Broken pencil

Remember last November when the Kauffman Foundation released a report that falsely implied, among other things, that exchange traded funds were responsible for the May 6th Flash Crash?  Well, Kauffman is back, and this time with a new report that is basically more of the same – namely, an inaccurate and misguided attempt to create fear about the structure of ETFs.  While their ultimate recommendation for regulators to reduce the number of fails might be well-intentioned, the path they took to get there is crooked and no thought seems to have been given to unintended consequences. But maybe they are saving that for their next report.

At the highest level, this new study is focused on the large and growing number of failed settlements in ETFs and mortgage-backed securities (MBS), meaning one party in the trade has failed to deliver the security within the standard 3-day timeframe to the other party in the trade. They then refer back to their first report where they implied that ETFs could implode due to phantom ETF trades being redeemed for real securities. This is where things begin to go off the tracks for me.   I can’t speak to the MBS market, but I do have some insight to share on the ETF marketplace relative to this report. Because the “researchers” once again play fast and loose with terminology, we have to begin by untangling what supposed problem we’re trying to solve.

Let’s start with some ETF 101.  ETFs are created in the primary market.  When an ETF share is “created” or “redeemed” (not “destroyed” as erroneously labeled in Kauffman #1), an Authorized Participant, or AP, delivers shares of the underlying securities for shares of the ETF, or vice versa.  If an AP is unable to deliver the necessary securities by the settlement date (typically trade date plus 3 days or T+3), the settlement is considered “failed.”  However, the whole trade itself has not failed as the name implies – rather the AP has been required to post collateral in order to delay delivery of the securities, which means the investors in the fund are protected until delivery occurs.  In our experience the occurrence of TRUE failed trades in an ETF create or redeem (meaning the AP’s collateral must be used to cover their failure to deliver securities) is virtually non-existent.  There are several logical reasons why an AP might have to delay delivery.  One of the most basic reasons is that an international market might be closed when international ETFs in the US are open for trading.

Now let’s talk about secondary market trading.  ETFs make up a relatively high volume of trades (about 23% of daily trades in the US), so it stands to reason that they’d represent a higher proportion of failed settlements.  Also on the secondary market, you have market makers who are permitted to settle on T+6 in order to facilitate their role in creating fair and efficient markets, which might be skewing the data.  Unfortunately, the failed trade reporting we see doesn’t differentiate between the age of the fail or the entity involved, it just notes the event. Given that, we have to question the validity of the data used in this report.

I completely agree with the premise that we should eliminate logical risks wherever we can.  ETFs are subject to the same SEC settlement rules as any stock in the equity market (the report mistakenly claims they are not).  So if the SEC wants to create stricter guidelines and punishments for ETF settlement failures, they can, and I might even agree that they should.  But first, I’d want to spend more time fixing the failed trade reporting and analyzing that data to see if something is even wrong.  I’m pretty sure there is a “Contact Us” link somewhere on our website.  Maybe someone at the Kauffman Foundation should make use of it before they release their third report.