Social Security Disability Insurance (SSDI) and Supplemental Security Income (SSI) pay about $200 billion a year in cash benefits to nearly 13 million Americans and give recipients access to public health insurance. The long-running public debate over these programs centers on people who receive benefits without a severe health condition.
Deshpande and Lockwood (2022) quantify the overall insurance value of U.S. disability programs, including the value of insuring nonhealth risk such as job loss, productivity shocks and changes in family structure. The paper focuses on “mismatches with respect to health”: people who receive benefits despite not having a health condition above a severity threshold (“less-severe recipients”) and people who do not receive benefits despite having one (“more-severe nonrecipients”). The main data source is the Panel Study of Income Dynamics, whose 2017 wave measures consumption and earnings for calendar year 2016. These data are supplemented with Social Security Administration records to measure mass layoffs, bankruptcy, eviction and foreclosure.
The paper’s positive analysis establishes that selection into disability receipt on nonhealth grounds is strong, and that less-severe recipients are a highly selected group. Before receiving benefits, they are 40 percent more likely than more-severe recipients to have experienced a mass layoff, 19 percent more likely to have experienced a foreclosure and 23 percent more likely to have experienced an eviction, and their consumption just prior to receipt ($17,000) is close to that of more-severe recipients ($16,400).
The paper estimates the MVPF of a reduction in receipt among individuals with less-severe health conditions. The paper separately estimates the MVPF of a marginal change in the level of disability benefits, and the MVPF of a reduction in allowance rate.
MVPF = 1.4
Net cost is measured as the change in government spending on the household, expressed per dollar of benefits that would be paid absent any behavioral response.
The mechanical component is $1.00: the $13,000 annual benefit that the government would no longer pay in the less-severe recipient states the reform removes from the rolls, counted in those states in which the household would have been below the earnings limit anyway. Across less-severe recipient households this comes to about $12,000 a year per household.
The fiscal externality is $0.34 per mechanical dollar. It has two parts:
Net cost is therefore $1.34 per mechanical dollar, or about $16,000 per less-severe recipient household per year. The health insurance component of disability benefits is assigned zero net cost in the baseline, on the grounds that it largely displaces other subsidized health care that recipients would otherwise have received.
Willingness to pay is the ex ante amount the household would give up, in states in which it does not receive disability benefits, to keep the benefits the reform denies: $1.9 per mechanical dollar, or about $22,750 per affected recipient household per year.
The willingness to pay estimate is determined as follows:
The ex post value of receiving benefits is the full $13,000 benefit in states where counterfactual no-benefit earnings fall below the substantial gainful activity limit, and zero elsewhere. As noted above, across affected recipient households this comes to about $12,000 a year per household. The paper notes that this treatment is conservative: someone whose counterfactual earnings were up to the limit plus the benefit could still gain by working less and consuming more, so setting their value to zero understates willingness to pay. The value also excludes any hassle or stigma costs of claiming.
The markup converts that ex post value into ex ante willingness to pay. One dollar of expected income in a recipient state is worth $(1 + M) of expected income in a nonrecipient state, where M is the proportional gap between marginal utility in the recipient state and mean marginal utility in nonrecipient states. Averaged across all less-severe recipient states, and across education-defined risk types, the estimated markup is 0.88 ≈ 0.9. A markup of this size corresponds, under the paper’s assumption of a coefficient of relative risk aversion = 2, to a consumption gap of roughly 27 percent between the affected recipient states and nonrecipient states. The paper measures recipients’ consumption while they are receiving benefits, which overstates their counterfactual no-benefit consumption and so understates the markup and the value of the benefit.
The estimated willingness to pay is then $12,000 * 1.9 = $22,750 (or $1.9 per dollar of direct cost).
Dividing willingness to pay by net cost gives $1.90 / $1.34 = 1.42 (equivalently: $22,800 / $16,000 per affected recipient household per year).
Each dollar the government saves by removing benefits from less-severe recipients costs those households about $1.42 of ex ante value; equivalently, each dollar of net government cost delivers about $1.42 of value. Because the reform is a contraction, the paper reports the corresponding social surplus as negative: −$5,164 per affected recipient household and −$292 per household per year.
Bound, John, Julie Berry Cullen, Austin Nichols, and Lucie Schmidt (2004). “The Welfare Implications of Increasing Disability Insurance Benefit Generosity.” Journal of Public Economics, 88(12): 2487–2514. DOI: https://doi.org/10.1016/j.jpubeco.2003.07.007
Deshpande, Manasi and Lee M. Lockwood (2022). “Beyond Health: Nonhealth Risk and the Value of Disability Insurance.” Econometrica, 90(4): 1781–1810. DOI: https://doi.org/10.3982/ECTA19668
French, Eric and Jae Song (2014). “The Effect of Disability Insurance Receipt on Labor Supply.” American Economic Journal: Economic Policy, 6(2): 291–337. DOI: https://doi.org/10.1257/pol.6.2.291