Looking at the structure of AT1 CoCos is a useful exercise in understanding the mechanics of extension risk. Extension risk refers to the fact that an AT1 is either called at par, after the requisite number of years, or the coupon is reset for a further period if not called. The risk is that as spreads widen, demand from investors lessens since they buy on the expectation the bond will be called at a specific date and can therefore impact market pricing. To illustrate, let us assume a bank issues a perpetual bond in USD today with an initial five-year non-call period and an 8% coupon. If the five-year swap rate in USD is 3%, this is equivalent to an initial spread of 5% (or 500 bps). After five years, if the bank decides not to call the bond, the new coupon will be the five-year swap rate at that time plus the initial 500 bps spread (“reset spread” or “back-end”)
It is important to consider what is likely to happen in five years. The actual decisions on whether to call or not are based on a wide range of factors and can be specific to each situation, but let us think about the cost of refinancing to the issuer. The issuer needs to consider the spread (cost for the issuer) of a potential new AT1 and compare this cost to the reset spread of the existing instrument. If the spread of a potential new AT1 is below the reset spread, the issuer will be incentivised to call (as the new instrument has a lower cost), while if the spread is above the issuer will likely leave the bond outstanding. Therefore, the reset spread is a key driver of extension risk.
For investors, the risk is linked to both unforeseen non-call events (not reflected in the price) and reassessing the risk of non-calls where prices drop as a result. To illustrate an extreme example of the former, Chart 1 shows the sharp 14% fall on the day Standard Chartered announced the non-call of a legacy Tier 1 instrument in late 2016, reflecting the market’s adjustment for a new coupon set at 151 bps over three-month Libor (the all-in coupon was circa 2.55% after the non-call compared to the previous level of 6.409%). Clearly investors had not anticipated this non-call event and the adjusted price reflected both a longer maturity (uncertainty of future calls) and the lower coupon.
Chart 1: Illustrating unexpected non-call re-pricing event – low back-end instrument
We believe the performance of AT1 CoCos in 2018 is a good example of the reassessment of non-call risks. As global AT1 CoCo spreads widened, by circa 160 bps, over the year1 there was a negative feedback loop (wider spreads led to increased pricing to perpetuity, which in turn pushed spreads wider) and this disproportionally impacted bonds with lower reset spreads.
AT1 CoCos are perpetual, with no final maturity, and can theoretically never be redeemed. The instruments carry no incentive to call and, as these are fully eligible as capital under the current regulatory framework, are an efficient source of capital compared to shareholders’ equity (estimated cost of equity circa 10-12% compared to an average 6.2% Yield on AT1 CoCos2). We expect the main driver of decisions to call to be refinancing economics – where issuers would call if they can issue a new AT1 at a cheaper level.
When spreads on AT1s significantly widened in 2018, a move that was exacerbated by increased pricing of instruments to perpetuity, there was a mechanical effect that decreased the likelihood of calls due to higher potential refinancing costs, which in turn raised fears of bonds being extended past their next call date. As investors require higher yields and spreads as compensation for a potentially longer holding period (ie extension past the next call) and a lower coupon (than on an equivalent AT1 issued) this caused further weakness in the market. This weakness was compounded by the fact that circa USD 20 billion of AT1s were issued from late 2017 to early 2018 (Q317-Q118) at the peak of the market with record low reset spreads.
Chart 2 illustrates the scale of this change: close to 70% of the AT1 CoCo market was pricing a non-call at the end of December 2018 compared to just 2% at the end of January 2018, suggesting the market expects more than half of instruments to be extended beyond the first call date.
Chart 2: Illustrating Extension Risk - % of AT1 CoCos priced to non-call
We believe extension risk will likely remain in focus in 2019 as there is a wave of instruments up for call currently priced to non-call. This marks a structural shift compared to 2018 where the AT1s up for call were generally refinanced at cheaper levels for issuers and had high back-ends. Despite the potentially negative initial impact on prices, should one or several instruments not be called, the market has in our view priced in extension risk to a large degree. However, instruments with lower reset spreads are now almost fully priced to perpetuity, which we think is wrong.
Overall, extension risk will likely remain an important driver of AT1 CoCo performance, but we feel the current re-pricing to perpetuity has been overdone as there is limited risk of instruments being left outstanding forever. Investors should also pay close attention to the specific structure of the instruments. We consider bonds that are callable annually, or more frequently after the first non-call period, to be more attractive as these strongly mitigate extension risk by increasing the optionality of the issuer to refinance at a more attractive level compared to instruments that are only callable every five years.
Finally, despite our belief that refinancing economics is a significant driver of call decisions, issuers will also consider other factors which ultimately leads to diverging approaches. Currently the market is fragmented between a purely “economic” approach (ie solely focused on refinancing cost) and a more debt-holder-friendly approach, which considers the reputational cost of not calling. In our view, this means there is still some potential upside from bonds being called despite pricing in some extension risk.
The European banks AT1 CoCo market has become a bellwether for financial subordinated debt and has evolved into a mature asset class of its own, yet it remains only a fraction of the overall financial subordinated debt market. The current circa USD 170 billion of CoCos in issuance compares to more than USD 800 billion of overall financial subordinated debt of European issuers (AT1 CoCos, Bank tier 2, bank legacy securities and subordinated insurance), therefore we believe it is important to take a step back and look at extension risk beyond AT1 CoCos. Yet, when considering bank legacy securities and subordinated insurance, we believe extension risk is a very different story.
Overall, we believe investors’ focus on extension risk in AT1 CoCos will likely continue in 2019 as the market is close to its end-state size3 and will therefore be entering a mature phase of refinancing – yet we view extension risk as excessively priced given that true perpetuity risk is limited and security selection can help mitigate extension risk. Moreover, looking at the broader subordinated financial debt universe, legacy bank capital securities and insurance subordinated debt carry less extension risk. In our view, investing across the capital structure in financial subordinated debt in combination with AT1 CoCos can help mitigate extension risk and potentially offers attractive valuations.