The strange effect of illiquidity on bond ETFs

I spotted some very strange goings on among fixed income exchange-traded funds (ETFs) last week. My LinkedIn post about them has gone viral (over 36,000 views at the time of writing), suggesting I am not alone in finding these developments thought provoking. They have brought to the fore some of the liquidity problems that have become an issue in credit markets since the financial crisis.

When you put together an asset which is somewhat illiquid and a vehicle which offers instant liquidity, you have the potential for things not to behave as you might expect.

What has happened?

The prices of all the big fixed income ETFs have started trading at surprisingly large discounts to their net asset values (NAVs). This is not supposed to happen. 

ETFs are liquid, tradable vehicles which are designed to track a basket of underlying securities, in this case bonds. For this reason, the price and the NAV should move in tandem with each other. If there is a gap, it should be possible for market makers to step in and buy the ETF while simultaneously selling the basket of underlying bonds, thereby banking a risk-free profit (known as arbitrage). If markets are efficient, it shouldn’t be possible to earn risk-free profits. This disconnect between prices and NAVs suggests something has gone awry.

The chart below illustrates the shortfall in underlying investment grade corporate bonds and ETFs, which is 5.0%, according to Refinitiv data.

But it’s not just investment grade corporate bonds that have been affected. Shortfalls cropped up in lots of areas at various points in the past week. For example, leveraged loans (3.8% shortfall) and emerging market debt (4.8% shortfall). Even Treasuries have not been immune (5.0% shortfall).

iShares iBoxx $ Inv Grade Corporate Bond ETF

Source: Schroders. Data from Refeinitiv correct to 12 March 2020.

What is behind this?

The root of this problem lies in the fact that liquidity in corporate bond markets has deteriorated considerably compared with the pre-financial crisis years – in simple terms, it is a lot harder to buy and sell corporate bonds than it used to be.

The credit market is a complicated beast. Unlike stock markets, where each company normally has a single share-class (or sometimes two), companies issue many different varieties of bond. They vary in issuance date, maturity, coupon, covenants, position in the capital structure, size and in other ways. For example, Bank of America, has more than 700 bonds outstanding.

This leads to fragmentation of trading across an issuer's bonds and contributes to more limited liquidity in each individual issue. Also, around 6-12 months after a bond is issued, trading levels drop precipitously. Prices for individual bonds which are infrequently traded may need to be inferred from the prices of similar bonds which have traded. An element of estimation is required.

Furthermore, unlike equities, which are traded on an exchange, credit is mostly traded over-the-counter via a network of broker-dealers. However, post-crisis regulatory changes have deterred these market makers from holding bonds as inventory on their balance sheets. Now if you want to sell a bond, a broker-dealer may be reluctant to buy it from you unless they know that they can easily offload it to another investor. The shock absorbers have been removed from the system.

What makes matters worse is that this reduction in the willingness of broker-dealers to provide liquidity has occurred at the same time that the size of the corporate bond market has ballooned.

The shock absorbers have been removed from the system

Source: Federal Reserve, Schroders. Data to 4 March 2020

Liquidity conditions have deteriorated. Trading volumes have failed to keep pace with the growth in the market and turnover has fallen. Liquidity, as measured by turnover, has fallen by 43% from its pre-crisis peak.

Corporate bond turnover has collapsed

12-month average daily traded volume / market value outstanding

Source: FINRA TRACE, Refinitiv, Schroders

How does this relate to ETFs?

There are two ways that this feeds into the odd behaviour in fixed income ETF-land:

• If lots of investors try to sell ETF holdings at once, but the market makers are unable to sell assets quickly enough to meet those withdrawal requests, then the price could fall below the asset value. In other words, if you want your money back immediately, you may be forced to take a haircut on what you get back. This has often been put forward as a problem with any vehicle where there is a mismatch between the liquidity offered by the vehicle and the liquidity of the underlying assets. In this view of the world, the problem is with the ETF.
• Infrequent trading of underlying bonds means that the price of the underlying bonds that the NAV is based on may not be the price you would get if you actually tried to buy or sell. This prevents the arbitrage mechanism from functioning. In this view of the world, the problem is with the bond prices, not the ETF. In this view of the world, the ETF price is a fairer reflection of prices than the NAV.

The reality is that it is probably a bit of one and a bit of the other. We know that corporate bond liquidity has deteriorated. We also know that parcelling up illiquid assets in vehicles which offer instant liquidity has the potential to lead to problems.

This is not exclusive to ETFs. Mutual funds which invest in credit assets are also exposed to the poorer liquidity in credit markets. Waves of redemptions could lead to challenging conditions for everyone. This is not a new development. All that is new is that markets have suddenly come alive to the risk.  Arguably credit spreads should have incorporated a higher liquidity premium for much of the past decade. But in a world where investors have been cossetted by central banks, this risk has been under-priced. No more.

The broader point is that we can expect to see more oddities emerge in the coming days, weeks and months. Financial ingenuity has brought forth a plethora of vehicles, structures and investment products to meet every possible demand from end investors. Most have never been tested in a volatile market environment. There are likely to be more cases where investors find their payoff is not quite what they were expecting or hoping for.  

As pointed out in a comment by my Australian colleague, Avik Roy, in a shared version of my original LinkedIn post: “Liquidity of the vehicle doesn't matter if the underlying has issues. Hope investors understood what they were buying.”