Extension risk has come at the forefront of the AT1 market in 2018 when the “extension risk tantrum” hit. The market came to the end of its iterative process of assessing the biggest risks in the market, starting with write-down risk, then coupon cancellation risk and ultimately realizing the first two are fairly irrelevant for the top tier issuers, and extension risk is by far your biggest risk.
For those not familiar, extension risk is the risk of prices declining due to a re-pricing to perpetuity. Prices decline as a you naturally need a higher yield to hold something for 100 years than for 5 years.
How you determine if a bond should be priced to the next call or not is simple.
AT1 coupons reset after the first call date (and periodically thereafter), at the spread at the time of issuance (called the reset spread or “back-end”), plus the 5-year government or swap rate. So the reset spread is your key driver. Put yourself in the shoes of the CFO or treasurer of bank XYZ, if issuing a new AT1 will come at a spread below the reset spread (i.e. you pay less) you’ll call and otherwise you won’t call. Other considerations come in such as the “reputational” risk of not calling, but let’s leave that aside.
Market pricing of extension risk is where that becomes interesting. As extension risk is your biggest risk, you prefer AT1s with higher resets as an investor (all else equal) and need to be compensated for a lower reset spread, as it is more likely that the bond will not be called (the call option of the issuer being out of the money). The lower the reset spread, the more optionality you are offering to the issuer and hence should be paid to take that risk.
Markets have tended to be very inefficient at pricing extension risk and drastically overshooting one way or the other. When markets are weak bonds to tend re-price to perpetuity as investors fear that all bonds will be left outstanding forever (or at least not called at the next call date). During the 2018 “extension risk tantrum” around 70% of the AT1 market was priced to perp – and 100% during the Covid-19 crisis, in which case there is extreme discrimination between low and high back-ends.
Otherwise during times of very strong markets (as is currently) the whole (or close) is priced to call like in early 2018, early 2020 or currently. And in this case the spread “premium” to buy low reset AT1s is extremely low.
As investors in the asset class, we need to make sure we’re compensated for the risk we’re taking. Buying bonds above par with limited upside potential but tens of points of downside if these re-price to perp is definitely not appealing in my view. Especially when you’re not compensated for the risk you’re taking. Better sacrifice a bit of yield/spread to be in more robust structures with less extension risk. And I’m not suggesting sacrificing 100bps of spread, this morning I was looking at a case where sacrificing 10bps of spread gets you >100bps of reset spread. At a time where spreads are around/at the tights, de-risking for free seems like a free lunch. When we’re seeing issuers print AT1 at <300bps of spread, you need to be comfortable holding those to maturity at that spread, or at least to be comfortable with those falling by 10pts more than the rest when the market sells off.
Anyways the very low back-ends are pretty much always a lose-lose game. They sell off harder when the market is risk-off and lag in the early stage of the recovery only to outperform at the very end when everything has compressed materially already.