The dynamic relationship between disability onset, earnings, and disability insurance application and receipt

The dynamic relationship between disability onset, earnings, and disability insurance application and receipt

Economics Letters 124 (2014) 374–377 Contents lists available at ScienceDirect Economics Letters journal homepage: www.elsevier.com/locate/ecolet T...

605KB Sizes 0 Downloads 32 Views

Economics Letters 124 (2014) 374–377

Contents lists available at ScienceDirect

Economics Letters journal homepage: www.elsevier.com/locate/ecolet

The dynamic relationship between disability onset, earnings, and disability insurance application and receipt Perry Singleton ∗ Department of Economics, Center for Policy Research, Syracuse University, United States

highlights • Earnings decrease sharply at the time of disability onset. • Simultaneously, disability insurance application and receipt increase. • At disability onset, the effect of disability insurance on earnings is limited.

article

info

Article history: Received 20 March 2014 Received in revised form 13 June 2014 Accepted 16 June 2014 Available online 21 June 2014

abstract At the time of disability onset, the effect of disability insurance on earnings is limited by the finding that work-prevented respondents, who account for the majority of benefit claims, have negligible earnings regardless of application status. © 2014 Elsevier B.V. All rights reserved.

JEL classification: H5 I3 J2 Keywords: Health Labor supply Disability insurance

1. Introduction Several recent studies measure the effect of health on labor productivity using longitudinal methods (Charles, 2003; Meyer and Mok, 2013; Mok et al., 2008; Singleton, 2012; Stephens, 2001). The effect of health is measured as the change in earnings at the time of disability onset. These studies consistently find that earnings decrease sharply at the time of disability onset, suggesting that the effect of health on labor productivity could be substantial. Although the decline in earnings may be due to health, other factors may contribute to this decline. One possible factor is the decision to apply for disability insurance benefits. This is because the two largest disability insurance programs in the US – Social Security’s Disability Insurance (DI) and Supplement Security Income (SSI) – require an applicant’s earnings to be below the ‘‘substantial

∗ Correspondence to: 426 Eggers Hall, Syracuse University, Syracuse, NY 13202, United States. Tel.: +1 301 437 7176; fax: +1 315 443 1081. E-mail address: [email protected]. http://dx.doi.org/10.1016/j.econlet.2014.06.016 0165-1765/© 2014 Elsevier B.V. All rights reserved.

gainful activity’’ amount to qualify for benefits. In 2014, the substantial gainful amount was $1070 per month for non-blind applicants. Thus, recently disabled workers may decrease their earnings – beyond the effect of health itself – to qualify for benefits. The goal of this article is to examine the scope to which disability insurance, and its disincentives to labor supply, could account for the decrease in earnings at the time of disability onset. This is accomplished by examining the dynamic relationship between disability onset, earnings, and disability insurance application and receipt. This is a contribution to related studies, which do not examine disability insurance applications. As the analysis shows, application and receipt increase substantially at the time of disability onset. However, the effect of disability insurance on earnings is limited by two factors. First, application and receipt are relatively low among workers who experience a work-limiting disability. And second, workers who experience a work-preventing disability have negligible earnings at the time of disability onset, regardless of benefit application status. The precise effect of application on earnings – particularly at the time of disability onset – remains unidentified.

P. Singleton / Economics Letters 124 (2014) 374–377

375

Table 1 Data summary by disability severity: males aged 26–57. Non-disabled

Work limited

Work prevented

Reference age

39.3 (0.04)

41.4 (0.24)

44.2 (0.30)

High school

58.0 (0.23)

64.2 (1.3)

52.6 (1.7)

College

29.9 (0.21)

13.1 (0.90)

6.2 (0.82)

White

86.8 (0.16)

86.1 (0.92)

76.9 (1.4)

Married

70.1 (0.21)

67.2 (1.3)

62.6 (1.6)

SSDI insured

82.8 (0.18)

80.1 (1.1)

74.4 (1.5)

Disabled while employed

93.7 (0.65)

88.1 (1.1)

Accident

51.0 (1.3)

43.6 (1.7)

Work accident

36.5 (1.3)

31.9 (1.6)

Musculoskeletal

39.5 (1.3)

35.1 (1.6)

Cardiovascular

13.6 (0.92)

12.0 (1.1)

1400

863

Observations

45 652

Figures are computed from the Survey of Income and Program Participation, pooled over survey years 1990, 1991, 1992, 1993, and 1996. Work prevented is defined as a one-year transition from non-disabled to work-prevented; work limited is defined as a one-year transition from non-disabled to work-limited, but a subsequent transition to being work-prevented is possible. The data do not include individuals disabled more than five years before the survey date. Married refers to marital status two years before the reference year. SSDI insured reflects insured status during the reference year and, due to lack of data, is only calculated for those whose reference year occurs on or after 1988.

2. Material and methods 2.1. Data The data come from the Survey of Income and Program Participation, a representative survey of the non-institutionalized US population. The sample is derived by pooling topical module two – which contains retrospective information on disability onset – across survey panels 1990, 1991, 1992, 1993, and 1996. Respondents first report whether they have a disability that limits work and, if so, the year and month of disability onset. Work-limited respondents then report whether their disability prevents work and, if so, the month and year of onset. The survey data are merged to administrative data on earnings and disability insurance application and receipt. The data on earnings come from the Detailed Earnings Record, which contains annual earnings from 1978 to 2004 derived from tax returns and W-2 filings. The data on disability insurance application and receipt come from three sources: the 831 file (DI and SSI application), the Supplemental Security Record (SSI receipt), and the Master Beneficiary Record (DI receipt). These data are maintained by the Social Security Administration. 2.2. Sample and summary The analysis sample is derived after imposing several data restrictions. One, the sample is limited to males. Two, disabled respondents are dropped if the date of disability onset is missing, if the work-prevention date precedes the work-limitation date, and if disability onset occurs more than five years before the survey.1

1 This restriction, which eliminates 51.9% of disabled respondents, ensures that earnings data are available prior to disability onset.

Four, the sample is limited to SIPP respondents that merge successfully with administrative data.2 And five, the sample is restricted to respondents whose age during the reference year is 26–57. The reference year for the disabled is the year of disability onset; the reference year for the non-disabled is the year of the survey. The final sample contains 45,652 non-disabled respondents and 2263 recently disabled respondents. Summary statistics of the sample by disability status are presented in Table 1. As shown, disabled respondents are older and less likely to be educated, married, or insured for DI benefits. Most disabilities occur while the respondent is employed. Close to half of disabilities are accidental, one-third are accidents related to work, and one-third are disabilities due to musculoskeletal conditions. 3. Results 3.1. Disability onset and earnings Fig. 1 illustrates the dynamic relationship between disability and earnings. Average annual earnings are plotted relative to the reference year.3 The figure reveals three notable patterns. First, earnings are lower among the disabled groups prior to disability onset. Second, earnings decrease sharply at the time of disability onset. Third, earnings recover slightly among the work prevented, but not the work limited. Similar to related studies, the decrease in earnings is estimated precisely using an event-study model. The model is given by,

2 Conditional on the preceding sample restrictions, the match rate of SIPP data to administrative data is 88.9%. 3 For the longitudinal analysis, the sample in each calendar year is conditioned on being alive at end of the year. This is determined by linking the data to the Numident file, which contains information on the date of death.

376

P. Singleton / Economics Letters 124 (2014) 374–377

Fig. 1. Earnings by disability onset status: males aged 26–57.

yit = αi + γt + β xit +

 s

δks Askit + εit .4

(1)

k

The unit of analysis is individual i and calendar year t. The model includes individual fixed-effects αi , calendar year fixed effects γt , and a vector of time-varying control variables xit . The vector of controls consists of age, its square, and the interaction of age and its square with education (indicators of high school diploma and college degree), race (an indicator of being white), and marital status (an indicator of being married). The dynamic change in earnings is measured by the disability indicators Askit . The superscript s denotes the two disability groups. The subscript k denotes the years relative to disability onset, from minus three to positive seven (the ‘‘left-out’’ years include period negative seven to negative four). Thus, the coefficient δks measures the average, within-person change in earnings for disability group s in period k. The estimation results confirm that disability onset is associated with a sharp decrease in earnings. By reference year two, earnings are $8.14 thousand (24.5%) lower among work-limited respondents and $19.9 thousand (80.2%) lower among work-prevented respondents. These estimates are comparable to Stephens (2001) and Meyer and Mok (2013) using the Panel Study of Income Dynamics.

Fig. 2. Disability application and receipt by disability onset status: males aged 26–57.

Fig. 2 illustrates the dynamic relationship between disability onset and disability insurance application and receipt. Disability insurance refers to both the DI and SSI programs. Panel A shows the percent of respondents who initiate an application in a given calendar year, and panel B shows the percent of respondents who receive disability benefits in a given year.5 As shown, both application and receipt increase at the time of disability onset, and the increases are greater among work-prevented respondents. Among work-prevented respondents, the rate of receipt plateaus at 65% four years after disability onset.6 These figures suggest that disability insurance may play an important role in explaining the changes in earnings at the time of disability onset.

ple is separated into three applicant groups: accepted applicants, rejected applicants, and non-applicants.7 Fig. 3 plots earnings by disability insurance status and by disability severity. Panel A corresponds to respondents who report a work-limiting disability. As shown, earnings at the time of disability onset are substantially lower among applicants relative to non-applicants. If the earnings of non-applicants serve as an upper bound to earnings of applicants had they not applied, then disability insurance could explain a significant proportion of the decline in earnings among applicants.8 However, the overall effect of disability insurance is limited by the fact that only 17.1% of work-limited respondents apply for benefits and only 10.9% are accepted. Panel B corresponds to respondents who report a workpreventing disability. Among this group, 63.1% apply for benefits and 55.1% are accepted. As shown, all three groups have negligible earnings around the time of disability onset. In reference year one, earnings among accepted applicants, rejected applicants, and nonapplicants were $2.7, $3.8, and $3.2 thousand, respectively. If the earnings of non-applicants serve as an upper bound to earnings of applicants, then the figure suggests that the effect of disability insurance on earnings is minimal.

3.3. Disability onset and earnings by application and receipt status

4. Discussion and conclusion

Next, the analysis considers the dynamic relationship between disability onset and earnings by disability insurance status. The goal is to determine whether the change in earnings among the disabled depends on the application decision and, furthermore, on the outcome of the application. For this purpose, the disabled sam-

According to the analysis, disability onset is associated with a substantial increase in disability insurance application and receipt.

3.2. Disability onset and disability application and receipt

4 The model is similar to the event-study method used by Jacobson et al. (1993). Consistent with Meyer and Mok (2013), earnings are in levels, and the model does not contain individual-specific time trends. 5 The percent only reflects initial applications, not appeals. 6 Meyer and Mok (2013) find that 50% of respondents receive DI or SSI benefits six to ten years after a ‘‘chronically-severe’’ disability.

7 To define these groups precisely, the sample is restricted to respondents who were not entitled to disability insurance benefits on or before two years prior to the reference year. This restriction eliminates 1.1% of the sample, yielding 45,204 nondisabled respondents and 2263 recently disabled respondents. Accepted applicants are those who applied for and received benefits; rejected applicants are those who applied for but did not receive benefits; non-applicants are the remainder of disabled respondents. As such, accepted applicants include respondents who were denied benefits at the initial application, but were awarded benefits on appeal. 8 This assumption is an extension to Bound (1989), where earnings of rejected applicants serves as an upper bound to the earnings of accepted applicants.

P. Singleton / Economics Letters 124 (2014) 374–377

377

This conclusion contrasts with studies that find a meaningful effect of benefit receipt on earnings—most recently, Maestas et al. (2013). However, these studies differ from the current study in important ways. First, this study considers non-applicants, whereas most related studies examine only rejected and accepted applicants. Second, the reference period in this study is the year of disability onset, whereas in related studies it is the year of application. And third, this study examines the effect of disability insurance on earnings at the time of disability onset, whereas related studies examine the effect on earnings several years after disability insurance application. Identifying the effect of disability insurance application on earnings – separate from the effect of benefit receipt on earnings – is an important direction for future research. Acknowledgments This project is supported by an inter-personnel agreement between Syracuse University and the Social Security Administration. The author would like to thank Kimberly Burham-Sacher and David Pattison for helpful comments and suggestions. The author is responsible for all errors. References

Fig. 3. Earnings by disability application and receipt status and disability status: males aged 26–57.

This suggests that disability insurance, and its disincentives to labor, may account for the observed decrease in earnings at the time of disability onset. However, if the earnings of non-applicants serve as an upper bound to the earnings of applicants had they not applied, then the effect of disability insurance on earnings is limited.

Bound, J., 1989. The health and earnings of rejected disability insurance applicants. Amer. Econ. Rev. 79 (3), 482–503. Charles, K.K., 2003. The longitudinal structure of earnings losses among worklimited disabled workers. J. Hum. Resour. 38 (3), 618–646. Jacobson, L.S., LaLonde, R.J., Sullivan, D.G., 1993. Earnings losses of displaced workers. Amer. Econ. Rev. 83 (4), 685–709. Maestas, N., Mullen, K.J., Strand, A., 2013. Does disability insurance receipt discourage work? Using Examiner Assignment to estimate causal effects of SSDI receipt. Amer. Econ. Rev. 103 (5), 1797–1829. Meyer, B.D., Mok, W.K., 2013. Disability, earnings, income and consumption. The Harris School of Public Policy Studies Working Paper #18869. Mok, W.K., Meyer, B.D., Charles, K.K., Achen, A.C., 2008. A note on ‘The longitudinal structure of earnings losses among work-limited disabled workers’. J. Hum. Resour. 43 (3), 721–728. Singleton, P.D., 2012. Insult to injury: disability, earnings, and divorce. J. Hum. Resour. 47 (4), 972–990. Stephens, M., 2001. The long-run consumption effects of earnings shocks. Rev. Econ. Stat. 83 (1), 28–36.