This proposal is chapter four of The Hamilton Project’s 15 Ways to Rethink the Federal Budget, and a segment in An Enduring Social Safety Net.
Disability insurance is the leading edge of the demographic tsunami that is starting to flood U.S. social insurance programs. Americans who are between the ages of fifty and sixty-five are four times more likely than those between the ages of twenty and forty-nine to be receiving disability insurance benefits. For the past decade, the same baby boomers who are just beginning to create fiscal challenges for Medicare and Social Security have been in their peak years of disability insurance receipt. Spending on disability benefits through the federal Disability Insurance (DI) and Supplemental Security Insurance (SSI) programs has increased from 0.7 percent of GDP in 1980 to 1.2 percent of GDP in 2013. Spending on Medicare and Medicaid benefits for DI and SSI recipients is also slightly more than 1 percent of GDP.
The good news is that spending on disability cash benefits appears to have peaked. With baby boomers transitioning off disability benefits and onto Social Security retirement benefits, and with the next cohorts slightly smaller than the baby boomers, the Congressional Budget Office (CBO) projects that spending on DI will fall by 0.1 percent of GDP between now and 2022 (CBO 2012).
But even though the fiscal burden of disability insurance is not expected to worsen, the program is in significant need of reform. This policy note summarizes the conclusions of a year-long research project designed to establish an evidence-based path to disability insurance reform. Our complete findings are available in Liebman and Smalligan (2013). The project was motivated by the observation that, while a consensus is emerging that changes are needed to the U.S. disability insurance system, there is no agreement around any specific reforms, nor does there appear to be a path in place that will lead to such agreement. Moreover, in most cases we lack the evidentiary base necessary to judge whether specific reforms would do more good than harm.
We therefore recommend a path that identifies promising reforms that are administratively realistic, pilots them or otherwise acquires the evidence necessary to judge their merits, and then rolls them out more broadly if proven benefits are established.
Two immediate steps are needed to start down this path. First, Congress should authorize three demonstration projects centered around early intervention. The key to reducing disability insurance costs is to intervene as early as possible to assist individuals in remaining at work. Waiting until after an individual has been approved for benefits is too late. Second, Congress should give the new Social Security commissioner the tools necessary to improve the disability determination system. Most important, funding for state disability determination services should be placed on the mandatory, rather than the discretionary, side of the budget. This will allow the Social Security Administration (SSA) to make investments in administrative capacity that will reduce spending on benefits—for example, by reducing the backlog of continuing disability reviews.
Like reforms to other social insurance programs, these changes will have a relatively small budget impact over the next ten years, but have the potential to produce much larger savings in later years. A package with these two reforms could save $10 billion to $20 billion over the coming decade, mostly through more thorough initial reviews. If the early intervention pilots are successful and taken to scale, annual savings of as much as 0.1 percent of GDP would be possible. 
1. We reach this estimate by assuming that roughly one-third of DI recipients are potentially able to be targeted for employment services and that the services enable one-third of that one-third to work rather than receive benefits. Net of the cost of the employment services, the savings would be around 0.1 percent of GDP.