Over at National Review Online, I write on how the benefits of immigration reform to Social Security may be overstated. My timing was good: since submitting the NRO piece, SSA’s actuaries released a score for the Gang of Eight’s reform legislation, concluding that it would reduce Social Security’s 75-year shortfall by around 8%. The White House today cited that report to declare that “commonsense immigration reform will strengthen Social Security over the long-term.”
But as I point out at NRO, the Social Security Administration’s actuarial model assumes that immigrants have the same earnings and life expectancies as US residents, when in fact immigrants tend to earn less and live longer than natives. This means they’ll pay less into Social Security, receive more generous benefits relative to their contributions (due to Social Security’s progressivity), and collect benefits longer. As a result, immigration won’t help Social Security as much as you’d think.
Over the weekend, I used the Policy Simulation Group’s Social Security microsimulation model to replicate as closely as possible the SSA’s assumptions regarding the number of new immigrants, the reduction in unauthorized immigration, the size of the future beneficiary rolls, and so on. Even out to 2085, I matched them fairly closely.
But the PSG model is different from SSA’s in that it uses immigrant-specific earnings and life expectancies to better catch how much immigrants will pay into Social Security and how much they’ll receive back. Using the PSG model, immigration reform cuts Social Security’s long-term deficit not by 8% but by less than 1%. It’s basically a wash. And as I point out at NRO, if immigration isn’t helping Social Security it’s downhill from there, budget-wise.
I’m sure I haven’t matched SSA’s assumptions perfectly — there are so many embedded in a simulation that it’s impossible to know. And the PSG model surely has shortcomings as well: modeling Social Security and immigration is really tricky. Yet when I tell the PSG model not to model immigrants’ earnings and life expectancies accurately — that is, when I introduce a deliberate error to the calculations — I generate results that aren’t far off SSA’s score. That tells me that members of Congress need to get answers to detailed questions about what is being modeled and what assumptions are being made — and before, not after, they vote on reform. Modeling immigration reform may not be rocket science, but it’s close enough to it that mistakes can really matter.