Greg Mankiw's blog
has a post reacting to Krugman's reaction to some comments by Joe Lieberman about Social Security's implicit debt.
And while I have the greatest respect for Greg as an economist (if only for the Mankiw, Romer and Wiel 'taking Solow seriously' paper) and use his textbook in my economics class, I think he's a little off-the-mark here.
Krugman, on the other hand, either is totally ignorant of what he's talking about (and is showing gross negligence by not phoning up Peter Diamond, left-of-center Social Security expert) or is engaging in deliberately deceptive spinning. I can only hope that the pressure of a weekly column is so large that a once-great scholar like Krugman just doesn't have time to take the 60 minutes I took to write this blog post.
Greg's explanation for the $600 billion growth in the Social Security implicit debt is that it's all about discounting. To put it very rudely, much like Krugman, he's outside his area of expertise, and he didn't look up the facts. However, Greg Mankiw and doesn't descend into ad hominem attacks. His larger point about civility and intellectual integrity require is 100% right though.The meat of the matter:
Social Security calculates its forecasts on a 75 year time horizon. All years beyond 202x show Social Security spending greater than revenue. The farther in the future you go, the bigger than gap is. So, when we go from Social Security calculation for 2004 to its calculation for 2005, we now include 2080 in our forecast. Since that's way beyond 202x, it adds a lot to the implicit debt. To paraphrase Greg's example, we're adding $2 owed in year 21 to the $100 we already owe in year 20.
There's a little table Social Security has drawn up to show "Reasons for Change in the 75-Year Actuarial Balance" at
Social Security prefers to report their actuarial imbalance in terms of percentage of total expected future payroll over their 75-year time period. I'd argue this is cheesy spinning it to make it look smaller, just as Cheney's discussion of the non-discounted shortfalls are scare-tactics spin to make it look larger.
The expected PDV of the shortfall is 1.68% of expected future payroll, up from 1.60 last year. That looks small. Of course, their expected future payroll is $275 TRILLION, so it's actually quite large. If we didn't add 2080 to our forecast window, the increase would only have been 0.03% of payroll rather than 0.08%. So, adding 2080 explains roughly 5/8ths of the increase in the gap.Discounting
2) Now, let me make myself look like an ass by criticizing Greg Mankiw's discussion of discounting. Greg's example implicitly assumes the liabilities in question are non-interest bearing.
But that's not the case for Social Security. Both its assets and its liabilities are interest-bearing.
If the federal government balances its budget and pays the interest on its debt, the debt stays unchanged. If the federal government borrows to cover the interest on the debt, but otherwise balances its budget, the debt increases.
If the discount rate and the interest rate are equal, which Social Security assumes they are (both 5.7% nominal), delaying paying back principal and interest leaves the PDV (as calculated from year zero) of the debt unchanged. Why? Because the PDV of owing $105.7 nine years from now is the same as owing $100 now. The growth in the amount owed and the decrease in the discount factor balance out.
I think Greg's example implicitly assumes we're comparing the PDV owed today to the PDV owed tomorrow when we haven't discounted for the fact that tomorrow is one day later
. But the Trustees report doesn't make that mistake. The 2004 report values the liabilities in 2004. The 2005 report values the liabilities owed in 2005. The change in the PDV of the liability is done correctly by just comparing the two reports.