Gilts go up. Gilts go down. But in financial market lingo, gilts are ‘risk-free’ sterling bonds. That’s because governments with their own currency — like Britain’s — can literally make up new IOUs to repay bonds.
As the Reserve Bank of Australia explains:
The expression ‘risk free’ is used because governments are not expected to fail to pay back the borrowing they have done by issuing bonds in their own currency.
Being risk-free, you might reasonably expect the yields on gilts to be lower than fixed rates offered for a comparable term by any other entity. You’d be wrong.
Gilt yields trade at a chunky spread over the fixed-rate leg of interest rate swaps — derivatives in which one party contracts to pay a fixed rate over the term of the swap in return for receiving floating rate payments and the other contracts to pay floating and receive fixed. (Side note: LCH cleared over a quadrillion dollars of notional of interest rate swaps in 2022. A QUADRILLION DOLLARS!!)
The swap spread is the difference between a gilt yield and the fixed rate leg of the interest rate swap. Five-year swap spreads refer to the difference between five-year gilts and five year fixed swap rates, ten-year swap spreads refer to … well, you get the picture.
There have been some odd things happening to swap spreads that have made us scratch our heads. Is the upshot free money for the UK government? We don’t know, but we think it’s worth asking. To understand why, you’re going to have to come with us by way of a quick history of how we got here, and an intro to bulk-purchase annuity insurers. Hooray!
How we got here
Back in the day, swap spreads were always positive. In the late nineties, investing in ‘credit’ meant mostly buying supranational and non-domestic currency sovereign securities, and as such making a call on the direction of swap spreads was maybe credit strategists’ number one task.
Here’s a chart showing twenty year UK swap spreads around the turn of the millennium. If you wanted to buy twenty-year bonds issued by the likes of the World Bank or the European Investment Bank around the turn of the millennium, you would have been able to pick them up at a spread over gilts maybe 10-20 basis points below this line:
Fast forward to the present day and the swap rate trades around 65 basis points *lower* than twenty-year gilt yields:
For wannabe fixed income geeks, there are a few things going on.
First, today’s swap spread is not your parents’ swap spread. The two charts above reference entirely different benchmark swap rates with entirely different credit characteristics. In 2000 the swap curve was the Libor swap curve. Today it’s the SONIA swap curve.
As the Bank explained, back in 2003 the swap spread represented a:
… perception of the systemic risk of the banking sector. This is because the risk of the systemic failure of the banking sector is embedded in Libor rates.
It’s probably not a great surprise that something as risky as a proxy for non-centrally-cleared bank credit risk, like old-school long-term Libor fixed swap rates, should yield more than Gilts.
But regulators have spent years (almost but not quite) killing off Libor after the mega-scandal in which banks paid huge fines and traders were convicted of manipulating the benchmark.
Its replacement, SONIA, the Sterling Overnight Index Average used in almost all modern day interest rate swaps, is by contrast described by the Bank of England as “the risk-free rate for sterling markets”. And interest rate swaps referencing it are overwhelmingly centrally-cleared (by the likes of LCH), and well-collateralised. This (almost) obviates credit risk.
And as this chart from a victory-lap speech on the transition away from Libor shows, SONIA now rules.
So comparing swap spreads of yesteryear (sort of, but not quite, a proxy for the additional yield you’d pick-up switching out of gilts into banking sector credit) with swap spreads today (the difference between the risk-free government bond yield and the fixed rate leg of an interest rate swap referencing a benchmark rate described by the Bank as risk-free) is to compare apples and oranges. The new spread is going to be a lot lower than the old spread.
But still there’s a puzzle. Or maybe three.
Puzzle #1: Why do gilts trade at a spread at all?
First, if you’re not a complete bond geek, how on earth can government bonds offer a chunky additional spread over and above the risk-free swap rate? That is to say, why does an IOU from HM Treasury pay more than an IOU on a derivative contract?
Laurence Mutkin, now Head of Strategic Analysis at the UK’s Debt Management Office, offered a speculative answer as long ago as 2010 when he was still running European Rates Strategy at Morgan Stanley. He reasoned that regulation in the wake of the GFC had repriced the cost of banks’ balance sheets — meaning that bondholders could charge substantial ‘rent’ for precious balance sheets versus the cost of just taking a view as to the path of rates through a less capital-intensive derivative position. His argument that this was a structural change was primarily about gilts versus fixed Libor swaps but can be applied more generally and has held up pretty well. The original note can still be spotted circulating among, and being commented on by, senior folks on LinkedIn to this day.
Puzzle #2: Why does the line go down?
Second, why has the line on the bottom panel of the second chart been heading down?
Moyeen Islam, Barclays’ gilt guru, points to a recurring theme within the gilt market: the question as to who will buy all the supply? With around £100bn net supply from HM Treasury, on top of anything coming out of the Bank’s QT programme, we’re talking large numbers. The combination of net supply and the normalisation of repo conditions underpinned by structural changes in maturity issuance by the DMO, Islam reckons, have pushed gilt spreads out at longer maturities by up to 25 bps. But, as he wrote in a note published last week:
Should pension hedging activity remain subdued, then current asset swap valuations can stabilise or even richen (bonds outperform swaps) due to both a lack of swap receiving and steady gilt buying.
Ah. Pension hedging activity.
We’ve learnt to our cost how important UK pensions are to the gilt market. Pension funds are big beasts, and if they want to grab higher yields with both hands to lock-in discount-rate-driven improvements in their funding ratios, one way is to pay SONIA and receive long-dated interest rate swap rates, bidding down long-term fixed swaps. This route has the advantage of avoiding having to sell illiquid assets at pace.
But ‘pension hedging’ also refers to bulk-purchase annuity insurers. Bulk purchase annuity insurers are the folks to whom defined benefit pension sponsors hand over the keys to their pension liabilities, along with big cheques (in the form of gilts and cash portfolios), then walk away.
Here’s a dynamic:
— Insurer business plans tend to involve earning gilts+x% on assets.
— Gilts don’t pay gilts+x%. 😢😢😢
Because of this, insurers tend to sell gilts, spend the cash on other assets, then use swaps to adjust their duration position. Here’s a copy of the insurers’ asset breakdown patched together from an LCP report published in October 2022. It’s a bit out of date, but gives a flavour:
If this allocation snapshot is any guide, at least two-thirds of buyout money is invested in things that are not gilts. Buying short-dated credit can deliver a yield pick-up, but doesn’t have much in the way of duration. Add in a fixed swap receiver and Bob’s your uncle.
2023 looks to have maybe been the biggest year yet for bulk-annuity insurance flow, and if LCP’s forecasts are right, there are many more record-breaking years to come:
We’ve noted before that the PPF reckon DB schemes collectively have the means to buyout with £150bn to spare, so the numbers don’t look preposterous.
Daniela Russell, head of UK rates strategy at HSBC, has been banging on about this for a while now. Back in May she published a ‘UK LDI/ BPA Special’ in which she forecast in relation to 30yr rates:
… a gradual cheapening of the range for long-end ASW [asset swap] spreads to take place over time. Our expected 50-65bp range could subsequently give way to a cheaper 70-80bp range once the accelerated de-risking phase has ended and buyout activity accelerates further through 2024.
With 30yr ASW hitting her forecast in the opening days of the year, Russell reckoned in a piece last week that:
the significant cheapening of gilts and linkers versus swaps may have led some insurers to switch back into gilts (selling swaps in the process).
Why would they do this? Tactically, insurers could be taking a view on the swap spread. But maybe it’s also because we are heading towards the limit of the economics of the gilts plus credit trade?
Let’s say that an insurer wants to buy USD high grade credit, swap out the rates market risk and whack on duration. Here is a sum setting out how the numbers might work for an insurer wishing to buy intermediate USD credit, strip it of US market risk, swap it back to GBP and correct for a long-dated UK duration position:
-
Receive AA 10yr USD Corp @ UST+56bps = 4.6%
-
Pay 10yr fixed swap and Receive SOFR @ 3.6%
-
Pay SOFR and Receive SONIA on cross-currency basis swap = c0.2%
-
Pay SONIA and Receive 20yr GBP fixed @ 3.65%
Total synthetic 20yr yield = 4.45 per cent
4.45 per cent is still more than buying a straight 20yr Gilt at a yield of 4.30 per cent. But 15 bps is really not much of a pick-up. There may also be some costs that we’ve missed out.
Puzzle #3: Why should non-bond-nerds care?
The third and final puzzle is why any of this might be at all interesting.
Aside from the sheer thrill of UK fixed income and the potential impact of all this stuff on buyout pricing and activity, the whole thing has led us to scratch our head about whether structural demand for UK swaps represents an opportunity.
Specifically — and yes it does sound a little weird — could there be an opportunity for the UK government to issue debt in non-GBP markets more cheaply than it does in GBP?
Rearranging the sum we outlined above, long-dated gilts, swapped to US dollars, offer pick-up of around 30 bps over US Treasuries. Supranational issuers like the World Bank and EIB trade at a spread of around US Treasuries +16bps. It doesn’t seem inconceivable that the UK would be able to issue at similar levels.
It’s entirely possible that we’ve missed a step somewhere, and we haven’t looked into costs that might be piled on by intermediaries. Assuming a 15bp saving on 10% of net issuance, we’d only be talking savings of only £15 million in the first year, so not huge numbers. But diversifying debt issuance across markets at a saving to the taxpayer is perhaps worth at least a ponder.
