A Cautionary Tale About Being Clear About Pronoun Antecedents...

Be careful with your pronoun antecedents, boys and girls!

Greg Mankiw writes:

Greg Mankiw's Blog: Reich on Taxes, Again: As I have noted before, former Labor Secretary Robert Reich has a way of expressing numbers that is so striking it makes me sit up and want to check the facts myself. Here is what he says at his blog now:

repeal [of the estate tax] would cost the U.S. Treasury $1 trillion in its first ten years. That's about equivalent to what's needed to save Social Security over the next 75 years.

Really? That is amazing.

Let's look a little harder at this and see if we can find some numbers that, at least approximately, back this up...

Now I hadn't noticed anything amazing when I read Bob Reich's weblog. So I went back to see why, and I thought as follows: In the late 1990s--before the Bush 2001 tax cut--estate tax collections were running at about 0.35% of GDP, with an expectation (by me at least) that under the then-current estate tax law that ratio would creep up to 0.4% of GDP because of the rising level of wealth inequality in America. If you go to Table 1-2 of the Congressional Budget Office's you will read that the CBO projects the 100-year Social Security deficit at a present value of 0.54% of the cumulative present value of GDP.

So Reich's statement that repealing the (pre-2001) estate tax will wind up costing us as much as the projected 75-year Social Security deficit did not seem amazing to me. The two money flows--estate tax collections under the pre-2001 law, and the Social Security deficit--do appear roughly equal in present value over the next three quarters of a century.

I, you see, had taken the antecedent of Reich's pronoun "that" to be "the cost of repeal of the estate tax."

But Mankiw takes the antecedent of Reich's pronoun "that" to be "the $1 trillion cost to the U.S. Treasury of the first ten years of estate tax repeal." So Mankiw goes on:

How much does repeal of the estate tax cost over its first 10 years? According to the Tax Policy Center, immediate repeal would cost about $300 billion over the first 10 years. That, however, appears to be an undiscounted number. The present value is probably around $250 billion. What does it cost to fill the social security shortfall over 75 years? This report from the Social Security Administration shows the present value of the 75-year shortfall for OASDI of about $5 trillion. Hmmm....Those two numbers don't seem "about equivalent" to me. In fact, the second one seems 20 times as big as the first.

Now I may have been unfair to Reich in the above calculations. Perhaps he means a different first ten years, rather than the ten years starting immediately. And maybe he prefers different Social Security projections than those I have cited. And maybe he prefers a different discounting convention (although I would insist that he treat the two numbers in a parallel fashion). I would not defend the factor of 20 to the death, or even to a brush burn. I can imagine whittling that number down to a factor of 10 or even 5. But can someone get these two numbers within the same ballpark? I doubt it.

Lastly, let me note that the comparison here is silly. Why compare a 75-year shortfall to a 10-year revenue loss? There is no reason to, other than for dramatic effect. But policy wonks are supposed to have less license in making up their drama than playwrights.