Institute for Disabiity and Rehabiitation Research Conference on Disabiity, Work, and Cash Benefits. Lahiri, K., D.R.Vaughan, and B. Wixon Modeing SSA s Sequentia Disabiity Determination Process Using Matched SIPP Data, Socia Security Buetin, Vo. 58, No. 4 (Winter), pp Lando, M The Effect of Unempoyment on Appication for Disabiity Insurance. Paper presented at the annua meeting of the American Statistica Association (St. Louis, MO). Washington, DC: American Statistica Association, pp Prevaence of Work Disabiity by State, Socia Securitv Buetin, Vo. 42, No. 5 (May), pp Lando, M., M. Coate, and R. Kraus Disabiity Benefit Appications and the Economy. Socia Securit?, Buetin. Vo. 42, No. 5 (May). pp Leonard, J.S The Socia Security Disabiity Program and Labor Force Participation. Washington, DC: Nationa Bureau of Economic Research. Working Paper No (August.) Levy, J.M., and A. Krute The Impact of the Loca Economy on the Disabiity Process: Further Resuts. Washington, DC: Socia Security Administration. Lewin-VHI Labor Market Conditions, Socioeconomic Factors and the Growth of Appications and Awards for SSDI & SSI Disabiity Benefits: Background and Preiminary Findings. Washington, DC: The Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services and the Socia Security Administration 1995a. Labor Market Conditions, Socioeconomic Factors and the Growth of Appications and Awards for SSDI & SSI Disabiity Benefits. Fina Report. Washington, DC: The Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services and the Socia Security Administration. Socia Security Buetin Vo. 58, No. 4 Winter

20 L 1995b. Impairment Trends in the Growth of Appications and Awards for SSA Disabiity Benefits. Fina Report. Washington, DC: The Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services and the Socia Security Administration ~. Case Studies of State- Leve Factors Contributing to DI and SSI Disabiity Appication and Award Growth. Fina Report. Washington, DC: The Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services and the Socia Security Administration, d. Longer Term Factors Af- fecting SSDI and SSI Disabiity Appica- tions and Awards. Fina Report. Wash- ington, DC: The Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services and the Socia Security Administration. Loprest, P., K. Rupp, and S.H. Sande Gender, Disabiities, and Empoyment. Journa ofhuman Resources, Vo. 30 (Suppement), pp. S293-S3 18. Moffitt, R. and B. Wofe The Effect of the Medicaid Program on Wefare Participation and Labor Suppy. The Review ofeconomics and Statistics, Vo. 124 (Apri), pp Muer, L.S The Impact of Loca Labor Market Characteristics on the Disabiity Process. Washington, DC: Socia Security Administration. Office of Research and Statistics, Working Paper No. 27. (Apri.) Muer, L.S. and P. Wheeer Disabiity Program Growth: Resuts from Socia Security s Survey of Fied Office Managers, presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (Juy.) Parsons, D.O The Decine of Mae Labor Force Participation. Journa of Poitica Economy, Vo. 88, No. 1, pp I99 1. Measuring and Deciding Disabiity. In C. Weaver (ed.), Disabiity and Work. Washington, DC: AEI Press. Rupp, K., S.H. Be, and L. McManus Design of the Project Network Returnto-Work Experiment for Persons with Disabiities, Socia Security Buetin, Vo. 57, No. 2 (Summer), pp Rupp, K. and C.G. Scott. 1995a. Length of Stay in the SSI Disabiity Program, Socia Security Buetin, Vo. 58, No. 1 (Spring), pp b. Determinants of Duration on the Disabiity Ros and Program Trends. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy. sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services. Washington, DC. (Juy.) Stapeton, D., K. Coeman, and K. Dietrich Demographic and Economic Determinants of Recent Appication and Award Growth for SSA s Disabiity Programs. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (Juy.) The Effects of the Business Cyce on Disabiity Appications and Awards. Paper presented at the 1995 annua conference of the Society of Government Economists, Aied Socia Sciences Associations meetings, Washington, DC. (January.) Stapeton, D. and G. Livermore Impairment Trends in Appications and Awards for SSA s Disabiity Programs. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services. Washington, DC. (Juy.) Stapeton, D., G. Livermore, and A. Zeuschner Lessons from Case Studies of Recent Program Growth in Five States. Paper presented at the conference. The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services. Washington, DC. (Juy.) Stapeton, D. and K. Dietrich Long- Term Trends and Cyces in Appication and Award Growth. Paper presented at the conference, The Socia Security Administration s Disabiity Programs: Expanations of Recent Growth and Impications for Disabiity Poicy, sponsored by the Socia Security Administration and the Office of the Assistant Secretary for Panning and Evauation, U.S. Department of Heath and Human Services, Washington, DC. (Juy.) Thompson, L.H. and P.N. Van de Water The Short Run Behavior of the Socia Security Trust Funds. Washington, DC: U.S. Department of Heath, Education, and Wefare. Technica Anaysis Paper No. 8. Tobin, J Poverty in Reation to Macroeconomic Trends, Cyces, and Poicies. In S.H. Danziger, G.D. Sandefur, and D.H. Weinberg (eds.), Confronting Poverty: Prescriptions for Change. Cambridge, MA: Harvard University Press, pp U.S. Congressiona Budget Office update. Projections of Nationa Heath Expenditures. Congressiona Budget Office Memorandum. Washington, DC. U.S. Department of Heath and Human Services The Socia Security Disabiity Insurance Program: An Anaysis. Report of the Department of Heath and Human Services pursuant to a request from the Board of Trustees of the Federa Od-Age and Survivors Insurance and Disabiity Insurance Trust Funds, Washington, DC. (December.) State Laws. Washing- U.S. Department of Labor. Workers Compensation ton, DC. U.S. Genera Accounting Office Socia Security: Rapid Rise in Chidren on SSI Disabiity Ros Foows New Reguations. Washington, DC: U.S. Genera Accounting Office. (GAOHEHS ) Suppementa Security Income: Disabiity Program Vunerabe to Appicant Fraud When Middemen are Used. Washington, DC: U.S. Genera Accounting Office. (GAOIHEHS-95-I 16.) 62 Socia Security Buetin Vo. 58, No. 4 Winter 1995