Under the international regulatory framework known as Basel III, banks are required to maintain a certain level of regulatory capital. This capital consists of three components: Common Equity Tier 1 (CET1) capital, Additional Tier 1 (AT1) capital, and Tier 2 capital. For a bond to be counted as AT1, it must be perpetual and cannot be called for at least five years. Many of these have fixed coupons until the first call date, and then switch to a floating rate. In contrast with previous regulation, securities issued under the new regime must not feature step-ups or other incentives to redeem, which significantly increases extension risk for investors. In this blog post we look at when and why these bonds are called—or, perhaps more importantly, not called. As we shall see, this is not an exact science, and different investors may have different views on whether these bonds will be called or not, which is why Axioma RiskTM now allows users to enter their own view on call dates at an individual security level.
First, a bit of history. In the earlier Basel Accords (namely Basel I and Basel II), Tier 1 capital included perpetual bonds with a fix-to-float structure, callable on the date when the rate switched from fixed to floating, often five years after issue (giving them the name ‘non-call 5s’) and featuring a step-up if they were not called. This was a popular structure, and a defining feature of these bonds was that the step from fixed coupon to floating-rate coupon was generally at a penal rate. Consequently, the call option on these bonds was almost always exercised. So great was the expectation that these bonds would be called—even when it did not seem to make economic sense—issuers would call it rather than suffer the negative press generated by not calling it.
That all changed with Basel III. While the fix-to-float perpetual non-call 5 structure has been maintained, the step-up has gone and banks are not permitted to do anything, which creates the expectation that a call will be exercised. And if they do call, they must continue to maintain a level of capital above their minimum capital requirement. Often, this means refinancing the called bond. In theory then, whether an AT1 bond is called or not all comes down to economics. And understanding the conditions under which a bond might not be called is important because of what is known as extension risk. Extension risk is the risk that an issuer will defer repayment of a security, which for bond investors generally means the portfolio will experience an increase in duration.
When deciding to refinance an AT1 bond, the pertinent question is whether the fixed rate payable on a new issue is more or less than the floating rate payable on the existing security. Although the first call date can be any time after five years, in an upward sloping interest-rate environment the cheapest option is likely to be the one with the nearest call, i.e., at five years. The most common floating rate index for Euro-denominated AT1 bonds is five-year swaps, so the two values we should compare (for a bond issued five years ago) are:
- Current floating rate = five-year swap + margin set at issuance
- Fixed rate of approximately five-year swap + five-year spread for the issuer
Since the margin at issuance would have been set somewhere near where the five-year spread for the issuer at the time, we need only to compare the five-year spread from today with the five-year spread from when the bond was issued.
While in euros, we estimate 96% of AT1 bonds are indexed to five-year swaps, there is a little more variety in USD AT1 bonds. According to our estimates, 75% of USD AT1 bonds are indexed to either five-year swaps or a five-year Federal Reserve benchmark rate, and 18% are indexed to three-month USD LIBOR. At a market level (but not at a security level), using the approximation of comparing spreads is probably reasonable for the two different five-year rates, but not great for bonds indexed to a three-month rate.
When assessing an individual security, it is important to consider specifics relating to the issuer, but for the market as a whole we can use this as a guide for when we would expect non-call 5 bonds to be called by comparing spread curves today with spread curves five years ago, or seven years ago for non-call 7s. Globally, we estimate that approximately two-thirds of AT1 debt is rated BBB3 or BB1, with the overall distribution skewed to lower rating bands. So, as a proxy for the EUR and USD AT1 markets, we use our Euro-denominated and USD-denominated BB1-rated Global Banks curves. Had we chosen BBB3-rated curves, or chosen European banks issuing in EUR and North American banks issuing in USD, this would not have qualitatively impacted our analysis.
The first AT1 capital bonds were issued in 2013 with an earliest possible call date in 2018, so this is when we start our analysis.
Data as of May 2021
Euros: The dark blue line in the chart above represents the five-year spread of global banks with a BB1 rating issuing in euros. The green and light blue graphs are the same time series, but lagged by five and seven years, respectively, to approximate the floating margins that would have been set at the time of issuance. When the green line is lower than the current five-year spread (dark blue line), then it is likely that any non-call 5 bond about to switch to the floating rate will do so at a rate lower than the fixed rate required for a new issue. Consequently, it makes little sense to refinance, and we would expect any such bonds not to be called. In the three years since the first non-call 5s reached their first call date in 2018, we see that the five-year lagged spread has been higher than, or at similar levels to, the five-year unlagged spread for all except some of 2019, and, more noticeably, for a few months in Q2 2020 after the world went into lockdown. Only during these periods would we expect to see non-call 5s left outstanding.
In the case of non-call 7s, the light blue line representing the five-year spread lagged by seven years has mostly been above the unlagged five-year spread, indicating that the fixed rate required to refinance is probably lower than the upcoming floating rate for bonds nearing their fix-to-float date. A call for these bonds seems likely. Coincidentally, the progression of the five-year spread seven years ago is following the same trend as the five-year spread since March 2020, as seven years ago we were recovering from the European sovereign debt crisis.
Data as of May 2021
Looking now at what actually happened, we see that in 2019 all EUR-denominated AT1 bonds were called. One large issue was not called at the first opportunity in 2019 when the bond switched to a floating rate, but a little later in Q1 2020. For bonds with periodic calls, the economics of making a call need to be re-assessed at each call opportunity, and as our lagged time series moves above our unlagged one, it is not surprising that bonds deemed uneconomic to call can now be refinanced more cheaply. Approximately 11% of AT1 bonds were left uncalled in 2020. In fact, these were all eligible to be called in mid-2020, when refinancing would have been expensive. As of May 2021, all bonds have been called.
Data as of May 2021
US dollars: Here it seems there is a possibility of not calling bonds in 2019, and as in euros, a clearer rationale for not calling in Q2 2020 when the refinancing rate implied by the unlagged five-year spread in blue is significantly higher than the lagged spread in green. Looking at the chart of calls, we see that that 20% of bonds were not called in 2019, and 21% were not called in 2020. However, if we consider only those bonds that are indexed to a five-year rate (the right-hand chart), we see that all bonds were called in 2019 and only 11% were not called in 2020. In fact, the bonds that were not called all had their first call date in Q2 2020.
Data as of May 2021
What all this implies is that banks are still keen to call their AT1 bonds, even when there is little obvious economic benefit in doing so. One possible explanation for this is in what is meant by “economic benefit”. In this analysis we have focussed on the dynamics of a single security. Bank treasurers are far more likely to take a long-term holistic view of financing costs. Maintaining a consistent policy of calling AT1 bonds reduces extension risk for investors, who will then demand lower premiums (in the form of coupons), which, in turn, will keep long-term capital costs down. This needs to be weighed against the short-term gain of calling a single bond and refinancing it at a lower cost.
What does this mean for investors?
- If spreads were to remain around their current low levels, it would seem reasonable to assume that most of the outstanding AT1 issues approaching their first call in the next 18 months or so will be called. The first time issuers might be tempted not to call would be late 2022/early 2023, when the bonds issued at the low margins prevalent in late 2017/early 2018 reach the ends of their fixed coupon periods.
- A pure bond valuation model is not enough to determine whether AT1 bonds will be called or not. Neither is it safe any longer to assume they will all be called. Not all investors are likely to agree on the most likely call date for a bond: that means your risk system must allow you to impose your own view, at a security level. Functionality introduced recently in AxiomaRiskTM allows users to override the expected call/maturity dates for bonds. To learn more, please contact your Qontigo representative.