AT1: when refinancing a 5% to 7.25% bond makes sense

One of the risks the AT1 investor faces is extension risk. Issuers have the ability to call these bonds at their discretion and there is no coupon increase if the bonds are not called. While banks typically consider a number of factors when considering whether to call and refinance versus not call and extend, the simple economic impact is of course among the most important.

It seems to us that some investors look at the COCO index, see that the average spot price for bonds in the index is in the mid-80s, and quickly conclude that the average AT1 bond might not be called. Moreover, they hear almost daily from central banks pointing out that higher rates for longer are the new norm. How can it be economical for a bank to call a 5% AT1 coupon and issue a more expensive replacement, if the fed funds rate is likely to peak in the mid-4% zone and stay elevated for a relatively long period of time? Some investors seem to conclude that a significant portion of the AT1 market will never be called.

However, one important factor is missing from this analysis: banks do not carry AT1s on their balance sheets at the fixed coupon that investors periodically receive. There are a few accounting considerations to this, but conceptually it can be thought of in the following way. When a bank issues an AT1 with a first call in five years, it is simultaneously entering into a five-year swap transaction where it receives floating rates and pays fixed rates. For example, if a bank issued a dollar AT1 a year ago with a spread of 400 bps, when the five-year swap rate was close to 1%, then the coupon would have been set at 5% per year until on the first call. At the time of issuance, the bank would have negotiated a five-year swap as described above; the bank. carries the AT1 as an instrument that pays the dollar floating rate plus 400 basis points.

Today, this hypothetical bond would most likely trade at a spot price in the 80s, given that five-year dollar rates have risen 300 basis points and spreads have widened significantly. Suppose that over the next four years, spreads normalize but rates remain relatively high as central banks need to maintain some pressure due to inflation. On the first call, let’s say this bank can refinance its AT1 by issuing a new five-year AT1 with a spread of 375 bps. Suppose, however, that the rates have only fallen slightly. If the five-year swap only fell from the current 4% to 3.5% by then, this bank would refinance an AT1 at 5% with an issued at 3.5% + 375bp = 7.25%. This apparently uneconomical refinancing is in fact economical. The bank carried the old AT1 at a variable rate of +400bp. Since AT1s do not have a step-up, if the bonds are not called, the coupon would be reset to five-year swaps on the day of the non-call + 400 bps (the spread at issuance). The new 7.25% 5-year issue would also be float-swapped, and would therefore be raised to a floating rate of +375bp.

The floating nature of the AT1 once swapped from fixed to floating means that as the floating rate rises, the interest cost to the bank also rises. This is offset from a margin perspective by the fact that when rates rise, banks charge higher rates to lend money to businesses and customers. Indeed, as we see at the moment, banks’ margins tend to increase when the underlying rates increase.

In conclusion, call economics is determined by spreads. Although we believe that some AT1 bonds are unlikely to be called on their first call dates over the next few quarters, it is important to keep in mind that it could be quite economical for banks to call a 5% bond and issue a new 7.25% one in its place. This analysis highlights that the dynamics of calls are very different for banks and for corporates (which are generally not interested in swapping their liabilities from fixed to floating).

It is important to note that it is impossible to draw conclusions simply by looking at bond prices or the relatively low coupons of outstanding AT1s. Indeed, assuming that a given AT1 bond will never be called is like assuming that spreads will never return to normalized levels. History shows that this is extremely unlikely.

Comments are closed.