Skip to main content

Advertisement

SpringerLink
Log in

Productive efficiency in processing social security disability claims: a look back at the 1989–95 surge

  • Published:
Empirical Economics Aims and scope Submit manuscript

“Social Security will always be a goal, never a finished thing,

because human aspirations are infinitely expandable—just

as human nature is infinitely perfectible”

—Arthur J. Altmeyer, the first SSA Commissioner (1945).

Abstract

Using panel data from a relatively volatile time period 1988–95, we have identified factors that account for over 80% of the longitudinal variation in the processing time of disability applications. Pending claims, workloads, percentage of SSI applications (children and adults), and the proportion of cases considered at different stages of the disability determination process explain a significant part of the variation. We found strong evidence that observed gains in organizational productivity were attained at the cost of timeliness in case dispositions. The dynamic panel data model estimated in this paper is used to compare the productive efficiency of different disability offices in a general econometric framework in which claim forecasts, staff allocations, and the number of adjudications are treated as endogenous. Our analysis suggests that there are persistent differences in the average processing time between states that can be attributed to organizational inefficiency. The importance of good forecasts of disability applications at sub-national levels is emphasized.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9

Similar content being viewed by others

Notes

  1. See National Academy of Social Insurance (1994), Lahiri et al. (1995a, b), Autor and Duggan (2006), French and Song (2014), and Leibman (2015) for commentaries on the history of these programs.

  2. See, for instance, GAO (1995, 1998, 2010, 2013), Social Security Advisory Board (1998, 2001), Green et al. (2005/2006), von Wachter et al. (2011), French and Song (2014), and Autor et al. (2015).

  3. The number of initial disability claims has continued to soar and has recently exceeded 5,000,000, with the waiting time for initial claim dispositions and hearings hovering around 120 and 515 days, respectively, cf. SSA (2020).

  4. See U.S. Department of Health and Human Services (1992).

  5. Our analytical sample was generated by merging a number of administrative data bases that are internal to the agency, and thus are not freely available to the public. The research was completed when the authors were affiliated with the Office of Policy, Division of Economic Research of SSA.

  6. “Applications,” “receipts,” and “claims” are used interchangeably in this paper.

  7. Waiting time is defined as the mean overall duration from the time of application to the date of initial decision, which includes time in queue and the actual times to process the cases by disability examiners. The SSA established guideline for the waiting time for this period was 60 days. In 1997, the percent of initial DI claims processed within 60 days of filing was 52.4 days. In recent years, waiting time has soared. It should be pointed out that there is a mandatory 5-month waiting period during which a prospective DI applicant cannot engage in any substantial gainful activity (SGA). Many authors including Halpern and Hausman (1986), Kreider (1999), Gruber and Kubik (1997), Rupp and Stapleton (1998), Parsons (1996) and Autor et al. (2015) have discussed the various implicit costs of waiting for a DI applicant.

  8. The increase in processing time during 1997–98 can be attributed to additional workloads due to the review of childhood, non-citizens, and drug addict and alcoholic (DA&A) cases and increased continuing disability reviews (CDRs) as mandated by the Congress in 1996. See SSA (1996, 1997).

  9. See OIG (2013). The number of field offices stayed almost the same till 2000. There have been 118 closings since then with approximately half of those after 2009. Despande and Li (2019) have used this information to study the importance of waiting time on self-screening for disability applications.

  10. To deal with sudden unforeseen changes in disability claims, SSA has a limited contingency budget at its disposal.

  11. Under the existing administrative arrangement, SSA’s Office of the Chief Actuary prepares the forecasts for disability applications and the Division of Budget of the Office of Finance, Assessment, and Management prepares the budgets and allocates FTEs for the FOs as well as the DDSs. The Office of Disability is the primary liaison between SSA and the state DDSs, so that it submits budget proposals for DDS operations to the Division of Budget based on inputs from the DDSs. For an analysis of forecasts generated by SSA’s Office of the Chief Actuary, see Kashin et al. (2015).

  12.  %FCST_ERR is defined as 100 × (actual receipts at the end of the year − projected receipts made at the beginning of year)/projected receipts made at the beginning of the year.

  13. It is not clear how the Office of Chief Actuary generates the yearly projections. Monthly workload data on disability claims for both DI and SSI for each state and the 10 SSA regions separately are publicly available (see https://www.ssa.gov/disability/data/ssa-sa-mowl.htm#DataSetDescription). These data can be used to construct optimal time series forecasts of yearly claims growth for each state or region using mixed-frequency methods with spatial dependence.

  14. The total number of DDSs is 54—one for each state, the District of Columbia, Puerto Rico, and Guam. South Carolina also has a separate agency for the blind. Our analysis includes 51 DDSs, one for each state and the District of Columbia.

  15. The number and type of field offices in each of the states are given in Table 5.

  16. WAGERATE is obtained after adjustment for local cost of living indices (1990 national average = 100). Even after adjusting for the regional cost of living, the average 1996 DDS staff compensation was $25,975 in NM compared to $67,627 in WA.

  17. Note that in addition to having a time-invariant unobserved νi, under certain additional assumptions, the time effects can be modeled more generally as λiηt, which allows each DDSi to absorb the common shock ηt differentially. See Artūras and Sarafidis (2018) and Sarafidis and Wansbeek (2020) for the scope of such dynamic panel data models with multifactor error structure.

  18. Lahiri (1993) and Kinal and Lahiri (1994) discuss estimation of simultaneous equations panel data models with lagged dependent and predetermined variables.

  19. From the standpoint of the cross-sectional component of the data, first differencing also reduces the problem of heteroskedasticity in errors due to size differences. We should, however, point out that our basic unit of observation is FO (not DDS), where the problem of size heteroskedasticity will be less problematic.

  20. We used the following instruments: FO_TYPE dummies; state dummies; second lags of logs of PT, FTE (at the DDSs), DO_FTE, and their squares; second lag of ln DISP and its square; second lag of PROJ and its square; second lag of FCST_ERR; current, first and second lagged values of (log of) total receipts at DDSs, and at FOs, and their squares; second lag of ln WKLOAD; all case-mix variables and their first differences;  %POOR,  %BLACK,  %HISP and  %UNEMPLOY, and their first differences; first difference of year dummies; all other exogenous variables in the equation and their lagged values. The validity of the over-identifying restrictions was accepted using the Sargan χ2 test at the 5% level of significance, cf. Baltagi (2013, p. 159).

  21. All subsequent analysis will be based on implied long-run coefficients calculated in a similar manner. The estimated coefficient 0.371 of ln PTt-1 suggests that the speed of adjustment is fairly fast—half of the adjustment takes place in approximately 7 months (i.e., median lag = ln 0.5/ln 0.371).

  22. We should emphasize that there is no definitional relationship between PPWY and PT.

  23. During several recent years, PPWY has been in excess of 300, cf. SSA (2020). Admittedly, steady advances in electronic case processing at FO and DDS offices and streamlining of the adjudication process might have increased the optimal 258 PPWY value we found based on 1989–95 data.

  24. The great disparity in PPWY between states continued over time. Thus, in 2000 Mississippi’s PPWY was 360, whereas the same for Michigan was 190. See Social Security Advisory Board (1998, 2001).

  25. See Lahiri et al. (1995a, b) and Hu et al. (2001) for details.

  26. They were: percent meeting or equaling the listings (%LISTING), percent not meeting the duration requirement (%DURATION), percent found able to do past work (%PASTWORK), percent found able to do any other work (%OTHERWORK), and percent of procedural denials (%PROCDEN).

  27. On the average, SSI adult applications face two weeks or more of additional adjudication time at the DDS level. See GAO (1994).

  28. Hu et al. (2001) found mixed evidence on the effect of workload pressure on denials. As a robustness check we also estimated (3) with  %DENIAL as another endogenous variable by pulling  %DENIALit and  %DENIALit-1 from the list of instruments and replacing them with  %DENIALit-2. The estimates were largely unchanged.

  29. Unfortunately, we could not locate data on the average level of education for each DDS over our sample.

  30. The attrition rate is normally around 10%; however, for some many states it can be as high as 35% per year. By simple regression analysis, we found that workload and the DDS wage rate (lagged one period) significantly affect attrition rate with expected positive and negative signs, respectively (R-Sq. = 0.13).

  31. Certainly, SSA is aware of the fact that disability claims in poorer areas take longer to adjudicate. In a re-classification (effective January 1997) of field offices, service area characteristics (e.g., percentage of people in poverty) have become “elements” for classification. See also OIG (2013).

  32. One possible explanation is that most of the case-mix variables change slowly over time. As a result, even though they are not strictly time-invariant, they do not contain much information in their first-difference form.

  33. FOs were classified as District Office (DO) Class I, DO Class II&III, Branch Office (BO) and Resident Station (RS). Since there were only seven RSs in our sample, our FO_TYPE2 dummy includes BOs and RSs. Both BOs and RSs offer a full range of services, but are subsidiaries of DOs. The DO Class I is the fundamental working unit of SSA’s organization, and is the center of operations, management and information for the DO service area. BOs and RSs tend to be smaller than their parent DOs in terms of workload, staff, and other areas of responsibility. See OIG (2013).

  34. In order to facilitate prompt case processing, some states (viz., AL, AZ, CA, FL, GA, KY, LA, MA, MI, MO, NJ, NY, PA, SC, VA, WA, and WV) have more than one DDS offices in the state. We introduced a dummy variable representing these decentralized states in this regression to determine if these states have lower processing time. However, the variable was not statistically significant at the 10% level of significance.

  35. Kumbhakar and Hjalmarsson (1995) have used a similar methodology to study the relative efficiency of Swedish social insurance field offices. Kumbhakar and Lovell (2003) and Greene (1997) present comprehensive surveys of the literature. See also Das and Kumbhakar (2012).

  36. Technically, the unobserved state effects can be correlated with ln PENDING, ln PPWY and (ln PPWY) 2. Our estimation strategy, laid out in Sect. 3, ensures consistent estimation of the effects of the last three variables on ln PT despite this correlation (cf. Baltagi 2013). The idea of keeping a part of the explained variation in ln PT as a component of unobserved productive efficiency can be found in Lovell (1993) and Kumbhakar and Lovell (2003).

  37. cf. Kumbhakar and Hjalmarsson, op.cit.

  38. In the last column of Table 3, we have reported the accuracy rate of all DDS decisions, averaged over 1988–96. Based on internal quality review, accuracy rate is the percentage of policy compliant initial disability determinations at a DDS and is calculated from the net error rate (i.e., the number of corrected deficient cases with changed disability decisions, plus the number of deficient cases not corrected within 90 days from the end of the period covered by the report, divided by the number of cases reviewed).

  39. In its several revisions to the 1994 Disability Process Redesign, SSA has continued to emphasize the need to develop a more comprehensive QA system that would promote uniform and consistent disability decisions across all geographic and adjudicative levels based on a broader concept of quality management. See GAO (2002).

References

  • Artūras J, Sarafidis V (2018) Fixed T dynamic panel data estimators with multifactor errors. Econom Rev 37:893–929

    Article  Google Scholar 

  • Autor DH, Duggan MG (2006) The growth in the social security disability rolls: a fiscal crisis unfolding. J Econ Perspect 20(3):71–96

    Article  Google Scholar 

  • Autor DH, Maestas N, Mullen KJ, Strand A (2015) Does delay cause decay? the effect of administrative decision time on the labor force participation and earnings of disability applicants. NBER Working Paper No. 20840

  • Baltagi BH (2013) Econometric analysis of panel data, 5th edn. Wiley, Chichester

    Google Scholar 

  • Benitez-Silva HM, Buchinsky HM, Rust CJ, Sheidvasser S (1999) Social security disability award process: how large are the classification errors?” national bureau of economic research working paper series w10219, NBER

  • Congressional Research Service Report to the Congress (1992) Status of the disability programs of the social security administration. 92–691 EPW. Washington, D.C.: The Library of Congress

  • Das A, Kumbhakar SC (2012) Productivity and efficiency dynamics in Indian Banking: an input distance function approach incorporating quality of inputs and outputs. J Appl Econom 27:205–234

    Article  Google Scholar 

  • Deshpande M, Li Y (2019) Who Is screened out? application costs and the targeting of disability programs. Am Econ J Econ Policy 11(4):213–248

    Article  Google Scholar 

  • French E, Song J (2014) The effect of disability insurance receipt on labor supply. Am Econ J Econ Policy 6(2):291–337

    Article  Google Scholar 

  • Green M, Eigen B, Lefko J, Ebling S (2005/2006) Addressing the challenges facing SSA’s disability programs. Soc Secur Bull, 66 (3): 29–39

  • Greene WH (1997) Frontier production functions. In: Pesaran MH, Schmidt P (eds) Handbook of applied econometrics. Blackwell Publishers, Malden, pp 81–166

    Google Scholar 

  • Gruber J, Kubik JD (1997) Disability rejection rates and the labor supply of older workers. J Public Econ 31:1314–1361

    Google Scholar 

  • Halpern J, Hausman JA (1986) Choice under uncertainty: a model of applications for the social security disability insurance program. J Public Econ 31:131–161

    Article  Google Scholar 

  • Hu J, Lahiri K, Vaughan DR, Wixon B (2001) A structural model of social security’s disability determination process. Rev Econ Stat 83(May):348–361

    Article  Google Scholar 

  • Kashin K, King G, Soneji S (2015) Systematic bias and nontransparency in US social security administration forecasts. J Econ Perspect 29(2):239–258

    Article  Google Scholar 

  • Kinal T, Lahiri K (1994) A simplified algorithm for estimation of simultaneous equations error components models with an application to a model of developing country foreign trade. J Appl Econom 8:81–92

    Article  Google Scholar 

  • Kreider B (1999) Social security disability insurance: applications, awards, and lifetime income flows. J Labor Econ 17:784–827

    Article  Google Scholar 

  • Kumbhakar SC, Hjalmarsson L (1995) Labour-use efficiency in Swedish social insurance offices. J Appl Econom 10:33–47

    Article  Google Scholar 

  • Kumbhakar SC, Hjalmarsson L (1998) Relative performance of public and private ownership under yardstick competition: electricity retail distribution. Eur Econ Rev 42:97–122

    Article  Google Scholar 

  • Kumbhakar SC, Knox Lovell CA (2003) Stochastic frontier analysis. Cambridge University Press, Cambridge

    Google Scholar 

  • Lahiri K (1993) Panel data models with rational expectations. In: Maddala GS, Rao CR, Vinod HD (eds) Handbook of statistics, Econometrics, vol 11. North Holland, Amsterdam, pp 721–737

    Google Scholar 

  • Lahiri K, Song J, Wixon B (1995a) A model of social security disability insurance using matched SIPP/Administrative data. J Econ 145(2008):4–20

    Google Scholar 

  • Lahiri K, Vaughan DR, Wixon B (1995b) Modeling social security’s sequential disability determination using matched SIPP data. Soc Secur Bull 58:3–42

    Google Scholar 

  • Liebman JB (2015) Understanding the increase in disability insurance benefit receipt in the united states. J Econ Perspect 29(2):123–150

    Article  Google Scholar 

  • Lovell CAK (1993) Production Frontiers and Productive Efficiency. In: Fried HO, Lovell CAK, Schmidt SS (eds) The measurement of productive efficiency. Oxford University Press, New York, pp 3–67

    Google Scholar 

  • Maestas N, Mullen KJ, Strand A (2015) Disability insurance and the great recession. Am Econ Rev 105(5):177–182

    Article  Google Scholar 

  • National Academy of Social Insurance (1994) Rethinking disability policy: the role of income, health care, rehabilitation and related services in fostering independence. Preliminary Status Report of the Disability Policy Panel, Washington

    Google Scholar 

  • Parsons DO (1996) Imperfect ‘Tagging’ in social insurance programs. J Public Econ 62:183–207

    Article  Google Scholar 

  • Rupp K, Stapleton D (eds) (1998) Growth in disability benefits: explanations and policy implications. W.E. Upjohn Institute for Employment Research, Kalamazoo

    Google Scholar 

  • Sarafidis V, Wansbeek T (2020) Celebrating 40 Years of panel data analysis: past, present and future. J Econom. https://doi.org/10.1016/j.jeconom.2020.06.001

    Article  Google Scholar 

  • Schmidt P, Sickles R (1984) Production Frontiers and Panel Data. Jf Bus Econ Stat 2:367–374

    Google Scholar 

  • Social Security Advisory Board 1998. How SSA’s disability programs can be improved. Washington, D.C

  • Social Security Advisory Board (2001) Charting the future of social security’s disability programs: the need for fundamental change. January. Washington, D.C

  • United States Department of Health and Human Services (1992) The social security disability insurance program: an analysis. report pursuant to a request from the board of trustees of the federal old-age and survivors insurance and disability insurance trust funds (referred to as the 709 Report). Washington, D.C

  • United States General Accounting Office (1993) Social security: increasing number of disability claims and deteriorating service. Report to the Chairman, Committee on Finance, U.S. Senate. GAO/HRD-94-11. Washington, D.C

  • United States General Accounting Office (1995) Social security administration: management action and program redesign needed to address long-standing problems. Testimony before the subcommittee on social security, committee on ways and means, house of representatives. GAO/T-HEHS-95-233. Washington, D.C

  • United States General Accounting Office (1997) Social security disability: SSA actions to reduce backlogs and achieve more consistent decisions deserve high priority.” Testimony before the subcommittee on social security, committee on ways and means, house of representatives. GAO/T-HEHS-97-118. Washington, D.C

  • United States General Accounting Office (1998) “SSA’s management challenges: strong leadership needed to turn plans into timely, meaningful action.” testimony before the subcommittee on human resources and the subcommittee on social security, committee on ways and means, house of representatives. GAO/T-HEHS-98-113. Washington, D.C

  • United States General Accounting Office (2002) Disappointing results from SSA’s efforts to improve the disability claims process warrant immediate attention. Report to the chairman, subcommittee on social security, committee on ways and means, house of representatives. GAO-02-322. Washington, D.C

  • United States General Accounting Office (2010) Social security disability: management of disability claims workload will require comprehensive planning, testimony before the sub-committee on social security, income security and family support, committee on ways and means, house of representatives, GAO-10-667T. Washington, D.C

  • United States General Accounting Office. 2013. Social security administration, long-term strategy needed to address key management challenges (GAO-13-459)

  • United States General Accounting Office (2018) SSA disability programs: better metrics and evaluation needed to inform decision making. Testimony before the sub-committee on social security, income security and family support, committee on ways and means, house of representatives, GAO-18-667T. Washington, D.C

  • United States Office of the Inspector General (OIG) (2013) Social Security Administration Audit Report, Field Office Performance, A-15-12-11290

  • United States Social Security Administration (1996). Report to the Congress on Rising Cost of Social Security Disability Insurance Benefits. February. Washington, D.C

  • United States Social Security Administration (1997) Social security: accountability report for fiscal year (FY) 1997. November. Washington, D.C

  • United States Social Security administration (2020) Annual performance report fiscal years 2019–2021. Washington, D.C

  • von Wachter T, Song J, Manchester J (2011) Trends in employment and earnings of allowed and rejected applicants to the social security disability insurance program. Am Econ Rev 101(7):3308–3329

    Article  Google Scholar 

  • Wunderlich GS, Rice DP (1998) The social security administration’s disability decision process: a framework for research. Second Interim Report. National Research Council, Institute of Medicine. Washington, D.C.: National Academy Press

Download references

Acknowledgements

We are grateful to Subal Kumbhakar, Vasilis Sarafidis, and two anonymous referees for many comments and suggestions on an earlier version of the paper. We, however, are responsible for any shortcomings remaining in the paper. The views expressed in this paper are the authors’ only and do not necessarily reflect the position of the Social Security Administration.

Author information

Authors and Affiliations

  1. Department of Economics, University at Albany: SUNY, Albany, NY, 12222, USA

    Kajal Lahiri & Jianting Hu

Authors
  1. Kajal Lahiri
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jianting Hu
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Kajal Lahiri.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest to disclose.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix

Appendix

See Tables 4 and 5.

Table 4 Overall DI processing time at the District Offices—state average (in days)
Full size table
Table 5 Type of field offices at DDS
Full size table

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Lahiri, K., Hu, J. Productive efficiency in processing social security disability claims: a look back at the 1989–95 surge. Empir Econ 60, 419–457 (2021). https://doi.org/10.1007/s00181-020-01943-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00181-020-01943-y

Keywords

JEL Classification

Search

Navigation