AT1 ETFs: should we be scared?


Mark Baker
Published on:

Talk of exchange-traded funds offering exposure to additional tier-1 debt may not be as worrying as it sounds

MB banner 660px

A recent Bloomberg report that banks like Citi and Goldman Sachs might be planning to launch exchange traded funds (ETFs) based on additional tier-1 (AT1) debt has sparked fears in some quarters about an increase in volatility in the asset class.

The introduction of a retail investor base, so the argument goes, will bring a very different dynamic to the market.

That sounds plausible, but is it true? And even if it is, does it matter?

Part of the answer depends on how we think about volatility, part depends on the distinction between primary and secondary liquidity (an Important Thing in the context of ETFs) and part depends on who might buy such a product and why.

Before getting into all that, it’s worth remembering the process by which ETF shares are created and redeemed – and the relationship between them and whatever benchmark they are trying to replicate. ETF shares appear and disappear through the actions of authorised participants (APs), financial institutions that enter into contracts with ETF issuers for this purpose.

When new ETF shares need to be created, an AP delivers a basket of relevant underlying securities to the ETF. There might be a discussion between the AP and the ETF manager over exactly what can go into that basket to ensure that the ETF will properly reflect whatever benchmark the manager has indicated to investors.

But once that’s agreed, the AP receives a ‘creation unit’ made up of a bundle of ETF shares – often something like 50,000 – that it can sell or hold. When it comes to a redemption, the AP hands over the same-sized bundle of ETF shares to the ETF and receives a basket of underlying securities in return. Again, the AP can do what it likes with these.

The creation process is primary liquidity – the forming of new ETF shares from scratch.

Secondary liquidity – the ability for an ETF investor to sell those shares – is provided through the fact of the fund being listed and free to trade. If an ETF investor wants to get out, it sells its holding to someone who wants to buy. The ETF doesn’t have to actually sell underlying securities to raise cash to pay out – one of the ways in which it differs from a mutual fund, for example, and pretty key for a manager who might be investing in a fairly illiquid underlying market.

So, if the creation and redemption process doesn’t directly reflect end-investor flows into and out of ETF positions, what is it for? The first thing to note is that it does indirectly reflect those movements – it’s just that creation and redemption are not required for an end-investor to put money into an ETF or take it out.

The creation and redemption process helps APs maintain the value of the ETF near to its net asset value (NAV), which is the value of its underlying assets, calculated and published at a certain time each day. That NAV is by definition a stale figure – if market moving news emerges during the trading day for the ETF, the ETF will reflect it quickly, and in the case of a bond-referenced ETF, often much more quickly than the underlying securities.

This is where divergence of the ETF from its underlying market can occur: heavy buying of the ETF will drive the ETF to a market value far higher than its NAV, while heavy selling will do the opposite. APs, through their power to create and redeem ETF shares, have the ability to reduce this arbitrage (and make money while doing so) by creating or redeeming ETF shares as needed.

Closing the gap

One of the great virtues of ETFs, particularly ones referencing fixed income securities, is that they can offer vastly more liquidity and trading flexibility than their underlying assets, which might be mostly traded over-the-counter. But if that’s the case, how can APs work their value-tweaking creation and redemption magic if the underlyings in question are illiquid?

To know this, we need to look at what happens in practice when an AP needs to create new ETF shares. It can source securities in the market every time it needs to do this, and if that were the only possibility then, yes, a drying up of liquidity in the underlying market could make it difficult to keep up.

But that’s often not what happens. As the head of bond ETF strategy at one of the biggest investment firms in the world tells Euromoney, an AP may well borrow the securities or simply use its own inventory.

When ETF shares are redeemed and an AP ends up with securities back from the ETF, it’s commonplace for that AP to hang onto them, using them for its own client-facing activities, before recycling them back into the ETF the next time new shares are needed. So there can be something of a buffer available before an AP has to source new securities in the market.

That doesn’t mean there are no potential challenges, particularly in illiquid markets or ones that are closed – or where inventory might be lacking.

Let’s say you are running a US-listed ETF benchmarked to Japanese equities. One of the great things about the ETF is that it can trade outside the trading hours of Japanese stocks. But that’s also one of the problems. If there is heavy selling or buying pressure on the ETF, then an imbalance will arise in the market. To counter that would require either buying/selling of the underlying or creation/redemption of ETF shares.

If the AP is not able to rely on inventory, it is likely to need to hedge a position that it will not be able to cover until the underlying market reopens. That hedging might be done in a number of ways, depending on the asset class – perhaps using futures, credit index derivatives, or even proxy hedging with other asset classes, such as using equity for a high-yield bond fund.

Even if trading hours are aligned, the AP might still find itself struggling to source underlying assets, meaning that it may not be able to affect the trading value of the ETF sufficiently to bring it back into line with the underlying assets, as reflected by the NAV.

What about panic?

So, what of the specific worry about an AT1 ETF: that a bunch of flighty retail investors will run for the hills at the slightest sign of trouble, injecting an unwelcome dose of disruption into the underlying AT1 market – which is already at times a feverish place?

It’s not so long ago (2016) that the whole edifice seemed to be collapsing around Deutsche Bank’s ears

A few things here. The concept of a liquid security being backed by illiquid underlying assets is by now rather commonplace. People who get jumpy about it have probably forgotten that this is what ordinary shares are, at least in part. Investment adviser Richard Bernstein – a former chief investment strategist at Merrill Lynch and now running his own firm – has a neat line on that: “A fear that ETFs might be more liquid than their underlying assets seems to ignore the long list of financial assets that are more liquid than their underlying assets.”

The fact that the ETF shares will potentially behave in a much more volatile way than the underlying is a reflection of the secondary liquidity offered by the listed asset class: it’s a lot of the point of it.

But here’s the thing: if APs regularly have to play in the underlying market in order to source inventory to allow them to intervene in the ETF via the creation of new shares (to dampen the ETF value), then surely some of this ETF ‘volatility’ is being translated into the underlying market?

That’s sort of true, but volatility might be the wrong word here. That kind of transmission looks a lot like price discovery, which is again a lot of the point of an ETF – the way that the iShares investment-grade and high-yield corporate bond ETFs effectively replaced the underlying when credit markets shut down during the Lehman Brothers crisis is just the most obvious example of that in action.

The ETF is reflecting actionable risk, in the words of one ETF professional, when the benchmark index is not actionable and nor is the NAV.

“If you tell me that the ETF is introducing volatility, then I would say it is indicating where the bonds would be trading if they were trading,” he says.

AT1 traders point out that APs are more likely to struggle to sell if there were big cash outflows on the ETF than they are to struggle to buy in the reverse situation. But one also notes that in practice the bid-offer would “go crazy”, helping to mitigate a run.

And finally – yes, the arbitrage might exist for a while. If it can be narrowed efficiently by the APs it will be, but if it can’t, or can only be closed slowly, then that can also happen. The fact that primary and secondary ETF liquidity are distinct things is why.

What’s left? Leverage, say ETF folk. All this theorising could be blown out of the water if the ETF was levered. So that’s a worry, perhaps.

Who’s buying anyway?

The worries have tended to coalesce around potential retail involvement. But a legitimate question is whether or not retail would even be the target market for something like this at all.

A lot may not be allowed to buy it. The UK’s Financial Conduct Authority (FCA) has a prohibition against marketing contingent convertible bonds (a big part of AT1 capital) to retail investors, and although Euromoney is still awaiting a definitive answer from the FCA on how it would treat an AT1 ETF, its view (from the results of its original consultation) seems pretty clear:

We consider that investment in a fund that invests wholly or predominantly in CoCos, while offering the potential for some diversification among issuers, still presents serious evaluative challenges for retail investors, including unusual market contagion risks which have not yet been tested by the market.

The background to the ban – and other similar pronouncements, such as the European Securities Markets Association’s warning to investors about CoCos – is to do with the peculiar risks of contingent convertible bonds, which can enter a self-perpetuating spiral down in value as conditions approach the point at which they would convert into equity.

Not just that, but investors holding paper at that time are likely to be receiving equity at precisely the time they don’t want it. Coupon deferral, principal risk, it’s all there and it’s why some regulators have got jumpy. Others, like France’s AMF, while not banning the product outright for retail investors, nonetheless require extra risk management procedures and disclosures.

The issue that a new ETF product would cause for a regulator like the FCA is that there is a loophole in its own prohibition for any products structured under a Ucits (Undertakings for Collective Investment in Transferable Securities) wrapper – a standardised form of regulated fund in Europe.

Without a change in the FCA’s wording, it would appear that an AT1 ETF could be freely bought by retail investors if it were packaged as a Ucits fund. It would need to comply with the various Ucits protections – but fund management professionals don’t think this would be difficult.

Institutional investors, or those retail investors deemed sufficiently sophisticated, already have plenty of options for exposure to AT1 securities via the increasing number of funds that have tweaked their mandates to allow a holding in the asset class, or are even specialising in it. Old Mutual is just one firm to have launched a dedicated CoCo fund last year.

There is another option, though: since Goldman and JPMorgan kicked it off late last year, a total return swap market on AT1 indices has grown swiftly, giving some investors a way to bet on AT1 performance while others use them to hedge AT1 bonds whose coupon deferral features make them unsuitable to be hedged via credit default swaps.

But anyone writing those total return swaps also needs a hedge – and a dedicated ETF might be just that.