When interest rates are high people are supposed to think twice about borrowing. When they’re low, they’re supposed to be whipping out the metaphorical (or literal) credit card. This is the central idea underpinning central banks‘ monetary policy fiddling.
Of course, central banks only set the short-term interest rate. Bond yields, while anchored by central bank rates, are established in the secondary market through a kind of a continual auction process. And in a world where government bond yields describe the likely course of future central bank action, a government should be indifferent to where it issues on this curve.
This, according to a couple of lengthy recent reports by rates strategists Moyeen Islam of Barclays and Mark Capleton of Bank of America — is not the world in which we live.
Before the global financial crisis, ten-year gilts often yielded more than thirty-year gilts (counter to the conventional logic that longer duration = more risk = more yield). Most bond-types understood this as a reflection of persistent demand from UK life insurers and pension funds pursuing Liability-Driven Investment strategies. As such it made sense for HM Treasury to shift their debt issuance longer, meeting this demand and reducing their own so-called “refinancing risk”. By doing this, the UK ended up with the highest average maturity government bond market in the world.
Fast forward to the post-Covid era, and longer-dated gilts trade with a yield not only higher than 10-year gilts, but also with a yield spread over 10s that is greater than those in the US or Germany. Charts via BofA:
(High-res)
Despite Capleton’s “scepticism about whether clever mathematical chromatography can really take a bond yield and deconstruct it into different unobservable (and perhaps hypothetical) component parts”, he reckons that this new yield gap represents a big ol’ chunky term premia — aka a sure sign that long-gilts are expensive for the government to issue. He backs up his view with a bunch of Gilts-SONIA forward spreads and cross-country forward spread charts that are too geeky even for FT Alphaville.
Barclays’ Islam has no such qualms about straight-up stating that rising term premia “explains the bulk of the move higher in [thirty-year] yields”.
Why might this be? A lot rests on the 800lb gorilla of a question hanging over the market for long-dated gilts: whether long-standing demand for them from UK pension funds is pretty much over. While ‘peak LDI’ has been called before, the arguments are worth spelling out.
First up, the massive so-called “de-risking” shift from equities to bonds that has been the bane of pretty much every UK equity manager’s existence for the past two decades has happened. So it cannot happen again. The amount of potential de-risking still to come is, according to Capleton, de minimis.
Second, with defined benefit pension schemes almost all closed to new entrants, people are retiring and ultimately dying. This is now showing up in shrinking membership data. As Capleton writes:
The ONS’s ‘Funded occupational schemes in the UK’ release shows that total membership fell 16% between 3Q 2019 and 1Q 2024, and within the total the composition is aging, with the proportion of pensioners rising from 42% to 49% over that same short period.
Third, the present value of pension fund liabilities has been absolutely cratered by … the rise in bond yields. So if every defined benefit scheme invested the entirety of its assets in gilts this would only represent around a trillion pounds of demand, down from a demand ceiling of around two trillion pounds a few years ago.
The bottom line is that there’s no new pension fund demand for long-dated gilts coming just around the corner. And this fact is showing up in bond prices.
Barclays makes this point with maths. While gilt auctions have gone pretty well, the market’s capacity to digest risk is, Islam argues, a function of underlying liquidity. One way in which Barclays takes a measure of this is by drawing lines along different parts of the gilt yield curve and calculating the root mean squared errors that are generated from these lines of best fit. The underlying intuition is that in a massively liquid market all the kinks will get arbitraged away.
But this is not happening. And the part of the curve where this measure of liquidity has been fast deteriorating is the long-end. Last year we pointed to the increasing kinkiness of the curve as maybe having something to do with a bid for tax-advantages attached to low-coupon gilts. But it looks like there’s something else going on too. Barclays:
(High-res)
As such, BofA’s Capleton reckons:
Gilt issuance needs to adapt, radically and rapidly. … This argues for a material reduction in long-dated Gilt issuance.
And this puts them basically on the same page as Barclays.
It’s not like governments haven’t made big changes to issuance before on the back of long-dated bonds becoming expensive to issue. Older readers will recall that on Halloween in 2001 the US Treasury caused a massive bond rally when it cancelled all long-bond issuance until further notice. Defending the decision, Peter Fisher, then-Undersecretary of the Treasury explained that:
This is about trying to manage taxpayers’ money prudently. … This is a relatively expensive borrowing tool that is simply not necessary for current financing requirements or those we expect.
Moreover — as Capleton reminds those of us under the state retirement age — the American move was an echo of Geoffrey Howe’s decision to cancel long-gilt issuance in his 1983 Budget. And while Barclays strikes a measured tone — encouraging the DMO to shorten the duration of its issuance to both improve market liquidity and reduce taxpayer costs — these more radical examples of wholesale long-bond issuance cancellation are the ones that BofA reckon the DMO should look closely at today.
Rather than issue expensive long-dated bonds whose prime beneficiaries are literally dying, BofA advocates shifting issuance towards UK Treasury bills. The UK’s T-bill market is pretty piddly by international standard, and Capleton makes the case that as quantitative tightening evolves there will be ample demand for bills from banks looking to find substitute assets that behave like Bank of England reserves.
(High-res)
Barclays and BofA both [ed: try saying that several times quickly] make a good case that the UK government, if it were a company, would be re-examining, and shortening, its debt issuance profile.
Capleton also raises the idea that they should buy-back long-dated gilts that are trading at a deep discount, “taking profits” on bonds sold into the market with low coupons, and perhaps knocking off as much as 16 per cent of debt/GDP in the process.
But he jumps the shark when he writes:
With some of these issues, the obvious temptation would be to consult the market. Our main worry with this is the risk that action is delayed, potentially for a long time, if it’s a very formal consultation process … we’d suggest experimental operations rather than full-blown consultations.
Experiment rather than consult? Sorry, we’re British.
Overall, these are two fascinating and imaginative reports full of interesting debt management ideas. But whether the UK government responds to bond market pricing signals in the way normal economic agents are supposed to do is another question.
Disclaimer: The author has direct gilt holdings among his personal investments, one of which is long-dated. 😢
