Tuesday, May 15, 2012
Meanwhile, in the real world
Posted by David Smith at 08:00 PM
Category: Thoughts and responses

The argument that very low gilt yields mean the government should be borrowing a lot more to fund capital or other spending has been around for a while, though nobody knows how much more borrowing it would take to trigger a flight from gilts and hence higher yields. Now Jonathan Portes of the National Institute has come up with a variation.

He claims to have made an extraordinary discovery, that for the amount of money the government is raising with the pasty tax, which he says is 150 million a year, the government could fund a 30 billion infrastructure programme. He accuses me of being confused for challenging it.

The real interest on index-linked gilts is 0.5%, so 30 billion of funding could be achieved for a "real" 150 million a year, as he says. It looks too good to be true. The pasty tax raises rather less than 150 million a year but let's leave that aside. Let us also leave aside that another 30 billion of index-linked issuance would probably cause market indigestion, almost doubling the amount the Debt Management Office is issuing at present. The question is what is the cost of index-linked funding?

There are two elements to an index-linked gilt. The inflation-adjusted coupon, or interest, and the inflation adjusted redemption value. It is the latter that we should concern ourselves with. It means, for example, that 30 billion of long-dated index-linked gilts issued now will be redeemed for something over 70 billion in 30 years time, using the DMO's 3% inflation assumption (2.75% is consistent with the Bank of England's 2% consumer price inflation target).

So what is the appropriate way to account for this inflation uplift? Is it to take a 40 billion hit in 30 years' time, or is it to allow for it annually? The latter approach has to be the only sensible one, so the first year's inflation uplift of our 30 billion of gilts has an effective cost of 900m, plus the 150m of interest, giving an overall cost of just over 1 billion, similar to the current annual cost of issuing 30-year conventional gilts, and lot of pasty taxes.

The DMO, in calculating nominal yields on index-linked gilts, does precisely this: "The real yields for index-linked issues have been converted into nominal yields using the Fisher Identity and a 3% inflation assumption", it says in its annual report on gilt issuance yields. It means, properly measured, yields on index-linked gilts are close to those on conventionals, as you would expect. There's a nice, and longer, critique, here, which makes some additional points.

You could argue, as Jonathan appears to, that the inflation element doesn't matter, though the persistent inflation overshoots of recent years mean index-linked has been an expensive form of funding. You could argue that the debt could simply be rolled over but that's just passing the buck to future generations.

I apply a simply test to this. If a politician said he could fund 30 billion of infrastructure spending for 150m a year people would accuse him of being a snake-oil salesman. It would be hard to disagree.