Productivity Growth and Social Security Solvency: What Did Jagadeesh Gokhale Really Say?

While Brad DeLong and Donald Luskin truly hate each other (Brad’s contention that Luskin is stupid is exactly right, while Luskin’s attack that Brad is fat was so pointless that Luskin even had to retreat from it), they both agree that Jagadeesh Gokhale’s Wage Growth and the Measurement of Social Security’s Financial Condition is a very good paper:

The Social Security trustees’ economic growth projections receive considerable attention because many people believe that higher growth would significantly improve the program’s actuarial balance (that is, reduce its actuarial deficit). This belief is validated by Social Security trustees’ calculations that show larger 75-year actuarial balances under faster assumed real wage growth rates. Since 2003 the trustees have reported the program’s actuarial balance measured in perpetuity. But they do not provide sensitivity analysis that examines the impact of various assumptions on the infinite-term actuarial balance. This paper shows analytically that faster wage growth may reduce Social Security’s infinite-term actuarial balance if the ratio of workers to retirees continues to decline rapidly beyond the 75th year. This result holds even if the decline in that ratio ceases after just two decades beyond the 75th year. The paper reports stylized calculations of the impact of real wage growth and demographic change – including time-varying rates of change based on official projections for the U.S. economy – on Social Security’s actuarial balance in a multi-period setting. Finally, the Social Security and Accounts Simulator (SSASIM) actuarial model of Social Security financing is used to estimate the degree to which increased wage growth could negatively affect the system’s infinite-term actuarial balance. These results raise questions about the conventional wisdom that holds that improved wage growth would affect Social Security’s financing, and how a widely used measure of Social Security’s financing captures those effects.

Luskin reads this and tries this summary:

For future participants, the present value of taxes and scheduled benefits are roughly equal. Faster growth would increase both the future revenues and the future benefits in approximately equal measure. The revenues would arrive before the higher benefits must be paid, so there’s an illusory improvement if you use a flawed metric like 75-year actuarial balance which counts the inflow but overlooks the outflow, but the actual total deficit, which is on the books already, isn’t really affected. The idea that faster economic growth will eliminate unfunded liability is simply incorrect

When Luskin starts with the premise that “the present value of taxes and scheduled benefits are roughly equal” for future participants – then one has to wonder why he claims there is a supposed shortfall. Of course, the real issue is that retirement of the baby boomers who have paid enough to build a Social Security Trust Fund reserve. Simple steady state propositions therefore don’t tell us that much. Which is why one needs to more carefully read what Jagadeesh Gokhale actually wrote. Brad’s take is as follows:

In a year-by-year pay-as-you-go-balance system, faster real wage growth makes the state of Social Security look better: because it raises each year’s wages relative to that year’s benefits, it means we need a lower tax rate in each future year less above their current levels in order to balance that year’s Social Security tax collections with that year’s Social Security benefits. In a substantially prefunded full-intertemporal-balance system with a sufficiently rapidly declining ratio of workers-to-beneficiaries, faster real wage growth makes the state of Social Security look worse: the declining worker-to-beneficiary ratio means that a larger share of workers than beneficiaries are in the near future (when faster wage growth makes relative wages lower) and a larger share of beneficiaries than workers are in the far future (when faster wage growth makes relative benefits higher), thus faster real wage growth raises average benefits more than average wages.

In other words, it depends on both your metrics (with Luskin being one of those pay-as-you-go-balance types – which means he’s hopelessly confused with his latest loony tune post) as well as one’s assumptions as to transition period before we return to some alleged steady state. Brad continues by noting what Jagadeesh Gokhale really said, but I fear we have gone way past the comprehensive skills of the clown known as Donald Luskin.