Wie funktionieren Anleihen-ETFs

[Ramstein]

Bei Bloomberg gibt es einen interessanten Artikel. Bevor der möglicherweise hinter der Bezahlschranke verschwindet, möchte ich die Informationen hier festhalten.

 

Zitat

It’s Hard to Run on Bond ETFs

By Matt Levine

1. März 2021, 18:31 MEZ

 

People are worried about bond market liquidity

We have talked a lot around here, over many years, about bond market liquidity and bond exchange-traded funds. People were worried about bond market liquidity, and they were worried that bond ETFs were somehow bad for it. For instance, last March, there was a lot of stress in the bond market, and some bond ETFs traded at discounts to their net asset value, and we talked a lot about that. I sort of thought we were out of things to say about bond market liquidity. But the March BIS Quarterly Review from the Bank for International Settlements has a cool special feature by Karamfil Todorov on “The anatomy of bond ETF arbitrage” that actually has something new and interesting to say about bond ETFs and bond market liquidity.

 

The way an S&P 500 exchange-traded fund works is roughly that the fund holds a big pile of all the 500ish S&P 500 stocks, and if anyone wants to buy or sell shares of the ETF they can trade with each other on the stock exchange. Sometimes people want to buy more shares than are for sale on the exchange, and so they will want to buy new shares from the ETF itself. But you can’t really buy new shares in the ETF, like you could in a mutual fund (you give the fund cash, it buys the underlying stocks, it gives you back shares). Instead there is a “creation” mechanism: There are firms called “authorized participants” (big broker-dealers and market makers) who go out and buy the underlying stocks and deliver them to the ETF in exchange for new shares of the ETF. An authorized participant will buy $10 million of S&P 500 stocks, hand them to the ETF, and get back $10 million of ETF shares. 

 

Similarly sometimes a lot of people will want to sell ETF shares, and an authorized participant will do a redemption trade in which it hands ETF shares back to the ETF and gets, not cash, but a basket of underlying stocks. (Which it can then sell for cash.) This creation/redemption mechanism—in which the ETF doesn’t buy or sell the underlying shares itself, but does in-kind trades with authorized participants—is good for the ETF’s expense efficiency (it doesn’t explicitly pay to trade) and for its tax efficiency (it doesn’t realize taxable gains). It is also good for the ETF’s pricing accuracy: If the ETF trades at a premium to the underlying stocks, there is an obvious arbitrage (buy the stocks, deliver them to the ETF for creation, get back ETF shares and sell them at a premium), which should keep the ETF’s price in line with the underlying index.

 

This is basic obvious stuff, but it isn’t really true for bond ETFs. For an S&P 500 ETF you can go buy all the S&P 500 stocks fairly efficiently and electronically, but for a 500-bond ETF it would be very difficult to go buy 500 bonds. The BIS explains:

Zitat

 

Whereas for equity ETFs baskets are usually almost identical to holdings, for bond ETFs they are systematically different and include a small share of the bonds in the actual holdings, eg less than 3% for the largest bond ETF. For bond ETFs, baskets also change significantly from day to day and creation baskets tend to have longer duration and higher liquidity than redemption baskets.

Several factors are behind this contrast between equity and bond ETFs. First, the nature of the underlying assets is different. Compared with equities, bonds are generally less liquid and trade in a market with fewer potential buyers and sellers. In addition, bonds mature, whereas equities do not. Second, the minimum trading amount of bonds is much larger than that of equities, which constrains the feasible trades. Given these specificities of the bond market, ETF sponsors need flexibility as regards the composition of baskets. Sponsors choose strategically which bonds to include among the available ones, with an eye on continuously matching key characteristics of the benchmark index. Likewise, APs influence the composition of baskets and could use them to accommodate demand from their own clients rather than to close arbitrage gaps.

 

 

There are three points here. One is that it is not, like, a structural requirement of ETFs that the creation and redemption baskets be a perfect sample of the underlying ETF: You could in theory do a creation trade with an S&P 500 ETF in which you give the ETF $10 million of only Tesla Inc. stock and get back $10 million of ETF shares, and in fact sometimes stock ETFs do handle index additions and removals that way. With bond ETFs that is just the normal approach: You could never give the bond ETF all the bonds, so any creation or redemption trade will involve some skewed sample.

The second point is that the ETF sponsor—the ETF itself—will want the creation and redemption baskets to be good for the ETF. So, for instance, it will want to get rid of short-maturity bonds (to avoid having them mature and having to reinvest the cash) and add longer-maturity bonds, so the creation basket will be longer-dated than the redemption baskets. It will ask its authorized participants to bring it the bonds it wants to create shares, and to take away the bonds it doesn’t want to redeem shares.

The third point is that the authorized participants will want the creation and redemption baskets to be good for them. Many APs are big banks and market makers who want to do portfolio trades for customers. A customer will come to a bank and say “I have these 30 bonds that I want to sell, how much will you pay for them?” And the bank will say “well I would pay a lot for them if I could just squeeze them into an ETF, get back ETF shares, and sell the ETF shares quickly on a liquid stock exchange.” So the bank will pick the most relevant bond ETF—if the customer has junk bonds, it will pick a junk-bond ETF, etc.—and call up the ETF and ask “hey would you take these 30 bonds as a creation basket?” And the ETF will look at the list and say “ahh that’s close enough to our index, sure, wave it in.” And a trade will get done. The BIS writes:

 

Zitat

ETF sponsors' portfolio optimisation and their incentives to maintain a long-term relationship with APs can lead to differences between baskets and holdings and to changes in baskets over time. Sponsors would adapt the composition of baskets based on the availability of bonds and would choose a subset of bonds that minimises tracking error. In turn, when an AP cannot deliver a bond ... it could propose some similar new bond ... that is easier to locate for the transaction and could even allow the AP to absorb a supply shock from its clients. While this bond is not part of the ETF holdings, a sponsor might accept the proposal if the new bond keeps the tracking error in check and helps maintain the relationship with the AP, whose market-making function provides valuable services to the sponsor.

 

There is a negotiation: The ETF wants a certain basket, the APs want a different basket, and they work together to get a basket they can both live with. 

Here’s what this has to do with bond market liquidity. People worry about bond ETFs creating a “liquidity illusion”: You can trade bond ETFs easily on the exchange, which makes you think they are more liquid than they are, but if everyone wants to redeem at once then there will be forced selling of the underlying bonds and a horrible death spiral.

With regular mutual funds, there is some basis for this worry. If everyone wants to redeem out of a bond mutual fund, they will all go to the mutual fund and ask for their money back. The mutual fund will have to sell bonds to raise money to give back to them. It will sell its most liquid bonds first, because they are easier to sell quickly to raise money. This will tend to leave the mutual fund with worse, less liquid bonds. Other investors will see this, realize that the mutual fund is getting worse, and also demand their money back. The fund will have to sell less liquid bonds, at a bigger discount, driving down its asset value and leading to more redemptions, etc.

With ETFs, though, the mechanism is different, perhaps even the opposite. If everyone wants to redeem out of a bond ETF, they will sell shares to an authorized participant, who will hand them back to the ETF sponsor. The ETF sponsor will hand the authorized participant back a chunk of its bonds. But it will hand them the worst bonds, the illiquid ones that it would have trouble selling, and keep the liquid ones for itself. From the BIS: 

 

Zitat

By selecting the composition of baskets, ETF sponsors could discourage runs by influencing the desirability of redemptions (Shim and Todorov (2021b)). If there is excessive selling of ETF shares in the secondary market, which puts redemption pressure on APs, the ETF sponsor can include only the riskier or less liquid securities from the pool of holdings in the redemption basket. The lower-quality bonds that APs obtain after redeeming ETF shares would in turn reassure non-running investors that their shares are now backed with holdings of higher average quality. This would discourage further runs and lead to ETF discounts during run episodes. In fact, such a stabilisation mechanism was arguably in place during the March–April 2020 episode … when some ETFs traded at a discount while redeeming baskets that were more illiquid than the holdings.

 

That is, according to the BIS, the reason that bond ETFs traded at a discount to the underlying bonds during last spring’s bond crash really is that the ETF arbitrage mechanism broke down, but in a good way, in an intended way, in a way that makes bond ETFs more robust and less prone to runs. In times of stress, bond ETFs will trade below net asset value, but the redemption mechanism means that you kind of can’t get your money back, or not efficiently, so you don’t redeem (that is, authorized participants don’t redeem to arbitrage away the discount), so there is no run on the ETF, so there is no need for it (or its authorized participants) to dump bonds at fire-sale prices, so there is no broader collapse in bond prices and no contagion. The ETF trades below its net asset value because it is, in effect, absorbing stress on the bond market, rather than transmitting it.

 

[oktavian]

Ein theoretisches Problem dieser Struktur wäre illegale Bereicherung des AP mit Bestechung von Mitarbeitern des ETF, um bessere baskets zu bekommen. Das kann man kaum erkennen im nachhinein. Als ETF Anleger muss man da eben etws Vertrauen haben. Ich glaube ETF Anleger würden das nie bemerken. Wenn es zu heftig wird, würde man es vielleicht an der underperformance des ETF erkennen.

 

Desweiteren ist es nachteilig, wenn man nicht steuern kann, wann man Liquidität benötigt. In turbulenten Marktphasen sollte der discount zum NAV für bond ETFs durchaus heftiger ausfallen als bei Aktienetfs, da das unerlying (bonds) wesentlich illiquider als Aktien ist. Als Kleinanleger hat man noch nicht einmal Zugriff auf den realtime NAV des ETF und der NAV ticker ist sowieso ein paar Sekunden verzögert. Der NAV unterliegt wegen des weiteren bid-ask spreads im Vergleich zu Aktien ohnehin einer höheren Ungenauigkeit. Also wenn ich jeden Monat cash benötigen würde, würde ich evtl. immer schon puffern und früher den Entnahmeplan durchführen und cash halten. Besonders bei den niedrigen bzw. negativen Zinsen, verliert man dadurch kaum und ist eben vor Überraschungen etwas sicherer, wenn man davon ausgeht, dass Liquidität nur temporär einbricht (wie z.B. im corona induzierten crash letztes Jahr. Ich hätte gerne einen Email Alarm wenn bestimmte ETFs xx% unter NAV notieren und habe es nicht kostenlos hinbekommen. Könnte man mit verzögerten Kursen programmieren. Auf jeden Fall ist mir das ganze ETF Zeug dann zu komplex gewesen und ich setze nicht auf Anleihen ETF, aber würde auch einsteigen mit discount zum NAV, aber eben nicht zum NAV oder was-auch-immer ich durch Sparplan/Börse bekomme. Sind aber alles bekannte Mechanismen.