It's being reported that George Osborne is planning to issue 100-year gilts or even perpetual gilts in order to 'lock in today's low borrowing rates.' From several perspectives, this makes no sense.

Put it this way. If the government issues a perpetual, it could 'lock in' a borrowing cost of at 3.8% per year (the current yield on War Loan). But why pay 3.8% when it could issue a five year gilt and pay 1.06%?

The answer's simple. Borrowing costs are expected to rise, so that when the five year gilt matures, the government will have to pay a higher interest rate.

But the total cost of a perpetual should be the same as a five year gilt, rolled over indefinitely. This is because the cost of borrowing to the tax-payer is just another way of saying the expected return to the investor. Investors should, in theory, expect the same returns on a perpetual as on five year gilts, rolled over; if they expected higher returns on the perp, they'd all buy the perp and none would buy the five year. Prices should adjust so that total expected returns on gilts of all maturities are the same. Which means total borrowing costs are the same. This is what the expectations theory of the yield curve predicts.

This has, in effect, been the view of the DMO for several years. It has tended to issue gilts fairly evenly across the curve, consistent with the theory that borrowing costs are the same for all maturities.

Now, you might object here that the expectations theory is wrong. You'd be right. But the ways in which it is wrong suggest that Osborne is even more wrong. I'm thinking of two things:

1. The 'preferred habitat' view of the yield curve. This says that the expectations theory fails, because it assumes that bonds of any maturity are perfect substitutes for each other. But this is not necessarily the case. A pension fund wishing to match long-term liabilities prefers long-dated gilts, whilst banks wanting liquid assets prefer short-dated ones.

This theory, though, says that the government should issue short-dated gilts, because yields on these are lower, implying that demand is greater.

2. Mistaken expectations. History shows that the expectations theory is wrong, because an upward-sloping yield curve, such as we have now, does not reliably lead to gilt yields rising; this is consistent with the US's evidence (pdf) over quite long periods (pdf).

But this too suggests Osborne should issue short-dated gilts, as it suggests that borrowing costs might not rise as much as the expectations theory implies.

All this poses the question: why, then, might Osborne want to issue long?

There's only one good possibility. It's that Osborne fears that market expectations are wrong in the opposite direction from what they have often been in the past, and that he believes gilt yields will rise even more than the expectations hypothesis implies. In other words, he expects the cost of government borrowing to rise even more than investors expect. It is this - and only this - that justifies 'locking in' today's low rates.

But there's a problem here. In announcing this, he has set the gilt market wondering whether he knows something they don't. As a result, gilt prices have fallen. And in this sense, Osborne has increased the costs of government borrowing .