Angry Bear thinks I made a mistake in my earlier post, where I expressed some skepticism about the claim that privatization of Social Security would do nothing more than change the accounts in which individuals hold their savings.  (OK, I actually said that I find that claim "highly dubious.")   As evidence of  my error, he points to a July 3, 2000 article by Robert Barro that appeared in Business Week. 

I guess I was unclear, because I fully agree with Barro (and AB by extension, I guess).  Here's Barro's version of one of my central observation (which both Barro and AB identify as a logical fallacy).

Compare first the real return under the present Social Security scheme-- about 2% per year-- with the risk-yield of 4% that personal accounts could guarantee by holding inflation-indexed U.S. Treasury securities.  The return to existing programs is so low because of the mechanics of a primarily pay-as-you-go system.  Today's contributions mainly go to pay for the pension benefits of today's retirees, who were the previous generation of workers.  The total return of the Social Security system corresponds to the  growth of overall wage income.  Thus, the real rate of return in an ongoing system is about 2% if the economy grows at that rate in the long run.

Right.  No fallacy there.  The problem is that the comparison  relates to a hypothetical world where we can start from scratch -- choose a pay-as-you-go (PAYGO) social security system or a system of private accounts.  That, of course, is not the world we live in, and things begin to look different when we factor in what has to happen as we make the transition from a PAYGO scheme to a privatized scheme.  That is where Barro and AB see the logical flaw.

So why is it a fallacy  to argue that a 4% yield on personal accounts is better? The return is low in the present system because workers start with a liability  -- providing for the retirees of the previous generation. 

Right again.  This is what I intended to communicate by this statement in my post.

At this point we can start arguing about the interest-sensitivity of saving behavior, the additional tax burdens implied by the transition costs of privatization versus those required to maintain the existing system, and so on.

The emphasis is added.  By that I meant to acknowledge that there is a real debate to have on what that transition might look like under a variety of privatization proposals, after which we can really decide on what system dominates (including, possibly, the present one).   

(By the way, this focus on the transition from one tax system to another was, in my opinion, one of the key lessons from the paper Simulating Fundamental Tax Reform in the United States (AER June 2001) that Angry Bear so generously cites.  In that paper we were studying the effects of a shift to a "flat tax" or consumption-based tax.  It turns out that, starting from scratch, consumption-based tax systems look like a pretty good deal, because they stimulate investment, increase the capital stock, and raise per capita incomes.  However, if you have to worry about shifting from the present income-based tax system, some people are going to be made worse off.  As with social security privatization, there is no free lunch)

To reiterate, here is my point.  I thought AB's post made it sound as if a private-account system and a pay-as-you-go system are equivalent.  They are not.  However, the choice of the former is complicated enormously by the fact that we start with the latter.  I fully concede that you cannot take any old private-account scheme off the shelf and claim that it dominates the current system.  But I think it should also be conceded that it is conceivable that a privatized system might dominate, if properly constructed. 

My former colleague Jagadeesh Gokhale and I did some preliminary arithmetic on this a while ago, asking the question: If we allowed people below some critical age the opportunity to shift to private accounts, while at the same time taxing them to pay promised liabilities to those who remain in the system, would it be in their interest to do so?   Our answer: yeah, maybe.

Hope that clears things up some.  Let the debate proceed.