Demonstrations to Promote Work Among Disability Beneficiaries Likely to Produce Limited Results
The October 2015 budget agreement extended the solvency of the Social Security Disability Insurance (DI) trust fund into 2022 and renewed the Social Security Administration’s (SSA’s) authority to conduct demonstration projects, allowing the agency to test ideas to encourage work among DI beneficiaries and applicants. This creates an opportunity to build further evidence on the efficacy of various ideas to encourage work among this population.
Few DI beneficiaries are able to work. Still fewer are likely to be able to return to self-supporting work on a sustained basis.New demonstrations will likely produce only limited results, however. SSA has conducted many work-incentive experiments over the past 25 years, and none has led to a significant number of beneficiaries earning enough to support themselves and leave DI. This result should not come as a surprise. DI’s eligibility criteria are strict. Few DI beneficiaries are able to work. Still fewer are likely to be able to return to self-supporting work on a sustained basis.
Moreover, the terms of the new authority constrain SSA’s ability to conduct demonstrations that will produce meaningful results. The budget deal keeps DI solvent for less than seven years, and we are unlikely to learn much during that relatively short timeline. It takes years to design, conduct, and evaluate a good experiment, especially for longer-term questions such as whether early intervention reduces DI participation. In addition, the budget deal specifies that beneficiaries may choose whether to participate in demonstrations and may drop out at any time, which undermines the experiments’ validity. Finally, the one demonstration that the law explicitly spells out, requiring SSA to test a new benefit offset, largely duplicates a project already underway but with some troubling new features.
Congress must replenish DI again by 2022. When it does, it will likely look to the results of SSA’s work demonstrations for evidence on which to base new policies. Though the renewed demonstration authority is flawed, SSA has the opportunity to learn something by testing new ideas. SSA is already designing one early intervention demonstration for younger workers with mental illnesses who are likely to receive DI. SSA could consider other early intervention ideas as well. Early medical interventions could help people recover from their health problems before they prevent substantial work. Early work interventions could keep people connected to jobs before their skills atrophy and their attachment to the job market weakens. SSA could also test transitional benefits, which would divert DI beneficiaries or applicants who are likely to recover or be able to work and provide them with medical, employment, and/or financial supports.
DI Beneficiaries Have Very Limited Work Capacity
By law, DI beneficiaries must have severe disabilities that limit their ability to work. Eligibility for DI is strict, and the rules are designed to weed out applicants who could support themselves through work. To qualify, applicants must have a severe, medically determinable impairment that has already lasted five months and is expected to last at least 12 months or result in death. Their impairment must make them unable to engage in substantial gainful activity at any work that exists in the national economy — regardless of whether such work is available where the applicant lives, whether a specific job vacancy exists, or whether he or she would be hired. SSA rejects a majority of applicants, and even rejected applicants fare poorly in the labor market afterward.
Research consistently finds that beneficiaries have limited work capacity and face significant barriers to work. They suffer from poor health and functional limitations, and their death rates are at least three times as high as the general population’s. Many also lack reliable transportation or face inflexible work arrangements, inaccessible workplaces, or employment discrimination.
Even if they can surmount these obstacles and return to work — which only a minority of DI beneficiaries are able to do — very few can earn enough to become self-sufficient. Of beneficiaries who were tracked during a ten-year period, roughly a quarter worked at some point after their DI application was approved, but generally episodically and at low earnings. Just 4 percent earned enough to work their way off the DI rolls, and about one-quarter of them subsequently returned to the program.
SSA Has Conducted Many Experiments to Encourage Work
SSA has carried out many experiments to encourage work over the past 25 years (see Table 1).  These pilots, studies, and demonstration projects have provided some DI beneficiaries with education and work supports, improved access to health care, and relaxed earnings limits. In some cases, they had positive effects: improved health outcomes and mental health status, greater rates of employment, and higher earnings. However, in none of the experiments did a significant number of beneficiaries earn enough to support themselves and to leave DI. This fact is no surprise, given DI’s strict criteria and beneficiaries’ limited work capacity.
|SSA Work-Incentive Experiments Have Shown Only Limited Results|
|Benefit Offset National Demonstration (BOND)||2009-2017||Testing a $1-for-$2 benefit offset for earnings above SGA level, with additional work supports for “Phase 2” beneficiaries||
|Accelerated Benefits Demonstration||2004-2011||Provided health care to DI beneficiaries during 24-month waiting period for Medicare, with additional medical and work supports for “AB Plus” beneficiaries||
|Mental Health Treatment Study||2003-2011||Provided medical and employment supports to beneficiaries with schizophrenia or affective disorders||
|Benefit Offset: Four-State Pilot||2003-2009||Replaced “cash cliff” with a $1-for-$2 offset for earnings above SGA level, with additional work supports||
|Youth Transition Demonstration||2001-2014||Waived SSI income and asset rules, provided state-designed employment and education supports for young DI and SSI beneficiaries||
|Ticket to Work||1999-present||Provides vocational rehabilitation and work support from employment networks (Ticket to Work is a change in law, not a demonstration)||
|State Partnership Initiative||1998-2006||Tested variety of state-designed interventions, including Medicaid waivers and employment services for DI and SSI beneficiaries||
|Project NetWork||1991-1999||Offered intensive outreach, work-incentive waivers, and case management services to DI and SSI applicants and recipients||
Notes: SSA=Social Security Administration; DI=Social Security Disability Insurance, SSI=Supplemental Security Income, SGA=Substantial Gainful Activity (varies by year, currently defined as earnings that exceed $1,130 a month in 2016).
Sources: CBPP analysis based on Scott D. Szymendera and William R. Morton, “Social Security Disability Insurance (SSDI) Demonstration Projects,” CRS Reports RL33585, April 10, 2014 and RL33585, July 20, 2011; Gina Livermore et al., “Finding Alternatives to Disability Benefit Receipt,” IZA Journal of Labor Policy, 3:14 (2014); Livermore et al., “Executive Summary of the Seventh Ticket to Work Evaluation Report,” Final Report, July 2013; Government Accountability Office, “Modernizing SSA Disability Programs: Progress Made, but Key Efforts Warrant More Management Focus,” GAO-12-420, July 2012; Robert Kornfeld and Kalman Rupp, “The Net Effects of the Project NetWork Return-to-Work Case Management Experiment on Participant Earnings, Benefit Receipt, and Other Outcomes,” Social Security Bulletin, 63:1 (2000).
Budget Agreement Allows More Demonstrations to Encourage Work
The October 2015 budget agreement extended Social Security DI solvency into 2022 by temporarily reallocating a portion of payroll taxes, thereby averting a nearly 20 percent cut in DI benefits. In addition, the law renews SSA’s authority to conduct demonstration projects, allowing the agency to test ideas to encourage work among DI beneficiaries and some applicants.
The renewal of SSA’s demonstration authority creates an opportunity to build further evidence on whether ideas to encourage work would help — or hurt — a very vulnerable population, and at what cost. These projects could inform future changes to DI. However, Congress reallocated enough funds to keep DI solvent for less than seven years; this restricted timeline will limit the scope of the demonstrations SSA undertakes and make it more difficult to provide meaningful results.
It takes years to design, conduct, and evaluate a good experiment. To use its new authority well, SSA must do most or all of the following:
- Design a demonstration;
- Gather input from stakeholders and experts — and, in some cases, from the public;
- Organize a review of the research plan by an expert panel;
- Conduct a pilot before scaling up the demonstration;
- Adjust the research design in light of the pilot’s results;
- Choose, through competitive bidding, a contractor to conduct and evaluate the experiment;
- Select sites for the experiment;
- Encourage beneficiaries or applicants to participate, and randomly assign them to treatment and control groups;
- Train people to provide the intervention;
- Conduct the experiment, which must run long enough to reach a meaningful number of subjects and observe key outcomes;
- Evaluate the experiment; and
- Complete a follow-up, especially if the demonstration is designed to change long-term outcomes such as delaying or forgoing DI receipt.
Given all that is involved in running a demonstration, it’s not surprising that most of SSA’s DI work demonstrations in the past 25 years have taken at least seven years from design to evaluation, as Figure 1 shows. Some experiments necessarily take longer. For example, testing whether early intervention would cause disabled workers to delay or forgo DI benefits would take years after the intervention to know whether it worked, and at what cost.
The other factor that limits the usefulness of SSA’s new demonstration authority is that participation in any demonstration must be voluntary. Beneficiaries may choose whether to participate and then may drop out of studies at any time — including the moment the provisions of the demonstration might reduce their benefits or terminate their eligibility. This self-selection undermines the validity of the demonstrations.
The gold standard for experiments is random assignment. If subjects are selected at random, those who get the intervention won’t differ systemically from those who don’t, so the demonstration can zero in on the effect of the intervention itself. However, beneficiaries who volunteer for a work demonstration likely will differ in important ways from other beneficiaries. Volunteers are more likely than typical disability beneficiaries to be able to work. In addition, few volunteers will stick with a demonstration if their benefits are reduced and/or terminated. Finally, not enough people may volunteer to stage a demonstration at all.
In any event, the results of a voluntary demonstration that allows opt-outs will not tell Congress what would happen if the rules applied to all beneficiaries, without exceptions.
New Benefit Offset Demonstration Has Serious Limitations
In addition to allowing SSA to design new ideas, the budget agreement specifically requires the agency to test a benefit offset intended to improve work incentives. This demonstration is unlikely to produce useful results. It largely duplicates SSA’s ongoing Benefit Offset National Demonstration (BOND), which has produced only limited results, and it has serious design flaws that will sharply limit its potential lessons.
DI already contains significant work incentives (see Figure 2):
- Under current law, DI beneficiaries can earn up to Social Security’s substantial gainful activity (SGA) threshold, set at $1,130 per month in 2016, while receiving full benefits.
- Beneficiaries may earn unlimited amounts and receive full benefits during a nine-month trial work period and three-month grace period.
- For beneficiaries who continue to earn more than the SGA amount after the trial work period and grace period, benefits are suspended. For the next three years, they maintain eligibility for DI and may return to DI benefits automatically if their monthly earnings sink below SGA.
- After this extended period of eligibility, eligibility ends for beneficiaries who continue to earn more than SGA. Beneficiaries remain eligible for expedited reinstatement for the next five years, without serving another five-month waiting period and with streamlined eligibility criteria, if their earnings fall below SGA and their original disability persists.
- In addition, beneficiaries may keep their Medicare coverage for up to 7½ years after their cash benefits stop.
The fact that beneficiaries lose their entire benefit for sustained earnings above SGA is known as the “cash cliff,” which some have criticized for discouraging work. This has led to proposals such as a benefit offset, under which DI benefits would gradually decline as earnings rise. There is little hard evidence, however, that the cash cliff keeps large numbers of beneficiaries from working to their full potential.
|Comparing Current Law, SSA’s Benefit Offset National Demonstration (BOND), and New Benefit Offset Demonstration|
|Current law||BOND||New offset demonstration|
|How much can a beneficiary earn monthly with no impact on benefits?||$1,130||$1,130||$810 or less|
|How long can earnings exceed the above threshold until benefits are affected?||12 months (9-month trial work period plus 3-month grace period)||12 months||0 months|
|What happens when a beneficiary earns more than the threshold?||Benefits are suspended||Benefits are offset by $1 for every $2 in excess earnings||Benefits are offset by $1 for every $2 in excess earnings|
|When is a person’s eligibility for DI benefits terminated?||After 45 months of earning above $1,130||After 45 months of earning above $1,130||The first month a person’s benefits are completely eliminated by the offset|
|Must beneficiaries participate in demonstration?||n/a||Mandatory (stage 1); Voluntary (stage 2)*||Voluntary|
|Can beneficiaries drop out of demonstration?||n/a||No||Yes|
*Stage 1 of BOND randomly selected DI beneficiaries for an offset; stage 2 invited only recruited and informed volunteers
Source: Social Security Administration
The new benefit offset demonstration will be similar to the ongoing BOND project, but with some important differences, as Table 2 shows. In BOND, the so-called cliff is replaced by an offset that reduces benefits $1 for every $2 in earnings over $1,130 (SGA in 2016). In the new demonstration, the offset must start at $810 or less, substantially lower than in BOND. The other major difference is that, in the new demonstration, beneficiaries’ DI eligibility terminates in the first month that their earnings completely offset their DI benefit. In BOND (and under current law), beneficiaries may earn unlimited amounts yet still receive full benefits for a year, and can maintain eligibility for nearly four years after earning above SGA, as explained above.
BOND’s lackluster results to date should temper expectations about the prospects for the new benefit offset demonstration. BOND’s offset is a pure incentive — it cuts benefits for no one compared to current law and raises benefits for anyone who manages to work above SGA. Yet BOND has produced only a very modest increase in earnings, a slight increase in benefit costs, and no reduction in the number of beneficiaries. There’s no reason to think that the new offset, which isn’t a pure incentive, would encourage any more work than BOND — and good reason to think that it could discourage work in some cases. This is because the new demonstration would actually penalize work for beneficiaries who earn more than the new threshold but less than the SGA level ($1,130 per month in 2016). Thus, while some people would have a financial incentive to work more, others would face a cut in benefits and lower total income and might decide to work less.
Such a benefit offset could also be costly, even if it doesn’t encourage any new work effort. An offset would allow beneficiaries with earnings above SGA to receive a partial DI benefit, whereas under the current-law “cash cliff,” they would stop receiving benefits after a year of earnings at this level.
Terminating benefits immediately when earnings rise above a certain level makes work riskier for beneficiaries. Offset advocates often say that the current rules make it too risky for DI beneficiaries to attempt work because they fear going over the “cliff” and losing their benefits. But beneficiaries don’t lose their benefits as soon as they earn more than SGA. Only after at least 45 months of trying to make it in the workforce — despite significant obstacles — do they have their eligibility terminated.
This contrasts with the benefit offset demonstration stipulated in the budget deal, which eliminates both the trial work period and the extended period of eligibility. Beneficiaries who earn enough to offset their entire benefits — even if only for a single month — have their benefits terminated immediately. They lose not only that month’s benefit, but their eligibility for DI in any future month. If for some reason the job doesn’t work out (which is common among people with severe disabilities), beneficiaries must reapply under the expedited reinstatement rules. Furthermore, a beneficiary’s extended Medicare coverage is shortened by nine months. All of these factors combine to make work riskier and less attractive for beneficiaries.
In addition, unlike the SGA threshold — which is the same for all beneficiaries and is less than full-time earnings at the minimum wage — the amount that a beneficiary will be able to earn under the new demonstration and still receive DI will vary dramatically for different beneficiaries, which will allow some beneficiaries to earn above-average wages. Under a $1-for-$2 offset, people with lower benefits have much lower earnings limits than those with higher benefits. Assuming the offset starts at $810 in monthly earnings, a person with an average monthly benefit of $500 could earn only $1,810 a month (about $22,000 a year) before being cut off DI, while someone with a $2,000 benefit could earn as much as $4,810 a month — or nearly $58,000 a year, well above the average American worker’s wages (see Figure 3).
This disparity matters because people with low DI benefits are worse off in other ways: they have lower past earnings and lower current income. Under the new offset demonstration, they also lose DI at much lower earnings levels than do other beneficiaries. This feature also raises questions about whether a person who can earn wages above the average for the U.S. economy should be eligible for a program designed for people unable to support themselves through work.
In addition, like all of the demonstrations permitted under the new legislative authority, the offset will be voluntary, and volunteers will be allowed to opt out at any time. Selective participation will severely limit the demonstration’s value. Volunteers will have greater work capacity than typical DI beneficiaries and are unlikely to stick with the demonstration if it means their benefits will be reduced or terminated. Given the risks of participating, not enough beneficiaries might volunteer to stage a demonstration at all. Because of all these limitations, the offset demonstration won’t show policymakers what would happen if this benefit offset were the law of the land.
Offset Demonstrations Won’t Examine Impact on Number of New Beneficiaries
One important question in analyzing changes to DI — particularly those that could expand eligibility, increase benefits, or offer new services — is whether they will increase the overall number of DI beneficiaries. For example, a benefit offset could attract new applicants who meet DI’s strict medical criteria but are able to earn more than SGA — which is not permitted under current law, but would be under a benefit offset. In fact, under the offset demonstration, beneficiaries could receive DI benefits and earn above-average wages, as discussed in this report.
Neither BOND nor the new offset demonstration is designed to test whether the offset will induce entry into the program. An advisory panel for BOND concluded that such testing would require a very large number of beneficiaries and communities, which would be “far too costly and unlikely to be able to accurately measure the induced entry effect.”a BOND’s design explicitly does not test the question of induced entry because of the expense and difficulty.b
A benefit offset would cost money if the number of new beneficiaries it attracted exceeded the number of current beneficiaries who worked their way off of benefits. There’s no reason to think the new benefit offset would result in a significant number of current beneficiaries leaving the program, given that BOND’s more generous offset hasn’t done so. However, there’s good reason to think that an offset policy — if it were a permanent feature of the law — would attract new beneficiaries. Millions of people meet the medical criteria for DI but don’t receive benefits; many of them earn more than SGA.c If even a small fraction of them became beneficiaries, it would increase DI’s cost and expand its scope.
a Hugo Benítez-Silva, Moshe Buchinsky, and John Rust, “Induced Entry Effects of a $1 for $2 Offset in SSDI Benefits,” May 3, 2011, http://ms.cc.sunysb.edu/~hbenitezsilv/jpam_2011new.pdf.
b David Stapleton et al., “BOND Implementation and Evaluation: BOND Final Design Report,” Abt Associates, Inc. and Mathematica Policy Research, Inc., December 3, 2010, https://www.ssa.gov/disabilityresearch/documents/BOND_Design%20Report_FINAL_Del2-2_12-17-10.pdf.
c Debra Dwyer et al., “Counting the Disabled: Using Survey Self-Reports to Estimate Medical Eligibility for Social Security’s Disability Programs,” ORES Working Paper No. 90, January 2001, https://www.ssa.gov/policy/docs/workingpapers/wp90.html.
Finally, the new offset demonstration won’t be able to assess whether the number of new beneficiaries attracted by a permanent, nationwide benefit offset would exceed the number of current beneficiaries who worked their way off of benefits. (See box.)
Critical to Get Next Demonstrations Right
Congress must revisit DI again by 2022. When it does, it will likely look to the results of SSA’s work demonstrations for evidence on which to base new policies. In addition to the offset demonstration, the law authorizes SSA to conduct others. Though the renewed demonstration authority is flawed, SSA can learn something by testing new ideas. Any new DI work incentive demonstration should be well-grounded in lessons from past experiments and what we know about Americans with disabilities; it also should be structured to get the best possible information. Two areas of interest are early intervention and transitional benefits.
Often, a person’s health and earnings start to deteriorate years before he or she meets the strict criteria for DI. Early medical interventions could help people deal with their health problems, and early work interventions could keep people connected to jobs before their skills atrophy and their attachment to the job market weakens. Such interventions could improve the quality of life for people with disabilities — and, potentially, reduce DI benefit receipt.
A recent experiment, the Demonstration to Maintain Independence and Employment, shows promise. It provided health and employment supports to people with potentially disabling health problems such as diabetes, HIV, or mental illness. In two of the four states involved, a rigorous analysis found the interventions lowered the likelihood that participants received DI and SSI (Supplemental Security Income) benefits.
SSA is already designing an Early Intervention Mental Health Demonstration. It will provide health and employment services for workers under age 50 with mental illness. It targets two groups likely to receive DI benefits in the future: people who receive state vocational rehabilitation services but not DI or SSI, and people recently denied DI or SSI benefits.
The President’s fiscal year 2016 budget proposed that SSA partner with existing state programs like vocational rehabilitation to provide wrap-around services focusing on long-term employment outcomes for workers likely to qualify for SSA disability benefits in the future. Other ideas for early interventions also merit attention.
Early intervention poses difficult tradeoffs. The earlier the intervention, the more likely it will help people with disabilities support themselves and delay or forgo disability benefits, possibly saving money in the long run. However, the wider the net is cast, the less the interventions are targeted toward the people most at risk.
Early intervention often requires coordination among different organizations: federal agencies (such as SSA and the Department of Health and Human Services), state agencies (such as vocational rehabilitation and workers’ compensation programs), and private organizations (such as employers and health care providers). These partnerships can yield significant gains but can be complicated to manage.
Another type of demonstration would divert DI beneficiaries or applicants who are more likely to recover or to be able to work and provide them with the support necessary to stay at or return to work instead of indefinite cash benefits. Diverted beneficiaries would receive medical and/or employment services, as well as financial incentives to work.
SSA could test transitional benefits by screening disability applicants and targeting those who appear likely to be determined eligible for benefits but who also have the potential for significant work. In exchange for suspending their disability insurance application, these applicants would be offered a package of benefits, such as targeted vocational and health interventions, an Earned Income Tax Credit (EITC)-like wage subsidy, and an emergency cash grant. The demonstration would aim both to improve applicants’ well-being and avoid or delay DI benefit receipt.
Transitional benefits pose challenges. For example, how should SSA target people for extra support — and, in some cases, for delayed or time-limited benefits? SSA’s method of categorizing beneficiaries by likelihood of medical improvement wasn’t designed for this task. What should SSA do at the end of the transitional benefit period if a person hasn’t improved medically and remains unable to support himself or herself through work? Implementing transitional benefits would also be labor-intensive, and appeals of agency decisions could further strain the adjudicative process. And extra supports could encourage more people to apply for DI, leading to the same “induced entry” issue that benefit offsets do. (See box.) Finally, some transitional benefit proposals would suspend or even terminate benefits, which would be impossible to test in a voluntary demonstration, when beneficiaries would be free to opt out of the experiment.
Empowering Americans with Disabilities Through Efforts Outside Social Security
SSA’s new demonstration authority provides an opportunity to test new ways of empowering Americans with disabilities to reach their full potential. However, Social Security is only one piece of the puzzle. Disabilities often begin long before a person applies for DI benefits — at birth, a catastrophic injury, or life-changing diagnosis, for example. There are many critical windows for intervention long before Social Security gets involved. Disabled Americans’ work limitations aren’t just a matter of incentives but of very real barriers, like discrimination, lack of accommodation, poor medical care, inadequate education and training, limited transportation, and so on.
Much room exists for improvement in supporting workers with disabilities before they apply for DI.  Expanding Medicaid in those states that have not yet done so would improve access to needed health care and reduce the financial burden of disability. Improving access to the long-term services and supports that many people with disabilities need in order to work could help some of them rise above poverty. Boosting the EITC, particularly for childless workers, could benefit workers with disabilities, who are more likely to earn low wages. Other possibilities outside Social Security worth considering are adequately funding vocational rehabilitation, creating subsidized employment opportunities, and updating the asset limits in SSI, Medicaid, and other public programs.
 The author acknowledges the extensive assistance of Cecile Murray.
 Kathy Ruffing, “It’s Hard to Qualify for Disability Insurance,” Center on Budget and Policy Priorities, August 7, 2015, https://www.cbpp.org/blog/its-hard-to-qualify-for-disability-insurance.
 Paul N. Van de Water, “Few Disability Insurance Beneficiaries Could Support Themselves by Working,” Center on Budget and Policy Priorities, March 9, 2015, https://www.cbpp.org/blog/few-disability-insurance-beneficiaries-could-support-themselves-by-working.
 Paul O’ Leary, Gina A. Livermore, and David C. Stapleton, “Employment of Individuals in the Social Security Disability Programs,” Social Security Bulletin, Vol. 71, No. 3, August 2011, http://www.ssa.gov/policy/docs/ssb/v71n3/index.html; Gina Livermore, David Wittenburg, and David Neumark, “Finding alternatives to disability benefit receipt,” IZA Journal of Labor Policy, March 1, 2014, http://www.izajolp.com/content/pdf/2193-9004-3-14.pdf.
 Kathy A. Ruffing, “No Surprise: Disability Beneficiaries Experience High Death Rates,” Center on Budget and Policy Priorities, April 4, 2013, https://www.cbpp.org/blog/no-surprise-disability-beneficiaries-experience-high-death-rates.
 Arif Mamun et al., “Employment among Social Security Disability Program Beneficiaries, 1996–2007,” Social Security Bulletin, Vol. 71, No. 3, August 2011, https://www.ssa.gov/policy/docs/ssb/v71n3/v71n3p11.html.
 Table 1 lists the project’s initiation as the first year of the design phase. An earlier version of this table used the first year of the implementation phase. Paul N. Van de Water, “Promoting Opportunity for Social Security Disability Insurance Beneficiaries,” Center on Budget and Policy Priorities, July 9, 2015, https://www.cbpp.org/social-security/paul-n-van-de-water-promoting-opportunity-for-social-security-disability-insurance.
 For further reading on DI experiments, see SSA’s overviews of six of the major demonstrations highlighted in Table 1 at https://www.ssa.gov/disabilityresearch/demos.htm. In addition, SSA has collected a number of papers and articles analyzing these and other experiments at https://www.ssa.gov/policy/research_sub12.html#sub28.
 Social Security Administration, Office of the Chief Actuary, “Estimate of the Bipartisan Budget Act of 2015, H.R. 1314,” October 27, 2015, https://www.ssa.gov/OACT/solvency/JBoehner_20151027.pdf.
 The new law grants SSA the authority to modify certain DI program rules through 2021 for purposes of demonstration projects under section 234. Prior law (section 1110) already allowed SSA to conduct other demonstrations that didn’t require such waivers; SSA is beginning an early intervention demonstration using that authority.
 For detailed information about DI and SSI work incentives, see Social Security Administration, 2015 Red Book, http://www.socialsecurity.gov/redbook/.
 SGA rises each year with the Average Wage Index. For blind beneficiaries, SGA is $1,820 in 2016.
 The trial work period is at least nine months (not necessarily consecutive) of work above the trial work period threshold ($810 in 2016) over a rolling 60-month period.
 Kathleen Romig, “Disability Insurance Benefit Offset Could Harm Beneficiaries and Discourage Work,” Center on Budget and Policy Priorities, September 4, 2015, https://www.cbpp.org/sites/default/files/atoms/files/9-3-15socsec.pdf.
 Dave Wittenburg et al., “BOND Implementation and Evaluation Third-Year Snapshot of Earnings and Benefit Impacts for Stage 1,” Social Security Administration, Deliverable 24.c.3 April 22, 2015, http://www.ssa.gov/disabilityresearch/documents/BONDS1Y3_Snapshot%20Rpt%20FINAL_042215_SSA.pdf and Daniel Gubits et al., “BOND Implementation and Evaluation First- and Second-Year Snapshot of Earnings and Benefit Impacts for Stage 2,” Social Security Administration, Deliverable 24c.5 August 15, 2014, http://www.ssa.gov/disabilityresearch/documents/BOND_Deliverable%2024c5_8-14-14.pdf.
 While SSA determines whether a beneficiary can be reinstated, SSA can pay provisional benefits for up to six months.
 The maximum threshold in the new offset demo is $810 in monthly earnings. If SSA set the earnings threshold below $810, the earnings cutoffs would be correspondingly lower.
 According to Social Security’s actuaries, average wages in 2014 were about $45,000; see https://www.ssa.gov/oact/cola/awidevelop.html.
 DI makes up the majority of most beneficiaries’ income. See: Michelle Stegman Bailey and Jeffrey Hemmeter, “Characteristic of Noninstitutionalized DI and SSI Program Participants, 2013 Update,” SSA Research and Statistics Note No. 2015-02, September 2015, https://www.ssa.gov/policy/docs/rsnotes/rsn2015-02.html.
 The experiment was run jointly by the Department of Health and Human Services and the Centers for Medicare and Medicaid Services and authorized by the Ticket to Work Act. See Denise Whalen et al., “Demonstration to Maintain Independence and Employment (DMIE),” Mathematica Policy Research, February 13, 2012, http://www.mathematica-mpr.com/our-publications-and-findings/publications/demonstration-to-maintain-independence-and-employment-dmie.
 For more details, see Social Security Administration, “Justification of Estimates for Appropriations Committees, Fiscal Year 2017,” https://www.ssa.gov/budget/FY17Files/2017FCJ.pdf. SSA will design the demonstration over the next year and then implement it over fiscal years 2017-2022.
 Jennifer Christian and coauthors propose to provide rapid-response case management for individuals experiencing health crises that could cause them to leave the workforce; see “A Community-Focused Health & Work Service (HWS),” http://ssdisolutions.org/document/christian-burton-and-wickizer-community-focused-health-and-work-service. A Bipartisan Policy Center task force has proposed establishing a private-public board to improve cross-sector collaboration to promote workforce attachment for people with disabilities. See “Improve the SSDI Program and Address the Impending Trust Fund Depletion: Consensus Recommendations of BPC’s Disability Insurance Working Group,” Bipartisan Policy Center, August 2015, http://bipartisanpolicy.org/wp-content/uploads/2015/08/BPC-Economy-SSDI-Program.pdf.
 David Stapleton et al., “Targeting Early Intervention to Workers Who Need Help to Stay in the Labor Force,” Mathematica Center for Studying Disability Policy, October 2015, http://www.disabilitypolicyresearch.org/news/~/media/publications/pdfs/disability/sawrtw_earlyintervention.pdf.
 Jeffrey B. Liebman and Jack A. Smalligan, “An Evidence-Based Path to Disability Insurance Reform,” 15 Ways to Rethink the Federal Budget, The Hamilton Project, February 26, 2013, http://www.brookings.edu/~/media/research/files/papers/2013/02/thp-budget-papers/thp_15waysfedbudget_prop4.pdf.
 Other proposals would offer temporary or partial DI benefits to some or all beneficiaries. For example, Kim Hildred and her coauthors have proposed giving temporary benefits to the small fraction of DI beneficiaries most likely to recover, along with rehabilitative and vocational services and no earnings threshold. See Kim Hildred et al., “Transitional Benefits for a Subset of the Social Security Disability Insurance Population,” http://ssdisolutions.org/sites/default/files/hildredmazerskikrentchristian.pdf. In another report, Jason Fichtner and Jason Seligman discuss making all new awards temporary and some partial. See “Beyond All Or Nothing: Reforming Social Security Disability Insurance To Encourage Work and Wealth,” http://ssdisolutions.org/sites/default/files/fichtnerseligman.pdf. The Bipartisan Policy Center’s Disability Insurance Working Group proposed partial DI benefits with more generous earnings thresholds, as well as making return-to-work benefits available after termination. See “Improve the SSDI Program and Address the Impending Trust Fund Depletion.”
 About 3.7 percent of DI and SSI beneficiaries are categorized as “medical improvement expected.” Of those, only 17-20 percent have their benefits terminated after a continuing disability review. 2014 Disability Policy Panel: Continuing Disability Reviews, Report to the Social Security Advisory Board, December 2014, http://www.ssab.gov/Portals/0/Documents/2014_DPP_Report_Final.pdf.
 Rebecca Vallas, Shawn Fremstad, and Lisa Ekman, “A Fair Shot for Workers with Disabilities,” Center for American Progress, January 28, 2015, https://www.americanprogress.org/issues/poverty/report/2015/01/28/105520/a-fair-shot-for-workers-with-disabilities/.