Kevin: On your definition of “social engineering,” it applies to policies that reduce the amount that the federal government influences behavior. That seems perverse.
The example I used in my previous post wasn’t “single-entry accounting”: I explicitly noted that children would be future beneficiaries as well as contributors. Nor did the example assume that entitlements were in actuarial balance. The point applies even if they are not. (I think this old post by Andrew Biggs, a colleague of mine at AEI who used to be a deputy commissioner for Social Security, underestimates how much a rise in the number of children would improve the finances of that program. But at least he gets the sign right. And he is well aware that the program has a long-term deficit, as he discusses in the post.)