On the Counter-Cyclicality of Disability Insurance

6/18/2013 10:55:00 AM
Harold Pollack tweeted about this paper which finds no correlation between expiration of extended unemployment benefits during the Great Recession and enrollment in social security disability insurance (DI).

One of the great puzzles of health economics is why, exactly, the number of people on DI is counter-cyclical, which is to say that the number of people applying to disability rises during recessions and falls during expansions. Here's a typical graph I lifted from Google (via FiscalTimes). Note that the graph shows DI applications, not approvals--we'll circle back to that point.

Indeed, republicans often present the counter-cyclical variation of the program as proof that there is widespread "waste" and fraud, since a requirement for the program is that you have to prove that you are completely unable to work and unlikely to ever be able to work again in your lifetime. Presumably the incidence of disabilities is exogenous to the business cycle, which suggests that some of the people applying would actually be able to work if they could find a job (I suppose real business cycle theorists could argue that recessions are caused by outbreaks of disabilities. But that would conflict with their ideological need to believe that DI is wasteful and over-generous).

The Rothstein paper linked above is valuable because it uses exogenous variation in the generosity of unemployment benefits during the great recession across states to test whether expiring unemployment insurance (UI) causes people to apply for, and get, DI. Basically, the legislatures screwed up several times during the recession, allowing UI benefits to expire and then get later renewed, leaving uneven gaps in coverage which are exogenous to the economy, allowing a causal analysis of the impact on DI. There are potential holes in the analysis, but it appears that DI is not an immediate (within a year) recourse for people who lose UI.

There are a couple additional points I want to make. First is that the graph above, which is most often the kind cited by critics of DI, overstate the magnitude of the issue. Here's a much more revealing graph from Rothestein:

In fact, almost all the alleged counter-cyclical variation disappears when you look at actual grants of DI (dashed line) rather than applications (solid line). And more generally, the business-cycle fluctuations account for a relatively small portion of the overall variation in DI variation. I will try to dig up the actual studies when I have time, but from what I recall of the research literature, total "endogenous" factors including the business cycle, demographics, and incidence of disabilities, account for only about 25% of the total variation in DI enrollees. In reality, most of the historical variation in DI cases is actually policy-driven. On the one hand, various presidents with anti-government ideologies occasionally go on anti-DI crusades, forcefully purging disabled workers from DI benefits--this is what caused the decline in DI cases around 1980. And it was the reversal of this policy--not the 1990 recession--that caused most of those workers to be added back into the program. Since 2000, most of the rise has actually reflected increases in the types of disabilities recognized by the Social Security administration, with mental disabilities--a category that never used to be eligible--comprising most of that increase.

Hence, I think the narrative of unemployed workers faking disabilities during recessions is basically false.