The Good ETF

What makes a good ETF in the long term?  My, what a question, driven by the ETFs challenging the limits of what is prudent.  Maybe it is easier to start with what makes a bad ETF, then:

  • Headline risk can be eclipsed by credit risk.  All ETNs, Currency ETFs, and ETFs that use non-exchange-traded swaps, sometimes for commodity funds, take credit risk.  Did you know you were taking credit risk?
  • Roll risk — for commodity funds, trying to replicate the returns of the spot market using the futures market works only when there aren’t a ton of funds trying to do so.  The flood of funds into front month futures contracts incites other funds to front-run the activity, capturing the profits that the commodity funds were trying to make.  (For storable commodities, better to take delivery and store.)
  • Market size risk — an ETF can become too large relative to liquidity or regulatory constraints of the market, and it no longer tracks its benchmark well — again, mainly a commodity fund problem.
  • Irreplication risk — This is mainly a bond market theme, but once the ETF defines the index, only index bonds can be bought in proportion to the index.  I ran into this personally in 2002, when I ask ed why a certain bond traded rich.  The answer came that it was in a common index, but it was a small bond issue in proportion to its weight in the index.  Many investment banks were short the note to provide liquidity, but could not source the bonds to cover the short because most were in index funds.  I would keep an eye out for those bonds, and would sell them to those short for a small markup when I found them.  For ETFs, the trouble is that arbitrage can’t take place, because bond buyers can’t find certain rare bonds in order to create new units in exchange for expensive ETF shares.  That is one reason whey NAVs get stretched versus market prices.
  • Abnormal or faddish theme — the risk is that they become too dominant in the trading of less liquid companies in their ETFs.  But away from structural risks is the faddish investment risk.  The ETF only gets created as the fad is about to go into decline.

In one sense, the market can reward non-consensus views, particularly when they are small compared to their relative advantage in their sub-markets.  In the same way, the market can punish those that become too large for the pond that they swim in.  Growth will be limited or negative.  Even the efforts to create more capacity, create it at the cost of credit risk.

Good ETFs are:

  • Small compared to the pool that they fish in
  • Follow broad themes
  • Do not rely on irreplicable assets
  • Storable, they do not require a “roll” or some replication strategy.
  • not affected by unexpected credit events.
  • Liquid in terms of what they repesent, and liquid it what they hold.

The last one is a good summary.  There are many ETFs that are Closed-end funds in disguise.  An ETF with liquid assets, following a theme that many will want to follow will never disappear, and will have a price that tracks its NAV.


  • Quints says:

    This is excellent. I read every article you write as soon as I can. You consistently provide value by having a position and explaining your reasoning.

    Keep up the great work! You have helped me tremendously.

  • Bob_in_MA says:

    That all makes perfect sense, but doesn’t it basically rule out most commodity ETFs?

    I’m a very conservative investor and I’ve been looking for some inflation hedges and, in a less sophisticated way, found most of the commodity ETFs too problematic. I’m thinking maybe royalty trusts are probably a safer way to go (but a pain in the butt at tax time.)

  • Quints — thanks ever so much.

    Bob, remember my fusion solution piece?

    This would save most commodity ETFs. They would track spot commodity moves over the long haul, but in the short run, the correlation with spot commodities would be diminished.

    That said, any investment strategy that becomes too popular will eventually underperform. That is the way of the world. Look for productive ideas that others are not following.

    The royalty trusts are not a bad idea, just keep the position sizing reasonable to account for downside risk.

  • bram says:

    A few thoughts on the piece (I generally agree with it, but have a few comments)

    Not sure I completely agree with #1. Or at least, not for the reason mentioned. An OTC swap could give rise to credit risk if there is no CSA in place between counterparties, rather than that this has to do with something being exchange traded I would say?

    As for the roll risk, an interesting comparison might be commodity indices such as for example the
    DJAIG or the SPGSxxx indices. ETFs should be more or less similar to the total return variants oof those. Now most of these index providers started out with rolling 1st future strategies. More recently however, they seem to be marketing a sort of “constant maturity” versions of those where they actually spread theire exposure over the curve and only roll the front aprt of it regularly. If there aren’t ETFs doing sometihng like this already, I could imagine that this type of innovation is something that is to be expected for ETFs too.

    Irreplication risk and roll risk sort of feel like the same type of risk to me but in different mmarkets. They both have to do with incurring a cost (be it a roll cost or a bond trading rich) because of self-induced demand by a succesfull ETF.

    As for faddish risk, I don’t consider that to be a risk associated with investing in ETFs themselves (though it is a general investment risk). Investing in a hype/bubble/fad might be poor judgment, but resulting losses from them don’t have to with ETF specific risk; unless you consider it a risk that ETFs might liquiditate themselves because after a fad they don’t have enough AuM to be economical ; however those should be able to liquiditate at (more or less) NAV. The part where an ETF is a relative big holder of underlyings in a specific part of the market would be more part of the market size risk
    and not faddish risk I’d say.

    As for market size risk, there are two sorts of risk that have recently shown themselves (or at least as far as my understanding goes, but I’m definitely no expert). The first is an ETF becoming so popular that it becomes so big that the SEC doesn’t allow for it to create more fund units (I think this happend to UNG recently). The ssecond is that an ETF becomes a sizeable player in the market and the entire market knows in what position the ETF is (relates both to the roll risk and your irreplacbility risk).