From a recent report by the Government Accountability Office (GAO):
Disability beneficiaries participate in the Social Security Administration’s (SSA) Ticket to Work and Self-Sufficiency program (Ticket to Work) by assigning a "ticket" to service providers who, in turn, provide help with employment. SSA compensates the service providers when Ticket to Work participants achieve designated levels of work and earnings. Using SSA data from 2002, when the program began, through 2018, the most recent year available, GAO estimated that 5 years after starting Ticket to Work, participants’ average earnings were $2,451 more per year than that of similar nonparticipants. However, the majority of participants remained unemployed 5 years after starting Ticket to Work.
Based on GAO’s analysis, the costs of Ticket to Work exceeded the savings in disability benefits to SSA by an estimated $806 million from 2002 through 2015, the most recent year with reliable savings data. Savings accrue when Ticket to Work participants receive lower benefits or leave the disability rolls due to earnings from work. GAO estimates that participants were slightly more likely to leave the rolls (9.7 percent) than nonparticipants who are similar across a range of characteristics such as age, gender, disability type, and education level (8.6 percent). A greater percentage of participants left the disability rolls due to work rather than for other reasons, such as medical improvement ...
GAO estimates that SSA incurred an additional $133 million to $169 million in costs (above the $806 million) from disability benefit overpayments to Ticket to Work participants. Overpayments can occur when beneficiaries who work do not report earnings to SSA or SSA delays in adjusting their benefit amounts. ...
This report ignores an important reality. No one makes claimants participate in the Ticket to Work program. Those who do participate are saying they believe they may have some capacity for work, meaning that they have better prospects for returning to work than claimants generally. Comparing the results in this group to claimants generally makes the Ticket to Work results look better than they are. Some of those who volunteered for Ticket to Work would have returned to work with or without Ticket to Work. We can't tell how many were only able to return to work due to Ticket to Work. If there were a control group, it would be some who volunteered for Ticket to Work but were arbitrarily denied Ticket to Work help but there is no such control group because none were turned away.
Note the ridiculous level of overpayments associated with Ticket to Work even among a group of beneficiaries who told Social Security they planned to return to work. Don't blame claimants for this. Social Security's system for handling reports of return to work just doesn't work.
I'll make a modest suggestion. Social Security should almost completely stop relying upon wage reports from claimants because they do a poor job of reporting and Social Security does a worse job of recording what they report. We should define Substantial Gainful Activity on a quarterly basis and base work deductions on quarterly wages reported by employers. This can mostly be done by the computers. Yes, there will be problems with "wages" that are reported for work done earlier (such as sick pay, vacation pay and residual commissions) and yes, there will be problems with self-employment income but we'd be in a much better place than we're in now if those are the only problems we have to deal with.